THE CONTROLLING FACTORS INVOLVED IN BIOMASS ...
-
Upload
khangminh22 -
Category
Documents
-
view
0 -
download
0
Transcript of THE CONTROLLING FACTORS INVOLVED IN BIOMASS ...
THE CONTROLLING FACTORS INVOLVED IN BIOMASS AQUEOUS PRETREATMENT:
FUNDAMENTALS TO APPLICATIONS
By
LIBING ZHANG
A dissertation submitted in partial fulfillment of
the requirements for the degree of
DOCTOR OF PHILOSOPHY
WASHINGTON STATE UNIVERSITY
Department of Biological Systems Engineering
JULY 2016
© Copyright by LIBING ZHANG, 2016
All Rights Reserved
ii
To the Faculty of Washington State University:
The members of the Committee appointed to examine the dissertation of LIBING ZHANG
find it satisfactory and recommend that it be accepted.
Bin Yang, Ph.D., Chair
Shulin Chen, Ph.D.
Manuel Garcia-Perez, Ph.D.
Melvin Tucker, Ph.D.
Hong-fei Wang, Ph.D.
iii
ACKNOWLEDGEMENTS
I would like to thank the agencies for funding my research through research grants,
including: Jet Fuel Production from lignin in Remote Locations – DOD DARPA YFA ,
Pretreatment of Sub-millimeter Biomass Particles – DOE-SBIR Phase I & II, Upgrading Lignin
to Aromatic Hydrocarbons- DOE-NREL, EAGER: Aqueous Phase Catalytic Processing of
Lignin to Hydrocarbons- NSF, and Catalytic Production of Aviation Fuel Hydrocarbons from
Lignocellulosic Biomass-Derived Lignin – Sun Gant- DOT. Also, I would like to thank the
Chinese Scholarship Council for providing the important Graduate Student Scholarship for me to
complete my thesis.
I sincerely appreciate my advisor, Professor Bin Yang, for his continued encouragement and
advice throughout these years. I would also like to acknowledge my committee members Dr.
Hongfei Wang, Dr. Melvin Tucker, Dr. Manuel Garcia-Perez, and Dr. Shulin Chen for their
supports in my research and PhD studies.
I am grateful to Drs. John R. Cort, Zheming Wang, Li Fu, Luis Velarde, Zhou Lu, Satish
Nune, and Shunli Chen from Pacific Northwest National Laboratory; Drs. Yuqiao Pu and Art J.
Ragauskas from BioEnergy Science Center; Dr. Shi-You Ding from Michigan State University;
and Dr. Jim Dooley and Lead Mechanical Engineer David Lanning from Forest Concepts, LCC
for their support in my PhD studies. Also, I would like to thank Marie S. Swita, Steve Jordan,
Daniel T. Howe, and Teresa Lemmon for helping me with my PhD research. I would like to
acknowledge Dr. Jie Xu and Mr. Dmitry Gritsenko from University of Illinois at Chicago for
their help in mass transfer modeling development.
iv
I’d like to specially thank the Biological Systems Engineering program at Washington State
University for providing me with the platform to develop research and leadership skills.
Additionally, I would like to give my appreciation to all my research colleagues for their helpful
discussions on the many research topics we collaborated on and valuable sharing of their career
development.
At last, I really want to thank my family members for their endless love and support,
especially my parents, Fuqing Zhang and Shuming Liang, for their encouragement and my
beloved husband, Dan C. Berghofer, for his continuous support.
v
THE CONTROLLING FACTORS INVOLVED IN BIOMASS AQUEOUS PRETREATMENT:
FUNDAMENTALS TO APPLICATIONS
Abstract
by Libing Zhang, Ph.D.
Washington State University
July 2016
Chair: Bin Yang
Of all sustainable resources, only lignocellulosic biomass can be transformed into organic
fuels and chemicals that can integrate well with our current transportation infrastructure due to
the inherent convenience, cost, and efficiency advantages of these organic fuels over current
fuels. However, the high cost of biofuel conversion technologies poses challenges in large-scale
commercialization. Pretreatment is the most expensive operation unit that is responsible for
disrupting the naturally recalcitrant lignocellulosic biomass to provide reactive intermediates for
production of renewable biofuels and bioproducts through biochemical or thermochemical
processes. A better fundamental understanding of biomass pretreatment is essential to bring
biofuels to market.
The goal of this dissertation is to study the dissolution chemistry of hemicellulose, cellulose,
and lignin for hardwood and softwood during flowthrough and batch pretreatment at
temperatures of 140 °C to 270 °C, initial pHs (pH 2-12), and flow rate of 0-25 mL/minute. The
complete solubilization of hardwood was achieved to produce insoluble lignin with high purity
and mild structural modification (i.e. changes of side chains and Cβ-C5). However, ~30%
softwood lignin was found to be undissolvable and was collected for analysis the first time.
Based on wet chemistry and NMR results, unlike hardwood, dilute acid flowthrough
pretreatment of softwood led to vigorous C-C5 recondensation resulting in undissolvable lignin
vi
structures. On the other hand, High Resolution Sum Frequency Generation Vibrational
Spectroscopy (SFG-VS) was developed to selectively and in-situ characterize cellulose for the
first time and new structural information was observed. The implementation of the simulation of
aqueous pretreatment of cellulose using a heated fluid test bed on cellulose and dynamic SFG-
VS analysis indicated a cellulose recrystallization hypothesis, leading to new insights on the
effectiveness of biomass pretreatment. In addition, Total Internal Reflection (TIR)-SFG-VS was
invented and tested to selectively characterize the cellulose surface of Avicel and the cellulose Iβ
crystalline particles. At last, the results of the effects of the biomass comminuting approach on
aqueous pretreatment indicated the associations among particle sizes, cutting methods, mass
diffusion and sugar yields in pretreatment and enzymatic hydrolysis. The findings of this
dissertation provide important fundamental insights to establish the effective biomass
pretreatment.
vii
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS ........................................................................................................... iii
ABSTRACT .................................................................................................................................... v
TABLE OF CONTENTS .............................................................................................................. vii
LIST OF TABLES ....................................................................................................................... xvi
LIST OF FIGURES ..................................................................................................................... xix
CHAPTER ONE
INTRODUCTION .......................................................................................................................... 1
1.1 Biochemical conversion of biomass to bioethanol ................................................................ 2
1.2 The essential role of aqueous biomass pretreatment ............................................................. 4
1.3 The key problems in aqueous pretreatment ........................................................................... 5
1.4 Thesis objectives and hypothesis........................................................................................... 6
1.5 Thesis organization ................................................................................................................ 8
CHAPTER TWO
FUNDAMENTALS OF BIOMASS AQUEOUS PRETREATMENT ..........................................11
2.1 Abstract .................................................................................................................................11
2.2 Introduction ..........................................................................................................................11
2.3 Plant cell wall macrostructure ............................................................................................. 12
2.4 Biomass aqueous pretreatment research progress ............................................................... 13
2.4.1 Hot water and dilute acid pretreatment ......................................................................... 16
2.4.2 Alkaline pretreatment .................................................................................................... 18
viii
2.4.3 The flowthrough aqueous pretreatment ........................................................................ 19
2.4.4 Summary of biomass degradation chemistry in aqueous pretreatment ........................ 23
2.5 Process characterization of flowthrough aqueous pretreatment .......................................... 28
2.5.1 Cellulose characterization in aqueous pretreatment ..................................................... 28
2.5.2 Lignin production and characterization in aqueous pretreatment ................................. 31
2.6 Preprocessing of feedstock for aqueous pretreatment ......................................................... 45
2.7 Conclusion ........................................................................................................................... 48
CHAPTER THREE
CHARACTERIZATION METHOD DEVELOPMENT FOR CRYSTALLINE CELLULOSE
USING HIGH RESOLUTION BROADBAND SUM FREQUENCY GENERATION
VIBRATIONAL SPECTROSCOPY IN BIOFUEL ..................................................................... 49
3.1 Abstract ................................................................................................................................ 49
3.2 Introduction ......................................................................................................................... 50
3.3 Materials and methods ......................................................................................................... 57
3.3.1 Cellulose samples.......................................................................................................... 57
3.3.2 HR-BB-SFG-VS ........................................................................................................... 58
3.3.3 Polarization Null angle measurement with picosecond scanning SFG-VS .................. 60
3.3.4 Atomic force microscopy (AFM).................................................................................. 60
3.4 Results and discussion ......................................................................................................... 61
3.4.1 Traditional Raman in cellulose characterization ........................................................... 61
3.4.2 Polarization Null Angle (PNA) testing ......................................................................... 62
3.4.3 HR-BB-SFG-VS spectra of cellulose Iα, Iβ and Avicel samples .................................. 64
ix
3.4.4 Morphology of cellulose Iα and cellulose Iβs from AFM image .................................. 69
3.4.5 Unique O-H signatures for cellulose Iα and cellulose Iβ .............................................. 71
3.4.6 CH Spectral features for cellulose Iβ samples from different sources .......................... 74
3.4.7 Characterization of Avicel probed by HR-BB-SFG-VS ............................................... 77
3.5 Conclusion ........................................................................................................................... 79
3.6 Acknowledgements ............................................................................................................. 80
CHAPTER FOUR
TOWARD DYNAMICAL UNDERSTANDING OF FUNDAMENTALS OF PRETREATMENT
AND ENZYMATIC HYDROLYSIS OF CELLULOSIC BIOMASS VIA A SFG-VS ............... 82
4.1 Abstract ................................................................................................................................ 82
4.2 Introduction ......................................................................................................................... 84
4.3 Materials and methods ......................................................................................................... 91
4.3.1 SFG-VS set-up .............................................................................................................. 91
4.3.2 Design of SFG-VS geometries to observe cellulose surface layers .............................. 92
4.3.3 The fluid heating cell system development................................................................... 94
4.3.4 Cellulose sample preparation and heating program ...................................................... 95
4.3.5 Molecular observation of cellulose structural changes in enzymatic hydrolysis using
SFG-VS .................................................................................................................................. 96
4.4 Results and discussion ......................................................................................................... 97
4.4.1 Structural characterization of cellulose surface ............................................................ 97
4.4.2 The characterization of thermal behaviors of dry cellulose by SFG-VS .................... 105
4.4.3 Structural changes of wet cellulose during heating process ........................................110
x
4.4.4 The different SFG-VS signals of dry and wet Avicel bulk crystals .............................113
4.4.5 Study of the temperature dependence of dry Avicel surface layers .............................114
4.4.6 Molecular structural changes of cellulose in enzymatic hydrolysis using SFG-VS ....116
4.5 Conclusion ..........................................................................................................................119
4.6 Acknowledgements ........................................................................................................... 121
4.7 Supplementary materials ................................................................................................... 122
CHAPTER FIVE
CHARACTERIZATION OF FLOWTHROUGH SOFTWOOD AQUEOUS PRETREATMENT
UNDER NEUTRAL AND ALKALINE PH AT ELEVATED TEMPERATURES .................... 129
5.1 Abstract .............................................................................................................................. 129
5.2 Introduction ....................................................................................................................... 129
5.3 Materials and methods ....................................................................................................... 132
5.3.1 Feedstock .................................................................................................................... 132
5.3.2 Flowthrough pretreatment system............................................................................... 132
5.3.3 Flowthrough pretreatment of pine wood chips ........................................................... 133
5.3.4 GC/MS analysis .......................................................................................................... 135
5.3.5 Enzymatic hydrolysis on pretreated whole slurries .................................................... 135
5.3.6 2-D 1H -13C HSQC NMR of ball milled pine wood lignin and pretreatment obtained
RISL ..................................................................................................................................... 135
5.4 Results and discussion ....................................................................................................... 136
5.4.1 pH changes of pretreatment solvent after pretreatment .............................................. 136
5.4.2 Hemicellulose removal and hemi-sugars recovery ..................................................... 139
xi
5.4.3 Cellulose removal and glucose recovery .................................................................... 140
5.4.4 Lignin removal and recovered soluble lignin yield .................................................... 142
5.4.5 2-D 1H -1C HSQC NMR of ball milled pine wood lignin and pretreatment obtained
RISL ..................................................................................................................................... 145
5.4.6 Pretreatment reactions kinetics at pH 6.4-12.0 ........................................................... 147
5.5 Conclusion ......................................................................................................................... 151
5.6 Acknowledgements ........................................................................................................... 151
5.7 Supplementary materials ................................................................................................... 152
CHAPTER SIX
CHARACTERIZATION OF LIGNIN DERIVED FROM WATER-ONLY AND DILUTE ACID
FLOWTHROUGH PRETREATMENT OF POPLAR WOOD AT ELEVATED
TEMPERATURES ...................................................................................................................... 155
6.1 Abstract .............................................................................................................................. 155
6.2 Introduction ....................................................................................................................... 156
6.3 Materials and methods ....................................................................................................... 159
6.3.1 Materials ..................................................................................................................... 159
6.3.2 Flowthrough pretreatment of poplar wood ................................................................. 159
6.3.3 Composition analysis of the solid residues and determination of lignin removal ...... 160
6.3.4 Ultraviolet visible spectroscopy (UV-Vis) quantification of RSL .............................. 161
6.3.5 Recovered insoluble lignin (RISL) isolation and purity measurement ....................... 161
6.3.6 Gel permeation chromatography analysis of molecular weight of flowthrough lignin
.............................................................................................................................................. 162
xii
6.3.7 Py-GC/MS analysis of ReL ........................................................................................ 162
6.3.8 Fourier transformed infrared (FTIR) spectroscopic analysis of ReL.......................... 163
6.3.9 Characterization of RSL after flowthrough pretreatment by GC/MS ......................... 163
6.3.10 RISL structural characterization by 2-D 1H -1C NMR HSQC NMR ........................ 163
6.4 Results and discussion ....................................................................................................... 164
6.4.1 Lignin removal and recovery in aqueous phase .......................................................... 165
6.4.2 Lignin purity determination of the isolated RISL ....................................................... 168
6.4.3 Py-GC/MS spectroscopic analysis of untreated poplar wood and pretreated solid
residues ................................................................................................................................ 170
6.4.4 FTIR spectroscopic analysis of ReL ........................................................................... 172
6.4.5 Characterization of the RSL by GC/MS ..................................................................... 174
6.4.6 2-D NMR of Ball-milled poplar lignin and RISL ....................................................... 175
6.5 Conclusion ......................................................................................................................... 179
6.6 Acknowledgements ........................................................................................................... 180
6.7 Supplementary materials ................................................................................................... 182
CHAPTER SEVEN
REVEALING THE MOLECULAR STRUCTURAL TRANSFORMATION OF HARDWOOD
AND SOFTWOOD IN DILUTE ACID FLOWTHROUGH PRETREATMENT ..................... 185
7.1 Abstract .............................................................................................................................. 185
7.2 Introduction ....................................................................................................................... 185
7.3 Materials and methods ....................................................................................................... 189
7.3.1 Materials ..................................................................................................................... 189
xiii
7.3.2 Dilute acid flowthrough pretreatment of pine wood ................................................... 190
7.3.3 Structural characterization of pretreatment recovered insoluble lignin by 2-D 1H-13C
HSQC NMR ......................................................................................................................... 191
7.3.4 Solid state CP/MAS 13C NMR characterization of dilute acid flowthrough pretreated
poplar and pine lignin .......................................................................................................... 192
7.3.5 Gel permeation chromatography molecular weight analysis of the flowthrough derived
pine lignin ............................................................................................................................ 192
7.3.6 Enzymatic hydrolysis .................................................................................................. 192
7.4 Results and discussion ....................................................................................................... 193
7.4.1 Delignification and cellulose removal of poplar and pine wood with dilute acid
flowthrough pretreatment ..................................................................................................... 193
7.4.2 Sugar recovery in dilute acid pretreatment of pine wood ........................................... 195
7.4.3 Characterization of pretreatment recovered insoluble lignin by 2-D 1H-13C HSQC
NMR .................................................................................................................................... 197
7.4.4 Solid state CP/MAS 13C NMR characterization of residual pine lignin in pretreated
solid residues ........................................................................................................................ 199
7.4.5 GPC analysis of the flowthrough derived lignin......................................................... 204
7.5 Conclusion ......................................................................................................................... 205
7.6 Acknowledgements ........................................................................................................... 205
7.7 Supplementary materials ................................................................................................... 206
CHAPTER EIGHT
HOT WATER AND DILUTE ACID PRETREATMENT OF SUB-MILLIMETER BIOMASS
xiv
PARTICLES ................................................................................................................................ 212
8.1 Abstract .............................................................................................................................. 212
8.2 Introduction ....................................................................................................................... 213
8.3 Materials and methods ....................................................................................................... 216
8.3.1 Materials ..................................................................................................................... 216
8.3.2 Batch tubular pretreatment .......................................................................................... 218
8.3.3 Digestibility evaluation of pretreated whole slurries by enzymatic hydrolysis .......... 220
8.3.4 Kinetic modeling of biomass degradation .................................................................. 221
8.4 Results and discussion ....................................................................................................... 223
8.4.1 Mass transfer modeling of hot water and dilute acid pretreatment of corn stover and
poplar wood ......................................................................................................................... 223
8.4.2 Effects of particle sizes and cutting approaches on mass transfer coefficients........... 226
8.4.3 Experimental results of hot water and dilute acid pretreatment of corn stover and
poplar wood ......................................................................................................................... 230
8.4.4 Optimization of the Douglas fir hot water and dilute acid pretreatment conditions ... 233
8.4.5 Pretreatment results of Douglas fir with various particle sizes and cuttings approaches
.............................................................................................................................................. 239
8.4.6 Hot water and dilute acid pretreatment of poplar wood ............................................. 244
8.4.7 Sugar yields from hot water and dilute acid pretreatment of poplar and corn stover and
sequential enzymatic hydrolysis .......................................................................................... 247
8.5 Conclusion ......................................................................................................................... 250
8.6 Acknowledgements ........................................................................................................... 251
xv
8.7 Supplementary materials ................................................................................................... 251
CHAPTER NINE
CONCLUSIONS AND FUTURE PERSPECTIVES ................................................................. 256
Future work ............................................................................................................................. 261
Achievements and perspectives ............................................................................................... 263
APPENDIX ................................................................................................................................. 265
REFERENCES ........................................................................................................................... 288
xvi
LIST OF TABLES
1. Table 2.1 Literature reviews of hemi-sugars recovery and total glucose yield operated under
flowthrough system; mono refers to monomeric sugars while oligo represents oligomeric
forms of sugars…………………………………………...…………….…………………….21
2. Table 2.2 Lignin characterization methods and targeted properties………………………....43
3. Table 3.1 Peak position and relative peak intensity parameters from curve fitting
using Lorentzian lineshape profiles (as in Equation 3.3 in the main text) of (a) Avicel, (b)
cellulose Iα from alga Valonia ventricosa (Glaucocystis (nostochinearum)), and (c) cellulose
Iβs from red reef tunicate and (d) Halocynthiaroretzi tunicate within wavelength of 2700 to
3050 cm-1 and 3200 cm-1 to 3450 cm-1………………………………………………………68
4. Table 3.2 Comparison of C-H, O-H peaks areas and their ratios of Avicel, cellulose Iβ
(different sources) and cellulose Iα .........................................................................................79
5. Table S4.1 Peak positions, amplitudes, and widths after curve fittings of SFG-VS spectra via
Lorentz profile convoluted with a Gaussian intensity distribution method in (a) TIR-SFG-VS
spectra of Avicel surface layers, (b) SFG-VS sepctra of Avicel bulks, (c) TIR-SFG-VS
spectra of cellulose Iβ surface layers, and (d) SFG-VS spectra of cellulose Iβ bulks within the
wavelength of 2800 to 3750 cm-1...........................................................................................122
6. Table 5.1 Distribution of the detected linkages of ball milled pine lignin and RISLs. BML:
ball milled wood lignin; A: RISL collected at pH of 11.0 and pretreatment severity of 6.1; B:
RISL collected at pH of 12 and pretreatment severity of 5.8. The percentages of inter
linkages of BML were based on all detected peaks………………………………………...146
7. Table S5.1 Main lignin 2-D 1H-13C Cross-peaks assignments in the HSQC
xvii
Spectra………………………………………………………………………..……….…….153
8. Table 6.1 FTIR spectra band assignments………………………..…………….…………..174
9. Table 6.2 Relative proportions of major structural linkages and S, G, H percentages in ball
milled lignin and flowthrough derived RISL; A: RISL collected at 240 °C, 25 ml/min, 10
min, with 0.05% (w/w) H2SO4; B: 270 °C, 25 ml/min, 10 min, with water-
only………………………………………………………..……………………...………...177
10. Table S6.1 Major GC/MS detected aromatic compounds in hydrolysates……..………….182
11. Table S6.2 Assignments of main lignin 1H- 13C cross-peaks in the HSQC Spectra of the
RISLs……………………………………………………………………….……...……….184
12. Table 7.1 General lignin difference in hardwood and softwood………………….…...…..186
13. Table 7.2 The linkage distribution of ball-milled and flowthrough-derived poplar wood and
pine wood lignin (flowthrough pretreatment conditions: 240 °C, 0.05% (w/w) H2SO4,
10min)………………………………………………………………………………...…….199
14. Table 7.3 Selected chemical shifts and signal assignments in a 13C NMR
spectrum………………………………………………………………………….…………202
15. Table 7.4 The molecular weight of pine lignin (flowthrough pretreatment at 240 °C, 0.05%
(w/w) sulfuric acid, and 10 min)………………….………………………………………..204
16. Table S7.1 Enzymatic hydrolysis of flowthrough pretreated pine wood whole
slurries………………………………….……………………………………...………… 206
17. Table S7.2 Assignments of main lignin 1H- 13C cross-peaks in the HSQC Spectra of the
pretreatment recovered insoluble lignin………………………………………..………….209
18. Table 8.1 Characteristics of feedstock used in the study………………………….……….217
xviii
19. Table 8.2 Selected pretreatment conditions performed in the study……………….…...…219
20. Table 8.3 Some unit volume of biomass solids (biomass density)……….…………...….. 222
21. Table 8.4 Kinetic modeling parameters (physical mass transfer coefficients) of xylan, glucan,
and lignin removal with consideration of mass transfer (a) 1% sulfuric acid pretreatment with
residence time of 0 to50 min at 140ºC and (b) hot water pretreatment with residence time of 0
to 30 min at 190ºC………………………………………………………………….……....227
22. Table 8.5 Sugar yields from pretreatment and enzymatic hydrolysis…………..……….…247
23. Table A1 Pyrolysis products of residual lignin derived from aqueous flowthrough
pretreatment by Py-GC/MS ……………..……………………………………...…………267
24. Table A2 The distribution of hydrodeoxygenation (HDO) products of flowthrough derived
lignin and the selectivity of HDO conversion with noble metal catalyst
matrix…………………………………………………….……………………...………….271
25. Table A3 Some acid diffusivity of various sized and cut biomass feedstock…………….286
xix
LIST OF FIGURES
1. Figure 1.1 Simplified process flowchart of biochemical conversion process for production of
ethanol from cellulosic biomass and cost distribution……………………………...…………4
2. Figure 1.2 (a) SFG-VS system and (b) flowthrough reactor……………....……………….....7
3. Figure 2.1 Plant cell wall structure and major chemical components matrix…………….....13
4. Figure 2.2 The aqueous degradation of major biomass components under acidic and alkaline
conditions; (a) cellulose, (b) hemicellulose, (c) lignin………………………………………27
5. Figure 3.1 (a) Raman spectra of cellulose Iα from alga Valonia ventricosa (Glaucocystis
(nostochinearum)) and (b) cellulose Iβ from Halocynthiaroretzi tunicate in the frequency
regions of 300 to 1600 cm-1 and 2500 to 3700 cm-1………………………………...………62
6. Figure 3.2 PNA for different visible polarization angles of cellulose Iα from alga Valonia
ventricosa (Glaucocystis (nostochinearum)), cellulose Iβ from Halocynthiaroretzi tunicate,
and Avicel using scanning SFG system at 2956 cm-1………………………………….……63
7. Figure 3.3 (a) Experimental HR-BB-SFG-VS spectra of Avicel, cellulose Iβ from
Halocynthiaroretzi tunicate, cellulose Iβ from red tunicate, and cellulose Iα from alga
Valonia ventricosa (Glaucocystis (nostochinearum)) at wavelengths of 2700 cm-1 to 3450 cm-
1. All spectra intensities were normalized and presented on the same intensity scale. (b)
Spectra for two cellulose Iβs after peak fitting via Lorentz profile convoluted with a Gaussian
intensity distribution method; (c) Spectra for cellulose Iα (left axis) and Avicel (right
axis)……………………………………………………...………………………………..…66
xx
8. Figure 3.4 Comparison of AFM images of (a) Avicel, (b) cellulose Iα from alga Valonia
ventricosa (Glaucocystis (nostochinearum), (c) cellulose Iβ from red tunicate and (d)
cellulose Iβ from Halocynthiaroretzi tunicate. The scale of all images is 2 x 2µm……...….70
9. Figure 3.5 HR-BB-SFG-VS spectra of cellulose Iβs from Halocynthiaroretzi tunicate and red
reef tunicate between wavenumber of 3100 cm-1 to 3500 cm-1………………………..…….72
10. Figure 3.6 HR-BB-SFG-VS spectra of cellulose Iβ from Halocynthiaroretzi tunicate and
cellulose Iα from alga Valonia ventricosa (Glaucocystis (nostochinearum)) between
wavenumber of 3200 cm-1 to 3450 cm-1……………………………………………………..73
11. Figure 3.7 Difference of cellulose Iβs from red reef tunicate and Halocynthiaroretzi tunicate
by HR-BB-SFG-VS…………………………………………………………………….……74
12. Figure 3.8 HR-BB-SFG-VS spectra of Avicel and cellulose Iα from alga Valonia ventricosa
(Glaucocystis (nostochinearum)) (a) and cellulose Iβ from Halocynthiaroretzi tunicate (b),
respectively, between wavelengths of 3200 cm-1 to 3450 cm-1…………………………...…77
13. Figure 4.1 a) TIR-SFG-VS design on cellulose surface observation and b) non TIR-SFG-VS
on cellulose bulk characterization. IR and VIS refers to infrared and visible 532nm beams,
respectively. c) total internal reflectance concept; d) simulation of SFG intensity with the
change of TIR and non TIR SFG-VS geometry, in which b) refers to non TIR SFG-VS while
TIR represents TIR-SFG-VS……………………………………………………………..….92
14. Figure 4.2 Cellulose heating high temperature fluid cell systems……………..……………94
15. Figure 4.3 SFG-VS spectra and peak fittings via Lorentz profile convoluted with a Gaussian
intensity distribution method of Avicel surface layers (red color) and bulk crystallines (blue
color) within wavelength of (a) 2800 to 3000 cm-1 and (b) 3000 to 3750 cm-1; Dots
xxi
represented experimental data while lines represented curve fittings. All intensities were
calibrated at the same voltage (1000 v) for the SFG-VS measurements…………………...97
16. Figure 4.4 SFG-VS spectra and peak fittings via Lorentz profile convoluted with a Gaussian
intensity distribution method of Iβ surface (red color) and bulk (blue color) within
wavelength of (a) 2800 to 3000 cm-1 and (b) 3000 to 3750 cm-1; Dots represented
experimental data while lines represented curve fittings. All intensities were calibrated at the
same voltage (700 v) for the SFG-VS measurements……………………..……………….103
17. Figure 4.5 Cellulose crystalline structural changes under different temperature during
heating process…………………………………………….……………………………….106
18. Figure 4.6 The recrystallization of dry cellulose polymer after cooling down from heating
process ………………………………………………….…………………………….…....108
19. Figure 4.7 Cellulose crystalline structural changes of wet Avicel heating
process...………………………………………………………………………………...…..111
20. Figure 4.8 Recrystallization process during cooling process of wet Avicel…………….....112
21. Figure 4.9 Impact of water on SFG-VS characterization of Avicel ...……………..…..…..113
22. Figure 4.10 Temperature dependence of Avicel surface layers characterized by SFG-VS (up:
structural changes of dry Avicel surface layers during heating process; down: heated dry
Avicel surface layers cooling down to room temperature)…………………………..……..114
23. Figure 4.11 SFG-VS spectra of the decay of enzymatic hydrolysis to study enzymatic
hydrolysis kinetics................................................................................................................117
24. Figure 4.12 SFG-VS characterization of Restart samples from interrupted enzymatic
hydrolysis ……………………………………………………………………………....….118
xxii
25. Figure S4.1 Whole spectra of Avicel and Iβ bulks and surface layers a) Avicel; b) Iβ (black:
blank control; blue: bulks; red: surface layers)…………………………………………..…124
26. Figure S4.2 Demonstrations of various molecular structures of Avicel surface layers and
relatively stable Iβ structure……………………………………………………..…….……125
27. Figure S4.3 TIR-SFG-VS spectra of (a) Avicel and (b) cellulose Iβ under two polarizations,
(red color) SSP and (Blue color) PPP; s-, s-, p- polarization corresponding to the polarization
of the optical fields of the SFG signal, visible and IR beams, respectively. p polarization is
defined when electric field vector parallel to the plane of incidence within the incident plane
formed by the incident beam direction and the surface normal, and s polarization is
perpendicular to the incident plane………………………………………………………....126
28. Figure S4.4 Cellulose structure (two glucose units) and different molecular vibrations......127
29. Figure S4.4 Heat disruption and recrystallization hypothesis……………………………...127
30. Figure 5.1 pH of the pretreatment hydrolysates at different severities after the hydrolysates
were cooled down to room temperature, pretreatment conditions: pH 6.4-12, 200-270°C, and
0-10 min……………………………………………….……………………………….…...137
31. Figure 5.2 Hemicellulose removal and hemi-sugars recovery in stage 1 (pretreatment only,
pH 6.4-12, 200-270°C, and 0-10 min)……………...………………………………………139
32. Figure 5.3 Cellulose removal and glucose recovery (Stage 1: pretreatment; Stage 2:
enzymatic hydrolysis), pretreatment conditions: pH 6.4-12, 200-270°C, and 0-10 min…...140
33. Figure 5.4 Lignin removal and recovered soluble lignin yield, pretreatment conditions: pH
6.4-12, of 200-270°C, and 0-10 min………………………………………………………..142
34. Figure 5.5 GC/MS results of the pretreatment hydrolysates collected at pH of 12 and
xxiii
pretreatment severity of 5.2………………………………………..……...………………..144
35. Figure 5.6 2-D 1H-13C HSQC NMR spectra of ball milled pine lignin and pretreatment
obtained RISLs at 11.0 and 12.0. (a,b) ball milled pine wood lignin, (c,d) RISL derived from
pH of 11.0 and pretreatment severity of 6.1, (e,f) RISL derived from the pH of 12.0 and
pretreatment severity of 5.8……………………………………………..……………….…145
36. Figure 5.7 Proposed biomass degradation mechanisms of softwood in aqueous flowthrough
pretreatment at initial pH 6.4-12, (a) hemicellulose; (b) cellulose; (c) lignin……...………149
37. Figure S5.1 The HPLC analysis of degradation products in pretreatment hydrolysates at pH
of 12.0 and pretreatment severity of 6.1…………………………….………….…………..152
38. Figure S5.2 Main detected lignin linkages…………………………..……………………..154
39. Figure 6.1 Scheme for lignin recovery and analytical analysis in this study………..…….165
40. Figure 6.2 Lignin removal and recovery as RSL and RISL. (a) water-only and (b) 0.05%
(w/w) H2SO4 flowthrough pretreatment (temperature: 160-270 °C, time: 0-12min, flow rate:
25mL/min)……………………………………………………………………………….…166
41. Figure 6.3 The RISL purity determined by AcBr method. (a) water-only; (b) 0.05% (w/w)
H2SO4 at the flow rate of 25 ml/min………………………………………….....………….168
42. Figure 6.4 Py-GC/MS results of untreated poplar wood (control) and solid residues after
flowthrough pretreatments. (a) Untreated poplar wood and solid residues pretreated at
240 °C, 0.05% (w/w) H2SO4 for 10 min (pretreatment severity ~5.0); (b) Untreated poplar
wood and solid residues pretreated at 270 °C, water-only for 10 min (pretreatment severity
~6.0)......................................................................................................................................170
43. Figure 6.5 Effects of pretreatment severity on S, G and H unit removal by water-only and
xxiv
0.05% (w/w) dilute acid pretreatment (poplar wood control sample: G:S:H= 7:11:2)……171
44. Figure 6.6 Characterization of ReL in pretreated solid residues by FTIR (800-2000 cm-1).
Control: Untreated Poplar wood; a) Solid residues from 0.05% (w/w) sulfuric acid
pretreatment at 240 °C for 2.6 min; b) Solid residues from water-only pretreatment at 230℃
for 3.8 min; c) Solid residues pretreated with 0.05% sulfuric acid at 240℃ for 5 min……173
45. Figure 6.7 2-D 13C - 1H HSQC correlation NMR spectra of aliphatic regions (left column)
and aromatic regions (right column), (a, b) Ball milled poplar wood lignin; (c, d) 240 °C,
0.05% (w/w) H2SO4 for 10 min (pretreatment severity ~5.0); (e, f) 270 °C, water-only for 10
min (pretreatment severity ~6.0)…………………………………………..……………….176
46. Figure 6.8 Plausible lignin β-O-4 cleavage and condensation reaction pathways in
hydrolysates…………………………………………………………………………...……179
47. Figure S6.1 Assigned interunit linkages of lignin, including different side-chain linkages,
and aromatic units: (A) β-O-4 aryl ether linkages; (B) resinol substructures (β-β′, α-O-γ′, and
γ-O-α′ linkages); (C) phenylcoumarane substructures (β-5′and α-O-4′ linkages); (D)
spirodienone substructures (β-1′ and α-O-α′ linkages); (G) guaiacyl units; (G′) oxidized
guaiacyl units with an Cα-ketone; (S) syringyl units; (S′) oxidized syringyl units with a Cα
ketone; (E) p-hydroxybenzoate substructures; (F) alcohol end groups; (K) cinnamaldehyde
end groups……………………………………………………………………………...…..183
48. Figure 7.1 Lignin and cellulose contents in pretreated solid residues of poplar and pine wood
after the flowthrough pretreatment (200-270°C, reaction time of 2-10min, and 0.05% (w/w)
H2SO4)………………………………………….…………………..………………………193
49. Figure 7.2 Sugar yields of pine wood at 0.05% (w/w) sulfuric acid flowthrough pretreatment
xxv
under different severity log R0 (a) Cellulose recovery; (b) Hemi-sugars recovery………...195
50. Figure 7.3 Characterization of (a,b) ball milled pine lignin; (c,d) flowthrough pretreatment
recovered insoluble pine lignin from liquid phase by 2-D 1H-13C HSQC NMR. (pretreatment
conditions: 240 °C, 0.05% (w/w) sulfuric acid, and 10min)……………………………….197
51. Figure 7.4 Solid state CP/MAS 13C NMR spectra of a) pine residual lignin; b) pine
pretreatment recovered insoluble lignin; c) poplar pretreatment recovered insoluble lignin
under 240°C with 0.05% (w/w) H2SO4 for 10min………………………………..……..….200
52. Figure 7.5 Possible recondensation mechanisms of pine lignin during dilute acid
pretreatment………………………………………………………………………...………203
53. Figure S7.1 Assigned interunit linkages of lignin, including different side-chain linkages,
and aromatic units: (A) β-O-4 aryl ether linkages; (B) resinol substructures (β-β′, α-O-γ′, and
γ-O-α′ linkages); (C) phenylcoumarane substructures (β-5′and α-O-4′ linkages); (D)
spirodienone substructures (β-1′ and α-O-α′ linkages); (G) guaiacyl units; (G′) oxidized
guaiacyl units with an Cα-ketone; (S) syringyl units; (S′) oxidized syringyl units with a Cα
ketone; (H) hydroxyphenyl groups; (E) p-hydroxybenzoate substructures; (K)
cinnamaldehyde end groups……………………………………………………...…………210
54. Figure S7.2 GPC analysis of lignin molecular weight of a) poplar flowthrough pretreatment
recovered insoluble lignin, b) pine wood flowthrough pretreatment recovered insoluble
lignin, c) pine wood flowthrough derived residual lignin under 240°C with 0.05% (w/w)
H2SO4 for 10min……………………………………………………………………...….....211
55. Figure 8.1 Kinetic diffusion modeling (equation 8.6, 7, 8) of pretreatment behaviors of major
components in six feedstock 1% (w/w) H2SO4 pretreatment at 140ºC with residence time of
xxvi
0-50 min including (a) xylan dissolution, (b) glucan dissolution, and (c) lignin dissolution,
and hot water pretreatment at temperature of 190ºC with residence time of 0-30 min
including (d) xylan dissolution, (e) glucan dissolution, and (f) lignin
dissolution…………………………………………………………………………….....….224
56. Figure 8.2 Biomass removal trend of Douglas fir in hot water and dilute acid pretreatment
(a) Biomass removal; (b) lignin removal; (c) cellulose removal, pretreatment severities 2.1-
4.5………………………………………………………………………………..…………235
57. Figure 8.3 Total sugar yields from Douglas fir (1% H2SO4), left axis is Absolute (hemi-
sugars+glucan)/hemi-sugars/glucan recovery (from 100g dry biomass) while right axis is
Relative (hemi-sugars +glucan) /hemi-sugars/glucan recovery (% of maximum), Stage 1:
sugar release during pretreatment (measured after posthydrolysis); Stage 2: sugar release
during enzymatic hydrolysis; Stage 1+2: sugar release during pretreatment and enzymatic
hydrolysis, 100 mg protein Ctec2 (93 FPU) with 20 mg Htec2/g (glucan +
xylan)…………………….…………………………………………………………..…….237
58. Figure 8.4 Biomass and major components removal of Douglas Fir (140°C, 1% (w/w)
sulfuric acid, 50min), Absolute biomass removal (from 100g dry biomass), crumbler (CB)
and hammer cut (HM)……………………………………………………………………..239
59. Figure 8.5 Recovery of Douglas fir sugars through dilute acid pretreatment (1% (w/w)
sulfuric acid), Stage 1: sugar release during pretreatment (measured after posthydrolysis);
Stage 2: sugar release during enzymatic hydrolysis; Stage 1+2: sugar release during
pretreatment and enzymatic hydrolysis, 20 and 200 mg protein Ctec2 with 4 and 40 mg
Htec2/g (glucan + xylan)………………………………....……………………...…………241
xxvii
60. Figure 8.6 Recovery of Douglas fir sugars through hot water and enzymatic hydrolysis (a)
biomass and lignin removal; (b) sugar recovery; Stage 1: sugar release during pretreatment
(measured after posthydrolysis); Stage 2: sugar release during enzymatic hydrolysis; Stage
1+2: sugar release during pretreatment and enzymatic hydrolysis, 20 and 200 mg protein
Ctec2 with 4 and 40 mg Htec2/g (glucan + xylan)...……………………………….………243
61. Figure 8.7 (a) Biomass removal, (b) Cellulose removal, (c) Lignin removal, (d) Manan and
xylan removal, and (e) sugar yields of poplar in dilute acid pretreatment, 140°C, 1% (w/w)
sulfuric acid, 40min, and enzymatic hydrolysis (72 hours), 100 mg protein Ctec2 with 20 mg
Htec2/g (glucan + xylan)…………………………………………………………………...245
62. Figure 8.8 Biomass removal, sugar yields of green poplar in hot water pretreatment, 190°C,
hot water, 15min (a) biomass/major components removal, (b) combined sugar yields after
pretreatment and enzymatic hydrolysis, 100 mg protein Ctec2 with 20 mg Htec2/g (glucan +
xylan)………………………………………………………………………………………246
63. Figure S8.1 Left) Crumbler M24 research prototype machine used with 4.8 and 1.6 mm
cutters, Right) Research scale Crumbler machine with 0.8 mm cutters (by ForestConcept,
LCC)…………………………………………………...…………..………...……………..251
64. Figure S8.2 Total sugar yields from Douglas fir of (a) 0.05% wt sulfuric acid (b) Water only,
left axis is Absolute (hemi-sugars+glucan)/hemi-sugars/glucan recovery (from 100g dry
biomass) while right axis is Relative (hemi-sugars +glucan) /hemi-sugars/glucan recovery (%
of maximum), Stage 1: sugar release during pretreatment (measured after posthydrolysis);
Stage 2: sugar release during enzymatic hydrolysis; Stage 1+2: sugar release during
pretreatment and enzymatic hydrolysis…………………..…………………….…………..254
xxviii
65. Figure S8.3 Hot water pretreatment of poplar wood chips for 5,10,15,25, and 30 min (left to
right), particle size of 1/8′′-16 mesh and cut by Crumbler…………………………………255
66. Figure A1 SFG-VS characterization of (a) Avicel 2600-3600cm-1; (b) Avicel 1000-1600cm-1;
(c) Cellulase and BSA 2600-3800 cm-1; (d) Cellulase and BSA 1500-1800 cm-1, Note: (A)-
Avicel without a flat CaF2 window on top of the sample; (B)-Avicel with a flat CaF2 window
on top of the sample………………………..............................................................………265
67. Figure A2 Isolation processes of the native and pretreatment derived
lignin ……………………………………………………………………………………….270
68. Figure A3 GC/MS determination of water/dilute acid soluble lignin fragments from 240°C
with 25 ml/min flow of 0.05% sulfuric acid………………………………………………..272
69. Figure A4 Determination of cellulose crystallinity before and after flowthrough pretreatment
by XRD………………………………………………………………………...…………...273
70. Figure A5 Our flowthrough system……………………………………….……………….274
71. Figure A6 Application of SFG-VS in observing pretreated biomass substrate and in
comparison with XRD results………………………………………………………..……..275
72. Figure A7 Designed measuring cell for mass transfer coefficients, (a) reported device (b)
designed device……………………………………………………………………….……279
73. Figure A8 Calibration of conductivity (a) Sulfuric acid (b) NaOH………………………281
74. Figure A9 Fractional uptake of NaOH, Douglas fir 1/8’’-16mesh, CB substrates………..285
xxix
DEDICATION
This dissertation is dedicated to my husband, parents and other
family members who provide love and support.
1
CHAPTER ONE
INTRODUCTION
The limitation of supply of fossil fuels can cause an energy crisis, which is why so much
research is on attractive and renewable energy resources. Along with hydro, geothermal, wind,
and solar energy, cellulosic biomass based biofuels are among the most promising and emerging
renewable energy sources that can have a pervading impact on energy use structures, especially
in transportation fuels. The biofuels produced from cellulosic biomass can enhance energy
security, reduce greenhouse gas emissions, improve global competitiveness, and create job
opportunities 1. The U.S. Department of Agriculture (USDA) and Department of Energy (DOE)
estimated that about 1.3 billion dry tons of cellulosic biomass could be available annually, which
is equivalent to 1.5 billion barrels of traditional petroleum 2. This amount of bio-oil can have
major effects on energy use structures in the U.S. without causing food shortage 3, 4. Currently,
E10, E85, and E20–E40 represent 10%, 85%, and 20-40% bioethanol blending with conventional
gasoline, respectively, while B20 indicates 20% biodiesel blending with petroleum refined diesel.
The research and commercialization of biofuels receive policy support from the U.S.
government. For example, the DOE’s Bioenergy Technologies Office has supported biorefinery
development such as INEOS Bio’s Indian River BioEnergy Center, POET-DSM’s Project
LIBERTY, and the Abengoa’s Bioenergy Biomass of Kansas. DuPont has implemented a
cellulosic ethanol biorefinery in Nevada, Iowa supported by government funding and
collaboration with National Renewable Energy Laboratory (DOE; Growing America’s Energy
Future: Bioenergy Technologies Office Successes of 2014). Biofuels have received considerate
attention from the public.
2
Among different types of biofuels, bioethanol production has been popular since the 1990s,
but it was only produced from corn grains back then. Nowadays, ethanol or butanol, which can
be produced from the fermentation of starches and carbohydrates, is considered to be the most
promising forms of biofuels to replace petroleum gasoline. The U.S. Secretary of Energy
proposed that 30% (~60 billion gallons) of transportation fuel should be ethanol, indicating the
need for about 60 billion gallons per year to meet the annual demand of 140 billion gallons of
gasoline. However, even though ethanol production from corn and grains has doubled from 4.5
to 7.5 billion gallons, the ethanol production is not enough to satisfy the growing demand for
transportation fuels 5. Therefore, cellulosic bioethanol can have a significant impact in helping to
meet the demand of the transportation fuel market. However, establishing full commercialization
of bioethanol is still a work in progress. The main challenge is the relatively high operation cost
of bioethanol production.
1.1 Biochemical conversion of biomass to bioethanol
After the energy crisis of the 1970s, the DOE initiated the Office of Alcohol Fuels to support
the increase of ethanol production. In the past few decades, the cost of cellulosic ethanol was
reduced from $4.00 to 1.2-1.5$/gallon 6. However, this price is still not low enough to compete
with the petroleum market. The main technique necessary to trigger this improvement is the
biochemical conversion process of biomass by biotechnology (Trichoderma reesei discovered
during World War II) 7. Biochemical conversion is one of the two main conversion R&D
processes (biochemical and thermochemical) of the DOE and Office of Energy Efficiency and
Renewable Energy (EERE) biomass program, which benefits greatly from advancements in
biotechnology 8. The biological conversion of cellulosic biomass through enzymatic hydrolysis
3
of cellulose and fermentation of sugars offers the potential for higher yields, higher selectivity,
lower energy costs, and milder operating conditions than thermochemical processes 9. The goal
of biochemical conversion R&D is to reduce the cost of biomass to fermentable sugars and other
usable intermediates, and further conversion of these intermediates to liquid transportation fuels
or other bioproducts 8. Thus, the main study focus of biochemical conversion is reducing the
cost.
A techno-economic analysis, particularly at large scale production to reduce the investment
risks, is required for commercialization of bioethanol production. Several techno-economic
analyses were carried out by the National Renewable Energy Laboratory 9-13, which gained
strong experience that can lower the risks to biofuel commercialization. In these analyses,
biochemical conversion to bioethanol is considered as one of the closest biofuel techniques to
commercialization. Also, the biochemical conversion of biomass includes mainly feedstock
preparation, pretreatment, enzymatic hydrolysis, and fermentation, presenting a cost distribution
of ~30%, ~20%, 12%, and ~10%, respectively (Figure 1.1). Therefore, pretreatment is the most
expensive operation process, and it impacts the cost of both upstream and downstream processes
14. As mentioned, the high cost of biochemical production of cellulosic ethanol is the major
challenge, thus, the reduction of pretreatment cost is significant to make cellulosic ethanol
competitive with starch ethanol and gasolines.
4
Figure 1.1 Simplified process flowchart of biochemical conversion process for production of
ethanol from cellulosic biomass and cost distribution
1.2 The essential role of aqueous biomass pretreatment
The direct biological treatments of cellulosic biomass to bioethanol produce a low yield of
bioethanol because lignocellulosic biomass is recalcitrant towards chemicals and microbiological
treatments due to its naturally complex structure and chemical composition. The sugar yield of
biomass of enzymatic hydrolysis is very low without pretreatment, which is not economically
feasible 15. Pretreatment can improve bioethanol yield, which refers to the process of opening
biomass structures to increase accessibility, reduce biomass recalcitrance, and produce more
fermentable sugars. The biomass pretreatment not only produces sugars for biochemical
conversion, but it also produces sugars, lignin or other chemical intermediates for
thermochemical, biological and catalytic upgrading to valuable materials, chemicals and fuels,
which improves biorefinery economics.
5
Biomass pretreatment methods have been categorized into chemical, physical, biological,
and thermal-chemical methods 14. In order to compete with the conventional fossil fuel market,
the sugar price should be, at most, 5 cents/lb to make bioethanol cost-effective. However, the
latest sugar price was projected to be 6.4 cents /lb in 2002 by National Renewable Energy
Laboratory (NREL) in the process design and economics study utilizing co-current dilute acid
pre-hydrolysis and enzymatic hydrolysis for corn Stover 11. To achieve such a low sugar cost, the
pretreatment technique should be oriented to low cost, which also affects the cost of feedstock
preprocessing and other downstream processes. The aqueous pretreatment based on water only or
dilute acid/alkali is a promising option to achieve low-cost-pretreatment due to the cheap and
recyclable steam, substantial industrial experience on steam treatments, and long history of
development. Many pretreatment reviews have been published in the past ten years 1, 5, 16-27, and
they have summarized the most significant achievements in biomass aqueous pretreatment
including pretreatment reactors, kinetics, process overview, economics, and biomass chemistry.
Thus, many efforts have focused on aqueous pretreatment in the past decades 17, however, more
efforts are still needed to advance pretreatment and reduce the cost of pretreatment to accelerate
commercialization .
1.3 The key problems in aqueous pretreatment
The insufficient studies of key fundamental issues of aqueous pretreatment result in a high
cost of aqueous pretreatment. Despite the importance of aqueous pretreatment technique, the
process characterization of aqueous pretreatment is poorly understood. Although both the
recovery of sugars and the removal of lignin are frequently investigated as pretreatment targets,
the mechanisms are not well understood. The effects of temperature, pH, and time on the
6
aqueous pretreatment performance are not well defined which might add extra cost on the
reactor, chemical and steam. Due to the limitation of cellulose analytical methods, cellulose
crystalline structure is rarely characterized in aqueous pretreatment. Without a clear
understanding of cellulose crystalline structural change during aqueous pretreatment, it is
difficult to develop strategies to improve sugar yields and reduce enzyme loading. This impedes
developing low-cost sugar production. The conversion of lignin to valuable materials, fuels, and
chemicals has received considerate attention in recent years. However, little effort has been made
to characterize lignin derived from aqueous pretreatment. This poses challenges on lignin
utilization and valorization which couldn’t provide extra credits and reduce the cost. Also, the
insufficient understanding of lignin chemistry in aqueous pretreatment impedes achieving the
high yield of sugars which increased sugar cost.
1.4 Thesis objectives and hypothesis
Against the above background, the goal of this thesis is to develop greater insight into the
key fundamental knowledge of aqueous pretreatment. The tools developed to pursue this
knowledge are SFG-VS and a flowthrough reactor (Figure 1.2). The specific research objectives
of this thesis are mainly as follows:
1. Study the key fundamental issues in aqueous pretreatment by process characterization of
the developed aqueous flowthrough pretreatment technique.
2. Study effects of pH, temperature, and reaction time on flowthrough aqueous
pretreatment.
3. Develop cellulose characterization tools and characterize original cellulose and structural
transformation in aqueous pretreatment.
7
4. Produce and characterize lignin from the aqueous pretreatment of poplar and pine wood
to study lignin characteristics and solubilization mechanisms for potential utilization. The
aqueous pretreatment derived lignins have been provided to upgrade to jet fuel, for
carbonization to carbon based materials to produce electrodes, and for biological
fermentation to produce bio-lipids but these are not the focus of this thesis.
5. Identify the most promising substrate particle sizes and cutting approaches (e.g. Shards,
Cubic prismatic) for aqueous pretreatment. Develop applicable mass transfer modeling to
describe solvent diffusion into biomass and better aqueous pretreatment kinetics.
Figure 1.2 (a) SFG-VS system and (b) flowthrough reactor
The main hypothesis corresponding includes:
1. The high yield of sugars and usable lignin (e.g. high purity, low molecular weight, and
mild structural modification) are produced by the aqueous flowthrough pretreatment
when elevated temperatures are applied at whole pH range because flowthrough is
fractionating both softwood and hardwood and moving products in a short time to avoid
further reactions.
8
2. The basic cellulose structure and structural transformation by the effects of aqueous
pretreatment heating temperature and media at the molecular level can be studied by
SFG-VS system with/without a cell to mimic aqueous pretreatment because SFG-VS is
in-situ and selectively characterizing crystalline structure of cellulose.
3. The fundamental issues of aqueous pretreatment in lignin properties, behaviors, and
chemistry can be understood by the flowthrough pretreatment system because the
flowthrough system can separate lignin (e.g. residual, soluble and insoluble lignins) from
carbohydrates for characterizing.
4. The energy-saving biomass particle sizes and cutting methods for aqueous pretreatment
can be studied by developing a kinetic modeling with mass transfer effects on hot water
and dilute acid pretreatment of various particle sizes and cutting methods because the
particle sizes and sharps of biomass particles affect the diffusion of pretreatment solvent
into biomass.
1.5 Thesis organization
The background and motivations of this thesis have been introduced in chapter 1. The
purpose of the rest of this chapter is to establish the organization of the thesis’ content. Chapter 2
reviews the fundamental knowledge and recent research progress of aqueous pretreatment. To do
a great job, one must first sharpen one’s tools. Due to the limitations demonstrated by traditional
methods of studying cellulose crystalline structure, chapter 3 describes the SFG-VS technique in
characterization cellulose crystalline structure to assist in cellulose characterization in aqueous
pretreatment. To investigate more detailed cellulose crystalline structure, High Resolution
Broadband SFG-VS (HR-BB-SFG-VS) is developed to characterize four cellulose crystals from
9
well-known sources. This work has been published in the journal Cellulose 28. This work lays the
foundation of the application of SFG-VS in studying cellulose structural characterization in
aqueous pretreatment.
In chapter 4, a high temperature liquid flowthrough cell with fluid pathways, heating system,
and temperature controller are used under the SFG-VS system to dynamically monitor cellulose
crystalline structure change with effects of high temperatures and chemicals, which simulates
practical aqueous pretreatment process. To study the cellulose crystalline surface layers, the laser
input angles and geometry (prism) are developed for the first time to observe the molecular
structure of cellulose crystalline surface layers. Also, the enzymatic hydrolysis process is
characterized by SFG-VS to demonstrate the capacity of SFG-VS in studying enzymatic
hydrolysis process.
Chapter 5 summarizes the effects of neutral and alkaline pH, and pretreatment severities on
pretreatment sugar yield and lignin recovery of softwood as well as the change of pH in aqueous
pretreatment. The softwood pretreatment kinetics and product distribution under neutral and
alkaline pH are investigated. Chapter 6 will take a closer look at hardwood lignin removal
mechanisms to reveal the depolymerization and repolymerization mechanisms of lignin in
aqueous pretreatment. The enhancement of sugar yield in flowthrough aqueous pretreatment as
well as lignin removal and properties (e.g. yield, purity, and molecular weight) are also well
documented in this chapter. Chapter 7 compares softwood and hardwood lignin removal
mechanisms and reveals their different recalcitrance in aqueous pretreatment. The different
hardwood and softwood structural transformations were revealed. This work is significant in
guiding pretreatment strategies on various feedstock, the utilization of lignins from various
10
sources and mixed feedstock. Feedstock preprocessing can have a significant impact on
pretreatment performance and biorefinery costs. In chapter 8, the pretreatment and enzymatic
hydrolysis behaviors of variously shaped and sized biomass are assessed to evaluate the
feasibility of energy saving shapes and sizes. Also, mass transfer modeling will be discussed to
determine the optimum particle sizes and cutting methods. At last, chapter 9 will cover the most
significant findings, conclusion, and perspectives of this thesis.
11
CHAPTER TWO
FUNDAMENTALS OF BIOMASS AQUEOUS PRETREATMENT
2.1 Abstract
Biomass pretreatment is required in the biochemical conversion of biomass to bioethanol.
The pretreatment process can assist enzymatic hydrolysis in producing high yield of stable sugar
intermediates for fermentation. Besides, new development of catalytic, biological, and
thermochemical valorization of lignin in the production of hydrocarbons has generated attention
to reactive lignin intermediates. It is expected to achieve a whole fractionation of biomass to
produce high yield of sugars and lignin and a clear fundamental understanding of cellulose and
lignin properties, behaviors, and chemistry in aqueous pretreatment. In this chapter, the research
progress of aqueous pretreatment is described. The recent research progress and characterization
of cellulose and lignin are summarized in aqueous pretreatment. At last, the preprocessing of
feedstock for aqueous pretreatment is reviewed.
2.2 Introduction
The aqueous pretreatment consists of a variety of pretreatment methods. In this thesis, it
refers to steam based pretreatment such as hot water, dilute acid, and alkaline pretreatment. The
advantages of aqueous pretreatment are the cheap and recyclable steam, the substantial industrial
experience using steam treatments, and the long historical development of aqueous pretreatment.
This chapter will first explore the research progress of aqueous pretreatment including hot water,
dilute acid, and alkaline pretreatment. The biomass degradation chemistry in aqueous
pretreatment will also be presented. Additionally, the application of flowthrough reactor system
in aqueous pretreatment will be discussed. The cellulose and lignin characterization and
12
chemistry in aqueous pretreatment will be reviewed as well. The aqueous pretreatment has
accumulated a great amount of publications. The chapter only covers key research references and
reviews for readers to pursue.
2.3 Plant cell wall macrostructure
Plant photosynthesis is the biological process for plants to convert solar energy into organic
biopolymers known as cell wall polysaccharides; primarily cellulose, hemicellulose, and lignin,
which all contain carbon and energy for plant survival. Only few microorganisms are capable of
decomposing these polymers, e.g. E. coli (a bacterium), Saccharomyces cerevisiae (yeast), and
Zvmomonas mobilis (bacterium). Lignin is believed to not only keep the mechanical strength of
the plant cell wall, but it is also thought to make carbohydrates resistant to microbial attack.
Besides, cellulosic biomass also comprises some amount of minerals, oils, free sugars, starches,
and other compounds. The chemical and physical characteristics of plant cell walls have a great
impact on their accessibility to enzyme and chemicals’ treatments, thus, the structural studies of
these organic biopolymers (e.g. cellulose and lignin) are crucial to resolving their interaction
with enzymes to improve hydrolysis efficiency. The plant cell wall microstructure is shown in
Figure 2.1, which shows the embedding of cellulose in hemicellulose and lignin to form a
complex matrix. Four main chemical bonds are observed in lignocellulosic biomass, namely,
ether, ester, C-C, and hydrogen bonds. The bonds in individual cellulose, lignin, and
hemicellulose are presented in Figure 2.1. In addition, these three biopolymers share chemical
bonds with each other that form a tight cell wall microstructure. The bonds for connecting
hemicellulose and lignin polymers are ether, ester, and hydrogen bonds, while cellulose and
lignin are connected by ether and hydrogen bonds. The hydrogen bonds with hemicellulose are
13
much weaker than intra and inter-chain hydrogen bonding network in cellulose since
hemicellulose lacks primary alcohol functional groups 16. In addition, cellulose and
hemicellulose are only connected by hydrogen bonds. Besides, lignin has covalent bonds with
polysaccharides, which causes three biopolymers bond tightly 16. The distribution of chemical
bonds in plant cell walls is important for pretreatment method design and degradation chemistry.
Also, the aqueous hydrolysis of these bonds is one of the mechanisms to break down the
biopolymers.
Figure 2.1 Plant cell wall structure and major chemical components matrix
2.4 Biomass aqueous pretreatment research progress
The successful pretreatment method design should have all the following key attributes 17, 21,
14
29-31:
1. Limited size reduction of biomass to decrease energy cost in biomass prehandling
2. Minimal and inexpensive chemical use
3. Low cost reactors (minimizing volume, pressure, temperature, and reaction time)
4. Compatible to downstream conversion
5. Reactive intermediates upgraded to value-added products to increase plant avenues
6. High solid loading operation to get high concentration of product stream
7. Easy products transportation
8. Fast heat transfer
9. Low capital and operational costs
10. High product yield and selectivity
To date, biological methods are hard to control, presenting poor products selectivity and long
residence time 17. Physical methods were not effective in achieving high sugar yields, and the
total pretreatment cost was high 32. Concentrated chemical pretreatment is deficient in high
chemical cost and tedious chemical recovery procedures. Aqueous pretreatment is the most
promising potential option that can achieve all the attributes described above. Aqueous
pretreatment heats biomass with steam, which is the simplest and cheapest pretreatment option
17. The aqueous based pretreatment methods are defined to be the process of heating cellulosic
biomass in water to convert structural biomass major components to reactive intermediates that
can be further biologically, catalytically, or thermochemically converted into fuels, chemical, and
biomaterials 17. Thus, the aqueous pretreatment occurs in pure water with or without the slight
addition of catalytic chemicals. The final products (e.g. sugars) from aqueous pretreatment of
15
biomass are not simply obtained for the purpose of biological conversion to ethanol. The
products can also be reactive intermediates for subsequent thermochemical and catalytic
upgrading, e.g. reactive lignin for hydrodeoxygenation upgrading to hydrocarbons, or converting
sugars to chemicals. Thus, the pretreatment methods can be diverse, depending on their various
purposes.
Many efforts have been devoted to biomass aqueous pretreatment in the past ten years,
which have been reviewed 14, 19, 20, 27, 33. Also, biomass aqueous pretreatment has been well
introduced and summarized in the book titled ′Aqueous Pretreatment of Plant Biomass for
Biological and Chemical Conversion to Fuels and Chemicals′ 17. Several leading aqueous
pretreatment methods, mainly acid and alkaline based aqueous pretreatment, have been
intensively investigated. The most extensively studied pretreatment was performed by the
Biomass Refining Consortium for Applied Fundamentals and Innovation (a national consortium
of universities and the DOE National Renewable Energy Laboratory) to use the same corn
stover, poplar wood, and switchgrass as feedstock and enzyme sources in ammonia, lime,
controlled pH, sulfur dioxide, and dilute sulfuric acid pretreatments 34-36. The advancement of
pretreatment techniques in recent years has achieved more than 95% of hemicellulose recovery,
which originally presents 20-30% in biomass (arabinose, galactose, glucose, mannose, and
xylose-hemi-sugars). However, the removal and recovery of cellulose and lignin are still limited,
which results in high enzyme dosage and poor utilization of carbohydrates and lignin. Secondly,
although the formation of acetic acid by autohydrolysis of biomass in hot water pretreatment can
act as a catalyst to biomass hydrolysis reaction, these sugar degradation products (e.g. furfural,
acetic acid, and HMF) can potentially inhibit a subsequent fermentation process. At last, the
16
degradation products from lignin inhibit downstream enzymatic hydrolysis and fermentation
efficiency.
In summary, it is important to find a pretreatment method that can fractionate biomass and
separate biomass degradation products into three streams (hemicellulose, lignin, and cellulose) or
two streams (sugar and lignin) with limited sugar degradation loss for high yield of ethanol
production and lignin utilization. The organosolv pretreatment was reported to achieve complete
separation of biomass components into solid lignin, an aqueous stream and pure cellulose 18.
However, this method suffers from the challenge of recovering pretreatment solvents. The
flowthrough reactor system can also achieve this goal by exiting sugars and solubilized lignin out
of the reactor after hydrolysis and depolymerization to avoid further sugar degradation loss.
2.4.1 Hot water and dilute acid pretreatment
In early 1898, sulfuric acid was used in the hydrolysis of cellulose and hemicellulose in
biomass to produce sugars; however, a high concentration of acid was used, which resulted in the
high operation cost and challenge in acid recovery. Currently, dilute sulfuric acid pretreatment is
greatly favored since it can recover more than 90% of the theoretical maximum sugars with more
monomeric sugars compared to less than about 65% in autohydrolysis, which occurs when
biomass is pretreated with water only. Several dilute acid pretreatment reviews were published to
summarize the pretreatment process and conditions, reaction kinetics, reactor systems, and
economics 26, 27, 33. The acidic reagents may be mineral or organic acids, such as hydrochloric,
sulfuric, phosphoric, peracetic, oxalic, and/or maleic acid. The liquid hot water is also a dilute
acidic method since the pH of water decreases with temperature. For example, under the same
pressure and ionic strength, pKw decreased with increasing temperatures until 250°C. Above
17
250°C, pKw increased along with increased temperature (Refers to Release on the Ionization
Constant of H2O". International Association for Properties of Water and Steam. August 2007.).
Thus, water acts as an acid in high temperature 37-39. At temperatures between boiling (100°C)
and critical temperature of water (374°C), water is less polar because temperature disrupts
hydrogen bonding of water, which provides a good solvent for pretreatment. The water/steam
treatment has been highly favored and gained a lot of experience in the pulp and papermaking
industry. The reactors used for this process in the industry include batch Masonite gun and the
continuous screw fed STAKE II reactor. Thus, dilute acid pretreatment technique gets support
from mature pulping industry. However, the challenge is to control the sugar degradation loss
and achieving high yield of sugars with low enzyme loading.
Hot water pretreatment is performed at temperatures of 200–230°C for up to 15 min. Hot
water pretreatment can achieve 40-60% of the total biomass solubilization with 4–22% of the
cellulose, 35–60% of the lignin, and all of the hemicellulose 16. A dilute acid pretreatment (0-2%
sulfuric acid) at temperatures of 140°C-280°C using different reactor systems was operated to
achieve more than 95% of hemi-sugars recovery and more than 80% of glucose yields followed
by enzymatic hydrolysis with over 90% of the hemicellulose as monomeric sugars. Steam
explosion pretreatment adds a sudden physical vapor pressure release of reactor to the hot
water/steam treatment of biomass, which increases enzymatic digestibility of pretreated
substrates. Controlled pH pretreatment was developed to reduce sugar degradation to obtain as
high as oligomeric hemi-sugar to reduce the effects of released organic acids in the sugar
degradation loss. However, the sugar degradation in hot water and dilute acid pretreatment is still
hard to control.
18
2.4.2 Alkaline pretreatment
Alkaline pretreatment is an important pretreatment method because of its unique
performance. In 1927, the Austrian chemist Skrabal published the importance of H+ and OH- in
biomass hydrolysis by indicating that the H+ and OH- concentration directly influenced
degradation of major biomass compounds. The study of the impact of OH- on biomass hydrolysis
occurred earlier than H+ because alkali had been used in the pulping industry for decades. The
Kraft process (NaOH 53% and sodium Na2S 21%; ~170°C; ~120min) was invented by Carl F.
Dahl in 1879 in Danzig, Germany. Alkalis can efficiently remove lignin to produce cellulosic
pulps. Nowadays, high concentration of alkali is used in pretreating biomass at low temperature
and pressures at ambient conditions for hours or days - to increase enzymatic hydrolysis sugar
yield. Several factors in alkaline pretreatment can have a great impact on pretreatment efficiency.
For example, in ammonia-based pretreatment methods, the ammonia loading, water loading,
temperature, time, blow down pressure, and number of treatments all have great impacts on
pretreatment performance 40. The aqueous ammonia swells the cellulose and causes a possible
transformation of the crystal structure at very mild conditions compared to other pretreatment
methods. Additionally, the common features of alkaline pretreatments can improve enzymatic
digestibility of biomass with the production of controllable sugar degradation loss (e.g. HMF and
furfural) 16. Meanwhile, the removal of the hemicellulose acetyl group is significant because it
reduces the inhibition of acetyl groups on fermentation, which can be achieved by alkaline
pretreatment 41. Ammonia Fiber Expansion (AFEX) was also performed to enhance enzymatic
hydrolysis with its benefits of fiber expansion. Ammonia Recycled Percolation (ARP) passes an
aqueous ammonia solution (usually 5-15 (w/w) %) through a column reactor packed with
19
biomass at temperatures between 80-180°C. The alkalis (e.g. ammonia, lime, and NaOH) can
depolymerize lignin, which reduces non-effective binding of lignin to enzymes (e.g. cellulase)
thereby improving the sugar yield of enzymatic hydrolysis. The alkaline pretreatment has
reached its bottleneck in achieving its low cost of commercialization.
The alkaline pretreatment increases biomass internal surface area, which exposes cellulose
for enzymes during the enzymatic hydrolysis process. Moreover, although the alkali is
ineffective at the removal of hemicellulose, the alkalis can solubilize lignin and break the bonds
between lignin-carbohydrates complex (LCC). Significantly, it can swell the cellulose crystalline
structure as mentioned before. Furthermore, the alkali can transform cellulose I to other
polymorphs such as Na-cellulose I and cellulose II, which are easily digested by enzymes as it
requires less activation energy to break down cellulose 42, 43. However, the fundamental
mechanism behind cellulose transformation in alkaline pretreatment is still not well investigated.
Thus, a reliable cellulose characterization method and study of cellulose crystalline structure,
especially on the molecular level, are vital to the development of alkaline pretreatment
techniques.
2.4.3 The flowthrough aqueous pretreatment
Recently, the flowthrough pretreatment reactor system has gained considerate attention due
to its capacity to fractionate biomass to produce higher yield of sugars with limited sugar
degradation loss than in the batch reactor system. This continuous pretreatment concept was
reported early in 1983 and pioneered by Bobleter and co-workers 44-47. The flowthrough
apparatus was described in a previous review on aqueous pretreatment, but it is still in its infant
stage 48. The current understanding of the flowthrough reactor system is that it can rapidly heat
20
up with a steam injection and cool-down within a short amount or/period of time 27 with a flow
of liquid passing through a packed bed of biomass, resulting in biomass hydrolysis 33. Also, the
flowthrough reactor provides the ideal mass and heat transfer by directing heating source-steam
with tubular reactor and fluidized sand bath instead of the oil/salts bath. It was reported that the
flowthrough can ensure that the same chemical reacts with biomass in pretreatment by pumping
a continuous pretreatment solution 27. Thus, the flowthrough reactor system is a promising tool to
study the interaction between chemicals and biomass and get accurate kinetic reaction data to
fundamentally study the biomass aqueous solubilization process.
Several pretreatment studies using flowthrough reactor have been reported. The hemi-sugars
recovery and total glucose yield operated under flowthrough system was examined (Table 2.1).
These previous studies showed higher sugar yield compared to the batch pretreatment. The dilute
sugar stream is a technical concern to the ones who want to commercialize, and it is raised when
discussing flowthrough pretreatment. A recent flowthrough pretreatment study reported a lower
minimum ethanol selling price than dilute acid, hot water, and steam explosion in a batch reactor
49. The extensive heat recovery, similar to an oil refinery, can be integrated, thereby sugar
dilution is no longer a limiting factor 49. In the early stage of flowthrough pretreatment study,
people focus on hemicellulose recovery while cellulose and lignin behaviors are untapped. Also,
the sugar (e.g. monomeric and oligomeric) releasing patterns and enzymatic hydrolysis
efficiency of pretreated slurries are not well understood. Additionally, little has been reported on
the chemistry of the major components of biomass, especially lignin chemistry, although people
have recently started to pay attention to lignin properties derived from flowthrough system.
21
Table 2.1 Literature reviews of hemi-sugars recovery and total glucose yield operated under flowthrough system; mono refers to
monomeric sugars while oligo represents oligomeric forms of sugars
Feedstock Reactor Temp.,
°C
Flow rate,
ml/min
Time,
min
Yield, % original hemi-sugars or
glucose
Lignin
Removal
Reference
Hemi-sugars Glucose
(Stage 1)
Glucose
(Stage 2)
Corn stover Hastelloy C-
276 tubing
(14.3 mL)
200, 2 or 25
Water
10 95 N/A N/A N/A 50
Cellulose Flow reactor
(3.6 mL)
295 10 Water 12 N/A 80mono+
20oligo
N/A N/A 51
Bamboo,
chinquapin(hardwood), and
Japan cedar(softwood)
Flow reactor
(3.6 mL)
180 for
20min
and
5°C/mi
n to
285
10 Water 60 >95 >95 N/A N/A 52
Corn stover 3.8 mL VCR
fitting
200 10 Water 16 90 <5 N/A 50 53
Corn stover 16.8 mL
tubular
200 10.5
Water
8 82 N/A N/A 30-40 54
Corn stover 3.8 mL 316
stainless steel
VCR fitting
180 0-10 (0.05
or 0.1
wt %
acid)
4-16 10-100 N/A N/A <50 55
Corn stover 14.3 mL 316
stainless steel
VCR fitting
180-
215
0-25 (0-
0.1 wt%
acid)
20 25-100 <5 50-96 10-70 56
22
Six woody and four
herbaceous
Hastelloy 2-
276 tubing
percolating
reactor
200-
230
1
water
0-15 100 (90
mono)
4-22 N/A 35-60 57
Sugar-cane bagasse and
leaves
316 stainless
steel 250 mL
190-
230
400-
500water
0.75-
4
100 (80
mono)
<10 N/A >60 58
Poplar Hastelloy 20.5
ml
200-
280
10-62.5
(water or
0.05% wt
acid)
2-10 90-100 10-100 Whole
slurries
to 100%
mono
60-100 59, 60
Poplar, sugarcane, bagasse 56 mL 316
stainless steel
180-
225
30 (water) 8-28 60-98 35-95 N/A 60-80 61
Miscanthus 20.5 ml
Hastelloy C
276
160 25 (water) 10-68 N/A N/A N/A Lignin
characteri
zation
performed
62
13C-enriched corn stover
stems
stainless steel
tube (D12.7
×L152mm)
170 20 (water) 0-80 <80 <20 N/A <50 63
Populus
trichocarpa×Populus
deltoides
stainless steel
tube (D12.7
×L152mm)
180 20 (water) 3-12 20-80 <5 N/A 20-55 64
Populus
trichocarpa×Populus
deltoids and Cellulolytic
Enzyme Lignin (CEL)
stainless steel
tube (D12.7
×L152mm)
140 or
180
20 12-
192
(wate
r)
N/A N/A N/A 5-75 65
23
In summary, the aqueous pretreatment played a significant role in biomass deconstruction
involving acids and alkalis. The high cost of aqueous pretreatment is still the main challenge for
commercializing. The purpose of the conventional pretreatment approaches is to increase the
digestibility of pretreated residual cellulose and hemicellulose. The substantial cellulose in the
residual solid residues resulted in high dosage use of enzymes, which are too expensive, resulting
in a cost of 0.17$ per gallon ethanol (9% production cost among total cost) 9, 11. The new concept
developed in our group is to fractionate the whole biomass, which can recover maximum high
yield of fermentable sugars (almost 100% of hemi-sugars and glucose) in the pretreatment stage
only, which potentially reduces enzyme dosage and achieves utilization of other usable by-
products (e.g. lignin) to increase profits. However, some aspects of this process, such as the
pretreatment process kinetics, mass transfer effect on pretreatment performance, cellulose
crystalline structural change in the process, and the characterization of derived lignin for
producing value-added products still requires further fundamental understanding. The
flowthrough reactor system can enable the study of these fundamentals.
2.4.4 Summary of biomass degradation chemistry in aqueous pretreatment
Biomass degradation chemistry is one of the fundamentals of aqueous pretreatment. It
possesses several reaction pathways. H+ and OH- concentration in pretreatment solution have a
great impact on the distribution of final products. The biomass in acidic conditions undergoes
mainly acid hydrolysis and dehydration reactions, and biomass degradation in alkaline conditions
includes peeling off reactions, termination stopping reactions, alkaline scission, and oxidative
alkaline degradation. Acidic and alkaline route conversion of biomass to bioproducts showed
potential abundant value-added products (See Figure 2.2). Firstly, cellulose hydrolysis can occur
24
under both acidic and alkaline conditions to produce monomeric and oligomeric sugars. Under
acidic conditions, sugars undergo further dehydration to form 5-HMF. Meanwhile, the sugars can
have reverse aldol and condensation reactions to generate aldehydes and humin respectively 66.
Under alkaline conditions, sugars suffer from severe peeling off reactions to form small organic
acids and aldehydes 67. Secondly, hemicellulose hydrolysis also can occur under both acidic and
alkaline conditions to produce sugars. The acidic degradation of sugars under acidic conditions
forms aldehydes (e.g. pruvaldehyde, glyceraldehyde, glycolaldehyde, acetaldehyde and
formaldehyde) and organic acids by reverse aldol reactions similar to sugars derived from
cellulose. Additionally, the alkaline degradation of sugars from hemicellulose mainly includes
various organic acids (e.g. glyceric, glycolic, formic, acetic, lactic and oxalic acids). At last, the
lignin chemistry under acidic conditions was reported to be predominantly centered about
depolymerization reactions via rupture of the β-O-4 linkages forming Hibbert ketones and
repolymerization by acid catalyzed condensation between the aromatic C6 and a carbonium ion
at Cα 68, 69, Cβ-C5 60, Cα-C5 70, and C4-O-C5. Under alkaline conditions, C5-C5 can be
generated through radical coupling reactions of lignin fragments. Also, enol ether structure is
easily formed the cleavage of β-aryl ether bonds in phenolic arylpropane units 71. However, the
chemistry showed is still not well defined, especially the cellulose and lignin degradation
mechanisms.
27
Figure 2.2 The aqueous degradation of major biomass components under acidic and alkaline
conditions; (a) cellulose, (b) hemicellulose, (c) lignin 60, 67, 71-74
28
2.5 Process characterization of flowthrough aqueous pretreatment
2.5.1 Cellulose characterization in aqueous pretreatment
Cellulose is the main target of aqueous pretreatment; thus, the characterization of its
structural and chemical changes is important. The hydrolysis of structural polysaccharides is still
the rate and cost limiting step in the conversion of lignocellulose into fuels. Although the
development of more efficient enzyme systems can help to resolve this issue in the long term, the
studies of the cellulose crystalline structure, surface accessibility, and interaction with enzymes
improves cellulose hydrolysis in the short term. Also, the study of cellulose structure can help
achieve the development of an effective cellulase enzyme system because the detailed cellulose
molecular structure is crucial to investigate its interaction with various enzymes.
2.5.1.1 Conventional characterization methods of cellulose
Cellulose is the most abundant and water insoluble biopolymer in the biomass cell wall
matrix. The growing and arrangement of plant amorphous and crystalline structures are affected
by species and environment, which makes cellulose structure from various sources complex and
different. Therefore, Cellulose from various sources possesses different degrees of
polymerization, crystallinity, surface structure, pore size, and heterogeneity of crystalline and
amorphous (or less crystalline) structures. The cellulose structure has been studied since the early
1900s. The bacterial and tunicate cellulose microcrystals have been studied as standard crystals
for investigating native cellulose crystalline structure. Thus, when developing a new tool to
characterize cellulose, one should always starts from these classical and standardized crystals.
The characterization of native cellulose in the plant cell wall is difficult due to the
interference of other matrix polymers (mainly hemicelluloses and lignin interlinks). Many
29
analytical tools have been seen employed to study cellulose structure recently. Advanced
imaging techniques, such as atomic force microscopy (AFM), have been used to observe the
cellulose surface structure. Also, other spectroscopic tools have demonstrated useful tools in the
characterization of cellulose molecular structure. These analytical tools of cellulose mainly
include X-ray diffraction (XRD), Fourier Transform Infrared Spectroscopy (FTIR), Raman,
Nuclear Magnetic Resonance (NMR), Neutron Fiber Diffraction, and computational calculation
and models. XRD has been used in studying cellulose crystalline structure for several decades.
Many studies have used XRD to investigate the cellulose crystal molecular and hydrogen
bonding network. In XRD analysis of cellulose, the crystallinity of cellulose samples can be
calculated by peak height ration, peak deconvolution, and amorphous subtraction methods.
However, crystallinity calculation depends on the method applied 75 and is not consistent in
different pretreatment studies. XRD has also been used in studying the lateral thermal expansion
coefficients (TECs) of cellulose crystals in the lateral direction up to 250°C to determine the
structural transformation of cellulose crystals to high temperature states 76. X-ray diffractograms
of cellulose crystals can be used to determine clinic phases of cellulose polymorphs as well as
crystal dimensional parameters (e.g. a, b, c crystal length, α, β, and γ crystal orientation angles)
77. In addition, Synchrotron X-ray and Neutron Fiber Diffraction in assistance with
computational calculation was also used to study hydrogen bonding and the unit cell parameters
(e.g. crystallographics, helical parameters n and h, glycosidic torsion angles φ=O-5---C-5—C-
1—O-1, ψ= C-5–C-1–O-1–C-4′ (φ, ψ), crystal density and dimensional parameters) in cellulose
two basic polymorphs 78, 79.
30
FTIR and Raman were used to determine the crystallinity of cellulose; however, the
calculation method was based on the peak ratio, which was inconsistent depending on the
selection of peaks 80. Also, the assignments of FTIR and Raman were not clear since the crowd
peak distribution 81, 82. NMR is another method of characterizing cellulose structure that can
identify C1-C6H in cellulose structure. Also, Iα and Iβ cellulose crystalline phases were first
demonstrated by solid-state NMR (13C CP/MAS). 13C NMR can detect and study C1 to C6.
However, the cellulose peaks in NMR spectra were hard to be distinguished from other polymers
in whole cell wall samples, e.g. hemicellulose and lignin, because of their complexity and
overlapping signals 83. NMR has also been used in cellulose amorphous or surface structural
study. However, the peak assignments of cellulose amorphous and surface regions have not yet
been well resolved 75.
2.5.1.2 New technique to characterize cellulose in aqueous pretreatment
In the characterization of cellulose in aqueous pretreatment, most studies focus on the
change of cellulose chemical compositions, degree of polymerization (DP) of cellulose, and
crystallinity. The molecular structure characterization has rarely been carried out; thus, the
cellulose crystalline structural transformation in aqueous pretreatment is not well understood.
The main limitation of these spectroscopic tools (NMR, FTIR, and Raman) is interference from
carbohydrates. Besides the spectroscopic tools, Scanning Electron Microscope (SEM) and AFM
were applied to image surface structure at the sub-nanometer resolution and morphological
information 84, but their primary application in cellulose characterization in aqueous pretreatment
is the observation of cellulose surface and crystal disruption. Also, the cellulose crystallinity
31
analysis is not always consistent when applying different calculation methods. The unclear
cellulose structure is a hurdle to understanding cellulose behaviors in aqueous pretreatment.
Due to the limitation of current cellulose characterization in aqueous pretreatment,
developing a new cellulose characterization method is needed to shred a light on a clear
understanding of cellulose characteristics in aqueous pretreatment. Also, the dynamic monitoring
of the structural transformation of cellulose during heating to mimic aqueous pretreatment is a
necessary foundation to achieve high C6 sugars yields as well as to help understand interaction
of cellulose crystalline cellulose with enzymes. Currently, the new emerging cellulose structural
characterization method is scanning SFG-VS (6cm-1) 85 as well as the high resolution SFG-VS
with a resolution of 0.6cm-1 28. The SFG-VS method can selectively characterize cellulose
crystalline structure without the interference of other biopolymers in biomass. Thus, the SFG-VS
tool has great potential to study cellulose structural change during biomass aqueous pretreatment.
2.5.2 Lignin production and characterization in aqueous pretreatment
Lignin is the most abundant natural source of aromatics, which are attractive as important
carbon precursors. The utilization of the second abundant biopolymer lignin in cellulosic
biomass to produce value added products can improve biorefinery economics. The conversion of
lignin to high value products can offset the high cost of pretreatment, which pushes the
biochemical conversion forward to commercialization. One of the Biochemical Conversion R&D
goals is to convert lignins from biomass into fuels, chemical intermediates, or materials but not
simply combusted to heat and utility supplies 8. The U.S. Energy Security and Independence Act
of 2007 mandates that 79 billion liters of cellulosic biofuels be produced annually by 2022, and,
for that to occur, about 62 million tons of lignin will be produced 86, 87. More importantly, besides
32
the lignin residues burned to meet the demand for the internal energy use, technologies are
needed to convert remaining lignin residues to value added products 87. The conversion of
carbohydrates to valuable products (e.g. ethanol, carboxylic acids, levulinic acid, glycerol, and
sorbitol) is well developed, but the utilization of lignin is not well performed. However, research
on lignin valorization has gained tremendous attention in the past five years. The aqueous
pretreatment can provide various lignin sources with different lignin chemical compositions and
structural characteristics (e.g. molecular weight, solubility, yield, purity, and functionality) for
valorization. The linkages abundance, monolignol types, molecular weight, and amount of lignin
are varied in different engineered biomass species. Thus, the characterization of various lignin in
aqueous pretreatment is very important for lignin valorization.
2.5.2.1 The significant role of lignin in aqueous pretreatment
Different biomass species present different lignin content, composition, and structural
information. For example, hardwood contains about 10% less lignin than softwood lignin;
eucalypt wood consists more syringyl units dominant in β-O-4' followed by β-β' linkages,
whereas the spruce shows more guaiacyl units and linked by mainly β-O-4' and phenylcoumaran-
type bonds 88, 89. Almost all lignin is linked to polysaccharides with covalent bonds, mainly
hemicellulose 90. Additionally, hemicellulose and lignin polymers are ether, ester, and hydrogen
bonds linked while lignin and cellulose have fewer bonds. Therefore, hemicellulose is like the
glue between lignin and cellulose. The separation of lignin from biomass requires the links
between polysaccharides to break down, which aids the difficulties of selective isolation of
certain lignin intermediates. Additionally, the challenges of studying native lignin include the
isolation process, interference of polysaccharides, and analytical method limitations. The lignin
33
isolation and purification methods usually caused structural modification on the lignin
macromolecular matrix. The dioxane/water extracted lignin was known as Bjorkman lignin or
ball milled wood lignin, which represents the native lignin prototype. This lignin extraction
method was believed to cause the least significant structural modification on native lignin;
however, the yield was up to 30%, as reported 91.
Lignin was demonstrated as the most recalcitrant compound in biomass aqueous
pretreatment. Lignin removal can not only improve pretreatment efficiency but also reduce
inhibition on subsequent enzymatic hydrolysis and fermentation processes. However, complete
lignin removal was reported insufficient for cellulose accessibility to enzymes as expected,
which was proposed to attribute to the aggregation of cellulose microfibrils 92. Thus, a detailed
study of lignin behaviors and sugar yields in aqueous pretreatment is not clear and still needed.
In aqueous pretreatment of biomass, the isolation of lignin can alter several lignin
characteristics, including depolymerization, repolymerization, molecular weight, solubility,
functionality, and interunit linkages. In addition, no reported tools are able to selectively and in-
situ observe lignin structure yet, which makes the lignin structure remain undefined.
Comprehensive reviews of the chemistry and structural properties of lignin were reported 89, 93-97.
Thus, different lignin isolation processes result in the diversity of lignin sources in molecular
weight, purity, solubility, and structural characteristics, which have a great impact on the
availability of lignin substrates, conversion to different value products, and economics.
The challenges of lignin production from cellulosic biomass are the complex three
dimensional structure, various and random interunit linkages, low product selectivity, and the
effects caused by bonded polysaccharides Besides, native lignin isolation, lignin has been
34
produced by several industrial processes, including pulping processes. Recently, only three lignin
sources have been available in the commercial market, namely, Kraft, lignosulfonate, and
organosolv lignin 98. These lignins make good comparisons to aqueous pretreatment derived
lignins in characteristics and market potential. These data of commercial lignin markets can help
with the development of aqueous pretreatment derived types of lignin. To explore further
utilization of lignin derived from aqueous pretreatment, the consistent characterization of lignin
intermediates and compared their properties to commercialized lignin are significant.
2.5.2.2 Lignin production and characterization in aqueous pretreatments
The aqueous pretreatment can cause the change of lignin structure and characteristics such
as the molecular weight, solubility, phenolic hydroxyl groups, aliphatic hydroxyl groups,
methoxyl groups, and inter-units linkages of lignin. The lignin characterization in aqueous
pretreatment is important to gain a fundamental understanding of aqueous pretreatment and
effective for producing lignin intermediates for different applications.
2.5.2.2.a Lignin derived from hot water and steam explosion pretreatment
The chemistry and structural transformation of lignin under hot water pretreatment are
similar to dilute acid pretreatment. This is because the pH of water decreases with increasing
temperature resulting in lightly acidic solutions. Also, releasing organic acid from acetyl groups’
hydrolysis from hemicellulose can result in the decreasing of pH of the pretreatment media. The
hot water flowthrough pretreatment was studied on Miscanthus and the mechanism was proposed
as the selective hydrolysis of β–O–4 sub-structures together with cinnamic and/or coumarylate
ester linkages cleavage followed by oxidative cleavage at C–C position, in conjunction with the
preferential removal of S-lignin during the delignification process 62. Little softwood lignin’s
35
chemistry or structural transformation is unknown. Thus, a detailed study of softwood lignin
chemistry and how it’s different from other species is significant.
Steam explosion pretreatment suddenly releases reaction pressure in steam gun/reactor to
degrade lignin. The recovery of lignin in steam exploded liquid is carried out by acidifying the
liquid to pH 2-3 to precipitate lignin followed by washing and filtration. The yield is about 70%
16. The steam explosion derived lignin has a wide range of polydispersity 4-38, while the average
molecular weight is about 3000Da 99. The reaction mechanism of lignin during steam explosion
was proposed by Li et al. 68, which described the generation of organic acids through the initial
hydrolysis of acetyl groups in hemicellulose, which leads to the formation of a carbonium ion
that can react with benzylic groups to eliminate ROH groups. Simultaneously, the carbonium ion
causes cleavage of β-O-4 inter-unit and causes the formation of Hibbert Ketones through
acidolysis reactions. However, besides the depolymerization reactions, unlike lignin
condensation C-C position in conventional pulping, the condensation of lignin also occurs in the
electron rich C-2 or C-6 positions during steam explosion. The steam pretreatment using a steam
gun can remove more than 95% of the xylan in the biomass. However, only a limited amount of
lignin is recovered because most of it (>85%) remains in the residual solids.
2.5.2.2.b Acid pretreatment lignin
The dilute acid pretreatment was usually carried out at temperature of 120-210°C with an
acid concentration of less than 4% for 0-60min. Besides the popularity of sulfuric acid,
hydrochloric, nitric, and phosphoric acids were also applied. The chemistry of acid degradation
of lignin is the acidolysis of aryl ether linkages (primarily β-O-4 linkages) and acid catalyzed
recondensation without significant cleavage of any other linkages. Dilute acid or steam explosion
36
pretreatment of biomass led to a decrease of β-O-4 68, 100, 101 as well as a decrease in the
protonated and oxygenated aromatic carbons per aromatic ring 100. Also, the molecular weight
distribution of lignin would be expected to decrease together with repolymerization or
condensation reactions, resulting in a more heterogeneous lignin structure. Several lignin
behaviors were reported in dilute acid pretreatment, namely, a) lignin droplet formation and
redeposit, b) pseudo lignin formation, and c) lignin migrating. Lignin droplets were formed at the
pretreatment temperatures higher than the phase transition temperature (e.g. 120-200°C 102),
which caused lignin to liquefy and coalesce in the pretreatment liquid, and the droplets might
travel in the cell wall matrix and deposit back to the biomass surface 103, 104. The droplets were
then reported to have a higher S/G ratio than the original lignin and showed inhibition on
enzymatic hydrolysis of pretreated slurries by blocking the cellulose surface from enzymes
attack 105. The pseudo lignin was formed by carbohydrates and lignin degradation products such
as furfural 106. The lignin migration and distribution in biomass plant cell walls were observed in
the dilute acid pretreatment system 107. Lignin has the tendency to form spherical droplets, which
causes deposition onto the biomass surface. It was observed that, during dilute acid pretreatment,
lignin droplets generated and stayed on the biomass surface 103, which suggests that the majority
of the lignin did not change significantly after pretreatment. The kinetics of lignin is not well
studied and mainly focuses on hemicellulose hydrolysis.
The utilization of acid pretreatment derived lignin depends on the separation and purification
of pretreatment slurries. Three main separation methods of lignin from acid pretreatment slurries
were reported, namely acid precipitation by gradually adjusting the pH of the solution 108. The
second method is to use various solvents such as organic solvents, enzymes, biomimetic catalysts
37
and ionic liquids to extract lignin 109. The third method is to use membrane technology, which is
good for sieving lignin fractions with various the molecular weights 110.
In high pretreatment severities, sugars suffer from sugar degradation loss. Therefore, the
traditional pretreatment methods tend to protect carbohydrates from degradation and remain in
biomass solid phase by performing low pretreatment severities pretreatment. However, lignin
removal from biomass is insufficient in low pretreatment severities. Substantial lignin remains as
residual lignin in biomass solids together with the majority of cellulose 35, 36. For example, at
0.1% wt sulfuric acid batch tubular pretreatment at 180°C for 20min, 71.7% xylan was removed
from the biomass solid while only 1.5% lignin was removed 56. Therefore almost 98.5% of lignin
and intact cellulose remained in biomass residues, which is not effective for cellulose and lignin
utilization as mentioned above. The direct recovery of this residual lignin from the biomass
residues adds extra cost. Furthermore, the recovery of lignin after enzymatic hydrolysis or
fermentation produced poor quality lignin and a low ethanol production yield.
The development of flowthrough reactor increased aqueous phase lignin recovery while also
ensuring high enzymatic hydrolysis digestibility. For example, by increasing the flow rate to
25ml/min, the pretreatment with 0.1% wt sulfuric acid pretreatment at 180°C for 20min achieved
100% xylan removal and 68.4% lignin removal as well as 30% more digestibility than in the
batch reactor system 56. Furthermore, high temperatures more than 200°C were used in
flowthrough reactor and reported to achieve almost 100% of biomass solubilization, in which
lignin was recovered almost 100% in the aqueous phase 59. However, the characterization of
lignin and cellulose in the aqueous flowthrough pretreatment is not well conducted. Also, the
flowthrough reactor suffers from high water consumption, resulting in a low sugar concentration
38
for downstream production. The lignin derived from flowthrough pretreatment can avoid
contamination of enzymes and chemicals to lignin intermediates, which greatly improves the
efficiency of using lignin. In summary, the flowthrough concept of biomass aqueous
pretreatment can greatly improve lignin recovery, isolation, and utilization.
2.5.2.2.c Alkaline pretreatment lignin
In carbohydrates, sodium hydroxide can disrupt hydrogen bonding network in cellulose and
hemicellulose, resulting in the swelling of carbohydrates. It was also reported that sodium
hydroxide can break the ester linkages between lignin and hemicellulose and the deprotonation
of phenolic groups 111. Sodium hydroxide is effective in lignin removal. Similar to sodium
hydroxide pretreatment, lime pretreatment can also remove lignin and improve enzymatic
digestibility of pretreated substrates 112, 113. However, the price of sodium hydroxide is higher
than other alkalis. For example, lime is less expensive than sodium hydroxide, and the recovery
of lime is easier when it reacts with carbon dioxide and insoluble calcium carbonate in water.
The use of ammonia hydroxide in biomass pretreatment has developed as Ammonia Freeze
Explosion (AFEX) and Ammonia Recycle Percolation (ARP). AFEX treats biomass by applying
ammonia and then suddenly releasing high pressure, which causes both chemical and physical
effects on the biomass 114. The ARP process is an improvement in pretreatment reactor design, in
which a flowthrough column reactor is used. This reactor design can eliminate away biomass
degraded products, which helps achieve high yields of depolymerized lignin intermediates.
However, the drawback of using ammonia is the possible inclusion of nitrogen into the lignin
structure, similar to sulfur being added to Kraft lignin structure 115.
The alkaline pretreatment caused saponification of intermolecular ester bonds crosslinking
39
xylan hemicelluloses and lignin 116. The delignification mechanism of alkaline pretreatment has
been the subject of many studies for years. The lignin depolymerization and isolation are difficult
due to the complexity of the lignin structure and the tendency of lignin to repolymerize.. In
alkaline pretreatment of biomass, the delignification mechanism involves an alkaline oxidative
by OH- anions on biomass caused by single electron/radical reactions, which caused cleavage of
both ether bonds and C-C bonds. The oxidative oxygen can be transformed to superoxide
radicals (-O2) by reacting with phenolic hydroxyl groups under pH>12. Several organic acids
were produced during the reaction. The nucleophilic attack also caused ring opening to produce
dicarboxylic acids. Meanwhile, condensation reactions happened in the C5 positions to have
Biphenyl methane structure 115. Thus, due to the different lignin removal mechanisms in alkaline
pretreatment compared to acid pretreatment, the produced forms of lignin should possess
different properties. The different properties make the tailoring of these lignins necessary for
utilization.
2.5.2.2.d Organosolv pretreatment lignin
The organosolv pretreatment can separate lignin from carbohydrates in biomass. Similar to
organosolv pulping, the biomass can be fractionated completely into lignin, aqueous sugar
stream, and solid cellulose 117. Organosolv pretreatment can solubilize lignin into the
pretreatment solvent or washing liquor. The amount of organic solvents together with water
applied varied in the pretreatment. For example, some studies applied low boiling point
methanol, ethanol, and acetone varied between 50-80%, which is miscible in water. The addition
of mineral acids (i.e., hydrochloric, sulfuric, and phosphoric) and organic acids (i.e., formic,
acetic, and peracetic acids, dimethyl sulfoxide, ethers, ketone, phenols, oxalic, acetylsalicylic,
40
and salicylic) can accelerate lignin delignification 18, 118. In some cases, several high boiling point
alcohols can be used at atmospheric pressure, such as ethylene glycol, glycerol, and
tetrahydrofurfuryl 18. The organosolv pretreatment produces a relatively pure, non-degraded, and
sulfur-free lignin. The drawbacks of the organosolv pretreatment are the high cost of solvents
and the difficulties of solvent recovery. Organosolv lignin is believed to keep even more of the
native lignin structure than ball milled lignin. Organosolv lignin can be precipitated by lowering
the solution pH for recovery as a separated fraction. The organosolv pretreatment is expensive
due to the high cost of solvents and solvent recovery. A process design and economics study
should be carried out to evaluate whether the profits of lignin as by-products converted to high
value products can offset the high cost. Also, the application of organic acid as a solvent might
cause subsequent enzymatic hydrolysis efficiency 119. However, the organosolv lignin is an
important native lignin representative.
The lignin isolation mechanism for alcohol extraction is OH¯, which attacks the linkages
between lignin and hemicellulose such as ester bonds as well as the ether and 4-O-
methylglucoronic bonds. The cleavage of β-O-4 linkages is the major lignin solubilization
mechanism in organosolv pretreatment 115. The cleavage of other ester bonds is also the main
mechanism of organosolv pretreatment (α-O-aryl ether> β-O-aryl ether). However, C-C
condensation reactions were observed in organosolv pretreatment (α-5 condensations) 120. This
C-C condensation position is different from that of other aqueous pretreatment. The lignin
degradation chemistry in organosolv pretreatment was reported 121 to be the initial solvolytic
cleavage of α-O-aryl ether linkages to form Quinone methide intermediate. It was followed by
the nucleophilic substitution in the benzylic position through the application of a SN2 mechanism
41
for the formation of a benzyl carbocation under acidic conditions 17, 121. In organosolv lignin
chemistry 121, 122, β-O-aryl ether linkages were broken by homolytic cleavage to form
formaldehyde and stilbenes. During organosolv pretreatment, a high yield of Hibbert’s ketones
with carbonyl groups were generated. The reactive benzyl carbocation intermediate from both α-
O-aryl and β-O-aryl ether cleavages can condense with another electron-rich carbon atom in a
neighboring lignin unit. The use of organic acids in organosolv pretreatment is different from
alcohols. For example, the hydroxonium ion can be generated from acids, which causes ring
hydroxylation in the lignin structure. The potential of using strong organic acid might have
oxidative reactions or ring opening besides the cleavage of ether bonds of lignin structure.
The molecular weight of organosolv lignin is lower than milled wood lignin which results in
the high solubility of solvents, which makes the lignin easy to use. Meanwhile, organosolv lignin
shows less sulfur content compared to Kraft pulping and lower oxygen content than ball milled
lignin. The hydroxyl (OH) functionalities, such as aliphatic, condensed and uncondensed
phenolics, and carboxylic OH groups in organosolv lignin were widely studied with 31P NMR
spectroscopy. These studies suggested the decreasing of aliphatic OH content while increasing of
the phenolic and carboxylic OH content 123, 124.The utilization of lignins depends on the
downstream industrial process needs on lignin properties. The modification of lignin in the lignin
separation process is required to meet the demand of the purity, solubility, molecular weight, and
polysaccharides contaminations of lignins.
The above description of aqueous pretreatment derived lignin behaviors indicated that the
fundamental understanding of lignin behaviors and chemistry in aqueous pretreatment are not
well performed, especially biomass lignin in the flowthrough system, which can reveal accurate
42
lignin chemistry by separating lignin with fast mass and heat transfer. The valorization of these
aqueous pretreatment derived lignins to materials, chemicals, and fuels relies on the
characterization of lignin during aqueous pretreatment.
2.5.2.3 Current lignin structural characterization methods
Lignin produced in an early stage of the process (i.e. after the biomass aqueous
pretreatment) can become a potential value-added co-product, improve enzymatic digestibility,
and reduce fermentation inhibition 125. Along with cellulose and hemicellulose, lignin presents
15–40% weight and contains the highest specific energy content among other biopolymers in
cellulosic biomass 126. Lignin consists of methoxylated phenylpropane units with three primary
monomers (monolignols): p-coumaryl, coniferyl, and sinapyl alcohols. The biosynthesis of lignin
is carried out by the oxidative coupling of monolignols to form a three dimensional biopolymer
through certain C-O-C and C-C linkages 89. The oxidative coupling is active in O–4, C–1, C–3,
C–5, and C–β, resulting in β–O–4, β–5, β–β, and β–1 formation 89. Due to the diversity of lignin
properties, lignin characteristics depend on different biomass origins, the environmental
conditions of plant growth, and various extraction methods. Lignins with the random formation
of linkages and the three dimensional amorphous structure of lignin are polar polymer, exhibiting
a poor solubility in a polar solvents such as polyolefin 127. Lignin characterization methods
advancement is crucial to track lignin structural modification during lignin recovery from
biomass aqueous pretreatment. Developing an analytical method to selectively characterize
lignin in the whole lignocellulosic material without isolating it is helpful in-situ study of native
lignin structures. Table 2.2 listed major lignin characterization methods applied in lignin
chemical composition and structural studies.
43
Table 2.2 Lignin characterization methods and targeted properties
Lignin characterization methods Target lignin characteristics Reference
Chemical
composition
K-lignin Acid insoluble lignin content; Lignin purity 128
Acetyl Bromide Lignin content; Lignin purity 129
Ultraviolet-visible
spectrophotometry
(UV-Vis)
Soluble lignin content 130
TAPPI method Acid insoluble lignin 131
Kappa number
(KMnO4 consumption)
Purity/lignin content 132
Thermogravimetric
analysis
Thermal stability 133, 134
Elemental analysis
(ICP)
Elemental composition (C, H, O content) 135
Pyrolysis gas
chromatography mass
spectrometry (Py-GC-
MS)
S/G ratio and intrinsic lignin fractions 136
Gel permeation
chromatography (GPC)
The number-average and weight-average
molecular weights (Mn and Mw,
respectively)
137
XPS Concentrations of functional groups, the
valence band structure and O/C atomic ratio
138, 139
Exclusion
chromatography (SEC)
Molecular size distribution 137
GC/GC-MS Soluble lignin identification and semi-
quantification
140
Structural
characteristics
Gel-state 2-D NMR
(HSQC)
Whole plant cell wall 2-D-NMR profiling
studies
83
31P NMR Hydroxyl group content (Aliphatic, phenols,
catechol , syringyl, Guaiacyl, condensed
lignin, carboxylic acid OH)
141
13C NMR qualitative and quantitative understanding of
lignin structure
142
1H NMR Identify functional groups: phenolic
hydroxyl (ArOH), aliphatic hydroxyl
143
44
(AlkOH), and methoxyl (MeO) groups and
quantify carboxylic acids, aromatic
hydrogens, and formyl and methoxyl groups
2-D HSQC NMR Identifying structural units 144
Solid state NMR In situ observe lignin structure without
lignin isolation, lignin content in biomass,
and identify condensed structure
145, 146
FTIR spectroscopy Structure, functionality, and lignin content 147, 148
Raman Structural information (aromatic, syringyl,
and guaiacyl)
149
Acidolysis (at 100 °C
in 0.2 M HCl in
dioxane±water (9:1,
v/v))
Monomeric and dimeric acidolysis products
to assist analysis by gas chromatography
after silylation, include β-O-4, β-5, β-β, β-1,
glyceraldehyde-2-aryl ether, 2-
aryloxypropiophenone, cinnamaldehyde,
cinnamic acid, benzaldehyde, benzoic acid
and quinoid types and selective cleavage of
arylglycerol-ethers and some other types of
labile ether linkages; only structural units
bound by arylglycerol b-ether bonds
137, 150
Thioacidolysis (boron
trifluoride in dioxane±
ethanethiol solution)
5-5 or 5-O-4 bonded dimers, monomeric
compounds to assist gas chromatography
detection after silylation; only structural
units bound by arylglycerol b-ether bonds
151, 152
Permanganate
oxidation (dimethyl or
diethyl sulfate
alkylation then
oxidation)
Oxidize lignin side chain to form carboxylic
acids (Di/Monocarboxylic acids) for gas
chromatography detection (e.g. biphenyls,
catechol and diphenyl ethers)
153
Nitrobenzene and
cupric oxide oxidation
(alkaline solutions at
high temperature)
Assist to spectroscopic studies 154
Ozonolysi Destroy double bonds and the aromatic
rings, leaving the side chains intact in the
form of carboxylic acids
137
NMR spectroscopy is a powerful and non-destructive tool in lignin structural
characterization that is widely used and developed. NMR spectroscopy technology is based on
the intrinsic spin property of nuclei. In a strong magnetic field, some molecule spins align with
45
the field, resulting in a slight net alignment parallel to the field. The energy of the spin is
detected as signal to reveal the information about individual atoms in a molecule and their
environment 155. Since lignin structure has no repeating linkages in the same environment, the
molecular structure can be accurately resolved by NMR. NMR was important in identifying
novel units such as dibenzodioxocins 156. However, the NMR characterization of lignin in
biomass suffers from the overlapping of polysaccharide peaks 83. Also, the quantification of
lignin interunit through either 1-D or 2-D spectra could not give absolute values 157, 158, which
only provides relative proportions of various peaks. 4D-, 5D-, and higher-D-NMR have been
applied to characterize labeled proteins 159. Thus, 3D- NMR will probably be an evolution of
current NMR tools for lignin structural characterization if it can be widely used to resolve
overlap and confusion of 2-D NMR. Besides structural characterization, the characterization of
several other lignin properties such as purity, molecular weight, OH content, and linkages
distribution is significant too since those lignin characteristics are important for further lignin
valorization. For example, the low molecular weight, narrow polydispersity, and total OH
content (aliphatic and phenolic) are beneficial for antioxidant activity 127. However, the majority
of lignin characterization only provides relative proportions or values compared to native lignin
rather than absolute values for each proportion. The lignin characterization methods displayed
above serve important roles in characterizing lignin in aqueous pretreatment to unravel lignin
behaviors and properties in aqueous pretreatment.
2.6 Preprocessing of feedstock for aqueous pretreatment
Mechanical refining before aqueous pretreatment is a critical supplement to aqueous
pretreatment in order to achieve a better fermentable sugars yield. Aqueous pretreatment
46
involves a penetration of catalyst into biomass 160, such as water, acid and alkali, to extract
dissolved biomass components. Such a solvent penetration process involves solvent wetting
biomass, impregnating biomass, and diffusing into biomass. Since the structure of biomass is
porous and heterogeneous 161, solvent diffusion during aqueous pretreatment is greatly
influenced by the pore size of particles and biomass shapes. However, intensive size reduction of
the biomass causes not only the energy-intensive and expensive grinding and milling before
chemical pretreatment, but it also might lead to the losses of carbohydrates during physical
treatment. The feedstock preparation cost represents 30% of the total cost in biological
conversion to bioethanol. To make the ethanol production economically feasible, the energy cost
from preprocessing should be reduced. It can be achieved by balancing the pretreatment
performance while using chips that are as large as possible in the pretreatment 162. The first step
towards this is the fundamental understandings on the effects of particle sizes and shapes of
biomass on pretreatment performances.
Recent studies on the effects of biomass sizes showed inconsistent results. Some studies
reported that smaller sizes led to better pretreatment performance. But some other studies found
the opposing results that those larger sizes can achieve better sugar yields. For example, the
steam explosion pretreatment has been shown to favor smaller sizes 163, while the steam
explosion with SO2 as a catalyst of Douglas Fir showed the best pretreatment sugar yields at size
of 5×5cm rather than smaller sizes (e.g. <0.4mm or 1.5×1.5cm) 164. Some studies on the effects
of biomass chip particle size on pretreatment effectiveness of woody biomass as well as
enzymatic hydrolysis effectiveness for biological processing were also conducted 165.
Furthermore, no studies have ever reported the optimization of biomass chips cutting in terms of
47
energy efficiency and pretreatment performance. The systematic study of particle sizes as well as
biomass cutting methods, which involves mass transfer effects, is important for cost effective
bioethanol conversion. Similar to most of chemical processes in large scale commercialization,
kinetic reactions, heat transfer, mass transfer, and momentum transfer are technical focuses.
Knowledge deficiencies in these four aspects of engineering are a hindrance for pretreatment
commercialization.
Kinetic and mass transfer modeling is a potential tool to study the effects of biomass sizes
and shapes on pretreatment. Saeman proposed the first acid hydrolysis model of cellulose based
on the assumption of a homogeneous two-step first order reaction in 1945. In the model,
cellulose was first converted to glucoses and then degraded to degradation products such as
hydroxymethylfurfural (HMF) 166. Further work found that first-order reactions were not
accurate for heterogeneous biomass degradation. Traditional kinetic modeling is deficient in
revealing true information of biomass reactions. Nevertheless, the heterogeneous structure of
hemicellulose in biomass was shown to have great effects on the reaction rate of hemicellulose
dissolution, in which mass transfer played an important role in pretreatment performance.
Solubility limitations, mass transfer, homogeneous or nonhomogeneous reaction types all
become important parameters to consider for kinetic modeling of aqueous pretreatment.
In summary, biomass particle size and cutting approaches are important factors in energy
effectiveness. Also, lignocellulosic biomass particle size, particle microscopic structure, porosity,
and particle shape have a significant impact on pretreatment solvent diffusion, even on
enzymatic hydrolysis efficiency 167. Thus, the mass transfer modeling is an effective approach to
optimize biomass cutting approaches and particle sizes.
48
2.7 Conclusion
This chapter has mainly reviewed current research status, process characterization, and the
significance of feedstock preprocessing of aqueous pretreatment. The key fundamental issues
and challenges are identified. First of all, the sugars undergo degradation loss with low aqueous
cellulose and lignin solubilization in aqueous pretreatment. The biomass aqueous pretreatment is
not well studied with different pH 35, 36. Although pH is an important factor to pretreatment sugar
yield, its effects on pretreatment performance were not were studied 168. Secondly, much less
effort has been devoted to cellulose structural transformation in aqueous pretreatment or
enzymatic hydrolysis. In addition, lignin structural modification and properties are not well
characterized, which impedes further utilization. At last, the effects of preprocessing feedstock
into different shapes and particle sizes in aqueous pretreatment in energy cost and pretreatment
performance are not well understood. The successful large scale production and
commercialization rely on resolving these key fundamental problems of aqueous pretreatment to
create cost competitive bioethanol versus conventional gasoline, which is the focus of this thesis.
49
CHAPTER THREE
CHARACTERIZATION METHOD DEVELOPMENT FOR CRYSTALLINE
CELLULOSE USING HIGH RESOLUTION BROADBAND SUM
FREQUENCY GENERATION VIBRATIONAL
SPECTROSCOPY IN BIOFUEL
3.1 Abstract
Here we conducted the first sub-wavenumber high-resolution broadband sum frequency
generation vibrational spectroscopy (HR-BB-SFG-VS) study on both the C-H and O-H region
spectra of crystalline cellulose. HR-BB-SFG-VS has about 10 times better resolution than the
conventional scanning SFG-VS and is known to be able to measure the intrinsic spectral
lineshape and to resolve much more spectral details. With HR-BB-SFG-VS, we found that in
cellulose from different sources, including Avicel and cellulose crystals isolated from algae
Valonia (Iα) and tunicates (Iβ), the spectral signatures in the OH regions were unique for
different allomorphs, i.e. Iα and Iβ, while the spectral signatures in the C-H regions varied in all
samples examined. Even though the origin of the different behaviors of the crystalline cellulose
in the O-H and C-H vibrational frequency regions is yet to be correlated to the structure of
cellulose, these results provided new spectroscopic methods and opportunities to classify and
understand the basic crystalline structure, as well as variations, in polymorphism of the
crystalline cellulose structure.
Keywords: Cellulose Iα • Cellulose Iβ • Avicel • High Resolution Broadband Sum Frequency
Generation Vibrational Spectroscopy (HR-BB-SFG-VS)
50
3.2 Introduction
Understanding the cellulosic biomass recalcitrance at the molecular level can be the key in
overcoming the fundamental barrier to make cellulosic biofuels cost-competitive 169. Although
plant cell walls are complex and dynamic, recent advances in analytical chemistry and genomics
have substantially enhanced our understanding, and also highlight how much the knowledge-gap
remains 170. The basic chemical constituents of biomass are well known. The predominant
polysaccharide of biomass is cellulose, which makes up about 30 to 50% of biomass in the form
of linear fibrils composed of approximately 30 to 40 hydrogen-bonded chains of -(1, 4)
glucopyranosides with a native degree of polymerization ranged from several hundreds to over
ten thousands, is the most abundant organic polymer on Earth. In cellulose structures, hydrogen
bond network exists in two conformations, i.e. intra-chain and inter-chain hydrogen bonding.
The intra-chain hydrogen bonds connect 6COH with the 2COH and 3COH with endocyclic
oxygen. Inter-chain hydrogen bonds link 3COH with 6COH 171. Intermolecular hydrogen bonds
are parallel to the glycosidic linkage for both types of chains. Due to its hydrogen bonding
diversity, cellulose can exhibit various supra-molecular structures, including amorphous, para-
crystalline and crystalline structures.
Although based on cellulose crystalline structures, at least four types of cellulose, namely
cellulose I, II, IIII,II, and IVI,II have been reported, cellulose I is the only existing form of natural
cellulose 172. Furthermore, native cellulose I has been shown to be composed of two distinct
crystalline forms, referred to as Iα and Iβ 173. Cellulose Iα is triclinic parallel crystalline with
only one repeating cellobiose in the unit cell while Iβ is monoclinic parallel-down crystalline
structure 174. The relative amounts of I and I have been found to vary significantly among
51
cellulose samples from different origins. For example, cellulose Iα is the dominant form in algae
and bacterial cellulose, while Iβ is found dominant in tunicate and higher plants 173. Cellulose
materials isolated from different organisms and the methods used to isolate them may result in
various different forms with different physiochemical properties, such as crystal size, shape, DP
and crystallinity 175. Previously, cellulose Iα and Iβ have been characterized by CP/MAS 13C
NMR 78, 79, 173, and molecular dynamics simulation. However, these studies only reflected
ordered crystalline and disordered amorphous structure. Other cellulose structures with different
crystalline packing forces and orders have not been studied in details 176, 177. These differences in
crystalline packing forces and orders have a broad impact on a variety of chemical and
enzymatic reactivity with cellulose 178. Thus, investigation of crystalline structural characteristics
in different cellulose polymorphs is a fundamental step to understand crystalline structure
recalcitrant to enzymatic hydrolysis in bioconversion of biomass.
Many cellulose characterization techniques provide capabilities to image and characterize
biomass, including: a) Atomic force microscopy (AFM) which allows researchers to not only
investigate chemical compositions of different layers but also characterize the topographic
structures of the cell wall surface 179, 180; b) High resolution solid state CP/MAS 13C NMR
powered in the quantitative analysis of structural features of cellulose, especially of the
crystalline allomorph and disordered domains. The line-shape spectral fitting analysis of solid
state NMR spectra allows a detailed comparison and characterization of the cellulose supra-
molecular structures and has been extensively used to investigate the structural characteristics of
cellulose 181, 182. Cellulose Iα and Iβ were first identified by 13C NMR 173. However, 13C CP/MAS
NMR results are usually interfered by lignin, hemicellulose and pectin 183.
52
In addition to solid-state nuclear magnetic resonance (NMR), native cellulose
polymorphism, crystallinity, hydrogen bonding, and mechanical strength were also studied by IR
and Raman vibration spectroscopy, and X-ray diffraction (XRD) 184-193. XRD has been widely
applied to quantify and identify cellulose crystalline structure. However, crystallinity index (CrI)
calculated from XRD varied depending on calculation methods 194. Infrared or Raman
vibrational spectroscopy of cellulose was also reported to be advantageous in understanding
cellulose structure among other analytical methods 195. FT-Raman was developed to quantify
cellulose crystalline structures. However, specific spectral signatures of cellulose with different
crystallinity have not been well established using IR or Raman spectra. Moreover, quantified
crystallinity results from different characterization techniques are not always unique or consistent
with each other when using different vibrational peaks. For example, Schenzel et al. employed a
spectral deconvolution method by using bands at 1462 and 1481 cm-1 80, while other researchers
applied different peaks as deconvolution bands thus resulted in inconsistent results. Due to
complex network of cellulose with other polymers as well as different growing environment of
cellulose species, cellulose structural information from these techniques is often interfered by
other amorphous polysaccharide components. Therefore, isolation and purification of cellulose
from biomass are performed before applying these techniques. However, the isolation and
purification process can affect cellulose structures, thus making characterization of native
cellulose difficult. Therefore, to avoid interference of other biopolymers from biomass and
tedious cellulose isolation process, a nondestructive spectroscopic technique that can selectively
probe the cellulose crystallinity is needed for characterization of biomass.
Recently SFG-VS has been used to study the physiochemical properties of crystalline
53
cellulose, as the crystalline cellulose exhibits enormous nonlinear optical response 85, 171, 196-201.
SFG-VS is a second-order nonlinear spectroscopic method that is sensitive to the structural
symmetry of the molecules and molecular assembly such as the cellulose crystalline units.
Therefore, it has shown great potential in studying cellulose microfibril arrangement and
orientation at different stages of plant growth or from different sources 197.
In SFG-VS measurement, usually a visible laser beam with a fixed frequency (𝑣𝑖𝑠) and an
infrared (IR) laser beam with tunable frequency or broadband (𝐼𝑅) simultaneously interact with
the material to generate a signal at the sum of the two frequencies (𝑆𝐹𝐺 = 𝑣𝑖𝑠 + 𝐼𝑅) that
give the vibrational spectra of the material. By tuning the IR frequency, the SFG vibrational
spectra of the interested material can be measured. SFG-VS is distinct from the IR and Raman
spectroscopy because SFG has unique symmetry properties as a second order nonlinear process.
Such symmetry properties require the molecule or molecular assembly in the materials that
generates SFG signal are with broken centrosymmetry, i.e. the material is either a non-
centrosymmetriuc crystalline material or a molecular surface/interface where the molecules are
aligned due to asymmetric forces at the surface/interface 202, 203. For bulk amorphous solids or
liquids, SFG-VS is surface/interface selective, i.e. forbidden for the bulk materials, because the
SFG-VS signals only come from the surface or interfaces. In the past two decades, SFG-VS has
been extensively used to probe molecular surfaces and interfaces 204-206. However, in recent
years, it was also found useful in selective probe or detection of crystalline materials such as
energetic and explosive materials as well as crystalline cellulose 207-209.
The signal of SFG is proportional to the intensity of the visible and IR laser intensity
(𝐼(𝜔𝑉𝐼𝑆) and 𝐼(𝜔𝐼𝑅) ) as well as the square of the effective susceptibility ( 𝜒𝑒𝑓𝑓(2) ) of the
54
material. That is 203, 206, 210,
(3.1)
, which contains all the structural and molecular nonlinear optical information of the
material, is proportional to the number density of the molecules (N) and is a linear function of
the SFG susceptibility tensor that is a structural and ensemble orientational average
( ) of the molecular polarizability tensor , that has a frequency dependence on the
intrinsic vibrational frequency of the molecules in the material 206, 210, 211. One has,
(3.2)
(3.3)
(3.4)
Here is the molecular polarizability tensor that consists a non-resonant contribution
and is frequency dependent on the vibrational frequency (with a homogeneous width
and an inhomogeneous width 2.35 ) of the qth vibrational modes of the molecule 211.
Here (i,j,k) and (i’,j’,k’) represent the axis of the (x,y,z) or (x,y,z) coordinates system for the
laboratory or the molecular frame, respectively. The mode specific amplitude factor is
proportional to the molecular Raman tensor and IR transition dipole vector , where Q
is the normal mode coordinates of the qth mode. Therefore, the SFG response can only come
2
2
SFG eff VIS IRI I I
2
eff
ijk
' ' 'i j k ' ' 'i j k
' ' '
' ' '
ijk s i j k
i j k
N
2
22, ' ' '
' ' ' , ' ' '
IR
qq i j k
i j k NR i j k
q IR q q
ei
' ' ', ' ' '
0
1
2
i j kq i j k
q q qQ Q
' ' 'i j k
, ' ' 'NR i j k q
2 q q
, ' ' 'q i j k
' 'i j
'k
55
from the mode that is both Raman and IR active. All these equations not only provide the way to
understand the molecular origin of the SFG spectra, but also provide the frame work to compute
or calculate the SFG response from the individual molecular property (through calculate )
and molecular unit structure and ensemble average (through calculate ). The values of the
and tensors are the result of the microscopic and macroscopic structure and symmetry
properties, thus they can be used to determine the structure and interactions of the molecular and
overall structure of the material.
Since the number of molecules contributing to the SFG-VS signal from the surface or
interface is much smaller than that of crystalline materials, usually the SFG-VS signal from
crystalline materials is at least one order or a few orders of magnitude stronger than that from the
surface. In addition, since the amorphous material generally has no SFG response, SFG-VS is
capable of selectively probe crystalline materials imbedded in the amorphous materials 85, 196.
This is the clear advantage of SFG-VS as a selective probe of crystalline cellulose than IR or
Raman spectroscopy that probes the crystalline and amorphous material indiscriminately. There
is also different polarization dependence between the SFG-VS signal from the surface and the
crystalline materials.
As the symmetry of crystalline material and surface is different, ensemble average of the
SFG responses from the crystalline material and the surface have very different polarization
dependence. Therefore, these properties can be used to distinguish and determine whether the
SFG signal is from the crystalline material or surface when the origin of the SFG signal cannot
be easily distinguished. In SFG-VS, the polarization angle corresponding to the zero (null) SFG
intensity is called the polarization null angle (PNA). In general, the SFG from crystalline
, ' ' 'q i j k
' ' 'i j k
' ' 'i j k ijk
56
material does not have a null angle, while the SFG signal from a surface usually has a clear null
angle (Velarde and Wang 2013b). Therefore, PNA measurement can be used to determine
whether the SFG signal is from the crystalline material or from the surface.
Kim and co-workers using a commercial picosecond scanning SFG-VS system with spectral
resolution about 6 cm-1 successfully demonstrated strong capacity of SFG in detecting crystalline
structure of cellulose 85, 171, 196. Also, density functional theory with dispersion corrections (DFT-
D2) was performed to simulate SFG vibration frequencies of laterally aligned cellulose Iα and Iβ
crystals 212. These works reported on the CH stretching vibrational modes even though the OH
stretching vibrational spectra with the resolution of SFG-VS at 6 cm-1 85, 171, 198, 212.
Recently, our team developed a high-resolution broadband sum frequency generation
vibrational spectrometer (HR-BB-SFG-VS) with sub-wavenumber spectral resolution that can
provide an accurate spectral lineshape and more spectral details in SFG-VS studies with
resolution of 0.6 cm-1 211, 213-215. This development has provided new opportunities in improved
measurement and better understanding of the SFG spectral and the surface/interface structure and
interactions. In this chapter, four different types of cellulose were probed with HR-BB-SFG-VS
in both the CH and OH stretching vibration frequency regions. We chose the samples with
known and well-characterized crystalline structures, such as Avicel, cellulose Iα and cellulose Iβs
from different sources, in order to establish explicit spectral signatures to differentiate the
cellulose Iα and the cellulose Iβ crystalline cellulose forms. The determination of the structural
difference and similarities of two basic polymorphs- cellulose Iα and cellulose Iβ as well as
different sources of cellulose Iβs can help establish solid foundation for developing reliable
methodology and metrics to identify cellulose structural variance and quantify the extent of
57
changes during cellulose chemical treatments or enzymatic hydrolysis. We expect the new
information in HR-BB-SFG-VS spectra to improve in-depth interpretation of basic cellulose
crystal structures. In this chapter, characteristic peaks were determined. Spectral information
provided clues for investigating Avicel composition of cellulose Iα and cellulose Iβ. PNA
measurement was operated to verify the crystal structure signal sources. Results from the current
study will aid in the investigation of complexity and uncertainty of cellulose I polymorphs by
revealing new information on cellulose structures.
3.3 Materials and methods
3.3.1 Cellulose samples
Cellulose crystals (cellulose crystals Valonia and cellulose crystals tunicate) were prepared
according to reported methods 78, 79, 212. Cellulose crystals Valonia (CCV) was isolated from alga
Valonia ventricosa (Glaucocystis (nostochinearum)) cell walls 216. Ethanol extraction, alkali and
acid treatment were carried out in sequence to produce cellulose Iα. 13C NMR analysis results
showed that cellulose Iα fraction was found to be as high as 90% 216. Two cellulose crystals
tunicate (CCT) samples from different sources were used in this chapter. One was produced from
tunicate (Halocynthiaroretzi) cellulose 217. The sample was initially treated with aqueous KOH
and sodium chloride for several times to make it colorless. Then strong sulfuric acid hydrolysis
was carried out at 40ºC for 8 hours with stirring afterwards. The suspension was further
processed to form cellulose microcrystals 186. Another source of cellulose Iβ was isolated from
red reef tunicate (Rhopalaea abdominalis) 218. The red reef tunicates (That Fish Place,
http://www.thatpetplace.com/pet/prod/235370/product.web) were slit open with a sharp knife
and their mantles were washed thoroughly with deionized water. Whole mantles were soaked
58
overnight in a 5% (w/v) aqueous KOH solution at room temperature. The mantles were then
rinsed with deionized water and bleached for 6 hours at 70°C with a bleaching solution,
exchanging the used bleaching solution with fresh solution every 2h. The bleaching solution
consisted of 300 mL of chlorite solution (containing 6g NaClO2) mixed with 5mL of glacial
acetic acid. After repeating the KOH/ bleaching treatment four times, the mantles became
completely white and were then washed thoroughly. Avicel PH-101 (11365) was purchased from
Fluka BioChemika (Ireland).
3.3.2 HR-BB-SFG-VS
The details of the experimental setup for sub-wavenumber HR-BB-SFG-VS and FID-SFG
intensity measurement were described previously 211, 215. Briefly, to make HR-BB-SFG-VS
possible, one needs to generate well synchronized sub- 100 fs IR pulses and ~ 100 ps 800 nm
pulses. The sub 100 fs IR pulses provide broad IR spectral coverage and the ~ 100 ps 800 nm
pulse provides the sub-wavenumber spectral resolution. This is achieved by electronically
synchronizing the two sets of Ti: Sapphire oscillators/amplifiers at 1 kHz repetition rate running
at 40 fs and ~ 100 ps pulses at the fundamental (800 nm), respectively. Such system was
constructed with lasers from Coherent, Inc. with synchronized 800 nm pulses with > 3.5 W
power at ~ 90 ps and > 7.5 W at ~ 40 fs, respectively. The overall timing jitter between the laser
pulses from the two amplifiers is estimated to be < 250 fs. In the HR-BB-SFG-VS experiment,
about half of the fundamental output generated by the ps amplifier (~ 90 ps as measured by
cross-correlation) was further attenuated by a variable neutral density filter to ~ 60 µJ/ pulse and
used as the source of VIS radiation. The incident angle of the VIS beam (800 nm) is βVIS=45º and
βIR =56º with respect to the surface normal. The SFG signal was collimated by an achromatic
59
lens and isolated from the VIS and IR beams by the combination of an optical iris, an 800 nm
notch filter, and a 785 nm shortpass filter (Semrock Inc.). The SFG signal was polarization
selected and imaged onto the spectrometer slit (typically open to < 100 µm) of a 750 mm
spectrograph (Andor Shamrock). The SFG signal was dispersed by a 1200 lines/mm grating and
recorded by a thermoelectrically cooled (-80°C) Electron-Multiplied CCD camera (Andor
Newton 971P, back-illuminated) containing a 1600x400, 16 µm2, pixel array. In the HR-BB-
SFG-VS measurement, usually a two-minute exposure for the ssp polarization combination was
used. Here the ssp polarization combination is defined as with s-, s-, p- polarization
corresponding to the polarization of the optical fields of the SFG signal, visible and IR beams,
respectively 213. p polarization is defined when electric field vector parallel to the plane of
incidence within the incident plane formed by the incident beam direction and the surface
normal, and s polarization is perpendicular to the incident plane. Background noise spectrum was
measured by delaying the ps VIS and fs IR pulses by > 1 ns. The spectra were normalized with
the SFG profile from the top surface of a thick Z-cut α-quartz crystal. In comparison to the
spectral resolution of the conventional BB-SFG-VS (~15 cm-1) and the scanning SFG-VS (~ 6
cm-1) in the literature, HR-BB-SFG-VS in this chapter has a spectral resolution of ~0.6 cm-1. As
demonstrated recently, such high order of magnitude improvement of the spectral resolution not
only enables resolving of closely overlapping peaks 215, but also provides accurate and intrinsic
spectral lineshapes that is crucial for the quantitative spectral and coherent vibrational dynamics
analysis of the molecular vibration- at surfaces/interfaces and in non-centrosymmetric crystalline
material 211, 213, 215. The chapter reported here is the first application of HR-BB-SFG-VS to study
crystalline materials, such as crystalline cellulose.
60
3.3.3 Polarization Null angle measurement with picosecond scanning SFG-VS
Polarization dependent scanning SFG-VS measurement was also performed with the
picosecond scanning SFG-VS system (EKSPLA, Inc.), pumped by an Nd:YAG laser with a 50
Hz repetition rate and ~30 ps pulse width. This system is similar to the system used in previous
cellulose studies 85. The resolution of the system was ~6 cm-1. The picosecond SFG was carried
out in the scanning mode. IR (energy 180µJ) and visible (energy 100µJ) beams overlapped on
the surface of the sample with IR incident angle of 56° and visible incident angle of 45°. The
SFG signal was detected by a monochromator and a photomultiplier controlled by a PC
computer.
In the polarization null angle measurement, the SFG signal polarization was kept at -45°
with respect to the p-polarization direction and IR polarization was fixed at 0° (p-polarized). IR
wavelength was kept at ~2956 cm-1. The polarization angle of the visible beam was changed with
a rotating half-wave plate from -45○ to 360○ 213. As shown below, there was no apparent
polarization dependence for the SFG-VS signal from all cellulose samples, indicating that the
SFG-VS was indeed from the bulk of the crystalline material instead of from the surface of these
materials.
3.3.4 Atomic force microscopy (AFM)
AFM was performed in air using the method reported previously 219. A multimode scanning
probe microscope (SPM) with NanoScope V controller (Veeco, Santa Barbara, CA. USA) was
utilized for all AFM measurements. To ensure stability, the AFM was situated in a specially
designed laboratory with acoustic and vibration isolation. A customized Nikon optical
microscope with deep focus (maximum 800 x magnification) was used to aid the positioning of
61
the AFM tip to the desired face and location on the cell wall. A standard 15-μm scanner was used
with the ScanAsyst™ imaging mode and probes SCANASYST-AIR was used for imaging in air,
and SCANASYST-FLUID+ for imaging under fluid (Bruker, Camarillo, CA USA). The software
Nanoscope Analysis v1.4 was used for AFM operation and later image processing. The AFM
imaging recorded many parameters, while all images presented in this study were height images,
which were flattened at 3rd order and filtered with low pass filter.
3.4 Results and discussion
3.4.1 Traditional Raman in cellulose characterization
Raman spectra were taken in NREL. Raman spectrometer was equipped with a video camera,
CW Nd: YAG diode laser, LabRamHR UV spectrometer, and CCD detector. Samples were
gently placed on the glass slides and focused with the video camera. The laser excitation from
diode laser was 2-20mW of 514.0 nm. The confocal hole was kept at 100 µm. Raman spectra
were analyzed using GRAMS/32or LabSpec data acquisition and analysis software 220.
Raman spectra of the (Figure 3.1a) cellulose Iα from alga Valonia ventricosa (Glaucocystis
(nostochinearum)) and (Figure 3.1b) cellulose Iβ from Halocynthiaroretzi tunicate. It is clear that
the Raman spectral features of the two crystalline cellulose forms are very similar to each other,
and do not show significant difference. This suggests the advantage of using HR-BB-SFG-VS as
the spectroscopic tool for cellulose structure characterization.
62
Figure 3.1 (a) Raman spectra of cellulose Iα from alga Valonia ventricosa (Glaucocystis
(nostochinearum)) and (b) cellulose Iβ from Halocynthiaroretzi tunicate in the frequency regions
of 300 to 1600 cm-1 and 2500 to 3700 cm-1
3.4.2 Polarization Null Angle (PNA) testing
PNA method in SFG-VS is a molecule orientation analysis to determine the relative phase
and amplitude ratio between the ssp and ppp polarization combination of SFG response 213.
Figure 3.2 shows the SFG polarization null angle measurements of cellulose Iα, cellulose Iβ, and
Avicel. The SFG intensity at 2956 cm-1 showed consistent flat response under different visible
beam polarization angles in Avicel, cellulose Iα and cellulose Iβ. The lack of null signal at all the
polarization angles of all the samples indicated that the SFG signals of Avicel, cellulose Iα and
63
cellulose Iβ materials were polarization-independent, e.g. not sensitive to the polarizations of the
incident laser beams. Such results suggested that the SFG signals detected from these samples
indeed originated from the bulk crystalline structure other than from the amorphous domain
surface 213. Another indication that these signals were from crystalline bulk is that the SFG signal
was more than two orders of magnitude stronger than the surface SFG signal with full monolayer
coverage, such as the air/DMSO interface. Since the SFG signal is not sensitive to polarization
angles and with bulk crystalline origin, polarization dependent measurement usually conducted
on molecular surfaces is not necessary, and the SFG measurements in this chapter were carried
out in the same ssp polarization consistently throughout this work.
Figure 3.2 PNA for different visible polarization angles of cellulose Iα from alga Valonia
ventricosa (Glaucocystis (nostochinearum)), cellulose Iβ from Halocynthiaroretzi tunicate, and
Avicel using scanning SFG system at 2956 cm-1
64
3.4.3 HR-BB-SFG-VS spectra of cellulose Iα, Iβ and Avicel samples
SFG-VS was applied to study cellulose microfibril assembling 212 and can serve as an
important tool in cellulose structure characterization. Uniaxial-aligned cellulose Iα and Iβ were
studied by resolution of 6 cm-1 212. It is critical to well understand two natural crystals Iα and Iβ
from different sources using high resolution SFG-VS with resolution of 0.6 cm-1. HR-BB-SFG-
VS was applied in the chapter to probe Avicel, cellulose Iβ from Halocynthiaroretzi tunicate,
cellulose Iβ from red tunicate, and cellulose Iα from alga Valonia ventricosa (Glaucocystis
(nostochinearum) in the frequency range of 2700 cm-1 to 3450 cm-1. The entire HR-BB-SFG-VS
spectra gave a comprehensive SFG profile of cellulose from different sources (Figure 3.3a).
Since SFG spectral features require a vibrational mode to be both IR and Raman active, not all
peaks seen in IR and Raman spectra were necessarily observed in SFG-VS 212. Spectral
differences in CH and OH stretch regions are evident, including peak positions, intensities, and
spectral shapes among different cellulose samples. These significant spectral differences enable
the identification of different types of cellulose as discussed below. Spectral fittings were
performed to reveal the accurate real peak positions, amplitude, and bandwidth (Fitting results
are listed in Table 3.1). Theoretically, SFG peaks could be fitted into Lorentz models convoluted
with Gaussian intensity distribution (Equation 3.3) 206, 211, which determines amplitude and other
peak shape information 221. However, in most cases Lorentzian lineshape is used in SFG-VS
fitting. Figure 3.3(b) and 3.3(c) display the fitting results of both CH and OH stretch regions of
Iβ, Avicel and Iα using commonly used Lorentzian lineshape function, respectively. Figure 3.3
showed significant difference as the Raman data (Figure 3.1) in the CH and OH stretching
frequency region.
66
Figure 3.3 (a) Experimental HR-BB-SFG-VS spectra of Avicel, cellulose Iβ from
Halocynthiaroretzi tunicate, cellulose Iβ from red tunicate, and cellulose Iα from alga Valonia
ventricosa (Glaucocystis (nostochinearum)) at wavelengths of 2700 cm-1 to 3450 cm-1. All
spectra intensities were normalized and presented on the same intensity scale. (b) Spectra for two
cellulose Iβs after peak fitting via Lorentz profile convoluted with a Gaussian intensity
distribution method; (c) Spectra for cellulose Iα (left axis) and Avicel (right axis)
Previous SFG-VS studies characterizing the polymorphism of crystalline cellulose mostly
focused on the CH stretching vibration frequency region. In the literature, it was shown that the
spectral features in the surface SFG-VS did not simply follow the IR and Raman spectral
assignments, while the assignments in the IR and Raman studies in these spectral regions were
not always correct, and the SFG-VS spectral assignment can be achieved with a set of
polarization selection rules 206, 222, 223. However, the SFG-VS peak assignments are challenging
for crystalline materials, because the polarization selection rules 206 developed for the surface
SFG-VS cannot be directly applied to the SFG-VS spectra from crystalline materials, as the SFG
67
signal from randomly oriented crystalline domains lacks polarization dependence. In previous
study, based on IR or Raman literatures, cellulose peak at 2945 cm-1 was assigned as CH2
asymmetric vibration of exocyclic 6CH2OH and peaks in the range of 3100-3500 cm-1 were
assigned to OH vibrational region 85. Therefore, some of these assignments may not be correct,
and the proper assignment of the cellulose SFG-VS spectral peaks is an issue that warrants
further studies.
Apparently, Figure 3.3 indicates more peaks are detected by HR-BB-SFG-VS spectra
compared to previously reported in previous studies 199, such as the 2913, 2930, 3007 and 3377
cm-1. This phenomenon is attributed to the higher SFG resolution in HR-BB-SFG-VS
measurement which allows detection of intrinsic spectral lineshape without instrument
broadening. As the result, more detailed vibrational stretch peaks can be identified. Curve fitting
results (See Table 3.1(a, b, c, d) also indicated that some fitted peaks are not shown obviously in
SFG spectrum. These data warrant future detailed studies in understanding the detailed
crystalline structure and interaction. In SFG-VS fitting, usually the Gaussian width is neglected.
One can see that not all the peaks in the fitting have the same certainty, and it can be seen from
the error bar of the fitting parameters. Usually the small peaks would have significantly larger
error bar. It needs to be very careful when discussing such detailed spectral features. We also
know that without the high-resolution spectra, one cannot identify as many peaks and the fitting
would be even less reliable. Nevertheless, one has to realize that so far the HR-BB-SFG-VS
represents the best effort one can have with the SFG-VS spectral measurement and analysis.
Table 3.1 Peak position and relative peak intensity parameters from curve fitting
using Lorentzian lineshape profiles (as in Equation 3.3 in the main text) of (a) Avicel, (b)
68
cellulose Iα from alga Valonia ventricosa (Glaucocystis (nostochinearum)), and (c) cellulose Iβs
from red reef tunicate and (d) Halocynthiaroretzi tunicate within wavelength of 2700 to 3050 cm-
1 and 3200 cm-1 to 3450 cm-1
Table 3.1a
Avicel
q /(cm-1) q /(a.u) q /(cm-1)
2846.4±1.5 2.1±0.2 32.49±2.6
2892.7±0.9 0.6±0.1 12.81±1.9
2945.6±0.2 1.9±0.3 9.17±0.5
2956.3±0.5 3.2±0.3 13.37±0.4
3200 cm-1 to 3450 cm-1
3332.3±0.5 7.7±0.3 33.67±0.9
3377.9±0.6 0.8±0.1 13.61±1.9
3432.4±2.6 5.9±1.1 47.76±6.9
Table 3.1b
Cellulose Iα from alga Valonia ventricosa
q /(cm-1) q /(a.u) q /(cm-1)
2830.8 ± 5.8 0.4±0.9 10.9±8.8
2849.2 ± 1.4 0.07±0.06 4.4±3.2
2884.6±0.8 0.08 ± 0.1 3.8±1.7
2902.3±0.5 0.12±0.1 4.2±1.5
2925.4±0.3 0.7±0.08 8.7±1.0
2947.9±0.1 3.2±0.06 6.9 ± 1.2
2972.3±0.1 1.4±0.05 8.1±0.2
3048.4±1.5 3.0±0.3 40.4±3.7
3232.8±0.5 1.0±0.1 15.4±1.5
3248±0.6 -0.35±0.1 10.0±1.0
3290±0.5 0.11±0.1 4.2±1.0
3304.2±0.9 0.20±0.1 8.7±1.8
3332.8±0.3 -1.5±0.3 16.3±0.6
3395.5±0.5 -2.5±0.1 26.3±1.3
Table 3.1c
Cellulose Iβ from red reef tunicate
q /(cm-1) q /(a.u) q /(cm-1)
2863.9 ± 1.3 4.7 ± 0.5 26.4 ± 2.0
2912.1 ± 0.5 -0.04 ± 0.1 6.3 ± 0.8
2929.8 ± 0.2 0.8 ± 0.1 5.6 ± 0.6
2949.3 ± 0.1 10.2± 0.7 6.0 ± 0.1
2973.3 ± 0.4 2.5 ± 0.1 15.5± 0.4
2990.3 ± 0.4 0.8 ± 0.2 6.9 ± 1.2
69
3200 cm-1 to 3450 cm-1
3280.8 ±1.6 10.6±0.5 18.8±1.4
3320.5±9.1 76.9±1.1 3.6±0.7
3368.9±6.6 4.4±0.7 24.8±0.8
Table 3.1d
Cellulose Iβ from Halocynthiaroretzi tunicate
q /(cm-1) q /(a.u) q /(cm-1)
2863.4±0.4 0.4±0.0 34.1± 6.8
2916.3 ± 0.3 -0.6 ± 0.1 3.1± 0.6
2920.8±0.3 0.5±0.02 2.9±0.8
2945.2 ± 8.9 8.5±0.3 7.9±0.7
2975.7±0.1 16.2 ± 3.7 7.8±0.1
2991.5±0.2 0.7±0.1 4.0±0.5
3007.7 ± 1.0 3.6 ± 1.8 19.8 ± 4.6
3276.8 ±1.5 10.1±1.8 19.5±1.7
3321.7±17.3 79.0±1.5 3.6±1.5
3369.4±6.6 6.6±0.4 23.6±1.7
3.4.4 Morphology of cellulose Iα and cellulose Iβs from AFM image
Atomic force microscopy (AFM) can obtain high-resolution images to reveal lateral packing
of the chains. AFM images of four cellulose samples were taken to provide detailed
morphological structure and arrangement of the microfibrils measured with HR-BB-SFG-VS.
Figure 3.4 compares AFM images of (a) Avicel, (b) cellulose Iα, and (c, d) two cellulose Iβs
taken in the same nano scale resolution. AFM images show similar morphological structures of
cellulose microfibrils from Valonia and tunicates (Figure 3.4 b-d), which are large crystals (20-40
nm in dimensions) that differ from Avicel with small microfibrils (3-5 nm). Cellulose sample
from red reef tunicate (C) appears to contain some amorphous structures. Also, Figure 3.4
presents different microfibril lateral packing information of Avicel, cellulose Iα and cellulose Iβs,
which showed randomly oriented cellulose crystals, different densities of microfibrils, and
microfibrils packing forms. The two cellulose Iβ samples showed similar microfibrils stacking
but different in microfibrils density. All those phenomena reveal the complexity of cellulose
microfibrils arrangements including fibrils overlapping, twisting and packing forms among four
70
samples. The packing of microfibrils provides information on the nature of the enzyme
complexes responsible for oriented cellulose synthesis during plants growth 177, 224, cellulose
species and cellulose extraction methods.
Figure 3.4 Comparison of AFM images of (a) Avicel, (b) cellulose Iα from alga Valonia
ventricosa (Glaucocystis (nostochinearum), (c) cellulose Iβ from red tunicate and (d) cellulose Iβ
from Halocynthiaroretzi tunicate. The scale of all images is 2 x 2µm
The differences in the microfibrils packing of the four celluloses are also reflected in the
HR-BB-SFG-VS spectra. As shown in Figure 3.3a, the SFG intensity of the cellulose Iβ from
Halocynthiaroretzi tunicate sample has the highest SFG intensity, followed by the cellulose Iβ
71
from red tunicate, the cellulose Iα from alga Valonia ventricosa, and Avicel. Since the SFG signal
should be roughly proportional to the density of the same microfibrils in the unit volume, it is
easy to understand why the cellulose Iβ from Halocynthiaroretzi tunicate has highest SFG
intensity and why the cellulose Iβ from red tunicate and the cellulose Iα from alga Valonia
ventricosa have smaller SFG intensity as their microfibril densities in the AFM images in Figure
3.4 followed the same decreasing order, i.e. from Figure 3.4d, Figure 3.4c, to Figure 3.4b. It is
interesting to see that even though the Avicel sample appears to have high density from its AFM
image (Figure 3.4a), its SFG intensity is the smallest and comparable to the cellulose Iα from
alga Valonia ventricosa. This fact can only suggest that certain amount of the materials in the
Avicel did not contribute to the SFG intensity. In another word, a certain part of the material in
the Avicel sample is most likely not in the crystalline cellulose form, or is in an amorphous form
that is SFG insensitive. These results suggested that SFG intensity and cellulose material density
correlation might be used to classify the SFG-sensitive crystalline and SFG-insensitive
amorphous cellulose materials, and it might also be used to quantify the amount of SFG-sensitive
crystalline material in the cellulose material.
3.4.5 Unique O-H signatures for cellulose Iα and cellulose Iβ
The unique spectral signatures for the different forms of crystalline cellulose materials were
shown in Figure 3.5. Results showed that the O-H spectra of two Iβ samples were almost
identical in their spectral lineshape even though they had significantly different spectral
intensities. In Figure 3.6, the Iα and Iβ samples clearly showed significantly different spectral
lineshapes. Figure 3.7 showed that the C-H spectra of the two Iβ samples had quite different
spectral lineshape even though the frequency positions of most of their C-H peaks were the
72
same. These results clearly indicated that the O-H spectra features were the unique signatures
that could be used to differentiate the Iα and Iβ forms of crystalline cellulose.
Figure 3.5 HR-BB-SFG-VS spectra of cellulose Iβs from Halocynthiaroretzi tunicate and red
reef tunicate between wavenumber of 3100 cm-1 to 3500 cm-1
Previous literature reported that the characteristic peaks of cellulose Iα in IR spectra located
at 3240 cm-1 79 while cellulose Iβ in IR spectra showed characteristic peak at 3270 cm-1 78. In
Figure 3.5, two cellulose Iβs from different sources presented almost identical peak positions and
shapes. It indicated characteristic peaks of cellulose Iβs locating in wavenumber range from 3100
to 3500 cm-1. From peak fitting results, the two cellulose Iβs showed peak positions at 3277,
3321, and 3369 cm-1.
73
Figure 3.6 HR-BB-SFG-VS spectra of cellulose Iβ from Halocynthiaroretzi tunicate and
cellulose Iα from alga Valonia ventricosa (Glaucocystis (nostochinearum)) between wavenumber
of 3200 cm-1 to 3450 cm-1
Figure 3.6 showed different spectral signatures of the cellulose Iα and cellulose Iβ in terms
of both the peak positions and lineshapes. For example, peaks of cellulose Iα were at 3232, 3248,
3290 3304, 3332, and 3395 cm-1, while peaks of cellulose Iβ presented at 3277, 3321, and 3369
cm-1. It appeared that 3232, 3248 and 3440 cm-1 peaks were unique to the cellulose Iα form,
while the main 3322 cm-1 peak of Iβ overlapped mostly with the 3332 cm-1 peak of the Iα form
although it is uncertain that how much Iβ or Iα form was present or dominant. Nevertheless,
these spectral signature differences of the OH vibrational peaks allowed clear identification of
cellulose Iβ and cellulose Iα and also indicated basic structural differences of cellulose Iβ and
cellulose Iα in terms of the OH groups structure and interactions in their unit cells. It was
reported that there were different OH bonds stretch in cellulose Iβ and cellulose Iα allomorphs,
74
which determined the different packing of cellulose chains 201, 225. The high-resolution SFG
spectral differences of the cellulose Iβ and cellulose Iα samples in this chapter also revealed
different cellulose microfibrils stacking and arrangements. The detailed connection and
quantitative correlation of these SFG spectral signatures to the specific crystalline structures of
the different forms of crystalline forms warrant further studies in the future.
3.4.6 CH Spectral features for cellulose Iβ samples from different sources
Figure 3.7 Difference of cellulose Iβs from red reef tunicate and Halocynthiaroretzi tunicate by
HR-BB-SFG-VS
SFG-VS with resolution of 0.6 cm-1 was applied to probe laterally packed and aligned films
of cellulose Iα nanowhiskers and cellulose Iβ Nanowhiskers which showed obvious peaks at
2850, 2944 cm-1, and slightly visible peaks at 2886, 2920, and 2988 cm-1. However, the HR-BB-
SFG-VS spectra results showed that even though the two cellulose Iβs had almost identical O-H
spectral features, their C-H stretching regions were significantly different (Figure 3.7). Dramatic
75
differences of the relative spectral intensities for the two Iβ forms could be identified. In order to
compare the relative intensity of different spectral features, the intensities of all peaks in the
cellulose Iβs were normalized by the intensity of peak at 2973 cm-1 that was the highest peak in
the CH region. Detailed peak positions, peak amplitudes, and peak widths are shown in Table
3.1. Cellulose Iβs from red reef tunicate and Halocynthiaroretzi tunicate showed similar peak
positions at 2863, 2913, 2925, 2946, 2973, and 2991. Several relative peak intensity of the two
cellulose Iβs was prominently similar, but some of them were quite different. For example, the
2946 cm-1/2973 cm-1 amplitude ratio of cellulose Iβ from Halocynthiaroretzi tunicate was 0.52
while ratio of cellulose Iβ from red reef tunicate was 4.08. Such differences indicated that the C-
H groups in the two cellulose Iβ samples were different. Such differences might be attributed to
different microfibrils stacking. However, the detailed structural differences need to be further
examined in future works.
The two cellulose Iβs were obtained from different species of tunicates and through different
cellulose isolation and purification methods, thus resulting in different modification on cellulose
microfibrils orientation. Meanwhile, during tunicate growth, microfibrils are also stacked
differently. It was demonstrated that the process of cell expansion and/or elongation that
occurred during plant development influenced the arrangement of the cellulose elementary fibril
in different plant cell wall 224. Therefore, microfibrils are responsible for oriented cellulose
synthesis and directional cell expansion during plants growth 177. Therefore, CH stretching
vibrational region might provide critical information in identification of cellulose Iβs from
different sources. However, such stacking differences should not alter the OH interaction in the
unit cell of the cellulose Iβ. Otherwise, the OH spectra of the two forms of Iβ samples should
76
also be quite different in the peak position or relative intensity.
It is clear that different microfibril stacking structures as well as glucose unit cell structure
can result in different vibrational environment for different CH groups in the cellulose. It is well
known that cellulose microfibrils aggregation significantly influences cellulose crystalline
structures. Microfibril is a cellulose morphological unit which contains a single cellulose
elementary fibril 224. In the primary plant cell wall, the cellulose elementary fibril is the nascent
fibril formed by cellulose synthase rosettes. It was reported that there are six C-H stretching
modes in glucose ring that is detectable with vibrational circular dichroism (VCD) in the
wavelength range of 2800 to 3000 cm-1 226, 227. Therefore, it is not surprising to identify the same
or even more number of peaks in the HR-BB-SFG-VS spectra of cellulose samples even though
the detailed assignment and understanding of their relative intensities can be a challenging task.
SFG signals would be canceled under the symmetry inversion, which leads to weaker or stronger
peak intensity. Carbons in glucose ring have different molecular vibration environment that can
result in different peak shape in SFG spectra. Even with high resolution SFG and more peaks are
able to be detected, detailed peak assignments and understandings on the six C-H stretchings are
still not quite clear.
77
3.4.7 Characterization of Avicel probed by HR-BB-SFG-VS
Figure 3.8 HR-BB-SFG-VS spectra of Avicel and cellulose Iα from alga Valonia ventricosa
(Glaucocystis (nostochinearum)) (a) and cellulose Iβ from Halocynthiaroretzi tunicate (b),
respectively, between wavelengths of 3200 cm-1 to 3450 cm-1
78
Avicel is considered mainly composed of Iβ and a small portion of Iα 85, 196. Figure 3.8
shows the Avicel OH stretching region SFG spectrum together with the spectra of both the
cellulose Iα and the cellulose Iβ. The peaks parameters of the three samples after curve fitting are
listed in Table 3.1. These results showed that Avicel appeared mostly composed of cellulose Iβ
peaks and a small portion of cellulose Iα peaks although it presented some different peak
positions and peak shapes compared with pure cellulose Iβ and cellulose Iα. In the early section
we discussed that OH stretch peaks were characteristic spectral signatures to distinguish
cellulose Iα and cellulose Iβ. While Avicel simply had peaks at 3332, 3378, and 3432 cm-1,
cellulose Iα had peaks at 3232, 3248, 3290, 3304, 3332, and 3395 cm-1, and cellulose Iβ had
peaks at 3277, 3322, and 3369 cm-1. The CH vibrational region (from 2700 to 3050 cm-1) of
cellulose Iα and cellulose Iβs both showed a peak at ~2973 cm-1 (Figure 3.3a, 3.3b, and 3.3c).
Peak fitting results of C-H region showed that four peaks in the Avicel CH stretch region at
~2846, ~2895, ~2946 and ~2956 cm-1 which present a more detailed structure than the previous
report. Since there were no Avicel spectral peaks in the 3232 and 3248 cm-1 region, it is unlikely
that there is much Iα form present in Avicel. It is therefore reasonable to conclude that Avicel
consists of mostly the Iβ form in addition to certain amount of SFG-insensitive amorphous
material. Thus, it appears that, as advantages of high resolution SFG-VS, O-H stretch region is a
signature to identify Avicel, cellulose Iα and cellulose Iβ, while C-H vibrational peaks reflect the
complex microfibril packing and microfibril composed sheets packing in cellulose structure
without manual alignment 200, 228-230.
Although both peak intensity and integrated peak ratios have been used to estimate
crystalline cellulose and cellulose lateral packing orientation 194, 196, 228, 231. However, intensity
79
ratio might not be accurate enough since some peaks were overlapped thus resulting in miscount
issues. For example, ~2913 and ~2925 cm-1 of cellulose Iβ are presented shoulders against main
peak at 2946 cm-1. Instead, the ratio (R) of the total C-H stretching vibrational peak areas over
the total integrated O-H peak areas seems to be useful to evaluate spectra difference of Avicel,
cellulose Iα and cellulose Iβs. Results showed that that the R values differed among Avicel,
cellulose Iα and cellulose Iβs, while they are similar among two cellulose Iβs (Table 3.2). For
example, cellulose Iβ from Halocynthiaroretzi tunicate showed R of 0.54 while cellulose Iβ from
red reef tunicate showed R of 0.52 via R of cellulose Iα is 1.1. The R of Avicel (0.67) was more
similar to that of cellulose Iβ. It is agreeable to previous reports that Avicel is composed of most
Iβ form 196. It indicates that integrated area ratio might be useful in characterizing cellulose
structures.
Table 3.2 Comparison of C-H, O-H peaks areas and their ratios of Avicel, cellulose Iβ (different
sources) and cellulose Iα
Samples C-H peaks area O-H peak area R
cellulose Iβ from Halocynthiaroretzi tunicate 147.0 273.0 0.54
cellulose Iβ from red reef tunicate 36.6 69.3 0.52
cellulose Iα from alga Valonia ventricosa
(Glaucocystis (nostochinearum))
12.5 11.4 1.10
Avicel 7.0 10.5 0.67
3.5 Conclusion
This is the first HR-BB-SFG-VS study on cellulose Iα and cellulose Iβ, and Avicel in order
to identify the spectral signatures of different cellulose microfibrils. The findings suggested that
the O-H stretching region was unique spectral signatures which can be used to identify cellulose
Iα and cellulose Iβ, while the C-H spectral features do not exhibit such explicit correlation to
different types of crystalline structures of the cellulose. We also found that the ratios of the total
80
C-H stretching vibrational peak areas over the total integrated O-H peak areas of Avicel,
cellulose Iα and cellulose Iβs correlated well to different types of cellulose microfibrils. In
addition, many more peaks were revealed in the C-H stretching region by the HR-BB-SFG-VS
than by the conventional SFG-VS. For example, the peak at ~2945 cm-1 was assigned to the
C6H2 asymmetric stretching peaks, however, HR-BB-SFG-VS detected another obvious peak at
~2973 cm-1 nearby, which is usually not present in the conversional SFG-VS. Such new insights
can help us define crystalline cellulose structures in more detail and can provide confirmation in
measuring spectral lineshape, resolving different spectral peaks of the high-resolution approach.
With the detailed spectral measurement with the HR-BB-SFG-VS, more future work should be
expected to focus on peak assignment, and on illustrating the cellulose structure changes in the
cellulose treatment or transformation processes as their detailed SFG-VS spectral features
change. These developments shall make SFG-VS a useful in-situ spectroscopic tool not only in
cellulose structure and kinetics studies, but also in other crystalline biopolymer material studies.
The significant differences of cellulose Iα, Iβ, and Avicel in peak positions, intensities and peak
ratios indicates that the traditional understandings of these celluloses are not enough and
accurate. Avicel, cellulose Iα, and cellulose Iβ might be completely different from each other and
the statement of Avicel is composed of major Iβ might not be true. More future work is still
needed to provide evidences and it will surely be a breakthrough in cellulose structures.
3.6 Acknowledgements
This work was made possible through the support of the DARPA Young Faculty Award
contract # N66001-11-1-414. The author acknowledges the Bioproducts, Sciences and
Engineering Laboratory, Department of Biosystems Engineering at Washington State University
81
and the BioEnergy Science Center, a DOE Bioenergy Research Center, and the Genomic Science
Program (ER65258), both supported by the Office of Biological and Environmental Research in
the DOE Office of Science. Part of this work was conducted at the William R. Wiley
Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility located
at the Pacific Northwest National Laboratory (PNNL) and sponsored by the Department of
Energy’s Office of Biological and Environmental Research (BER). The theis author was partially
supported by the grant from the Chinese Scholarship Council (CSC). The author would like to
express special appreciation to Drs. Hongfei Wang and Bin Yang for their encouragements and
providing opportunities to do this research. The author of this thesis would like to thank Drs.
Hong-Fei Wang, Bin Yang, Zhou Lu, Luis Velarde, Li Fu, Yunqiao Pu, Shi-You Ding, Arthur J.
Ragauskas, and Dr. Seong Kim (Penn State University) for their contribution to this work and
publish it in the Cellulose journal.
82
CHAPTER FOUR
TOWARD DYNAMICAL UNDERSTANDING OF FUNDAMENTALS OF
PRETREATMENT AND ENZYMATIC HYDROLYSIS
OF CELLULOSIC BIOMASS VIA A SFG-VS
4.1 Abstract
In this chapter, the SFG-VS was applied to dynamically characterize the structural changes
of cellulose during biomass aqueous pretreatment and enzymatic hydrolysis. To mimic the
aqueous pretreatment system, a SFG-VS with a fluid reactor/cell system was developed to study
the effects of thermal heating and water media from aqueous pretreatment on the disruption of
cellulose crystals. To selectively characterize the molecular structures in surface layers of Avicel
and cellulose Iβ crystals, this work developed Total Internal Reflection Sum Frequency
Generation Vibrational Spectroscopy (TIR-SFG-VS) for the first time. Combined with the
conventional non TIR-SFG-VS, our study demonstrated the significant difference in molecular
orientations of surface layers and bulks of these two types of cellulose. The spectral difference of
cellulose surface layers and bulks suggested that the peaks at ~2948 cm-1 and ~2857 cm-1 were
from C-H vibrations on cellulose carbon rings and side chains -6CH2OH, respectively. Also, the
surface layers of Avicel and Iβ were much less crystallized than their bulks while Iβ surface
layers and bulks were more uniform than that in Avicel. The SFG signals in O-H stretching
region showed the different intermolecular, intramolecular, or free hydrogen bonding
distributions in surface layers and bulks of these two types of celluloses. This work demonstrated
the capacity of TIR-SFG-VS and conventional Non TIR-SFG-VS in selectively studying the
surface layers and bulk structures for cellulose crystalline materials. Therefore, the molecular
83
characterization of the surface layers of cellulose crystals is performed for the first time and an
important step to fundamental understanding the cellulose surface layers in biomass
pretreatment.
By the capacities of studying cellulose surface layers by SFG-VS, the molecular structural
changes of the surface layers and bulks of Avicel in aqueous pretreatment were studied under
dynamic heating program from room temperature to 280C for the first time. The high
temperature fluid cell was used to simulate the aqueous pretreatment up to 280°C with/without
pretreatment solvents (e.g. DI water). It was found that the bulk crystalline structures of dry and
wet Avicel were dramatically disrupted with the increasing of temperatures, and wet Avicel was
indicated faster. The possibilities of the gradual conversion of cellulose Iα and Iβ in Avicel to
disordered metastable intermediates with the increasing of high temperatures were proposed.
Significantly, a possible recrystallization process of disrupted Avicel bulks was hypothesized for
the first time to explaining the cellulose structural transformation in the process of and after
cooling down to room temperature. The thermal stability of surface layers was studied and the
recrystallization was found not occurred. Additionally, the surface layers needed 240-280°C to be
completely disrupted which was higher than the disruption of bulks required. Thus, the
temperature dependences of bulks and surface layers of cellulose as well as the recrystallization
of disrupted bulk crystals might be correlated with the requirements of elevated temperatures to
achieve high pretreatment effectiveness. These initial findings are important for not only in the
thermal design of pretreatment but also other thermal processing of cellulose. Based on our
results and analysis, a cellulose recrystallization hypothesis was proposed that heating-disrupted
cellulose bulk microfibrils are recrystallized during and after cooling down to room temperature
84
which is driven by the rearrangements of the disrupted cellulose microfibrils to a stable
crystalline state. Although cellulose presents highly ordered crystalline structure, the
recrystallization is not a complete recovery process. Thus, the recrystallization results in new
crystals which are differed from original crystals but show similarities. However, the surface
layers of cellulose were not recrystallized during and after cooling off due to their less
crystallized structures. The hypothesis still needs more experimental work to validate besides
this chapter indicated.
SFG-VS was also found useful in studying enzymatic hydrolysis system. Interestingly, with
the proceeding of enzymatic hydrolysis, Avicel molecular structure was changing during
enzymatic hydrolysis according to the changes of the peak area ratios. The C-H vibrational
region was indicated a crucial region to study the enzymatic hydrolysis. SFG-VS was also
proposed to study the interactions of enzymes and cellulose without peak overlaps and separating
enzymes and cellulose. Thus, SFG-VS demonstrated as an important tool in enzymatic
hydrolysis study but more work is still needed. These aspects of work will greatly contribute to
the fundamental understandings of aqueous pretreatment and enzymatic hydrolysis processes.
Keywords: Cellulose surface layers • Cellulose bulk crystals • Sum frequency generation •
vibrational spectroscopy • Total internal reflectance • High temperature • Cellulose
recrystallization • Enzymatic hydrolysis
4.2 Introduction
The interests of seeking a renewable energy are growing in recent years. Lignocellulosic
biomass serves as a sustainable source for biofuel production, which can be a promising option
for the renewable energy. The efficient utilization of cellulose is crucial in biorefinery
85
economics, which relies on the fundamental understandings of cellulose structural changes
during biorefinery processes. Cellulose is the major component in biomass plant cell wall. The
cellulose structural information is very important to cellulose accessibility to various treatments
such as enzymatic hydrolysis 232, chemical treatments 233, 234, and other industrial applications
(e.g. nanocellulose materials production) 235. Cellulose is the only crystalline structure source of
wood and believed to be thermally stable. It is commonly believed that cellulose microfibrils
form both crystalline and amorphous structure although their assembling is not clear 236-238.
Nowadays, the fundamental studies of cellulose structural changes during aqueous pretreatent
enzymatic hydrolysis are crucial to a successful biorefinery.
The cellulose behaviors in high temperatures have been studied. It was reported that
cellulose Iα and Iβ can be converted to a high temperature phase, and after cooling to room
temperature produces cellulose Iβ 239-243. This was explained by two proposed elucidations
including the relative free energy of cellulose Iα and Iβ predicted by molecular simulation which
cannot take into accounts of all influencing factors 244 and high temperature phase structure (I-
HT) intermediated pathways 245. However, experimental evidences to clarify these proposals are
still required. The two models of structural transformation mechanisms from cellulose Iα and Iβ
were a transformation/break-slip model by moving these chains up or down by one glucose
residue 246, and a rotation model with 180 degrees rotation along cellulose chain axis due to the
different crystal arrangements and hydrogen bonding of cellulose Iα and Iβ 240. The
computational molecular dynamics (MD) simulations with two force fields were applied to
elucidate the high temperature behaviors of cellulose Iβ suggesting the changes in layer spacing
and hydrogen bond patterns of I-HT.
86
Experimental work on cellulose thermal behaviors as a function of temperatures is not
widely reported. The spectroscopic tools have been used in the characterization of cellulose
thermal behaviors. For example, the infrared spectroscopy with a perturbation-correlation
moving-window two-dimensional correlation spectroscopy was used in studying the
temperature-dependent changes of hydrogen bonding in cellulose Iα and Iβ from 30-260C, and
proposed that the cellulose Iα and Iβ phase transformation was induced by the changes of
hydrogen bonding networks (e.g. 220C as hydrogen bond transition temperature) 247, 248. The
(CP/MAS) 13C NMR spectroscopy has been used to studying the structural transformation of
cellulose I at 230 to 280C for 30min suggesting that both cellulose Iα and Iβ were transformed
into a new crystal cellulose Iα similar to cellulose Iβ 243, 249. It was reported the thermal
expansion of cellulose crystal occurred during cellulose heating from room temperature to 250C
studied by XRD analysis, and found a high temperature cellulose phase at 220-230C 239.
However, the molecular structural transformation of cellulose Iα and Iβ during a dynamic heating
process, the molecular structure of I-HT, and the transformation of cellulose during cooling
process were not illustrated. More efforts are still needed to determine the molecular level details
of the high-temperature phase. Besides, the crystallinity changes of different wood cellulose was
reported with oven heating dry/wet cellulose at 220 ºC for 0-240min suggesting the different
crystallinity changes of heated wet/dry cellulose 250. Similarly, the effects of moisture in cellulose
were investigated on cellulose crystallinity changes under 180, 200, and 220 ºC, however, the
molecular changes were not well studied 251. Our SFG-VS system is able to characterize both the
surface layers and the interior transformations of cellulose Iα to Iβ, and study the molecular
structural transformations of cellulose I polymorphs during and after cooling process.
87
The enhancement of enzymatic hydrolysis efficiency can be achived by biotechnology and
improving cellulose substrates accessibility. The biotechnology is the long term goal while the
cellulose substrates study is a short term goal. The understanding of cellulose structure and its
interaction with enzymes is the key for the short term goal. In this chapter, the enzymes and pure
cellulose involving in the regular enzymatic hydrolysis system will be studied, and this was a
start of conducting the interactions of enzymes with cellulose surface layers and bulks.
The cellulose in pretreatment and enzymatic hydrolysis system mainly involves in surface
reactions. To assist in studying the cellulose surface layers, a new tool is urgently needed.
Recently, the research progress of cellulose bulk structure has been advanced with the
development of analytical methods. For exmple, the difference of Iα and Iβ was distinguished by
hydrogen bondings and cellulose chain arrangements, e.g monoclinic and triclinic, as well as the
difference in side chain -6CH2OH orientations 78, 79, 171, 252. However, the understanding of
cellulose surface layers (e.g. Iα and Iβ surface layers) at molecular level is limited. This
limitation might be due to the lack of quantitative in situ surface-selective tools for analyzing the
complex cellulose structures.
The crysalline surface/surface layers are still not well defined yet. Some literature mentions
cellulose surface studies but not clearly defined. For example, X-ray diffraction has been used to
simulate wide and narrow surfaces and surface hydrogen bonding of cellulose Iβ 253 and the
homogeneity of the cellulose crystalline surface with bulk 254-256. Also the 13C NMR was applied
to suggest a splitting of broad C4 and C6 signals of different celulose samples to be the
anhydroglucoses on the surface of cellulose elementary fibrils 257 or the surface molecules of
crystalline domains (indicating by the broad C-4 peak) 258, 259. Some studies assigned the peaks to
88
the mixture of Iα and Iβ forms 254. These controversia peak assignments were the basics of most
cellulose surface studies. It was reported that the celluose surface has a different conformation in
C6 with the bulk by solid state NMR and meawhile showed the reduction of intramolecular
hydrogen bonding on surface 260. Although the study of cellulose surface by solid state 13C NMR
was well recognized, the peak assignments were still doubtful 254. Atomic Force Microscopy
(AFM) was another tool to directly oberserve the morphology of cellulose. The surface
roughness and corrugations on the surface were found 254. Besides, the ultra-high resolution
AFM combining with computer modeling was applied to image the surface of native cellulose
and was proposed to study triclinic and monoclinic crystals on surface 256. Electron
microdiffraction study indicated the surface might only exist Iα form 261, while it required more
evidences 255. Additionally, a new imaging apporach of simulated Raman scattering (SRS)
microscopy was developed to image lignin and cellulose in vasular bundle with subcellular
resolution at 1600 and 1100cm-1 262. However, the AFM is limited in observing surface
morphological information and not applicable for molecular level study. Some studies even
demonstrated the distribution of glucan chains 260 or other β(1,4)-glycan polymers onto the
surface of cellulose263. However, these statements lack strong proofs. Unitil now, the cellulose
surface is not well understood, especially at molecular scale and how microfibrils on surface are
assembled. The interaction of cellulose crystalline surface with water or other solvents was
unknown as well, which impeded the research progress of aqueous pretreatment and enzymatic
hydrolysis.
The study of cellulose crystalline surface was greatly influcenced by characterization
methods. The traditional characterization methods of cellulose surface at the molecular level
89
were interefered by the bulk crystalline signals. The cellulose surface spectra were reported to
obtain from subtraction of crystalline spectra by cellulose whole spectra, which was not in-situ
study for cellulose surface. For example, the celery collenchyma cellulose was believed as
cellulose surface standard 264, consisting of 60% of surface chains which was twice amount of
surface chains found in textile celluloses like flax. Also, an another challenge of the study of
cellulose surface is that the majority of spectroscopic assignments were not clear yet. For
example, 87-91 ppm was assigned to crystalline structure C4 while 82-86 ppm (a broader peak)
was assigned to surface structure C4. However, 82-86ppm was also proposed from amorphous
cellulose instead of surface structure 265. Whether the amorhpous cellulose has similarities with
the cellulose surface is unknown.
In our study, the SFG-VS is a second-order nonlinear spectroscopy, which has been
exploited as a significant tool in monitoring surfaces or interfacial systems, such as solid–liquid,
liquid–liquid, solid–gas, and solid–solid interfaces 266, 267. The SFG-VS studies of cellulose
structures are popular recently 85, 171, 268 but all focusing on measuring the bulk crystalline
structures of cellulose. It was reported that SFG-VS can selective characterize cellulose
crystalline molecular structures without the influence of amorphous structures or polymers from
biomass 85. Furthermore, our previously developed high resolution broad band (HR-BB) SFG-
VS was first time applied to observe structural signatures in different cellulose sources and
polymorphs, which obtained more structural information than conventional SFG-VS settings 268.
The objectives of this chapter were firstly to design laser input angles and geometries of
SFG-VS to selectively characterize cellulose surface layers. We integrate total internal reflection
geometry in SFG-VS (TIR-SFG-VS) for the purpose of selectively measuring the cellulose
90
surface molecular structure and conformation, while suppresses the SFG contribution from the
crystalline bulk. This surface information is more from surface layers (length depends on the
geometry and laser wavelength) and may or maynot the same with previously reported surface
due to the undefined cellulose surface previously. In traditional characterization, the total internal
reflection (TIR) geometry in spectroscopic techniques was reported that can significantly
enhance the spectroscopic sensitivity on sample surface layers, thus, greatly increased signals
with several orders of magnitude 269. In TIR configuration, the evanescent wave from TIR beam
can be utilized to address the interface molecular structures 270-272. For example, the TIR
geometry has been widely used in IR and SFG investigation of various material surfaces or
interfaces, such as metal catalysts surface characterization, surface adsorption phenomena, and
nanoparticle surfaces or interfacial characterization 273-276. These studies concentrate on the
characterization of sample surface information of non cellulose crystalline materials because
these samples usually don’t have signals from bulk besides the surface. The goal of TIR setting
in our study is to use SFG-VS in the characterization of the surface layers of cellulose crystalline
material.
Additionally, the chapter investigated the dynamic thermal stabilities of Avicel surface layers
and bulk with/without water up to 280ºC which is a temperature limit used in biomass aqueous
pretreatment. At last, SFG-VS was used to study the fundamentals of enzymatic hydrolysis
system including molecular and crystal structural changes of cellulose with/without enzymes free
during enzymatic hydrolysis.
91
4.3 Materials and methods
4.3.1 SFG-VS set-up
The picosecond scanning SFG-VS system was a commercial EKSPLA SFG-VS
spectrometer utilized a 50 Hz and 29 ps Nd:YAG laser (EKSPLA PL2251A-50) which was
reported before 28. The visible beam at 532 nm though a KD*P crystal was produced by a portion
of the fundamental output (1064 nm). And the rest of the Nd:YAG output pumped an optical
parametric generation/amplification and difference frequency generation system (EKSPLA
PG401/DFG), generating an infrared (IR) beam tunable between 650 and 4300 cm−1. The
spectral resolution is ∼6 cm−1 277. The details of HR-BB-SFG-VS spectrometer system were
reported in previous work 268, 278. Briefly, HR-BB-SFG-VS can achieve sub-wavenumber
resolution to 0.6 cm−1, made possible by the precise synchronization of two sets of Ti:sapphire
oscillators/amplifiers that can achieve precisely timed 1 kHz pulse sequences of (nominally) 40
fs and 100 ps pulses at the fundamental (∼800 nm). The two mode locked oscillators are actively
synchronized in a master-slave configuration by a Synchrolock-AP (Coherent Inc.) system. The
master radio frequency (r.f.) pulse train originates from a ∼100 ps oscillator (TIGER, Time-
Bandwidth, Inc.) and its repetition rate (76 MHz) is used as the frequency reference for the
stabilization of a sub-40 fs oscillator (Micra-5, Coherent, Inc.). A convenient phase delay can be
applied within the Synchrolock-AP controls providing ∼12 ns of continuous relative time delay
between the slave and the master pulse trains while maintaining a fixed phase/frequency
relationship between them. This continuous delay adjustment is maintained in the amplified
pulses and becomes extremely useful when finding the temporal overlap between the two
amplified pulses and changing SFG experimental geometries 278.
92
4.3.2 Design of SFG-VS geometries to observe cellulose surface layers
In particular, the Avicel or cellulose Iβ particles are placed under a equilateral CaF2 triangle-
prism to achieve the total internal reflection for both the IR and Visible laser lights. The
evanescent waves at the surface can only penetrate the depth of λ/2 of the incident light, e.g.
roughly <250nm, and thus the TIR-SFG-VS only probes the very top surface layers of the
cellulose materials, without being overwhelmed by the SFG signal from the crystalline bulk.
This is completely different from previous TIR studies for the surface materials (non cellulose
crystalline materials) and SFG characterization for the bulk crystalline cellulose. Instead, our non
TIR and TIR-SFG-VS are built to characterize the surface and bulk cellulose separately for the
first time. The TIR-SFG-VS study of cellulose surface is investigated by examining both C-H
and O-H stretching regions of cellulose vibrations. The study is the first time to observe cellulose
surface molecular structure, which facilitates understandings of critical cellulose surface
molecular information that could not be achieved by conventional characterization methods.
Figure 4.1 a) TIR-SFG-VS design on cellulose surface observation and b) non TIR-SFG-VS on
93
cellulose bulk characterization. IR and VIS refers to infrared and visible 532nm beams,
respectively. c) total internal reflectance concept; d) simulation of SFG intensity with the change
of TIR and non TIR SFG-VS geometry, in which b) refers to non TIR SFG-VS while TIR
represents TIR-SFG-VS
Two SFG designs and their mechanisms were displayed in Figure 4.1a,b,c,d. To observe
cellulose surface layers, TIR-SFG-VS is equipped with a CaF2 material based prism (equilateral
triangular prism) to specifically monitor the cellulose surface information without bulk cellulose
interference (Figure 4.1a). To achieve TIR geometry, the incident angles should be above the
critical angle on prism surface which resulted in none transmission lights through the sample.
Meanwhile, evanescent wave was formed which allowed the characterization of the sample
surface layers up to λ/2 thickness (<250 nm in our study) 270-272. The thickness of the cellulose
samples being probed by TIR-SFG-VS can be approximately estimated by comparing the
intensity difference between bulk and surface spectra if the linear correspondence of sample
thickness with SFG-VS signal intensity is assumed regardless of detailed structural variance and
the increasing magnitude order of TIR on surface signals.
The direct contact of visible and IR beams was widely used and showed great advantages of
SFG-VS in detecting cellulose crystalline structure selectively 85, 171, 196, 268, 279 similar to Figure
4.1b where the bulk cellulose crystalline is prominent in the SFG signals. The bulk cellulose
crystalline structural information can be monitored due to both reflection and transmission lights.
Also, the SFG signal from the surface structure of cellulose is overwhelmed by the bulk signals,
and thus the structural difference between the surface and bulk cellulose remains largely unclear.
According to TIR-SFG-VS, the TIR set-up should enhance surface SFG signals to more than 100
94
times by SFG-VS intensity simulation (100 vs 0.8; Figure 4.1d). Detailed simulation method is
described in supplemental materials (Calculation of the Fresnel factor in the SFG measurement).
Therefore, if SFG-VS signals of samples from prism (Figure 4.1a,c) are below or more than 0-
100 times of the signals from flat window (Figure 4.1b), the surface and bulk are different.
4.3.3 The fluid heating cell system development
The fluid heating cell system was applied in the study of cellulose thermal behaviors and
interactions with solvents in aqueous pretreatment. Several other reactor systems were reported
to study cellulose heating process such as 50ml micro reactor using oven and oil bath as heating
source 251. In our study, we used the reactor designed with an accurate temperature monitoring
and short heating time (Figure 4.2). The cell was connected to three thermal heaters, a thermal
couple (connected to the little hole above three heaters) and a temperature controller (OMEGA).
Besides, the cell is connected to a nitrogen purging device and a pressure regulation system with
a back pressure regulator. The temperature applied to the cell is limited within ambient-300C
while the pressure performed should be ambient-500 psi. The flow rate limite for gas is 0-6
mL/min and liquid is 0.001-5 mL/min.
Figure 4.2 Cellulose heating high temperature fluid cell systems
The dry/wetted samples were filled up the cell sample holder under the prism or CaF2
95
window. SFG-VS system was well warmed up and sample signals were well normalized to the
same voltage (1000 PMT) otherwise indicated at the similar visible (120 µJ) and infrared (200
µJ) beam energy inputs. The incident angles of visible and infrared lights applied were 65° and
55° respectively otherwise notified. The sample covers were cylindrical CaF2 window (Red
Optronics; WIN-3104; Diameter: 25.4 mm; Thickness: 3.0mm) and CaF2 based prism (60°
triangular).
4.3.4 Cellulose sample preparation and heating program
The cellulose samples used in the study were Avicel and cellulose Iβ. Avicel (11365
SIGMA-ALDRICH Avicel PH-101) with ~50 μm particle size was purchased from Sigma
Aldrich. The cellulose Iβ was provided by National Renewable Energy Laboratory. It was
prepared according to the reported method (Tunicate (Halocynthiaroretzi) cellulose 217) and the
same sample used in our previous study 28. Avicel PH-101 (11365) was purchased from Fluka
BioChemika (Ireland). The wetted Avicel paste was prepared by adding 200ul DI water with
80mg Avicel powder. The dry or wetted samples were filled out in the cell sample holder and
covered with the prism or CaF2 window. SFG-VS system was well warmed up and sample
signals were well normalized to the same voltage (1000 PMT) at the similar visible (120 µJ) and
infrared (200 µJ) beam energy inputs. The incident angles of visible and infrared lights applied
were 65° and 55° respectively. All spectra were normalized to the same voltages and laser
intensities displayed in the same spectrum. The spectra were fitted by Lorentz profile convoluted
with a Gaussian intensity distribution method.
The thermal heating of Avicel is programed to higher temperatures continuously and
monitored by SFG-VS in the fixed time intervals at each temperature. The cellulose heating
96
process was programed with heating rate 1°C per 5s and stayed at 65, 100, 130, 175, 200, 220,
240, 260 and 280°C for 2min before taking SFG-VS spectra. The cool-down of heating cellulose
was performed under room temperature with the heating device off and temperature monitor on.
4.3.5 Molecular observation of cellulose structural changes in enzymatic hydrolysis using
SFG-VS
The restart samples used in our study were referring to the interrupted samples from
enzymatic hydrolysis followed by well cleaning and drying of cellulose to remove enzymes in
each sample 280. After withdrawing some amount clean cellulose sample for analysis, the left
well-washed samples were proceeding to enzymatic hydrolysis by adding in new enzymes and
others. The restart samples were used in the study to study the cellulose structural changes in
enzymatic hydrolysis. The other materials used in the uninterrupted batch hydrolysis were Avicel
PH101 and cellulase (Ctec 2 from Novozymes). The uninterrupted batch hydrolysis operation
was carried out at 2% w/v cellulose and enzyme loadings 60 FPU/g cellulose (65 mg protein
Ctec/g Avicel) with 50 mL of 50 mM acetate buffer (pH 4.8) in 50 °C incubator with a rotation
speed of 160 rpm. The pre-incubation of the slurries to the desired temperature was performed
for 15 min before adding enzymes. 10µl aliquots was withdrawn to a flat window after
enzymatic hydrolysis for 0-120 h. SFG-VS was used to probe each aliquot along with the
measurement of the Avicel conversion yields.
97
4.4 Results and discussion
4.4.1 Structural characterization of cellulose surface
Figure 4.3 SFG-VS spectra and peak fittings via Lorentz profile convoluted with a Gaussian
intensity distribution method of Avicel surface layers (red color) and bulk crystals (blue color)
within wavelength of (a) 2800 to 3000 cm-1 and (b) 3000 to 3750 cm-1; Dots represented
experimental data while lines represented curve fittings. All intensities were calibrated at the
same voltage (1000 v) for the SFG-VS measurements
Conventionally as discussed, TIR-SFG-VS was used to enhance the SFG signal (up to >100
times) in non-crystalline materials 269, 276. However, the signal of Avicel using TIR-SFG-VS was
98
not increased at all compared to that in non-TIR-SFG-VS, and in contrast, the signal was reduced
more than 10 times (Figure 4.3). The whole spectra comparison presenting a scope of Avicel
surface layers and bulks were shown in Figure S4.1. Interestingly, the spectra of Avicel surface
layers varied from different batches of Avicel since the surface was non-uniformly arranged.
Thus, the spectrum of surface layers above is a representative of many spectra of Avicel surface
layers which can present common peak characteristics of Avicel layers (other spectra shown in
Figure S4.2). Figure 4.3 showed our TIR-SFG-VS signals of Avicel were different from non-
TIR-SFG-VS signals. This was because the traditional TIR-SFG-VS in other non-crystalline
materials characterization only focus on the surface signal regardless of the bulk, and the purpose
of TIR setting was to enhance the surface signal of the non-crystalline materials. However, the
particle size (~50µm diameter) of Avicel (a crystalline material) in our study was much bigger in
diameter than the wavelength. Thus, TIR-SFG-VS signals of Avicel was prominent from Avicel
surface layers. If 100 times of intensity increasing of surface signals was assumed by TIR setting
like in conventionals non-crystalline material characterization, it was estimated that only ~50nm
thickness of the surface layers of Avicel were characterized in our study according to 10 times
intensity difference (Figure 4.3a,b). Beyond doubt, the TIR-SFG-VS provided an approach to
selectively probe the surface layers versus the crystalline bulk. The SFG signal of Avicel from
the TIR-SFG-VS setting was greatly dominant by Avicel surface structure while the non TIR-
SFG-VS could dominantly detect the bulk crystalline structures. Non TIR and TIR-SFG-VS
provide crucial tools to systematically study Avicel structures and its changes in various
processes, such as cellulose surface changes and bulk crystalline disruptions.
The TIR-SFG-VS spectrum of Avicel was significantly different from bulk crystalline
99
spectrum, which indicated the structural difference between Avicel surface and bulk crystals
(Figure 4.3). The reduction of SFG signal intensity of Avicel surface compared to bulk
crystalline structure signal by 10 times suggested that Avicel surface was not as crystalline as
bulk Avicel but the surface was still crystalline since it has signals in SFG-VS. Thus, the results
demonstrated layers of less ordered or less crystalline cellulose distributed on Avicel surface
which surrounded cellulose core crystalline structure. The cellulose microfibrils assembling is
still controversial and several models were proposed. For example, microfibrils were reported to
be consist of bands of crystalline cellulose which was surrounded by amorphous cellulose 281.
But the fringed micellar hypothesis suggests that both crystalline and amorphous areas appeared
at any one cross section of the microfibrils alternatively 237, 282. In addition, the folded-chain
structure demonstrated that microfibrils were helical along the fibril axis, and the model
indicated the distortion or twisted of cellulose chain exists as less crystalline areas without
amorphous areas 282, 283. Thus, our results tended to support 'the surrounding hypothesis'.
The TIR-SFG-VS spectra of Avicel surface under two different polarization settings
indicated slightly but noticeable polarization dependence (Figure S4.3a). Bulk Avicel crystalline
structure was reported to be complete polarization independence 268. The difference in
polarization dependence of Avicel surface and bulk crystalline region also demonstrated that
Avicel surface was not as crystalline as bulk Avicel. Avicel surface was composed of less
crystalline cellulose which surrounded around bulk crystalline structure. Since Avicel surface
showed same peak distribution with the bulk in SFG-VS spectra, those less crystalline cellulose
on surface was probably because of the disordered arrangement of cellulose chain at surfaces.
The disordered arrangement of crystalline cellulose surface referred to the less organized or
100
packed cellulose microfibrils than the bulk orderly and linearly arranged crystalline structures,
which caused different molecular orientations (C-H and O-H in Figure S4.4). This disorder might
be because of the reported twisted cellulose fibrils, paracrystalline celluose or simply randomly
distributed cellulose chains.
In addition, the comparison of non TIR and TIR-SFG-VS spectra of Avicel can help resolve
some mysteries of peak assignments. The curve fitting via Lorentz profile convoluted with a
Gaussian intensity distribution method provides theoretical peak position, amplitude and width
(Figure 4.3; Table S4.1a,b). Avicel surface spectra of TIR-SFG-VS showed peaks at 2872, 2937,
2968, 3494, and 3696 cm-1. Avicel surface and bulk crystalline spectra showed similar peak
positions at ~2857, ~2948, and ~2964 cm-1 but completely different at ~3328, ~3467, ~3508 and
~3700 cm-1. Although Avicel surface and bulk crystalline spectra shared some similar peaks in C-
H vibirational region, the peak intensities and width were differed greatly. All those difference
revealed significant structural difference between Avicel surface and bulk crystalline. The peak
assignments of C-H region are challenging because of Fermi resonance effects. The peaks at
~2857 and ~2948 were assigned to be -CH2 symmetric stretching by Fermi resonance effects 284,
285. Also, the peaks at ~2857 and ~2948 cm-1 in previous Avicel crystalline structure study were
assigned to the -CH2 symmetric and asymmetric vibrations, of the exocyclic -6CH2OH group,
respectively 85. Our results suggested these peak assignments should be revisited. If both 2857
and 2948 cm-1 were from same C-H stretching of -6CH2OH group, these two peak relative ratios
in Avicel surface and bulk crystalline spectra should be the same. However, our results showed
significant difference of peak relative intensity at 2857 and 2948 cm-1 of Avicel surface and bulk
crystal spectra, which firmly suggested that these peaks were from different C-H vibrational
101
sources. Since in glucose unit, C-H vibration from carbon ring is more abundant than in side
chain -6CH2OH. The carbon ring was usually believed in ordered and linear distribution 286.
Given in bulk crystalline spectra, peak at 2948 cm-1 was a lot higher than 2857 cm-1, 2948 cm-1
was more likely to originate from C-H stretching of carbon rings while 2857 cm-1 was from C-H
vibration of the exocyclic -6CH2OH group (Figure S4.4). Our peak assignment at ~2948 cm-1
also agreed with some IR or VCD studies of glucose indicated that ~2948 cm-1 was likely to be
C-H vibration mode from glucose carbon ring 227, 287. Additionally, due to the higher SFG peak
intensity of 2857 cm-1 compared to 2948 cm-1 on Avicel surface, -6CH2OH groups showed more
crystalline signals than glucose carbon ring on surface layers. This indicated that glucose carbon
ring might not well oriented on the surface. Thus, the arrangement of glucose units on Avicel
surface was less crystalline or less ordered distribution than the bulk, which was consistent with
previous discussion. In O-H vibrational region (Figure 4.3b), the peak at 3328 cm-1 could not be
seen in the TIR-SFG-VS spectra of Avicel surface, however, it was a major peak in Avicel bulk
crystalline structure. Peak at 3328 cm-1 was usually assigned to intermolecular O-H vibration
such as O(6)H…O(3) 81, 288. Thus, Avicel bulk contained strong intermolecular hydrogen bonding
while Avicel surface was not, which also agreed with previously less crystalline on surface
discussed. The intramolecular hydrogen bonding and free O-H appeared more likely in high
frequency of wavenumbers 81, 288-290. Figure 4.3b showed the prevalence of intramolecular
hydrogen bonding in low intensity and free O-H on Avicel surface. This also endorsed that Avicel
surface layers were less crystalline than the bulk crystals.
Avicel is mainly composed of Iβ. To test our concept of characterizing cellulose surface
layers and bulks separately, Iβ from tunicate (Halocynthiaroretzi species) was probed by TIR-
102
SFG-VS and Non TIR-SFG-VS. Detailed peak fitting parameters were shown in Table S4.1c,d.
The non TIR-SFG-VS spectra of Iβ showed common peaks with TIR-SFG-VS spectra at ~2919,
~2947, ~2970, ~3263, and ~3372 cm-1, which indicated the similarities of Iβ surface and bulk
structures. However, the intensities of these peaks in Iβ surface were about two times more than
Iβ bulk, which indicated that Iβ surface glucan chains presented more ordered orientation than
Avicel surface versus Avicel bulk. Although the peak intensities of Iβ surface were two times
more than Iβ bulk, Iβ surface glucan chains were still less ordered in orientation or crystalline
than Iβ bulk since TIR was supposed to enhance the signals for more than 100 times in surface if
the surface and bulk were uniformed or crystalline at the same level (Figure 4.1d). The results
were consistent with previous conclusion of Avicel on the significant difference of cellulose
surface layers and bulks characterized by Non TIR and TIR-SFG-VS. The difference of relative
peak ratios of these common peaks in Iβ surface and bulk can predict different molecular
orientations on surface and bulk. The peak ratios of ~2947 cm-1 with other peaks were greatly
differed in Iβ surface and bulk. The peak at ~2947 cm-1 was assigned to C-H vibrations from
carbon ring as discussed. The peak ratios difference in the surface and bulk indicated
significantly different C-H orientations in Iβ surface and bulk. These different C-H orientations
on the cellulose surface probably were caused by disordered arrangements of glucan chain,
which confirmed previous discussion. The peak of Iβ surface at 2864 cm-1 was right-shifted
compared with the peak in Iβ bulk. This peak was assigned to C-H vibrations from -6CH2OH
group. The difference of Iβ surface and bulk in peak from -6CH2OH groups indicated the
different –CH2 orientations in Iβ surface and bulk. The peaks at ~2919 and ~2970 cm-1 were
proposed to be C-H vibrations from different carbons on carbon ring of glucose units in
103
cellulose, which was difficult to be distinguished exact carbon locations from each other. Their
different peak ratios with peak at ~2947 cm-1 in Avicel bulk and surface were attributed to the
complex disordered arrangement of carbon ring, which cause peak cancellation. Compared to
Avicel C-H vibrational region, Iβ surface and bulk demonstrated more peaks in C-H region. The
difference of Avicel and Iβ in these peaks might be because of peak cancellation of the
composition of Iα in Avicel. Instead of polarization independence of Iβ bulk crystalline as
reported 268, Iβ showed obvious polarization dependence (Figure S4.3b), which also confirmed
the difference of Iβ surface and bulk.
Figure 4.4 SFG-VS spectra and peak fittings via Lorentz profile convoluted with a Gaussian
intensity distribution method of Iβ surface (red color) and bulk (blue color) within wavelength of
104
(a) 2800 to 3000 cm-1 and (b) 3000 to 3750 cm-1; Dots represented experimental data while lines
represented curve fittings. All intensities were calibrated at the same voltage (700 v) for the SFG-
VS measurements
In O-H vibration region, Iβ bulk showed peak at 3322cm-1 while Iβ surface shifted right to
3337cm-1. 3322cm-1 was assigned to intermolecular hydrogen bonding while 3337cm-1 was
attributed to intramolecular hydrogen bonding such as O(3)H…(O)5 81, 288. This shift was
consistent with previous Avicel surface hydrogen bonding that surface was distributed
intramolecular hydrogen bonding while lack of intermolecular bonding. 3322cm-1 was assigned
to intra-chain/molecular hydrogen bonding while peaks at less than 3300 cm-1 were more likely
to be intermolecular hydrogen bonding 291. The difference implied the difference of hydrogen
bonding network in surface and bulk of cellulose Iβ.
In summary, the surface layers of Avicel and Iβ are selectively probed by our developed
TIR-SFG-VS. The significant spectra difference between the surface and bulk domains of these
two cellulose particles indicated that Avicel and Iβ surfaces were less crystalline in comparison
to their bulk crystals. However, Iβ was more uniformed than Avicel on surface and bulk and thus
the surface of Iβ showed more crystalline or orderly oriented than Avicel surface layers. The
peak assignments of the surface layers and the crystalline cores of the two cellulose suggested
that the main peak around 2948cm-1 was likely to originate from C-H vibration on the carbon
ring of the glucose units instead of reported from side chain –6CH2OH mode while peak at
~2860cm-1 was from C-H vibration from side chain –6CH2OH. The difference of C-H vibrational
regions in Avicel surface layers and bulks suggested the different –CH and –CH2 molecular
orientations which might be attributed to the disordered arrangments of cellulose surface
105
microfibrils. Also, the O-H vibrational region observation indicated cellulose crystalline surface
layers mainly distribted intramolecular hydrogen bonding and free OH, while the bulk contained
mainly intermolecular hydrogen bonding. These valuable information helps the understanding of
cellulose bulk and surface, which contributes to future cellulose transformation studies. Non TIR
and TIR-SFG-VS can allow performing in situ observation of chemical and enzyme degradation
experiments on the cellulose surface and bulk, separately.
4.4.2 The characterization of thermal behaviors of dry cellulose by SFG-VS
The effects of high temperature on cellulose stability during biomass aqueous
pretreatment are important to help improve sugar yields. In our study, SFG-VS was used to
observe the molecular structural changes of cellulose crystalline bulks and surface layers in
dynamic heating program from room temperature to 280°C.
0.004
0.002
0.000
SF
G I
nte
nsity (
a.u
.)
36003400320030002800
Wavenumber (cm-1
)
Dry Avicel (Under Flat Window)Open air800 V; SSPVIS 65°C; IR 55°; IR 200µJ; VIS 110µJ
Room temperature (RT) fit_RT 65°C fit_65°C 100°C fit_100°C 130°C fit_130°C 175°C fit_175°C 200°C 220°C
106
Samples R (C-H peaks area/ O-H peak area)
Original Avicel 0.73
65C 0.69
100C 0.60
130C 0.53
175C 0.55
200C N/A
Figure 4.5 Cellulose crystalline structural changes under different temperatures during heating
process
The molecular structural characterization of the dynamic heating process of cellulose
(Avicel) is for the first time reported in this study. As the increasing of temperature, all peak
intensities at ~2850, ~2945cm-1 and ~3310 cm-1 decreased suggesting the disruption of cellulose
crystalline structures (Figure 4.5). In the previous discussion, peaks at ~2945 cm-1 and ~2850 cm-
1 were from C-H vibration on cellulose carbon rings and side chain -6CH2OH, respectively. At
temperature >175°C, all peaks disappeared indicating the complete disruption of cellulose
crystals. It was reported that cellulose Iα and Iβ can be converted to a I-HI 240, 241. However, the
molecular structure of this I-HI is unknown. Our results showed that with the dynamic increasing
of heating temperatures, this metastable intermediate was less and less crystalline suggested by
the peak decreasing in Figure 4.5. When reaching 280C, the crystalline features of this high
temperature intermediates were gone. Besides, the mild hydrolysis and dehydration of cellulose
might occurred in this thermal disruption on dry cellulose 292. Interestingly, peak at ~3310 cm-1
was shifted to a higher wavenumber along with the increasing of heating temperatures. This
change of peak intensity and position at ~3310 cm-1 indicated the gradually break-down of
hydrogen bonding with the thermal effects as well as transformation of hydrogen bond. Thus, we
hypothesize that the heating temperatures induced the dislocation, random-arrangement and
disordering of cellulose Iα and Iβ microfibrils resulting in the gradually loss of crystalline
107
structure to form a complete amorphous high temperature metastable intermediates eventually.
More future work is still needed to confirm this hypothesis.
Additionally, the peak area ratios of C-H and O-H vibrations decreased with the increasing
of heating temperatures till close to the previously reported ratio signature of cellulose Iβ 28
(Figure 4.5). Thus, the resulting new cellulose intermediates at heating temperatures have some
similarities with cellulose Iβ before turning to amorphous. The Iα may be converted to Iβ besides
to the high temperature intermediates during the dynamic heating process to cause these changes
of the peak area ratios. It was proposed that the conversion of cellulose Iα to Iβ was through
converting cellulose Iα to the high temperature phase and then followed by the conversion of the
intermediates to cellulose Iβ 240. Our results indicated that the cellulose Iβ was aggressively
transformed to the metastable intermediates at > 175C. However, a more comprehensive
conclusion still needs more work, such as the test of cellulose Iα and Iβ individually in dynamic
heating process.
The heated Avicel samples were cooled down to room temperature and characterized by
SFG-VS (Figure 4.6). Surprisingly, Figure 4.6 presented the regaining of crystalline structures
after cellulose was cooled down to room temperature. When cellulose was heated to 220°C, all
peaks disappeared compared to the original spectra (not heated sample).
108
Figure 4.6 The recrystallization of dry cellulose polymer after cooling down from heating
process
After the sample was cooled down to room temperature for one hour, the peaks’ intensity
was recovered 2/3 in comparison with the original peak intensity. The observation of the sample
after cooling overnight at room temperature showed almost the same intensity levels with the
original spectrum. When observing the different spots of the sample, the spectra showed different
peak intensities and peak ratios between ~2945 and 3310cm-1 among them, suggesting that the
regained crystalline structure was heterogeneous. When compared the spectra of newly regained
crystals to the original Avicel, the crystals were not exactly back to the original state of
0.005
0.004
0.003
0.002
0.001
0.000
SF
G I
nte
nsity (
a.u
.)
36003400320030002800
Wavenumber (cm-1
)
Recrystallization of heated dry cellulose after cooling Under Flat Window800 V; SSPVIS 65°C; IR 55°C
Original sample at RT (before heating process) Heated to 220°C RT (cooling from 220°C for 1 hour) RT after cooling overnight RT after cooling overnight different sample spot
Samples R (C-H peaks area/ O-H peak area)
Original Avicel 0.73
RT after cooling overnight 0.63
RT after cooling overnight 2nd spot 0.57
109
crystalline structure, for example, the peak area ratios of new crystals were 0.63 and 0.57
compared to the original 0.73. Also, the O-H region showed slightly shift to higher wavelength.
Significantly, the peak area ratios of new crystals were 0.63 and 0.57 which suggested that the
recrystallized crystals were showing more similarities to cellulose Iβ which indicated that the
transformation of cellulose Iα was irreversible. It was reported that after heated samples cooling
down to room temperature, the heated cellulose was transformed to cellulose Iβ 249. Our results
indicated that the new formed crystals were not exactly the same with Iβ. We hypothesize that
the cooling process of heated cellulose samples undergoes a cellulose recrystallization process
which refers to the rearrangement of dislocated/disordered cellulose microfibrils during heating.
The recrystallization process during cooling might be triggered by the tendency of cellulose
microfibrils to reach stable structure/microfibrils orientation. Cellulose crystals are high ordered
crystalline structure, which makes the recrystallization of disrupted microfibrils back
substantially but still presenting differences. This hypothesis should still be validated via many
experimental results and observations. More work is needed to study the differences of the
recrystallized crystals with the original cellulose crystals.
The recrystallization and the related annealing are the thermomechanical processing of
materials which are known to occur in all types of crystalline materials. The recrystallization has
been widely studied in crystalline metals or metallic materials 293. The recrystallization of most
crystals (grains) has been widely studied. It was reported that the recrystallization may occur
during or after deformation (during cooling or a subsequent heat treatment) driven by a
thermodynamic driving force to reduce the internal energy of the system 293. It is noteworthy to
mention that most of recrystallization of metal crystals is random on deformed crystals but the
110
cellulose crystalline materials were mostly recovered although they are still different from
originals (Figure 4.6). However, the disrupted cellulose microfibrils were still not a complete
recovery process which may be because of the irreversible cellulose Iα. The cellulose crystalline
structure is the highly ordered structure but its recrystallization of disrupted crystals is not well
defined and understood before, especially at the molecular structure level. The molecular
recrystallization phenomenon of cellulose is not seen reported, although some studies might be
trying to describe the phenomena, such as the existence of cellulose Iα or Iβ after cooling down
heated cellulose Iα and Iβ 249, and the returning back of XRD reflections of cooled cellulose to
the originals 239. Our SFG-VS can study detailed molecular structural changes during the
recrystallization process of crystalline bulks and surface layers. However, more future work
should be carried out to confirm the hypothesis and provide more details.
4.4.3 Structural changes of wet cellulose during heating process
Water plays a significant role in biomass hydrolysis in aqueous pretreatment. However,
the fundamental studies of the impact of moist on cellulose crystalline changes during dynamic
heating are little, probably, due to the limitation of analytical tools. The wet cellulose crystalline
structural change during dynamic heating process was studied by SFG-VS (Figure 4.7). As the
increase of heating temperatures, similar to the dry cellulose dynamic heating process, the peak
intensities at ~2850, and 2950 cm-1 decreased. All peaks disappeared when the temperature
reached 170°C which was lower than the transformation temperature of dry cellulose. This might
be because of the assistance of water in the disruption of cellulose crystalline structures by
hydrolysis. Also, this was consistent that a previous report that high moist cellulose was easier to
decrystalline than oven-dry cellulose 239.
111
Figure 4.7 Cellulose crystalline structural changes of wet Avicel heating process
Similar to the recrystallization of heat-disrupted dry cellulose, when the sample was
cooled down to room temperature, the peaks 2850 and 2950 cm-1 went back to the similar
lineshapes of original cellulose peaks (Figure 4.7). However, the water has great impact on the
SFG characterization which will be discussed in next few sections. But the peak intensity at
~3310cm-1 increased more than three times. Thus, this recrystallized cellulose structure was
different from the original crystalline structure. This indicated that the recrystallization process
of wet cellulose occurred at room temperature after it was cooled down from high heating similar
to dry cellulose recrystallization process discussed above. In the study of the transformation of
native cellulose crystals with saturated steam at high temperatures, it was reported that both
cellulose Iα and Iβ were transformed into a cellulose Iα' or Iβ after cooling down 249. This Iα' or
0.0020
0.0015
0.0010
0.0005
0.0000
SF
G I
nte
nsity (
a.u
.)
380036003400320030002800
Wavenumber (cm-1
)
Wet Avicel (Under Flat Window)Open air1200 VVIS 65°; IR 55°
Room temperature (RT) fit_RT 65°C fit_65°C 100°C fit_100°C 135°C 170°C Cool down to RT
112
Iβ might be originated from the recrystallization of disrupted cellulose crystals indicated in our
study. Our study provides an alternative explanation of the formation of crystalline structure Iα'
or Iβ in the cooling process of heat-disrupted cellulose crystals. Nevertheless, the
recrystallization of disrupted cellulose crystals is a significant consideration in designing
effective aqueous pretreatment techniques.
Figure 4.8 Recrystallization process during cooling process of wet Avicel
The recrystallization of heated wet cellulose observed was similar to the recrystallization of
heated dry cellulose. Figure 4.8 dynamically monitored recrystallization process during the
cooling process from high heating temperature of 280°C to room temperature. As the
temperature slowly decreased, the peak intensities increased back to almost same level as
original cellulose peak intensities. The O-H vibrational peaks were returning back fast suggesting
0.0020
0.0015
0.0010
0.0005
0.0000
SF
G I
nte
nsity (
a.u
.)
380036003400320030002800
Wavenumber (cm-1
)
Wet Avicel cooling280°C to room temperature1200V; SSPVIS 65°; IR 55°
Original sample at RT (Before heating process) At 280°C 175 to125°C 125 to 90°C 90 to 70°C 57 to 45°C 32 to 30°C RT after cooling
113
that the hydrogen bonding was easier to be rearranged with water presence. These new crystals
by recrystallization during cooling process showed differed peak intensity ratios with originals
which was probably attributed to different molecular structure and orientation.
4.4.4 The different SFG-VS signals of dry and wet Avicel bulk crystals
Figure 4.9 Impact of water on SFG-VS characterization of Avicel
The peak intensity of wet cellulose by SFG-VS was 140 times less than dry cellulose
peak intensity (Figure 4.9). This was because the water can absorb IR and VIS lasers which
reduced the intensity of peaks. Also the peak intensity ratio between peak ~2945 and ~3310cm-1
was different in dry and wet Avicel bulks. The peak ratio/area ratio of peak 2945cm-1 and
3310cm-1 (-OH) in dry Avicel spectrum was smaller than in wet spectrum. This also might be
because the impact of water on the hydrogen bonding of cellulose, which can be an interesting
0.05
0.04
0.03
0.02
0.01
0.00
SF
G I
nte
nsity (
a.u
.)
380036003400320030002800
Wavenumber (cm-1
)
x140
Avicel Bulk (CaF2 Window)
Room temperature1000V; SSPVIS65°, IR55°
Dry Wet
114
story to study in the future.
4.4.5 Study of the temperature dependence of dry Avicel surface layers
0.002
0.001
0.000
SF
G I
nte
nsity (
a.u
.)
3000295029002850280027502700
Wavenumber (cm-1
)
Dry Avicel (Under Prism)950 V; SSPVIS65°; IR55°
room temperature 65°C 100°C 130°C 175°C 200°C 220°C 240°C 280°C
0.0030
0.0025
0.0020
0.0015
0.0010
0.0005
0.0000
SF
G I
nte
nsity (
a.u
.)
3000295029002850280027502700
Wavenumber (cm-1
)
Dry Avicel (Under prism)Cooling process950 V; SSPVIS65°; IR55°
Original untreated At 280°C 280°C-121°C 120°C-70°C Room temperature Room temperature overnight
115
Figure 4.10 Temperature dependence of Avicel surface layers characterized by SFG-VS (up:
structural changes of dry Avicel surface layers during heating process; down: heated dry Avicel
surface layers cooling down to room temperature)
The molecular structural characterization of cellulose surface layers in a dynamic heating
process has not been reported before. With the increasing of temperatures, the peak intensities of
Avicel surface layers (at 2872, 2940cm-1) slightly increased until 175-200°C, which might be
because some cellulose amorphous gained more crystalline structure at elevated temperature 294.
When temperatures were higher than 200C, the cellulose surface layers were disrupted.
However, at 280C, the signals of the surface layers were still weakly present. Thus, the cellulose
surface layers need a higher temperature (280C) to completely disrupt surface layers of
cellulose than cellulose crystalline bulks (Figure 4.10). Unlike Avicel bulk crystals, the heat-
disrupted surface layers did not recrystallize during/after cooling down from high heat (Figure
4.10). This probably because the cellulose bulks possessed high stored energy of recrystallization
while not the amorphous surface layers although the surface layers present some ordered crystals
293. Since the interference of DI water on SFG-VS signals of cellulose surface layers, more future
work is still needed to analyze the results and will not be discussed in this thesis.
In aqueous pretreatment, the complete disruption of some ordered microfibrils in cellulose
surface layers requires temperature higher than 240C according to our observations. Our results
might explain some publications including our previous publications that the optimal
temperatures were higher than 240C in dissolving cellulose completely 59, 60, 243. Even the bulk
crystals can be disrupted under less than 200C, the recrystallization of these disrupted to
crystalline structures again may lead to the difficulties of aqueous pretreatment. More
116
experimental results or observation should be conducted to validate this hypothesis.
4.4.6 Molecular structural changes of cellulose in enzymatic hydrolysis using SFG-VS
Cellulose enzymatic hydrolysis is a solid-liquid interface surface reaction to hydrolyze
carbohydrate biopolymers, which is slower than direct enzyme and bulk cellulose reactions 295,
296. Thus, the rate of cellulose hydrolysis is controlled by several factors, including the cellulose
surface and structure after aqueous pretreatment, the assay cellulosic enzymes and the surface
interaction between them 296. Enzymatic hydrolysis technology requires a better understanding of
the fundamental mechanisms underlying the bio-interactions of enzymatic cellulose hydrolysis,
which includes the tracking of cellulose structural change after biomass aqueous pretreatment
and potential affinities with enzymes to facilitate disrupting crystallizing cellulose during
enzymatic hydrolysis. Further efforts should focus on the interaction of cellulase with biomass
substrates at molecular-level hydrolysis with the enhancement of cellulose decrystallization.
These goals rely on a reliable characterization tool that is highly sensitive to cellulose crystalline.
This tool will be helpful in characterizing aqueous pretreated cellulose structure and its
interaction with various enzymes.
The SFG-VS peak decay of the samples withdrawn from enzymatic hydrolysis was explored
in this study to correlate with cellulose enzymatic hydrolysis efficiency (Figure 4.11). The life
decay of O-H was about two times of life decay of C-H, which indicated that C-H vibrational
regions are important in studying enzymatic hydrolysis system. Figure A1 showed the SFG-VS
can study enzymatic hydrolysis without the concerns of peak interferences between enzymes and
cellulose. Here, the feasibility of the application of SFG-VS in characterizing enzymes and
cellulose together was proposed for the first time. This is important to study the molecular
117
structural changes of enzymes and cellulose and their interaction in enzymatic hydrolysis.
Figure 4.11 SFG-VS spectra of the decay of enzymatic hydrolysis to study enzymatic hydrolysis
kinetics
118
Samples R (C-H peaks area/ O-H peak area)
Original Avicel 0.70
1 h 0.77
2 h 0.83
3 h 0.94
4 h 0.92
13 h 0.93
20 h 1.07
Figure 4.12 SFG-VS characterization of Restart samples from interrupted enzymatic hydrolysis
The cellulose restart samples (preparation method explained in material section) from
enzymatic hydrolysis slurries were characerized by SFG-VS to determine the crystalline
structural changes. With the increasing of enzymatic hydrolysis time, the peak area ratios of
vibrational C-H and O-H regions increased from 0.70 to 1.07 (Figure 4.12). Compared to the R
SF
G inte
nsity (
a.u
.)
40003800360034003200300028002600
Wavenumbers (cm-1
)
Avicel restart samples (Flat Window)850 V; SSPVIS 45°; IR 56°
Avicel fit_Avicel 1 h fit_1 h 2 h fit_2 h 3 h fit_3 h 4 h fit_4 h 13 h fit_ 13 h 20 h fit_ 20 h
119
values of Avicel, cellulose Iα and cellulose Iβ our previous study reported 28, the cellulose
crystals were closer and closer to cellulose Iα with the time goes. Thus, cellulose Iβ might be
more reactive in enzymatic hydrolysis and left cellulose Iα behind. However, the enzymatic
hydrolysis rate of cellulose Ia was found higher than cellulose Ib due to the differences in crystal
structures 297, 298. This indicated that the statement that cellulose Iβ may have better reactivity
than cellulose Iα during enzymatic hydrolysis might be wrong. More experimental work is
needed to clarify this standpoint.
4.5 Conclusion
The selective characterization of the surface layers of Avicel and Iβ is achieved by our
developed TIR-SFG-VS for the first time. The significant spectra difference between the surface
layers and bulk domains of Avicel and Iβ particles indicated that the surface layers of these two
celluloses were less crystalline compared to their bulk crystals. However, Iβ was more uniformed
than Avicel on surface layers and bulks and thus the surface layers of Iβ showed more crystalline
or orderly oriented than Avicel surface layers. The peak assignments of the surface layers and the
crystalline bulks of the two celluloses suggested that the main peak around 2948cm-1 was likely
to originate from C-H vibration on the carbon ring of the glucose units instead of reported from
side chain –6CH2OH mode while peak at ~2860cm-1 was from C-H vibration from side chain –
6CH2OH. The difference of C-H vibrational regions in Avicel surface and bulk suggested the
different –CH and –CH2 molecular orientations which might be attributed to disordered
arrangments of cellulose surface microfibrils. Also, the O-H vibrational region observation
indicated cellulose crystalline surface mainly distributed intramolecular hydrogen bonding and
free O-H, while the bulks contained mainly intermolecular hydrogen bonding. These valuable
120
information helps the understanding of cellulose bulk and surface, which contributes to future
cellulose transformation studies. Non TIR and TIR-SFG-VS can allow performing in situ
chemical and enzyme degradation experiments on the cellulose surface layers and bulks,
separately.
The TIR-SFG-VS enabled the molecular structural characterization of cellulose surface
layers for the first time, which is important for the study of cellulose behaviors in aqueous
pretreatment and enzymatic hydrolysis. In this chapter, a fluid cell was applied to mimic aqueous
pretreatment so that it allows the study of the molecular structural changes of the surface layers
and bulks of cellulose in a dynamic heating program with/without water. Both wet and dry
cellulose crystalline bulks were disrupted with the increasing of temperatures but wet cellulose
bulks were degraded faster. It was proposed that the microfibrils of cellulose Iα and Iβ were
possibly dislocated/disrupted/random-oriented by heating to form an microfibril-disordered
metastable intermediate during dynamic heating process and cellulose Iα may be converted to Iβ
besides converted to the intermediate which is not reversible. Based on the results of thermal
stability of heat-disrupted cellulose crystals during and after cool-down to room temperature, a
hypothesis of a possible cellulose recrystallization was proposed that the rearrangments of
dislocated/disrupted cellulose microfibrils possibly occured and were possibly drived by the
tendancy of disrupted cellulose microfibrils to a stable structure. The recrystallization was not a
complete recovery process which resulted in new rearranged crystals. These crystals were similar
to the orignal crystals but different in the orientation of microfibrils, probably, due to the
orignally highly ordered crystalline structure of cellulose. Due to the amorphous of cellulose
surface layers, some ordered crystals in surfaca layers were not recrystallined during/after
121
cooling to room temperature. This hypothesis still needs more future work to get stronger
evidences besides this chapter indicated, such as the structural characterization of newly
recrystallized crystals by HR-BB-SFG-VS and other tools, weight and wet chemistry analysis of
heated cellulose, dynamic heating test on cellulose Iα and Iβ, and applying other analytical tools
to test the hypothesis. This work is helpful in designing an effective pretreatment technology and
other thermal processing of cellulose. At last, our SFG-VS study on enzymatic hydrolysis system
suggested that SFG-VS can be possibly developed to study cellulose crystalline structure
degradation during enzymatic hydrolysis, such as the interaction of enzymes and cellulose
without separating them. Results indicated that C-H structural changes are important in
enzymatic hydrolysis since the decay of C-H was two times slower than O-H vibrational peaks.
Thus, the application of SFG-VS in the study of fundamental issues of aqueous pretreatment and
enzymatic hydrolysis is important to gain new insights of cellulose transformation and properties
as a disordered crystalline material.
4.6 Acknowledgements
The author acknowledges the support of Bioproducts, Sciences and Engineering Laboratory,
Department of Biosystems Engineering at Washington State University. The thesis author was
partially supported by the grant from the Chinese Scholarship Council (CSC). The author thanks
Dr. Shi-you Ding for insightful discussions and providing one of the celluose samples at the
BioEnergy Science Center, a DOE Bioenergy Research Center, and the Genomic Science
Program (ER65258), both supported by the Office of Biological and Environmental Research in
the DOE Office of Science. Part of this work was conducted at the William R. Wiley
Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility located
122
at the Pacific Northwest National Laboratory (PNNL) and sponsored by the Department of
Energy’s Office of Biological and Environmental Research (BER). The author acknowleges
Zizwe Chase from Washington State University for his help in completing some of the
experiments. The author would like to present appreciations to Drs. Hongfei Wang and Bin Yang
for providing great platforms and have insightful discussions with the author. The author of the
thesis would like to thank Drs. Li Fu, Shunli Chen, HongFei Wang, and Bin Yang for
contributing to this work and helping in preparing this work for submission to a journal.
4.7 Supplementary materials
Table S4.1 Peak positions, amplitudes, and widths after curve fittings of SFG-VS spectra via
Lorentz profile convoluted with a Gaussian intensity distribution method in (a) TIR-SFG-VS
spectra of Avicel surface layers, (b) SFG-VS sepctra of Avicel bulks, (c) TIR-SFG-VS spectra of
cellulose Iβ surface layers, and (d) SFG-VS spectra of cellulose Iβ bulks within the wavelength
of 2800 to 3750 cm-1
(a)
position/(cm-1) amplitude/(a.u) width/(cm)
2872 ± 2.9 3.3 ± 0.2 49 ± 2
2940 ± 1.3 0.3 ± 0.1 16 ± 3
2965 ± 0.8 -0.1± 0.03 5.6 ± 1.4
3508 ± 3.7 0.3 ± 0.06 22 ± 2
3700 ± 3.4 0.2 ± 0.05 17 ± 2
(b)
position/(cm-1) amplitude/(a.u) width/(cm)
2857 ± 2 1.8 ± 0.2 38 ± 3
2948 ± 0.2 2.8 ± 0.1 13.4 ± 0.4
2964 ± 0.5 0.49 ± 0.1 9.0 ± 1.2
3328 ± 1.3 3.4 ± 0.3 41.0 ± 2
3467±7 7.3±1.1 143 ±19
3686 ± 3 4.2 ± 0.0.5 76 ± 6
(c)
position/(cm-1) amplitude/(a.u) width/(cm)
2864 ± 1.2 0.75 ± 0.06 23.2 ± 2
2919±1.2 0.19±0.05 10.9±3
123
2947 ± 0.3 1.2 ± 0.07 15.2 ± 0.7
2968 ± 0.2 0.23± 0.02 6 ± 0.4
3263±1.6 -2.55±0.1 37±1.7
3337±0.6 3.75±0.2 34±1.1
3375 ± 2 0.13 ± 0.07 14 ± 5
(d)
position/(cm-1) amplitude/(a.u) width/(cm)
2851 ± 1 1.0 ± 0.04 34.1 ± 1.5
2919±1 0.27±0.04 14.4±1.9
2947 ± 0.5 0.28 ± 0.04 11.9 ± 1.3
2970 ± 0.1 0.37± 0.01 7.3 ± 0.2
3260 ± 1.6 -1.1 ± 0.06 26.7 ± 1.5
3322±0.8 1.63±0.05 30.4±0.8
3372 ± 1.1 0.04 ± 0.01 6.0 ± 1.9
124
Figure S4.1 Whole spectra of Avicel and Iβ bulks and surface layers a) Avicel; b) Iβ (black:
blank control; blue: bulks; red: surface layers)
125
Figure S4.2 Demonstrations of various molecular structures of Avicel surface layers and
relatively stable Iβ structure
126
Figure S4.3 TIR-SFG-VS spectra of (a) Avicel and (b) cellulose Iβ under two polarizations, (red
color) SSP and (Blue color) PPP; s-, s-, p- polarization corresponding to the polarization of the
optical fields of the SFG signal, visible and IR beams, respectively. p polarization is defined
when electric field vector parallel to the plane of incidence within the incident plane formed by
the incident beam direction and the surface normal, and s polarization is perpendicular to the
incident plane
127
Figure S4.4 Cellulose structure (two glucose units) and different molecular vibrations (cellulose
adpated from 286 )
Figure S4.5 Heat disruption and recrystallization hypothesis (Rollett, A., et al. (2004).
Recrystallization and related annealing phenomena, Elsevier.)
128
Calculation of the Fresnel factor in the SFG measurement
In our experiment, we used ssp polarization setting (s-polarized SFG, s-polarized visible and
p-polarized IR) for SFG measurement, with the effective second order susceptibility as
(2)
ssp ( ) ( ) ( )sinyy SFG yy Vis yy IR IR yyzL L L ,
where ( ) ( ) and ( )yy SFG yy Vis yy IRL L L , ,
are Fresnel factors, and IR is the incident anlge of IR
beam. The Fresnel factors can be calculated as follows:
1
1 2
2 ( )cos( )
( )cos ( )cos
SFG SFGyy SFG
SFG SFG SFG SFG
nL
n n
1
1 2
2 ( )cos( )
( )cos ( )cos
Vis Visyy Vis
Vis Vis Vis Vis
nL
n n
2
2 1
1 2
2 ( )cos ( )( )
( ) cos ( )cos '( )
IR IR IRyy IR
IR IR IR IR IR
n nL
n n n
in which i is the refractive angle from medium 1 into medium 2 defined by
1 2( )sin ( )sini i i in n . '( )IRn is the effective refractive index parameter of the interface layer,
whose definition and physical meaning was elucidated by Shen et al.
All the angles has been calculated and incorporated into the Figure 4.1, with the simulated
SFG intensity plotted in Figure 4.1d. In particular, the relative ratio of the SFG intensity of the
two geometries used in our experiments can be determined. Under the TIR reflection geometry
using the prism (Figure 4.1a), the SFG intensity from the interface should be ~100 times of that
from the geometry used in Figure 4.1b.
129
CHAPTER FIVE
CHARACTERIZATION OF FLOWTHROUGH SOFTWOOD AQUEOUS
PRETREATMENT UNDER NEUTRAL AND ALKALINE PH AT
ELEVATED TEMPERATURES
5.1 Abstract
To well understand the fundamentals of aqueous pretreatment of softwood, the flowthrough
reactor was used and run pretreatment at 200-270 °C and 6.4-12.0 pH for 2-10 min with the flow
rate of 25 mL/min. Significantly, the starting pHs of pretreatment media have great impact on
pretreatment kinetics and chemistry. It was found that the initial pretreatment pHs decreased
from 6.4-9.0 to about 4.0 after pretreatment, which led to the acid hydrolysis kinetics of biomass.
However, high initial pHs (around 11.0) showed not significant changes of pHs after the
pretreatment, which resulted in the moderate alkaline pretreatment kinetics. When the starting
pH was ~12, most of the carbohydrates underwent severe alkaline degradation reactions resulting
in the formation of organic acids with the substantial loss of sugars. This makes the analysis of
the mixtures of products difficult. Moreover, 2-D 1H-13C NMR analysis of pretreatment
recovered insoluble lignin revealed that the cleavage of C-O-C linkages in lignin was the main
alkaline depolymerization mechanism. Thus, the hot water flowthrough pretreatment kinetics and
chemistry by controlling initial pretreatment pHs can add more fundamental understandings of
aqueous pretreatment and help make wise pretreatment designs.
Keywords: Softwood • Flowthrough pretreatment • Neutral and Alkaline • pH • Lignin
5.2 Introduction
Cellulosic biomass pretreatment is the key step to overcome the natural recalcitrance of
130
biomass in biofuel production. Softwood is one of the main types of woody biomass for biofuel
production and is well distributed in western United States. However, several leading
pretreatment methods were proposed and intensively studied on biomass such as poplar, corn
stover and switchgrass but not softwood 31. Softwood is more challenging to pretreat using the
leading pretreatment methods than hardwood and other herbaceous species 299, 300. This is
attributed to softwood lignin contents, compositions, and structures by most literatures 302, 304.
For example, in softwood lignin, 90-95% of Guaiacyl (G unit) is present while softwood lignin
contained less than 1% and 4% of Syringyl (S unit) and hydroxyphenyl (H unit), respectively 301,
302. Thus, it was proposed that softwood lignin possessed more available reactive C5 and more
condensed β-5 and 5-5 sub-units than hardwood lignin 302, 303. These characteristics add the
difficulties of softwood lignin removal in pretreatment. Besides, softwood was reported having
dense, strong and lacking pores 304, which makes less attention on application of softwood in
pretreatment.
Due to the lack of sufficient efforts on softwood pretreatment, the aqueous pretreatment of
softwood was poorly understood. Currently, there are very few effective pretreatment options
available for softwood 300. Some most commonly reported pretreatment methods of softwood
include dilute acid 305, post-alkaline (peroxide) after steam explosion 306 and acid/SO2 steam
explosion pretreatment 307, 308. In softwood dilute acid pretreatment, the recovery yield of
hemicellulose is achieved by more than 90% of maximum compared to the hemicellulose sugar
yield of ~65% of the maximum possible from autohydrolysis. Also, it was reported that in 0.4%
sulfuric acid pretreatment, 90-95% hemicellulose of the Douglas fir and 20% of the cellulose
were removed to liquid phase at 200-230°C for 1-5 min and 90% of C6 sugars was recovered in
131
enzymatic hydrolysis process 305. Besides, 90% of glucose yield was achieved by enzymatic
hydrolysis of pretreated Douglas-Fir at 200°C with 4% (w/w) SO2 for 5min 308. However, these
studies were facing the difficulties in lignin removal and generating inhibitions on enzymatic
hydrolysis, which impedes an efficient pretreatment technique on softwood. To utilize softwood
for sugar production, the effective softwood pretreatment is necessary.
The alkaline pretreatment was reported to be effective in lignin removal and the potential to
achieve pretreatment economic viability 309. The alkaline pretreatment is using high dosage of
alkalis (0.5-30 wt%) at low temperature (e.g. room temperature or < 200 °C) for a long residence
time. For example, the NaOH pretreatment was operated at room temperature to 180 °C, 0.5-
10% (w/v) loading, and reaction time of 5 min to several hours 310. The pine wood lignin was
extracted by alkali 1–8% NaOH, for 30–120 minutes at temperature 130–150°C and the
produced lignins were characterized regarding lignin removal efficiency and molecular weight of
derived lignin 311. However, it was reported sodium hydroxide is more effective with non-woody
biomass species indicated by calculating the activation energy and the Arrhenius constant of
pretreatment kinetics 312. It was also reported that the alkaline pretreatment methods could
remove lignin effectively, swollen cellulose fibers, release acetyl groups, reduce sugar
degradation loss and improve sugar yields in enzymatic hydrolysis 313, 314. Thus, the alkaline
pretreatment of softwood is still not well performed and understood yet.
The objective of this study is to improve fundamental understandings of neutral and alkaline
pretreatment of softwood by implementing the aqueous flowthrough pretreatment with starting
pH of neutral to 12. The flowthrough reactor system can serve as a significant tool to provide
great mass and heat transfer with a short residence time which warranties our accurate studies of
132
the patterns, kinetics, and mechanisms of biomass hydrolysis 27. It was reported a nearly
complete hydrolysis of poplar wood was achieved in aqueous and dilute acid flowthrough system
at elevated temperatures 59. Therefore, the study postulated that high lignin removal and yields of
sugars can be achieved when dilute alkalis and elevated temperatures (>200ºC) were applied.
The flowthrough pretreatment of softwood has not been reported yet. A detailed
understanding of the softwood flowthrough pretreatment at several neutral and alkaline
conditions is significant. The dilute alkali pretreatment is also not well seen in softwood 300. In
this study, aqueous flowthrough pretreatment of softwood at neutral-pH 12 at temperature of
200-270C was carried out to evaluate the hydrolysis kinetics of cellulose, hemicellulose, and
lignin. The study gains a fundamental knowledge of softwood aqueous pretreatment which can
be applied to improve pretreatment technology on softwood.
5.3 Materials and methods
5.3.1 Feedstock
Beetle-killed Lodge pole pine was provided by Forest Concepts, LLC, and milled to 40-60
mesh. The standing dead pine was harvested after death for 2-4 years in watershed creek
drainage near Walden, CO (no needles included). The pine wood chips contain 37.89±0.59% of
glucan, 5.25±0.047% of xylan, 29.95±0.78% of lignin, 3.62±0.093% of galactan, 2.25±0.018%
of arabinan and 9.89±0.021% of mannan, according to NREL composition analysis protocol 128.
5.3.2 Flowthrough pretreatment system
The same flowthrough pretreatment system was reported previously 59, 60. Briefly, the
flowthrough pretreatment system contains a 4kW fluidized sand bath, a high-pressure pump with
a flow rate of 0-100 ml/min, a cooling tank, solvent flowing tubing and connection fittings, and a
133
tubular reactor (1.3cm i.d. × 15.2cm length with an internal volume of 20.2 mL). Other
accessories include a pressure gauge, a thermal monitor, a stainless steel thermocouple, and a
back-pressure regulator.
5.3.3 Flowthrough pretreatment of pine wood chips
Pretreatment solutions were prepared using H2SO4 and NaOH for different pH solutions (6.4
(DI water), 8, 9, 11, and 12). The pH meter (pH 510 series, Oakton) was used for pH
measurement before and after pretreatment at room temperature. 0.5 g biomass was loaded into
the tubular reactor. The pretreatment solvents with various pHs were pumped by 25 ml/min
through the reactor. The reactor was placed in the fluidized sand bath which was heated to the
target temperatures (200-270 °C). The sand bath was set 10 ºC higher than targeted temperatures
and monitored by a thermal meter. After 2-10 min completion of experiments, the pretreated
hydrolysates were collected and 2ml of them was injected to HPLC for monomeric sugar
analysis. In addition, 10 ml of the hydrolysates were subjected to the posthydrolysis analysis at
121°C autoclave for 1 h with 4% sulfuric acid for total sugar recovery in pretreatment
(monomeric plus oligomeric sugars). The solid residues in tubular reactor were frozen dried for
K-lignin analysis which determined the compositions of residues after pretreatment 128. For the
recovered insoluble lignin collection (RISL), the pretreatment hydrolysates precipitated
overnight and followed by DI water washing and freeze drying. 2-D 1H-13C HSQC NMR of the
RISL structure was carried out following the same method reported 60. Pretreatment severity
(equation 5.1) was used to evaluate pretreatment performance.
LogR0 = Log[t × exp(T−100
14.75)] (5.1)
T: the temperatures applied in the pretreatment; t: the pretreatment time.
134
Besides, several parameters were used to evaluate the effectiveness of softwood flowthrough
pretreatment. The parameters were divided into two categories which were the removal and
recovery. Biomass, hemicellulose, cellulose and lignin removal results represented the removal
of biomass, hemicellulose, cellulose and lignin out of solid biomass loaded in the reactor
respectively, which was obtained by determination of dry mass weight change of biomass, K-
lignin analysis of the solid residues followed by a dry mass weighing after acid hydrolysis and
HPLC measurement of sugars in the solid residues. The water HPLC system (model 2695) is
equipped with a 410 refractive detector and a Waters 2695 auto-sampler using Waters Empower
Build 1154 software; Waters Co., Milford, MA, USA).The recovery result was from the direct
measurements of pretreatment or enzymatic hydrolysis liquid through HPLC and UV on sugar
contents in the liquids. In summary, the parameters included hemicellulose removal (%),
hemicellulose recovery (%), cellulose removal (%), lignin removal (%), and recovered soluble
lignin (RSL): (Equation details in previous study 60). Hemi-sugars recovery used in this study
refers to the recovery of C6 and C5 sugars from hemicellulose including xylose, mannose,
galactose, and arabinose. Other parameters included the following:
Cellulose recovery in stage 1 as C6 sugars%=mCRecovery
mC× 100 (5.2)
mCRecovery: HPLC detected glucose mass weight recovered in liquid phase of pretreatment after
posthydrolysis (Stage 1); mC: mass weight of glucose in the original loaded biomass.
Total cellulose recovery as C6 sugars%=mCRecovery
mC× 100 (5.3)
mCRecovery: glucose mass weight recovered in the liquid phases of pretreatment (Stage 1) and
enzymatic hydrolysis (Stage 2); mC: mass weight of glucose in the original loaded biomass.
135
5.3.4 GC/MS analysis
The hydrolysates after pretreatment were extracted with ethyl acetate, following vigorous
mixing and the subsequent suspension. 1 ml solvent phase was injected into an Agilent gas
chromatography mass spectrometer (GC/MS; GC, Agilent 7890A; MS, Agilent 5975C) equipped
with a fused silica capillary column (DB-5MS column:30 m × 320 µm × 0.25 µm). The carrier
gas was Helium at a flow rate of 1.3 ml/min. The splitter/injector was kept at 300oC with a
splitless mode. The oven temperature was programmed from 45°C to 270°C at ramping rate of
5°C /min. The solvent delay was held for 5 minutes.
5.3.5 Enzymatic hydrolysis on pretreated whole slurries
The whole slurries of pretreated samples enzymatically hydrolyzed in 0.05 M citric acid
buffer at pH 4.8 supplemented with 10mg/mL sodium azide, in which Novozymes Cellic®
CTec1 (220 mg protein/mL, preserve 200 mg glucose/mL, 205 FPU/mL) and Novozymes
Cellic® HTec2 (230 mg protein/mL, preserve 180 mg xylose/mL) at a ratio of 5:1. The
experiment was controlled at 50 °C for 72 h. The cellulase loading was 100 mg protein/ g
(glucan and xylan) to get maximum theoretical sugar recovery. In addition, 1% (w/v) bovine
serum albumin (BSA) was used to block lignin on enzymes.
5.3.6 2-D 1H -13C HSQC NMR of ball milled pine wood lignin and pretreatment obtained
RISL
50 mg recovered insoluble lignin (RISL) was dissolved in 600 μL deuterated DMSO
(Cambridge Isotope Laboratories). The resulting liquid sample was placed in 5 mm Wilmad 535-
PP NMR tubes. NMR spectra were collected at 25° C on 500 and 600 MHz Agilent (Varian)
Inova NMR spectrometers equipped with z-axis pulsed-field triple-resonance HNCP probes.
136
Samples contained 0.05% (v/v) TMS for chemical shift referencing. Two-dimensional 1H-13C
HSQC spectra of the aliphatic and aromatic regions were collected separately using the BioPack
gchsqc pulse sequence, with 1H spectral width of 17 ppm and 13C spectral widths of 100 or 60
ppm for the aliphatic or aromatic regions, respectively. Spectra were collected with 1024 points
(Varian parameter np) and 61 ms acquisition time with 128 or 256 transients and 128 or 96
complex points (Varian parameter ni in States-TPPI mode) in the indirect dimension, for
aliphatic and aromatic spectra, respectively. Adiabatic WURST decoupling was applied during
acquisition. Delayed times tCH and lambda for 1/4*JCH, were 1.8 ms and 1.6 ms for aliphatic
spectra, and 1.45 ms and 1.3 ms for aromatic spectra, respectively. Reference one-dimensional
1H spectra were collected with 32k points and 128 transients. HSQC spectra were processed and
analyzed with Felix 2007 (FelixNMR, Inc) or MestReNova 6.0.4 (Mestrelab Research), with
matched cosine-bell apodization in both dimensions, 2X zero filling in both dimensions, and
forward linear prediction of 30% more points in the indirect dimension. One-dimensional 1H
spectra were processed with no apodization or linear prediction and 2X zero filling. Relative
peak integrals were measured in MestReNova.
5.4 Results and discussion
5.4.1 pH changes of pretreatment solvent after pretreatment
The pH of pretreatment media is an important factor in pretreatment, which were reported to
influence biomass pretreatment sugar yields 315. Some efforts have been made on investigating
the effects of pH on biomass pretreatment, such as in the previous reviews on acidic, neutral and
alkaline conditions 316. For example, the fundamentals of pretreatment at higher and lower pHs
were reviewed 317, 318. The continuous pH monitoring during pretreatment was attempted in
137
pressure cooking water pretreatment of yellow poplar wood sawdust 38. Additionally, the
pretreatment of corn bran for enhanced enzymatic arabinoxylan degradation has been performed
at various pHs (1.7-9.8). Also, the catalyst types (different acids and alkalis) at various pHs in
mild thermal pretreatment were employed to pretreat wheat straw to achieve a high yield of
monomeric sugars and a high lignin removal 315. However, the comprehensive studies of
pretreatment at pHs ranged from specific 6 to 12 in elevated temperature (> 200 °C) was not
carried out, especially in flowthrough pretreatment of softwood. In specific, the detailed kinetic,
chemistry and products distribution were not well discussed in flowthrough pretreatment of
softwood at neutral and alkaline conditions.
138
Figure 5.1 pH of the pretreatment hydrolysates at different severities after the hydrolysates were
cooled down to room temperature, pretreatment conditions: pH 6.4-12, 200-270°C, and 0-10 min
In the flowthrough reactor system, the pretreatment solvent was pumped and passed through
the tubular reactor continuously. The pH of pretreatment media could be affected by the complex
biomass pretreatment reactions and different water dissociations at various heating temperatures.
However, there was no pH measuring method during pretreatment right at high temperatures.
The final pH values were measured after the pretreatment hydrolysates cooling down to room
temperature. It was found that the final pHs of pretreatment hydrolysates decreased to ~4.0 when
the initial pHs ranged from 6.4 to 9.0 regardless of the operating severities (Figure 5.1). This
pretreatment system acted like a buffering system which buffered pH to 4.0 when the initial pHs
were not too high (≤ 9.0). This can be caused by the following factors, such as the water self-
ionization at elevated temperatures (200-270 °C) 319, the biomass self-buffering system 320,
and/or a releasing of organic acids during pretreatment. Surprisingly, we found the initial pHs
were related to pretreatment chemical reactions. For example, the pretreatment reaction
pathways followed acid hydrolysis instead of alkaline degradation when initial pHs were ≤ 9.0.
At the initial pH of 11.0, the pHs slightly decreased but remained alkali conditions (pH > 10.0)
after pretreatment. When the initial pH was 12.0, the final pHs of hydrolysates presented no
significant changes at the different pretreatment severities. Probably, the high concentration of
NaOH might break the buffering system which led to strong alkaline conditions.
139
5.4.2 Hemicellulose removal and hemi-sugars recovery
Figure 5.2 Hemicellulose removal and hemi-sugars recovery in stage 1 (pretreatment only, pH
6.4-12, 200-270°C, and 0-10 min)
Hemicellulose is usually easier to be degraded than cellulose and lignin. In Figure 5.2a & b,
the hemicellulose removal could reach 100% when the initial pHs of pretreatment solutions were
≤ 9.0 in all tested pretreatment severities (4.0-6.2). This is probably because of the acid
hydrolysis of hemicellulose. Hemi-sugars can be recovered more than 90%. Only a portion of
hemi-sugars (<10%) was degraded when pretreatment severities were larger than 4.8. However,
when the pretreatment pHs were higher than 11.0 at the severities less than 5.2, hemicellulose
could not be completely removed (Figure 5.2a). Up to 50% of hemi-sugars could be recovered in
pH 11.0 but none was recovered in pH of 12 under tested pretreatment severities (Figure 5.2b).
Thus, at pHs 11.0 (severities>4.5) and 12.0 (all tested severities), hemi-sugars underwent severe
degradation reactions resulting little hemi-sugar yields. The low hemi-sugars recovery yield in
pH of 11.0 was because of the degradation of hemi-sugars 71. At pH 12.0, no hemi-sugar was
observed, which might due to the direct and fast degradation of hemicellulose to small fractions.
140
5.4.3 Cellulose removal and glucose recovery
Figure 5.3 Cellulose removal and glucose recovery (Stage 1: pretreatment; Stage 2: enzymatic
hydrolysis), pretreatment conditions: pH 6.4-12, 200-270°C, and 0-10 min
In Figure 5.3a, when the initial pHs were less than 9.0, glucan removal decreased slightly
from pH 6.4 to 9.0. When the initial pHs were higher than 11.0, However, the glucan removal
increased with pH increasing. This was maybe attributed to pretreatment reactions under initial
dilute alkali (pH less than 9.0); the reactions were H+ dominant acid hydrolysis indicating by the
final pHs (Figure 5.1). The slight difference of final pHs (initial 6.4-9.0) resulted in slightly
difference of glucan removal. However, high pH (≥ 11.0) broke the buffering range which led to
141
the alkaline hydrolysis of biomass. It was reported that under alkaline conditions, carbohydrates
underwent both hydrolysis and peeling off reactions 48. The peeling off reactions of
carbohydrates increased with the increase of NaOH concentration, which might be the main
reason to resulting in high glucan removal. When pretreatment severities were higher than 5.2 at
pH of 12.0, glucan was abruptly removed at 100% which was because of the severe peeling off at
very high temperature and relatively strong alkalis.
Total glucose recovery (stage 1 + stage 2) included glucose yield obtained from both the
pretreatment and enzymatic hydrolysis processes. In Figure 5.3, the glucose recovery in stage 1
and stage (1+2) was similar when initial pHs were less than 9.0 (e.g. 6.4, 8, and 9), and the total
glucose yield can be higher than 80%. It was indicated that the similar glucose yield was because
of the similarities of the final pH of pretreatment hydrolysates at room temperature. The total
glucose yields at the initial pHs of 11.0 and 12.0 were comparable with others under severities
less than 4.8. However, at severities larger than 4.8, the glucose yield was much lower than
others presenting less than 50% (Figure 5.3b). The difference between the high cellulose removal
and low glucose recovery at pH of 11.0 and 12.0 might be the degradation of cellulose or glucose
during pretreatment. Moreover, these alkaline degradation products might inhibit enzymatic
hydrolysis sugar yields. In the pretreatment stage only (Figure 5.3c), cellulose was recovered as
monomeric and oligomeric sugars with similar and high yields when the initial pHs were less
than 9.0. However, at pH 11.0, less than 10% of glucose was observed, and no glucose was
obtained at pH 12.0 in pretreatment stage. Thus, these results showed the possibilities of severe
degradation of carbohydrates in pH 11.0-12.0. The initial pHs of pretreatment hydrolysates were
proposed to be correlated with these tremendously different pathways.
142
5.4.4 Lignin removal and recovered soluble lignin yield
Figure 5.4 Lignin removal and recovered soluble lignin yield, pretreatment conditions: pH 6.4-
12, of 200-270°C, and 0-10 min
Lignin can be recovered from pretreatment hydrolysates as recovered soluble and insoluble
phases. Lignin structural changes in acids and alkalis were different. In this study, lignin removal
showed no significant increase when the initial pHs increased from 6.4 to 11.0 at the same
severities (Figure 5.4a). When the initial pH reached 12.0 at pretreatment severities lower than
4.5, lignin removal was close to that at pH 11.0 under the same pretreatment severities. Probably,
the alkali stable lignin structure in softwood lignin limited its solubilization in alkaline
conditions. However, when the pretreatment severities were increased to higher than 4.5, lignin
was removed significantly until 100% of lignin removal. The initial pHs presented the similar
influence on changes of recovered soluble lignin recovery (Figure 5.4b). The increasing of initial
pHs and/or pretreatment severities improved lignin solubilization. It was reported that lignin
depolymerization mechanism under acidic conditions was mainly the cleavage of β-O-4 linkages
68. Nevertheless, lignin degradation under alkaline conditions first induced the generation of -
Quinone methide intermediates, which led to the subsequent cleavage of C-O-C linkages and
143
even ring open reactions 321, 322. The different mechanisms of lignin acidic and alkaline
degradations might cause the different lignin solubilization and removal in pretreatments at pHs
ranged from 6.4 to 12.0. It was observed in Figure 5.4a & b, the alkaline degradation of lignin
removed more lignin than in the acidic depolymerization of lignin. Additionally, the alkaline
degradation of lignin might lead to ring open reactions 321, 322. Thus, the different initial pHs of
pretreatment media can result in various lignin degradation mechanisms.
In summary, the results showed at initial pHs 6.4-9.0, the pretreatment reactions followed
the acid hydrolysis kinetics while presented alkaline degradation mechanisms at pH 11.0 and
12.0. These results were consistent with the final pHs of pretreatment hydrolysates which have a
close relation with pretreatment reactions mechanism (Figure 5.1).
5.5 Determination of the degradation products from carbohydrates and lignin by GC/MS and
HPLC
The determination of the degradation products in pretreatment hydrolysates derived from pH
11.0 and 12.0 was really difficult. It was because the carbohydrates were degraded to a mixture
of small fragments of organics. The characterization method of this mixture has not been
reported yet and the challenges were reported that alkaline pretreated liquor is a complicated
mixture of acids (derived from carbohydrate degradation reactions that occur in alkaline
conditions), partially depolymerized polysaccharides, monosaccharides, aromatic monomers
(derived from lignin), high molecular weight lignin, and acetate. At present, no single method
exists to completely identify and quantify all components present in the liquor 310, 323. In our
study, we also faced the same challenges so that the study could not provide a comprehensive
analysis of products from alkaline degradation of carbohydrates.
144
Firstly, the HPLC was applied to detect possible degradation products from carbohydrates at
pH 11.0-12.0. Only glycolic acid, acetic acid, and formic acid were detected (Figure S5.1), which
might be from alkaline degradation of carbohydrates. Besides, a trace furfural was detected when
the initial pHs ranged 6.4 to 9.0 at high pretreatment severities. Figure 5.5 showed some
monomers derived from lignin alkaline degradation in some selected hydrolysates. These
compounds were recovered in soluble lignin phase. The vanillin was found in lignin degradation
products under alkaline conditions, which was originated from lignin oxidations. Also, the
compounds similar to vanillin structure were detected (compound 8-12). The yield of recovered
soluble lignin can be achieved more than 50% of maximum lignin recovery (Figure 5.4b). These
monomeric lignin aromatics can play a significant role in producing value added aromatics.
Figure 5.5 GC/MS results of the pretreatment hydrolysates collected at pH of 12 and
pretreatment severity of 5.2
145
5.4.5 2-D 1H -1C HSQC NMR of ball milled pine wood lignin and pretreatment obtained
RISL
Note: the assignments and NMR detected lignin linkages were shown in supplementary materials
(Table S5.1 and Figure S5.2 60).
Figure 5.6 2-D 1H-13C HSQC NMR spectra of ball milled pine lignin and pretreatment obtained
RISLs at 11.0 and 12.0. (a,b) ball milled pine wood lignin, (c,d) RISL derived from pH of 11.0
and pretreatment severity of 6.1, (e,f) RISL derived from the pH of 12.0 and pretreatment
severity of 5.8
The cleavage of lignin C-O-C linkages is important in lignin solubilization under alkaline
solutions. The quantification of the basic linkages referred to the reported method 157, 324. In this
study, lignin linkages such as β-O-4, β-β and β-5 were cleaved at pH 11.0-12.0 (Figure 5.6a, c, &
d; Table 5.1). The table 5.1 showed that the cleavage of β-5 was slower than the cleavages of β-
O-4 and β-β. The cleavages of these linkages were increased from pH of 11.0 to 12.0 resulting in
high lignin solubility in liquid phase. At pH 12.0, the β-β and β-5 was not presented (Figure 5.6e,
146
Table 5.1). The aromatic regions of the two RISL showed the majority aromatic rings remaining
even in this high tested pretreatment severity (5.8) at pH of 12.0 (Figure 5.6b, d & f). The
majority of lignin ether and C-C bonds were cleaved as the increasing of strong alkalis. The
quantification of each linkage is from the volume-integration of cross-peak contours in HSQC
spectra. The oxidized G units were completely disrupted during pretreatment and not seen in 2-D
NMR spectra of RISLs. The peak integrals of cinnamaldehyde and p-hydroxybenzoate found
decreasing.
Table 5.1 Distribution of the detected linkages of ball milled pine lignin and RISLs. BML: ball
milled wood lignin; A: RISL collected at pH of 11.0 and pretreatment severity of 6.1; B: RISL
collected at pH of 12 and pretreatment severity of 5.8. The percentages of inter linkages of BML
were based on all detected peaks
Inter linkages (%) β-O-4 Resinol
(β-β)
Phenylcoumarane
(β-5, α-O-4)
BML 77.5 5.4 17.1
A 58.36 2.74 38.90
B 100.00 Gone Gone
149
Figure 5.7 Proposed biomass degradation mechanisms of softwood in aqueous flowthrough
pretreatment at initial pH 6.4-12, (a) hemicellulose; (b) cellulose; (c) lignin 60, 67, 68, 71-74, 321, 322
The mechanisms of aqueous flowthrough pretreatment at neutral-12 pH on softwood were
proposed in Figure 5.7 to suggest the impact of initial pHs. The proposed chemical products
include some of our detected compounds in this study. The hemicellulose and cellulose was
proposed to experience an acid hydrolysis and dehydration in the aqueous pretreatment at pH
6.4-9.0 which produced sugars and subsequent limited sugars degradation if high severities were
applied. At the initial pH of 11.0, the final pH decreased to the alkali (pH > 10.0) after
pretreatment. Thus, the degradation of these conditions was proposed to follow a moderate alkali
kinetics which most of biomass solid was preserved while alkaline degradation of sugars
occurred at high severities (>5.2). Additionally, it was reported that the carbohydrates
150
isomerization, dehydration, fragmentation, and condensation can occur in forming a mixture of
sugars, aldehydes, acids and humins in the water media at elevated temperature 66. However,
when pH increased to 12.0, the carbohydrates were degraded in fast alkaline reaction pathways
and none sugars were detected in products. The carbohydrates in cellulosic biomass can undergo
severe degradation reactions such as the peeling off reactions to a mixture of carboxylic acids as
well as the isomerization reactions to a mixture of different sugars 67. These mechanisms make
biomass chemistry complex and attractive in producing an abundant of bio-chemicals. However,
the study encountered difficulties in the characterization of the alkali pretreated liquor. Thus, the
comprehensive determination of products in the hydrolysates derived from aqueous pretreatment
at pH 12 is still not resolved, which needs more future work. Furthermore, the alkaline
delignification chemistry has been discussed and reviewed before 71. In the study, when pH was
6.4-9.0, the lignin was proposed to be depolymerized through the cleavage of C-O-C bonds. At
pH of 11.0 and 12.0, our 2D HSQC 1H-13C NMR results indicated that the main mechanism of
lignin depolymerization was still the cleavage of β-O-4. Under pH 12.0, all β-β and β-5/α-O-4
were disappeared while some β-O-4 remained although β-O-4 showed the fastest degradation
rate. Besides, although the ring open reactions were reported to occur, our results suggested the
substantial aromatic rings remained and no significant changes were observed. In summary, the
proposed pathways of the aqueous pretreatment of softwood were greatly affected by the initial
pHs applied. However, more future efforts still need to be made to investigate the comprehensive
distribution of aromatic chemicals in alkali pretreated liquor. Our study gained more
understandings of the application of pretreatment pHs and help in design wise production of
aromatic chemicals.
151
5.5 Conclusion
The study demonstrated several highlights include a) the aqueous flowthrough pretreatment
of softwood was studied for the first time; b) the detailed pretreatment mechanisms was proposed
in neutral and alkaline pHs; c) the biomass solubilization was achieved at pH 12 with a mixture
of major carboxylic acids and aromatic chemicals. However, more work is needed to
characterize this mixture of compounds. To be specific, the initial pretreatment pH has
significant effects on sugar yields, lignin solubilization, and pretreatment kinetics. When the
initial pHs were 6.4-9.0, the biomass degradation was indicated to follow the acid hydrolysis
suggested by the final pH of ~4.0 at room temperature after pretreatment terminated if the initial
pHs were 6.4-9.0. This buffering capacity of aqueous flowthrough pretreatment system was
proposed to originate from the water self-ionization at elevated temperatures, the biomass self-
buffering system and/or a releasing of organic acids during pretreatment. However, when the
initial pHs were 11.0 and 12.0, the biomass degradation pathways proceeded with an moderate
and strong alkaline hydrolysis under applied pretreatment severities (3.8-6.0), respectively.
Besides, the 2-D 13C-1H HSQC NMR analysis showed the majority of C-O-C cleaved during the
alkaline degradation of lignin, resulting the increasing of lignin solubilization, while most of
aromatic rings still remained but the β-O-4, resinol, and phenylcoumarane linkages were
completely degraded at pH of 12.
5.6 Acknowledgements
The author acknowledges the Sun Grant-DOT Award, # T0013G-A- Task 8 for supporting
this work. Also, the author thanks Department of Biological Systems Engineering at Washington
State University. The author thanks Mr. Gildardo Soto for his assistance in conducting some
152
experiments. The thesis author appreciated Drs. Yucai He, John R. Cort, and Bin Yang for their
insightful discussions.
5.7 Supplementary materials
`
Figure S5.1 The HPLC analysis of degradation products in pretreatment hydrolysates at pH of
12.0 and pretreatment severity of 6.1
153
Table S5.1 Main lignin 2-D 1H -13C Cross-peaks assignments in the HSQC Spectra
Lignin linkages and monolignols Chemical shift
-OCH3 55.47(C) 3.70(H)
A: β-O-4 71.70(S-Cα) 4.84(S-Hα) 59.50-59.70(Cγ) 3.40-3.63(Hγ) 85.90(S- Cβ)
4.09(S-Hβ) 83.49(G/H-Cβ) 4.28(G/H-Hβ)
B: Resinol 84.85(Cα) 4.62(Hα) 53.30(Cβ) 3.46(Hβ) 70.85(Cγ) 4.14/3.78(Hγ)
C: Phenylcoumaran 86.79(Cα) 5.41(Hα) 53.3(Cβ) 3.05(Hβ) 62.52(Cγ) 3.68(Hγ)
D: Spirodienone 59.67(Cβ) 3.19(Hβ)
G: Guaiacyl 111.02(C2) 6.95(H2) 115.05(G5) 6.74(H5) 119.01(G6) 6.78(H6)
G: Oxidized (Cα=O) guaiacyl 111.56(C2) 7.50(H2) 123.55(C6) 7.54(H6)
H: p-hydroxyphenyl 128.0(C2/6) 7.2 (H2/6)
E: p-hydroxybenzoate 131.33(C2/6) 7.62(H2/6)
K: Cinnamaldehyde 126.23(Cβ) 6.74(Hβ)
Note: G, S, H-C or G, S, H-H refers to C and H in the lignin sub-units, guaiacyl, syringyl and p-
hydroxyphenyl
155
CHAPTER SIX
CHARACTERIZATION OF LIGNIN DERIVED FROM WATER-ONLY
AND DILUTE ACID FLOWTHROUGH PRETREATMENT OF
POPLAR WOOD AT ELEVATED TEMPERATURES
6.1 Abstract
Flowthrough pretreatment of biomass is a critical step in lignin valorization via conversion
of lignin derivatives to high-value products, a function vital to the economic efficiency of
biorefinery plants. Comprehensive understanding of lignin behaviors and solubilization
chemistry in aqueous pretreatment such as water-only and dilute acid flowthrough pretreatment
is of fundamental importance to achieve the goal of providing flexible platform for lignin
utilization. In this study, the effects of flowthrough pretreatment conditions on lignin separation
from poplar wood were reported as well as the characteristics of three sub-sets of lignin
produced from the pretreatment, including residual lignin in pretreated solid residues (ReL),
recovered insoluble lignin in pretreated liquid (RISL), and recovered soluble lignin in
pretreatment liquid (RSL). Both the water-only and 0.05 % (w/w) sulfuric acid pretreatments
were performed at temperatures from 160 to 270 °C on poplar wood in a flowthrough reactor
system for 2–10 min. Results showed that water-only flowthrough pretreatment primarily
removed syringyl (S units). Increased temperature and/or the addition of sulfuric acid enhanced
the removal of guaiacyl (G units) compared to water-only pretreatments at lower temperatures,
resulting in nearly complete removal of lignin from the biomass. Results also suggested that
more RISL was recovered than ReL and RSL in both dilute acid and water-only flowthrough
pretreatments at elevated temperatures. NMR spectra of the RISL revealed significant β-O-4
156
cleavage, α-β deoxygenation to form cinnamyl-like end groups, and slight β-5 repolymerization
in both water-only and dilute acid flowthrough pretreatments. Elevated temperature and/or dilute
acid greatly enhanced lignin removal to almost 100 % by improving G unit removal besides S
unit removal in flowthrough system. Only mild lignin structural modification was caused by
flowthrough pretreatment. A lignin transformation pathway was proposed to explain the
complexity of the lignin structural changes during hot water and dilute acid flowthrough
pretreatment.
Keywords: Aqueous • Dilute acid • Flowthrough pretreatment • Poplar • Lignin •
Characterization
6.2 Introduction
Lignin is a main constituent of lignocellulosic biomass (15-30% by weight, up to 40% by
energy) and the second most abundant biopolymer on Earth. Recently, lignin has received broad
attention as its intermediates have found wide use in various carbon products, such as electrodes,
carbon fibers, jet fuels, biochemicals, plastics, and antioxidants 87, 325-330. Thus, the utilization of
waste lignin in pretreatment as a feedstock for conversion to fuels and chemicals offers a
promising opportunity for enhancing the overall operational efficiency, carbon conversion,
economic viability, and sustainability of biorefineries.
Lignin is an amorphous, cross-linked biopolymer that consists of three phenylpropanoid
units, namely, p-hydroxyphenyl (H units), guaiacyl (G units), and syringyl (S units). These units
are derived from the three monolignol building blocks: p-coumaryl, coniferyl, and sinapyl
alcohols. The major inter-unit linkages in lignin are β-O-4, α-O-4, β-5, β-β, 5-5, β-1 and 4-O-5
331. Selective depolymerization of lignin is important to lignin utilization and also considered to
157
be very challenging due to the broad distribution of bond energies in the various C-O and C-C
linkages within the structure of lignin and the tendency for uncontrollable thermal and catalytic
fragmentation when trying to cleave them 87. Most lignin extraction processes usually cause
many lignin structural changes and modifications. As removing lignin is favorable to the process
of enzymatic hydrolysis for sugar conversion, great efforts have been made to boost lignin
extraction from the solid biomass to the liquid phase during the aqueous pretreatment 332. Most
water-only and dilute acid batch pretreatments were reported to achieve limited lignin removal
333, 334 dictated by the lignin depolymerization and condensation chemistry 68, 335, 336. Under acidic
pretreatment conditions, the predominant reactions associated with lignin are fragmentation by
acidolysis of aryl ether linkages and acid catalyzed recondensation, while linkages like resinol
and phenylcoumaran subunits are fairly stable. The β-O-4 linkages in lignin are susceptible to
acidic hydrolysis and the pretreatments generally result in their lower relative content in the
pretreated biomass 337. The dilute acid pretreatment led to an increase of phenolic OH groups in
lignin apparently resulting from cleavage of aryl ether linkages 100, 182, 338. In addition, lignin in
different biomass species has various characteristics, which influence pretreatment sugar yield
and enzymatic hydrolysis effectiveness. It was reported that sugar yield was related to lignin
content and structural characteristics, especially S/G ratio in the biomass substrate 339. Also,
changes in cellulose structure were influenced by lignin content 340. Biomass genetic
modification and manipulation were performed to reduce lignin recalcitrance, modify S/G ratio
and improve sugar yields 341-344. However, biomass pretreatment is still challenging for high
sugar recovery with maximum usable lignin recovery since sugars are easily degraded under
high pretreatment severities. High yields of sugars and lignin are critical to achieve favorable
158
pretreatment economics 345.
Sugar yields and removal of lignin by pretreatment can be greatly improved by flowing hot
water through the biomass 345, 346. Use of a flowthrough reactor system was reported to increase
lignin removal as much as 98% 345, and the addition of dilute acid further increased the removal
extent 53, 56, 345, 347. The main advantage of a flowthrough reactor system is its ability to restrict
condensation reactions by constantly removing lignin into the aqueous phase and reducing the
chance of pseudo-lignin formation 348, 349 and lignin deposition 350. In addition, pretreatment was
shown to mitigate lignin droplets on the cellulose surface 107 thus it led to effective lignin
deconstruction and hemicellulose recovery 59, 351. In pretreatment, lignin is believed to
depolymerize via both homolytic and acidolytic cleavage into low molecular weight lignin
globules 337, 352. Thus, solubilized lignin derivatives continuously exit the flowthrough system,
thus further thermal or chemical treatments on depolymerized lignin fractions can be avoided. In
this case, flowthrough-derived lignin is believed to only exhibit mild structural change compared
to the native original lignin. Effective recovery of sugars and lignin with whole biomass
solubilization by flowthrough pretreatment at elevated temperatures under tested pretreatment
conditions was previously reported 345. However, application of lignin derivatives to produce
added value products cannot be realized unless the corresponding lignin recovery yield and
chemistry under these pretreatment conditions are well understood. In this study, the three lignin
sub-sets, including residual lignin in pretreated solid residues (ReL), the purity of recovered
insoluble lignin in pretreated liquid (RISL) and recovered soluble lignin in pretreatment liquid
(RSL) from flowthrough pretreatment were measured following acetyl bromide treatment, and
their structure was analyzed using Fourier transform infrared (FTIR) spectroscopy, GC/MS,
159
pyrolysis GC/MS, and two-dimensional (2-D) 1H-13C heteronuclear single-quantum coherence
(HSQC) NMR spectroscopy. Structural characteristics of three lignin isolates were investigated
in order to reveal intrinsic structural modification of lignin under the tested pretreatment
conditions, which will benefit the utilization of these lignin sources in the context of a
biorefinery dependent on high yields of sugars.
6.3 Materials and methods
6.3.1 Materials
Hybrid poplar wood chips were kindly provided by the Forest Concepts Corporation, located
in Auburn, Washington. The chips were air-dried and then milled to 0.4-0.8 mm. The resulting
poplar particles were stored at -20 ℃ for experimental use with their moisture content at about
5%. Compositions of poplar wood were determined using the NREL Laboratory Analytical
Procedure (LAP; NREL, 2008) 353. Poplar wood contains 48.5±0.6% glucan, 17.2±0.5% xylan,
and 23.4±0.4% lignin. Ball-milled poplar lignin was kindly provided by Georgia Institute of
Technology, Atlanta, Georgia.
6.3.2 Flowthrough pretreatment of poplar wood
Poplar chips (0.5 g dry weight) were loaded into a tubular reactor (1.3 cm i.d. ×15.2 cm
length with an internal volume of 20.5 mL). Two silver-plated snubber gaskets with 5 µm were
applied in both ends of the tubular reactor. The reactor was then connected to our flowthrough
system. The flowthrough system was equipped with a 4-kW fluidized sand bath (model SBL-2D,
Omega Engineering, Inc., Stamford, CT). Distilled water or 0.05% (w/w) sulfuric acid at room
temperature was pumped through a preheated coil in sand bath and then passed through the
reactor continuously. The pretreatment hydrolysates were quenched by submerging the coil in ice
160
water. The pretreatment was performed at 160 °C to 280 °C and the temperature was monitored
by a thermometer (Omega Engineering Co., Stamford, CT) located in the reactor outlet. The
back pressure was regulated by a pressure gauge in the range of 10-70 bar, which corresponded
to the saturated steam pressures. The sand bath temperature was controlled at 5-10°C above the
reaction temperature using a temperature controller (Omega, HH501AJK), with consideration of
heat loss and transfer. The flow rate was set at 25 mL/min through the high-pressure pump
(Acuflow Series III Pumps, Fisher, and Pittsburgh, PA, USA). After the pretreatment was
completed, the reactor was immediately transferred to ice water. The undissolved poplar solid
residues were collected and freeze-dried for analysis. In this study, the pretreatment severity
factor was applied to evaluate pretreatment conditions (equation 6.1).
Log R0= Log t*exp (TH−TR
14.75) (6.1)
Where 𝑡 is the reaction time in minutes, 𝑇𝐻 is the reaction temperature in °C, 𝑇𝑅 is the reference
temperature in °C (100°C) 354.
6.3.3 Composition analysis of the solid residues and determination of lignin removal
Composition analysis of solid residues was carried out by the NREL procedure 353. 0.03g of
solid residues underwent 72% sulfuric acid hydrolysis in a 30°C water bath for one hour,
followed by 4% sulfuric acid in autoclave with pressure at 121°C for one hour. The resulting
slurries were filtrated and dried in the 105°C oven. According to this method, the remaining
oven-dried solids were ReL, known as acid insoluble lignin. Lignin removal refers to the
percentage of original lignin released to aqueous phase (equation 6.2). This parameter presents
the effectiveness of delignification by pretreatment.
Lignin removal % = 100 −mRe
mL × 10 (6.2)
161
Where mRe refers to the dry weight mass of remaining lignin after composition analysis of solid
residues; mL represents the mass weight of lignin in the original loaded biomass (mL= the loaded
biomass mass weight×the lignin content obtained from section 6.3.1).
6.3.4 Ultraviolet visible spectroscopy (UV-Vis) quantification of RSL
The RSL content (%) was determined by the UV absorbance measurement of the pretreated
hydrolysates at 320 nm according to the NREL procedure 128. RSL was calculated according to
the following equations (equation 6.3, 6.4).
RSL% =UVabs×Volumepretreatment liquor×Dilution
ε×ODWsample×Pathlength× 100 (6.3)
Where 𝑈𝑉𝑎𝑏𝑠 is the average UV-Vis absorbance at 320 nm; Volumepretreatment liquor refers to
the volume of pretreatment hydrolysate; ε represents molar absorptivity of biomass at 320 nm
(30 L/g cm); ODWsample is the weight of sample in milligrams; Pathlength is the pathlength of
UV-Vis cell in cm.
Dilution =Volumesample+Volumediluting solvent
Volumesample (6.4)
6.3.5 Recovered insoluble lignin (RISL) isolation and purity measurement
The pH of some collected pretreatment hydrolysate samples that were not at pH 2-3, was
adjusted to pH 2-3 to precipitate the RISL. The resulting precipitates were freeze-dried to obtain
dry, solid RISL 325. The RISL yield was calculated according to equation 6.5.
RISL%=mp
mL × 100 (6.5)
Where, mp is the mass weight of precipitated lignin; 𝑚𝐿 represents the mass weight of the lignin
in the original loaded biomass (mL = the loaded biomass mass weight × the lignin content
calculated in section 6.3.1).
162
The RISL purity analysis was performed by determining the ultraviolet absorbance at 280
nm after subjecting the lignin sample to acetyl bromide (AcBr method), followed by acetic acid
treatments 355. The purity was calculated by using the equation of Morrison (equation 6.6).
AcBr lignin = (3.37 × Uabs
CL− 1.05) × 100 (6.6)
Uabs refers to the UV absorbance at 280 nm; CL is the lignin concentration before sending to UV
measurement.
6.3.6 Gel permeation chromatography analysis of molecular weight of flowthrough lignin
The molecular weight analysis of the flowthrough derived lignin was carried out using gel
permeation chromatography (GPC). Lignin samples were individually acetylated by stirring with
2 ml of acetic anhydride– pyridine (1/1, v/v) at room temperature for 24 hours. After acetylation,
the acetylated lignin samples were dissolved in THF for GPC analysis using an Agilent 1200 LC
equipped with an ultraviolet (UV) detector. The sample was filtered through a 0.45 μm
membrane filter prior to injection of a 20 μl sample. GPC analyses were carried out using a UV
detector (280 nm) on a 4-column sequence of WatersTM Styragel columns (HR0.5, HR2, HR4
and HR6) at 1.00 ml min−1 flow rate. Polystyrene standards were used for calibration. WinGPC
Unity software (Version 7.2.1, Polymer Standards Service USA, Inc.) was used to collect data
and determine molecular weight profiles 325.
6.3.7 Py-GC/MS analysis of ReL
Py-GC/MS was performed with a CDS 5000 pyrolysis autosampler (Oxford, PA, USA)
attached to a Thermo Trace GC (6890N/MSD, 5975B) gas chromatography/mass spectrometry
system (Bellevue, WA, USA). Approximately 0.5 mg of the solid residues was loaded into a
quartz tube that was filled with a thin layer of quartz wool in advance. The initial temperature
163
performed in the process was 250°C for 6 seconds and then the sample was pyrolyzed up to a
temperature of 610°C within 1 min. The generated vapor was separated by a 30 m 0.25 μm inner
diameter (5% phenyl)-methylpolysiloxane column. The pyrolyzed gas was held in the pyrolysis
oven for 56 min and then sent to the mass spectrometer that was operated in EI mode (70 ev) for
analysis.
6.3.8 Fourier transformed infrared (FTIR) spectroscopic analysis of ReL
The spectra (4500-800 cm-1) were obtained with a Bruker IFS 66v/s spectrometer equipped
with an IR-microscope using about 2mg of each sample. Spectra were obtained using the
triangular apodization with a resolution of 4 cm-1 and an interval of 1 cm-1. 64 scans were
conducted for each background and the sample spectra.
6.3.9 Characterization of RSL after flowthrough pretreatment by GC/MS
10 mL of diethyl ether was used to extract RSL derivatives in 5 mL of hydrolysates,
following vigorous mixing and subsequent suspension and separation. A 1mL aliquot of the
solvent phase was injected into an Agilent gas chromatography mass spectrometer (GC/MS; GC,
Agilent 7890A; MS, Agilent 5975C) equipped with a fused silica capillary column (DB-5MS
column: 30m × 320µm × 0.25µm). The carrier gas was helium at a flow rate of 1.3 mL/min. The
splitter/injector was kept at 300°̊C with a split ratio of 5:1. The oven temperature was
programmed from 100 °C to 270 °C at a ramping rate of 5 °C /min. Both the initial and final
temperatures were held for 5 minutes.
6.3.10 RISL structural characterization by 2-D 1H -1C NMR HSQC NMR
50 mg recovered insoluble lignin (RISL) was dissolved in 600 μL deuterated DMSO
(Cambridge Isotope Laboratories). The resulting liquid sample was placed in 5 mm Wilmad 535-
164
PP NMR tubes. NMR spectra were collected at 25° C on 500 and 600 MHz Agilent (Varian)
Inova NMR spectrometers equipped with z-axis pulsed-field triple-resonance HNCP probes.
Samples contained 0.05% (v/v) TMS for chemical shift referencing. Two-dimensional 1H-13C
HSQC spectra of the aliphatic and aromatic regions were collected separately using the BioPack
gchsqc pulse sequence, with 1H spectral width of 17 ppm and 13C spectral widths of 100 or 60
ppm for the aliphatic or aromatic regions, respectively. Spectra were collected with 1024 points
(Varian parameter np) and 61 ms acquisition time with 128 or 256 transients and 128 or 96
complex points (Varian parameter ni in States-TPPI mode) in the indirect dimension, for
aliphatic and aromatic spectra, respectively. Adiabatic WURST decoupling was applied during
acquisition. Delayed times tCH and lambda for 1/4*JCH, were 1.8 ms and 1.6 ms for aliphatic
spectra, and 1.45 ms and 1.3 ms for aromatic spectra, respectively. Reference one-dimensional
1H spectra were collected with 32k points and 128 transients. HSQC spectra were processed and
analyzed with Felix 2007 (FelixNMR, Inc) or MestReNova 6.0.4 (Mestrelab Research), with
matched cosine-bell apodization in both dimensions, 2X zero filling in both dimensions, and
forward linear prediction of 30% more points in the indirect dimension. One-dimensional 1H
spectra were processed with no apodization or linear prediction and 2X zero filling. Relative
peak integrals were measured in MestReNova.
6.4 Results and discussion
In this study, poplar carbohydrates were mostly recovered as sugars under tested
pretreatment conditions 48, 59. Results showed that flowthrough pretreatment at elevated
temperatures enhanced sugar recovery to over 90% and limited degradation loss. This chapter
focuses on understanding lignin recovery and structural characteristics are the recovered lignin,
165
which are crucial for conversion to high-value products. To characterize the recovery of the three
lignin sub-sets, a comprehensive process was followed as summarized in Figure 6.1. Recovered
insoluble lignin was collected by settling pretreatment liquid at pH 2-3 overnight to precipitate
solid lignin at the bottom followed by fast filtration to collect lignin solids, avoiding filter paper
clogging.
Figure 6.1 Scheme for lignin recovery and analytical analysis in this study
6.4.1 Lignin removal and recovery in aqueous phase
Lignin in its pretreated aqueous phase was recovered as the RSL and RISL. Lignin recovery
by flowthrough pretreatment increased as the pretreatment severity increased (LogR0) under both
water-only and dilute acid conditions (Figure 6.2). The pretreatment severity (LogR0) was used
to evaluate the severity of the pretreatment process by combining two variables, including the
166
temperature and reaction time (Equation 6.1 in the method section). Nearly 100% of lignin was
released into the aqueous phase at around LogR0=6.0 in water-only pretreatment or at LogR0=5.0
in dilute acid pretreatment. Under such conditions, it was previously reported that nearly whole
biomass was solubilized into the liquid phase and carbohydrates were almost completely
recovered as monomeric and oligomeric sugars 345.
Figure 6.2 Lignin removal and recovery as RSL and RISL. (a) water-only and (b) 0.05% (w/w)
H2SO4 flowthrough pretreatment (temperature: 160-270 °C, time: 0-12min, flow rate: 25mL/min)
Dilute acid was found to be more effective for lignin recovery than water-only pretreatment
at the same severities (Figure 6.2). Addition of 0.05% (w/w) sulfuric acid enhanced
delignification and solubilization into the aqueous phase. Considering that the pH of the water at
220 °C was 5.5, and that it dropped to 3.2-4.0 at temperatures higher than 220°C 48, 59, water-only
pretreatment at increased temperatures is similar to acid-catalyzed pretreatments. Different RSL
and RISL yields were observed in water-only and dilute acid pretreatments (Figure 6.2a, b). The
RSL yields were less than 20% and 5% in water-only pretreatment and dilute acid pretreatment,
respectively. Higher yields of RSL were found in water-only than in dilute acid pretreatment at
the same LogR0. In addition, the dilute acid-derived RSL was observed to gradually increase
(a) (b)
167
with increasing LogR0, followed by a slight decline at about LogR0=4.6, which differed from the
climbing trend of RSL in water-only pretreatment. These suggested that reactions shifted from
lignin depolymerization to counterproductive RSL condensation or repolymerization in dilute
acid pretreatment, thereby limiting the removal of depolymerized lignin derivatives. Since
flowthrough pretreatment possibly limited the number of condensation reactions, little decrease
in RSL yield was observed (Figure 6.2b). More than 60% of the recovered RISL was found in
flowthrough pretreatment at LogR0>4.0, and dilute acid addition further improved RISL
formation. The pretreatment severities at ~6.0 and ~5.0 appeared to be the most promising
conditions tested to produce high yield of RISL as well as high sugar yields by water-only and
dilute acid pretreatments, respectively 345. The lignin structural analysis in this study mainly
focused on these two conditions.
168
6.4.2 Lignin purity determination of the isolated RISL
Figure 6.3 The RISL purity determined by AcBr method. (a) water-only; (b) 0.05% (w/w)
H2SO4 at the flow rate of 25 ml/min
The flowthrough pretreatment hydrolysates contained a mixture of degradation products
from carbohydrates and lignin derivatives. The RISL was isolated by precipitating hydrolysates
at pH 2-3, followed by filtration, washing, and freeze-drying. The RISL purity was found to be
higher than 50% in both water-only and dilute acid pretreatment hydrolysates under the tested
conditions (Figure 6.3). The highest purity of RISL was achieved at about 80% to 90% in water-
only and dilute acid pretreatments, respectively. RISL pretreated using dilute acid was found to
have higher purity than that in water-only pretreatment. This was because most carbohydrates
(a)
(b)
169
were hydrolyzed into monomeric sugars and washed away in dilute acid flowthrough
pretreatment 345.
RISL recovered at high temperatures (240°C and 270°C for 0.05% wt sulfuric acid and hot
water pretreatment, respectively) were treated with pyridine/acetic anhydride. THF was used as
eluent. Individual lignin samples were dissolved in THF. GPC analysis was carried out in Agilent
GPC system with UV detector. The set flow rate is 1 ml/min. Polystyrene was used for
calibration curve. The number average molecular weight (Mn) and the weight average molecular
weight (Mw) for lignin obtained under 240°C, residence time of 10mins, 0.05% sulfuric acid and
flow rate of 25ml/min were 1083 and 1955 respectively. However, Mn and Mw for lignin
obtained under 270°C, residence time of 10mins, water-only and flow rate of 25ml/min were
1197 and 2661, respectively (Figure S6.2). The polydispersity (D; M̅w/M̅n) of lignin derived
from dilute acid condition (240°C, 10mins, 0.05% sulfuric acid and flow rate of 25ml/min,) and
hot water condition (270°C, 10mins, water-only and flow rate of 25ml/min) was 1.81 and 2.22,
respectively. The molecular weight distribution provides evidence of low molecular weight
nature of our lignins, which offers great opportunities in lignin utilization.
170
6.4.3 Py-GC/MS spectroscopic analysis of untreated poplar wood and pretreated solid
residues
Figure 6.4 Py-GC/MS results of untreated poplar wood (control) and solid residues after
flowthrough pretreatments. (a) Untreated poplar wood and solid residues pretreated at 240 °C,
0.05% (w/w) H2SO4 for 10 min (pretreatment severity ~5.0); (b) Untreated poplar wood and
solid residues pretreated at 270 °C, water-only for 10 min (pretreatment severity ~6.0)
Py-GC/MS analysis of ReL revealed structural information in the form of pyrolytic products.
Selected py-GC/MS results of residual solids pretreated at 240 °C with 0.05% (w/w) H2SO4 for
10 min (LogR0 ~5.0) and at 270 °C by water-only for 10 min (LogR0 ~6.0) are shown in Figure
6.4. The pyrolysis products of residual lignin can be found in Table A1. As seen from the
differences among peak intensities in Figure 6.4, solid residues after pretreatment showed much
less lignin remaining than that in the untreated poplar wood due to the enhanced lignin
depolymerization.
(a) (b)
171
Figure 6.5 Effects of pretreatment severity on S, G and H unit removal by water-only and 0.05%
(w/w) dilute acid pretreatment (poplar wood control sample: G:S:H= 7:11:2)
The ratios of the S, G, and H units contain structural information about lignin. The S/G ratio
was used as a criterion for evaluating delignification352, 356. Py-GC/MS was applied to investigate
S/G ratio change during flowthrough pretreatment. The preferential release of sub-units (G and
S) of lignin into the aqueous phase was reported in flowthrough pretreatment at temperatures up
to 200°C 357. In this study, nearly complete lignin removal was achieved under severities of
higher than ~5.0 and ~6.0 in dilute acid and water-only pretreatments, respectively. The dynamic
changes of the S, G and H units during flowthrough pretreatment are shown in Figure 6.5. The
spectra in Py-GC/MS results were normalized and the relative percentages of the S, G, and H
172
units or their reduction based on untreated poplar lignin are shown.
Figure 6.5 shows the high extent of release of S, G, and H units into the aqueous phase at
severities higher than 4.2 and 4.5 in dilute acid and water-only pretreatments, respectively. The
removal rate of the S unit was higher than that of the G units in flowthrough pretreatment,
indicating the preferential removal of S units versus G units. Such preference was consistent with
previous results of water-only pretreatment at temperatures lower than 200°C 62, 358. The H unit,
constituting about 10% of lignin in untreated poplar wood, showed the fastest removal, which
might due to greater reactivity at the C3 and C5 positions. The elevated temperatures and
addition of 0.05% (w/w) H2SO4 enhanced lignin depolymerization, thus, it not only led to
removal of the S units but also promoted solubilization of the G unit.
6.4.4 FTIR spectroscopic analysis of ReL
FTIR was carried out to evaluate ReL structural modification in functional groups by
flowthrough pretreatment. The assignments of the functional groups of lignin in this study were
shown in Figure 6.6 and Table 6.1. Peaks at 1260 cm-1-1760 cm-1 were assigned to lignin
structural spectral vibration regions 62. In the selected spectra of solid residues obtained from
both flowthrough pretreatments, ~80% reduction of peak intensity at 1594, 1455, and 1423 cm-1
was observed, indicating removal of the aromatic rings to liquid phase (Figure 6.6). The shape
and intensity changes in the peak at 1115 cm-1 were also observed to correspond with the
modification of the aromatic ring and cleavage of the ester linkages, which was especially
obvious in sample C (solid residues by 0.05% sulfuric acid pretreatment at 240℃ for 5 min). The
peak at 1735 cm-1 disappeared in all samples, which indicated the removal of oxidized side
chains. In addition, the peak intensity at 1655 cm-1 increased, indicating the formation of
173
conjugated carbonyl groups, particularly under acidic conditions.
Figure 6.6 Characterization of ReL in pretreated solid residues by FTIR (800-2000 cm-1).
Control: Untreated Poplar wood; a) Solid residues from 0.05% (w/w) sulfuric acid pretreatment
at 240 °C for 2.6 min; b) Solid residues from water-only pretreatment at 230℃ for 3.8 min; c)
Solid residues pretreated with 0.05% sulfuric acid at 240℃ for 5 min
Changes of the S and G units were also observed in the IR spectra. Peaks at 1325, 1244, and
1268-1270 cm-1 decreased in solid residues, demonstrating the release of the syringyl and
guaiacyl derivatives into the liquid phase. These results were consistent with the observation by
Py-GC/MS analysis in which addition of dilute acid enhanced releasing of S and G units to
aqueous phase. The FTIR results showed that more than 90% of S and G units from biomass
solids could be removed.
(a)
(b)
(c)
174
Peaks at 1375, 1120 and 1030-1086 cm-1 was assigned to bond stretching vibrations in
cellulose and hemicellulose. The decrease of these peak intensities was attributed to the
solubilization of cellulose and hemicellulose into the aqueous phase during pretreatment.
Particularly, sample C presented almost a 90% peak decrease in the carbohydrates region.
Table 6.1 FTIR spectra band assignments 124, 359-364
No. Wavenumber(cm-1) Assignment
1 1708-1738 Stretching of C=O unconjugated to aromatic rings (oxidized side-chains,
non-conjugated carbonyl)
2 1655 Stretching of C=O conjugated to aromatic rings (conjugated carbonyl)
3 1590-1609 Aromatic ring vibrations and C=O stretching
4 1500-1515 Aromatic ring vibrations
5 1455/1425 C-H deformation and aromatic ring vibration
6 1420-1424 Aromatic ring vibrations
7 1375 C-H stretching in cellulose and hemicellulose
8 1330-1325 Syringyl nuclei (C-O stretching)
9 1270-1268/1244 Guaiacyl nuclei (C-O stretching)
10 1221 C-C, C-O, C=O stretching in G ring
11 1160 Deformation vibrations of C-H bonds on benzene rings
12 1120 Carbon ring stretching of cellulose
13 1115 Vibrations of ester linkage
14 1086 C-O stretch of secondary alcohols and aliphatic ethers
15 1048 C-O stretch in cellulose and hemicellulose
16 1060/1040/1030 C-O stretch (primary alcohols)
6.4.5 Characterization of the RSL by GC/MS
GC/MS can identify lignin derivatives in soluble phase but it is less sensitive to most
oligomeric lignin polymers (>1000) because of their low volatility. As discussed above, RSL was
shown only less than 20% and 5% of the total yields of original lignin in water-only and dilute
acid flowthrough pretreatments, respectively. However, profiles of lignin derivatives in the RSL
phase can provide evidence of the lignin depolymerization mechanism.
In this study, the identified derivatives from water-only and 0.05% (w/w) sulfuric acid
pretreatments were mainly vanillin, benzaldehyde, hydroxybenzaldehyde, 2-methoxy-4-
vinylphenol, 2,6-dimethoxy-phenol, 4-hydroxy-3,5-dimethoxy-benzaldehyde, and benzoic acids.
175
These compounds indicated possible Cα oxidation and cleavage of β-O-4 linkages 357. In
addition, these products suggested that dehydroxylation or demethoxylation reactions were
present in both flowthrough pretreatments, which is consistent with FTIR results. Other
identified derivatives included, interestingly, the compounds 2,2'-methylenebis[6-(1,1-
dimethylethyl)-4-methyl-phenol-], 4-(1,1,3,3-tetramethylbutyl)-phenol, and 2,6-bis(1,1-
dimethylethyl)-4-methyphenol (butylated hydroxytoluene) (Table S6.1).
6.4.6 2-D NMR of Ball-milled poplar lignin and RISL
Two-dimensional 13C- 1H NMR (2-D NMR) provided evidence of lignin alteration by
flowthrough pretreatments. Both the aliphatic (δC/δH 50-90/2.5-6.0 ppm) and aromatic (δC/δH
100-135/5.5-8.5 ppm) regions’ HSQC spectra were collected (Figure 6.7). The assignments of
the NMR signals are shown in Figure 6.7 and summarized in detail in the supplemental material
(Table S6.2; Figure S6.1). Relative peak area integrations for ball milled lignin from the same
poplar wood sample and RISLs were measured to determine each lignin linkage and monolignol
change (Table 6.2). The quantification of each linkage is from the volume-integration of cross-
peak contours in HSQC spectra according to previous publication 62. Monolignol quantification
was performed by peak integrations of peak C2/6 from S and H units and peak C2 from G units
(doubled). The quantification of the basic linkages, Aα, Bα, and Cα contours were integrated to
represent A, B, and C, respectively, among which Cα and Cβ contours in C represented α-O-4´
and β-5´ linkages 62, 365.
176
Figure 6.7 2-D 13C - 1H HSQC correlation NMR spectra of aliphatic regions (left column) and
aromatic regions (right column), (a, b) Ball milled poplar wood lignin; (c, d) 240 °C, 0.05%
(w/w) H2SO4 for 10 min (pretreatment severity ~5.0); (e, f) 270 °C, water-only for 10 min
(pretreatment severity ~6.0); Label A: β-O-4 aryl ether linkages; Label B: resinol substructures
(β-β, α-O-γ, and γ-O-α linkages); Label C: phenylcoumaran substructures (β-5’and α-O-
4linkages); Label D: spirodienone substructures ( β-1 and α-O-α linkages); Label G: guaiacyl
(a) (b)
(d) (c)
(f) (e)
177
units; Label G: oxidized guaiacyl units with an Cα-ketone; Label S: syringyl units; Label S:
oxidized syringyl units with a Cα ketone; Label E: p-hydroxybenzoate substructures; Label F:
cinnamyl alcohol end groups; Label K: cinnamaldehyde end groups. Their structures can be
found in supplementary material in Figure S6.1
Table 6.2 Relative proportions of major structural linkages and S, G, H percentages in ball
milled lignin and flowthrough derived RISL; A: RISL collected at 240 °C, 25 ml/min, 10 min,
with 0.05% (w/w) H2SO4; B: 270 °C, 25 ml/min, 10 min, with water-only
Samples β-O-4 Resinol
(β-β´)
Phenylcoumaran Spirodienone
( β-1´, α-O-α´)
S/G
β-5´ α-O-4´
Ball milled 73.07 10.71 3.81 12.01 0.40 0.602
A 46.57 8.18 5.72 8.88 0.00 0.674
B 49.38 8.19 5.94 8.49 0.00 0.669
The NMR spectra of pretreatment derived lignins showed the basic linkages of lignin as in
ball milled wood lignin. Comparing pretreatment-derived lignin and ball milled wood lignin
indicates that flowthrough pretreatment only caused mild modification on original lignin. These
structural changes include, first in the aliphatic region (in Figure 6.7a, c, e), loss of structure D
(spirodienone) which was not seen in both RISLs pretreated by water-only and dilute acid,
indicating that flowthrough pretreatment decomposed this structure. Second, in the aromatic
region, the RISL showed formation of propenyl end group structures F (cinnamyl alcohol and its
O4 ether and 3- and/or 5-methoxy analogues) and K (like cinnamaldehyde and analogues), which
were not found in ball milled poplar lignin. A plausible explanation for these propenyl end
groups is β-O-4´ cleavage by a mechanism involving eliminative dehydration to remove the α-
OH and produce an enol ether, which could then be cleaved by protonation of the enol, addition
178
of water, and formation of a carbonyl at the α or β position upon cleavage of the aliphatic ester
bond. Carbonyls at Cβ were not observed, while Cα carbonyls were observed, although either of
these could be reduced and eliminated (dehydration) to yield the alkene. Oxidation of the γ-OH
to the aldehyde or carboxylic acid could be a source of electrons for reduction of these carbonyls
(Figure 6.8).
Quantification of lignin linkages and substructures was obtained from 1-D 13C NMR
spectra and by measuring contour volume integrals of 2-D HSQC peaks 157, 366. The
aromatic region provided S and G ratios in RISL while the aliphatic region indicated an
abundance of C-O-C and C-C linkages (Table 6.2). Ball-milled poplar lignin was mainly
composed of β-O-4 ether, resinol, and phenylcoumaran structures with a small portion of
spirodienone structure, p-hydroxybenzoate, and the p-hydroxycinnamyl alcohol. After
pretreatment, the RISL showed different ratios of these linkages. However, the
flowthrough pretreatment only caused a modest change in the ratio, especially for peak
volumes in the aromatic region. First, the decline in percentages of β-O-4´ linkage, α-O-
4´, and resinol structures indicated the depolymerization of these structures in the original
lignin by pretreatment. Secondly, β-5´ structure slightly increased compared to ball milled
wood lignin indicating the occurrence of lignin β-5´ condensation reactions. The flowthrough
system prevented severe condensation reactions that occur in batch reactors65, 367. Finally,
cinnamyl-like structures appear in the NMR spectra, suggesting a β-O-4´ cleavage mechanism
linked to dehydration at the α-position and oxidation at γ-OH. A possible mechanism is
illustrated in Figure 6.8.
179
Figure 6.8 Plausible lignin β-O-4 cleavage and condensation reaction pathways in hydrolysates
6.5 Conclusion
Our previous work has shown great advantages in recovering sugars for our flowthrough
pretreatment system operating at elevated temperature 345. Understanding the fundamental
chemistry of lignin during pretreatment is crucial to the development of advanced biorefinery
strategies. In the flowthrough system, elevated temperature and addition of dilute acid was found
to lead to cleavage of primarily C-O-C as well as C-C linkages in lignin, and effective
180
solubilization of lignin derivatives into the aqueous phase with mild modification of lignin
aromatic structures. Results indicated that 0.05% (w/w) dilute sulfuric acid improved RISL yield
as well as its purity. Together with elevated temperatures, 0.05% (w/w) dilute sulfuric acid
enhanced removal of not only S units but also most of G units, thus resulting in almost 100%
removal of lignin. Modest structural changes in side chains were observed, with formation of
benzylic carbonyl groups at the α-position, as well as α-β unsaturation and oxidation of the γ-OH
to the aldehyde. A new plausible explanation for these propenyl end groups during the
pretreatment is β-O-4´aryl ether cleaved by dehydration at the α-position and oxidation at γ-
OH. However, besides basic lignin linkages in RISL, such as resinol, β-O-4´, and the
phenylcoumaran structures observed, slight repolymerization occurred in the form of new Cβ-
C5´ linkages in RISL. Thus, flowthrough pretreatment of biomass has the potential to enable
high-yield recovery of lignin derivatives for further upgrading, a potentially vital factor in the
economics of biorefinery operations 325, 330, 337.The lignin derived in this chapter has been tested
on the conductive electrode materials application 328.
6.6 Acknowledgements
The author is grateful to the DARPA Young Faculty Award # N66001-11-1-414, DOE-EERE
Award # DE-EE0006112, The Sun Grant-DOT Award # T0013G-A- Task 8, and the National
Science Foundation Award # 1258504 for funding this research. Part of this work was conducted
at the William R. Wiley Environmental Molecular Sciences Laboratory (EMSL), a national
scientific user facility located at the Pacific Northwest National Laboratory (PNNL) and
sponsored by the Department of Energy’s Office of Biological and Environmental Research
(BER). Thank Drs. Yunqiao Pu, Hongliang Wang, and Hongfei Wang for insightful discussions.
181
The thesis author would like to thank Drs. Lishi Yan, Zheming Wang, Dhrubojyoti D. Laskar,
Marie S. Swita, John R. Cort, and Bin Yang for their support to this work and publish in
Biotechnology for Biofuels journal.
182
6.7 Supplementary materials
Table S6.1 Major GC/MS detected aromatic compounds in hydrolysates
a:0.05%(w/w) H2SO4, 62.5mL/min; b: Water-only, 25mL/min; c: 0.05%(w/w) H2SO4, 25mL/min; d: pretreatment severity
factor. + 0-10%; ++ 10%-20%; +++ 20%-50%; - not available.
Major lignin derivatives Relative abundance Structure MW
4.2d
ACIDa
4.4d
ACIDa
4.7d
ACIDa
4.5d
HWb
6.0d
HWb
4.8d
ACIDc
5.1 d
ACIDc
Vanillin
+
-
-
-
-
+
+
152
Butylated Hydroxytoluene (BHT)
++
++
++
++
+++
++
+
220
Phenol, 4-(1,1,3,3-tetramethylbutyl)-
+
+
+++
+
++
+++
+++
206
Coniferyl alcohol
-
-
-
-
-
+
-
180
Phenol, 2,2'-methylenebis[6-(1,1-
dimethylethyl)-4-
+++
+++
+++
++
++
+++
+++
340
183
Figure S6.1 Assigned interunit linkages of lignin, including different side-chain linkages, and
aromatic units: (A) β-O-4 aryl ether linkages; (B) resinol substructures (β-β′, α-O-γ′, and γ-O-α′
linkages); (C) phenylcoumarane substructures (β-5′ and α-O-4′ linkages); (D) spirodienone
substructures (β-1′ and α-O-α′ linkages); (G) guaiacyl units; (G′) oxidized guaiacyl units with an
Cα-ketone; (S) syringyl units; (S′) oxidized syringyl units with a Cα ketone; (E) p-
hydroxybenzoate substructures; (F) cinnamyl alcohol end groups; (K) cinnamaldehyde end
groups
184
Table S6.2 Assignments of main lignin 1H- 13C cross-peaks in the HSQC Spectra of the RISLs
144, 157, 368-371
Lignin linkages and monolignols Chemical shift
-OCH3 55.47(C) 3.70(H)
A: β-O-4 71.70(S-Cα) 4.84(S-Hα) 59.50-59.70(Cγ) 3.40-3.63(Hγ) 85.90(S-
Cβ) 4.09(S-Hβ) 83.49(G/H-Cβ) 4.28(G/H-Hβ)
B: resinol 84.85(Cα) 4.62(Hα) 53.30(Cβ) 3.05(Hβ) 70.85(Cγ) 4.14/3.78(Hγ)
C: phenylcoumaran 86.79(Cα) 5.41(Hα) 53.3(Cβ) 3.46(Hβ) 62.52(Cγ) 3.68(Hγ)
D: spirodienone 59.67(Cβ) 3.19(Hβ)
G: guaiacyl 111.02(C2) 6.95(H2) 115.05(G5) 6.74(H5) 119.01(G6) 6.78(H6)
G: oxidized (Cα=O) guaiacyl 111.56(C2) 7.50(H2) 123.55(C6) 7.54(H6)
S: syringyl 103.95(C2/6) 6.67(H2/6)
S: oxidized (Cα=O) syringyl 106.52(C2/6) 7.29(H2/6)
E: p-hydroxybenzoate 131.33(C2/6) 7.62(H2/6)
F: cinnamyl alcohol 128.39(Cβ) 6.20(Hβ) 128.59(Cα) 6.42(Hα)
K: cinnamaldehyde 126.23(Cβ) 6.74(Hβ)
a Note: G, S, H-C or G, S-H refers to C and H in the lignin sub-units, guaiacyl, syringyl and p-
hydroxyphenyl
185
CHAPTER SEVEN
REVEALING THE MOLECULAR STRUCTURAL TRANSFORMATION
OF HARDWOOD AND SOFTWOOD IN DILUTE ACID
FLOWTHROUGH PRETREATMENT
7.1 Abstract
To better understand the intrinsic recalcitrance of lignocellulosic biomass, the main hurdle to
its efficient deconstruction, the effects of dilute acid flowthrough pretreatment on the dissolution
chemistry of hemicellulose, cellulose, and lignin for both hardwood (e.g. poplar wood) and
softwood (e.g. lodgepole pine wood) were investigated at temperatures of 200 °C to 270 °C and
a flow rate of 25 mL/minute with 0.05% (w/w) H2SO4. Results suggested that the softwood
cellulose was more readily to be degraded into monomeric sugars than that of hardwood under
same pretreatment conditions. However, while the hardwood lignin was completely removed into
hydrolysate, ~30% of the softwood lignin remained as solid residues under identical conditions,
which was plausibly caused by vigorous C5-active recondensation reactions (C-C5). Unique
molecular structural features that pronounced the specific recalcitrance of hardwood and
softwood to dilute acid pretreatment were identified for the first time in this study, providing
important insights to establish the effective biomass pretreatment.
Keywords: Softwood and Hardwood • Lignin • Structural transformation • Recondensation •
Solid State NMR • 2-D HSQC NMR • Dilute acid flowthrough pretreatment
7.2 Introduction
Cellulosic biomass has a rigid plant cell wall mainly composed of cellulose, hemicellulose,
and lignin. Biomass is recalcitrant to chemical or microbiological conversions and a pretreatment
186
process is usually needed to disrupt the cell wall matrix and reduce the recalcitrance of biomass
to improve sugar yields for fermentation. The characterization of lignin in pretreatment is
important for understanding biomass recalcitrance and developing effective pretreatment
methods. Among various biomass species, hardwood and softwood vary greatly in lignin content,
monolignol types, and interunit linkages abundance 89, 372, 373. For example, softwoods generally
contain more lignin than hardwoods, and softwood lignin is composed of mostly G units while
hardwood lignin contains both S and G units. In addition, softwood presents more C-C linkages,
and fewer C-O-C linkages (Table 7.1). Unfortunately, pretreatment of softwood is more
challenging for reducing the recalcitrance and improving sugar recovery than hardwood, and the
mechanism still remains to be fully understood.
Table 7.1 General lignin difference in hardwood and softwood
Biomass Linkages
Abundance (%) 89
Monolignol types
(%)
Molecular weight
(Bjorkman lignin) 372, 373
Amount of
lignin (%)
Softwood β-O-4: ~50
α-O-4: ~8
5-O-5: ~5
β-5: ~11
5-5: ~18
β-1: ~7
β-β: ~2
G: 90-95
S: 0-1
H: 0.5-3.5
e.g. spruce
Mn: ~3000
Mw: ~10000
PD: ~3.33
~28
Hardwood β-O-4: ~60
α-O-4: ~8
5-O-5: ~5
β-5: ~6
5-5: ~10
β-1: ~7
β-β: ~2
G: 25-50
S: 50-75
H: Trace
e.g. Eucalyptus
globulus
Mn: ~2600
Mw: ~6700
PD: ~2.6
~20
Generally, softwood pretreatment achieved less sugar yield and lignin removal than
hardwood pretreatment. Softwood pretreatment methods previously reported include biological,
187
steam explosion with or without acid/SO2, alkaline, hot water, dilute acid, and organosolv
pretreatment 300, 374-384. The dilute acid pretreatment is known for achieving more than 80%
recovery yield of hemicellulose sugars. However, most dilute acid pretreatments ranging up to
200°C can only achieve a limited improvement of cellulose sugar yield (i.e, glucose yield) from
softwood. For example, the dilute acid pretreatment of various body parts of pine wood achieved
a glucose yield of 8.7-42.3% of original glucan in the pretreatment stage at 180°C with 2-4 %
(w/w) sulfuric acid for 6-12 min and a glucose yield of 13.6-32.6% in the a 72 h enzymatic
hydrolysis of pretreated solid residues 385. Also, lodgepole pine was pretreated at 180°C with
2.2 % (w/w) sulfuric acid for 30min and achieved 18.3% yield of original glucose on pretreated
solid residues after a 72 h of enzymatic hydrolysis 386. In addition, diluted acid pretreatment was
also observed to have a limitation for lignin removal in softwood. Li et al. reported a 2.2% lignin
removal in lodgepole pine wood at 180 °C. At higher temperatures of 200-230°C, with 90 – 95%
of the hemicellulose and ~ 20% of the cellulose solubilized, residual Klason lignin remained in
pretreated solid residues was reported at levels of more than 129% of original lignin content,
which was plausibly due to the formation of pseudo lignin from carbohydrates 305, 387. When
compared to pretreatment of softwood, dilute acid pretreatment of hardwood showed different
results in sugar yield, enzymatic digestibility and lignin removal. For example, when poplar
wood was pretreated with steam heated to 210°C for 5 min, ~68% of the original glucose was
recovered while 10-18% lignin was removed 388. In the Consortium for Applied Fundamentals
and Innovation (CAFI) study, polar wood was subjected to dilute acid pretreatment at 190°C
with 2.0% (w/w) sulfuric acid for 1.1 min, resulting in a 23.9% of the original glucose recovery
in the pretreatment stage, 62.8% of the original glucose in enzymatic hydrolysis stage (with
188
enzyme loading of 15 FPU/g glucan), and lignin removal of ~28% 389.
The diverse pretreatment methods and conditions on hardwood and softwood pose
challenges to resolving the recalcitrance of softwood and hardwood in pretreatment. Most of
pretreatment approaches have been carried out in batch systems, which have limitations in
selectively characterizing pretreatment derived biomass components. Under batch process, the
lignin chemistry during dilute acid or hot water pretreatment was reported to be predominantly
centered about depolymerization reactions via rupture of the β-O-4 linkages forming Hibbert
ketones and repolymerization by acid catalyzed condensation between the aromatic C6 and a
carbonium ion at Cα 68, 69. In addition, literature reports have documented that during batch dilute
acid pretreatments a) lignin droplet formation and redeposit occurs, b) pseudo lignin formation,
c) lignin coalesce and migration within and out of cell wall 103, 104, 106, 390. In contrast, a
flowthrough system can reduce the above mentioned phenomena by exiting fractionated products
out of the reactor in a very brief time and providing effective mass and heat transfer 27. This
continuous pretreatment concept was pioneered by Bobleter et al. and reported in 1983 44-47. The
hot water flowthrough systems can carry soluble materials away from the reaction zone and limit
the opportunity for further reaction or degradation. These result in high distributions of
oligomers and low prevalence of degradation products 50, 391. Furthermore, a sizable portion of
the hemicellulosic and lignin sheath surrounding cellulose macrofibrils is solubilized, and the
solubilized components are rapidly removed from the reaction zone. Similar results to
flowthrough pretreatments with just hot water are obtained if very dilute sulfuric acid is used as
the flowing solvent, with the exception that more cellulose is solubilized. Increasing the
temperatures of hot water flowthrough pretreatments to 225-270°C within or above saturated
189
steam pressure also solubilizes more cellulose 60, 345, 392. Thus, the flowthrough pretreatment
system is the promising tool to study the unique molecular structural features that pronounce the
specific impacts on hardwood and softwood during dilute acid pretreatment.
A recent study in our group reported almost 100% whole poplar biomass solubilization in
the dilute acid flowthrough system 60. The flowthrough system allowed separation of lignin from
biomass to study the lignin characteristics, and the results indicated the predominant β-O-4
cleavage and slight β-5 recondensation mechanisms in dilute acid pretreatment, which improves
our understanding of poplar lignin chemistry and structural transformation in dilute acid
pretreatment on a molecular structural basis. In this study, dilute acid pretreatment of softwood
loblolly pine was carried out in the flowthrough system for the first time to investigate the
behaviours of its major components as compared to those of hardwood poplar under same
conditions 59, 60. The effectiveness of the flowthrough pretreatment on softwood was assessed by
sugar yields after pretreatment and enzymatic hydrolysis. The softwood lignin structural
transformation was investigated through two-dimensional heteronuclear single quantum
coherence solution nuclear magnetic resonance spectroscopy (2-D 1H-13C HSQC NMR) and
solid state CP/MAS 13C NMR.
7.3 Materials and methods
7.3.1 Materials
Beetle-killed Lodge pole pine was provided by Forest Concepts, LLC., and milled to a
particle size of 40-60 mesh. The feedstock compositions were analyzed according to the K-lignin
analysis protocol 128. The pine wood chips contained 37.89±0.59% of glucan, 5.25±0.047% of
xylan, 29.95 ±0.78% of lignin, 3.62±0.093% of galactan, 12.14 ±0.018% of arabinan and
190
mannan, and 0.62±0.015% of ash.
7.3.2 Dilute acid flowthrough pretreatment of pine wood
The flowthrough pretreatment of Beetle-killed Lodge pole pine was conducted with a 0.05%
(w/w) sulfuric acid solution at 200-270°C for 2-10 min (3.7-6.0 pretreatment severities) 59, 60.
The pretreatment severity refers to the variable incorporating temperature and time (equation
7.1) 354.
LogR0 = Log[t × exp (T−100
14.75)] (7.1)
T refers to the temperatures applied in the pretreatment and t represents the pretreatment time.
The flow rate was set to 25ml/min. 0.5 g biomass was loaded into the tubular reactor (0.5′′ O.D.
× 6′′ length; 0.035′′thickness; 20.2 ml), which was attached to the flowthrough system. The
0.05% (w/w) sulfuric acid solution was pumped through the reactor continuously while the
reactor was placed in a fluidized sand bath with a thermocouple (Omega Engineering Co.,
Stamford, CT) to monitor the actual temperature. The flowthrough system was set up as
previously reported 59, 60. After pretreatment, the solid residues remaining in the reactor were
collected, and their composition was analyzed according to the K-lignin protocol 128. Monomeric
sugar in pretreated hydrolysate was analyzed by HPLC (Biorad Aminex HPX-87P column). In
addition, the pretreated hydrolysate was posthydrolysed with 4% (w/w) sulfuric acid and
autoclaved at 121 °C for 1 h to determine total sugar recovery yield and oligomeric sugars yield
in pretreatment. For pretreatment recovered insoluble lignin collection, the pretreatment at the
selected condition was repeated 10 times to accumulate hydrolysates to precipitate lignin at pH
2-3 followed by washing, filtration, and freeze drying for analysis 325.
191
7.3.3 Structural characterization of pretreatment recovered insoluble lignin by 2-D 1H-13C
HSQC NMR
The pretreatment recovered insoluble lignin (50 mg) was dissolved in 600 μL deuterated
DMSO (Cambridge Isotope Laboratories). The resulting pretreatment recovered insoluble lignin
was placed in 5 mm Wilmad 535-PP NMR tubes. NMR spectra were collected at 25° C on 500
and 600 MHz Agilent (Varian) Inova NMR spectrometers equipped with z-axis pulsed-field
triple-resonance HNCP probes. Samples contained 0.05% (v/v) TMS for chemical shift
referencing. Two-dimensional 1H-13C HSQC spectra of the aliphatic and aromatic regions were
collected separately using the BioPack gchsqc pulse sequence, with 1H spectral width of 17 ppm
and 13C spectral widths of 100 or 60 ppm for the aliphatic or aromatic regions, respectively.
Spectra were collected with 1024 points (Varian parameter np) and 61 ms acquisition time in 1H
dimension and with 128 or 256 transients and 128 or 96 complex points (Varian parameter ni in
States-TPPI mode) in the indirect 13C dimension, for aliphatic and aromatic spectra, respectively.
Adiabatic WURST decoupling was applied during acquisition. Delay times tCH and lambda for
1/4*JCH, were 1.8 ms and 1.6 ms for aliphatic spectra, and 1.45 ms and 1.3 ms for aromatic
spectra. Reference one-dimensional 1H spectra were collected with 32k points and 128 transients.
HSQC spectra were processed and analyzed with Felix 2007 (FelixNMR, Inc) or MestReNova
6.0.4 (Mestrelab Research), with matched cosine-bell apodization in both dimensions, 2X zero
filling in both dimensions, and forward linear prediction of 30% more points in the indirect
dimension. One-dimensional 1H spectra were processed with no apodization or linear prediction
and 2X zero filling. Relative peak integrals were measured in MestReNova 60. The same method
has been reported in our previous study 60.
192
7.3.4 Solid state CP/MAS 13C NMR characterization of dilute acid flowthrough pretreated
poplar and pine lignin
The solid-state CP/MAS 13C NMR experiments were performed on a Bruker Avance III 400
MHz spectrometer operating at frequencies of 100.59 MHz for 13C using a Bruker double-
resonance 4-mm MAS probe head at ambient temperature. The samples were packed in a 4 mm
ZrO rotor fitted with a Kel-F cap and spun at 8,000 Hz. CP/MAS 13C data were acquired with a
Bruker CP pulse sequence under the following acquisitions: pulse delay 4s, contact pulse
2000ms, and 2k to 4k numbers of scans.
7.3.5 Gel permeation chromatography molecular weight analysis of the flowthrough
derived pine lignin
The lignin molecular weight distribution was analyzed with gel permeation chromatography
(GPC) after acetylation. The lignin samples were dissolved in a mixture of acetic anhydride/
pyridine (1:1 v/v) and stirred at room temperature for 24 h. After acetylation, ethanol was added
to the reaction mixture and then removed with a rotary evaporator. The addition and removal of
ethanol was repeated until until all traces of acetic acid and pyridine were removed from the
sample. The resulted acetylated lignin samples were analysed on a PSS-Polymer Standards
Service (Warwick, RI, USA) GPC Security 1200 system featuring Agilent HPLC 1200
components equipped with four Waters Styragel columns (HR0.5, HR1, HR4 and HR6) and an
UV detector (270 nm). Tetrahydrofuran was used as the mobile phase, and flow rate was 1.0
mL/min. Standard narrow polystyrene samples were used for calibration 325.
7.3.6 Enzymatic hydrolysis
Enzymatic hydrolysis of pretreated whole slurries was carried out by adding Novozymes
193
Cellic® CTec1 (220 mg protein/mL, preserve 200 mg glucose/mL, 93 FPU/mL) and Novozymes
Cellic® HTec2 (230 mg protein/mL, preserve 180 mg xylose/mL) at a ratio of 5:1. The enzyme
loadings were 20 or 100 mg protein/ g (glucan+xylan), in which the higher enzyme loading can
evaluate the theoretical glucose yield while the lower enzyme loading was designed to compare
the digestibility of substrates from different pretreatment conditions. The pretreated whole slurry
was adjusted to pH 4.8-5.0 with 0.1 N NaOH. 1% (w/w) bovine serum albumin (BSA) was
added and supplemented with 10mg/mL sodium azide. The resulting mixture was placed in a
shaker at a temperature of 50°C with shaking speed of 180 mph for 72 h.
7.4 Results and discussion
7.4.1 Delignification and cellulose removal of poplar and pine wood with dilute acid
flowthrough pretreatment
Figure 7.1 Lignin and cellulose contents in pretreated solid residues of poplar and pine wood
after the flowthrough pretreatment (200-270°C, reaction time of 2-10min, and 0.05% (w/w)
H2SO4) (Poplar pretreatment data was adapted from previous publications 59, 60)
The dilute acid flowthrough pretreatment was studied for the effectiveness in delignification
194
and carbohydrates solubilization in loblolly pine during pretreatment. The lignin content in
pretreated solid residues was 0-25% and 25-65% for poplar and pine under tested pretreatment
conditions, respectively (Figure 7.1a). Under the same tested pretreatment conditions, 20-40%
more poplar lignin was removed than pine lignin, based strictly on mass corrected for different
levels of starting lignin, indicating a higher resistance to removal of pine lignin than poplar
lignin. The difficulty in softwood lignin removal during dilute acid pretreatment has been well
documented, probably because of the unique structural features and complexity of softwood
lignin behaviors in dilute acid pretreatment. For example, more than 100% of the original pine
lignin reportedly remained in pretreated solid residues during batch dilute acid pretreatment,
partially attributed to the generation of pseudo lignin 305. The flowthrough system can reduce the
formation of pseudo lignin by exiting solubilized products in a short time, resulting in a higher
lignin removal than batch system. Only less than 5% of original poplar lignin reportedly
remained in solid residues with flowthrough pretreatment 59. Thus, flowthrough pretreatment
provides a unique approach to characterize lignin behaviors during pretreatment on a molecular
structural basis.
Comparing lignin and cellulose contents in pine solid residues under pretreatment conditions
(Figure 7.1a, b), pine lignin removal was 20-40% lower than pine cellulose removal. Under
pretreatment severities higher than 5.2, pine cellulose was completely solubilized (0% in solid
residues), but nearly 30% lignin was still present in solid residues even at the highest tested
pretreatment severities. This 30% remaining lignin was collected for structural analysis in order
to understand the pine lignin chemistry and structural transformation in pretreatment. In
comparison, poplar lignin was nearly completely removed when the pretreatment severity was
195
higher than 5.3 while complete removal of poplar cellulose required higher severity over 5.8. At
the tested range of pretreatment severity (3.7-6.0), ~5% less poplar cellulose was solubilized in
general than pine cellulose while 25-40% more poplar lignin removal than pine lignin was
observed under identical conditions (Figure 7.1b). The difference between poplar and pine lignin
solubilization was bigger than the difference between their cellulose solubilization. Therefore,
poplar and pine showed different solubilization behaviors during dilute acid pretreatment.
7.4.2 Sugar recovery in dilute acid pretreatment of pine wood
Figure 7.2 Sugar yields of pine wood at 0.05% (w/w) sulfuric acid flowthrough pretreatment
under different severity log R0 (a) Cellulose recovery; (b) Hemi-sugars recovery
The sugar recovery pattern of pine wood was different than reported hardwood poplar sugar
recovery under the identical flowthrough pretreatment conditions 59. Pine wood sugar recovery
was higher than poplar wood under the same pretreatment conditions (see Figure 7.2). For
example, when the pretreatment severity was 5.2, the total sugar recovery (monomeric and
oligomeric sugars) from pine was ~90% of glucose and 90% of hemi-sugars. However, under the
same pretreatment conditions, the poplar glucose was recovered ~75% with a xylose recovery of
196
80% 59. Pine wood carbohydrates were mostly recovered as monomeric sugars and much less
oligomers. Conversely, poplar carbohydrates were recovered mostly as oligomers and less as
monomeric sugars under the same dilute acid flowthrough pretreatment conditions 59. It appeared
that hardwood cellulose showed more resistance to hydrolysis compared to softwood cellulose
during pretreatment.
The hemi-sugars in pine wood were recovered as high as 95% of original hemi-sugars.
When the pretreatment severity was higher than 4.7, the degradation of hemi-sugars to furfural
and formic acid occurred (Figure 7.2b). The results also showed that the dilute acid flowthrough
pretreatment of pine wood achieved high total sugar yield. For example, under pretreatment
severities higher than 4.8, the glucose recovery yield (monomeric sugar yield) was higher than
60% of original glucose (Figure 7.2a) while the hemi-sugars (monomeric sugars) was as high as
90% but declined when the pretreatment severities were higher than 5.2. Besides, under the same
conditions, our previous results reported 40-50% of monomeric glucose recovery yield in poplar
59, which was 10-30% less than those of pine wood in this study. This might have some relation
with the chemical compositions of hardwood and softwood. Hardwood has higher glucan
content, more xylan, and much less arabinan and mannan than softwood. Therefore, results
indicated that a higher yield of monomeric sugars could be obtained from softwood through
dilute acid pretreatment than that from hardwood under same conditions, which showed better
feasibility of softwood in producing monomeric sugars.
Results showed that softwood pine wood cellulose was hydrolyzed to a high yield of glucose
in the pretreatment stage regardless of its retarded delignification (Figure 7.1a). For example,
under the pretreatment severities higher than 4.8, the monomeric glucose yield from pine was as
197
high as 70% even though still 30% of original lignin was left in the solid residues. Thus, pine
cellulose is easier to be hydrolyzed than pine lignin by dilute acid flowthrough pretreatment.
7.4.3 Characterization of pretreatment recovered insoluble lignin by 2-D 1H-13C HSQC
NMR
Figure 7.3 Characterization of (a,b) ball milled pine lignin; (c,d) flowthrough pretreatment
recovered insoluble pine lignin from liquid phase by 2-D 1H-13C HSQC NMR. (pretreatment
conditions: 240 °C, 0.05% (w/w) sulfuric acid, and 10min; poplar wood data by 2-D 1H-13C
HSQC NMR under the identical analytical conditions has been published 60
The pretreatment recovered insoluble lignin refers to the removed lignin fragments in the
pretreatment liquid, which were collected through the precipitation of pretreatment liquid at pH
2-3 followed by water washing and freeze-drying. In this study, pretreatment recovered insoluble
198
pine lignin was collected for structural analysis along with ball milled pine lignin, a
representative of native lignin. To study the difference of poplar and pine pretreatment recovered
insoluble lignin, pretreatment recovered insoluble pine lignin was collected under the identical
pretreatment conditions performed on poplar previously 60, which was 240 °C with 0.05% (w/w)
sulfuric acid for 10min. The characterization of poplar wood pretreatment recovered insoluble
lignin along with ball milled poplar lignin was investigated by 2-D 1H-13C HSQC NMR in
previous publication 60, suggesting the modest modification of lignin structure. However, in this
study, results on pretreatment recovered insoluble pine lignin characterized under the same NMR
characterization conditions (Figure 7.3) showed distinct pine wood lignin spectra compared with
that of poplar wood as reported 60. The lignin linkages distribution of poplar 60 and pine before
and after pretreatment are displayed in Table 7.2. The quantification of each lignin linkage is
from the volume-integration of cross-peak contours in HSQC spectra according to previous
publications 60, 65, 365. The quantification of the β-O-4 ether, resinol group, and phenylcoumarane
was based on Aα, Bα, and Cα contours integration, respectively (peak assignments see Table
S7.2 and Figure S7.1).
The ratios of the three linkages of poplar wood and pine wood ball milled lignin were
81:11:9 60 and 78:5:17 in the order of β-O-4 ether, resinol group, and phenylcoumarane (Table
7.2), respectively. Compared to pine and poplar ball milled lignin, the β-O-4 ether in pine and
poplar flowthrough recovered insoluble lignin decreased ~52% and ~46%, respectively. This
indicates the cleavage of β-O-4 ether in pine wood lignin similar to poplar wood lignin under
acidic conditions. The slight difference of β-O-4 cleavage might be because S units were
reported to be less difficult to removal than G and hydroxyphenyl-like units 393. Similar to
199
hardwood poplar lignin, the β-O-4 cleavage was also the main depolymerization mechanism of
pine lignin. Additionally, the resinol groups in recovered insoluble pine wood lignin showed an
almost three times increasing in percentage more than in poplar lignin, suggesting their different
changes of resinol groups, and resinol groups were degraded slower than β-O-4. Significantly,
the phenylcoumarane groups in poplar pretreatment recovered insoluble lignin increased, which
was proposed due to the recondensation at β-5 in our previous study 60, while the results of pine
pretreatment recovered insoluble lignin also showed that β-5 condensation might occur because
phenylcoumarane groups in pine wood increased despite of the percentage of β-O-4 decreasing.
Thus, the results indicated the main β-O-4 depolymerization mechanism and possible β-5
condensation mechanism of pine pretreatment recovered insoluble lignin.
Table 7.2 The linkage distribution of ball-milled and flowthrough-derived poplar wood and pine
wood lignin (flowthrough pretreatment conditions: 240 °C, 0.05% (w/w) H2SO4, 10min)
Detected linkages β-O-4 ether Resinol group Phenylcoumarane
Poplar wood Ball milled 80.6 10.5 8.9
Flowthrough derived 38.9 17.5 43.6
Pine wood Ball milled 77.5 5.4 17.1
Flowthrough derived 41.8 16.3 41.8
7.4.4 Solid state CP/MAS 13C NMR characterization of residual pine lignin in pretreated
solid residues
In the dilute acid flowthrough pretreatment of this study, the pine carbohydrates were
completely removed at 240°C with 0.05% (w/w) H2SO4 for 10min as demonstrated previously,
which isolated the residual lignin for structural characterization. This undissolved portion of pine
residual lignin remained as solids in the reactor labeled as pine residual lignin. This residual
200
lignin was collected and freezing dried for solid-state CP/MAS 13C NMR analysis (Figure 7.4),
providing opportunities to study the unique structural transformation of pine lignin towards
dilute acid pretreatment. Instead, poplar lignin was completely solubilized, thus no residual
poplar lignin was available for comparison under the identical pretreatment conditions.
Additionally, both pine and poplar pretreatment recovered insoluble lignins in the pretreatment
liquid were analyzed by solid state CP/MAS 13C NMR for comparison to comprehensively
understand the pine wood lignin pretreatment mechanisms (Figure 7.4).
Figure 7.4 Solid state CP/MAS 13C NMR spectra of a) pine residual lignin; b) pine pretreatment
recovered insoluble lignin; c) poplar pretreatment recovered insoluble lignin under 240°C with
0.05% (w/w) H2SO4 for 10min
Solid state CP/MAS 13C NMR can determine the abundance of condensed structures in
lignin. The characterization of pine residual lignin and two pretreatment recovered insoluble
lignins from pine and poplar are displayed in Figure 7.4. The peak assignments of the spectra in
Figure 7.4 are shown in Table 7.3. Peaks at 146-147 ppm and within 100-120 ppm are assigned
to chemical shifts of carbons in G units. Results indicated the high abundance of G units in pine
lignin while less G units in poplar lignin (Figure 7.4a,b,c), which was consistent with high G
units content in original softwood lignin. The poplar pretreatment recovered insoluble lignin
201
showed a sharp peak at 143-144ppm, indicating a possible β-5 recondensed structure, consistent
with our previously reported poplar lignin recondensation mechanism 60.
The pine wood pretreatment recovered lignin in pretreated liquid and residual lignin in solid
residues showed different spectra. For example, the pine residual lignin contained a large amount
of saturated aliphatic chain (15-40ppm) and rare ether linkages (60-90ppm), suggesting that the
pine residual lignin was pure C-C linked interunit. However, the pine pretreatment insoluble
lignin demonstrated none saturated aliphatic chain while high abundance of ether linkages. Thus,
the pine residual lignin presented complete condensed C-C linkages in the structure, which
explained peak absence in testing of residual lignin in 2-D 1H-13C HSQC NMR using DMSO as
a solvent. Additionally, the residual lignin was consisted of a considerable amount of aliphatic R-
OR or R-OH linkage/groups, while these groups were almost depleted in pine pretreatment
recovered insoluble lignin. Instead, the pretreatment recovered insoluble lignin contained much
higher alkane groups in its aliphatic region. Therefore, the structures of pine residual lignin were
dramatically different from pine pretreatment recovered insoluble lignin.
The above-mentioned complete C-C condensed structures of pine residual lignin can be
observed in 120-140ppm. To be more specific, pine residual lignin presented a great amount of
C-C5 recondensed structures indicated by the broad peak at 120-140ppm. The peaks at 131-132
and 125-126ppm were possible from C5-C5. The C5-C5 in residual lignin was from
recondensation of solubilized lignin fractions since the majority of native 5-5 structure exited the
reactor along with dilute acid stream during the lignin depolymerization by the cleavage of β-O-
4, staying inside corresponded pretreatment recovered lignin. In addition, the most native C5-C5
linkages in lignin were reported to be etherified by the α and β carbons of another phenyl
202
propane unit to give the existing format of the eight-membered dibenzodioxocin ring instead of
simply C5-C5 156. Thus, this raised intriguing conclusion that C5-C5 recondensation is one of
mechanisms of forming pine residual lignin.
The chemical shift of C5 in pine residual lignin (115ppm) was abundant, indicating the
possibly intensive C-C5 recondensation reactions to form new C-C5 linkages. The broad peak
between 120-140ppm can be originated other condensed structures related to C5 position
(Labeled as C-C5) 144. The C-C5 linkages could possibly be Cα-C5, Cβ-C5, C6-C5 or C4-C5
structures besides C5-C5. The Cα and Cβ are relatively reactive under acidic conditions.
Therefore, the C-C recondensation active sites were proposed to be possibly located at stable Cα-
C5 and Cβ-C5 besides other possible C-C5 recondensation. Because the model compounds of
those new C-C5 linkages were not available, the study is unable to get direct proof through NMR
analysis of structural transformation of the model compounds. Cα-C5 recondensation
mechanisms were reported in Kraft lignin structure derived from the Kraft pulping and from
cellulolytic enzyme lignin repolymerization in 0.1 M H2SO4 dilute acid pretreatment at 160°C
for 0−20 min 70, 71, 322, 394, 395. Also, Cα-C5 recondensation pathways were mentioned possible in
the steam explosion of aspen wood but not proposed due to the lack of straightforward evidences
68, 396. Thus, the possible recondensation mechanisms of Cα-C5 and Cβ-C5 were proposed in
Figure 7.5. The chemical pathways of C5-C5, C6-C5 or C4-C5 recondensation under acidic
conditions are still not clear and need further research.
Table 7.3 Selected chemical shifts and signal assignments in a 13C NMR spectrum 144, 397-399
δ (ppm) Assignment
146-147 C-3/C-4 in (non-)etherified G units (β-O-4
type)
143-144 C-4 in β-5 units
131-132 C-1/C-5/C-5′ in (non-) etherified 5-5 units
203
134.6 C-1 in etherified G units
125-126 C-5/C-5′ in non-etherified 5-5 units
100-120 C-1,2,5,6 in G units
115 C-2, 5, 6 in G units
86-87 Cα in G type β-5 units
60-85 Cα,β,γ in G type (β-O-4)
15-40 CH3 and CH2 in saturated aliphatic chain
56 -OCH3
Figure 7.5 Possible recondensation mechanisms of pine lignin during dilute acid pretreatment
60,70
204
7.4.5 GPC analysis of the flowthrough derived lignin
The molecular weight of pine and poplar pretreatment recovered insoluble lignin and pine
residual lignin were measured by GPC (Figure S7.2). The results showed that the pretreatment
recovered insoluble and residual pine lignin in this study possessed relatively small molecular
weight lignin with limited polydispersity (Table 7.4). In addition, the molecular weight of
pretreated pine wood residual lignin sub-unit (Mn) in this study was smaller than pine wood
recovered lignin in hydrolysate, poplar wood recovered lignin, and other reported pretreated
lignin 400.
The molecular weight of pine residual lignin was lower than pine and poplar pretreatment
recovered insoluble lignin (Table 7.4). Also, the polydispersity of pine residual lignin was
much higher than both pine and poplar pretreatment recovered insoluble lignin. These results
indicated that the condensed C-C5 in pine residual lignin originated from recondensed structures
of small lignin fractions regardless of the C-C5 in the original lignin. This fact confirmed our
proposed recondensation mechanisms (Figure 7.5) and eliminated the possibility of condensation
structures deriving from the original lignin structure. The molecular weight of poplar
pretreatment recovered insoluble lignin was higher than pine pretreatment recovered
insoluble lignin. It might be attributed to the more β-5 recondensation reactions that occurred as
reported in our previous study 60. This also indicated that poplar lignin is relatively easy to
remove compared to pine lignin.
Table 7.4 The molecular weight of pine lignin (flowthrough pretreatment at 240 °C, 0.05%
(w/w) sulfuric acid, and 10 min)
Sample name Mn Mw Polydispersity (D;
M̅w/M̅n)
205
Pine liquid phase lignin 542.0 1556.0 2.87
Pine residual lignin 224.5 1205.5 5.37
Poplar liquid phase lignin 60 1083 1955 1.81
7.5 Conclusion
The dilute acid flowthrough pretreatment effectively removes cellulose and hemicellulose
fractions from both hardwood and softwood. For the hardwood lignin, the structure’s sight
condensation in Cβ-C5 and the loss of the γ–methyl group in the completely removed lignin
fraction occurred 60. However, 30% of the softwood lignin with relatively low molecular weight
remained in the pretreated solid residues due to C-C recondensation in reactive C-C5 sites (e.g.
Cα-C5, Cβ-C5, and 5-5) even when carbohydrates were completely hydrolyzed. These unique C-
C condensation mechanisms were reported for the first time with straightforward proofs in
causing the significant retention of undissolved softwood lignin in the reactor as solid residues
and the difficulties of removing softwood lignin by dilute acid pretreatment. However, under
identical conditions, the poplar lignin was completely solubilized. Interestingly, pretreated
softwood hydrolysate contained more monomeric sugars and less oligomeric sugars than that of
hardwood under the same reaction conditions. The pretreated softwood whole slurries showed
decent enzymatic digestibility. The insights in different chemistry of softwood and hardwood
during dilute acid flowthrough pretreatment contribute to the fundamental understanding of
biomass pretreatment technology.
7.6 Acknowledgements
We are grateful to the Sun Grant-DOT Award # T0013G-A- Task 8, DOE-EERE Award #
DE-EE0006112 for funding this research. We acknowledge the Bioproducts, Sciences and
Engineering Laboratory, Department of Biosystems Engineering at Washington State University.
206
L. Zhang was partially supported by the Chinese Scholarship Council (CSC). Part of this work
was conducted at the William R. Wiley Environmental Molecular Sciences Laboratory (EMSL),
a national scientific user facility located at the Pacific Northwest National Laboratory (PNNL)
and sponsored by the Department of Energy’s Office of Biological and Environmental Research
(BER). We also thank Dr. Haisheng Pei and Ms. Marie S. Swita for insightful discussions.
We acknowledge the Bioproducts, Sciences and Engineering Laboratory, Department of
Biosystems Engineering at Washington State University. L. Zhang was partially supported by the
Chinese Scholarship Council (CSC). Part of this work was conducted at the William R. Wiley
Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility located
at the Pacific Northwest National Laboratory (PNNL) and sponsored by the Department of
Energy’s Office of Biological and Environmental Research (BER).
7.7 Supplementary materials
Table S7.1 Enzymatic hydrolysis of flowthrough pretreated pine wood whole slurries
Pretreatment conditions Enzyme loading
(mg/g biomass solid residues)
Glucose yield (% of
original glucosea)
Temp. t H2SO4 LogR0 72h
220 2 0.05 3.83 20 70.9
100 73.6
220 4 0.05 4.14 20 75.5
100 80.1
220 6 0.05 4.31 20 85.7
100 91.7
220 8 0.05 4.44 20 87.7
100 92.6
220 10 0.05 4.53 20 85.9
100 94.6
240 2 0.05 4.61 20 85.8
100 93.7
240 4 0.05 4.72 20 87.4
100 90.8
240 6 0.05 4.90 20 90.4
100 92.7
240 8 0.05 5.03 20 90.9
207
100 96.7
240 10 0.05 5.12 20 92.1
100 97.2
270 2 0.05 5.20 20 82.4
100 90.7
270 4 0.05 5.61 20 80.7
100 89.5
270 6 0.05 5.78 20 79.4
100 88.9
270 8 0.05 5.91 20 80.4
100 86.7
270 10 0.05 6.01 20 75.8
100 82.1
a based on original glucose content in loaded biomass; Temp. refers to the pretreatment
temperature (°C), t refers to the pretreatment time (min), and H2SO4 refers to the concentration
of dilute acid, % (w/w). 20 mg protein Ctec2 (12 FPU) with 4 mg protein Htec2/g glucan +
xylan; 100 mg protein Ctec 2 (60 FPU) with 20mg Htec2/g glucan + xylan.
Pretreated pine whole slurry, which contained solid residues and pretreatment liquid
fractions was hydrolyzed by enzymes to test their digestibility for glucose generation. The
enzyme loading employed in this study was 100 mg protein CTec2 with 20 mg HTec2/g
glucan + xylan and 20 mg protein Ctec2 with 4 mg protein Htec2/g glucan + xylan,
respectively. Table S7.1 shows enzymatic glucose yields from pretreated pine reached in
the range of 70% to 97% (LogR0 from 3.83 to 6.01, 220oC-270oC, 2-10mins, 0.05%
(w/w) acid) after 72h enzymatic hydrolysis. This elevated efficiency of glucose
generation from the dilute sulfuric acid flowthough pretreatment could be attributed to the
high yield of monomeric carbohydrates and low inhibitors in the pretreated whole slurry
401. The total glucose yield increased along with the increasing of pretreatment severities
until reaching more than 92% with only 12FPU enzymes at the pretreatment severity of
5.12 (240oC, 10mins, 0.05% (w/w) acid). However, when pretreatment severities were
208
higher than 5.12, the glucose yields started decreasing which might be attributed to the
generation of more inhibitors to enzymatic hydrolysis under higher severities.
209
Table S7.2 Assignments of main lignin 1H- 13C cross-peaks in the HSQC Spectra of the
pretreatment recovered insoluble lignin 144, 157, 368-371
Lignin linkages and monolignols Chemical shift
-OCH3 55.47(C) 3.70(H)
A: β-O-4 71.70(S-Cα) 4.84(S-Hα) 59.50-59.70(Cγ) 3.40-3.63(Hγ)
85.90(S-Cβ) 4.09(S-Hβ) 83.49(G/H-Cβ) 4.28(G/H-Hβ)
B: resinol 84.85(Cα) 4.62(Hα) 53.30(Cβ) 3.05(Hβ) 70.85(Cγ)
4.14/3.78(Hγ)
C: phenylcoumaran 86.79(Cα) 5.41(Hα) 53.3(Cβ) 3.46(Hβ) 62.52(Cγ) 3.68(Hγ)
D: spirodienone 59.67(Cβ) 3.19(Hβ)
G: guaiacyl 111.02(C2) 6.95(H2) 115.05(G5) 6.74(H5) 119.01(G6)
6.78(H6)
G: oxidized (Cα=O) guaiacyl 111.56(C2) 7.50(H2) 123.55(C6) 7.54(H6)
S: syringyl 103.95(C2/6) 6.67(H2/6)
S: oxidized (Cα=O) syringyl 106.52(C2/6) 7.29(H2/6)
E: p-hydroxybenzoate 131.33(C2/6) 7.62(H2/6)
F: cinnamyl alcohol 128.39(Cβ) 6.20(Hβ) 128.59(Cα) 6.42(Hα)
K: cinnamaldehyde 126.23(Cβ) 6.74(Hβ)
a Note: G, S, H-C or G, S, H-H refers to C and H in the lignin sub-units, guaiacyl, syringyl and p-
hydroxyphenyl.
210
Figure S7.1 Assigned interunit linkages of lignin, including different side-chain linkages, and
aromatic units: (A) β-O-4 aryl ether linkages; (B) resinol substructures (β-β′, α-O-γ′, and γ-O-α′
linkages); (C) phenylcoumarane substructures (β-5′and α-O-4′ linkages); (D) spirodienone
substructures (β-1′ and α-O-α′ linkages); (G) guaiacyl units; (G′) oxidized guaiacyl units with an
Cα-ketone; (S) syringyl units; (S′) oxidized syringyl units with a Cα ketone; (H) hydroxyphenyl
groups; (E) p-hydroxybenzoate substructures; (K) cinnamaldehyde end groups
211
Figure S7.2 GPC analysis of lignin molecular weight of a) poplar flowthrough pretreatment
recovered insoluble lignin, b) pine wood flowthrough pretreatment recovered insoluble
lignin, c) pine wood flowthrough derived residual lignin under 240°C with 0.05% (w/w) H2SO4
for 10min
0.0
0.2
0.4
0.6
0.8
Molar mass [Da]
1*102
1*103
1*104
W(lo
g M
)
PS
S W
inG
PC
Un
iCh
rom
, B
uild
48
15
, P
C1
01
11
3,
Insta
nce
#1
Sample :Integration from:Integration to :Calibration File : Eluent :MHK - A (Cal.): MHK - K (Cal.):Int.stand.-cal.: Int.stand.-sam.:Pump : Flowrate :Concentration : Inject volume :Column 1 : Temperature :Column 2 : Temperature :Column 3 : Temperature :Column 4 : Temperature :Detector 1 : Delay volume :Detector 2 : Delay volume :Operator : Acquisition interval :
Vial 14: Pu3'-240A-HW-FL - 1Thursday 11/12/15 21:04:43 27.323 mlThursday 11/12/15 21:18:42 41.309 mlYoo_09232015.CAL THF 0.000E+0 1.000E+0 ml/g 50.000 ml ------- mlPSS SECcurity 1.000 ml/min 1.000 g/l 25.000 ulWaters Styragel HR6 30.000 °CWaters Styragel HR4 30.000 °CWaters Styragel HR1 30.000 °CWaters Styragel HR0.5 30.000 °CPSS SECcurity UV 0.000 mlPSS SECcurity RI 0.006 mlAdministrator 1.000 sec
PSS SECcurity UV
Mn :
Mw :
Mz :
Mv :
D :
[n]:
Vp :
Mp :
A :
< 37
w% :
> 55717
6.8168e2
2.0745e3
5.8308e3
0.000000
3.0432e0
0.000000
3.3326e1
1.2840e3
3.1425e0
0.00
100.00
0.00
g/mol
g/mol
g/mol
g/mol
ml/g
ml
g/mol
ml*V
Project : Account :Date : Sign :
C:\ORNL GPC\Data\Mi Li\Mi Li.LDXTuesday 11/17/15 15:10:25
0.0
0.2
0.4
0.6
0.8
Molar mass [Da]
1*102
1*103
1*104
W(lo
g M
)
PS
S W
inG
PC
Un
iCh
rom
, B
uild
48
15
, P
C1
01
11
3,
Insta
nce
#1
Sample :Integration from:Integration to :Calibration File : Eluent :MHK - A (Cal.): MHK - K (Cal.):Int.stand.-cal.: Int.stand.-sam.:Pump : Flowrate :Concentration : Inject volume :Column 1 : Temperature :Column 2 : Temperature :Column 3 : Temperature :Column 4 : Temperature :Detector 1 : Delay volume :Detector 2 : Delay volume :Operator : Acquisition interval :
Vial 12: Pu2-240A-SW-FL - 1Thursday 11/12/15 19:22:27 28.217 mlThursday 11/12/15 19:35:38 41.390 mlYoo_09232015.CAL THF 0.000E+0 1.000E+0 ml/g 50.000 ml ------- mlPSS SECcurity 1.000 ml/min 1.000 g/l 25.000 ulWaters Styragel HR6 30.000 °CWaters Styragel HR4 30.000 °CWaters Styragel HR1 30.000 °CWaters Styragel HR0.5 30.000 °CPSS SECcurity UV 0.000 mlPSS SECcurity RI 0.006 mlAdministrator 1.000 sec
PSS SECcurity UV
Mn :
Mw :
Mz :
Mv :
D :
[n]:
Vp :
Mp :
A :
< 36
w% :
> 30194
5.4193e2
1.5560e3
3.8169e3
0.000000
2.8712e0
0.000000
3.3602e1
1.1133e3
2.1867e0
0.00
100.00
0.00
g/mol
g/mol
g/mol
g/mol
ml/g
ml
g/mol
ml*V
Project : Account :Date : Sign :
C:\ORNL GPC\Data\Mi Li\Mi Li.LDXTuesday 11/17/15 15:09:38
0.0
0.2
0.4
0.6
0.8
Molar mass [Da]
1*102
1*103
1*104
W(log
M)
PS
S W
inG
PC
UniC
hro
m,
Bu
ild 4
81
5,
PC
10
11
13
, In
sta
nce
#1
Sample :Integration from:Integration to :Calibration File : Eluent :MHK - A (Cal.): MHK - K (Cal.):Int.stand.-cal.: Int.stand.-sam.:Pump : Flowrate :Concentration : Inject volume :Column 1 : Temperature :Column 2 : Temperature :Column 3 : Temperature :Column 4 : Temperature :Detector 1 : Delay volume :Detector 2 : Delay volume :Operator : Acquisition interval :
Vial 15: Pu4-SWSR-240A - 1Thursday 11/12/15 21:58:42 29.681 mlThursday 11/12/15 22:10:35 41.553 mlYoo_09232015.CAL THF 0.000E+0 1.000E+0 ml/g 50.000 ml ------- mlPSS SECcurity 1.000 ml/min 1.000 g/l 25.000 ulWaters Styragel HR6 30.000 °CWaters Styragel HR4 30.000 °CWaters Styragel HR1 30.000 °CWaters Styragel HR0.5 30.000 °CPSS SECcurity UV 0.000 mlPSS SECcurity RI 0.006 mlAdministrator 1.000 sec
PSS SECcurity UV
Mn :
Mw :
Mz :
Mv :
D :
[n]:
Vp :
Mp :
A :
< 33
w% :
> 11340
2.1706e2
1.1282e3
3.3035e3
0.000000
5.1973e0
0.000000
3.7832e1
1.6854e2
7.253e-2
0.00
100.00
0.00
g/mol
g/mol
g/mol
g/mol
ml/g
ml
g/mol
ml*V
Project : Account :Date : Sign :
C:\ORNL GPC\Data\Mi Li\Mi Li.LDXTuesday 11/17/15 15:11:51
a) b)
c)
212
CHAPTER EIGHT
HOT WATER AND DILUTE ACID PRETREATMENT OF SUB-
MILLIMETER BIOMASS PARTICLES
8.1 Abstract
The hot water and dilute acid pretreatment studies of a variety of feedstock with different
shapes and sizes are significant towards the optimal sub-millimeter biomass particles for sugars
production at low cost. Physical modifications on biomass, such as milling and cutting, result in
different biomass shapes and sizes, thus have significant impact on pretreatment performances
and sequential enzymatic hydrolysis sugar yields. The kinetics of hot water-only and dilute acid
pretreatments considering mass transfer effects and the following enzymatic hydrolysis process
were evaluated using hardwood (e.g. poplar wood) and softwood (e.g. Douglas fir) substrates
with four different particle sizes by, two different cutting approaches, hammer milling and
crumbling (e.g. Shards, Cubic prismatic). Results showed that different biomass substrates,
particle sizes, cutting approaches and pretreatment methods (e.g. hot water, and dilute acid) led
to different pretreatment performances. Interestingly, Biomass particles with 12-30/16-40 mesh
size instead of finer particle size led to a better total sugar release and pretreatment lignin
removal. Also the crumbled biomass chips showed comparable pretreatment and enzymatic
hydrolysis sugar yields but consumed three times less energy in comparison with the traditional
hammer milling method. To further explore the mass transfer effects of particle sizes and cut
methods on pretreatment effectiveness, the measurements of the mass transfer coefficients were
proposed using reported and designed devices. It was proposed that various particle sizes and
cutting methods on biomass resulted in their different mass transfer capacities (coefficients) and
213
the pretreatment performances were corresponded to the mass transfer diffusion capacities of
biomass. However, more future work is needed in order to provide more conclusions and
evidences. This work is significant to the preprocessing of feedstock for aqueous pretreatment.
Keywords: Biomass particles • Size • Biomass cutting • Pretreatment • Mass transfer • Diffusion
coefficient
8.2 Introduction
Cellulosic biomass is low in cost, with biomass at $60/dry ton about the same cost as
petroleum at $20/barrel on the basis of equivalent energy content 169, 402. Furthermore,
technologies are well established and low in cost for fermenting sugars that can be derived from
cellulose and hemicellulose in biomass into ethanol and other products 403. The challenge to
lower costs has been to achieve high yields of sugars from biomass with cost competitive
technologies. In other words, we must find ways to overcome the biomass recalcitrance as a
barrier to economic production of fuels from biomass 404, 405. The costs of pretreatment and
enzymes are the primary barriers to release sugars with low cost, and it is vital to apply low cost
pretreatments to open up the structure of biomass to enzymes 32, 169, 405. In particular,
pretreatments are needed to achieve high sugar yields from cellulose and hemicellulose through
the coupled operations of pretreatment and enzymatic hydrolysis while requiring little if any
chemical inputs, minimal need for inhibitor removal, low sugar losses in
pretreatment/conditioning, low energy demands, low operating pressures, short reaction times,
inexpensive materials of construction, and simple operation to avoid expenses for chemical
recovery and recycle 32, 406, 407. In light of these demands, only pretreatments that require none or
minimal chemical additions can provide the lowest cost route provided yields from pretreatment
214
and enzymatic hydrolysis are high. Thus, aqueous pretreatments that only use steam or hot water
are preferred, backed up with dilute sulfuric acid approaches if aqueous yields are too low to be
cost competitive. High sugar yields must be achieved with low enzyme loadings for these or any
other pretreatment options to be economically viable 32.
Mechanical treatments, such as chipping, milling and grinding, were commonly used to
reduce the size of the particles, essentially increasing the surface area of the lignocellulose before
aqueous pretreatment 408. The size reduction can influence the pretreatment efficiency by
producing different biomass particle sizes and shapes via various cutting methods, and affect
pretreatment and total cost. The size reduction can also influence enzymatic hydrolysis
efficiency. For example, the cellulose crystalline structure and its degree of polymerization (DP)
might be reduced during mechanical treatment. Hammer mills and ball mills were commonly
used to reduce particle size and have been demonstrated to enhance the digestibility of the
lignocellulose in enzymatic hydrolysis 409. However, high energy consumption of mechanical
pretreatment hindered its large scale application 410. Thus, developing low energy mechanical
processes for biomass size reduction is essential for economic biorefinery production 411.
The impact of biomass mechanical prehandling on biorefinery processes was not widely
studied although it has been reported before. Some studies focus on the effects of sub-millimeter
substrates on the thermal conversion. For example, the effects of sub-millimeter scale and
macroscopic wood particles on the pyrolysis under different conditions were determined 412-414.
Also, some studies showed the effects of biomass particle characteristics on enzymatic
hydrolysis efficiency 161, 415. However, the impact of particle size reduction on pretreatment
performance is not well understood yet 416. Some studies have been carried out to study effects of
215
physical treatments or impregnations of lignocellulosic biomass on wood structure alteration and
solvent mass transfer, which influenced subsequent pretreatment performance 168, 417, 418. For
example, a diffusivity data of different biomass from batch pretreatment was fitted into the
Arrhenius equation for extrapolation to higher temperatures, which was subsequently
incorporated into a theoretical model to determine acid concentration profiles within biomass
matrix 160. Additionally, a study developed a model based on the diffusion of liquid into biomass
by considering the effects of the particle size to reduce energy consumption 419. Biomass
pretreatment was typically based on particle size of 10 mm or less 420, 421. However, the effects of
sub-millimeter scale biomass particle size on pretreatment diffusion have not been reported 422,
not mentioning the effects of biomass cutting approaches such as crumbling (Figure S8.1) and
hammer-milling.
The study of effects of biomass particles with sub-millimeter scales on mass transfer
diffusion in aqueous based pretreatment is critical to achieve the large scale production of
biomass pretreatment. Bench-top batch tubular reactors were widely applied in many
pretreatment studies for a long time 423. The gaps between bench-top batch reactors and pilot-
scale reactors are mass and heat transfer issues 424. However, few pretreatment studies focused
on resolving mass and heat transfer in literature. Also, most pretreatment studies apply relatively
lower solid loadings than industrial scale, which is an issue directly related to mass transfer
effects. The high solid loading with excellent mass transfer is usually preferred in industrial
application. First of all, high solid loading offers many advantages, such as reducing production
and capital costs and increased sugar and ethanol concentrations for downstream processes.
Secondly, a consideration of high solid loading are particle shapes and sizes which have been
216
reported to have a significant impact on the viscosity of pretreatment slurry because particle
properties influence the particle networking and packing in bulk slurry 425-427. However, high
viscosity of pretreated slurry will cause mixing difficulties as well as the problems of solvent
pumping and flowing, which leads to poor mass transfer. At last, biphasic reaction rate behaviors
of biomass dissolution might be also attributed to mass transfer limitation as reported 428.
Therefore, the motivations of understanding impacts of particle sizes and cutting shapes of
substrates in pretreatment mass transfer and diffusion are critical for a successful large-scale
production.
The objective of this chapter is to identify the most promising substrate particle sizes and
cutting approaches (e.g. Shards, Cubic prismatic) through comparative study of water only and
dilute acid pretreatment as well as the digestibility of pretreated biomass particles to generate
sugars. Also, investigate mass transfer differences during pretreatment among biomass substrates
with different particle sizes and by different cutting approaches.
8.3 Materials and methods
8.3.1 Materials
All samples used in this study were provided by the Forest Concepts Corporation. Poplar
chips, Douglas fir and corn stover were cut through two approaches, namely, crumbling and
hammer milling. Specifically, poplar and corn stover underwent 0.8 mm hammer milling and
passed 16 mesh but not 40 mesh, which formed 16-40 mesh shards shape of poplar chips and
corn stover. Besides, polar wood and corn stover went through 0.8 mm crumbling and were
orbital screened by passing 16 mesh but not 40 mesh, which formed 16-40 mesh cubic prismatic
poplar chips and corn sotver. In addition, poplar was crumbled by 1.6 mm crumble and passed
217
1/8 inch sieving and no pass 16 mesh, which formed 1/8′′-16 mesh cubic prismatic poplar chips
and corn stover. Thus, the hammer milling and crumbling resulted in different particle sizes and
shapes. Table 8.1 listed the details of feedstock used in the study. The composition analysis was
carried out to determine the contents of glucan, xylan and lignin in the feedstocks according to
NREL procedure -Determination of Structural Carbohydrates and Lignin in Biomass (NREL/TP-
510-42618). Poplar contained 43.25±0.32% of glucan, 17.25±0.52% of xylan, 26.12±0.21% of
lignin and 1.16±0.04% of ash. Corn stover contained 33.60±0.31% of glucan, 20.71±0.24% of
xylan, 18.44±0.38% of lignin and 5.50±0.03% of ash. Douglas fir presented 45.55±1.10% of
glucan, 5.14±0.14% of xylan, 31.16±0.12 of lignin, 7.81±0.45% of galactan, 8.66±0.24% of
arabinan and mannan, and 0.44±0.02% of ash.
Table 8.1 Characteristics of feedstock used in the study
Biomass Size Cutting approaches Sieve size Particle geometry
Poplar wood 16-40 mesh 0.8mm Hammer milled Pass 16 mesh
No Pass 40 mesh
Shards
Poplar wood 16-40 mesh 0.8mm Crumbled Pass 16 mesh
No Pass 40 mesh
Cubic prismatic
Poplar wood 1/8′′-16 mesh 1.6mm Crumbled Pass 1/8 inch
No Pass 16 mesh
Cubic prismatic
Corn Stover 16-40 mesh 0.8mm Hammer Milled Pass 16 mesh
No Pass 40 mesh
Shards
Corn Stover 16-40 mesh 0.8mm Crumbled Pass 16 mesh
No Pass 40 mesh
Cubic prismatic
Corn Stover 1/8′′-16 mesh 1.6mm Crumbled Pass 1/8 inch
No Pass 16 mesh
Cubic prismatic
Biomass Size Cutting approaches Sieve size Particle geometry
Green poplar 30-60 mesh Hammer milled N/A Shards
Green poplar 12-30 mesh Hammer milled N/A Shards
Green poplar 1/8′′-16 mesh Hammer milled N/A Shards
218
Green poplar 30-60 mesh Crumbled N/A Cubic prismatic
Green poplar 12-30 mesh Crumbled N/A Cubic prismatic
Green poplar 1/8′′-16 mesh Crumbled N/A Cubic prismatic
Douglas fir 30-60 mesh Hammer milled N/A Shards
Douglas fir 12-30 mesh Hammer milled N/A Shards
Douglas fir 1/8′′-16 mesh Hammer milled N/A Shards
Douglas fir 30-60 mesh Crumbled N/A Cubic prismatic
Douglas fir 12-30 mesh Crumbled N/A Cubic prismatic
Douglas fir 1/8′′-16 mesh Crumbled N/A Cubic prismatic
8.3.2 Batch tubular pretreatment
Batch pretreatment conditions were selected based on water only and dilute acid conditions
that were reported to achieve a better enzymatic hydrolysis yield as presented in Table 8.2.
Pretreatment was carried out in a fluidized sand bath system (model SBL-2D, Omega
engineering, Inc., CT) equipped with a temperature controller. Tubular reactors were made as 1.3
cm diameter ×15.2 cm long with volume of 13.4 ml. 0.5 g feedstock was loaded into the batch
tubular reactor, which was followed by an addition of 10 ml water or 1% sulfuric acid. Both
reactor ends were sealed tightly with Teflon plugs and stainless cups. Reactors were gently
placed in the heated sand bath with a temperature controller setting of desired temperature
monitored by a thermometer. The pretreated solid residues and liquors were separated through
filtration, and 2 ml liquid fraction was analyzed using HPLC to determine mono sugars in
hydrolysates. The Waters HPLC system (model 2695) is equipped with a 410 refractive detector
and an auto-sampler (Waters 2695) using a Waters Empower Build 1154 software (Waters Co.,
Milford, MA, USA). A Bio-Rad Aminex HPX-87H column (Bio-Rad Laboratories, Hercules,
CA, USA) was operated under 60°C. The liquid fraction was further post hydrolyzed by the 4%
219
of sulfuric acid in 121ºC autoclave to determining the amount of oligomeric sugars in
hydrolysates after pretreatment following by a NREL procedure 429. The solid residues were
freeze dried and experienced a solid residue composition analysis following by NREL procedure
353.
Table 8.2 Selected pretreatment conditions performed in the study
Feedstock Pretreatment
methods T (℃) Residence time
(min)
Chemical
(%)
Solid loading
(w/w)
Log R0 Reference
Poplar/corn stover Dilute acid 140 10,20,30,40,50 1% H2SO4 5% 2.1-2.9 423
poplar/corn stover Water only 190 5,10,15,25,30 N/A 5% 3.3-4.0 430
Douglas fir Dilute acid 140-
200
3-70 1% H2SO4 5% 2.6-3.7 N/A
Douglas fir Water only 190-
220
1-70 N/A 5% 3.6-4.5 N/A
Poplar wood was pretreated at 140°C with ~1% wt sulfuric acid for 40min and 190°C with
DI water for 15min, yielding nearly 100% of xylose recovery 431, 432. Therefore, in our study,
dilute acid pretreatment of poplar wood was carried out with 1% (w/w) H2SO4 for 10, 20, 30, 40,
and 50 min at temperature of 140°C (Log R0=2.1-2.9), and hot water pretreatment was
performed with DI water for 5, 10, 15, 20, and 25min at temperature of 190°C (Log R0=3.3-4.0).
However, no comparable data could be found in softwood acid/hot water pretreatment under the
same pretreatment conditions. In this study, we applied conditions similar to these for poplar
wood on Douglas fir and evaluate the softwood pretreatment sugar yields. The selection of the
pretreatment conditions for Douglas fir was based on literature reviews and the criteria of more
than 95% of hemicellulose recovery in most of Biomass Refining Consortium for Applied
Fundamentals and Innovation (CAFI) pretreatment methods. Because the softwood pretreatment
is reported challenging, especially in delignification, this study applied reaction times from 30 to
70 mins (pretreatment severities at 2.6-3.7). In addition to 140°C, 180°C was applied to pretreat
220
Douglas fir for 3-20min with controlled severities at the same range 2.6-3.7 in order to determine
the impact of temperatures. Besides, the current softwood pretreatment is mainly performed by
dilute acid or acid/SO2 catalyzed steam explosion at temperature of about 200°C 300, 380, 433. Thus,
190 and 200°C dilute acid were selected to pretreat Douglas fir with 0.5% wt sulfuric acid for 2-
10min (severities at 3.6-4.5). In summary, dilute acid pretreatment of Douglas fir was performed
with 0.5, or 1% (w/w) H2SO4, for 3-70 min at 140-200°C (Log R0= 2.6-3.7). Similarly, hot water
pretreatment conditions were selected using DI water for 1-70min at 190-220°C (Log R0=3.6-
4.5).
8.3.3 Digestibility evaluation of pretreated whole slurries by enzymatic hydrolysis
The whole slurries of pretreated samples including hydrolysates and solid residues were
hydrolyzed with CCTEC cellulase (Genencor, Palo Alto, CA, USA) and HTEC2 Novozyme 188
β-glucosidase (Novozyme, Franklinton, NC, USA)- Novozymes Cellic® CTec1 (220mg
protein/mL, preserve 200mg glucose/mL, 93 FPU/mL) and Novozymes Cellic® HTec2 (230mg
protein/mL, preserve 180mg xylose/mL). All enzymatic hydrolysis experiments were conducted
in duplicates under the standard conditions (50 °C, 0.05 M citric acid buffer at pH 4.8). The pH
of the pretreated whole slurries from batch reactors was adjusted to pH4.8 with 0.1N NaOH. A
mixture of CCTEC2 and HTEC2 enzymes at a ratio of 5:1 based on protein weight was added
with enzyme loading of 20, 100, or 200 mg protein Ctec2 (93 FPU) with 4, 20, and 40 mg
Htec2/g (glucan + xylan). In the study, 10 mg/L sodium azide and 1% (w/v) boving serum
albumin (BSA) were added. The resulting liquid samples were withdrawn at 72 hours and
injected to HPLC for monomeric sugars analysis. Yields of enzymatic hydrolysis can be
evaluated through the following equations (see equation 8.1, 8.2).
221
Enzymatic glucose yield% =WG2−WG1
WTG× 100% (8.1)
Enzymatic xylose yield% =WX2−WX1
WTX× 100% (8.2)
In these equations, WG1 and WX1 are the glucose and xylose released in the pretreatment (stage
1); WG2 and WX2 are the glucose and xylose released in enzymatic hydrolysis (stage 2); WTG and
WTX are the total theoretical glucose and xylose released (glucose and xylose in untreated
biomass). The unit of W consistently refers to g/100g DW raw biomass.
8.3.4 Kinetic modeling of biomass degradation
A leaching model was reported to depict the degradation behaviors of xylan, glucan and
lignin during the pretreatment process.
(8.3)
where As is the unit surface area of biomass particles (cm2/g), Vc is the unit volume of the solid
(cm3/g), Xn is dissolved xylan (g), kM is the mass transfer coefficient (cm/min).
(8.4)
where As is the unit surface area of the particles (cm2/g), Vc is the unit volume of the solid
(cm3/g), C is dissolved glucan (g), kM is the mass transfer coefficient (cm/min).
(8.5)
MkXylan Dissolved Xylan
( )sM
c
AdXnk Xn
dt V
MkCellulose Dissolved Cellulose
( )sM
c
AdCk C
dt V
MkLignin Dissolved Lignin
( )sM
c
AdLk L
dt V
222
where As is the unit surface area of the particles (cm2/g), Vc is the unit volume of the solid
(cm3/g), L is dissolved lignin (g), kM is the mass transfer coefficient (cm/min) 434. After
integration, equations applied in our study can be presented as following (See equation 8.6, 7, 8).
G = 100e(−KM
AsVc
)t (8.6)
H = 100e(−KM
AsVc
)t (8.7)
L = 100e(−KM
AsVc
)t (8.8)
here G, H, and L refer to the percentage of solid residues after pretreatment in untreated biomass
(unit: %).
The surface area of feedstocks in this study was measured by exposing biomass particles to
nitrogen gas after drying samples under 200ºC for 1 hour. The absorption of nitrogen was
determined. The surface area was calculated by applying Brunauer, Emmett and Teller-BET
hypothesis equations 435. The volume of solids is measured by packing 5g biomass in a graduated
cylinder. The volume occupied by loaded samples was read and divided by 5g to get unit volume
of biomass solids (m3/g) 436. The volume of solids refers to the biomass density which the ratio of
the mass of a quantity of a substance to its volume and was expressed in terms of weight per unit
volume. Table 8.3 showed that crumbled particles and smaller sizes have smaller density than
hammer milled particles and large sizes, respectively.
Table 8.3 Some unit volume of biomass solids (biomass density)
Sample information Volume of solids, cm3/g
30-60 mesh, Cubic prismatic, poplar 6.2
16-40 mesh Shards, Poplar 7.2
16-40 mesh Cubic prismatic, Poplar 6.8
12-30 mesh, Cubic prismatic, poplar 6.8
12-30 mesh, Shards, poplar 7
223
8.4 Results and discussion
8.4.1 Mass transfer modeling of hot water and dilute acid pretreatment of corn stover and
poplar wood
The surface area of biomass particles is important in chemical kinetics. The surface area of
corn stover was 2.8m2/g and poplar had 0.7 m2/g surface area (BET). The surface area of a solid
object is a measure of the total area that the surface of an object occupies. Here surface area
refers to internal surface area of particles accessible for chemical interactions. Increasing the
surface area of a substance generally increases the rate of a chemical reaction. Results showed
that corn stover has a higher surface area than poplar, which may be attributed to a denser wood
structure of polar than corn stover structure. Corn stover particles are more loosen and shows
more accessible surface than that of poplar wood chips. Volume of solids was calculated as
shown in Table 8.3. Particle sizes and cutting approaches both affect solid volume. The solid
volume is a factor that impacts the mass transfer capacities. Smaller and Cubic prismatic -shaped
particles can occupy space efficiently which gave smaller solid volumes. Both biomass surface
area and density factors are significant factors in kinetic modeling considering mass transfer (See
equation 8.6, 8.7, 8.8) 434. The different mass transfer capacities of biomass particles with various
cutting methods and particle sizes might be caused by the different volume of solids.
1/8′′-16 mesh, Shards, Poplar 8.4
1/8′′-16 mesh Cubic prismatic, Poplar 7.2
16-40 mesh Shards, Corn stover 7.2
16-40 mesh Cubic prismatic, Corn stover 6.8
1/8′′-16 mesh Cubic prismatic, Corn stover 7.2
1/8′′-16 mesh, Cubic prismatic, Douglas fir 6.4
12-30 mesh, Shards, Douglas fir 7.2
225
Figure 8.1 Kinetic diffusion modeling (equation 8.6, 7, 8) of pretreatment behaviors of major
components in six feedstock 1% (w/w) H2SO4 pretreatment at 140ºC with residence time of 0-50
min including (a) xylan dissolution, (b) glucan dissolution, and (c) lignin dissolution, and hot
water pretreatment at temperature of 190ºC with residence time of 0-30 min including (d) xylan
dissolution, (e) glucan dissolution, and (f) lignin dissolution
Experimental results in Figure 8.1 (plotted with different markers) showed the effects of
particle sizes and cutting approaches on major biomass components removal. Results showed
that different particle sizes and cutting approaches resulted in different removal pattern of
biomass components (i.e. cellulose, hemicellulose, and lignin) during pretreatment. Since
biomass structure is porous and heterogeneities 161, solvent diffusion during pretreatment was
influenced by the size of particles and cutting approaches of biomass substrates. Meanwhile, our
experimental results were consistent with reported statements that different physical
modifications of the biomass (e.g. size reductions and particle cutting methods) which have
different levels of compression and shear forces affected pretreatment performance 416. Results of
(e) (f)
226
hot water and dilute acid pretreatment of Poplar and corn stover indicated that the release of
glucan, xylan and lignin follows the leaching model with consideration of mass transfer effects
(See equation 8.6, 7, 8) 434. Experimental data for glucan, xylan and lignin recovery and removal
percentages in hot water and dilute acid pretreatment were fitted into equation 8.6, 8.7, and 8.8,
which is shown in Figure 8.1. Figure 8.1 shows the experimental and modeling fitting results of
xylan, glucan, and lignin removal for some of the samples by hot water and dilute acid
pretreatment. The accuracy of modeling results in this study fitting to experiments is acceptable
with average chi square of about 50. The differences of fitting results with the experimental data
points were maybe corresponded to the complicated biomass dissolution mechanisms which
cannot be simply simulated by a kinetic model that solely aims at mass transfer impacts.
However, for this study, mass transfer is the key factor to evaluate the effects of particle size and
cutting approaches. Therefore, such kinetic modeling simulation would reveal insights for the
purpose of this study.
8.4.2 Effects of particle sizes and cutting approaches on mass transfer coefficients
The effects of particle sizes and cutting approaches on mass transfer of poplar and corn
stover during hot water and dilute acid pretreatment were investigated by integration of
experimental data into the mass transfer leaching model described in the method section. In
equations 8.6, 8.7, and 8.8, KM was calculated by integrating measured AS, and VC into fitting
equations. Table 8.4 showed mass transfer coefficients for the selected six biomass samples
performed with two different pretreatment methods under different pretreatment conditions.
Table 8.4 Kinetic modeling parameters (physical mass transfer coefficients) of xylan, glucan,
and lignin removal with consideration of mass transfer (a) 1% sulfuric acid pretreatment with
227
residence time of 0 to50 min at 140ºC and (b) hot water pretreatment with residence time of 0 to
30 min at 190ºC
(a)
Sample information Mass transfer coefficients (KM
)
Xylan removal Glucan removal Lignin removal
16-40 mesh Shards, Poplar 0.0391 0.0050 0.0062
16-40 mesh Cubic prismatic, Poplar 0.0298 0.0035 0.0044
1/8′′-16 mesh Cubic prismatic, Poplar 0.0286 0.0029 0.0047
16-40 mesh Shards, Corn stover 0.3394 0.0484 0.0505
16-40 mesh Cubic prismatic, Corn stover 0.2474 0.0305 0.0412
1/8′′-16 mesh Cubic prismatic, Corn stover 0.2229 0.0293 0.0376
(b)
Sample information Mass transfer coefficients (KM
)
Xylan removal Glucan removal Lignin removal
16-40 mesh Shards, Poplar 0.0593 0.0179 0.0182
16-40 mesh Cubic prismatic, Poplar 0.0436 0.0089 0.0116
1/8′′-16 mesh Cubic prismatic, Poplar 0.0431 0.0084 0.0100
16-40 mesh Shards, Corn stover 0.4221 0.1269 0.1389
16-40 mesh Cubic prismatic, Corn stover 0.2863 0.0840 0.0827
1/8′′-16 mesh Cubic prismatic, Corn stover 0.2856 0.0825 0.0823
For dilute acid pretreatment at 140ºC, results showed that xylan was dissolved with the
highest mass transfer coefficients for all six samples than glucan and lignin (Table 8.4a, b). It
was consistent with previous reports that xylan was released in the initial phase of pretreatment
56. For all samples, mass transfer coefficients of lignin degradation appeared slightly to moderate
higher than those of glucan degradation. The explanation was amorphous structure was easy for
solvents to diffuse compared to dense crystalline cellulose structure 420. Meanwhile, another
reason for better lignin dissolution than glucan dissolution was probably attributed to the existing
of LCC (lignin carbohydrates complex) that deconstructed some lignin together with
hemicellulose dissolution. On the other hand, it was consistently found that in biomass
dissolution including glucan, xylan, and lignin degradation, medium particle sizes (16-40mesh)
228
showed better mass transfer during pretreatment process. Also the shard shape particles
presented some higher mass transfer coefficients. It might be because smaller biomass particle
has more accessible surface for solvents and shards cutting has a sharp easy ending for solvent
attack. However, the crumbling method can achieve comparable results with hammer milling
because of crumbled particles possessed smaller volume of solids. Results also indicated that
different cutting approaches have more significant effects on mass transfer coefficient than
particle size. Some studies reported the effects of particle size on pretreatment behaviors 411, 421,
434, 437, however, no study has reported effects of biomass cutting and resulted biomass shapes on
solvent diffusion and pretreatment. Our study showed that the cutting approaches were important
since they resulted in different shapes of particles which maybe determine the packing of particle
when loaded in tubular reactors. The packing influenced interaction between solvents and
biomass. For example, 16-40 mesh Shards Corn stover showed best mass transfer among other
samples with the value of 0.3394 cm/min. As different cutting approaches were applied, 16-40
mesh Shards poplar and corn stover showed higher mass transfer coefficients for all three
biomass main components than same particle size. The medium particles (16-40 mesh Shards) of
poplar and corn stover with same cutting approach (i.e. Crumbler) only resulted in slightly higher
mass transfer coefficients than larger particle size of 1/8-16 mesh. Interestingly, 16-40 mesh
Shards and Cubic prismatic poplar presented higher mass transfer coefficients than that of 1/8′′-
16 mesh Cubic prismatic Poplar which indicated the advantage of smaller particle size of 16-40
mesh (For example, 0.391 Versus 0.0298 and 0.0286 cm/min in xylan dissolution). Although the
mass transfer coefficients of hammer milled particiles were larger than crumbled particles, the
volume of solids of hammer milled was ~1.2 times of crumbled particles. Thus, according to
229
equations (8.6, 7, 8), the kinetic coefficients of crumbled particles were comparable to hammer
milled particles. These results indicated that both Cubic prismatic and Shards shaped particles
achieved comparable biomass dissolution for 1% sulfuric acid pretreatment at 140ºC.
For hot water pretreatment at 190ºC, mass transfer coefficients of glucan, xylan and lignin
degradation showed similar pattern as those for 1% sulfuric acid pretreatment at 140ºC. Xylan
was first dissolved with high mass transfer and degradation rate. Lignin deconstruction rate of
most samples had better mass transfer than glucan dissolution. Results suggested that particle
sizes play less significant role in hot water pretreatment at 190ºC than roles of biomass cutting
methods. For example, comparing 16-40 mesh Cubic prismatic poplar and corn stover with 1/8′′-
16 mesh Cubic prismatic poplar and corn stover, mass transfer coefficients of biomass
dissolution (i.e. glucan, xylan, and lignin)showed slight difference (For example, 0.2856 Versus
0.2863 cm/min in xylan dissolution). However, the cutting approaches appeared a more
determined factor as the mass coefficient increased from 0.2856 for 16-40 mesh Cubic prismatic
corn stover to 0.4221cm/min for the same particle size of 16-40 mesh corn stover but using
Shards approach. Biomass cutting methods were usually ignored in pretreatment. The study
sheds lights on understanding of significance of biomass cutting, not only for energy
consumption but also for pretreatment and enzymatic hydrolysis yields.
For all tested particle sizes and cutting approaches, corn stover showed much higher mass
transfer coefficients than poplar under both hot water and dilute sulfuric acid. Results indicated
better solvent diffusion into corn stover. For example, 1/8′′-16 mesh Cubic prismatic corn stover
showed 0.2229 cm/min coefficient for 1% sulfuric acid pretreatment at 140ºC, while similar
particle size poplar prepared by the same cutting approach only has mass transfer coefficient of
230
0.0286% cm/min. It can be corresponded to the less dense structure of corn stover compared to
poplar.
Comparing different pretreatment methods, all samples in hot water pretreatment (e.g.
190ºC) indicated better mass transfer coefficients for all three biomass main components (i.e.
cellulose, hemicellulose, and lignin) than those in dilute acid pretreatment at 140ºC. This
revealed advantageous solvent diffusion in 190ºC hot water pretreatment over dilute acid
pretreatment at 140ºC, especially for lignin deconstruction by hot water pretreatment at 190ºC,
mass transfer coefficients of all six samples were consistently higher than those by 1% sulfuric
acid pretreatment at 140ºC. For example, corn stover sample (16-40 mesh Shards, Corn stover)
presented 0.1389 cm/min mass transfer coefficient for hot water pretreatment at190ºC. However,
the mass transfer during lignin dissolution of the same corn stover sample only showed 0.0412
cm/min mass transfer coefficient by 1% sulfuric acid pretreatment at 140ºC. The phenomenon
was attributed to recondensation and repolymerization of dissolved lignin occurred at 140ºC for
1% sulfuric acid pretreatment while 190ºC hot water pretreatment showed less lignin
recondensation than that for 1% sulfuric acid pretreatmentat140ºC 438, 439.
8.4.3 Experimental results of hot water and dilute acid pretreatment of corn stover and
poplar wood
For dilute acid pretreatment at 140ºC (See Figure 8.1 (a,b,c)), xylan removal was the slowest
for the poplar sample with larger particles (1/8′′-16 mesh) and prepared by Cubic prismatic
cutting while it is the fastest for the corn stover sample with finer particle size (16-40 mesh) and
prepared by Shards cutting (Figure 8.1a). Smaller particle size has higher accessible surface area
which led to high accessibility of chemicals to biomass. For both poplar and corn stover samples,
231
when cutting approaches changed from Cubic prismatic to Shards, xylan removal was improved
5-10%. Shards cutting might result in sharp edges of biomass particles which were easy for
chemicals to interact. When the particle size decreased for Cubic prismatic cut samples, xylan
removal was increased to the slightly lower degree than changing cutting methods. Results
indicated that xylan removal was enhanced with smaller particle size. It was consistent with the
mass transfer coefficient’s increasing trend (Table 8.4). The sub-millimeter biomass pretreatment
behaviors were different from biomass pretreatment performance with conventional particle size.
It was reported larger steam-exploded particle (8-12 mm) resulted in higher cellulose and
enzymatic digestibility because biomass particle was broken down during pretreatment process
421, 440. Interestingly, our results showed the biomass removal and mass transfer coefficient
changed obviously with cutting approaches and only slightly with particle size. This indicated
the significant effects of cutting methods on pretreatment performance. Xylan removal was
altered slightly as these two factors changed. The explanations were that xylan was easy to be
degraded due to amorphous structure of xylan 56. On the other hand, xylan removal of corn
stover samples was insignificantly higher than that of poplar samples with similar particle size
and by similar cutting methods. This was correlated to the mass transfer coefficient differences
between poplar and corn stover. This was because corn stover was easier to be hydrolyzed than
poplar under dilute acid and hot water conditions 428, 437, 441. For glucan hydrolysis, glucan
removal was significantly improved 10-20% when cutting methods was changed from Cubic
prismatic to Shards, with more obvious effects with poplar than with corn stover. It might be
because wood is difficult to be pretreated and slightly positive change of mass transfer can
enhance obvious pretreatment performance. As particle size became finer, both poplar and corn
232
stover showed about 5% higher glucan removal. Similar to xylan removal, corn stover samples
resulted in higher glucan removal than poplar did. Glucan removal results correlated strongly
with the mass transfer coefficients (Table 8.4) which indicated that mass transfer could be a key
factor to glucan removal. Similar to glucan, lignin removal was more significantly affected by
cutting methods than particle size with 10-20% improvement of lignin removal with Shards and
about 5% improvement with finer particles at 30 min pretreatment. In summary, faster mass
transfer was effective force to promote efficient pretreatment. For all three biomass components,
corn stover showed higher removal than poplar and reached the highest removal of 42.6%
glucan, 90.9% xylan and 44.7% lignin (16-40 mesh Shards) at 30 min while the highest removal
of 35% glucan, 80% xylan and 30% lignin for poplar (16-40 mesh Shards) at 30 min. Compared
to best pretreatment performance under similar conditions reported 441, 442, sub-millimeter
particles and modified cutting approaches showed advantageous performance.
For hot water pretreatment at 190ºC, hydrolysis of three biomass main components followed
the same pattern as dilute acid pretreatment: the particle sizes and cutting approaches showed
less effect on xylan removal but cutting approach of Shards improved lignin and glucan removal
at a degree of 5-10%. The removal extent and rate of three biomass components for corn stover
were higher than poplar. The highest removal of xylan, glucan and lignin was about 90%, 54%
and 56%, respectively; using 16-40 mesh Shards corn stover. However, the highest removal for
poplar was about 13% glucan, 70% xylan and 15% lignin at 30 min using 16-40 mesh Shards
poplar. The results confirmed our finding that different pretreatment mass transfer caused by
various particle sizes and cutting approaches influenced pretreatment performance significantly.
These results provided evidences that mass transfer during pretreatment affected
233
carbohydrates recovery and lignin removal. Higher mass transfer coefficients lead to better
biomass removal. It turned out that biomass with particle size 16-40 mesh in the study has better
performance, such as resulting high carbohydrates recovery and high amount of lignin removal.
Meanwhile, shards cutting methods showed better pretreatment performance than Cubic
prismatic cutting method. In dilute acid pretreatment, xylan, glucan and lignin removal could
reach as high as 90.86%, 42.64% and 44.66% of original xylan, glucan and lignin in raw biomass
(16-40 mesh mm Shards, corn stover), respectively. In hot water pretreatment, xylan, glucan and
lignin can achieve 90.73%, 54.4% and 56.6%, respectively. In summary, particle size of 16-40
mesh and Shards cutting method showed great advantages in carbohydrates recovery and lignin
removal during pretreatment.
8.4.4 Optimization of the Douglas fir hot water and dilute acid pretreatment conditions
The softwood pretreatment is found challenging. The detailed hot water and dilute acid
pretreatment of Douglas fir is rarely studied. Our study optimized the pretreatment conditions for
Douglas fir on the substrate with particle size of 40-60 mesh cut by hammer mill. The optimized
conditions were optimized and selected to be 1% wt sulfuric acid pretreatment at 140°C for
50min and hot water pretreatment at 200°C for 15min, which was assessed by lignin removal,
sugar yield and pretreated substrates digestibility (Figure 8.2; Figure 8.3; Figure S8.2).
235
Figure 8.2 Biomass removal trend of Douglas fir in hot water and dilute acid pretreatment (a)
Biomass removal; (b) lignin removal; (c) cellulose removal, pretreatment severities 2.1-4.5
Biomass removal is a parameter that can allow the evaluation of pretreatment effectiveness
directly. To achieve the same biomass removal, the less pretreatment time was needed when the
higher temperatures and concentrations of sulfuric acids applied (Figure 8.2a). Figure 8.2a
suggested that one can achieve similar biomass removal when the same range of biomass
pretreatment severities was applied. The biomass removal under all tested conditions was 10-
45% of original loaded biomass. In Figure 8.2b, lignin removal was less than 10% at all tested
conditions. Even in some conditions, lignin content in solid residues was increasing with the
increasing of the reaction time under the same pretreatment temperatures, such as the hot water
pretreatment at 190 and 220°C. This was probably attributed to pseudo-lignin formation 348, 349
form carbohydrates and/or lignin deposition 350 on solid residues under hot water and dilute acid
0 10 20 30 40 50 60 70
0
5
10
15
20
25
30
35
40
45
50
Original cellulose weight
in 100g biomass
140oC-1% (w/w) H
2SO
4
180oC-1% (w/w) H
2SO
4
160oC-0.5% (w/w) H
2SO
4
190oC-0.5% (w/w) H
2SO
4
200oC-0.5% (w/w) H
2SO
4
190oC-hot water
200oC-hot water
220oC-hot water
Ce
llulo
se
re
mo
va
l (g
in
10
0g
ori
gin
al b
iom
ass)
Pretreatment time (min)
(c)
236
conditions. The dilute acid pretreatment at 140C with 1% (w/w) sulfuric acid achieved the least
lignin removal among all tested conditions, while the hot water pretreatment at 190 and 200C
can obtain comparable lignin removal with the other conditions. The cellulose removal was in
the range of 7-20% of original cellulose under the tested conditions (Figure 8.2c). All tested
pretreatment conditions under the same pretreatment severities can achieve the similar cellulose
removal. Thus, the optimal pretreatment conditions for Douglas fir depend on the sugar recovery
under these conditions.
237
Figure 8.3 Total sugar yields from Douglas fir (1% H2SO4), left axis is Absolute (hemi-
sugars+glucan)/hemi-sugars/glucan recovery (from 100g dry biomass) while right axis is
Relative (hemi-sugars +glucan) /hemi-sugars/glucan recovery (% of maximum), Stage 1: sugar
release during pretreatment (measured after posthydrolysis); Stage 2: sugar release during
238
enzymatic hydrolysis; Stage 1+2: sugar release during pretreatment and enzymatic hydrolysis,
100 mg protein Ctec2 (93 FPU) with 20 mg Htec2/g (glucan + xylan)
The sugar recovery in stage 1 (pretreatment) and stage 2 (enzymatic hydrolysis) was shown
in Figure 8.3. In stage 1, cellulose derived sugars increased with pretreatment time except the
conditions under 1% (w/w) H2SO4 at 140C after 60min. However, in dilute acid pretreatment
with 1% (w/w) H2SO4, hemi-sugars underwent degradation at 180C for all tested reaction time.
The highest sugar recovery yield was at 140C for 50min (more than 90% of original
hemisugars) until hemisugars degraded afterwards. Thus, dilute acid with 1% (w/w) H2SO4 at
140C can achieve higher sugar yields than at 180C. The condition using 1% (w/w) H2SO4 at
140C for 50min in Douglas fir was applying 10mins more than the poplar’s optimal conditions
(1% (w/w) H2SO4 at 140C for 40min) in CAFI projects. Other sugar recovery results were
shown in Figure S8.2. The 0.05% sulfuric acid pretreatment achieved about a 90% sugar yield at
160C for 30min which was comparable to the yield at 1% (w/w) sulfuric acid at 140C for
50min (See Figure S8.2a). In the hot water pretreatment, the optimal conditions were 200C for
15min and 190C for 30min, which obtained up to 95% of sugar yields (See Figure S8.2b). In
comparison with the reported and optimized conditions for CAFI poplar wood discussed in
Section 3.2, Douglas fir needed higher temperature (200 vs 190C) to achieve high sugar yields.
In our study, the pretreatment conditions of 140C for 50min with 1% (w/w) sulfuric acid and
hot water at 200C for 150min were applied to evaluate the effects of biomass particle sizes and
cutting methods (particle shapes) on dilute acid and hot water pretreatment, respectively.
The dilute acid and hot water pretreatment we conducted is comparable to the target of
NREL in sugar yields and concentrations. The final maximum sugar concentration of Douglas fir
239
pretreatment was ~1.5% wt water (5% loading) and NREL Process Design and Economics (30%
loading) presented ~11.86% wt water in NREL process design 443. In Stage 2- enzymatic
hydrolysis, the final maximum sugar yield was ~60% (w/w) C5+C6 sugar Douglas fir while corn
stover in NREL Process Design and Economics demonstrated ~63.44% (w/w) 443. Thus, our
pretreatment technique on Douglas fir achieved comparable theoretical final sugar concentration
with the reported leading processes.
8.4.5 Pretreatment results of Douglas fir with various particle sizes and cuttings approaches
The appearance of poplar wood chips (particle size of 1/8′′-16 mesh and cut by crumbler)
under hot water pretreatment for 5,10,15,25, and 30 min were presented in Figure S8.3. The
colors of the chips were darker along with the increasing of pretreatment reaction times. In dilute
acid pretreatment, the hemicellulose was completely removed (~20% of original biomass) in all
tested substrates. The Douglas fir feedstock with different combinations of two cut methods (e.g.
crumbling and hammer mill) and three particle sizes (e.g. 1/8’’-16mesh; 12-30mesh; 30-60
0
20
40
60
80
100
CB 30-60
CB 12-30
CB 1/8-16
HM 30-60
HM 12-30
HM1/8-16
Control
Bio
mas
s re
mo
val (
%
of
ori
gin
al b
iom
ass)
Douglas fir feedstock
0
20
40
60
80
100
CB 30-60
CB 12-30
CB 1/8-16
HM 30-60
HM 12-30
HM 1/8-16
ControlCe
llulo
se r
em
ova
l (%
o
f o
rigi
nal
bio
mas
s)
Douglas Fir feedstock
0
20
40
60
80
100
CB 30-60
CB 12-30
CB 1/8-16
HM 30-60
HM 12-30
HM 1/8-16
Control
Lign
in r
em
ova
l (%
of
ori
gin
al b
iom
ass)
Douglas fir feedstock
(a) (b)
(c) Figure 8.4 Biomass and major
components removal of Douglas Fir
(140°C, 1% (w/w) sulfuric acid,
50min), Absolute biomass removal
(from 100g dry biomass), crumbler
(CB) and hammer cut (HM)
240
mesh) showed similar biomass (~30% of original biomass), cellulose (~15% of original
cellulose) and lignin removal (<5% of original lignin) (Figure 8.4). Thus, in Douglas fir
pretreatment, the lignin removal is much less than cellulose removal in the same substrates.
However, in the Consortium for Applied Fundamentals and Innovation (CAFI) study, when polar
wood was subjected to dilute acid pretreatment at 190°C with 2.0% (w/w) sulfuric acid for 1.1
min, the lignin removal of ~28% was higher than its cellulose removal (<25%) 389. Due to this
difficulty of removing lignin in Douglas fir, it is significant to study the effects of biomass
particle sizes and cutting methods on lignin removal in dilute acid pretreatment.
241
Figure 8.5 Recovery of Douglas fir sugars through dilute acid pretreatment (1% (w/w) sulfuric
acid), Stage 1: sugar release during pretreatment (measured after posthydrolysis); Stage 2: sugar
release during enzymatic hydrolysis; Stage 1+2: sugar release during pretreatment and enzymatic
hydrolysis, 20 and 200 mg protein Ctec2 with 4 and 40 mg Htec2/g (glucan + xylan)
The majority of sugars were recovered in monomeric forms under acidic conditions. The
242
recoveries of hemi-sugars of all tested substrates were similar in dilute acid pretreatment (Figure
8.5). In the pretreatment stage, the cellulose recovery was comparable on Douglas fir with all
combinations of three sizes and two cutting methods. However, the pretreated substrates with
various cutting methods and particle sizes showed larger difference of sugar yields in the
subsequent enzymatic hydrolysis than in pretreatment stage. Thus, the biomass particle sizes and
cutting methods have both effects on pretreatment and enzymatic hydrolysis. In our study, we
achieved very compatible sugar recovery yields in crumbler (CB; Cubic prismatic shape of
particles) and hammer cut (HM, shards shape of particles) (Figure 8.5). However, crumbler
consumed three times less energy than hammer cut, which showed potential biomass
preprocessing in the future large scale biorefinery.
0
20
40
60
80
100
CB30-60 CB12-30 CB1/8''-16 HM30-60 HM12-30 HM1/8''-16 Control
Bio
mas
s/m
ajo
r co
mp
on
en
ts
rem
ova
l (%
of
ori
gin
al b
iom
ass)
Dougals fir feedstock
Biomass removal Lignin removal(a)
243
Figure 8.6 Recovery of Douglas fir sugars through hot water and enzymatic hydrolysis (a)
biomass and lignin removal; (b) sugar recovery; Stage 1: sugar release during pretreatment
(measured after posthydrolysis); Stage 2: sugar release during enzymatic hydrolysis; Stage 1+2:
sugar release during pretreatment and enzymatic hydrolysis, 20 and 200 mg protein Ctec2 with 4
and 40 mg Htec2/g (glucan + xylan)
(b)
244
Nearly complete glucose recovery and over 90% hemi-sugars recovery from medium (12-
30) and finer (30-60) particle size substrates cut by both approaches were achieved through acid
pretreatment at milder temperature followed by sequential enzymatic hydrolysis (Figure 8.5). All
tested substrates had similar hemi-sugars recovery through both pretreatments while crumbler
cutting showed slight benefit over hammer milling, especially with bigger particle size
substrates. When pretreated at higher temperature with hot water only, to achieve near complete
combined glucose recovery (pretreatment+enzymatic hydrolysis), finer particle size (less than
12-30 size) was required for crumbler cutting than for hammer milling (Figure 8.6). In enzymatic
hydrolysis, Douglas fir particles with crumbler cut (&12-30 mesh size) showed compatible sugar
yields with other cut and size combination particles.
8.4.6 Hot water and dilute acid pretreatment of poplar wood
Effects of particle sizes and cutting approaches on poplar wood were shown in Figure 8.7
and 8.8. All hemicellulose was completely removed (~20.62% of original biomass). Compatible
pretreatment performance with different particle sizes and cutting, e.g. crumbler (CB) and
hammer cut (HM), was found. The final maximum sugar concentration (pretreatment) of the
sugar stream of poplar wood hot water or dilute acid pretreatment was ~1.01% wt (5% loading)
water in comparison with ~11.86% wt water in NREL Process Design and Economics (30%
loading) 443.
245
Figure 8.7 (a) Biomass removal, (b) Cellulose removal, (c) Lignin removal, (d) Manan and
xylan removal, and (e) sugar yields of poplar in dilute acid pretreatment, 140°C, 1% (w/w)
sulfuric acid, 40min, and enzymatic hydrolysis (72 hours), 100 mg protein Ctec2 with 20 mg
Htec2/g (glucan + xylan)
Biomass, cellulose, hemicellulose and lignin removal were comparable in dilute acid
pretreatment of poplar wood (Figure 8.7a,b,c,d). The sugar recovery was also consistently similar
0
20
40
60
80
100B
iom
ass
rem
ova
l (%
of
ori
gin
al b
iom
ass)
Poplar feedstock
0
20
40
60
80
100
Ce
llulo
se r
em
ova
l (%
of
ori
gin
al b
iom
ass)
Poplar feedstock
0
20
40
60
80
100
CB 30-60
CB 12-30
CB 1/8-16
HM 30-60
HM 12-30
HM1/8-16
Control
Lign
in r
em
ova
l (%
of
ori
gin
al
bio
mas
s)
Poplar feedstock
0
20
40
60
80
100
Bio
mas
s/m
ajo
r co
mp
on
en
ts
rem
ova
l (%
of
ori
gin
al
bio
mas
s)
Poplar feedstock
Mannan Xylan
(a) (b)
(c) (d)
(e)
246
among tested substrates. In hot water pretreatment, although hammer milling (HM) showed
slightly higher biomass and cellulose removal than crumbler (CB), their lignin removal was
comparable. The medium size (12-30mesh) poplar achieved slightly higher lignin removal than
other substrates (Figure 8.8a). Additionally, medium size (12-30mesh) presented ~10% higher
sugar recovery than others.
Figure 8.8 Biomass removal, sugar yields of green poplar in hot water pretreatment, 190°C, hot
water, 15min (a) biomass/major components removal, (b) combined sugar yields after
pretreatment and enzymatic hydrolysis, 100 mg protein Ctec2 with 20 mg Htec2/g (glucan +
xylan)
In summary, the crumbler (CB) and medium size (12-30mesh) achieved 5-15% higher
sugars than other biomass feedstock. The biomass removal and sugar recovery yield of hammer
0
10
20
30
40
Biomass Removal Lignin Removal Cellulose Removal
Bio
mas
s/m
ajo
r co
mp
on
en
ts
rem
ova
l (%
of
ori
gin
al
bio
mas
s)
Poplar feedstock
CB 30-60 CB 12-30 CB 1/8-16 HM 30-60 HM 12-30 HM 1/8-16(a)
(b)
247
milled and crumbler cut particles were compatible in dilute acid and hot water pretreatment and
particles with 12-30 mesh gained 5% more sugars than others. The crumbler cut poplar particles
with 12-30 mesh size showed better sugar yields than other cut and sizes particles in enzymatic
hydrolysis.
8.4.7 Sugar yields from hot water and dilute acid pretreatment of poplar and corn stover
and sequential enzymatic hydrolysis
Table 8.5 Sugar yields from pretreatment and enzymatic hydrolysis
Size and biomass particle shape
Pretreatment sugar yield
Stage 1
Enzymatic hydrolysis sugar
yield Stage 2 Total sugar yield
Glucose Xylose Glucose Xylose Glucose Xylose
16-40 mesh Shards, Poplar 16.86 71.14 70.79 5.42 87.65 76.56
16-40 mesh Cubic prismatic,
Poplar 12.03 67.81 69.83 3.61 81.86 71.42
1/8′′-16 mesh Cubic prismatic,
Poplar 9.71 67.04 68.88 2.21 78.59 69.25
16-40 mesh Shards, Corn
stover 33.66 90.78 61.30 3.77 94.97 94.55
16-40 mesh Cubic prismatic,
Corn stover 28.79 87.58 55.00 3.57 83.79 91.15
1/8′′-16 mesh Cubic prismatic,
Corn stover 24.81 82.02 54.92 2.98 79.73 85.00
Notes
1.With 100 mg protein Ctec2 (93 FPU) with 20mg Htec2/g (glucan+xylan)
2. Pretreated whole slurry was obtained from 1% dilute acid pretreatment at140ºC with
retention time of 40 min
Size and biomass particle shape
Pretreatment sugar yield
Stage 1
Enzymatic hydrolysis sugar
yield Stage 2 Total sugar yield
Glucose Xylose Glucose Xylose Glucose Xylose
16-40 mesh Shards, Poplar 15.75 72.27 61.86 10.00 77.61 82.27
16-40 mesh Cubic prismatic,
Poplar 18.52 69.77 50.70 7.51 69.22 77.28
1/8′′-16 mesh Cubic prismatic,
Poplar 15.39 63.19 37.52 3.60 52.91 66.79
16-40 mesh Shards, Corn
stover 37.43 79.82 56.94 19.80 94.36 99.62
16-40 mesh Cubic prismatic,
Corn stover 49.24 90.04 40.80 5.08 90.04 95.12
1/8′′-16 mesh Cubic prismatic,
Corn stover 40.21 86.25 40.22 4.53 80.43 90.78
Notes
1.With 100 mg protein Ctec2 (93 FPU) with 20mg Htec2/g (glucan+xylan)
2. Pretreated slurry was obtained from hot water pretreatment at 190ºC with retention
time of 25 min
248
Enzymatic hydrolysis of pretreated biomass whole slurries was carried out with enzyme
loading of 100 mg protein Ctec2 (93 FPU) plus 20 mg protein Htec2/ g (glucan+xylan). Table
8.5 showed sugar yields of pretreatment and enzymatic hydrolysis of pretreated biomass by hot
water pretreatment at 190ºC for 25 min and 1% dilute acid pretreatment at 140ºC for 40 min. All
sugar yields were calculated based on the original content of glucan or xylan in untreated
biomass. The total sugar yields in stage 1 pretreatment included yields of both monomeric and
oligomeric sugars. Total xylose and glucose were monomeric sugar yield because oligomeric
sugars were hydrolyzed into monomeric sugars during stage 2 enzymatic hydrolysis. It was
found that the highest sugar yields from pretreatment could be achieved at 25 min for hot water
pretreatment at 190ºC and at 1% dilute acid pretreatment for 40 min at 140ºC, respectively.
Results of sugars’ yields from both pretreatments showed that higher glucose and xylose
yields were observed from corn stover samples than those of poplar samples. 82-90% xylose
yield from corn stover samples and 67-71% xylose yield from poplar samples were obtained
through 1% dilute acid pretreatment. Although similar xylose yields are found for corn stover
and poplar samples using hot water pretreatment, higher portion of xylan degradation products in
hydrolysate was xylan oligomers with hot water pretreatment. The pretreatment glucose yields
are strongly correlated to the mass transfer coefficient of the samples (Table 8.4).
Effective pretreatment is able to disrupt biomass plant cell wall and results in high enzymatic
digestibility. Enzymatic yields can be criteria to reveal effectiveness of pretreatment methods.
For pretreated biomass by 1% sulfuric acid pretreatment at140℃, the enzymatic hydrolysis yield
appeared highly dependent on particle sizes, biomass types and cutting approaches of pretreated
biomass. 16-40 mesh Shard scorn Stover showed the highest enzymatic digestibility of 94.97%
249
of glucose yield and 96.81% of xylan yield based on glucan and xylan content of pretreated
biomass. Enzymatic hydrolysis of 16-40 mesh Shard scorn stover pretreatment by hot water
pretreatment at 190ºC attained 94.36% of glucose and 99.62% of xylose yield. This indicated an
effective pretreatment on 16-40 mesh shards corn stover. Hot water pretreatment showed relative
better enzymatic hydrolysis yield, which might be because of more lignin was removed in hot
water pretreatment. Lignin remained in pretreated slurries after dilute acid pretreatment at 140ºC
inhibited enzymatic hydrolysis 164. Results from both hot water and dilute acid pretreatment also
indicated that samples with high mass transfer coefficients obtained higher pretreatment sugar
yields and achieved higher enzymatic hydrolysis yield too (Table 8.4, 8.5).
The total sugar yield, which was the combined yield of stage 1 (pretreatment) and stage 2
(enzymatic hydrolysis) showed strong correlation to the mass transfer coefficients of substrates.
16-40 mesh particle size and Shards cutting resulted in the highest total sugar yield. It was
reported increased solvent diffusion during pretreatment by changing particle size can increase
pretreatment severity which led to a good pretreatment yield 419. Among three poplar samples,
16-40 mesh Shards poplar showed the highest total glucose yield of 87.65% and total xylose
yield of 76.56% for dilute acid pretreatment, 77.61% total glucose yield and 82.27% total xylose
yield for hot water pretreatment. 16-40 mesh Shards corn stover also resulted in the highest total
glucose yield of 94-95% and total xylose yield of 94.55-99.62% for both pretreatments. Results
showed particle size and biomass cutting approaches (i.e. Shards and Cubic prismatic) have
significant impact on the enzymatic hydrolysis yield. One reason was their positive effects on
pretreatment process which disrupted plant cell wall, increased porosity and removed lignin 161,
169, 444. Those chemical and physical change of biomass affected enzymatic hydrolysis yield.
250
Second reason was the effects of particle sizes and shapes on biomass accessibility for enzymes.
Our results showed medium particle size (16-40/ 12-30 mesh) shards and Cubic prismatic cutting
were comparable in enzymatic hydrolysis and higher compared to other combinations of particle
sizes and cuttings.
8.5 Conclusion
The hot water and dilute acid pretreatment of corn stover, poplar and Douglas fir with four
particle sizes and two cutting methods was carried out to determine their different degradations
of xylan, glucan and lignin. Results showed that biomass species and cutting methods have more
impact on pretreatment mass transfer and yields than particle sizes. Also, results of enzymatic
hydrolysis on hot water pretreatment and 1% (w/w) dilute acid pretreatment indicated the
significant roles of particle sizes and cutting approaches in enzymatic hydrolysis yield. It was
found that the sugar yields of biomass particles with two cutting methods and four particle sizes
were compatible. Although shards shaped (hammer milling) biomass particles showed a bit
higher sugar yields, the crumbling cut still achieved comparable sugar yields with hammer
milling. Interestingly, the crumbler cutting showed 5-15% advantages in getting better sugar
yield in enzymatic hydrolysis. Addtionally, the energy consumption of crumbling was three times
less than traditional hammer milling. Biomass particles with 12-30/16-40 mesh size could
achieve similar combined sugar yields as finer particle size substrates. The mass transfer
modeling was proposed in Appendix (proposed mass transfer modeling). The preliminary work
showed that larger particle sizes and crumbling cutting have better mass transfer diffusion
capacities. Thus, the crumbling cut biomass has advantages in mass diffusion. However, more
future work is needed such as the improvements of measurements of mass transfer coefficients
251
and integrating these values into the proposed models, and study of the effects of particle sizes
and cutting shapes on mass diffusion. This study is significant for choosing wise preprocessing
methods and pretreatment methods for biomass in the future.
8.6 Acknowledgements
I would like to thank Forestconcepts, LCC for the funding support for this work. I would
like to thank Dr. Jim Dooley and Leading Mechanical Engineer David Lanning from
Forestconcepts for their support. Also, I want to acknowledge Dr. Jie Xu and Dmitry Gritsenko
from University of Illinois at Chicago for their help in mass transfer modeling development.
Additionally, I thank Mr. Pei-yu Leu and Sohrab H. Mood for his assistance in completing
experiments.
8.7 Supplementary materials
Figure S8.1 Left) Crumbler M24 research prototype machine used with 4.8 and 1.6 mm cutters,
Right) Research scale Crumbler machine with 0.8 mm cutters (by ForestConcept, LCC)
254
(b)
Figure S8.2 Total sugar yields from Douglas fir of (a) 0.05% wt sulfuric acid (b) Water only, left
axis is Absolute (hemi-sugars+glucan)/hemi-sugars/glucan recovery (from 100g dry biomass)
while right axis is Relative (hemi-sugars +glucan) /hemi-sugars/glucan recovery (% of
maximum), Stage 1: sugar release during pretreatment (measured after posthydrolysis); Stage 2:
sugar release during enzymatic hydrolysis; Stage 1+2: sugar release during pretreatment and
enzymatic hydrolysis
255
Figure S8.3 Hot water pretreatment of poplar wood chips for 5,10,15,25, and 30 min (left to
right), particle size of 1/8′′-16 mesh and cut by Crumbler
256
CHAPTER NINE
CONCLUSIONS AND FUTURE PERSPECTIVES
Lignocellulosic biomass is a renewable source for the production of liquid transportation
fuels which potentially can cause a major impact on energy use architecture. The recalcitrant
nature of biomass is the main challenge to overcome. The goal of pretreatment is to reduce the
biomass recalcitrance and improve yields of sugars and ethanol. Recent aqueous pretreatment
fractionation of biomass in producing sugar and lignin intermediates facilitates the utilization of
the whole biomass to achieve cost-competitive biorefinery.
The topics of this thesis focus on the study of key fundamental issues in biomass aqueous
pretreatment. In chapter 3, the HR-BB-SFG-VS was developed to assisting the characterization
of crystalline cellulose in aqueous pretreatment for the first time. New structural information
(e.g. more peaks in C–H and O–H regions) of cellulose was found, and this information has not
been achieved by traditional methods, which demonstrated a significant role of SFG-VS in
cellulose structure study. It was found that O–H vibrational regions were the signatures of
various cellulose polymorphs while C–H vibrations distinguished celluloses from various
sources. Additionally, the peak integrated area ratio of C–H and O–H was developed as a factor
to distinguish different crystals, such as cellulose Iα, Iβ, and Avicel. Furthermore, the chapter 4
investigated the structural changes of cellulose bulks and surface layers of crystals in a dynamic
heating process with/without water in aqueous pretreatment using the SFG-VS equipped with a
fluid reactor. In order to characterize the significant surface layers, an innovative design by
changing the laser angles and geometries was proposed and successfully selectively characterize
the cellulose surface layers for the first time. The results showed the distinguished structural
257
differences between crystals’ surface layers and bulks of Avicel and Iβ, and the SFG assignments
of cellulose C-H and O-H peaks were revisited. In aqueous pretreatment, the high temperature is
a crucial factor in influencing pretreatment efficiency. The chapter 4 showed the cellulose bulk
crystalline structures were completely disrupted at elevated temperatures (170-220C).
Interestingly, it was proposed that both cellulose Iα and Iβ in Avicel were converted gradually to
an intermediate which possessed dislocated/disordered/random-oriented microfibrils with the
dynamic increasing of heating temperatures, and the transformation of cellulose Iα to the
intermediates were not reversible and Iα may or may not be converted to cellulose Iβ. In
comparison, the surface layers of cellulose were not recrystallized to gain crystalline structures.
The possible hypothesis was proposed that the disrupted cellulose microfibrils recrystallize
during and after cooling down to room temperature. This was possibly triggered by the
rearrangements of dislocated crystals to reach a stable phase. However, this recrystallization is
not a recovery process so that the newly formed crystals were differed from the original
structures but showing similarities. Because cellulose bulks are highly disordered crystals, the
new crystals showed considerate similarities with the original cellulose structures and molecular
orientations. However, due to the amorphous of the surface layers of crystalline cellulose, the
recrystallization of the surface layers is not occurred. This hypothesis still needs more
experimental work to validate, such as a detailed characterization of the newly formed crystals
by HR-BB-SFG-VS, the application of other analytical tools to assist analysis, the weight and
wet chemistry analysis of the heated cellulose and the tests of cellulose Iα and Iβ in dynamic
heating process. Besides, the enzymatic hydrolysis of cellulose was also characterized by the
SFG-VS which suggested the potential of SFG-VS in characterizing the interaction of celluloses
258
and enzymes without separating them. Also, the C-H vibrational region with slow peak decay is
important for the study of enzymatic hydrolysis in SFG characterization. Surprisingly, with the
undergoing of enzymatic hydrolysis, it was proposed that cellulose crystals were changing
suggested by the characterization of peak area ratios of enzymatic restart samples. However, this
conclusion remains open. The thesis provided some preliminary work on the cellulose structure
characterization and application of SFG-VS in characterization of cellulose structural changes.
We found significant spectra differences between cellulose Iα, Iβ, and Avicel such as peak
position, intensity and ratios. The traditional understandings of cellulose structure are not enough
and some assumptions might not be accurate. More research work is needed to give the final
conclusions but our preliminary results are significant and point out the research direction. The
SFG-VS system has evidently demonstrated its significant roles in studying cellulose crystalline
structural changes in biorefinery.
The flowthrough reactor system provides a useful tool to study the fundamental lignin
behavior and chemistry in aqueous pretreatment. The flowthrough aqueous pretreatment of
softwood and hardwood in chapter 5, 6, and 7 introduced the effectiveness of flowthrough
reactor in fractionating biomass to produce high yields of fermentable sugars and usable forms of
lignin by exiting products in time to avoid sugar degradation and lignin condensation. The liquid
phase lignin derived from flowthrough aqueous pretreatment after acidifying, precipitation, and
purification was provided for the hydrodeoxygenation to produce high yield of hydrocarbons
(Table A2) which is not the focus on this thesis although some work has been published in Green
Chemistry journal 445. The aqueous flowthrough pretreatment has helped understand several
fundamental questions. In chapter 5, the softwood flowthrough pretreatment is studied at pH
259
neutral-12 for the first time. The results showed the great impact of pretreatment initial pHs on
products distribution, kinetics and biomass chemistry. The solubilization of the whole biomass
was found possible with proper controlling of temperature and pHs. The value-added
biochemical was produced (pH≥11.0) while presented difficulties in analysis. However, the
potential utilization of the mixture of aromatics or other compounds is an interesting topic in the
future. The pretreatment mechanisms with impact from initial pHs of pretreatment media were
proposed. The chapter 6 comprehensively studied the lignin characteristics derived from
hardwood flowthrough pretreatment with main lignin depolymerization mechanism by the
cleavage of β-O-4, producing the proper properties of lignin with low molecular weight, very
mild C-C condensation (Cβ-C5), and a high purity and yield. The lignin was provided in
carbonation conversion for the production of carbon Nanoporous materials which were found
promising as electrodes in supercapacitor. Some of the work has been published in
ChemSusChem journal 328. The lignin characterization of aqueous flowthrough pretreatment in
the thesis provides significant information on the properties of aqueous pretreated lignins, which
are the basics of utilizing them. The results showed that the flowthrough system and optimized
pretreatment conditions can separate and isolate lignin in aqueous phase for more than 95% yield
and 85% of purity while still ensured high recovery of sugars. Lignin was produced through
three subsets which includes little residual lignin, substantial acid precipitated insoluble solid
lignin and less than 20% of soluble lignin. The structural analyses of the flowthrough derived
lignin showed the cleavage of β-O-4 and lignin condensation reactions, which contributes the
lignin reaction chemistry of aqueous pretreatment. The flowthrough moved depolymerized lignin
fragments out of reactor in a short time to preventing from further reactions, which produced
260
lignin that possessed similarities to the original native lignin. Due to the high purity and yield of
the lignin as well as the modest modification of lignin structure in flowthrough with low
molecular weight, the valorization or utilization of flowthrough aqueous pretreatment derived
lignin is attractive in the future.
In chapter 7, the softwood and hardwood lignin structural transformation at molecular level
was revealed for the first time starting with their magnificent difference of lignin removal under
the same flowthrough pretreatment conditions. The advanced aqueous and solid phase NMR
analytical methods supported the severe recondensation mechanisms (proposed at Cα-C5, C5-
C5, Cβ-C5 and C5-C4) of softwood lignin resulting in the remaining of 30% softwood lignin as
residual lignin in solid residues even when all softwood carbohydrates were completely
dissolved. The GPC analysis of the softwood residual lignin revealed the low molecular weight
distribution of residual lignin which confirmed the recondensed mechanisms in forming the
softwood residual lignin. This study is the first time to find this opportunity (complete poplar
solubilization while 30% softwood lignin remained as solids) to study the different structural
transformations between softwood and hardwood, and the mechanisms of difficult softwood
lignin removal have remained for a long time in the leading pretreatment.
The pretreatment process can have a great impact on the preprocessing methods and
costs. Conversely, the preprocessing of feedstock can affect the pretreatment efficiency and total
biorefinery cost. In chapter 8, the pretreatment performances of new preprocessed feedstock
substrates (e.g. new cutting approach crumbling and particle sizes) with three times lower energy
consumption than traditional methods, were assessed in aqueous pretreatment. It was found that
the aqueous pretreatment of the feedstock cut by crumbling, the new approach, achieved the
261
comparable pretreatment performance with traditional hammer milling. More importantly, it was
found that the cutting methods which have rarely received attentions were more important than
particle sizes in mass diffusion. Also the biomass species and cutting approaches are more
influencing in pretreatment and enzymatic hydrolysis performances than the effects from particle
sizes. Surprisingly, our results suggested that finer particle sizes did not always lead to high
pretreatment and enzymatic hydrolysis sugar yields. The mass transfer modeling and the
measurements of diffusion capacities (acid/alkali diffusivity) of various feedstocks with different
cutting methods, biomass species, and particles were proposed by reported and designed
apparatus. The preliminary results indicated the particle sizes and cutting methods were
correlated with mass diffusion capacities. However, more efforts are still needed to improve the
mass transfer model. All these results are important for preprocessing of feedstocks at low cost
and energy, especially in large scale biorefinery in the future.
Future work
The SFG-VS is demonstrating a significant role in helping study the structural changes of
cellulose crystals in biomass aqueous pretreatment. However, this thesis only focuses on the
fundamental study of cellulose structural changes with/without the effects of water. More work
can be done on the understandings of the effects from other chemicals such ad acids and alkalis.
Also, the SFG-VS system can be developed to study the enzymatic hydrolysis system, such as
the interactions of cellulose with enzymes in enzymatic hydrolysis. Additionally, the
recrystallization hypothesis of cellulose crystalline materials has been proposed in the thesis.
More experimental work is needed to provide solid proofs to enhance the hypothesis, such as
applying HR-BB-SFG-VS in the characterization of the newly formed crystals, using other
262
analytical tools to assist in studying the recrystallization process, and studying thermal behaviors
of other cellulose crystalline materials.
The flowthrough reactor system is indicating the advantages in fundamental studies of
pretreatment processes by separating the three major components for characterization in this
thesis. However, the thesis did not cover the utilization of the recovered sugar streams into
applications. More efforts in the future can be devoted to the application of the sugar streams to
produce chemicals, materials or fuels. Besides, the commercialization study of flowthrough
reactor is necessary considering its strong capacities in producing a high yield and purity of
intermediates. Additionally, the flowthrough reactor can be expanded to producing more
intermediates from biomass besides sugars and lignins. The alkaline pretreated liquor in the
thesis was indicated hard to analyze but it contains a valuable mixture of chemicals. Thus, the
future work also can focus on the design of the analysis methods and utilize these compounds in
producing biochemical. In chapter eight, the mass transfer modeling and the measurements of
mass transfer efficient were proposed but more future work is still needed to improve the
modeling.
The thesis highlighted the process characterization of aqueous pretreatment, focusing on the
characterization of cellulose and lignin. In the future, most of research related to lignin and
cellulose can be implemented by the capacities and skills demonstrated in the thesis. The author
is interested in the following aspects of work. Firstly, continue the fundamental studies of
aqueous pretreatment and enzymatic hydrolysis. For example, study the interaction between
enzymes with cellulose can be studied by various spectroscopic tools and molecular dynamic
simulation. Study mass transfer issues in aqueous pretreatment. Continue research on alkaline
263
pretreatment triggering by the preliminary results of excellent mass diffusion capacities at
<100C. Second, the author is interested in applying lignin and cellulose as carbon sources to
produce value-added products. Apply intermediates, such as cellulose, lignin, chemicals, and
aromatic compounds, indicated in this thesis to produce various chemicals, materials, and fuels.
For example, the conversion of sugars and lignin to biolipids, the conversion of cellulose and
lignin to hydrocarbons, the carbonization of cellulose and lignin to Nanoporous carbon materials,
carbon fibers or other materials, and the conversion of lignin to valuable aromatic compounds.
Third, develop new pretreatment techniques to tailor cellulose and lignin properties for different
applications. At last, learn commercialization knowledge to facilitate our processes to
commercialization.
Achievements and perspectives
This research has revealed new structural information of cellulose by HR-BB-SFG-VS for
the first time. The in-situ characterization of the molecular structure of cellulose surface layers
was also achieved for the first time. The fundamental study of cellulose structure helps reduce
the cost of pretreatment and enzymatic hydrolysis by fine-tuning the loading of chemicals and
enzymes results in needing fewer of these chemicals for the purpose of increasing the cellulose
accessibility. Our research found that the traditional understanding of cellulose structure should
be should be adjusted and our study pointed out the future research direction to do this. An
accurate understanding of the recrystallization phenomena of cellulose materials strongly affects
the design of thermal heating in pretreatment, and ensures that the most cost-effective amount of
heat in the large-scale pretreatment of cellulose. Our research also uncovered new insights into
the thermal modification of cellulose in industry. The poplar lignin was comprehensively
264
characterized in water-only and dilute acid flowthrough pretreatment and a mechanism of side
chain transformation and Cβ-C5 condensation was proposed based on the evidence. The recent
lignin valorization studies insufficiently studied the connection between lignin characteristics
and the effectiveness and efficiency of lignin conversion to value-added products. The thorough
characterization of pretreatment produced lignin facilitates the future conversion of lignin
substrates to useful commercial products and controls their qualities. The 30% pure undissolved
residual softwood lignin was isolated for the first time and strong evidence was found that this
residual pure C-C5 condensed structure was caused by severe recondensation mechanisms. This
can be used to study softwood lignin removal mechanisms which have been historically
challenging. This is significant to the application of mix-feedstock or the use of multiple biomass
types in biorefinery. The mix-feedstock is a solution to solve the shortage of biomass since
different types of biomass can be distributed differently by seasons and locations. Since
undissolved C-C lignin is less reactive to conversion it can instead be used to heat the factories
while converting the more reactive portions to value-added products. The impact on pretreatment
mechanisms by different levels of the initial pHs of pretreatment media was studied and the
impact was found to be very significant. This will help alkaline and dilute acid pretreatment to
better produce sugars and guide the selection of methods for producing sugars and biochemicals.
The significance of cutting methods of biomass was researched and it was found that finer
particle sizes did not always lead to high pretreatment and enzymatic hydrolysis sugar yields. A
new research direction of the study of cost-effective cutting methods and the biorefinery
performances of cut biomass is needed because of this finding. Using optimal particle sizes can
reduce the cost of expensive preprocessing to finer sizes.
265
APPENDIX
Figure A1 SFG-VS characterization of (a) Avicel 2600-3600cm-1; (b) Avicel 1000-1600cm-1; (c)
Cellulase and BSA 2600-3800 cm-1; (d) Cellulase and BSA 1500-1800 cm-1, Note: (A)-Avicel
without a flat CaF2 window on top of the sample; (B)-Avicel with a flat CaF2 window on top of
the sample
The application of a flat CaF2 window showed similar signals of cellulose samples with
the one without a flat CaF2 on top. The wavelength of 1000-1600 cm-1 is the fingerprint region of
cellulose crystalline structures. Thus, the current SFG-VS characterization of cellulose only
focuses on the wavelength of 2600-4000cm-1due the complexity of the fingerprint region.
In order to study the complex enzymatic hydrolysis system, the functionality of various
cellulase components should be closely inspected firstly. To assist the study of the interaction of
cellulose with enzymes in enzymatic hydrolysis system, various proteins (i.e. cellulase, lignin
(a) (b)
(d) (c)
266
blocking agent-BSA) were characterized by SFG-VS. The cellulase and BSA were characterized
in solution with 1.2g/cm-3 and 40mg/mL, respectively, which are their maximum solubility to
guarantee the molecules locating on the surface of the solution so that SFG-VS can detect their
signals. As seen in Figure A1, the SFG-VS characterization of enzymatic hydrolysis system can
avoid the overlaps of peaks of cellulose and cellulase. Thus, the SFG-VS presented a significant
tool to study the enzymatic hydrolysis without separating cellulase and cellulose for
characterization.
In the SFG-VS characterization of proteins, the region of wavelength of 1500-1800 cm-1
is usually a signature of proteins, thus, these peak characteristics located in this region can be
developed to study the enzymes in the enzymatic hydrolysis system. However, the cellulase and
BSA presented the same peak positions in ~2870 and ~2940 cm-1, however, their peak ratios
were differed. Additionally, both cellulase and BSA showed strong SFG polarization
dependence. In the same enzymatic hydrolysis system, when cellulase and BSA are both applied,
their peaks might interfere with each other. Even though, this problem can still be avoided by
good design in experiments. The SFG-VS is a valuable tool to characterize molecular structural
changes in enzymatic system.
267
Table A1 Pyrolysis products of residual lignin derived from aqueous flowthrough pretreatment
by Py-GC/MS
Number Retention time (min) Compounds MW Sources
1 8.74 Phenol 94 H
2 10.65 Phenol, 2-methyl- 108 H
3 11.27 Phenol, 3-methyl- 108 H
4 11.38 Phenol, 2-methoxy-
(Guaiacol)
124 G
5 11.81 Phenol, 2,6-dimethyl- (2,6-
Xylenol)
122 H
6 12.68 Phenol, 2-ethyl- 122 H
7 12.93 Phenol, 2,4-dimethyl- (2,4-
Xylenol)
122 H
8 13.41 2,3-Dihydroxybenzaldehyde 138 G
9 13.44 Phenol, 3,5-dimethyl- (3,5-
Xylenol)
122 H
10 13.89 Phenol, 2-methoxy-4-methyl-
(4-methylguaiacol)
138 G
11 13.98 Benzoic acid 122 H
12 14.14 Phenol, 2,4,6-trimethyl- 136 H
13 15.01 Phenol, 4-ethyl-3-methyl- 121 H
14 15.10 1,2-Benzenediol 110 G
15 15.30 Phenol, 2-(1-methylethoxy)-,
methylcarbamate
G
16 15.52 Phenol, 2-propoxy- 152 G
17 15.81 Benzaldehyde, 2-hydroxy-5-
methoxy-
152 G
18 15.87 Phenol, 4-ethyl-2-methoxy-
(4-ethylguaiacol)
152 G
19 15.91 1,2-Benzenediol, 3-methoxy- 140 S
20 15.97 1,4-Benzenediol, 2-methoxy- 140 S
21 16.19 Phenol, 2-ethyl-4,5-dimethyl- 150 H
22 16.73 2-Methoxy-4-vinylphenol 150 G
23 16.83 1,2-Benzenediol, 3-methyl- 124 G
24 16.99 1,2-Benzenediol, 4-methyl- 124 G
25 17.37 1,3-Benzenediol, 2-methyl- 124 G
26 17.46 Phenol, 4-(2-propenyl)- 134 H
27 17.64 Phenol, 2,6-dimethoxy-
(Syringol)
154 S
28 18.50 Phenol, 3,4-dimethoxy- 154 S
29 17.84 Phenol, 2-methoxy-4-(1-
propenyl)- (Isoeugenol)
164 G
30 18.84 Vanillin 152 G
31 18.97 Phenol, 2-methyl-6-(2-
propenyl)-
148 H
32 19.24 1,4-Benzenediol, 2,6-
dimethyl-
138 G
268
33 19.31 1,3-Benzenediol, 4,5-
dimethyl-
138 G
34 20.02 Phenol, 2-methoxy-4-propyl- 166 G
35 22.63 Phenol, 2,6-dimethoxy-4-(2-
propenyl)-
194 S
36 23.97 Benzaldehyde, 4-hydroxy-3,5-
dimethoxy-
182 S
37 26.35 Phenol, 4-(ethoxymethyl)-2-
methoxy
182 G
38 23.69 Benzoic acid, 4-ethoxy- 166 G
39 11.10 Phenol, 4-methyl- 108 H
40 11.72 Phenol, 2,3-dimethyl- 122 H
41 12.77 Phenol, 2-propyl- 136 H
42 13.38 Phenol, 4-ethyl- 122 H
43 13.43 Phenol, 3-ethyl- 122 H
44 17.58 Eugenol 164 H
45 18.67 Phenol, 2-methoxy-6-(2-
propenyl)-
164 H
46 23.11 Benzene, 1,4-dimethoxy-2-
methyl-
152 S
47 23.48 Benzoic acid, 3-hydroxy- 138 H
48 25.19 Benzaldehyde, 3,4,5-
trimethoxy-
196 S
49 25.42 4-((1E)-3-Hydroxy-1-
propenyl)-2-methoxyphenol
(coniferol)
180 G
50 25.61 Benzaldehyde,-2-propoxy,-5-
methoxy
194 G
51 28.68 Benzoic acid, 3,4-dihydroxy- 154 G
52 20.34 2,3-Dimethoxytoluene 152 G
53 28.99 Phenol, 3,5-dimethoxy- 154 S
54 8.04 Benzaldehyde 106 H
55 16.94 3-Methoxy-5-methylphenol 138 G
56 17.77 Ethyl Vanillin 166 G
57 17.90 2,4-Dimethoxyphenol 154 S
58 18.15 Resorcinol 110 G
59 18.21 Phenol, 3-methoxy- 124 G
60 19.54 Benzoic acid, 4-hydroxy-3-
methoxy- (Vanillic acid)
168 G
61 20.81 4-Hydroxy-2-
methoxybenaldehyde
152 G
62 22.63 2,5-Dimethoxy-4-
ethylbenzaldehyde
182 S
63 24.15 Benzoic acid, 3-ethoxy- 166 G
64 24.38 Benzoic acid, 4-hydroxy- 138 S
65 9.21 Benzene, ethoxy- 122 G
66 10.70 Benzene, 1,2,3-trimethyl- 120 H
67 18.09 1,4-Benzenediol, 2-methyl- 124 G
68 18.91 3,4-Dimethoxytoluene 152 G
69 20.23 Benzaldehyde, 2-ethoxy- 150 H
70 20.68 3,5-Dimethoxybenzaldehyde 166 S
71 20.86 4-Hydroxy-2- 152 G
269
methoxybenaldehyde
72 22.75 4-Propyl-1,1'-diphenyl 196 H
73 24.59 Phenol, 3-methoxy-2-methyl- 138 G
74 25.19 Ethanone, 1-(4-hydroxy-3,5-
dimetho
196 S
75 27.97 1,1'-Biphenyl, 2-methyl- 168 H
76 29.80 Benzene, 1-phenoxy-2-(2-
propenyl)-
210 H
77 10.13 Benzeneacetaldehyde 120 H
78 15.12 Phenol, 4-ethoxy- 138 G
79 17.02 Benzene, 1,3-dimethoxy- 138 G
80 18.11 3,5-Dihydroxytoluene
(Orcinol)
124 G
81 23.30 Benzaldehyde, 2,4-
dimethoxy-
166 G
82 9.49 Benzene, 1-methoxy-4-
methyl-
122 H
83 10.12 Benzaldehyde, 4-hydroxy- 122 H
84 13.45 2-Methoxy-6-methylphenol 138 G
85 14.95 Benzene, 1-ethyl-4-methoxy- 136 H
86 15.01 Phenol, 2-ethoxy- 138 G
87 18.02 1,2,3-Trimethoxybenzene 168 S
88 18.06 2,4-Dimethoxytoluene 152 G
89 19.42 Benzoic acid, 4-methoxy- 152 H
90 20.57 Benzoic acid, 5-hydroxy-2-
methoxy-
168 G
91 21.13 Benzene, 1,2,3-trimethoxy-5-
methyl
180 S
92 25.35 Phenol, 4-(3-hydroxy-1-
propenyl)-2-methoxy-
180 G
271
Table A2 The distribution of hydrodeoxygenation (HDO) products of flowthrough derived lignin
and the selectivity of HDO conversion with noble metal catalyst matrix
272
NO. Retention time (min) Compounds
2 13.049 Vanillin
4 15.606 Butylated Hydroxytoluene (BHT)
6 18.055 Phenol, 4-(1,1,3,3-tetramethylbutyl)- (antioxidants)
13 38.079 Phenol, 2,2'-methylenebis[6-(1,1-dimethylethyl)-4-methyl-
(antioxidants)
18 14.516 Phenol, 2-methoxy-4-propyl-
19 16.771 Phenol, 4-(1,1-dimethylpropyl)-
20 19.004 Phenol, 2-methyl-4-(1,1,3,3-tetramethylbutyl)-
26 9.856 Phenol, 4-tert-butyl-
27 10.503 Phenol, 3-methyl-5-(1-methylethyl)-
28 17.138 Phenol, 2,6-bis(1,1-dimethylethyl)-4-methyl-
29 22.155 Phenol, 2,5-bis(1-methylpropyl)-
Figure A3 GC/MS determination of water/dilute acid soluble lignin fragments from 240°C with
25 ml/min flow of 0.05% sulfuric acid; the compounds 4 and 13 are the potential high value
intermediates for biodiesel production
273
Figure A4 Determination of cellulose crystallinity before and after flowthrough pretreatment by
XRD
275
Figure A6 Application of SFG-VS in observing pretreated biomass substrate and in comparison
with XRD results
276
Proposed mass transfer modeling
The most common kinetic modeling is proposed by Saeman in 1945 446.
𝐻𝑒𝑚𝑖𝑐𝑒𝑙𝑙𝑢𝑙𝑜𝑠𝑒𝑘1→ 𝑋𝑦𝑙𝑜𝑠𝑒
𝑘2→ 𝐷𝑒𝑔𝑟𝑎𝑑𝑎𝑡𝑖𝑜𝑛 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠 (A1)
where 𝑘 is Arrhenius temperature dependence and can be calculated based on equation (A2).
𝑘𝑖 = 𝑘𝑖0 × 𝐴𝑚𝑖 × exp (−𝐸𝑖/𝑅𝑇) (A2)
where 𝑘𝑖0 is the preexponential factor; 𝐴 is the concentration of acid (wt %); 𝑚𝑖 is the power,
and 𝐸𝑖 is the activation energy.
𝐶𝑒𝑙𝑙𝑢𝑙𝑜𝑠𝑒 𝑘1→ 𝐺𝑙𝑢𝑐𝑜𝑠𝑒
𝑘2→ 𝐷𝑒𝑔𝑟𝑎𝑑𝑎𝑡𝑖𝑜𝑛 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠 (A3)
𝑑𝐶
𝑑𝑡= −𝑘1 × 𝐶 (A4)
𝑑𝐺
𝑑𝑡= 𝑘2 × 𝐺 − 𝑘1 × 𝐶 (A5)
Considering the dependence of Arrhenius temperature,
𝑘𝑖 = 𝑘𝑖0 × 𝐴𝑚𝑖 × exp(−𝐸𝑖/𝑅𝑇) (A6)
Thus, the glucose yield followed equation (A7).
𝐺 = 𝐶0 [𝑘1
𝑘1−𝑘2] × (𝑒−𝑘2𝑡 − 𝑒−𝑘1𝑡) + 𝐺0 × 𝑒−𝑘2𝑡 (A7)
in which G is the fraction of the total potential glucose; 𝐶0 𝑎𝑛𝑑 𝐺0 are the initial fractions of
cellulose and glucose, respectively. However, these modeling equations are assumed first order
and no mass transfer effects considered. Thus, the development of mass transfer modeling in
aqueous pretreatment is important since mass diffusion is closely related to pretreatment
performances.
Gas mass transfer coefficients are more accurate by experimental measurements than using
empirical equations. Liquid mass transfer coefficients were calculated mostly by empirical
277
equations summarized by experiences involving several factors, such as the viscosity of the
solution, molar mass, the chemicals, and solvent dissociation factor (Equation A8).
𝐷𝐴𝐵 =7.4×10−8(𝑎𝑀𝐵)0.5𝑇
𝜇𝑉𝐴0.6 (A8)
DAB-the mass transfer coefficient of liquid compound A in B, cm2/s; T-absolute temperature, K;
µ-the viscosity of the solution, mPa·s; MB-molar mass of B, kg/kmol; VA-component A molar
volume under the boiling point, cm3/mol. VA(water)=75.6, α-the solvent association factor,
α(water)=2.6.
This study proposed to measure liquid mass transfer coefficients experimentally.
Conductivity is a measurement of ability of electrolyte solution to conduct electricity (current).
Its unit is S/m (1 S/m=1 mho/m) and usually refers to the value at 25°C. The measurement of
solution conductivity is a fast, cheap and reliable way of measuring the ionic contents in a
solution. Conductivity is in direct proportion to total dissolved solids (T.D.S). It is well known
that 0.54-0.96 1µS/cm equals 0.64 mg of NaCl per kg of water. The measuring of solution
conductivity is through AC resistance of the solution between two electrodes. Usually, dilute
solution meets Kohlrausch’s laws of concentration dependence and additivity of ionic
contributions. If electrolytes dissociate completely in solution, Kohlrausch’s laws can be shown
as following 447.
𝛬𝑚 = 𝛬𝑚0 − 𝐾√𝑐
𝛬𝑚0 -limiting molar conductivity; K-empirical constant; c-electrolyte concentration
𝛬𝑚0 = 𝜐+𝜆+
0 +𝜐−𝜆−0
𝜐+ 𝜐− -the number of moles of cations and anions, respectively (1 mole of the dissolved
electrolyte); 𝜆+0 𝜆−
0 -limiting molar conductivities of the individual ions (For H+: 34.96; OH-:
278
19.91; Na+:5.011; SO42-:15.96).
Nowadays, the cell for measuring conductivity is built as electrical conductivity meter. The
cell includes a solution of an electrolyte, two flat or cylindrical electrodes, a fixed distance, an
alternating voltage. The conductivity is revealed by the resistance of the solution in cell which
can be expressed as followed.
𝑅 =𝑙
𝑆𝜌
𝜌-resistivity (material property); l and s are distance and surface area of two electrode poles.
Calibration of cell constant can be performed to avoid knowing l and s.
𝑅∗ = 𝐶𝜌∗
C-cell constant; 𝑅∗ and 𝜌∗ -known resistance and resistivity of standard compound. Specific
conductance, k (kappa):
𝜅 =1
𝜌=
𝐶
𝑅; 𝛬𝑚 =
𝜅
𝑐
c-solution concentration; 𝛬𝑚-molar conductivity. Therefore, the conductivity has a linear relation
with the concentration of the solution or the ionic content in the solution. The study will use
conductivity meter to determine the acid/alkali concentration change during their diffusion into
biomass particles. The solution concentration can’t be calculated simply by the above equations
since K is unknown. Thus, a diffusion model is needed to simulate solution concentration with t
to calculate De.
279
Figure A7 Designed measuring cell for mass transfer coefficients, (a) reported device (b)
designed device
The measurements of mass transfer coefficients were reported to use the reported device
(Figure 8.1a) 160. The unsteady state acid diffusion into a solid sphere in a well agitated tank was
used in this device and is a classical problem that the solution has already been reported shown
in equation (A9).
280
𝐹(𝑡) = 1 − ∑6𝛼(1+𝛼)
9+9𝛼+𝑞𝑛2𝛼2
exp (−𝑞𝑛
2𝐷𝑒𝑡
𝑅2)∞
𝑛=1 (A9)
Where 𝑞𝑛 are the non-zero roots of tan 𝑞𝑛 = 3𝑞𝑛/(3 + 𝛼𝑞𝑛2) , 𝛼 = 3𝑉/(4𝜋𝑅3) is the ratio of the
volumes of liquid and solid spheres, 𝐹 = (𝐶0 − 𝐶(𝑡))/(𝐶0 − 𝐶∞) is the fractional uptake, 𝐶(𝑡) is
the concentration at time 𝑡 and 𝐶∞ is the concentration at infinite time (Crank, 1983). Our study
applied C0/2 as the 𝐶∞ while the concentration at 30min was measured as 𝐶∞ in one study 168.
The designed system can avoid the reported apparatuses which biomass chips were directly
merged in bulk solution. In the designed device (Figure 8.1b), the cell was equipped with two
individual 50ml tubes held by two tube holders. The PDMS molds were prepared by mixing
elastomer and curing agents at a 10 parts to 1 part ratio (10:1) and dried at 105°C. The two
chambers were filled with 50ml acid (1% wt)/alkali (0.41% wt NaOH (0.0102 mol/L)) and DI
water, respectively, and well mixed. The conductivity meter (EXTECH instruments; Model
number: DO700) was used to measure the change of DI conductivity caused by acid or alkali
mass transfer from left chamber to right chamber. At room temperature, 4mm poplar wood chip
was molded in a PDMS plate and dried in 105°C.
The conductivity calibration curve was made by measuring conductivities of different
acid/alkali concentrations (Figure A9). The concentration change of right chamber during
measurements was calculated by referring to calibration curve on detected conductivities.
281
Figure A8 Calibration of conductivity (a) Sulfuric acid (b) NaOH
Different mass transfer models were selected to describe the two devices of measuring mass
transfer coefficients. In the designed device, a non-steady-state and one-dimensional mass
transfer equation can be expressed and used as followed:
𝜕𝐶
𝜕𝑡= 𝐷(𝑡)
𝜕2𝐶
𝜕𝑥2 (A10)
where C is the concentration of sulfuric acid or sodium hydroxide, D is the diffusion coefficient
of sulfuric acid or sodium hydroxide, and the initial and boundary conditions to solve the
equation include the following:
𝐶 = 𝐶0 = 0 when t=0
𝐶0 = 𝐶 at the edge of biomass particle when x=0.
The final solution is:
𝐶(𝑥, 𝑡) =1
2𝑐0 + ∑ 𝑐𝑛
∞𝑛=1 cos(
𝑛𝜋𝑥
𝐿)𝑒−𝐷𝑒(𝑛𝜋/𝐿)2𝑡 (A11)
Where 𝑐𝑛 =2
𝐿∫ cos (
𝑚𝜋𝑥
𝐿) 𝐷(𝑥, 0)𝑑𝑥
𝐿
0, C is the chemical concentration in wood but we measure
chemical concentration in water chamber. D is the coefficient. L is the particle length in
diffusion direction. X is the diffusion distance to the edge of biomass particle. (Source: Weisstein,
282
Eric W. "Heat Conduction Equation." From MathWorld--A Wolfram Web Resource.
http://mathworld.wolfram.com/HeatConductionEquation.html). C1,2,3 are the concentrations of
left chamber, wood sample and right chamber, respectively. What we measured is dC3/dt. The
concentrations C1(t) and C3(t) are constant while C2(t) decreased linearly. A refers to the tube
cross area (28.26cm2). L refers to the length of 10 cm. Then the mass rate was:
𝑑𝐶1
𝑑𝑡= −𝑘1𝐶1 + 𝑘2𝐶20
𝜕𝐶2
𝜕𝑡= 𝐷
𝑑2𝐶2
𝑑𝑥2
𝑑𝐶3
𝑑𝑡= 𝑘3𝐶3 − 𝑘2𝐶21
𝐶1 = 𝐶1(𝑡), 𝐶2 = 𝐶2(𝑥, 𝑡), 𝐶3 = 𝐶3(𝑡)
𝐶1(0) = 𝐶0, 𝐶2(𝑥, 0) = 0, 𝐶2(0, 𝑡) = 𝐶1, 𝐶3(0) = 0
𝐶20 = 𝐶2(0, 𝑡), 𝐶21 = 𝐶2(𝐿, 𝑡)
To simplify the model, we assumed that the volume of biomass was small in comparison with the
volume of chambers. Thus,
𝐷 ∗ 𝐴 ∗∆ 𝐶
𝐿= −𝑉 ∗
𝑑𝐶1
𝑑𝑡= 𝑉 ∗
𝑑𝐶3
𝑑𝑡
∆𝐶 = 𝐶1 − 𝐶3
𝐶1 + 𝐶3 = 𝐶0 = 1
𝑑𝐶3
𝑑𝑡=
𝐷∗𝐴
𝐿∗𝑉(𝐶0 − 2𝐶3)
𝐶3(0) = 0
𝐶3(𝑡) =𝐶0
2(1 − 𝑒−2𝑎𝑡), a=
𝐷∗𝐴
𝐿∗𝑉
Plot: 2𝐷𝐴
𝐿𝑉𝑡 = ln(
𝐶0
𝐶0−2𝐶3)
283
𝑑𝐶3
𝑑𝑡=
𝐷∗𝐴
𝐿∗𝑉(𝐶0 − 2𝐶3)
Bring known values in, we can get 0.113D=ln(1
1−2𝐶3). However, the designed device has a
limitation on the biomass particle sizes. The applicable particle size is larger than 4cm to avoid
the difficulties of imbedding particles with PDMS. Thus, the sub-millimeter particles applied in
our study applied the reported device to measure both the acid and alkali concentration changes
during their diffusion into the sub-millimeter particles.
The diffusion model in equation A9 was used to describe the diffusion process in the
reported device. The temperature dependence of diffusion coefficients can be expressed by the
simplest and commonly used Arrhenius model.
D=D0 exp (-Eact/kT) (A.12)
where Eact is activation energy, k is Boltzmann constant, T is temperature. The different values of
D at different T were measured. Polynomial expansion is ebT=1-bT+0.5b2T2+…. Thus, when
T=140, 160, 180, 190, and 200°C in our pretreatment conditions, D can be obtained to represent
the true mass transfer coefficient during pretreatment.
Cellulose and hemicellulose followed the similar kinetics. For hemicellulose (H) hydrolysis
to hemicellulose oligomers (HO) and then to hemicellulose monomers (HM), the kinetic
modeling equation can be expressed as following:
𝐻𝑘1→𝐻𝑂
𝑘2→𝐻𝑀
𝑘3→ 𝐻𝐷
𝑑[𝐻]
𝑑𝑡= −𝑘1[𝐻]
𝑑[𝐻𝑂]
𝑑𝑡= 𝑘1[𝐻] − 𝑘2[𝐻𝑂]
284
𝑑[𝐻𝑀]
𝑑𝑡= 𝑘2[𝐻𝑀] − 𝑘3[𝐻𝐷]
𝑘1 = 𝐴1[𝐶�̅�(𝑡)]𝑛1𝑒−𝐸1𝑅𝑇
𝑘2 = 𝐴2[𝐶�̅�(𝑡)]𝑛2𝑒−𝐸2𝑅𝑇
𝑘3 = 𝐴3[𝐶�̅�(𝑡)]𝑛3𝑒−𝐸3𝑅𝑇
𝜕𝐶𝑠(𝑟, 𝑡)
𝜕𝑡= 𝐷
𝑑2𝐶𝑠(𝑟, 𝑡)
𝑑𝑟2
Solutions:
𝐶𝐻(𝑡) = 𝐶𝐻0𝑒−𝑘1𝑡
𝐶𝐻𝑂(𝑡) = 𝐶𝐻0
𝑘1
𝑘2−𝑘1[𝑒−𝑘1𝑡 − 𝑒−𝑘2𝑡]
𝐶𝐻𝑀(𝑡) = 𝐶𝐻0
𝑘1𝑘2
(𝑘2−𝑘1)(𝑘3−𝑘1)[𝑒−𝑘1𝑡 − 𝑒−𝑘3𝑡] − 𝐶𝐻0
𝑘1𝑘2
(𝑘2−𝑘1)(𝑘3−𝑘2)[𝑒−𝑘2𝑡 − 𝑒−𝑘3𝑡]
𝐶𝐻(𝑡) = 𝐶𝐻0𝑒−𝑘1𝑡
𝐶𝐻𝑂(𝑡) = 𝐶𝐻0
𝑘1
𝑘2−𝑘1[𝑒−𝑘1𝑡 − 𝑒−𝑘2𝑡]
𝐶𝐻𝑀(𝑡) = 𝐶𝐻0
𝑘1𝑘2
(𝑘2−𝑘1)(𝑘3−𝑘1)[𝑒−𝑘1𝑡 − 𝑒−𝑘3𝑡] − 𝐶𝐻0
𝑘1𝑘2
(𝑘2−𝑘1)(𝑘3−𝑘2)[𝑒−𝑘2𝑡 − 𝑒−𝑘3𝑡]
𝐶𝑠(𝑟, 𝑡) = 𝐶0erf (𝑟
2√𝐷𝑡), where 𝐶0 =
2𝐶0√𝜋𝐷
𝐴; erf (r) =
2
√𝜋∫ 𝑒−𝑡2
𝑑𝑡𝑡
0
Notes:
H-hemicellulose; can get it from experiment
HO-hemi-sugars in oligomeric form; can get it from experiment
HM-hemi-sugars in monomeric form; can get it from experiment
HD-degradation of hemi-sugars; can get it from experiment
k1, k2, k3-kinetic reaction rate constants; unknown we should get from simulation
285
E1, E2, E3-activation energy for kinetics; unknown we should get from simulation
A1, A2, A3-pre-exponential factors for kinetic equations; unknown we should get from
simulation
n1, n2, n3-chemical concentration factor
R, T-gas constant (8.314) and temperature (T=413.15 K))
t-reaction time, 30,40,50,60,70 min
D-mass transfer coefficients, 0.25
r-biomass particle diameter
A-surface area of biomass particle
𝐶𝑠(𝑟, 𝑡) –concentration of pretreatment solvent inside the biomass particles, such as acid and
alkalis
Some preliminary results
Figure A9 Fractional uptake of NaOH, Douglas fir 1/8’’-16mesh, CB substrates
The conductivity changes in the solution were measured by the reported design (Figure A8a).
Thus, the fractional uptake was calculated to simulate with time to calculate mass transfer
diffusion coefficients according to equation A9. It was found that NaOH was much easier to
diffuse into biomass particles than sulfuric acid under tested temperatures (room temperature, 35,
1.0
0.8
0.6
0.4
0.2
0.0
Fra
ctio
na
l U
pta
ke
F(t
)
6005004003002001000
Time (s)
(b)
22°C
35°C
50°C
286
and 50; Figure A10). This was indicated by the high fractional uptakes of NaOH regardless of
time increasing. This conclusion can be used to explain the usual NaOH pretreatment that is
carried out under low temperatures. Additionally, the diffusion capacities of NaOH were not
showing significant changes with the increasing of tested temperatures. Thus, the NaOH
diffusion phenomenon was not obeying the uptake modeling (equation A9) and Arrhenius
temperature dependence of mass transfer coefficients. More detailed mass transfer study of
NaOH in pretreatment is still needed in the future. The acid diffusion coefficients using the same
equation were reported 1-20× 10−6 in the study of 0.1N sulfuric acid pretreatment of Bagasse,
Corn stover, Rice straw and Yellow poplar with particle sizes of 14-20 and 20-24 mesh at 25, 50,
and 75C 168. The fitting results of some feedstocks in our study under 1% (w/w) sulfuric acid
were shown in Table A3. The difference of our coefficients and the reported number might be
because the selection of concentration on solid surface at infinite time. Table A3 showed
crumbled particles possessed about four times better diffusion capacities than hammer milled
ones. The larger sizes showed better diffusion than finer particles. For biomass species, poplar
presented around two times higher diffusivities than Douglas fir. However, this work still need
more future work such as verification of the modeling feasibilities, improving data collection
method, and integrating mass transfer coefficients with kinetic modeling in simulating
experimental data of pretreatment.
Table A3 Some acid diffusivity of various sized and cut biomass feedstock
Biomass Cuttings Size Diffusivity (cm2/s) × 10−7
22C 35C 50C 75C
Douglas fir Crumbler 12-30mesh 0.53±0.055 0.90±0.076 9.96±1.24 19.65±1.73
287
Douglas fir Crumbler 30-60mesh 0.068±0.00081 0.11±0.011
Poplar Crumbler 12-30mesh 0.62±0.042 1.62±0.040 1.466±0.55 2.765±0.73
Poplar Crumbler 30-60mesh 0.13±0.013 0.28±0.013 0.377±0.20 0.598±0.38
Poplar Hammer 12-30mesh 0.37±0.020 1.73±0.075 3.00±0.11
288
REFERENCES
1. B. Yang, Y. Lu and C. E. Wyman, Biofuels from Agricultural Wastes and Byproducts. HP Blaschek, TC
Ezeji and J. Scheffran, Eds., Wiley-Blackwell: USA, 2010, 175-200.
2. R. D. Perlack, L. M. Eaton, A. F. Turhollow Jr, M. H. Langholtz, C. C. Brandt, M. E. Downing, R. L.
Graham, L. L. Wright, J. M. Kavkewitz and A. M. Shamey, 2011.
3. R. D. Perlack, L. L. Wright, A. F. Turhollow, R. L. Graham, B. J. Stokes and D. C. Erbach, Biomass as
feedstock for a bioenergy and bioproducts industry: the technical feasibility of a billion-ton annual supply,
DTIC Document, 2005.
4. L. R. Lynd, M. S. Laser, D. Bransby, B. E. Dale, B. Davison, R. Hamilton, M. Himmel, M. Keller, J. D.
McMillan and J. Sheehan, Nature biotechnology, 2008, 26, 169-172.
5. B. Yang and C. E. Wyman, APBN, 2007, 11(9), 555-563.
6. C. E. Wyman, Annual Review of Energy and the Environment, 1999, 24, 189-226.
7. C. Wyman, An overview of ethanol production for transportation fuels, National Renewable Energy Lab.,
Golden, CO (United States), 1993.
8. US. DOE, Office of Energy Efficiency and Renewable Energy, Washington, DC, 2012.
9. R. Wooley, M. Ruth, J. Sheehan, K. Ibsen, H. Majdeski and A. Galvez, Lignocellulosic biomass to ethanol
process design and economics utilizing co-current dilute acid prehydrolysis and enzymatic hydrolysis
current and futuristic scenarios, DTIC Document, 1999.
10. R. Davis, L. Tao, E. Tan, M. Biddy, G. Beckham, C. Scarlata, J. Jacobson, K. Cafferty, J. Ross and J. Lukas,
Process design and economics for the conversion of lignocellulosic biomass to hydrocarbons: Dilute-acid
and enzymatic Deconstruction of biomass to sugars and biological conversion of Sugars to Hydrocarbons,
National Renewable Energy Laboratory (NREL), Golden, CO., 2013.
11. L. Montague, A. Slayton and J. Lukas, Lignocellulosic biomass to ethanol process design and economics
utilizing co-current dilute acid prehydrolysis and enzymatic hydrolysis for corn stover, Citeseer, 2002.
12. R. Davis, A. Aden and P. T. Pienkos, Applied Energy, 2011, 88, 3524-3531.
13. L. Tao, A. Aden, R. T. Elander, V. R. Pallapolu, Y. Y. Lee, R. J. Garlock, V. Balan, B. E. Dale, Y. Kim, N. S.
Mosier, M. R. Ladisch, M. Falls, M. T. Holtzapple, R. Sierra, J. Shi, M. A. Ebrik, T. Redmond, B. Yang, C.
E. Wyman, B. Hames, S. Thomas and R. E. Warner, Bioresource Technology, 2011, 102, 11105-11114.
14. B. Yang and C. E. Wyman, Biofuels Bioproducts & Biorefining-Biofpr, 2008, 2, 26-40.
15. T. Hsu, Applied Energy Technology Series, Taylor & Francis, Washington DC, 1996, 179.
16. P. Harmsen, W. Huijgen, L. Bermudez and R. Bakker, 2010.
17. C. E. Wyman, Aqueous pretreatment of plant biomass for biological and chemical conversion to fuels and
chemicals, John Wiley & Sons, 2013.
18. X. Zhao, K. Cheng and D. Liu, Appl Microbiol Biotechnol, 2009, 82, 815-827.
289
19. R. S. Ramirez, M. Holtzapple and N. Piamonte, Aqueous Pretreatment of Plant Biomass for Biological and
Chemical Conversion to Fuels and Chemicals, 2013, 145-167.
20. H. L. Trajano and C. E. Wyman, Aqueous Pretreatment of Plant Biomass for Biological and Chemical
Conversion to Fuels and Chemicals, 2013, 103-128.
21. B. Yang and C. E. Wyman, Biofuels, Bioproducts and Biorefining, 2008, 2, 26-40.
22. N. Mosier, C. Wyman, B. Dale, R. Elander, Y. Y. Lee, M. Holtzapple and M. Ladisch, Bioresource
Technology, 2005, 96, 673-686.
23. M. Foston and A. J. Ragauskas, Industrial Biotechnology, 2012, 8, 191-208.
24. M. E. Himmel, S.-Y. Ding, D. K. Johnson, W. S. Adney, M. R. Nimlos, J. W. Brady and T. D. Foust,
science, 2007, 315, 804-807.
25. J. Zheng and L. Rehmann, International journal of molecular sciences, 2014, 15, 18967-18984.
26. Y. Lee, P. Iyer and R. W. Torget, in Recent progress in bioconversion of lignocellulosics, Springer, 1999,
pp. 93-115.
27. B. Yang and M. Tucker, Aqueous Pretreatment of Plant Biomass for Biological and Chemical Conversion
to Fuels and Chemicals, 2013, 489-521.
28. L. Zhang, Z. Lu, L. Velarde, L. Fu, Y. Pu, S.-Y. Ding, A. Ragauskas, H.-F. Wang and B. Yang, Cellulose,
2015, DOI: 10.1007/s10570-015-0588-0, 1-16.
29. J. D. McMillan, 1994.
30. N. Mosier, C. Wyman, B. Dale, R. Elander, Y. Lee, M. Holtzapple and M. Ladisch, Bioresource technology,
2005, 96, 673-686.
31. C. E. Wyman, B. E. Dale, R. T. Elander, M. Holtzapple, M. R. Ladisch and Y. Lee, Bioresource technology,
2005, 96, 1959-1966.
32. B. Yang and C. E. Wyman, Biofuels, Bioproducts and Biorefining, 2008, 2, 26-40.
33. B. Yang and C. E. Wyman, in Biofuels, Springer, 2009, pp. 103-114.
34. Y. Kim, N. S. Mosier, M. R. Ladisch, V. R. Pallapolu, Y. Y. Lee, R. Garlock, V. Balan, B. E. Dale, B. S.
Donohoe, T. B. Vinzant, R. T. Elander, M. Falls, R. Sierra, M. T. Holtzapple, J. Shi, M. A. Ebrik, T.
Redmond, B. Yang, C. E. Wyman and R. E. Warner, Bioresource Technology, 2011, 102, 11089-11096.
35. C. E. Wyman, B. E. Dale, R. T. Elander, M. Holtzapple, M. R. Ladisch and Y. Y. Lee, Bioresource
Technology, 2005, 96, 1959-1966.
36. R. T. Elander, B. E. Dale, M. Holtzapple, M. R. Ladisch, Y. Lee, C. Mitchinson, J. N. Saddler and C. E.
Wyman, Cellulose, 2009, 16, 649-659.
37. J. Weil, A. Sarikaya, S.-L. Rau, J. Goetz, C. Ladisch, M. Brewer, R. Hendrickson and M. Ladisch, Applied
Biochemistry and Biotechnology, 1997, 68, 21-40.
38. J. Weil, M. Brewer, R. Hendrickson, A. Sarikaya and M. Ladisch, Applied Biochemistry and Biotechnology,
290
1998, 70-72, 99-111.
39. J. Weil, A. Sarikaya, S.-L. Rau, J. Goetz, C. Ladisch, M. Brewer, R. Hendrickson and M. Ladisch, Applied
Biochemistry and Biotechnology, 1998, 73, 1-17.
40. M. Holtzapple, J.-H. Jun, G. Ashok, S. Patibandla and B. Dale, Applied Biochemistry and Biotechnology,
1991, 28-29, 59-74.
41. X. Chen, J. Shekiro, M. A. Franden, W. Wang, M. Zhang, E. Kuhn, D. K. Johnson and M. P. Tucker,
Biotechnology for Biofuels, 2012, 5, 1.
42. K. Kobayashi, S. Kimura, U.-J. Kim, K. Tokuyasu and M. Wada, Cellulose, 2012, 19, 967-974.
43. M. Wada, M. Ike and K. Tokuyasu, Polymer Degradation and Stability, 2010, 95, 543-548.
44. R. W. Torget, Patent US6228177, 2001.
45. R. W. Torget, K. L. Kadam, T.-A. Hsu, G. P. Philippidis and C. E. Wyman, Patent US5424417, 1996.
46. O. Bobleter, G. Bonn and R. Concin, Alternative Energy Sources, 1980, 3, 323-332.
47. H. Hörmeyer, W. Schwald, G. Bonn and O. Bobleter, Holzforschung-International Journal of the Biology,
Chemistry, Physics and Technology of Wood, 1988, 42, 95-98.
48. O. Bobleter, Progress in polymer science, 1994, 19, 797-841.
49. V. Archambault‐Léger, Z. Losordo and L. R. Lynd, Biofuels, Bioproducts and Biorefining, 2015, 9, 95-108.
50. B. Yang and C. E. Wyman, Bioresource Technology, 2008, 99, 5756-5762.
51. T. Sakaki, M. Shibata, T. Sumi and S. Yasuda, Industrial & Engineering Chemistry Research, 2002, 41,
661-665.
52. H. Ando, T. Sakaki, T. Kokusho, M. Shibata, Y. Uemura and Y. Hatate, Industrial & Engineering Chemistry
Research, 2000, 39, 3688-3693.
53. C. Liu and C. E. Wyman, Industrial & Engineering Chemistry Research, 2003, 42, 5409-5416.
54. C. G. Liu and C. E. Wyman, Applied Biochemistry and Biotechnology, 2004, 113, 977-987.
55. C. Liu and C. E. Wyman, Industrial & Engineering Chemistry Research, 2004, 43, 2781-2788.
56. B. Yang and C. E. Wyman, Biotechnology and bioengineering, 2004, 86, 88-98.
57. W. S. L. Mok and M. J. Antal, Industrial & Engineering Chemistry Research, 1992, 31, 1157-1161.
58. S. G. Allen, L. C. Kam, A. J. Zemann and M. J. Antal, Industrial & Engineering Chemistry Research, 1996,
35, 2709-2715.
59. L. Yan, L. Zhang and B. Yang, Biotechnology for Biofuels, 2014, 7, 76.
60. L. Zhang, L. Yan, Z. Wang, D. D. Laskar, M. S. Swita, J. R. Cort and B. Yang, Biotechnology for Biofuels,
291
2015, 8, 1-14.
61. V. Archambault-Leger, X. Shao and L. R. Lynd, Biotechnol. Biofuels, 2012, 5, 49.
62. D. D. Laskar, J. Zeng, L. Yan, S. Chen and B. Yang, Industrial Crops and Products, 2013, 50, 391-399.
63. M. Foston, H. L. Trajano, R. Samuel, C. E. Wyman, J. He and A. J. Ragauskas, Bioresource technology,
2015, 197, 128-136.
64. H. L. Trajano, S. Pattathil, B. A. Tomkins, T. J. Tschaplinski, M. G. Hahn, G. J. Van Berkel and C. E.
Wyman, Bioresource Technology, 2015, 179, 202-210.
65. H. L. Trajano, N. L. Engle, M. Foston, A. J. Ragauskas, T. J. Tschaplinski and C. E. Wyman, Biotechnology
for biofuels, 2013, 6, 110.
66. M. J. Antal, W. S. Mok and G. N. Richards, Carbohydrate research, 1990, 199, 91-109.
67. C. J. Knill and J. F. Kennedy, Carbohydrate Polymers, 2003, 51, 281-300.
68. J. Li, G. Henriksson and G. Gellerstedt, Bioresource Technology, 2007, 98, 3061-3068.
69. M. Wayman and J. H. Lora, Journal of Applied Polymer Science, 1980, 25, 2187-2194.
70. Q. Sun, Y. Pu, X. Meng, T. Wells and A. J. Ragauskas, ACS Sustainable Chemistry & Engineering, 2015, 3,
2203-2210.
71. F. S. Chakar and A. J. Ragauskas, Industrial Crops and Products, 2004, 20, 131-141.
72. M. J. Antal Jr, W. S. Mok and G. N. Richards, Carbohydrate research, 1990, 199, 91-109.
73. R. S. Ramirez, M. Holtzapple and N. Piamonte, Aqueous Pretreatment of Plant Biomass for Biological and
Chemical Conversion to Fuels and Chemicals, 145-167.
74. M. Möller and U. Schröder, RSC Advances, 2013, 3, 22253-22260.
75. S. Park, J. O. Baker, M. E. Himmel, P. A. Parilla and D. K. Johnson, Biotechnology for biofuels, 2010, 3, 1.
76. M. Wada, Journal of Polymer Science Part B: Polymer Physics, 2002, 40, 1095-1102.
77. K. Meyer and L. Misch, Helv. Chim. Acta, 1937, 20, 232-244.
78. Y. Nishiyama, P. Langan and H. Chanzy, Journal of the American Chemical Society, 2002, 124, 9074-9082.
79. Y. Nishiyama, J. Sugiyama, H. Chanzy and P. Langan, Journal of the American Chemical Society, 2003,
125, 14300-14306.
80. K. Schenzel, S. Fischer and E. Brendler, Cellulose, 2005, 12, 223-231.
81. B. Hinterstoisser and L. Salmén, Vibrational Spectroscopy, 2000, 22, 111-118.
82. K. Schenzel and S. Fischer, Cellulose, 2001, 8, 49-57.
292
83. H. Kim and J. Ralph, RSC Advances, 2014, 4, 7549.
84. S.-Y. Ding and Y.-S. Liu, in Biomass Conversion: Methods and Protocols, ed. E. M. Himmel, Humana
Press, Totowa, NJ, 2012, DOI: 10.1007/978-1-61779-956-3_3, pp. 23-30.
85. A. L. Barnette, L. C. Bradley, B. D. Veres, E. P. Schreiner, Y. B. Park, J. Park, S. Park and S. H. Kim,
Biomacromolecules, 2011, 12, 2434-2439.
86. M. Langholtz, M. Downing, R. Graham, F. Baker, A. Compere, W. Griffith, R. Boeman and M. Keller, SAE
International Journal of Materials & Manufacturing, 2014, 7, 115-121.
87. A. J. Ragauskas, G. T. Beckham, M. J. Biddy, R. Chandra, F. Chen, M. F. Davis, B. H. Davison, R. A.
Dixon, P. Gilna and M. Keller, Science, 2014, 344, 1246843.
88. J. Rencoret, G. Marques, A. Gutiérrez Suárez, L. Nieto Garrido, J. Jiménez-Barbero, Á. T. Martínez and J.
C. d. Río Andrade, 2008.
89. C. Heitner, D. Dimmel and J. Schmidt, Lignin and lignans: advances in chemistry, CRC press, 2010.
90. M. Lawoko, G. Henriksson and G. Gellerstedt, Biomacromolecules, 2005, 6, 3467-3473.
91. A. Bjorkman, Nature, 1954, 174, 1057-1058.
92. C. I. Ishizawa, T. Jeoh, W. S. Adney, M. E. Himmel, D. K. Johnson and M. F. Davis, Cellulose, 2009, 16,
677-686.
93. W. G. Glasser and S. Sarkanen, Lignin, properties and materials, Washington, DC (USA); American
Chemical Society, 1989.
94. H. Hatakeyama and T. Hatakeyama, in Biopolymers, Springer, 2009, pp. 1-63.
95. H. Hatakeyama and T. Hatakeyama, Lignin and Lignans: Advances in Chemistry. CRC Press, New York,
2010, 301-316.
96. T. Hatakeyama and H. Hatakeyama, Thermal properties of green polymers and biocomposites, Springer
Science & Business Media, 2006.
97. J. A. Schmidt, Lignin and lignans: Advances in chemistry, 2009, 49.
98. J. Lora, B. Naceur and A. Gandini, Industrial commercial lignins: sources, properties and applications,
Elsevier, Amsterdam, 2008.
99. J. Li, G. Gellerstedt and K. Toven, Bioresource technology, 2009, 100, 2556-2561.
100. P. Sannigrahi, A. J. Ragauskas and S. J. Miller, Bioenergy Research, 2008, 1, 205-214.
101. S. V. Pingali, V. S. Urban, W. T. Heller, J. McGaughey, H. O'Neill, M. Foston, D. A. Myles, A. Ragauskas
and B. R. Evans, Biomacromolecules, 2010, 11, 2329-2335.
102. S. Y. Lin and C. W. Dence, Methods in lignin chemistry, Springer, 1992.
103. M. J. Selig, S. Viamajala, S. R. Decker, M. P. Tucker, M. E. Himmel and T. B. Vinzant, Biotechnology
293
Progress, 2007, 23, 1333-1339.
104. B. S. Donohoe, S. R. Decker, M. P. Tucker, M. E. Himmel and T. B. Vinzant, Biotechnol Bioeng, 2008, 101,
913-925.
105. H. Li, Y. Pu, R. Kumar, A. J. Ragauskas and C. E. Wyman, Biotechnology and bioengineering, 2014, 111,
485-492.
106. P. Sannigrahi, D. H. Kim, S. Jung and A. Ragauskas, Energy & Environmental Science, 2011, 4, 1306-
1310.
107. B. S. Donohoe, S. R. Decker, M. P. Tucker, M. E. Himmel and T. B. Vinzant, Biotechnology and
Bioengineering, 2008, 101, 913-925.
108. R. Sun and J. Tomkinson, Separation and Purification Technology, 2001, 24, 529-539.
109. A. Toledano, A. García, I. Mondragon and J. Labidi, Separation and Purification Technology, 2010, 71, 38-
43.
110. K. R. Colyar, J. Pellegrino and K. Kadam, Separation Science and Technology, 2008, 43, 447-476.
111. Y. Chen, R. R. Sharma-Shivappa, D. Keshwani and C. Chen, Applied Biochemistry and Biotechnology,
2007, 142, 276-290.
112. V. S. Chang, B. Burr and M. T. Holtzapple, in Biotechnology for Fuels and Chemicals, Springer, 1997, pp.
3-19.
113. S. Kim and M. T. Holtzapple, Bioresource Technology, 2005, 96, 1994-2006.
114. F. Teymouri, L. Laureano-Perez, H. Alizadeh and B. E. Dale, Applied Biochemistry and Biotechnology,
2004, 113, 951-963.
115. O. Sánchez, C. J. Alméciga-Díaz and R. Sierra, Delignification process of Agro-Industrial wastes an
alternative to obtain fermentable carbohydrates for producing fuel, INTECH Open Access Publisher, 2011.
116. Y. Sun and J. Cheng, Bioresource technology, 2002, 83, 1-11.
117. X. Zhao, K. Cheng and D. Liu, Applied microbiology and biotechnology, 2009, 82, 815-827.
118. R. W. Thring, E. Chornet and R. P. Overend, Biomass, 1990, 23, 289-305.
119. X. Pan, N. Gilkes and J. N. Saddler, Holzforschung, 2006, 60, 398-401.
120. P. Sannigrahi and A. J. Ragauskas, Aqueous pretreatment of plant biomass for biological and chemical
conversion to fuels and chemicals, 2013, 201-222.
121. T. J. McDonough, The chemistry of organosolv delignification, IPST technical paper series; number 455,
1992.
122. R. El Hage, N. Brosse, P. Sannigrahi and A. Ragauskas, Polymer Degradation and Stability, 2010, 95, 997-
1003.
294
123. P. Sannigrahi, A. J. Ragauskas and S. J. Miller, Energy & Fuels, 2009, 24, 683-689.
124. R. El Hage, N. Brosse, L. Chrusciel, C. Sanchez, P. Sannigrahi and A. Ragauskas, Polymer Degradation
and Stability, 2009, 94, 1632-1638.
125. L. J. Jönsson, B. Alriksson and N.-O. Nilvebrant, Biotechnology for biofuels, 2013, 6, 1.
126. K. V. Sarkanen and C. H. Ludwig, Lignins: occurrence, formation, structure and reactions., 1971.
127. C. Pouteau, P. Dole, B. Cathala, L. Averous and N. Boquillon, Polymer Degradation and Stability, 2003,
81, 9-18.
128. A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter and D. Templeton, National Renewable Energy
Laboratory, Golden, CO, 2006.
129. K. Iiyama and A. F. Wallis, Journal of the Science of Food and Agriculture, 1990, 51, 145-161.
130. A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton and D. Crocker, Report N. TP-510-
42618, 2011, 17.
131. T. Standard, Acid-insoluble lignin in wood and pulp Tappi, T222 om-02, 2002.
132. C. W. Dence, in Methods in lignin chemistry, Springer, 1992, pp. 33-61.
133. F. Beall, Wood Fiber, 1969, 1, 215-226.
134. M. G. Grønli, G. Varhegyi and C. Di Blasi, Industrial & Engineering Chemistry Research, 2002, 41, 4201-
4208.
135. Z. Strassberger, P. Prinsen, F. van der Klis, D. S. van Es, S. Tanase and G. Rothenberg, Green Chemistry,
2015, 17, 325-334.
136. F. Martin, C. Saiz-Jimenez and F. Gonzalez-Vila, Holzforschung, 1979, 33, 210-212.
137. G. Brunow, Biopolymers Online, 2005.
138. L.-S. Johansson, J. M. Campbell, K. Koljonen and P. Stenius, Applied Surface Science, 1999, 144–145, 92-
95.
139. T. Haensel, M. Reinmöller, P. Lorenz, W. J. Beenken, S. Krischok and S. I.-U. Ahmed, Cellulose, 2012, 19,
1005-1011.
140. D. Takada, K. Ehara and S. Saka, Journal of Wood Science, 50, 253-259.
141. Y. Pu, S. Cao and A. J. Ragauskas, Energy & Environmental Science, 2011, 4, 3154-3166.
142. J. Ralph and L. L. Landucci, 2010.
143. S. Li and K. Lundquist, Nordic Pulp & Paper Research Journal, 1994, 9, 191-195.
144. Y. Pu, B. Hallac and A. J. Ragauskas, in Aqueous Pretreatment of Plant Biomass for Biological and
Chemical Conversion to Fuels and Chemicals, 2013, pp. 369-390.
295
145. L. Fu, S. A. McCallum, J. Miao, C. Hart, G. J. Tudryn, F. Zhang and R. J. Linhardt, Fuel, 2015, 141, 39-45.
146. A. T. Martínez, A. González, M. Valmaseda, B. Dale, M. Lambregts and J. Haw, Holzforschung-
International Journal of the Biology, Chemistry, Physics and Technology of Wood, 1991, 45, 49-54.
147. S. Raiskila, M. Pulkkinen, T. Laakso, K. Fagerstedt, M. Loija, R. Mahlberg, L. Paajanen, A.-C. Ritschkoff
and P. Saranpaa, Silva Fennica, 2007, 41, 351.
148. A. Karmanov and O. Y. Derkacheva, Russian Journal of Bioorganic Chemistry, 2013, 39, 677-685.
149. U. P. Agarwal and S. A. Ralph, Holzforschung, 2008, 62, 667-675.
150. A. W. Pelzer, M. R. Sturgeon, A. J. Yanez, G. Chupka, M. H. O’Brien, R. Katahira, R. D. Cortright, L.
Woods, G. T. Beckham and L. J. Broadbelt, ACS Sustainable Chemistry & Engineering, 2015, 3, 1339-
1347.
151. C. Rolando, B. Monties and C. Lapierre, in Methods in Lignin Chemistry, eds. S. Y. Lin and C. W. Dence,
Springer Berlin Heidelberg, Berlin, Heidelberg, 1992, DOI: 10.1007/978-3-642-74065-7_23, pp. 334-349.
152. K. M. Holtman, H.-M. Chang, H. Jameel and J. F. Kadla, Journal of agricultural and food chemistry, 2003,
51, 3535-3540.
153. G. Gellerstedt, in Methods in Lignin Chemistry, Springer, 1992, pp. 322-333.
154. J. I. Hedges and J. R. Ertel, Analytical Chemistry, 1982, 54, 174-178.
155. E. Säwén, 2011.
156. P. Karhunen, P. Rummakko, J. Sipilä, G. Brunow and I. Kilpeläinen, Tetrahedron Letters, 1995, 36, 169-
170.
157. J. Ralph, T. Akiyama, H. Kim, F. Lu, P. F. Schatz, J. M. Marita, S. A. Ralph, M. S. Reddy, F. Chen and R.
A. Dixon, Journal of Biological Chemistry, 2006, 281, 8843-8853.
158. L. Zhang and G. Gellerstedt, Magnetic Resonance in Chemistry, 2007, 45, 37-45.
159. J. Ralph, J. M. Marita, S. A. Ralph, R. D. Hatfield, F. Lu, R. M. Ede, J. Peng and L. L. Landucci, 1999.
160. S. B. Kim and Y. Y. Lee, Bioresource Technology, 2002, 83, 165-171.
161. J. S. Luterbacher, J.-Y. Parlange and L. P. Walker, Biotechnology and Bioengineering, 2013, 110, 127-136.
162. J. Y. Zhu and X. J. Pan, Bioresource Technology, 2010, 101, 4992-5002.
163. Z.-H. Liu, L. Qin, F. Pang, M.-J. Jin, B.-Z. Li, Y. Kang, B. E. Dale and Y.-J. Yuan, Industrial Crops and
Products, 2013, 44, 176-184.
164. I. F. Cullis, J. N. Saddler and S. D. Mansfield, Biotechnology and bioengineering, 2004, 85, 413-421.
165. J. D. DeMartini, M. Foston, X. Meng, S. Jung, R. Kumar, A. J. Ragauskas and C. E. Wyman,
Biotechnology for Biofuels, 2015, 8, 1-16.
296
166. J. F. Saeman, Industrial & Engineering Chemistry, 1945, 37, 43-52.
167. S. Harun, V. Balan, M. S. Takriff, O. Hassan, J. Jahim and B. Dale, Biotechnology for biofuels, 2013, 6, 40.
168. S. B. Kim and Y. Lee, Bioresource Technology, 2002, 83, 165-171.
169. B. Yang, Z. Dai, S.-Y. Ding and C. E. Wyman, Biofuels, 2011, 2, 421-450.
170. C. Somerville, S. Bauer, G. Brininstool, M. Facette, T. Hamann, J. Milne, E. Osborne, A. Paredez, S.
Persson, T. Raab, S. Vorwerk and H. Youngs, Science, 2004, 306, 2206-2211.
171. C. M. Lee, A. Mittal, A. L. Barnette, K. Kafle, Y. B. Park, H. Shin, D. K. Johnson, S. Park and S. H. Kim,
Cellulose, 2013, 20, 991-1000.
172. J. Hayashi, A. Sufoka, J. Ohkita and S. Watanabe, Journal of Polymer Science: Polymer Letters Edition,
1975, 13, 23-27.
173. R. H. V. Atalla, David L., Science, 1984, 223, 283-285.
174. B. H. Henri Chanzy, FEBS Lett, 1985, 184, 285–288.
175. Y. Habibi, L. A. Lucia and O. J. Rojas, Chemical reviews, 2010, 110, 3479-3500.
176. A. N. Fernandes, L. H. Thomas, C. M. Altaner, P. Callow, V. T. Forsyth, D. C. Apperley, C. J. Kennedy and
M. C. Jarvis, Proceedings of the National Academy of Sciences, 2011, 108, E1195-E1203.
177. L. H. Thomas, V. T. Forsyth, A. Sturcova, C. J. Kennedy, R. P. May, C. M. Altaner, D. C. Apperley, T. J.
Wess and M. C. Jarvis, Plant physiology, 2013, 161, 465-476.
178. M. E. Himmel, S. Y. Ding, D. K. Johnson, W. S. Adney, M. R. Nimlos, J. W. Brady and T. D. Foust,
Science, 2007, 315, 804-807.
179. S.-Y. Ding and Y.-S. Liu, in Biomass Conversion, ed. M. E. Himmel, Humana Press, 2012, vol. 908, ch. 3,
pp. 23-30.
180. R. R. Lahiji, X. Xu, R. Reifenberger, A. Raman, A. Rudie and R. J. Moon, Langmuir, 2010, 26, 4480-4488.
181. Y. Pu, B. Hallac and A. J. Ragauskas, in Aqueous Pretreatment of Plant Biomass for Biological and
Chemical Conversion to Fuels and Chemicals, John Wiley & Sons, Ltd, 2013, DOI:
10.1002/9780470975831.ch18, pp. 369-390.
182. R. Samuel, Y. Pu, M. Foston and A. J. Ragauskas, Biofuels, 2010, 1, 85-90.
183. H. E. Larsson PT, Wickholm K, Pettersson E, Iversen T, Solid State Nuclear Magnetic Resonance, 1999,
15, 31-40.
184. M. Dick-Pérez, Y. Zhang, J. Hayes, A. Salazar, O. A. Zabotina and M. Hong, Biochemistry, 2011, 50, 989-
1000.
185. A. Isogai, M. Usuda, T. Kato, T. Uryu and R. H. Atalla, Macromolecules, 1989, 22, 3168-3172.
186. C. Y. Liang and R. H. Marchessault, Journal of Polymer Science, 1959, 39, 269-278.
297
187. P. J. H. Lynette M. Daviesa, Roger H. Newman, Carbohydr Res, 2002, 337, 587-593.
188. J. Mann and H. J. Marrinan, Journal of Polymer Science, 1958, 32, 357-370.
189. H. J. Marrinan and J. Mann, Journal of Polymer Science, 1956, 21, 301-311.
190. H. Morikawa, R. Hayashi and M. Senda, Plant Cell Physiol., 1978, 19, 1151-1159.
191. M. L. Nelson and R. T. O'Connor, Journal of Applied Polymer Science, 1964, 8, 1311-1324.
192. R. H. Newman, M.-A. Ha and L. D. Melton, Journal of Agricultural and Food Chemistry, 1994, 42, 1402-
1406.
193. M. Tsuboi, Journal of Polymer Science, 1957, 25, 159-171.
194. S. Park, J. O. Baker, M. E. Himmel, P. A. Parilla and D. K. Johnson, Biotechnology for biofuels, 2010, 3,
10.
195. P. Zugenmaier, 2007, DOI: 10.1002/bbpc.19850891120.
196. A. L. Barnette, C. Lee, L. C. Bradley, E. P. Schreiner, Y. B. Park, H. Shin, D. J. Cosgrove, S. Park and S. H.
Kim, Carbohydrate Polymers, 2012, 89, 802-809.
197. K. Kafle, X. Xi, C. M. Lee, B. R. Tittmann, D. J. Cosgrove, Y. B. Park and S. H. Kim, Cellulose, 2013, 21,
1075-1086.
198. S. Kim, C. Lee and K. Kafle, Korean J. Chem. Eng., 2013, 30, 2127-2141.
199. C. M. Lee, N. M. A. Mohamed, H. D. Watts, J. D. Kubicki and S. H. Kim, Journal of Physical Chemistry
B, 2013, 117, 6681-6692.
200. Y. B. Park, C. M. Lee, B. W. Koo, S. Park, D. J. Cosgrove and S. H. Kim, Plant physiology, 2013, 163,
907-913.
201. C. M. Lee, K. Kafle, Y. B. Park and S. H. Kim, Physical Chemistry Chemical Physics, 2014, 16, 10844-
10853.
202. Y. R. Shen, The Principles of Nonlinear Optics, Wiley-Interscience, New York, 1984.
203. Y. R. Shen, Nature, 1989, 337, 519-525.
204. K. B. Eisenthal, Abstracts of Papers of the American Chemical Society, 1997, 213, 69-COLL.
205. Y. R. Shen, Surface spectroscopy by nonlinear optics, 1994.
206. H.-F. Wang, W. Gan, R. Lu, Y. Rao and B.-H. Wu, Int. Rev. Phys. Chem., 2005, 24, 191-256.
207. W. E. Asher and E. Willard-Schmoe, Applied Spectroscopy, 2013, 67, 253-260.
208. H. Kim, A. Lagutchev and D. D. Dlott, Propellants Explos. Pyrotech., 2006, 31, 116-123.
209. E. Surber, A. Lozano, A. Lagutchev, H. Kim and D. D. Dlott, J. Phys. Chem. C, 2007, 111, 2235-2241.
298
210. X. Zhuang, P. B. Miranda, D. Kim and Y. R. Shen, Physical Review B, 1999, 59, 12632-12640.
211. L. Velarde and H.-F. Wang, Physical Chemistry Chemical Physics, 2013, DOI: 10.1039/C3CP52577E.
212. C. M. Lee, N. M. Mohamed, H. D. Watts, J. D. Kubicki and S. H. Kim, Journal of Physical Chemistry B,
2013, 117, 6681-6692.
213. L. Velarde and H. F. Wang, Chemical Physics Letters, 2013, 585, 42-48.
214. L. Velarde and H. F. Wang, Journal of Chemical Physics, 2013, 139, 084204.
215. L. Velarde, X. Y. Zhang, Z. Lu, A. G. Joly, Z. M. Wang and H. F. Wang, Journal of Chemical Physics, 2011,
135, 241102.
216. T. Imai, J. Sugiyama, T. Itoh and F. Horii, Journal of Structural Biology, 1999, 127, 248-257.
217. S. K. Yoshiharu N, Masahisa W, Takeshi O, Macromolecules, 1997, 30, 6395–6397.
218. J. Sugiyama, H. Chanzy and G. Maret, Macromolecules, 1992, 25, 4232-4234.
219. S.-Y. Ding, Q. Xu, M. K. Ali, J. O. Baker, E. A. Bayer, Y. Barak, R. Lamed, J. Sugiyama, G. Rumbles and
M. E. Himmel, BioTechniques, 2006, 41, 435.
220. A. R. Felmy, O. Qafoku, B. W. Arey, J. Z. Hu, M. Hu, H. Todd Schaef, E. S. Ilton, N. J. Hess, C. I. Pearce,
J. Feng and K. M. Rosso, Geochimica et Cosmochimica Acta, 2012, 91, 271-282.
221. H.-F. Wang, W. Gan, R. Lu , Y. Rao and B.-H. Wu, Int. Rev. Phys. Chem., 2005, 24, 191-256.
222. R. Lu, W. Gan, B. H. Wu, H. Chen and H. F. Wang, Journal of Physical Chemistry B, 2004, 108, 7297-
7306.
223. R. Lu, W. Gan, B. H. Wu, Z. Zhang, Y. Guo and H. F. Wang, Journal of Physical Chemistry B, 2005, 109,
14118-14129.
224. S.-Y. Ding, S. Zhao and Y. Zeng, Cellulose, 2013, 21, 863-871.
225. Kovalenko, Russian Chemical Reviews, 2010, 79, 231-241.
226. M. G. Paterlini, T. B. Freedman and L. A. Nafie, Journal of the American Chemical Society, 1986, 108,
1389-1397.
227. T. Taniguchi and K. Monde, Organic & biomolecular chemistry, 2007, 5, 1104-1110.
228. C. M. Lee, K. Kafle, Y. B. Park and S. H. Kim, Physical chemistry chemical physics : PCCP, 2014, 16,
10844-10853.
229. K. Kafle, R. Shi, C. M. Lee, A. Mittal, Y. B. Park, Y.-H. Sun, S. Park, V. Chiang and S. H. Kim, Cellulose,
2014, 21, 2219-2231.
230. Y. B. Park, C. M. Lee, K. Kafle, S. Park, D. J. Cosgrove and S. H. Kim, Biomacromolecules, 2014, 15,
2718-2724.
299
231. G. Ma and H. C. Allen, The Journal of Physical Chemistry B, 2003, 107, 6343-6349.
232. U. Bornscheuer, K. Buchholz and J. Seibel, Angewandte Chemie International Edition, 2014, 53, 10876-
10893.
233. H. Zhao, J. H. Kwak, Z. C. Zhang, H. M. Brown, B. W. Arey and J. E. Holladay, Carbohydrate polymers,
2007, 68, 235-241.
234. K. Igarashi, T. Uchihashi, A. Koivula, M. Wada, S. Kimura, T. Okamoto, M. Penttilä, T. Ando and M.
Samejima, Science, 2011, 333, 1279-1282.
235. D. Klemm, F. Kramer, S. Moritz, T. Lindström, M. Ankerfors, D. Gray and A. Dorris, Angewandte Chemie
International Edition, 2011, 50, 5438-5466.
236. H. Chanzy, Journal, 1993.
237. D. Klemm, B. Heublein, H. P. Fink and A. Bohn, Angewandte Chemie International Edition, 2005, 44,
3358-3393.
238. L. H. Sperling, Introduction to physical polymer science, John Wiley & Sons, 2005.
239. M. Wada, Journal of Polymer Science Part B: Polymer Physics, 2002, 40, 1095-1102.
240. M. Wada, T. Kondo and T. Okano, Polym J, 2003, 35, 155-159.
241. M. Wada, R. Hori, U.-J. Kim and S. Sasaki, Polymer Degradation and Stability, 2010, 95, 1330-1334.
242. E. Debzi, H. Chanzy, J. Sugiyama, P. Tekely and G. Excoffier, Macromolecules, 1991, 24, 6816-6822.
243. H. Yamamoto, F. Horii and H. Odani, Macromolecules, 1989, 22, 4130-4132.
244. S. L. Price, Physical Chemistry Chemical Physics, 2008, 10, 1996-2009.
245. J. F. Matthews, M. E. Himmel and M. F. Crowley, Cellulose, 2012, 19, 297-306.
246. J. Sugiyama, J. Persson and H. Chanzy, Macromolecules, 1991, 24, 2461-2466.
247. A. Watanabe, S. Morita and Y. Ozaki, Biomacromolecules, 2007, 8, 2969-2975.
248. A. Watanabe, S. Morita and Y. Ozaki, Biomacromolecules, 2006, 7, 3164-3170.
249. F. Horii, H. Yamamoto, R. Kitamaru, M. Tanahashi and T. Higuchi, Macromolecules, 1987, 20, 2946-2949.
250. T. R. Bhuiyan and N. Hirai, Journal of Wood Science, 2005, 51, 42-47.
251. M. T. R. Bhuiyan, N. Hirai and N. Sobue, Journal of Wood Science, 46, 431-436.
252. J. Sugiyama, R. Vuong and H. Chanzy, Macromolecules, 1991, 24, 4168-4175.
253. O. Biermann, E. Hädicke, S. Koltzenburg and F. Müller‐Plathe, Angewandte Chemie International Edition,
2001, 40, 3822-3825.
300
254. A. A. Baker, W. Helbert, J. Sugiyama and M. J. Miles, Journal of structural biology, 1997, 119, 129-138.
255. A. Baker, W. Helbert, J. Sugiyama and M. Miles, Biophysical Journal, 2000, 79, 1139-1145.
256. A. Baker, W. Helbert, J. Sugiyama and M. Miles, Applied Physics A, 1998, 66, S559-S563.
257. W. L. Earl and D. VanderHart, Macromolecules, 1981, 14, 570-574.
258. K. Wickholm, P. T. Larsson and T. Iversen, Carbohydrate Research, 1998, 312, 123-129.
259. H. Yamamoto and F. Horii, Macromolecules, 1993, 26, 1313-1317.
260. R. J. Viëtor, R. H. Newman, M. A. Ha, D. C. Apperley and M. C. Jarvis, The Plant Journal, 2002, 30, 721-
731.
261. T. Imai and J. Sugiyama, Macromolecules, 1998, 31, 6275-6279.
262. B. G. Saar, Y. Zeng, C. W. Freudiger, Y. S. Liu, M. E. Himmel, X. S. Xie and S. Y. Ding, Angewandte
Chemie International Edition, 2010, 49, 5476-5479.
263. H. Chanzy, M. Dube, R. Marchessault and J. Revol, Biopolymers, 1979, 18, 887-898.
264. M. C. Jarvis and D. C. Apperley, Plant physiology, 1990, 92, 61-65.
265. S. Park, J. O. Baker, M. E. Himmel, P. A. Parilla and D. K. Johnson, Biotechnol Biofuels, 2010, 3.
266. H.-F. Wang*, W. Gan†‡, R. Lu†‡ §, Y. Rao†‡¶ and B.-H. Wu†, International Reviews in Physical
Chemistry, 2005, 24, 191-256.
267. Y. Shen, Applied Physics A Solids and Surfaces, 1994, 59, 541-543.
268. L. Zhang, Z. Lu, L. Velarde, L. Fu, Y. Pu, S.-Y. Ding, A. J. Ragauskas, H.-F. Wang and B. Yang, Cellulose,
2015, 22, 1469-1484.
269. J. Löbau and K. Wolfrum, JOSA B, 1997, 14, 2505-2512.
270. M. Flörsheimer, C. Brillert and H. Fuchs, Langmuir, 1999, 15, 5437-5439.
271. M. S. Yeganeh, S. M. Dougal and B. G. Silbernagel, Langmuir, 2006, 22, 637-641.
272. Y. Liu and M. C. Messmer, The Journal of Physical Chemistry B, 2003, 107, 9774-9779.
273. S. B. Waldrup and C. T. Williams, Catalysis Communications, 2007, 8, 1373-1376.
274. C. T. Williams, Y. Yang and C. D. Bain, Langmuir, 2000, 16, 2343-2350.
275. S. Kweskin, R. Rioux, H. Song, K. Komvopoulos, P. Yang and G. Somorjai, ACS Catalysis, 2012, 2, 2377-
2386.
276. K. Gautam, A. D. Schwab, A. Dhinojwala, D. Zhang, S. Dougal and M. Yeganeh, Physical review letters,
2000, 85, 3854.
301
277. Z. Lu, A. Karakoti, L. Velarde, W. Wang, P. Yang, S. Thevuthasan and H.-f. Wang, The Journal of Physical
Chemistry C, 2013, 117, 24329-24338.
278. L. Velarde, X.-y. Zhang, Z. Lu, A. G. Joly, Z. Wang and H.-f. Wang, The Journal of chemical physics, 2011,
135, 241102.
279. Y. B. Park, C. M. Lee, B.-W. Koo, S. Park, D. J. Cosgrove and S. H. Kim, Plant physiology, 2013, 163,
907-913.
280. B. Yang, D. M. Willies and C. E. Wyman, Biotechnology and Bioengineering, 2006, 94, 1122-1128.
281. R. D. Preston and J. Cronshaw, Nature, 1958, 181, 248-250.
282. L. H. Sperling, Introduction to physical polymer science, John Wiley & Sons, 2015.
283. H. Chum, L. Douglas, D. Feinberg and H. Schroeder, Evaluation of pretreatments of biomass for enzymatic
hydrolysis of cellulose.[Organosolv process, wet oxidation, and steam explosion of wood chips] , Solar
Energy Research Inst., Golden, CO (USA); Colorado State Univ., Fort Collins (USA), 1985.
284. X. Zhuang, P. Miranda, D. Kim and Y. Shen, Physical Review B, 1999, 59, 12632.
285. R. Lu, W. Gan, B.-h. Wu, H. Chen and H.-f. Wang, The Journal of Physical Chemistry B, 2004, 108, 7297-
7306.
286. P. Zugenmaier, Crystalline Cellulose and Derivatives, Springer Berlin Heidelberg, 2008.
287. M. G. Paterlini, T. B. Freedman and L. A. Nafie, Biopolymers, 1986, 25, 1751-1765.
288. C. Y. Liang and R. H. Marchessault, Journal of Polymer Science, 1959, 37, 385-395.
289. Y. Guo and P. Wu, Carbohydrate Polymers, 2008, 74, 509-513.
290. T. Kondo, Cellulose, 1997, 4, 281-292.
291. K. Tashiro and M. Kobayashi, Polymer, 1991, 32, 1516-1526.
292. F. Shimazu and C. Sterling, Journal of Food Science, 1966, 31, 548-551.
293. A. Rollett, F. Humphreys, G. S. Rohrer and M. Hatherly, Recrystallization and related annealing
phenomena, Elsevier, 2004.
294. F. Shafizadeh, Thermal uses and properties of carbohydrates and lignins, Elsevier, 2012.
295. B. Henrissat, Cellulose, 1994, 1, 169-196.
296. B. Yang, Z. Dai, S.-Y. Ding and C. E. Wyman, Biofuels, 2011, 2, 421-449.
297. K. Igarashi, M. Wada, R. Hori and M. Samejima, Febs Journal, 2006, 273, 2869-2878.
298. N. Hayashi, J. Sugiyama, T. Okano and M. Ishihara, Carbohydrate Research, 1997, 305, 261-269.
299. W. E. Mabee, D. J. Gregg, C. Arato, A. Berlin, R. Bura, N. Gilkes, O. Mirochnik, X. Pan, E. K. Pye and J.
302
N. Saddler, Applied Biochemistry and Biotechnology, 2006, 129, 55-70.
300. M. Galbe and G. Zacchi, Applied microbiology and biotechnology, 2002, 59, 618-628.
301. J. Sun, S.-Y. Ding and J. Doran-Peterson, Biological Concerstion of Biomass for Fuels and Chemicals:
Exploration from Natural Utilization Systems, Royal Society of Chemistry, 2013.
302. D. Fengel and G. Wegener, Wood: chemistry, ultrastructure, reactions, Walter de Gruyter, 1983.
303. E. Adler, Wood Sci. Technol., 1977, 11, 169-218.
304. A. J. Panshin and C. d. Zeeuw, Textbook of wood technology. Volume I. Structure, identification, uses, and
properties of the commercial woods of the United States and Canada., 1970.
305. Q. A. Nguyen, M. P. Tucker, F. A. Keller and F. P. Eddy, Applied Biochemistry and Biotechnology, 2000,
84-6, 561-576.
306. B. Yang, A. Boussaid, S. D. Mansfield, D. J. Gregg and J. N. Saddler, Biotechnology and bioengineering,
2002, 77, 678-684.
307. L. Kumar, V. Arantes, R. Chandra and J. Saddler, Bioresource technology, 2012, 103, 201-208.
308. L. Kumar, R. Chandra, P. A. Chung and J. Saddler, Bioresource technology, 2010, 101, 7827-7833.
309. Y.-C. He, Y. Ding, Y.-F. Xue, B. Yang, F. Liu, C. Wang, Z.-Z. Zhu, Q. Qing, H. Wu and C. Zhu, Bioresource
technology, 2015, 193, 324-330.
310. e. Karp, M. G. Resch, B. S. Donohoe, P. N. Ciesielski, M. H. O'Brien, J. E. Nill, A. Mittal, M. Biddy and G.
T. Beckham, ACS Sustainable Chemistry & Engineering, 2015.
311. J. Lehto, H. Pakkanen and R. Alén, Journal of Wood Chemistry and Technology, 2015, 35, 337-347.
312. R. Sierra, M. T. Holtzapple and C. B. Granda, AIChE Journal, 2011, 57, 1320-1328.
313. V. S. Chang and M. T. Holtzapple, 2000.
314. R. Gupta and Y. Y. Lee, Bioresource Technology, 2010, 101, 8185-8191.
315. M. Pedersen, A. Viksø-Nielsen and A. S. Meyer, Process Biochemistry, 2010, 45, 1181-1186.
316. M. Pedersen and A. S. Meyer, New Biotechnology, 2010, 27, 739-750.
317. R. S. Ramirez, M. Holtzapple and N. Piamonte, in Aqueous Pretreatment of Plant Biomass for Biological
and Chemical Conversion to Fuels and Chemicals, John Wiley & Sons, Ltd, 2013, DOI:
10.1002/9780470975831.ch8, pp. 145-167.
318. H. L. Trajano and C. E. Wyman, in Aqueous Pretreatment of Plant Biomass for Biological and Chemical
Conversion to Fuels and Chemicals, John Wiley & Sons, Ltd, 2013, DOI: 10.1002/9780470975831.ch6,
pp. 103-128.
319. A. V. Bandura and S. N. Lvov, Journal of Physical and Chemical Reference Data, 2006, 35, 15-30.
303
320. B. S. Dien, H.-J. G. Jung, K. P. Vogel, M. D. Casler, J. F. Lamb, L. Iten, R. B. Mitchell and G. Sarath,
Biomass and Bioenergy, 2006, 30, 880-891.
321. D. Guay, B. Cole, R. Fort Jr, J. Genco and M. Hausman, Journal of Wood Chemistry and Technology, 2000,
20, 375-394.
322. J. Gierer, Wood Sci. Technol., 1985, 19, 289-312.
323. E. M. Karp, B. S. Donohoe, M. H. O’Brien, P. N. Ciesielski, A. Mittal, M. J. Biddy and G. T. Beckham,
ACS Sustainable Chemistry & Engineering, 2014, 2, 1481-1491.
324. E. A. Capanema, M. Y. Balakshin and J. F. Kadla, Journal of agricultural and food chemistry, 2004, 52,
1850-1860.
325. D. D. Laskar, M. P. Tucker, X. Chen, G. L. Helms and B. Yang, Green Chemistry, 2014, 16, 897-910.
326. Y.-m. Zhang, Y. Peng, X.-l. Yin, Z.-h. Liu and G. Li, Journal of Chemical Technology & Biotechnology,
2014, 89, 1954-1960.
327. D. A. Baker and T. G. Rials, Journal of Applied Polymer Science, 2013, 130, 713-728.
328. J.-W. Jeon, L. Zhang, J. L. Lutkenhaus, D. D. Laskar, J. P. Lemmon, D. Choi, M. I. Nandasiri, A. Hashmi,
J. Xu, R. K. Motkuri, C. A. Fernandez, J. Liu, M. P. Tucker, P. B. McGrail, B. Yang and S. K. Nune,
ChemSusChem, 2015, 8, 428-432.
329. H. Wang, H. Ruan, H. Pei, H. Wang, X. Chen, M. p. Tucker, J. R. Cort and B. Yang, Green Chemistry,
2015, DOI: 10.1039/C5GC01534K.
330. Application: WO2014163652 A1 Pat., 2013-US38927
2014163652, 2014.
331. J. Ralph, K. Lundquist, G. Brunow, F. Lu, H. Kim, P. F. Schatz, J. M. Marita, R. D. Hatfield, S. A. Ralph
and J. H. Christensen, Phytochemistry Reviews, 2004, 3, 29-60.
332. S. D. Mansfield, C. Mooney and J. N. Saddler, Biotechnology progress, 1999, 15, 804-816.
333. S. I. Aronovsky and R. A. Gortner, Industrial & Engineering Chemistry, 1930, 22, 264-274.
334. W. S. L. Mok and M. J. Antal Jr, Industrial & Engineering Chemistry Research, 1992, 31, 1157-1161.
335. D. Robert, M. Bardet, C. Lapierre and G. Gellerstedt, Cellulose chemistry and technology, 1988.
336. R. Samuel, M. Foston, N. Jiang, L. Allison and A. J. Ragauskas, Polymer Degradation and Stability, 2011,
96, 2002-2009.
337. D. D. Laskar, B. Yang, H. Wang and J. Lee, Biofuels, Bioproducts and Biorefining, 2013, 7, 602-626.
338. S. Cao, Y. Pu, M. Studer, C. Wyman and A. J. Ragauskas, Rsc Advances, 2012, 2, 10925-10936.
339. M. H. Studer, J. D. DeMartini, M. F. Davis, R. W. Sykes, B. Davison, M. Keller, G. A. Tuskan and C. E.
Wyman, Proceedings of the National Academy of Sciences, 2011, 108, 6300-6305.
304
340. Q. Sun, M. Foston, X. Meng, D. Sawada, S. V. Pingali, H. M. O’Neill, H. Li, C. E. Wyman, P. Langan and
A. J. Ragauskas, Biotechnology for biofuels, 2014, 7, 150.
341. C. Fu, J. R. Mielenz, X. Xiao, Y. Ge, C. Y. Hamilton, M. Rodriguez, F. Chen, M. Foston, A. Ragauskas and
J. Bouton, Proceedings of the National Academy of Sciences, 2011, 108, 3803-3808.
342. H. Hisano, R. Nandakumar and Z.-Y. Wang, In Vitro Cellular & Developmental Biology-Plant, 2009, 45,
306-313.
343. A. Ragauskas, Y. Pu, R. Samuel, N. Jiang, C. Fu and Z.-Y. Wang, Frontiers in Energy Research, 2014, 1,
14.
344. H. Shen, M. Mazarei, H. Hisano, L. Escamilla-Trevino, C. Fu, Y. Pu, M. R. Rudis, Y. Tang, X. Xiao and L.
Jackson, The Plant Cell Online, 2013, 25, 4342-4361.
345. L. Yan, L. Zhang and B. Yang, Biotechnology for biofuels, 2014, 7, 76.
346. O. Bobleter, G. Bonn and R. Concin, 1980.
347. C. Liu and C. E. Wyman, Bioresource Technology, 2005, 96, 1978-1985.
348. F. Hu and A. Ragauskas, Rsc Advances, 2014, 4, 4317-4323.
349. H. Shen, C. R. Poovaiah, A. Ziebell, T. J. Tschaplinski, S. Pattathil, E. Gjersing, N. L. Engle, R. Katahira,
Y. Pu, R. Sykes, F. Chen, A. J. Ragauskas, J. R. Mielenz, M. G. Hahn, M. Davis, C. N. Stewart, Jr. and R.
A. Dixon, Biotechnology for Biofuels, 2013, 6.
350. A. J. Ragauskas, G. T. Beckham, M. J. Biddy, R. Chandra, F. Chen, M. F. Davis, B. H. Davison, R. A.
Dixon, P. Gilna, M. Keller, P. Langan, A. K. Naskar, J. N. Saddler, T. J. Tschaplinski, G. A. Tuskan and C.
E. Wyman, Science, 2014, 344, 709-+.
351. S. G. Allen, D. Schulman, J. Lichwa, M. J. Antal, E. Jennings and R. Elander, Industrial & Engineering
Chemistry Research, 2001, 40, 2352-2361.
352. D. D. Laskar, J. Zeng, L. Yan, S. Chen and B. Yang, Industrial Crops and Products, 2013, 50, 391-399.
353. A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, D. Templeton and D. Crocker.
354. R. Overend, E. Chornet and J. Gascoigne, Philosophical Transactions of the Royal Society of London.
Series A, Mathematical and Physical Sciences, 1987, 321, 523-536.
355. K. Iiyama and A. F. A. Wallis, Journal of the Science of Food and Agriculture, 1990, 51, 145-161.
356. H. L. Trajano, N. L. Engle, M. Foston, A. J. Ragauskas, T. J. Tschaplinski and C. E. Wyman, Biotechnology
for Biofuels, 2013, 6, 110.
357. D. D. Laskar, J. Zeng, L. Yan, S. Chen and B. Yang, Industrial Crops and Products, 2013, 50, 391-399.
358. H. L. Trajano, N. L. Engle, M. Foston, A. J. Ragauskas, T. J. Tschaplinski and C. E. Wyman, Biotechnology
for Biofuels, 2013, 6.
359. K. K. Pandey and A. J. Pitman, International Biodeterioration & Biodegradation, 2003, 52, 151-160.
305
360. M. Åkerholm, B. Hinterstoisser and L. Salmén, Carbohydrate Research, 2004, 339, 569-578.
361. O. Faix and J. H. Böttcher, Holz als Roh- und Werkstoff, 1992, 50, 221-226.
362. A. T. Martı́nez, G. Almendros, F. J. González-Vila and R. Fründ, Solid State Nuclear Magnetic Resonance,
1999, 15, 41-48.
363. A. Gutierrez, P. Bocchini, G. C. Galletti and A. T. Martinez, Applied and environmental microbiology,
1996, 62, 1928-1934.
364. D. Ibarra, J. C. del Río, A. Gutiérrez, I. M. Rodríguez, J. Romero, M. J. Martínez and Á. T. Martínez,
Journal of Analytical and Applied Pyrolysis, 2005, 74, 116-122.
365. S. D. Mansfield, H. Kim, F. Lu and J. Ralph, Nature protocols, 2012, 7, 1579-1589.
366. S. D. Mansfield, H. Kim, F. Lu and J. Ralph, Nat. Protocols, 2012, 7, 1579-1589.
367. J. H. Lora and M. Wayman, Canadian Journal of Chemistry, 1980, 58, 669-676.
368. T. Q. Yuan, S. N. Sun, F. Xu and R. C. Sun, J Agric Food Chem, 2011, 59, 10604-10614.
369. J. C. Del Río, J. Rencoret, P. Prinsen, A. n. T. Martínez, J. Ralph and A. Gutiérrez, Journal of agricultural
and food chemistry, 2012, 60, 5922-5935.
370. J.-L. Wen, S.-L. Sun, B.-L. Xue and R.-C. Sun, Materials, 2013, 6, 359-391.
371. J. Rencoret, G. Marques, A. Gutierrez, D. Ibarra, J. Li, G. Gellerstedt, J. I. Santos, J. Jimenez-Barbero, A.
T. Martinez and J. C. del Rio, Holzforschung, 2008, 62, 514-526.
372. A. Guerra, L. A. Lucia and D. S. Argyropoulos, Holzforschung, 2008, 62, 24-30.
373. A. Guerra, I. Filpponen, L. A. Lucia and D. S. Argyropoulos, Journal of Agricultural and Food Chemistry,
2006, 54, 9696-9705.
374. C. Olsen, V. Arantes and J. Saddler, Bioresource technology, 2015, 187, 288-298.
375. W.-S. Lim and J.-W. Lee, Bioresource technology, 2013, 140, 306-311.
376. M. Paananen and H. Sixta, Bioresource Technology, 2015, 193, 97-102.
377. S. Agnihotri, I. A. Johnsen, M. S. Bøe, K. Øyaas and S. Moe, Wood Sci. Technol., 2015, 1-16.
378. C. Asada, C. Sasaki, Y. Uto, J. Sakafuji and Y. Nakamura, Biochemical engineering journal, 2012, 60, 25-
29.
379. W. Zhu, J. Zhu, R. Gleisner and X. Pan, Bioresource Technology, 2010, 101, 2782-2792.
380. Q. Nguyen, M. Tucker, B. Boynton, F. Keller and D. Schell, in Biotechnology for Fuels and Chemicals,
Springer, 1998, pp. 77-87.
381. A. Fissore, L. Carrasco, P. Reyes, J. Rodríguez, J. Freer and R. T. Mendonça, Journal of industrial
microbiology & biotechnology, 2010, 37, 893-900.
306
382. X. Pan, J. F. Kadla, K. Ehara, N. Gilkes and J. N. Saddler, Journal of Agricultural and Food Chemistry,
2006, 54, 5806-5813.
383. X. Pan, D. Xie, R. W. Yu, D. Lam and J. N. Saddler, Industrial & Engineering Chemistry Research, 2007,
46, 2609-2617.
384. B. Yang, A. Boussaid, S. D. Mansfield, D. J. Gregg and J. N. Saddler, Biotechnology and Bioengineering,
2002, 77, 678-684.
385. M. Normark, S. Winestrand, T. A. Lestander and L. J. Jönsson, BMC biotechnology, 2014, 14, 1.
386. X. Li, X. Luo, K. Li, J. Zhu, J. D. Fougere and K. Clarke, Applied biochemistry and biotechnology, 2012,
168, 1556-1567.
387. Q. A. Nguyen, M. P. Tucker, F. A. Keller, D. A. Beaty, K. M. Connors and F. P. Eddy, Applied biochemistry
and biotechnology, 1999, 77, 133-142.
388. R. P. Chandra, Q. Chu, J. Hu, N. Zhong, M. Lin, J.-S. Lee and J. Saddler, Bioresource technology, 2016,
199, 135-141.
389. R. T. Elander, B. E. Dale, M. Holtzapple, M. R. Ladisch, Y. Y. Lee, C. Mitchinson, J. N. Saddler and C. E.
Wyman, Cellulose, 2009, 16, 649-659.
390. F. Hu, S. Jung and A. Ragauskas, Bioresource technology, 2012, 117, 7-12.
391. B. Yang and E. Wyman Charles, Biotechnol Bioeng, 2004, 86, 88-95.
392. L. Yan, Y. Pu, M. Bowden, A. J. Ragauskas and B. Yang, Acs Sustain Chem Eng, 2015, DOI:
10.1021/acssuschemeng.5b01021.
393. M. Foston, H. L. Trajano, R. Samuel, C. E. Wyman, J. He and A. J. Ragauskas, Bioresource Technology,
2015, 197, 128-136.
394. J. Gierer, The reactions of lignin during pulping. A description and comparison of conventional pulping
processes, DTIC Document, 1970.
395. J. Gierer, Wood Sci. Technol., 1980, 14, 241-266.
396. M. Wayman and J. Lora, Tappi Tech Assoc Pulp Paper Ind, 1978.
397. M. Drumond, M. Aoyama, C.-L. Chen and D. Robert, Journal of wood chemistry and technology, 1989, 9,
421-441.
398. X. Pan, D. Lachenal, V. Neirinck and D. Robert, Journal of wood chemistry and technology, 1994, 14, 483-
506.
399. D. Robert, in Methods in lignin chemistry, Springer, 1992, pp. 250-273.
400. A. Tolbert, H. Akinosho, R. Khunsupat, A. K. Naskar and A. J. Ragauskas, Biofuels, Bioproducts and
Biorefining, 2014, 8, 836-856.
401. C. Tengborg, M. Galbe and G. Zacchi, Enzyme and microbial technology, 2001, 28, 835-844.
307
402. B. Yang, Y. Lu and C. E. Wyman, Biofuels from Agricultural Wastes and Byproducts, 2010, 175-200.
403. L. R. Lynd, M. S. Laser, D. Bransby, B. E. Dale, B. Davison, R. Hamilton, M. Himmel, M. Keller, J. D.
McMillan, J. Sheehan and C. E. Wyman, Nat Biotechnol, 2008, 26, 169-172.
404. L. R. Lynd, C. E. Wyman and T. U. Gerngross, Biotechnol Prog, 1999, 15, 777-793.
405. C. E. Wyman, Trends Biotechnol, 2007, 25, 153-157.
406. S. P. S. Chundawat, G. T. Beckham, M. E. Himmel and B. E. Dale, Annual Review of Chemical and
Biomolecular Engineering, 2010, 2, 1-25.
407. L. da Costa Sousa, S. P. S. Chundawat, V. Balan and B. E. Dale, Current Opinion in Biotechnology, 2009,
20, 339-347.
408. M. Galbe and G. Zacchi, in Biofuels, ed. L. Olsson, Springer Berlin Heidelberg, 2007, vol. 108, ch. 70, pp.
41-65.
409. P. Alvira, E. Tomás-Pejó, M. Ballesteros and M. J. Negro, Bioresource Technology, 2010, 101, 4851-4861.
410. J. Zhu, X. Pan and R. S. Zalesny Jr, Applied microbiology and biotechnology, 2010, 87, 847-857.
411. L. Cadoche and G. D. López, Biological Wastes, 1989, 30, 153-157.
412. H. Lu, E. Ip, J. Scott, P. Foster, M. Vickers and L. L. Baxter, Fuel, 2010, 89, 1156-1168.
413. T. Pattanotai, H. Watanabe and K. Okazaki, Fuel, 2013, 104, 468-475.
414. K. R. Gaston, M. W. Jarvis, P. Pepiot, K. M. Smith, W. J. Frederick and M. R. Nimlos, Energy & Fuels,
2011, 25, 3747-3757.
415. S. P. S. Chundawat, B. Venkatesh and B. E. Dale, Biotechnology and Bioengineering, 2007, 96, 219-231.
416. E. M. K. Joseph Shekiro III*, Nicholas J Nagle, Melvin P Tucker, Richard T Elander and Daniel J Schell,
Biotechnology for biofuels, 2014, 7.
417. K. H. Kim, M. P. Tucker and Q. A. Nguyen, Biotechnology Progress, 2002, 18, 489-494.
418. K. M. F. Kazi, P. Jollez and E. Chornet, Biomass and Bioenergy, 1998, 15, 125-141.
419. S. A. Hosseini and N. Shah, Bioresour Technol, 2009, 100, 2621-2628.
420. P. Kumar, D. M. Barrett, M. J. Delwiche and P. Stroeve, Industrial & Engineering Chemistry Research,
2009, 48, 3713-3729.
421. I. Ballesteros, J. M. Oliva, M. J. Negro, P. Manzanares and M. Ballesteros, Process Biochemistry, 2002, 38,
187-192.
422. A. M. Elshafei, J. L. Vega, K. T. Klasson, E. C. Clausen and J. L. Gaddy, Bioresource Technology, 1991,
35, 73-80.
423. B. Yang and C. E. Wyman, Methods in molecular biology, 2009, 581, 103-114.
308
424. J. Shekiro III, E. Kuhn, N. Nagle, M. Tucker, R. Elander and D. Schell, Biotechnology for biofuels, 2014, 7,
23.
425. M. R. Ehrhardt, T. O. Monz, T. W. Root, R. K. Connelly, C. T. Scott and D. J. Klingenberg, Appl Biochem
Biotechnol, 2010, 160, 1102-1115.
426. N. Szijarto, M. Siika-aho, T. Sontag-Strohm and L. Viikari, Bioresour Technol, 2011, 102, 1968-1974.
427. S. Viamajala, J. D. McMillan, D. J. Schell and R. T. Elander, Bioresour Technol, 2009, 100, 925-934.
428. A. Esteghlalian, A. G. Hashimoto, J. J. Fenske and M. H. Penner, Bioresource Technology, 1997, 59, 129-
136.
429. A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter and D. Templeton.
430. Y. Kim, R. Hendrickson, N. S. Mosier and M. R. Ladisch, Methods in molecular biology, 2009, 581, 93-
102.
431. A. Esteghlalian, A. G. Hashimoto, J. J. Fenske and M. H. Penner, Bioresource Technology, 1997, 59, 129-
136.
432. T. A. Lloyd, Dartmouth College, 2005.
433. W. Mabee, D. Gregg, C. Arato, A. Berlin, R. Bura, N. Gilkes, O. Mirochnik, X. Pan, E. Kendall Pye and J.
Saddler, Applied Biochemistry and Biotechnology, 2006, 129, 55-70.
434. W. C. Brennan MA, Applied Biochemistry and Biotechnology, 2004, 113-116.
435. K. Hammes, R. J. Smernik, J. O. Skjemstad and M. W. I. Schmidt, Applied Geochemistry, 2008, 23, 2113-
2122.
436. S. Brown, Estimating Biomass and Biomass Change of Tropical Forests: a Primer. (FAO Forestry Paper -
134), 1997.
437. R. Kumar, G. Mago, V. Balan and C. E. Wyman, Bioresour Technol, 2009, 100, 3948-3962.
438. Y. Pu, F. Hu, F. Huang, B. H. Davison and A. J. Ragauskas, Biotechnology for biofuels, 2013, 6, 15.
439. J. Li, G. Henriksson and G. Gellerstedt, Bioresour Technol, 2007, 98, 3061-3068.
440. S. J. B. Duff and W. D. Murray, Bioresource Technology, 1996, 55, 1-33.
441. T. A. Lloyd and C. E. Wyman, Bioresour Technol, 2005, 96, 1967-1977.
442. C. E. Wyman, B. E. Dale, V. Balan, R. T. Elander, M. T. Holtzapple, R. S. Ramirez, M. R. Ladisch, N. S.
Mosier, Y. Y. Lee, R. Gupta, S. R. Thomas, B. R. Hames, R. Warner and R. Kumar, in Aqueous
Pretreatment of Plant Biomass for Biological and Chemical Conversion to Fuels and Chemicals, John
Wiley & Sons, Ltd, 2013, DOI: 10.1002/9780470975831.ch12, pp. 239-259.
443. R. Davis, L. Tao, C. Scarlata, E. Tan, J. Ross, J. Lukas and D. Sexton, 2015.
444. L. Laureano-Perez, F. Teymouri, H. Alizadeh and B. Dale, in Twenty-Sixth Symposium on Biotechnology
309
for Fuels and Chemicals, eds. B. Davison, B. Evans, M. Finkelstein and J. McMillan, Humana Press, 2005,
DOI: 10.1007/978-1-59259-991-2_91, ch. 91, pp. 1081-1099.
445. H. Wang, L. Zhang, T. Deng, H. Ruan, X. Hou, J. R. Cort and B. Yang, Green Chemistry, 2016, DOI:
10.1039/C5GC02967H.
446. J. F. Saeman, Industrial & Engineering Chemistry, 1945, 37, 43-52.
447. J. R. Gray, in Environmental Instrumentation and Analysis Handbook, John Wiley & Sons, Inc., 2005, DOI:
10.1002/0471473332.ch23, pp. 491-510.