Reaction Pathway Analysis for the Acid
Catalysed Transformation of Hexose
Carbohydrates to Advantaged Biofuel
Components
Thesis presented for the award of Doctor of Philosophy (PhD)
Thomas Flannelly
University of Limerick
Supervisors
Prof. J.J Leahy and Dr Stephen Dooley
Submitted to the University of Limerick
March-2016
Acknowledgements
Firstly, I would like to thank my supervisor, Prof. J.J. Leahy, for his unwavering
support, guidance and constant encouragement for the duration of my time at the
University of Limerick. I will be forever grateful for the opportunities you have
afforded me.
Secondly, I would like to thank Dr Stephen Dooley for his honest feedback,
guidance, and his commitment to ensuring all work was done to the highest standard.
Thank you for broadening my horizons and for all the conversations about sport that
kept me sane during my most difficult moments.
I would also like to thank Prof Michael Hayes for his constant encouragement and
for the great interest that he took in my work. I would like to thank all my friends
that I made whilst working in the lab here at the University of Limerick. I would
like to particularly thank Francesco and Mícheál in this regard.
I owe a lot to my late Uncle Noel Nicholson, without Noel I would have never
attended the University of Limerick. Thanks Noel for always being there for me, for
all the dinners, and for the comfort of knowing there was someone I could call if I
ever got into difficulties whilst in Limerick.
Finally to my parents Joe and Lucy for their unconditional love and support that they
gave me since the day I was born, 25 years ago. Throughout my PhD I always
looked forward to going home to Knocknakillew at the weekend, and it always
provided me shelter from the pressures of my studies.
Declaration
The work presented in this Thesis is the original work of the author, conducted under
the supervision of Prof. J.J. Leahy and Dr Stephen Dooley, and due reference has
been made, where necessary, to the work of others. No part of this thesis has been
previously submitted to this or any other University.
Thomas Flannelly
Date
Abstract
The valorisation of readily available and abundant lignocellulosic derived hexose
carbohydrates by reaction with ethanol can provide an alternative to petroleum
derived fossil fuels. The main barrier in effectively doing so is the lack of
mechanistic knowledge regarding the reaction pathways linking the hexose sugar to
the intended fuel component.
In pursuit of the chemical mechanisms responsible for the hydrolysis of; D-fructose,
D-galactose, D-glucose, D-mannose to levulinic acid in aqueous systems using 2.5
wt% H2SO4 at 423 K is deciphered upon. The mechanistic comprehension gained is
used as constraints to evaluate the more complex
ethanolysis
(ethanol/H2SO4/hexose) system. It is also comprehensively shown that formic and
levulinic acids are not formed stoichiometrically from lignocellulosic derived
hexoses, as is widely believed in the literature. At steady-state conversions of the
reactant, the formic and levulinic acid ratio for D-fructose, D-glucose, D-mannose
and D-galactose is shown to be 1.08 ±0.04, 1.15 ±0.05, 1.20 ±0.10 and 1.19 ±0.04
respectively.
Next, the ethanolysis process is introduced as a superior alternative to the aqueous
hydrolysis systems. Two advantaged fuel components 5-ethoxymethylfurfural and
ethyl levulinate are identified and numerical modelling is utilised to test the
feasibility of carefully designed mechanistic propositions at 351 K catalysed by
hydrogen cations. It is shown that the hydrogen cation is consumed by reaction with
ethanol and that the overall system is ‘‘pseudo’’ catalytic. Condensed phase
conditions (331-351 K) are deemed more suitable for the formation of 5ethoxymethylfurfural, whilst biphasic conditions (>353 K) favours the formation of
ethyl levulinate. A kinetic model is developed for the ethanolysis of D-fructose to 5ethoxymethylfurfural (331-351 K), which includes kinetic contraints derived from
conducting reactions with all key chemical intermediates in the system.
The reaction mechanism for the ethanolysis of D-glucose to ethyl levulinate (351423 K) catalysed by H2SO4 (0.015-0.075) mol/L is deciphered upon, with no
significant amounts of ethyl levulinate formed below 393 K. Significantly the main
reaction flux for ethyl levulinate formation from D-glucose does not advance through
any furan intermediates as it does in the D-fructose ethanolysis mechanism, which is
not the proposed pathway in the literature. Finally, the fuel properties of the range of
fuel molecules produced in ethanolysis systems are presented. It is shown that by
integrating mechanistic understanding, kinetic parameters and fuel properties of the
synthesized molecules, drop-in tailor-made fuel additives can be synthesized in a
highly flexible “one-pot” process designed for specific purposes be it for ‘‘diesel’’
or ‘‘gasoline’’ fuels.
Table of contents
List of Achievements .................................................................................................. I
Abbreviations .......................................................................................................... III
List of Figures ............................................................................................................ V
List of Tables .......................................................................................................... XV
Chapter 1: Introduction ............................................................................................ 1
1.1.
Global dependence on fossil fuel derived energy ............................................ 2
1.2.
Energy consumption in Ireland ........................................................................ 3
1.3.
Fossil fuels and CO2......................................................................................... 4
1.4.
Transportation fuels ......................................................................................... 6
1.5.
Biomass as an energy source of liquid fuels .................................................... 7
1.6.
From 1st to 2nd generation biofuels................................................................... 8
1.7.
The conversion of carbohydrates to liquid transportation fuel ........................ 9
1.8.
Acid hydrolysis .............................................................................................. 10
1.9.
Ethanol ........................................................................................................... 11
1.10. Synthesising liquid transportation fuel components by reacting hexose
carbohydrates with ethanol ........................................................................................ 14
1.11. Scope of the research ..................................................................................... 16
1.12. Outline of the Thesis ...................................................................................... 17
1.13. References ...................................................................................................... 20
Chapter 2: Literature Review ................................................................................. 26
2.1.
Lignocellulosic derived hexose sugars: D-fructose and D-glucose ............... 27
2.2.
D-Fructose/D-glucose isomerisation .............................................................. 29
2.3.
Mechanisms and kinetics for the dehydration of D-fructose in aqueous
systems ....................................................................................................................... 31
2.4.
Mechanism for the dehydration of D-glucose in aqueous systems ................ 32
2.5.
Mechanisms and kinetics of hexose carbohydrates in non-aqueous solvents 34
2.6.
Mechanism and kinetics of hexose dehydration in alcohols .......................... 36
2.7.
Ethyl levulinate is an advantaged oxygenated hydrocarbon .......................... 37
2.8.
Production pathways for the synthesis of ethyl levulinate ............................. 41
2.8.1.
The esterification of levulinic acid ....................................................... 41
2.8.2.
The alcoholysis of furfuryl alcohol to ethyl levulinate ........................ 44
2.8.3.
The formation of alkyl levulinates from hexose carbohydrates in a one-
pot synthesis........................................................................................................ 46
2.9.
5-Ethoxymethylfurfural as an advantaged fuel component ........................... 51
2.9.1.
Production pathways for 5-ethoxymethylfurfural ................................ 52
2.9.2.
The ethanolysis of 5-hydroxymethylfurfural ....................................... 53
2.9.3.
The one-pot synthesis of 5-ethoxymethylfurfural from hexose
carbohydrates ...................................................................................................... 54
2.10.
Novelty of this Thesis .................................................................................... 57
2.11.
References ...................................................................................................... 58
Chapter 3: Materials and Methods ........................................................................ 69
3.1.
Experimental procedure ................................................................................. 71
3.1.1.
Experimental configuration for experiments conducted under reflux
conditions ........................................................................................................... 71
3.1.2.
3.2.
3.3.
Experimental configuration for high pressure experiments ................. 72
Detection of analytes using ion chromatography........................................... 72
3.2.1.
Electrochemical detector ...................................................................... 75
3.2.2.
Diode array detector ............................................................................. 76
3.2.3.
Detection of carbohydrates and carbohydrate derived derivatives ...... 76
3.2.4.
Detection of furanic compounds and organic acids ............................. 79
Detecting analytes using gas chromatography ............................................... 79
3.3.1.
Sample preparation .............................................................................. 80
3.3.2.
Sample injection................................................................................... 81
3.3.3.
Column employed ................................................................................ 81
3.3.4.
Flame ionisation detector (FID) ........................................................... 83
3.3.5.
Quantification of sample species ......................................................... 84
3.3.6.
Gas chromatography-mass spectrometry ............................................. 84
3.4.
Estimation of hydrogen cations in solution ................................................... 87
3.5.
Quantification of black insoluble materials (humins) .................................... 89
3.6.
Computational methods employed ................................................................ 89
3.6.1.
Rates, rate constants and reaction kinetics ........................................... 89
3.7.
3.6.2.
Reaction mechanism ............................................................................ 90
3.6.3.
Methods for estimating rate constants ................................................. 91
3.6.4.
Evaluation of kinetic parameters and kinetic model operation ............ 93
References ...................................................................................................... 95
Chapter 4: Non-Stoichiometric Formation of Formic and Levulinic Acids from
the Hydrolysis of Biomass Derived Hexose Carbohydrates ................................. 97
4.1.
Abstract .......................................................................................................... 98
4.2.
Introduction .................................................................................................... 99
4.3.
Experiment ................................................................................................... 105
4.4.
4.5.
4.3.1.
Materials ............................................................................................. 105
4.3.2.
Experiment configuration ............................................................... 105
4.3.3.
Calculations and uncertainty analysis ................................................ 108
Results and discussion ................................................................................. 110
4.4.1.
Hexose conversion ............................................................................. 110
4.4.2.
Levulinic acid yields .......................................................................... 110
4.4.3.
Ratio of formic to levulinic acid ........................................................ 112
4.4.4.
Pathways to formic acid formation .................................................... 114
4.4.5.
Excess formic acid from direct hexose transformations .................... 115
4.4.6.
Excess formic acid from hexose derived intermediates. .................... 119
4.4.7.
Outlook............................................................................................... 121
Conclusion ................................................................................................... 123
4.6.
References .................................................................................................... 123
Chapter 5: Reaction Pathway Analysis of Ethyl Levulinate and 5Ethoxymethylfurfural from D-Fructose Acid Hydrolysis in Ethanol ............... 129
5.1.
Abstract ........................................................................................................ 130
5.2.
Introduction .................................................................................................. 131
5.3.
Experimental configuration ......................................................................... 135
5.4.
5.3.1.
Materials ............................................................................................ 135
5.3.2.
Experimental ...................................................................................... 135
5.3.3.
Analytical methods ............................................................................ 137
5.3.4.
Measurement uncertainties ................................................................ 139
5.3.5.
Identification of unknown species ..................................................... 142
5.3.6.
Quantification of unidentified species ............................................... 146
Results and discussion ................................................................................. 147
5.4.1.
Experimental observations and reaction mechanism ......................... 147
5.4.2.
Trace species, not considered in mechanistic analysis ...................... 147
5.4.3.
Major species, considered in mechanistic analysis ............................ 148
5.4.4.
Hydrogen cation concentrations ........................................................ 158
5.4.5.
Kinetic model ..................................................................................... 161
5.4.6.
Computational methods ..................................................................... 164
5.4.7.
Chemical reaction mechanism and kinetics discussion ..................... 167
5.4.8.
Yield analysis ..................................................................................... 174
5.4.9.
Volumetric energy density ................................................................. 174
5.5.
Conclusions .................................................................................................. 175
5.6.
References .................................................................................................... 176
Chapter 6: A Low Temperature Kinetic Model for the Ethanolysis of DFructose to 5-Ethoxymethylfurfural in a One-Pot-Synthesis ............................ 184
6.1.
Abstract ........................................................................................................ 185
6.2.
Introduction .................................................................................................. 186
6.3.
Experimental .............................................................................................. 191
6.4.
6.3.1.
Experimental configuration ................................................................ 191
6.3.2.
Materials and methods ....................................................................... 192
6.3.3.
Analytical procedure ....................................................................... 193
6.3.4.
Incorporating the effect of H2SO4 on reaction rates .................... 194
6.3.5.
Estimation of kinetic parameters.................................................... 195
6.3.6.
Kinetic modeling .............................................................................. 196
Results and discussion ................................................................................. 199
6.4.1.
Experimental observations and kinetic modelling ............................. 199
6.4.1.1.
5-Ethoxymethylfurfural experimental observations ................... 199
6.4.1.2.
5-Ethoxymethylfurfural sub-model ............................................ 201
6.4.1.3.
5-Hydroxymethylfurfural experimental observations ................ 202
6.4.1.4.
5-Hydroxymethylfurfural sub-model. ......................................... 203
6.4.1.5.
D-Fructose experimental observations ........................................ 206
6.4.1.6.
6.4.2.
D-Fructose sub-model ................................................................. 206
Implications of the kinetic model ...................................................... 210
6.5.
Conclusions .................................................................................................. 218
6.6.
References .................................................................................................... 218
Chapter 7: Synthesis of “Diesel” or “Gasoline” Oxygenated Fuel Components
from Hexose Carbohydrates by Reaction with Ethanol..................................... 224
7.1.
Abstract ........................................................................................................ 225
7.2.
Introduction .................................................................................................. 226
7.3.
Experimental ................................................................................................ 230
7.4.
7.3.1.
Materials ............................................................................................ 230
7.3.2.
Experimental configuration ............................................................... 231
7.3.3.
Analytical methods ............................................................................ 231
7.3.4.
Kinetic modelling............................................................................... 233
Results and discussion ................................................................................. 234
7.4.1.
Reaction mechanism for Bronsted acid catalysed ethanolysis of D-
glucose………………………………………………………………………...234
7.4.2.
Conversion of ethanol to diethyl ether ............................................... 240
7.4.3.
Manipulating the reaction mechanism to produce fuels of tailor-able
properties, be it ‘‘diesel’’ or ‘‘gasoline’’.......................................................... 243
7.4.4.
7.5.
Outlook, limitations and future work ................................................. 247
Conclusion ................................................................................................... 248
Chapter 8: Conclusions ......................................................................................... 254
8.1.
Conclusions .................................................................................................. 255
8.2.
Recommendations for future work............................................................... 259
8.2.1.
Utilisation of more sophisticated analytical techniques ..................... 259
8.2.2.
Incorporating the effect of water concentration and reactant
concentration as parameters in the kinetic model. ............................................ 259
8.2.3.
Employing computational chemistry to determine the validity
mechanistic propositions .................................................................................. 260
8.2.4.
Formulating kinetic models incorporating glucose to fructose
isomerisation ..................................................................................................... 261
8.2.5.
8.3.
Conducting fuel property analysis on ethanolysis fuel products ....... 261
References .................................................................................................... 261
List of Achievements
Peer reviewed publications
Flannelly T., Dooley S., Leahy J.J, (2015) “Reaction pathway analysis of ethyl
levulinate and 5-ethoxymethylfurfural from D-fructose acid hydrolysis in ethanol,”
Energy & Fuels, 29 (11), pp 7554–7565
Flannelly T., Lopes M., Kupiainen L., Dooley S., Leahy J.J., (2015) “Nonstoichiometric formation of formic and levulinic acids from the hydrolysis of
biomass derived hexose carbohydrates,” RSC Advances, 6, 5797-5804
Severini F., Flannelly T., Nolan D.O., Leahy J.J., Kwapinski W. (2015)
“Development of heterogeneous acid catalysts produced from the carbonization of
Miscanthus x giganteus for the esterification of butyric acid to butyl butyrate with nbutanol,” Journal of Chemical Technology and Biotechnology.
Conference presentations
Flannelly T., Dooley S., Leahy J.J. (2015) ‘‘Mechanism and kinetics of advantaged
synthesis from D-fructose’’ Paper No 20, ACS Spring National Meeting, Denver
Colorado, March 22-26, 2015.
Flannelly T., Howard M., Dooley S., Leahy J.J. (2015) ‘‘Synthesis of “Diesel” or
“Gasoline” Oxygenated Fuel Components from Hexose Carbohydrates by Reaction
with Ethanol ’’ Cost Action CM1404 ‘‘Smart Energy Carriers as Liquid
Transportation Fuels’’ 27th August 2015, Thessaloniki, Greece.
I
Flannelly T., Howard M., Dooley S., Leahy J.J. (2015) ‘‘Mechanism and Kinetics of
Advantaged Biofuels Synthesis from D-Fructose’’ 67th Irish University Chemistry
Research Colloquium, 25th June 2015, NUI Maynooth.
Selected poster presentations
Flannelly T., Howard M., Dooley S., Leahy J.J. (2015) ‘‘Synthesis of “Diesel” or
“Gasoline” Oxygenated Fuel Components from Hexose Carbohydrates by Reaction
with Ethanol’’ 22nd of June 2015, Aachen, Germany.
II
Abbreviations
5-EMF
5-Ethoxymethylfurfural
5-HMF
5-Hydroxymethylfufural
ACS
American Chemical Society
AFF
2,6-Anhydro-β-D-fructofuranose
AGF
1,6-Anhydro-β-D-glucofuranose
AGP
1,6-Anhydro-β-D-glucopyranose
AL
Alkyl Levulinate
AS
Auto Sampler
B3LYP
Becke 3 Lee Yang Parr
DAD
Dioxide Array Detector
DFT
Density Functual Theory
DOE
Department of Energy
eNRTL
electrolyte Non-Random Two-Liquid
Ea
Activation Energy
EL
Ethyl Levulinate
EPA
Environmental Protection Agency
EU
European Union
FA
Formic Acid
FAL
Furfuryl Alcohol
FETFE
Fluoroelastomer with Tetrefluorolethylene Additives
FID
Flame Ionisation Detector
G4MP2
Fourth Generation Second order Moller-Plesset
Perturbation
GC
Gas Chromatography
GC-MS
Gas Chromatography Mass Spectrometry
III
GVL
γ-Valerolactone
HPAs
Heteropoly Acids
IC
Ion Chromatography
IEA
International Energy Agency
Ka
Dissociation Constant
LA
Levulinic Acid
LGN
Levoglucosan
NIST
National Institute of Standards and Technology
NMR
Nuclear Magnetic Resonace
NREL
National Research Energy Laboratory
ODE
Ordinary Differintial Equation
PES
Potiential Energy Surfaces
Ppmv
Parts per Million by Volume
PTFE
Polytetraflouroethylene
RSC
Royal Society of Chemistry
SA
Sulphuric Acid
SEAI
Sustainable Energy Authority of Ireland
ZSM-5
Zeolite Socony Mobil-5
IV
List of Figures
Figure 1.1 Total world primary energy supply 2013 (a), data extracted from (IEA
2014) and projected future fossil fuel extraction (b) data taken from (Watts 2011).... 2
Figure 1.2 Projected global CO2 emissions from the United States EPA based on
four projects scenarios, (US EPA, 2015). .................................................................... 5
Figure 1.3 Ireland’s total energy requirements by sector (a) Primary energy
consumption CO2 emissions by sector (b) Total CO2 emissions by sector (SEAI
2014a)........................................................................................................................... 6
Figure 1.4 Suggested routes for the conversion of biomass to liquid fuels taken from
(Serrano-Ruiz and Dumesic 2011). ............................................................................ 10
Figure 1.5 Predicted Global annual ethanol production by country taken from ....... 13
Figure 1.6 Valorisation of hexose carbohydrates by reaction with ethanol producing
energy dense oxygenated hydrocarbons..................................................................... 15
Figure 1.7 A hierarchical scheme for robust mechanistic and kinetic reaction
modelling. .................................................................................................................. 17
Figure 2.1 Tautomeric distributions of D-fructose/D-glucose reactant in solution at
383 K in aqeous systems with no catalyst as derived by (Kimura et al 2011). .......... 27
Figure 2.2 Predicted Gibbs free energy (298 K) required to protonate the various
oxygen sites of glucopyranose, fructopyranose, and fructofuranose as adopted
calculated by (Assary et al 2012) using G4MP2 theory. .......................................... 28
Figure 2.3 Proposed pathway from cellulose to D-glucose to D-fructose to 5hydroxymethylfurfural. .............................................................................................. 31
V
Figure 2.4 Two-step D-fructose dehydration scheme as suggested by (Swift et al
2013). ......................................................................................................................... 32
Figure 2.5 Reaction path analysis of major elementary processes in Bronsted acidcatalysed D-glucose dehydration from (Yang et al 2015). ......................................... 33
Figure 2.6 Structures of 2,6-anhydro-β-D-fructofuranose (AFF), 1,6-anhydro-β-Dglucofuranose (AGF), 1,6-anhydro-β-D-glucopyranose (AGP), and levoglucosenone
(LGN), as adopted from (Qi et al 2014). ................................................................... 35
Figure 2.7 Possible reaction pathways for the dehydration of D-fructose in alcohol
media adapted from (Tucker et al 2013). ................................................................... 37
Figure 2.8 Ethyl levulinate an advantaged oxygenated hydrocarbon. ...................... 38
Figure 2.9 ∆H
Combustion
and research octane number of ethyl levulinate and 5-
ethoxymethylfurfural compared to ethanol and conventional gasoline. .................... 39
Figure 2.10 A suggested balanced reaction equation for the esterification of levulinic
acid to ethyl levulinate. .............................................................................................. 42
Figure 2.11 Suggested reaction pathway for the alcoholysis of furfuryl alcohol to
ethyl levulinate. .......................................................................................................... 44
Figure 2.12 Simplified mechanism for the one-pot synthesis of alkyl levulinates in
an ethanol/acidic media.............................................................................................. 47
Figure 2.13 5-Ethoxymethylfurfural a promising liquid fuel transport component. 52
Figure 2.14 A suggested balanced reaction for the ethanolysis of 5hydroxymethlfurfural. ................................................................................................ 53
Figure 2.15 General mechanism for the one-pot synthesis of 5-ethoxymethylfurfural
from hexose carbohydrates. ....................................................................................... 55
VI
Figure 3.1 Reflux apparatus for condensed phase reaction synthesis (Chapter 5). ... 71
Figure 3.2 Apparatus for high pressure synthesis reactions. ..................................... 72
Figure 3.3 The ICS-3000 Dionex system at the University of Limerick. ................. 73
Figure 3.4 Schematic diagram of the chromatographic system for the analysis of
monosaccharides and furanic compounds: 1) Sampling needle, 2) Sample syringe,
3) Vial tray, 4) Flush reservoir, 5) Injection valve, 6) Six-port loading valve, 7)
Guard column, 8) Analytical column, 9) Three way-valve, 10) Electrochemical
detector, 11) Pre-column pump, 12) Post-column pump, diode array detector. ........ 74
Figure 3.5 Amperometric pulse for electrochemical detection of monosaccharides.76
Figure 3.6 Exemplar chromatographs employing the CarboPac PA1 column. a)
Exemplar chromatographs generated for the acid catalysed transformations of Dfructose in an ethanol media (Chapter 5&6). b) Exemplar chromatogram of mixed
carbohydrates in aqueous media (Chapter 4). c) Exemplar chromatogram of the acid
catalysed transformations of D-glucose in ethanol media (Chapter 7). ..................... 78
Figure 3.7 The Agilent 7820 A GC system at the University of Limerick. .............. 80
Figure 3.8 Exemplar chromatogram illustrating species separation that are present in
a typical ethanolysis reaction mixture. ....................................................................... 82
Figure 3.9 Schematic of a flame ionisation detector (FID) as extracted from
(Agilient 2011). .......................................................................................................... 83
Figure 3.10 FID detector response for ethanolysis mixture species of interest. ....... 84
Figure 3.11 The gas chromatography mass spectrometer at the University of
Limerick. .................................................................................................................... 85
Figure 3.12 Features of a mass spectrometer. ........................................................... 85
VII
Figure 3.13 Schematic of a quadrupole mass spectrometer detector that is a feature
of the Agilient 5975C MSD at the University of Limerick extracted from (Gates et al
2016). ......................................................................................................................... 86
Figure 3.14 Mass spectrum of 5-hdroxymethylfurfural diacetal. ............................. 87
Figure 3.15 Flow chart of the method for estimation of rate constants and
mechanism evaluation employed in Chapter 5. ......................................................... 93
Figure 3.16 Example of sensitivity analysis conducted for the estimation of the preexponential factor in this Thesis. ............................................................................... 94
Figure 4.1 The historical understanding of the stoichiometric formation of formic
and
levulinic
acids,
from
cellulosic
and
hemicellulosic
derived
hexose
carbohydrates. .......................................................................................................... 100
Figure 4.2 Detector response for carbohydrates and their derivatives using the
Carbopac™ PA1 column. ........................................................................................ 107
Figure 4.3 Detector responses for compounds of interest using the Dionex Acclaim®
Organic Acid column. .............................................................................................. 107
Figure 4.4 Typical chromatogram from the acid catalysed degradation of hexose
sugars using an Acclaim® Organic Acid column. .................................................... 108
Figure 4.5 Typical chromatogram from the acid catalysed degradation of hexose
sugar using a PA1 Dionex CarboPac PA1 carbohydrate column. ........................... 108
Figure
4.6
Experimental
data
for;
(a)
Reactant
conversions,
(b)
5-
hydroxymethylfurfural (5-HMF) yields, (c) Formic acid yields, (d) Levulinic acid
yields. All experiments use 2.5 wt% H2SO4 in water at 423 K with a reactant loading
of 0.3 mol/L. ............................................................................................................ 111
VIII
Figure 4.7 Stability of formic acid in the presence of 2.5 wt% H2SO4 at 423 K.
Note the variability of conversion measurement is within experimental uncertainty.
.................................................................................................................................. 112
Figure 4.8 Formic/levulinic acid ratios per time considering each model compound
as reactant. ................................................................................................................ 113
Figure 4.9 Possible pathways for formic acid formation that have been reported in
the literature. ............................................................................................................ 118
Figure 4.10 (a) Stability of model compounds at 453 K catalysed by 2.5 wt% H2SO4
b) selectivities of 5-hydroxymethylfurfural and furfuryl alcohol to levulinic acid at
453 K catalysed by 2.5 wt% H2SO4. Note only formic acid is detected from Derythrose. .................................................................................................................. 119
Figure 4.11 Molar yields of dihydroxyacetone (a), the relative abundance of acetic
acid (b) and the relative abundance of furfural (c) detected for all model hexose
carbohydrates. .......................................................................................................... 122
Figure 5.1 Molecular structures of ethyl levulinate and 5-ethoxymethylfurfural. .. 131
Figure 5.2 An exemplar ion chromatogram typical of those obtained for Tests #1-7.
.................................................................................................................................. 140
Figure 5.3 The pulsed ampermetric detector response for carbohydrate standards.
.................................................................................................................................. 143
Figure 5.4 Suggested reaction mechanism for the hydrolysis of D-fructose in
condensed phase ethanol/H2SO4.* Depicts species detected with heterogeneous
catalysis. ................................................................................................................... 148
IX
Figure 5.5 Species fractions for Test #1, D-fructose/H2SO4 (0.29/0.09 mol/L) at 351
K, in 100% ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines). .................................................................................................. 155
Figure 5.6 Species fractions for Test #2, D-fructose/H2SO4 (0.29/0.22 mol/L) at 351
K, in 100% ethanol. Mechanism #4.
Experimental values (symbols), model
calculations (lines). .................................................................................................. 155
Figure 5.7 Species fractions for Test #3, D-fructose/H2SO4 (0.29/0.32 mol/L) at 351
K, in 100% ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines). .................................................................................................. 156
Figure 5.8 Species fractions for Test #4, D-fructose/H2SO4 (0.15/0.11 mol/L) at 351
K, in 100% ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines). .................................................................................................. 156
Figure 5.9 Species fractions for Test #5, D-fructose/H2SO4 (0.43/0.34 mol/L) at 351
K, in 100 % ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines). .................................................................................................. 157
Figure 5.10 Species fractions for Test #6, D-fructose/H2SO4 (0.29/0.22 mol/L) at
351 K, in 88/12 mass % ethanol/water. Mechanism #4. Experimental values
(symbols), model calculations (lines). Note, water concentrations shown are
additional to the 12% water media........................................................................... 157
Figure 5.11 Species fractions for Test #7, D-fructose/H2SO4 (0.29/0.22 mol/L) at
351 K, in a 76/24 ratio by mass % ethanol/water ratio. Mechanism #4. Experimental
values (symbols) model calculations (lines). ........................................................... 158
X
Figure 5.12 Hydrogen cation (H+) behaviour at reaction conditions of Table 5.2
where brackets donate ethanol/water mass ratio. [H2SO4] dissociation equilibrium is
calculated by the relations of Que et al. ................................................................... 160
Figure 5.13 Reaction mechanism derived from experimental observations and
kinetic modelling. k7 and k8 are derived from Mechanism #3 for Tests #6-7 where
water is added to the reaction media. ....................................................................... 163
Figure 6.1 Suggested reaction mechanism for the ethanolysis of D-fructose in
homogenous H2SO4 catalysed conditions. ............................................................... 197
Figure 6.2 The seven mechanisms considered for kinetic modelling. The kinetic
parameters derived for the 5-ethoxymethylfurfural (Mechanism 2) and 5hydroxymethylfurfural (Mechanism 4) are used as constraints to evaluate the
parameters for the D-fructose mechanism. ............................................................... 198
Figure 6.3 Temperature dependent behaviour of 5-ethoxymethylfurfural hydration to
ethyl levulinate. ........................................................................................................ 200
Figure 6.4 Reaction order for the consumption of 5-ethoxymethylfurfural (H2SO4/5ethoxymethylfurfural 0.035/0.29 Mol/L). Note the overall reaction conforms to first
order kinetics. ........................................................................................................... 200
Figure 6.5 A parity plot, illustrating the difference between experimental
measurements and model predictions for the 5-ethoxymethylfurfural sub-model
(Mechanism #2). ...................................................................................................... 202
Figure 6.6 Reaction order for the consumption of 5-hydroxymethylfurfural (5hydroxymethylfurfural/H2SO4/0.035/0.29 Mol/L)................................................... 203
XI
Figure 6.7 A parity plot, illustrating the difference between experimental
measurements and model predictions for the 5-hydroxymethylfurfural sub-model
(Mechanism #4). ...................................................................................................... 205
Figure 6.8 A sample comparison of the main reaction species fractions between
experimental measurements (symbols) and model calculations (lines) catalysed by
H2SO4 (0.035-0.13 mol/L H2SO4) with a starting material of (a b) 5ethoxymethylfurfural (5-EMF) (c d) 5-hydroxymethylfurfural (5-HMF) (e g) Dfructose. Note 5-ethoxymethylfurfural hydration to ethyl levulinate could not be
fitted in the kinetic model (a b). ............................................................................... 209
Figure 6.9 A parity plot, illustrating the difference between experimental
measurements and model predictions, Mechanism 6. ............................................. 212
Figure 6.10 Three Dimensional (a c) and contour plots (b d) depicting the maximum
yields of 5-ethoxymethylfurfural obtainable after 480 minutes, from 5hydroxymethylfurfural (a b) reactants and D-fructose (c d) as reactants with the range
of [H2SO4] employed in this study. Note the optimum temperatures for 5ethoxymethylfurfural production are found to be 353 K and 343 K from D-fructose
and 5-hydroxymethylfurfural as reactants. .............................................................. 215
Figure 6.11 Sensitivity analysis for the derived pre-exponential factors as reported in
Table 6.2. ................................................................................................................. 216
Figure 6.12 Sensitivity analysis for the activation energies as reported in Table 6.2.
.................................................................................................................................. 217
Figure 7.1 A potentially more practical approach for the synthesis of liquid
transportation fuels................................................................................................... 227
XII
Figure 7.2 The difference between the (a) D-glucose and (b) D-fructose Bronsted
acid mechanism at 438 K catalysed by 0.04 mol/L H2SO4 in ethanol. Note no 5hydroxymethylfurfural is produced in the D-glucose ethanolysis system. .............. 235
Figure 7.3 Concentrations as a function per time for the ethanolysis of D-glucose to
ethyl levulinate, as for select conditions of Table 7.1 Note each reaction is presented
with different timescales to ensure the glucose to ethyl levulinate reaction has
reached steady-state. ................................................................................................ 238
Figure 7.4 Detailed mechanism for the Bronsted acid catalysed ethanolysis of Dglucose. Note all exciting fuel components produced are highlighted in green boxes.
1-3 describes the main reaction flux for the formation of ethyl levulinate. Each reach
reaction assumes a first order relationship with [H+]. The detection of ethyl acetate,
ethyl pyruvate and ethyl lactate is not pursued in this preliminary study but have
been reported in other Bronsted acid ethanolysis systems. ...................................... 241
Figure 7.5 A parity plot illustrating the accuracy of the proposed model for the
formation of diethyl ether from ethanol (423-453 K). ............................................. 243
Figure 7.6 Manipulating species mole fractions of ethanolysis products to produce a
fuel mixture to mimic the centane number of conventional diesel (a) and gasoline
(b), as is highlighted by a white line on both figures. Note the white line corresponds
to the centane number of diesel (54) (a) and the research octane number (94) of
gasoline (b). .............................................................................................................. 245
Figure 7.7 A comparison of the cetane numbers and energy densities of the flux of
fuel components produced per time between condensed phase and biphasic reaction
conditions. Note it is assumed that all the water has been removed. Condensed phase
XIII
conditions D-fructose/H2SO4 0.29/0.1 mol/L, biphasic conditions D-fructose/H2SO4
0.115/0.04 mol/L. ..................................................................................................... 246
Figure 7.8 The chemical mechanism and fuel chemistry of a D-fructose/ethanol
synthetic fuel system interpreted as gasoline or diesel fuel additives. .................... 247
XIV
List of Tables
Table 1.1 Irelands total energy requirement by fuel, 1990-2013 (SEAI 2014c). ........ 3
Table 1.2 Classification of biomass feed-stocks for second generation bio-refining
adopted from (Slade et al 2011) ................................................................................... 9
Table 2.1 Details of the pertinent studies, regarding the esterification of levulinic
acid (LA) to alkyl levulinate (AL). ............................................................................ 43
Table 2.2 A summary of pertinent catalysts and methods used for the alcoholysis of
furfuryl alcohol (FAL) to alkyl levulinates (AL). ...................................................... 45
Table 2.3 Various methods and catalysts reported, for the one-pot production of
alkyl levulinates from monosaccharides. ................................................................... 50
Table 2.4 Various catalysts used for the synthesis of 5-ethoxymethylfurfural (5EMF) from 5-hydroxymethylfurfural. ....................................................................... 54
Table 2.5 Literature summary of catalysts and methods for the one-pot synthesis of
5-ethoxymethylfurfural from hexose carbohydrates. ................................................. 57
Table 3.1 Summary of materials and methods employed in the construction of this
Thesis ......................................................................................................................... 70
Table 4.1 Literature overview of formic/levulinic acid ratios reported using 5hydroxymethylfurfural and hexoses as reactant. Note water is the medium unless
stated. ....................................................................................................................... 102
Table 5.1. Enthalpies of combustion for proposed transportation fuel components
.................................................................................................................................. 132
Table 5.2 Experimental variables for 8 hour reflux reactions at 351 K. ................. 137
XV
Table 5.3 Exemplar gas chromatography and ion chromatography data for Test #1
(0.29 mol/L D-fructose, 0.22 mol/L H2SO4, 351 K, in 100% ethanol). This data is
also representative of Tests #2-7. Note abbreviations 5-hydroxymethylfurfural (5HMF) and 5-ethoxymethylfurfural (5-EMF). .......................................................... 141
Table 5.4.Carbohydrate pKa verses retention time at the ion exchange
chromatography conditions of this study. ................................................................ 142
Table 5.5 Mass fractions of species detected but not considered in the mechanistic
analysis (with the exception of unknown #5) as the detected quantities are not
appreciable,( <6 mass % in the worst case). Data reported per time, per test and
presented as a mass % of the initial D-fructose used. .............................................. 149
Table 5.6 Glossary of carbohydrate structures and notes on terminology used in the
text............................................................................................................................ 154
Table 5.7 Rate constants derived from the modelling of Mechanism #1. Rate
Constants (k) identification numbers are as described in Figure 5.13. .................... 170
Table 5.8 Rate constants derived from the modelling of Mechanism #2. Rate
Constants (k) identification numbers are as described in Figure 5.13. .................... 170
Table 5.9 Rate constants derived from the modelling of Mechanism #3. Rate
Constants (k) identification numbers are as described in Figure 5.13. .................... 171
Table 5.10 Reaction rate constants derived from modelling of experimental data
with Mechanism #4. Chemical reactions are assigned to each rate constants (kx) in
Figure 5.13.The yield of 5-ethoxymethylfurfural (5-EMF) and ethyl levulinate (EL)
refers to their combined mass at 480 minutes relative to the mass of stoichiometric
reaction of D-fructose with ethanol. ......................................................................... 172
XVI
Table 6.1 Experimental variables for low temperature ethanolysis at 331-351 K
catalysed
by
H2SO4.
Abbreviations:
5-ethoxymethylfurfural
(5-EMF),
5-
hydroxymethylfurfural (5-HMF). ............................................................................ 192
Table 6.2 Estimated activation energies (Ea), Pre-exponential factors (A) with
corresponding rate constants at 351 K. Error margins correspond to 95% confidence
intervals. See Figure 6.2, Mechanism 7 for reaction numbers.* Calculated at a mean
temperature of 383 K. .............................................................................................. 211
Table 7.1 Table of experimental conditions.* depicts experiments carried on for
longer periods until reactions have proceeded to steady-state. ................................ 233
Table 7.2 Derived kinetic parameters for the derived for the reaction of ethanol... 242
Table 7.3 Fuel properties of all pertinent fuel molecules present produced in the
ethanolysis system in this Thesis, obtained from Sigma Aldrich (Sigma Aldrich
2015). Cetane numbers are obtained from Murphy (Murphy et al 2004) and
calculated research octane numbers are estimated using the correlations of Hass and
Dryer (Hass and Dryer 2013) ................................................................................... 244
XVII
Chapter 1
Chapter 1: Introduction
1. intro
1
Chapter 1
1.1. Global dependence on fossil fuel derived energy
Energy is essential to life and cannot be created or destroyed. Without it, many
billions of people would be left cold and hungry. Fossil derived carbon supplies the
energy to keep the heartbeat of our modern world ticking over. Clearly, we as a
society are over reliant on this source, illustrated by the International Energy Agency
recently stating that 81.2% of the world’s energy supply is reliant on fossil fuels
(IEA 2014) (See Figure 1.1). All economically developed societies are heavily
reliant on fossil fuels, not only as a source of energy but also for the production of a
vast array of petrochemicals and plasticers. Global demand for this vital resource is
projected to be more severe with the passing of each year, with world energy
consumption anticipated to grow by 56% between 2010 and 2040, from 552 KJ to
quadrillion 861 KJ (IEA 2014). This expected increase in demand can be attributed
to rapid human population growth and the industrialisation of the developing world.
Energy use in developing countries, for example, is projected to increase by 90% by
2040, while industrialized nations will see a projected increase of 17%. By 2040,
China's energy demand is expected to be twice that of the U.S (IEA 2014).
pr
Figure 1.1 Total world primary energy supply 2013 (a), data extracted from (IEA
2014) and projected future fossil fuel extraction (b) data taken from (Watts 2011).
2
Chapter 1
Fossil fuels are finite resources with supply expected to peak by the middle of this
century (See Figure 1(b)). This coupled with geopolitical tensions in the Middle-East
and legally binding environmental legislation such as the Kyoto Protocol and
Biofuels Directive mean the securement of a stable energy supply has never been
more vulnerable.
1.2. Energy consumption in Ireland
Ireland’s reliance on fossil fuels is even more alarming; with imported fossil fuels
currently accounting for 90% of Ireland’s total primary supply (SEAI 2014a).
Worringly, from an economic point of view Ireland spends €6 billion per year on the
importing of such fossil fuels (SEAI 2014a).
Total Primary Energy Requirement (ktoe)
Share %
1990
2000
2005
2010
2012
2013
1990
2013
Coal
2085
1815
1886
1241
1493
1324
22
9.9
Peat
1377
803
786
791
802
723
14.5
5.4
Oil
4422
7859
9130
7294
5246
5262
46.6
47
Gas
1446
3059
3477
4692
4023
3872
15.2
29
Renewables
168
235
373
688
853
911
1.8
6.8
Wastes
-
-
-
9
44
58
-
0.4
Total
9,497
13,772
15,652
14,715
13296
13314
Table 1.1 Ireland’s total energy requirement by fuel, 1990-2013 (SEAI 2014c).
3
Chapter 1
This creates risk and haemorrhages large amounts of money from the domestic
economy in times of economic uncertainty (SEAI 2014b). Of the generic fossil
fuels, Table 1.1 illustrates that Ireland is most reliant on oil for its total primary
energy requirements (46.6% in 2013). Encouragingly there is some strong evidence
that Ireland is beginning to make real progress in diversifying its energy supply from
fossil fuel dependence. In 1990, 98% of Ireland’s energy requirements were supplied
by fossil fuels; in 2013 this dependence has dropped to 90% (SEAI 2014a). This can
be attributed to the development of a more diversified energy supply with an
increase in the use of renewables from 1.8% to 6.8% between the two time periods
(SEAI 2014a). Whilst this represents progress, clearly there is scope for further
major energy supply diversification, by further investing in renewable technologies.
1.3. Fossil fuels and CO2
Carbon dioxide (CO2) is the primary greenhouse gas emitted through anthropogenic
activities and is the main greenhouse gas responsible for global warming. In 2013,
CO2 accounted for about 82% of all U.S. anthropogenic greenhouse gas emissions
from human activities (US EPA 2015).
Atmospheric CO2 concentrations have
increased by more than 40% since pre-industrial times, from approximately 280 parts
per million by volume (ppmv) in the 18th century to 398ppmv as of August 2015
(US EPA 2015). The main human activity responsible for CO2 emissions is the
combustion of fossil fuels (coal, natural gas, and oil) for energy and transportation,
although certain industrial processes and land use changes also emit CO2. The
potential impacts of increased CO2 levels are both devastating and far reaching.
4
Chapter 1
Figure 1.2 Projected global CO2 emissions from the United States EPA based on
four projects scenarios, (US EPA, 2015).
These potential impacts include, melting of the world’s polar icecaps, rising sea
levels and changes to current ecological systems that may drastically affect global
food supplies. Figure 1.2 depicts four scenarios for estimated concentrations of
atmospheric CO2 over the next century. The highest emission scenario (RCP 8.5)
considers large scale economic and human growth coupled with no effort to find
sustainable alternatives to fossil fuels, resulting in projected atmospheric CO2 levels
of 1300ppm by 2100 accelerating the global impact of climate change (Nordhaus
2014). To counteract this, most major industrial countries have signed the legally
binding Kyoto Protocol (including Ireland) in a global effort to control spiralling
global CO2 emissions (Kyoto Protocol 1997). Under the Kyoto Protocol, Ireland is
required to limit total national greenhouse gas emissions to 13% above Ireland’s
1990 baseline value (Irish EPA, 2013).
5
Chapter 1
1.4. Transportation fuels
Although energy usage from other sectors appears to be stabilising to an extent, this
is not the case for the transportation sector. For example, from 1990 to 2010, the
average growth of energy use in the Irish transport sector was 4.3% per annum,
which sums to 132% over 20 years (Gusciute et al 2014). The combustion of fossil
fuels for transportation purposes produces 31 % of global CO2 emissions and is the
second biggest contributor to global CO2 emissions after electricity (37 %). Ireland’s
transport sector demands 32.7% of Ireland’s total energy requirements producing
34.7% of Ireland’s total CO2 emissions (See Figure 1.3).
Figure 1.3 Ireland’s total energy requirements by sector (a) Primary energy
consumption CO2 emissions by sector (b) Total CO2 emissions by sector (SEAI
2014a).
Fossil derived diesel (60%) and gasoline (40%) are used to power automobiles for
road transportation in Ireland (Hamelinck et al 2004). This reliance on diesel and
gasoline is not sustainable and Ireland has made legally binding commitments for the
introduction of biofuels into transportation fuel combustion as part of the Biofuels
Directive 2003/30/EC (EU-Commission 2003). The Biofuels Directive 2003/30/EC
entered into force in May 2003 and is primarily concerned with the promotion of the
6
Chapter 1
use of biofuels in the transport sector. Each member state was required to replace 5.7
% of all transport fossil fuels with biofuels by 2010. The directive also set an
intermediate target of 2% by December 2008. This was replaced with Directive
2009/28/EC detailing targets of 10% by 2020 for every member EU state.
As a result of the above, Ireland needs to develop strategies for the provision of
renewable (non-fossil) alternative liquid transportation fuels for substitution of fossil
diesel and gasoline fuels. Current alternatives that are being considered in order to
comply with the Biofuels Directive 2003/30/EC include the use of electric vehicles,
bioethanol, waste vegetable oils, and grass gasification for biomethane for natural
gas vehicles (Gusciute et al 2014). However another promising such alternative is
the conversion of lignocellulosic derived hexose carbohydrates to oxygenated liquid
fuel components.
1.5. Biomass as an energy source of liquid fuels
Currently, a great deal of attention is focused on the production of biofuels (Chheda
et al 2007; Stӧcker 2008; Alonso et al 2010). Unlike fossil fuels, biofuels are fuels
derived from biomass or animals and are renewable on a practical timescale.
Combustion is one of the oldest forms of energy generation, but liquid and gaseous
fuel types may also be biomass derived (Demirbas 2005). Today, the desired
products are liquids, largely due to the dominance of the internal combustion engine
in world transportation. Often, biomass sources are remote relative to both the
processing plant and the consumer, therefore liquid fuels are ideal. Liquids typically
have much higher energy densities compared to solid biomass and are appealing due
to the ease of handling, storage and transportation. Thus the production of dense high
7
Chapter 1
energy liquid fuel components from biomass represents an enormous opportunity to
develop alternatives to fossil derived gasoline and diesel.
1.6. From 1st to 2nd generation biofuels
Currently most renewable liquid fuel production is focused on the conversion of
starch, fats, animal’s fats, and vegetable oil to biofuels (so called first generation
biofuels). However there are significant ethical concerns with the production of fuels
from ‘‘food’’ feed-stocks (Fairley 2011). To alleviate these concerns, there has been
a recent shift away from first generation feed-stocks, towards the use of non‐food
carbohydrates and lipids such as waste cooking oil. This can be attributed to the
recent emergence of lignocellulosic feed-stocks which are the most abundant
renewable materials available (Naik et al 2010; Sims et al 2010). It is estimated that
the global annual production of lignocellulosic materials exceeds 1×1010 MT
annually (Harmsen et al 2010). This enormous resource can be sourced from waste
streams such as municipal green waste, forestry and agricultural waste and from
specialised energy crops as outlined in Table 1.2. Significantly none of these feedstocks can be used as a significant human food source. Thus the synthesis of
chemicals, potential fuels and fuel additives from non‐food feed-stocks reduces
competition for food commodities. As a result, this eliminates many of the
objections raised by the debate between ‘food versus fuel’ for which first generation
biofuels were highly criticised.
8
Chapter 1
Classification
Source
Energy crops
Perennial crops
Short rotation coppices
(Willow poplar) Plantation
trees (Eucalptus) Energy
grasses (Miscanthus)
Residues
Forestry residues
Wood left on field (small
trees, branches), poor quality
stem wood
Agricultural crop residues
Straw from cereals (wheat
rye), oil, seed rape and
residues from other crops
Processing residues
Wood chips, sawdust and
bark from sawmill,
agricultural arising’s (from
municipal tree surgery
operations
Waste wood
Clean and contaminated
wood wastes
Organic waste
Paper/card, food wastes,
textiles wastes
Sewage sludge
Derived from waste water
treatment
Animal manures
Pig manure, poultry litter,
animal slurries and farmyard
manure
Wastes
Table 1.2 Classification of biomass feed-stocks for second generation bio-refining
adopted from (Slade et al 2011).
1.7. The conversion of carbohydrates to liquid transportation fuel
In general lignocellulosic biomass contains three main components, cellulose,
hemicellulose and lignin (Hayes 2013). Cellulose is a major structural component of
cell walls, and it provides mechanical strength and chemical stability to plants. It is a
polymer of glucose, and can be represented by the chemical formula (C6H10O5)n
(Harmsen et al 2010; Hayes 2013). Hemicellulose is a copolymer of different hexose
9
Chapter 1
and pentose sugars that also exists in the plant cell wall, whilst lignin is the third
component which is a polymer of aromatic compounds (Harmsen et al 2010).
However, it is the hexose and pentose carbohydrates components of the
lignocellulosic biomass in the form of hemicellulose and cellulose that are of
particular interest to chemists. Figure 1.4 illustrates that there are a wide variety of
possible thermochemical methods used for the conversion of lignocellulosic biomass
derived carbohydrates to liquid fuels, including gasification, pyrolysis and acid
hydrolysis.
Figure 1.4 Suggested routes for the conversion of biomass to liquid fuels taken from
(Serrano-Ruiz and Dumesic 2011).
1.8. Acid hydrolysis
The focus of this Thesis is acid hydrolysis (as elaborated upon later). It is employed
to liberate the hexose and pentose carbohydrates from the cellulose and
10
Chapter 1
hemicellulose components of lignocellulosic biomass. Once liberated the pentose
and hexoses carbohydrates can undergo acid catalysed transformations to form
furfural in the case of pentose carbohydrates and 5-hydroxymethylfurfural in the
case of hexoses. The attraction in hydrolysing hemicellulose and cellulose to
monomeric glucose is the versatility that it exhibits as a starting material and is ready
availability and inexpensiveness. Upon isolation glucose can undergo a vast range
of chemical reactions such as reduction reactions to form sorbitol (Davda and
Dumesic 2004), fermentation, to form products such as succinic acid (Meynial-Salles
et al 2008) and ethanol (Sun and Cheng 2002), or its dehydration to form 5hydroxymethylfurfural which is pertinent to the focus of this Thesis (Kuster 1990).
5-hydroxymethylfurfural is historically obtained from cellulose through acid
hydrolysis using mineral acids (Ackerson et al 1981; Helm et al 1989) in which
yields of up to 50 mol% were obtained. The yield limiting factor being, the
prevalence of undesirable reactions producing the polymeric humic material (Dee
and Bell 2011).
The potential of 5-hydroxymethylfurfural as a bulk platform
chemical in a bio-based economy was heighted by the US Department of Energy. In
this report, 5-hydroxymethylfurfural is listed in the top ten value chemicals from
biomass (Werpy et al 2004). This Thesis seeks to synthesise transportation liquid
fuel components from derivatives of 5-hydroxymethylfurfural derived from hexose
carbohydrates by reaction with ethanol.
1.9. Ethanol
Ethanol is a biofuel produced from agricultural feed-stocks that are high in sugar or
starch such as sugarcane, sugar beet, corn (maize), and wheat, as well as
lignocellulosic crops, and is the biggest industrial bio-refining product globally with
11
Chapter 1
approximately 25 million gallons produced in 2015 (Baier et al 2009). Ethanol is an
attractive fuel as it is appropriate for blending requirements in terms of compliance
with the Biofuels Directive 2003/30/EC. Currently ethanol is suitable for use in
conventional spark ignition engines, without any modification when blended up to
5% v/v with fossil gasoline. It can be blended in 85% v/v mixtures for use in
engines, which have been specifically modified (NREL 2015). As a result of this
global ethanol production has increased dramatically over the past 20 years (See
Figure 1.5). The United States is the largest producer of ethanol having produced 14
billion gallons of ethanol alone in 2014. Together the US and Brazil produce 83% of
global ethanol (Baier et al 2009).
Brazil are leading the way in utilising ethanol as a liquid transportation fuel. All their
filling stations are required to sell blends of gasohol (25% ethanol blends with
gasoline) and pure ethanol (Rüther 2015). Tax incentives have been given by the
Brazilian government to encourage the use of 100% ethanol in vehicles. Ethanol as a
fuel has been further strengthened by the introduction of so called flex fuel vehicles
by most of the local automakers, allowing for various mixtures of ethanol/gasoline to
be used at any time (25-100% by volume). Currently the production costs of ethanol
in Brazil are approximately $0.20/L compared to $0.55/L in Europe (Rüther 2015),
with the current price of pure ethanol at the pump being $ 1.61/L in Europe as of
October 2015 (NASDAQ 2015). In spite of all this, there is no ethanol production
industry in Ireland.
However there are efforts now to revive sugar beet production in Ireland and to
develop a bioethanol facility that requires 1 million tonnes of sugar beet and 56,000
tonnes of grain to produce 154,000 tonnes of sugar and 50 million L of ethanol at an
estimated plant construction cost of €350–€400 million (Gusciute et al 2014).
12
Chapter 1
Figure 1.5 Predicted Global annual ethanol production by country taken from (Baier
et al 2009).
Despite these recent technological advancements, it cannot be ignored that ethanol is
a relatively poor fuel. For example it is not as easy to transport to the pump as
gasoline, due to its tendency to absorb water, and is also a corrosive liquid (NREL
2015). There are also considerable concerns associated with its storage and transport
due to its high volatility (boiling point 351 K). As well as this its high octane number
restricts its usefulness as a blend-stock with petroleum derived fuels (NREL 2015).
However above all, its major disadvantage is its low volumetric energy density,
which is a key parameter for a transportation fuel. Its
∆HCombustion at 24 MJ/L is
75% that of conventional gasoline (32 MJ/L) and 61% that of diesel (38 MJ/L)
(NIST 2016).
13
Chapter 1
1.10.
Synthesising liquid transportation fuel components by reacting hexose
carbohydrates with ethanol
This Thesis focuses on the synthesis of high energy density liquid transportation
components fuels by the reaction of lignocellulosic derived carbohydrates with
ethanol in the presence of conventional mineral acids. As demonstrated above both
the lignocellulosic carbohydrates and ethanol are available in significant quantities
and are renewable in source in terms of replacing existing fossil derived liquid fuels
and reducing global CO2 emissions. Lignocellulosic derived hexose such as Dglucose and D-fructose although energy dense materials (∆HCombustion 26.3 kJ/mol)
are present in a solid form containing six oxygen atoms per molecule. Reaction with
ethanol converts the solid hexoses into liquid form, removing oxygen molecules,
producing energy dense liquid oxygenated hydrocarbons that can be used as ‘‘drop
in manner’’ in a one-pot synthesis. Ethanol acts as solvent as well as a reactant
partner with which to valorise the hexose carbohydrates by acid hydrolysis
(“ethanolysis”). Their reaction can result in the formation of ethyl levulinate and 5ethoxymethylfurfural
both of which show considerable promise as oxygenated
hydrocarbons due to their high volumetric energy densities (Dautzenberg and Gruter
2012; Wang et al 2012) and are the focus of this Thesis. Thus the overall process
valorises both the lignocellulosic derived hexose and ethanol simultaneously (See
Figure 1.6).
14
Chapter 1
Figure 1.6 Valorisation of hexose carbohydrates by reaction with ethanol producing
energy dense oxygenated hydrocarbons.
Acid hydrolysis of hexoses, in ethanol in a one-pot synthesis offers several other
advantages compared to conventional acid hydrolysis in aqueous systems. For
example the hydrolysis of hexoses in an alcohol media also tends to produce a lot
less insoluble humins than in aqueous systems, which are a waste product of reaction
and thus negatively affect yields of the intended reaction products (Maldonado et al
2012). Additionally hydrolysis rates of sugars in ethanol systems is at least an order
of magnitude faster than in aqueous systems (Flannelly et al 2015).
15
Chapter 1
1.11. Scope of the research
At present yields of the ethyl levulinate and 5-ethoxymethylfurfural from hexoses
have been poor (<60%), threatening the economic viability of the proposition due to
an unsatisfactory understanding of the mechanism and chemical kinetics of the
synthetic system. Reaction kinetics and mechanistic understanding for the hydrolysis
of hexoses in even conventional aqueous systems are beyond state-of-art (Swift et al
2013). Even less kinetic information regarding the acid catalysed transformations of
hexose in non-aqueous systems are available in the literature. A recent state-of-theart review has stated a kinetic and mechanistic understanding for the hydrolysis of
sugars, is key for the development of a sustainable biorefinig industry (Saha et al
2015). It is envisioned that these technical and scientific difficulties can be resolved
by the provision of reliable mechanistic analysis and by the development of high
fidelity kinetic models to describe the synthesis of these fuel components from
lignocellulosic derived hexoses. For reasons discussed in later chapters, any efficient
process for the production of fuel components through furanic intermediates such as
5-ethoxymethylfurfural and ethyl levulinate are envisioned go through a D-fructose
intermediate. Higher yields of furanics such as 5-hydroxymethylfurfural are more
achievable from D-fructose than from D-glucose (Assary et al 2012). D-Fructose is an
isomerisation product of glucose and the optimisation of this transformation is
currently the subject of extensive investigation (Choudhary et al 2013;
Saravanamurugan et al 2013). Consequently, this Thesis focuses on achieving a
mechanistic and kinetic understanding of the acid catalysed transformation of Dfructose in acidic ethanol conditions. Once understood, this understanding can be
used as constraint to evaluate the simpler glucose and cellulose substrates in a
hierarchical manner eventually comprehending the global reaction mechanism of
16
Chapter 1
lignocellulosic biomass to the hexose sugar to the energy dense fuels (See Figure
1.7).
Figure 1.7 A hierarchical scheme for robust mechanistic and kinetic reaction
modelling.
1.12.
Outline of the Thesis

Chapter 1 outlines the motivation and background for this Thesis.

Chapter 2 identifies the properties of 5-ethoxymethylfurfural and ethyl
levulinate that make them advantaged liquid transportation fuel components.
Attention is given to the various methods employed for their systhesis using
homogenous and hetergenous catalysts. As well as this mechanisms for the
hydrolysis of D-fructose and D-glucose to 5-hydroxymethylfurfural are
explored in detail, all of which is pertient to this PhD Thesis.
17
Chapter 1

Chapter 3 briefly describes the fundamentals of the experimental and
analytical techniques used in the generation of data for this PhD Thesis. The
techniques used for conducting reaction mechanism analysis and kinetic
modelling are also described.

Chapter 4 focuses on conventional acid hydrolysis processes for the
formation of levulinic acid from hexose carbohydrates in aqueous systems.
Here the mechanistic content is set in relation to the remainder of the Thesis.
It outlines the variety of reaction pathways present in conventional acid
hydrolysis systems that are not net contributers for the formation of the
desired furanics and levulinic acid. Particular attention is given to formic
acid. Historical mechanistic understanding suggests formic and levulinic
acids are formed stoichiometrically from
the acid hydrolysis of
lignocellulosic hexose carbohydrates. This chapter focuses on acessing this
common assumption.

Chapter 5 introduces ethanol as an alternative reaction mediator to aqeous
systems. Here, an extensive reaction pathway analysis of the acid hydrolysis
of D-fructose to 5-ethoxymethylfurfural and ethyl levulinate in ethanol
catalyed by hydrogen cations at 351 K is studied. An extensive mechanistic
study is conducted, focusing in particular on the variety of intermiediates
formed between D-fructose and 5-hydroxymethylfurfural. A methodology
with which to access the accuracy of mechanistic propositions is developed
by the elucidation of the mechanism fidelity index. Special emphasis is
placed on the behavaviour of H2SO4 in ethanol and a comparison of its
dissociation compared to conventional aqeous systems is investigated.
Finally, phenomenological kinetic modelling is conducted on four particular
18
Chapter 1
mechanistic propositions to a give a phenomological estimation of reaction
rate constants for eight separate reaction pathways using a mass conserved
modelling approach.

Chapter 6 builds on the results of the mechanistic analysis conducted in
Chapter 5. It focuses on building a low temperature (331-351 K) kinetic
model to predict the formation of 5-ethoxymethylfurfural from D-fructose in
a one-pot-system. To do this a hierarchial approach is employed utilising data
produced from 181 individual experiments conducted with the main reaction
intermiediates in the system.

Chapter 7 consists of three compoments. In the first part a provisional
mechanistc study is conducted on the ethanolysis of D-glucose to ethyl
levulinate at temperatures between 351 K and 460 K. The second core
component focuses on the consumption of ethanol to diethyl ether at
temperatures above 373 K. Finally the third component focuses on the fuel
properties of all the potiential fuel molecules produced in the Bronsted acid
ethanolysis system. As well as this, the fuel properties of all the molecules
present in conventional acid hydrolysis processes is examined. It is then
illustrated how the molecular diversity present can be leveraged to produce
tailorable fuel properties (either diesel or gasoline) in a one-pot-synthesis.

Chapter 8 presents the key findings of this work and directions for future
work.
19
Chapter 1
1.13. References
Ackerson, M., Ziobro, M., Gaddy, J. (1981) “Two-stage acid hydrolysis of biomass,”
in Biotechnol. Bioeng. Symp.;(United States).
Alonso, D.M., Bond, J.Q., Dumesic, J.A. (2010) “Catalytic conversion of biomass to
biofuels,” Green Chemistry, 12(9), 1493–1513.
Assary, R.S., Kim, T., Low, J.J., Greeley, J., Curtiss, L.A. (2012) “Glucose and
fructose to platform chemicals: understanding the thermodynamic landscapes of
acid-catalysed reactions using high-level ab initio methods,” Physical Chemistry
Chemical Physics, 14(48), 16603–16611.
Baier, S.L., Clements, M., Griffiths, C.W., Ihrig, J.E. (2009) “Biofuels impact on
crop and food prices: using an interactive spreadsheet,” FRB International
Finance Discussion Paper, (967).
Chheda, J.N., Huber, G.W., Dumesic, J.A. (2007) “Liquid-phase catalytic processing
of biomass-derived oxygenated hydrocarbons to fuels and chemicals,”
Angewandte Chemie International Edition, 46(38), 7164–7183.
Choudhary, V., Pinar, A.B., Lobo, R.F., Vlachos, D.G., Sandler, S.I. (2013)
“Comparison of homogeneous and heterogeneous catalysts for glucose-to-fructose
isomerization in aqueous media,” ChemSusChem, 6(12), 2369–2376.
Dautzenberg, F.; Gruter, G. J. M. Method for the synthesis of 5-alkoxymethyl
furfural ethers and their use. Patent 8, 133, 289 B2, May, 2012.
Davda, R.R., Dumesic, J.A. (2004) “Renewable hydrogen by aqueous-phase
reforming of glucose,” Chem. Commun., (1), 36–37.
20
Chapter 1
Dee, S.J., Bell, A.T. (2011) “A study of the acid-catalyzed hydrolysis of cellulose
dissolved in ionic liquids and the factors influencing the dehydration of glucose
and the formation of humins,” ChemSusChem, 4(8), 1166–1173.
Demirbas, A. (2005) “Potential applications of renewable energy sources, biomass
combustion problems in boiler power systems and combustion related
environmental issues,” Progress in energy and combustion science, 31(2), 171–
192.
EU-Commission, others (2003) “Directive 2003/30/EC of the European Parliament
and of the Council of 8 May 2003 on the promotion of the use of biofuels or other
renewable fuels for transport,” Official Journal of the European Union, 5.
Fairley, P. (2011) “Introduction: Next generation biofuels,” Nature, 474(7352), S2–
S5.
Flannelly, T., Dooley, S., Leahy, J.J. (2015) “Reaction pathway analysis of ethyl
levulinate and 5-ethoxymethylfurfural from D-fructose acid hydrolysis in
ethanol,” Energy & Fuels.
Gusciute, E., Devlin, G., Murphy, F., McDonnell, K. (2014) “Transport sector in
Ireland: can 2020 national policy targets drive indigenous biofuel production to
success?,” Wiley Interdisciplinary Reviews: Energy and Environment, 3(3), 310–
322.
Hamelinck, C., van den Broek, R., Rice, B., Gilbert, A., Ragwitz, M., Toro, F.
(2004) Liquid Biofuels Strategy Study for Ireland, ISI.
Harmsen, P., Huijgen, W., Bermudez, L., Bakker, R. (2010) “Literature review of
physical
and chemical pretreatment processes for lignocellulosic biomass”.
Wageningen UR, Food & Biobased Research.
21
Chapter 1
Hayes, D.J. (2013) “Second-generation biofuels: why they are taking so long,” Wiley
Interdisciplinary Reviews: Energy and Environment, 2(3), 304–334.
Helm, R.F., Young, R.A., Conner, A.H. (1989) “The reversion reactions of d-glucose
during the hydrolysis of cellulose with dilute sulfuric acid,” Carbohydrate
Research, 185(2), 249–260.
Irish EPA (2014), Ireland Greenhouse Gas Emissions 2013, [online]
http://www.epa.ie/pubs/reports/air/airemissions/GHGprov.pdf [accessed 14
October 2015].
Kuster, B. (1990) “5-Hydroxymethylfurfural (HMF). A review focussing on its
manufacture,” Starch-Stärke, 42(8), 314–321.
Maldonado, G.M.G., Assary, R.S., Dumesic, J.A., Curtiss, L.A. (2012) “Acidcatalyzed conversion of furfuryl alcohol to ethyl levulinate in liquid ethanol,”
Energy & Environmental Science, 5(10), 8990–8997.
Meynial-Salles, I., Dorotyn, S., Soucaille, P. (2008) “A new process for the
continuous production of succinic acid from glucose at high yield, titer, and
productivity,” Biotechnology and Bioengineering, 99(1), 129–135.
Naik, S., Goud, V.V., Rout, P.K., Dalai, A.K. (2010) “Production of first and second
generation biofuels: a comprehensive review,” Renewable and Sustainable
Energy Reviews, 14(2), 578–597.
NASDAQ (2015) 'Ethanol future Prices', [online], available:
http://www.nasdaq.com/markets/ethanol.aspx [acessed 21 October 2015].
22
Chapter 1
NIST (2016) 'NIST Standard Reference Database Number 69' [online], available:
http://webbook.nist.gov/chemistry/ [acessed 22 March 2016].
Nordhaus, W.D. (2014) A question of balance: weighing the options on global
warming Policies, Yale University Press.
NREL 2015 'Ethanol Blended Fuels',[online], available:
http://www.nrel.gov/education/pdfs/educational_resources/high_school/teachers_
guide_ethanol.pdf [acessed 11 October 2015].
Protocol, K. (1997) “United Nations framework convention on climate change,”
Kyoto Protocol, Kyoto.
Rüther, R. (2015) “Renewable Energy Policies in Brazil: Bioenergy, Photovoltaic
Generation, and Transportation,” Energy Efficiency and Renewable Energy
Handbook, 109.
Saha, B., Bohre, A., Dutta, S., Abu-Omar, M.M. (2015) “Upgrading furfurals to
drop-in biofuels: An Overview,” ACS Sustainable Chemistry & Engineering.
Saravanamurugan, S., Paniagua, M., Melero, J.A., Riisager, A. (2013) “Efficient
isomerization of glucose to fructose over zeolites in consecutive reactions in
alcohol and aqueous media,” Journal of the American Chemical Society, 135(14),
5246–5249.
SEAI (2014 a) ' Energy in Ireland',[online], available:
http://www.seai.ie/Publications/Statistics_Publications/Energy_in_Ireland/Energyin-Ireland-1990-2013-report.pdf [ accessed 10 October 2015].
SEAI (2014 b) ' SEAI Press Release', [online],
availablehttp://www.seai.ie/News_Events/Press_Releases/2014/Renewable-
23
Chapter 1
energy-has-saved-Ireland-over-%E2%82%AC1-billion-in-fossil-fuel-imports-inpast-five-years.html [accessed 11 October 2015].
SEA1 (2014 c)' Energy in Ireland Key Statistics 2014',[online]
http://www.seai.ie/Publications/Statistics_Publications/Energy_in_Ireland/Energy
_in_Ireland_Key_Statistics/Energy-in-Ireland-Key-Statistics-2014.pdf [accessed
11 October 2015].
Serrano-Ruiz, J.C., Dumesic, J.A. (2011) “Catalytic routes for the conversion of
biomass into liquid hydrocarbon transportation fuels,” Energy & Environmental
Science, 4(1), 83–99.
Sims, R.E., Mabee, W., Saddler, J.N., Taylor, M. (2010) “An overview of second
generation biofuel technologies,” Bioresource Technology, 101(6), 1570–1580.
Slade, R., Saunders, R., Gross, R., Bauen, A. (2011) “Energy from biomass: the size
of the global resource,” Imperial College Centre for Energy Policy and
Technology and UK Energy Research Centre, London.
Stӧcker, M. (2008) “Biofuels and biomass-to-liquid fuels in the biorefinery:
Catalytic conversion of lignocellulosic biomass using porous materials,”
Angewandte Chemie International Edition, 47(48), 9200–9211.
Sun, Y., Cheng, J. (2002) “Hydrolysis of lignocellulosic materials for ethanol
production: a review,” Bioresource Technology, 83(1), 1–11.
Swift, T.D., Bagia, C., Choudhary, V., Peklaris, G., Nikolakis, V., Vlachos, D.G.
(2013) “Kinetics of homogeneous Brønsted acid catalyzed fructose dehydration
and 5-hydroxymethylfurfural rehydration: A combined experimental and
computational study,” ACS Catalysis, 4(1), 259–267.
24
Chapter 1
US EPA (2015)' Global Green Gas Emissions Data', [online]
http://www3.epa.gov/climatechange/ghgemissions/global.html [accessed 11
October 2015].
Wang, Z., Lei, T., Liu, L., Zhu, J., He, X., Li, Z. (2012) “Performance investigations
of a diesel engine using ethyl levulinate-diesel blends,” BioResources, 7(4), 5972–
5982.
Watts (2011)' The Fate of all Carbon',[online], available:
http://wattsupwiththat.com/2011/11/13/the-fate-of-all-carbon/ [accessed 18 October
2015].
Werpy, T., Petersen, G., Aden, A., Bozell, J., Holladay, J., White, J., Manheim, A.,
Eliot, D., Lasure, L., Jones, S. (2004) “Top value added chemicals from biomass.
Volume 1-Results of screening for potential candidates from sugars and synthesis
gas.
25
Chapter 2
Chapter 2: Literature Review
2. Literature review
26
Chapter 2
2.1. Lignocellulosic derived hexose sugars: D-fructose and D-glucose
Hexose sugars are present in lignocellulosic biomass in the form of cellulose and
hemicellulose. Cellulose consists of D-glucose subunits, linked by β-1,4 glycosidic
bonds, whilst hemicellulose consists of D-glucose with minor amounts of D-galactose
and D-mannose, however the composition is largely dependent on the feedstock
(Hendriks and Zeeman 2009). Of the hexoses mentioned, D-glucose is of primary
concern to this Thesis. For carbohydrates the ‘‘D’’ and ‘‘L’’ notation refers to the
asymmetric carbon farthest from the aldehyde and the α/β refers to positions (axial
and equatorial) of the anomeric acetal carbon (Whistler et al 1964). In solution, the
open-chain form of D-glucose (either ‘‘D’’ or ‘‘L’’) exists in equilibrium with
several cyclic isomers, each containing a ring of carbons closed by one oxygen atom.
In aqueous solutions, more than 99% of glucose molecules exist as pyranose
structures. D-Glucose can also exist in the open chain form which is limited to about
0.25% of the total glucose structures and the furanose form, which exists in
negligible amounts (Kimura et al 2011).
Figure 2.1 Tautomeric distributions of D-fructose/D-glucose reactant in solution at
383 K in aqeous systems with no catalyst as derived by (Kimura et al 2011).
27
Chapter 2
Fructose in solution also shows tautomeric behaviour (See Figure 2.1) Of the
tautomers β-D-fructopyranose is the preponderant tautomer, followed by β-Dfructofuranose, and then α-D-fructofuranose in equilibrium with water (Kimura et al
2011). These tautomers have previously been determined to account for 69.6%,
21.1% and 5.7% of the solubilised sugar at room temperature employing in-situ
NMR (Angyal 1969). It has been shown that at higher temperatures the equilibrium
shifts in favour of furanose structures (Kimura et al 2013). The presence of furanose
structures is pertinent as in means that D-fructose can be more easily converted to 5hydroxymethylfurfural which is envisioned to be a key intermediate for the
formation of oxygenated hydrocarbon as is the aim of this Thesis. The reason for this
is that the furanose structures are more thermodynamically driven to dehydrate than
the equivalent pyranose structures that make up the composition of that D-glucose.
For instance, Assary et al. conducted a thermodynamic study in which they predicted
the Gibbs free energy required to protonate the various oxygen atoms on
glucopyranose, fructopyranose and fructofuranose structures (See Figure 2.2).
Figure 2.2 Predicted Gibbs free energy (298 K) required to protonate the various
oxygen sites of glucopyranose, fructopyranose, and fructofuranose as adopted
calculated by (Assary et al 2012) using G4MP2 theory.
28
Chapter 2
They found that the Gibbs free energy required to protonate the primary oxygen sites
on the fructofuranose structure to be 8 kcal/mol which is significantly lower than
both the fructopyranose (11.9 kcal/mol) and glucopyranose structures (15.0
kcal/mol) (Assary et al 2012).
In the literature there is sometimes confusion between the naming of the hexose
sugars, hence forth α/β-D-fructopyranose and α/β-D-glucopyranose are termed ‘‘Dfructose’’ and ‘‘D-glucose respectively’’. The correct description is provided in
Figure 2.1
2.2. D-Fructose/D-glucose isomerisation
To date, the yields of 5-hydroxymethylfurfural achieved from D-glucose have been
modest (typically <50 mol%) due to the electronic configuration of its hydroxyl
groups which increases its propenisty to form by-products and in particular the yield
limiting insoluble humins (Choudhary et al 2012). In addition, the stability of its ring
structure makes processing into critical platform furanic molecules such as 5hydroxymethylfurfural
difficult
(Nikolla
et
al
2011).
Yields
of
5-
hydroxymethylfurfural reported in conventional acid hydrolysis systems have been
significantly higher from D-fructose with frequent reporting of selectivities of over
80% using a variety of catalysts (Fan et al 2011). D-fructose is not an ideal biomass
derived feed-stock due to the fact that it is quite expensive and is not readily
available from lignocellulosic biomass unlike D-glucose. Indeed the cost of Dglucose is about 50% that of D-fructose (Torres et al 2012). Configuring hexose
dehydration processes to go through D-fructose, in theory should improve reaction
yields of the desired fuel components from cellulosic materials and thus make such
processes economically viable. Therefore it is envisioned that the state-of-the-art
29
Chapter 2
scientific approach for the optimal synthesis of 5-hydroxymethylfurfural from
biomass includes three mains steps (Choudhary et al 2013):
1. The hydrolysis of biomass into hexose carbohydrates (mainly D-glucose).
2. D-Glucose to D-fructose isomerisation.
3. D-Fructose dehydration to 5-hydroxymethylfurfural and to other products.
As a result of this, the development of catalysts in aqueous media is an important
milestone in societies ability to efficiently utilise lignocellulosic biomass. Typically
D-glucose to D-fructose isomerisation occurs in three stages:
1. D-Glucose ring opening.
2. D-Glucose isomerisation to D-fructose (unclear whether this is a
fructopyranose or fructofuranose structure).
3. Ring closure of D-fructose (See Figure 2.3).
Consequently this isomerisation reaction is the subject of numerous researchers
endeavours by employing a variety of heterogeneous (Moliner et al 2010; Despax et
al 2013) and homogenous (Pagan-Torres et al 2012; Choudhary et al 2013) catalysts.
Significantly for this Thesis successful D-glucose/D-fructose isomerisation has been
reported in an ethanol media. For instance, Saravanamurugan et al. achieved Dfructose yields of 55 mol% from D-glucose using a ZSM-5, and mordenite
heterogenous catalyst at 393 K (Saravanamurugan et al 2013). However, at present
the most promising approach for D-glucose/D-fructose isomerisation in terms of the
formation of 5-hydroxymethylfurfural is simultaneous use of Bronsted and Lewis
acids to catalyse the reaction. The Lewis acid catalyses the D-glucose to D-fructose
isomerisation reaction and the Bronsted acid dehydrates D-fructose to 5-
30
Chapter 2
hydroxymethylfurfural. Using this approach, Pagan-Torres et al. achieved 5hydroxymethylfurfural yields of 62 mol% using a combination of HCl and AlCl3
from D-glucose as reactant (Pagan-Torres et al 2012). Similarly Choudhary et al.
using a combination of HCl and CrCl3 achieved yields 5-hydroxymethylfurfural of
59 mol% (Choudhary et al 2013).
Figure 2.3 Proposed pathway from cellulose to D-glucose to D-fructose to 5hydroxymethylfurfural.
2.3. Mechanisms and kinetics for the dehydration of D-fructose in aqueous
systems
The mechanisms and kinetics responsible for the dehydration of D-fructose to 5hydroxymethylfurfural is not a well understood process. Despite their being
numerous kinetic studies conducted in the literature, a definite mechanism for the
dehydration of D-fructose has yet to be unambiguously determined. For example a
recent gas-phase study of Bronsted acid catalysed D-fructose and D-glucose
dehydration screened and identified more than 100 possible reaction intermediates
by focusing only on the energetics of stable intermediates (Yang et al 2012). The
large number of pathways highlights the complexity of the problem. The vast
majority of detailed kinetic studies conducted on D-fructose dehydration focuses on
the formation of cyclic or acyclic intermediates (Akien et al 2012; Yang et al 2012).
31
Chapter 2
Caratzoulas and Vlachos conducted a kinetic study of the closed ring members
considering quantum mechanics (Caratzoulas and Vlachos 2011). They suggest that
D-fructose dehydration to 5-hydroxymethylfurfural proceeds via multiple elementary
steps, some of which consider intramolecular proton transfers. Building on the
findings of Caratzoulas and Vlachos, Swift et al. developed a practical skeleton
model that also captures D-fructose tautomer distribution to quantitatively describe
an extensive set of experimental data (Swift et al 2013). D-Fructose dehydration was
found to be first order dependent on proton and D-fructose concentration. Their
kinetic model suggests that for optimum yields of 5-hydroxymethylfurfural higher
temperatures are favourable due to the tendency of D-fructose to form by-products at
lower temperatures. Despite this recent progress, identification of intermediates,
reaction rates and selectivities remain poorly understood (Caratzoulas et al 2014).
Figure 2.4 Two-step D-fructose dehydration scheme as suggested by (Swift et al
2013).
2.4. Mechanism for the dehydration of D-glucose in aqueous systems
Unlike D-fructose, fundamental kinetic and mechanistic studies for the acid catalysed
transformations of D-glucose to 5-hydroxymethylfurfural are difficult to find. To
date the most prominent fundamental study has been conducted by Yang et al. (Yang
et al 2015). Here they formulate a computional theory micro-kinetic model,
constrained by experimental NMR measurements. The reaction model considers two
32
Chapter 2
main pathways for D-glucose conversion to 5-hydroxymethylfurfural. The main
pathway considers the direct transformation of the pyranose ring to a five membered
ring intermediate. The alternative acyclic mechanism proposes D-glucose
isomerization
to
D-fructose
before
its
subsequent
dehydration
to
5-
hydroxymethylfurfural. They suggest that the dehydration rate depends strongly on
the reaction temperature and [H+] and that little D-glucose isomerisation occurs
during Bronsted acid catalysed transformation of D-glucose (See Figure 2.5).
Figure 2.5 Reaction path analysis of major elementary processes in Bronsted acidcatalysed D-glucose dehydration from (Yang et al 2015).
There have been numerous empirical ‘‘engineering’’ type models produced to
develop useable kinetic parameters for the production of levulinic acid from Dglucose (Weingarten et al 2012; Girisuta et al 2006; Tarabanko et al 2002). For
33
Chapter 2
example Weingarten et al. developed a kinetic model for the aqueous-phase
production of levulinic acid from D-glucose using HCl at temperatures of 413-473 K
(Weingarten et al 2012). Their kinetic model considered four key steps:
1. D-Glucose dehydration to form 5-hydroxymethylfurfural.
2. D-Glucose degradation to form humins.
3. 5-Hydroxymethylfurfural rehydration to form levulinc aid.
4. 5-Hydroxymethlfurfural degradation to form humins.
Using this non-mass conversed approach an accurate kinetic model was developed,
but a fundamental mechanistic analysis was not acquired.
2.5. Mechanisms and kinetics of hexose carbohydrates in non-aqueous solvents
In recent times hexose dehydration in organic solvents such as dimethyl sulfoxide, γvalerolactone and alcohols have emerged as competitors to conventional aqeous
systems. Unfortunately hexose dehydration to 5-hydroxymethylfurfural often isn’t
the major acid catalysed degradation product (Choudhary et al 2012). Higher
selectivities to 5-hydroxymethylfurfural can be achieved in non-aqueous solvents
minimising the formation of unwanted by-products. For example selectivities for 5hydroxymethylfurfural formation of up to 100% from D-fructose can be achieved
(Shimizu et al 2009). Unfortunately D-glucose does not show the same behaviour.
For instance Chheda et al. reported 5-hydroxymethylfurfural selectivity of 53% with
D-glucose conversions of 48% (Chheda et al 2007).
34
Chapter 2
Figure 2.6 Structures of 2,6-anhydro-β-D-fructofuranose (AFF), 1,6-anhydro-β-Dglucofuranose (AGF), 1,6-anhydro-β-D-glucopyranose (AGP), and levoglucosenone
(LGN), as adopted from (Qi et al 2014).
Not surprisingly changing the solvent changes the reaction mechanism of hexose
dehydration. In addition to the pyranose, furanose cyclic and non-cyclic intermediate
structures between D-fructose/D-glucose and 5-hydroxymethylfurfural, the use of
organic solvents such as dimethyl sulfoxide and γ-valerolactone can lead to the
formation of anhydro-hexoses.
When investigating the mechanism for sucrose hydrolysis to levulinic acid in γvalerolactone catalysed by HCl, Qi et al. detected, 6-anhydro-β-D-fructofuranose
(AFF),
1,6-anhydro-β-D-glucofuranose
(AGF),
1,6-anhydro-β-D-glucopyranose
(AGP), and levoglucosenone in significant quantities (See Figure 2.6) (Qi et al
2014). Choudary et al. found that that the formation of anhydro-glucose increases
with temperature when investigating D-glucose dehydration in dimethyl sulfoxide,
and that the addition of water to the reaction solvent lowers the selectivity to
anhydro-hexose formation (Choudhary et al 2012). This all illustrates how the
reaction medium can manipulate the reaction mechanism to produce higher yields
and selectivities of the desired components.
35
Chapter 2
As the mechanistic behaviour of hexose dehydration in non-aqueous solvents is at an
early stage in terms of its development no kinetic studies of note have been
conducted to date regarding hexose dehydration using alcohols as solvent.
2.6. Mechanism and kinetics of hexose dehydration in alcohols
The investigation of the mechanisms and kinetics responsible for the dehydration of
hexoses in alcohol solvent is of particular interest to this Thesis for reasons outlined
in Chapter 1. The
most pertinent study conducted to date regarding the
transformation of hexoses carbohydrates in alcohol solvent was performed by
Kimura et al. Conducting an NMR study they detected 1,6-anhydro-α-Dfructofuranose,
2,6-anhydro-β-D-fructofuranose
and
3,6-anhydro-α-D-
fructofuranose in addition to the furanose and pyranose structures detected in
aqueous systems (Kimura et al 2013). As well as this, in an alcohol solvent Dfructose/D-glucose can react with the alcohol solvent forming alkyl fructosides and
glucosides respectively. When conducting a study on the etherification of D-glucose,
Hue at al., found it to form ethyl glucosides under homogeneous Bronsted acid
catalysed conditions (Hu et al 2013). Likewise alkyl fructosides have been reported
from the etherification of D-fructose (Tucker et al 2013; Saravanamurugan et al
2013). Tucker et al. have also reported the formation D-fructose dihydrides in
ethanol as is depicted in Figure 2.7.
36
Chapter 2
Figure 2.7 Possible reaction pathways for the dehydration of D-fructose in alcohol
media adapted from (Tucker et al 2013).
Obviously the formation of fructosides and or difructose anhydrides in the presence
of ethanol results in lower selectivity to 5-hydroxymethylfurfural at earlier reaction
times as one might expect due to the presence of an organic solvent. As the reaction
proceeds, the reversible reactions between ethyl fructosides/difructose anhydrides
and fructose favours fructose resulting in an increased selectivity to 5hydroxymethylfurfural with fructose conversion. Of further interest is that they
estimated that the rate of tautomerisation of D-fructose in 90/10 v/v% ethanol to be
250 times faster than in conventional aqueous systems.
Thus it appears the
ethanolysis process to be a much more efficient way of hydrolysing carbohydrates.
2.7. Ethyl levulinate is an advantaged oxygenated hydrocarbon
Alky levulinates are potential oxygenated hydrocarbons which have generated
considerable attention regarding their use as liquid transportation fuel blending
components. The main attraction in introducing alkyl levulinates into transportation
fuel mixtures is that they can be a superior derived alternative to conventional
biomass derived fuel components (ethanol, methyl-tert-butylether). They have been
reported as blend additives for diesel (Christensen, et al 2011; Wang et al 2012),
37
Chapter 2
gasoline (Fagan et al 2003) and biodiesel (Joshi et al 2014) fuels. Among the
possible levulinate fuel candidates, long alkyl chain levulinates (4 to 10 carbons)
such as ethyl levulinate are currently mainly studied due to their greater solubility in
the hydrocarbons resulting in a lower propensity to be soluble in water (Christensen
et al 2011). Of the alkyl levulinates, ethyl levulinate in particular has attracted
considerable interest from leading oil companies. For example the prominent oil
company Texaco showed that mixtures containing 20% ethyl levulinate, 79% diesel
and 1% of other co-additives can be used as a fuel with reduced sulphur emissions in
diesel engines (Hayes et al 2006).
Figure 2.8 Ethyl levulinate an advantaged oxygenated hydrocarbon.
In general, when fabricating an ideal fuel component (gasoline/diesel component) it
is desirable for the component to accomplish a variety of functions such as helping to
maintain the cleanliness of engine parts, temper fuel gelling and nozzle choking,
prevent corrosion and incomplete combustion of the fuel, improve fuel economy and
reduce greenhouse gases and particulate emissions. It is desirable that the fuel
component is energy dense and does not cause environmental problems upon
combustion, or if leaked into local watercourses. In this sense, ethyl levulinate
presents
considerable
promising
characteristics
as
well
as
non-desirable
characteristic as a fuel additive. On the positive side, it has a density of 1.01g/cm3
38
Chapter 2
and an ∆HCombustion comparable of that of conventional gasoline and over 20% higher
than that of ethanol (See Figure 2.9) (NIST 2016).
160
HCombustion
140
Research Octane Number
120
HCombustion MJ/L
30
100
80
20
60
40
10
Reserach Octane Number
40
20
0
0
Gasoline
Figure 2.9 ∆H
Combustion
5-Ethoxy
methyl
furfural
Ethanol
Ethyl
Levulinate
and research octane number of ethyl levulinate and 5-
ethoxymethylfurfural compared to ethanol and conventional gasoline.
As well as this, ethyl levulinate can be used as a diesel additive at 20% (by volume)
blends without the necessity for modifications to a tested diesel engine (Wang et al
2012). Other attractive fuel properties include its tendency to provide a reduction in
particulate matter emissions, soot formation, and smoke emissions (Wang et al.
2012). Alkyl levulinates also contain characteristics that make them appropriate for
use as cold flow improvers in biodiesel (Joshi et al 2011). For instance Christensen
et al. mixed 10% by volume ethyl levulinate with diesel in a 2008 Cummins engine,
and they found that its addition resulted in the reductions of the engine cut out smoke
number by 41% (Christensen et al 2011a). Importantly from an environmental
perspective, ethyl levulinate also exhibits positive characteristics in relation to its
environmental fate in the atmosphere upon combustion. Regarding this Farkas et al.
conducted a kinetic study of the atmospheric decomposition of ethyl levulinate. They
39
Chapter 2
concluded that it degraded in a short time indicating its likely negligible impact on
global warming and on the deterioration of air quality (Farkas et al. 2011).
Ethyl levulinate also has undesirable properties pertaining to its use as a liquid
transportation fuel.
Christensen et al. compared the relative stability of ethyl
levulinate in various volume % mixtures with conventional road diesel. They
demonstrated that if added in a 33 volume % proportion in ULSD diesel, crude ethyl
levulinate separates from the blends at low temperature (283 K), but this
phenomenon can be limited in the presence of a biodiesel that serves as a co-solvent
(Christensen et al 2011a). Christensen et al. also concluded that although these two
compounds improved the lubricity and conductivity of the fuel, their low cetane
number and poor diesel solubility would limit their use for this application. Thus a
compromise between ethyl levulinate concentration and the envisioned co-blending
partner has to be formulated in order to utilise it efficiently as a diesel fuel
component. A further potential disadvantage of its use as a fuel component has been
recently highlighted by Koivisto et al., who conducted compression ignition
experiments on ethyl levulinate in a single cylinder compression ignition engine.
They found that ethyl levulinate displayed significantly longer ignition delay times
in comparison to conventional diesel molecules (Koivisto et al 2015), which
suggests modifications to a conventional diesel engines may have to be made before
added to diesel in significant quantities. Despite these recent developments there still
is considerable confusion around whether ethyl levulinate is best served as diesel or
a gasoline additive.
The majority of existing literature supposes the use of ethyl levulinate as a
sustainable blend-stock for use with diesel fuel. In this regard, Windom et al. have
determined the volatility behaviour of (1-20 vol%) ethyl levulinate blends with
40
Chapter 2
ULSD and with biodiesel (Windom et al 2011). They show that ethyl levulinate is
more volatile than either fuel, being entirely removed in the first 60% of the distillate
volume fraction, and thus preferentially vaporised to ULSD. However, in a second
study, Christensen et al. also examined the compatibility of ethyl levulinate as a
blend-stock for gasoline. They note that the use of ethyl levulinate (at 3.7 wt%
oxygen, ~5.9 mol%) as a gasoline extender gives favourable attributes over other
possible oxygenated additives, including ethanol and propanols, by; lowering the
volatility of the fuel; increasing blended octane numbers by ~15 (Christensen et al
2011b). Perhaps most notably, ethyl levulinate appears to be more miscible in the
tested gasolines than the diesels. Christenson show that at 10 vol% ethyl levulinate
did not separate from the gasoline until 233 K, in contrast to the more polar, methyl
levulinate which separated at 273 K (Christensen et al 2011a).The position of the
fuel and combustion literature on the prospective uses of ethyl levulinate may thus
be summarised as suggesting it to be intermediate in physical property between
gasoline and diesel distillate, but of a kinetic propensity much lower than
conventional diesel and even lower than conventional gasoline.
2.8. Production pathways for the synthesis of ethyl levulinate
At present there are three main methods, for the synthesis of ethyl levulinate:
1. Alkylation of levulinic acid.
2. Ethanolysis of furfuryl alcohol.
3. Ethanolysis of lignocellulosic biomass derived hexoses.
2.8.1. The esterification of levulinic acid
Esterification reactions like this are normally carried out in the liquid phase using
mineral acids such as H2SO4, H3PO4 or HCl as depicted in Figure 2.10 below:
41
Chapter 2
Figure 2.10 A suggested balanced reaction equation for the esterification of levulinic
acid to ethyl levulinate.
The reaction of ethanol with levulinic acid occurs at room temperature but it can be
accelerated by using high temperatures, pressures and catalysts (Fernandes et al
2012). Whilst mineral acids are effective for the preparation of alkyl levulinates
there are issues regarding their use in commercial applications, with product
separation and reactor corrosion issues being the main concerns. As a result of this
there has been a lot of interest in the use of heterogeneous catalysis in the literature.
Heterogeneous catalysts can be easily separated from reaction mixtures for
recyclability, reusability and also provide advantages in relation to reduced corrosion
effects and easier disposal. Among the solid acid catalysts reported to date,
heteropoly acids (HPAs) and zeolites are the most developed in the literature. For
example, Pasquale et al. achieved yields of 78 mol% of ethyl levulinate using
reusable silica-Wells-Dawson heteropolyacid as catalyst (Pasquale et al 2012) whilst
Yan et al. achieved yields of 67 mol% ethyl levulinate using
mesoporous
H4SiW12O4 (Yan et al 2013). In relation to zeolites, in a recent study conducted by
Patil et al. (Patil et al 2014), the authors showed that there is a difference between
micro and mesoporous H-BEA zeolites for the synthesis of ethyl levulinate. The best
activity was observed on the mesoporous zeolites which also contained sufficient
acid sites for conversion (Patil et al 2014). Other prominent studies conducted
include the investigations by Li et al. and Fernandes et al. Fernandes et al. prepared
42
Chapter 2
sulfated stannia and sulphated titania that resulted in ethyl levulinate yields of 44 and
40 mol% respectively with respect to the levulinic acid starting material. This was in
comparison to the 54% figure achieved by the commercially available Amberlyst-15
at the same reaction conditions (Fernandes et al 2012). Li et al. achieved high yields
of ethyl levulinate from the esterification of levulinic acid using TiO2 nanorods and
ZrO2-modified TiO2 nano composites that had been prepared by hydrothermal
synthesis and a deposition-precipitation method.
LA Conversion
AL Yield
(Mol %)
(Mol %)
378 K, 24h
N/a
Et, ~ 65
(Sah and Ma
1930)
H3PMo12O40–
SiO2
351 K, 5h
76
Et, 76
(Pasquale et al
2012)
H4SiW12O40 SiO2,
348 K, 6h
79
Me, 73
Et, 77
(Yan et al
2013)
H-BEA
zeolites
351 K, 5h
40
Et, 39
(Fernandes et
al 2012)
SnO2-SO3H,
343 K, 10h
45
Et, 45
(Fernandes et
al. 2012)
90
Et, 91
(Li et al. 2012)
Catalyst
Condition
H2SO4
Sulfonated
ZrO2-TiO2
nanorods,
378 K, 3.5h
Reference
Table 2.1 Details of the pertinent studies, regarding the esterification of levulinic
acid (LA) to alkyl levulinate (AL).
Using these methods yields of up to 91 mol% of ethyl levulinate can be achieved at a
reaction temperature of 378 K after 210 minutes (Li et al. 2012).
Table 2.1
43
Chapter 2
summarises the details of the major pertinent studies conducted regarding the
conversion of levulinic acid to ethyl levulinate.
2.8.2. The alcoholysis of furfuryl alcohol to ethyl levulinate
Alkyl levulinates can be formed from the ethanolysis of furfuryl alcohol which is in
itself formed by the reduction of furfural, obtained from the xylan hemicellulose
component of lignocellulosic biomass. Thus, furfuryl alcohol can then undergo
hydrolysis in the presence of an alcohol by the global reaction as is shown in Figure
2.11 below:
Figure 2.11 Suggested reaction pathway for the alcoholysis of furfuryl alcohol to
ethyl levulinate.
Like from the esterification of levulinic acid, high yields of levulinates can be
obtained using conventional mineral acids (Lange et al 2009) with furfuryl alcohol as
reactant. However for the same reasons as the esterification of levulinic acid the use
of alternatives catalysts is being pursued, with solid acid resins and zeolites being the
most successful to date. The most noted study conducted to date in this area was
conducted by Maldonado et al., in which they studied the mechanism responsible for
the ethanolysis of furfuryl alcohol using ab initio quantum chemical calculations.
They also achieved yields of >84 mol% ethyl levulinate using solid acid catalysts
such as Amberlyst-15 and benzenesulfonic acid (Maldonado et al. 2012). Neves et
al. conducted an interesting study when investigating the effective of the porosity of
44
Chapter 2
zeolites and amberlyst resins on alkyl levulinate yields and concluded that the
accessibility of active sites to the reactants is the single most important characteristic
for catalytic furfuryl alcohol transformation to ethyl levulinate (Neves et al 2013).
Finally, the use of metal chloride catalysts has also been reported to perform
adequately in terms of acquiring high yields of ethyl levulinate from furfuryl
alcohol. For example Khusnutdinov and co-workers reported that the reaction of
furfuryl alcohol with an aliphatic alcohol in the presence of an iron‐containing
catalyst resulted in high yields of alkyl levulinates (Khusnutdinov et al 2007). Of the
iron‐containing catalysts investigated in their study, the use of iron (III)
acetylacetonate resulted in the highest yields (80‐90 mol%) of the products (methyl,
ethyl, isopropyl and propyl levulinate).
FAL Conversion
AL Yield
(Mol%)
(Mol%)
343 K, 4 h
100
Me, 80
IPr, 98
(Khusnutdino
v et al 2007)
Benzenesulfonic
Acid
318 K, 10
m
100
Et, 93
(Maldonado
et al 2012)
H-ZSM-5
Zeolite
398 K
100
Et, 80
(Lange et al
2009)
Al-TUD-1
Zeolite
413 K, 24 h
100
Et, 80
(Neves et al
2013)
Benzenesulfonic
Acid
318 K,10
min
100
Et, 93
(Maldonado
et al 2012)
Catalyst
Conditions
Fe(acac)3
Reference
Table 2.2 A summary of pertinent catalysts and methods used for the alcoholysis of
furfuryl alcohol (FAL) to alkyl levulinates (AL).
45
Chapter 2
More information regarding the aforementioned methods and catalysts are obtained
in Table 2.2. Like from, levulinic acid esterification the main difficulty with this
technique, is the difficulty in obtaining furfuryl alcohol as reactant which would
entail several costly separation and purification steps. Therefore an alternative
approach is necessary.
2.8.3. The formation of alkyl levulinates from hexose carbohydrates in a onepot synthesis
In theory the synthesis of alkyl levulinates from hexose carbohydrates possess
several advantages compared to the aforementioned methods. Firstly a one-pot
approach avoids the complications of isolating reactive substances such as furfuryl
alcohol and levulinic acid. As well as this there are considerable economic
consequences of configuring energy intensive separation processes that would be
required for such isolations. Therefore this Thesis focuses on the preparation of alkyl
levulinates from hexose carbohydrates in a one-pot synthesis, with a particular
emphasis on deriving realistic reaction mechanisms and kinetics. Figure 2.12
outlines a simple mechanism for the acid catalysed production of alkyl levulinate
from D-fructose and D-glucose in ethanol. Note that more detailed reaction
mechanisms are suggested and investigated in Chapters 5, 6 and 7.
46
Chapter 2
Figure 2.12 Simplified mechanism for the one-pot synthesis of alkyl levulinates in
an ethanol/acidic media.
The formation of ethyl levulinate from D-glucose as reactant is the rate limiting step
in terms of producing high yields of alkyl levulinates from lignocellulosic biomass.
This is the case as cellulose is made up of individual monomers of D-glucose linked
by glycosidic bonds (Fengel and Wegener 1983). Thus, to date the hydrolysis of Dglucose in the presence of ethanol has been challenging, owing to the necessity of
having to use high temperatures and corrosive mineral acids to achieve any
significant conversions. Alkylated glucosides are typically the main D-glucose
ethanolysis product and their transformation is known to be rate limiting in
achieving high yields of oxygenated hydrocarbons. Alkyl levulinate yields achieved
to date from D-glucose have been modest, for example Zhu et al. reported yields of
only 45% using 0.16 mol/L H2SO4 at 473 K (Zhu et al 2014). As a result of using
such harsh conditions the unwanted and explosive by-product dialkyl ether is formed
typically by ethanol solvent reacting with H2SO4 or other mineral acids. For
example, Peng et al. showed that during the methanolysis of D-glucose in the
presence of 0.1 mol/L H2SO4 at 473 K, up to 58% of methanol was lost in the form
47
Chapter 2
of dimethyl ether (Peng et al 2012), however the presence of ZSM-5 zeolites has
been reported to lower the formation of dialkyl ether to some extent (Xu et al 2013).
To date, the utilisation of solid acid catalysts for the ethanolysis of D-glucose has not
been as successful as in their employment for the alcoholysis of furfuryl alcohol or
for the esterification of levulinic acid. This is so, as the ethanolysis of D-glucose to
alkyl levulinates requires high temperatures. Therefore the utilisation of acidic
Amberlyst resins that have been successfully applied to hydrolyse other
monosaccharides (ie D-fructose) cannot be employed in the case of D-glucose,
because of their lower thermal stability at high temperatures. However, there has
been some reporting of heterogeneous catalysis for the ethanolysis of D-glucose, for
example Peng et al. using TiO2-SO3H achieved yields of 32 mol% of ethyl levulinate
from D-glucose (Peng et al. 2011). However the recent more pertinent literature
focuses on the promoting of the isomerisation between D-fructose and D-glucose.
As elaborated in Section 2.2, it envisioned that successful isomerisation of D-glucose
to D-fructose is key to the future feasibility for bio-refinery processes involving the
transformations of hexose sugars. As a consequence of this, there are several studies
published on the catalytic ethanolysis using D-fructose as a starting material in the
quest for optimising alkyl levulinate yields. For example, Liu et al. using various
types of p-styrene sulfonic acids grafted on nanotubes obtained levulinate yields of
80 mol% after 24 hours at 393 K (Liu et al 2013).
Saravanamurugan et al.
(Saravanamurugan et al 2011) evaluated the use of acid ionic liquids as catalysts
resulted in yields of 74 mol% obtained from D-fructose. However arguably the most
interesting study conducted to date regarding the ethanolysis of D-fructose was
conducted by Balakrishnan et al. (Balakrishnan et al 2012) who performed the
48
Chapter 2
alcoholysis reactions of D-fructose with ethanol and n-butanol using H2SO4, and
different acid resins such as Amberlyst 15, and Dowex 8 . The significance of this
study is that it produces other ethanolysis side products such as alkoxymethylfurfural
dialkylacetals and 1,1-dialkoxyethane and provides the first kinetic data of any
consequence in this topic area.
Perhaps somewhat surprisingly, the one pot ethanolysis of pentose sugars to ethyl
levulinate has been recently reported. Hu et al. reported the first direct synthesis of
alkyl levulinates (~20 mol% yield) from a five carbon carbohydrate, xylose (Hu et al
2013) . They showed that the co-presence of Amberlyst-70 and Pd/Al2O3 under H2
pressure allows hydrogenation of furfural resulting in furfuryl alcohol which
undergoes alcoholysis to form alkyl levulinates. The practicalities of this approach
present significant concerns because of the necessity of addition of external
hydrogen. Never the less the study provides food for further thought. More
information regarding the methods dicussed and catalysts are presented in Table 2.3.
49
Chapter 2
AL Yield
Catalyst
Conditions
Carbohydrate
H2SO4
(0.016
mol/L)
473 K, 2.5h
D-Glucose
Et, 45
(Zhu et al 2014)
H-USY(6),
60 wt%
433 K, 20h
D-Glucose
Me, 49
Et, 41
(Saravanamurugan et
al 2013)
H2SO4,
0.005 mol/L
373 K 4h
D-Glucose
Me, 50
(Peng et al 2012)
TiO2-SO3H,
2.5 wt%
473 K, 2h
D-Glucose
Me, 33
(Peng et al 2011)
H2SO4
10 mol%
393 K, 24h
D-Fructose
Et, 56
Bu, 64
(Balakrishnan et al
2012)
Amberlyst
15
10 mol%
383 K, 24h
D-Fructose
Et, 16
(Balakrishnan et al
2012)
H2SO4,
10 mol%
393 K, 30h
D-Fructose
Et, 56
Bu, 64
(Balakrishnan et al
2012)
(Mol%)
Reference
Sulfonic Acid
Et, 89
(Liu, et al 2013)
D-Fructose
Carbon Nano 393 K, 24h
tubes
Amberlyst428-448 K,
(Hu, Song, et al
70,
Et, 18
D-Xylose
2h
2013)
Pd/Al2O3
Table 2.3 Various methods and catalysts reported, for the one-pot production of
alkyl levulinates from monosaccharides.
From a more applied prospective it has also been demonstrated that alkyl levulinates
can be synthesised directly from lignocellulosic biomass. Garves was the first to
investigate the formation of methyl levulinate from biomass achieving yields of 1631 mol% using bagasse barley, waste paper and wheat meal (Garves 1988).
Following on from the methods developed by Garves, Olson et al. studied ethyl
levulinate production from waste chip board and plywood achieving yields of 31
mol% (Olson et al 2001). The highest yield to date from biomass was obtained by
50
Chapter 2
Chang et al. who achieved yields of 51 mol% using wheat straw as a feed-stock
(Chang et al 2012). Interestingly in a study conducted by Grisel et al. they found that
the delignification of wheat straw was found to have no effect on the yields of alkyl
levulinates (Grisel et al 2014). This is significant as lignin is said to significantly
hinder the hydrolysis of hexose sugars. Their findings suggest that a costly a pretreatment for lignin removal may not be required for the conversion of similar feedstocks to alkyl levulinates.
2.9. 5-Ethoxymethylfurfural as an advantaged fuel component
5-(ethoxymethyl)-furfural carboxaldehyde (5-ethoxymethylfurfural) is another
promising potential liquid fuel component that can be synthesised by reaction of
hexose carbohydrates with ethanol. In this Thesis Chapters 5 and 6 focus on its
formation from D-Fructose. It is part of the promising fuel group termed ‘furanics’
and may show considerable promise for a range of diesel and jet applications
(Dautzenberg and Gruter 2012). In comparison to ethyl levulinate, information
purporting to its use as a transportation fuel is less well reported and thus research
regarding the merits of its performance as a fuel is at an earlier stage of development.
Nevertheless, 5-ethoxymethylfurfural has attractive qualities as a fuel additive
having a density of 1.1 g/cm3 and a ∆H Combustion (MJ/L) that is 15%, 17% and 50%
higher than conventional gasoline, ethyl levulinate and ethanol respectively. Recent
combustion tests have also shown that 5-ethoxymethylfurfural performs well as a
diesel blend additive with its addition to diesel resulting in a significant reduction in
soot and SOX formation (Wang et al 2013). Additionally there is further opportunity
to improve the fuel properties of 5-ethoxymethylfurfural by hydrogenation further
increasing the carbon/hydrogen ratio and thus improving its properties as a fuel.
51
Chapter 2
Figure 2.13 5-Ethoxymethylfurfural a promising liquid fuel transport component.
2.9.1. Production pathways for 5-ethoxymethylfurfural
At present the most efficient chemical process for the synthesis of 5ethoxymethylfurfural is the method developed by Mascal and Nikitin (Mascal and
Nikitin 2008). This involves the production of 5-chloromethylfurfural from hexose
carbohydrates using 12 mol/L. For example 5-chloromethylfurfural can be produced
at yields of 71 mol% from cellulose. The aforementioned is then converted into 5ethoxymethylfurfural by reaction with ethanol in the presence of acid catalysis.
Although this process is highly efficient in terms of achieving high product yields,
there are corrosion issues with using such as concentration of HCl as well as the
potential to have chorine in the synthesis system. This is highly undesirable with
regard to environmental concerns but also in terms of process optimisation. As well
as this having chlorine in fuel combustion systems can give rise to the formation of
dioxins. Thus, at present the two most viable production routes for 5ethoxymethylfurfural can be categorised as:
1. The direct ethanolysis of 5-hydroxymethylfurfural.
2. The one-pot synthesis of 5-ethoxymethylfurfural from hexose carbohydrates.
52
Chapter 2
2.9.2. The ethanolysis of 5-hydroxymethylfurfural
5-Hydroxymethylfurfural has a reactive hydroxyl group and quite readily loses a
molecule of water to form 5-ethoxymethylfurfural using homogenous catalysts like
H2SO4 or strong acid sulfonic resins as depicted in Figure 2.14.
Figure 2.14 A suggested balanced reaction for the ethanolysis of 5hydroxymethlfurfural.
For example, both Lanzafame et al. and Balakrishnan et al. achieved almost 100%
conversion of 5-hydroxymethylfurfural to 5-ethoxymethylfurfural and ethyl
levulinate using H2SO4 and strong sulphonic resins such as Amberlyst-15.
Lanzafame et al. investigated the use of mesoporous materials such as ziroconia or
sulphated ziroconia over mesoporous silica (SMA-15) to good effect (Lanzafame et
al 2011), whilst Balakrishnan et al. found that sulfonic acid functionalized resins,
Amberlyst-15 and Dowex DR2030 showed exceptional reactivity and selectivity for
the aforementioned reactions (Balakrishnan et al 2012). Finally Che et al.
investigated the use of H4SiW12O40 /MCM-41 nanospheres as solid acid catalysts and
achieved yields of 81.2 mol% 5-ethoxymethylfurfural after 4 hours (Che et al 2012).
More details regarding pertinent catalysts and methods used are displayed in Table
2.4
53
Chapter 2
Catalyst
Conditions
5-EMF Yield (Mol%)
Reference
Z-SBA-15
413 K, 2h
76
(Lanzafame et al
2011)
H2SO4
(10 mol%)
413 K, 2h
3
(Lanzafame et al
2011)
MCM-41(50)
413 K, 2h
68
(Lanzafame et al
2011)
H4SiW12O40/MC
M-41
nanospheres
373K, 24h
81
(Che et al 2012)
Amberlyst 15
(5 mol%)
348 K, 25h
55
(Balakrishnan et al
2012)
DowexDR2030
348 K, 24h
57
(Balakrishnan et al
2012)
Fe3O4@C-SO3H
353 K, 12h
88
(Yuan et al 2015)
Table 2.4 Various catalysts used for the synthesis of 5-ethoxymethylfurfural (5EMF) from 5-hydroxymethylfurfural.
Finally Che et al. investigated the use of H4SiW12O40/MCM-41 nanospheres as solid
acid catalysts and achieved yields of 81.2 mol% 5-ethoxymethylfurfural after 4 hours
(Che et al 2012). More details regarding pertinent catalysts and methods used are
displayed in Table 2.4.
2.9.3. The
one-pot
synthesis
of
5-ethoxymethylfurfural
from
hexose
carbohydrates
In the long term it appears unlikely that the ethanolysis of 5-hydroxymethylfurfural
will be viable on a commercial scale. Obtaining 5-hydroxymethylfurfural as a
primary reactant is challenging due to its reactive nature and is a costly endeavour.
For the same reasons as outlined for the synthesis of ethyl levulinate from hexose
carbohydrates the one-pot synthesis approach, at present, is the most feasible
54
Chapter 2
technique for the formation of 5-ethoxymethylfurfural from lignocellulosic derived
carbohydrates.
Figure 2.15 General mechanism for the one-pot synthesis of 5-ethoxymethylfurfural
from hexose carbohydrates.
Like ethyl levulinate, the ethanolysis studies involving D-fructose are more common
than from D-glucose. Relatively high yields of 5-ethoxymethylfurfural can be
achieved by using conventional mineral acids. For instance Balakrishnan et al.
reported yields of 71 mol% using H2SO4 at 353 K after 24 hours. Liu et al. using
H2SO4 in a dimethyl sulfoxide/ethanol solvent achieved yields of 58 mol% at 413 K
(Liu et al 2015). A considerable drawback to such methods is that ethyl levulinate is
formed in significant qauntities as a by-product, which has a negative effect on
trying to optimise yields of the intented 5-ethoxymethylfurfural product in such
systems. As a result a wide range of hetergeneous catalysts have been employed with
the quest of optimising yields of 5-ethoxymethylfurfural and minmising the
formation of ethyl levulinate. To this effect Kraus and Guney using
methylimidazolebutylsulfate phosphotungstate ([MIMBS]3 PW12O40) as a catalyst
obtained 5-ethoxymethylfurfural yields of 90.5 mol% after 24 hours at 363 K (Kraus
55
Chapter 2
and Guney 2012). Whilst Wang et al. achieved yields of 71 mol% using graphene
oxide as catalyst. Numerous other studies have been conducted employing a range of
heterogeneous catalysts.
The formation of 5-ethoxymethylfurfural from D-glucose has been challenging. The
Bronsted acid catalysed ethanolysis of D-glucose typically results in the formation of
ethyl glucosides which at appropriate temperatures dehydrate three times to form
ethyl levulinate. Significantly 5-ethoxymethylfurfural is not a significant
intermediate in such systems. Therefore studies focusing on its formation from Dglucose have focused on promoting D-glucose/D-fructose isomerisation and then
dehydrating the resultant D-fructose. Lew et al. achieved this, using a Sn-Beta
Zeolite in tandem with Amberlyst-131 and detected 5-ethoxymethylfurfural yields of
31 mol% at 363 K from D-glucose (Lew et al 2012). Using a tandem Bronsted/Lewis
heterogeneous acid system Li et al. observed yields of 44 mol% at 369 K using a
combination of DeAl-H-beta + Amberlyst-15 as catalyst (Li et al 2015). However,
the most promising study of this kind using homogenous catalysis was conducted by
Liu et al. who achieved yields of 38 mol% from D-glucose at 373 K in a using the
Lewis acid AlCl3 (Liu et al 2013). A summary of all pertinent studies for the one-pot
synthesises of 5-ethoxymethylfurfural from hexose carbohydrates is presented in
Table 2.5. Similar to the formation of ethyl levulinate from hexoses, no kinetic data
regarding the one-pot synthesis of 5-ethoxymethyl from D-fructose has been
reported.
56
Chapter 2
5-EMF Yield
Catalyst
Conditions
Hexose
H2SO4
(10 mol%)
373 K, 24h
D-Fructose
70
(Balakrishnan et
al 2012)
Amberlyst-15
(10 mol%)
393 K, 24h
D-Fructose
71
(Balakrishnan et
al 2012)
H2SO4 (10 mol%)
Ethanol / DMSO
413 K, 8h
D-Fructose
58
(Liu et al 2015)
Graphene Oxide
403 K, 24h
D-Fructose
71
(Wang et al 2013)
([MIMBS]3
PW12O40)
363 K, 24h
D-Fructose
90
(Kraus and Guney
2012)
Sn-BEA Zeolite
Amberlyst-131
363 K, 16h
D-Glucose
33
(Lew et al 2012)
DeAl-H-beta
Amberlyst 15
369 K, 11h
D-Glucose
44
(Li et al 2015)
AlCl3
373 K, 11h
D-Glucose
38
(Liu et al 2013)
(Mol%)
Reference
Table 2.5 Literature summary of catalysts and methods for the one-pot synthesis of
5-ethoxymethylfurfural from hexose carbohydrates.
2.10.
Novelty of this Thesis
From all of the above it is clear that there is a significant gap in the literature
regarding mechanisms, kinetics and reaction pathway analysis for the conversion of
hexose carbohydrates to furanic derived fuel components such as ethyl levulinate
and 5-ethoxymethylfurfural in non-aqueous solvents. This critical information is
57
Chapter 2
absent even in the more frequently studied aqueous systems. As Chapter 1 outlines
this Thesis aims to help close this gap in the literature by carefully designing
experiments to:
1. Establish a global reaction mechanism for the hydrolysis of hexoses in
aqueous systems, and using this as a baseline to inform the more complicated
ethanolysis mechanism (Chapter 4).
2. Develop an understanding of the D-fructose ethanolysis mechanism and
deriving an approach for determining the feasibility of mechanistic
propositions.
3. Conduct temperature dependent kinetic modelling for the D-fructose
ethanolysis system.
4. Develop an understanding of the D-glucose ethanolysis mechanism and an
appreciation of the fuel properties from the resultant mixture.
The experimental regime required to do this is outlined in the next Chapter (Chapter
3).
2.11. References
Akien, G.R., Qi, L., Horváth, I.T. (2012) “Molecular mapping of the acid catalysed
dehydration of fructose,” Chemical Communications, 48(47), 5850–5852.
Angyal, S.J. (1969) “The composition and conformation of sugars in solution,”
Angewandte Chemie International Edition in English, 8(3), 157–166.
Assary, R.S., Kim, T., Low, J.J., Greeley, J., Curtiss, L.A. (2012) “Glucose and
fructose to platform chemicals: understanding the thermodynamic landscapes of
58
Chapter 2
acid-catalysed reactions using high-level ab initio methods,” Physical Chemistry
Chemical Physics, 14(48), 16603–16611.
Balakrishnan, M., Sacia, E.R., Bell, A.T. (2012) “Etherification and reductive
etherification of 5-(hydroxymethyl) furfural: 5-(alkoxymethyl) furfurals and 2, 5bis (alkoxymethyl) furans as potential bio-diesel candidates,” Green Chemistry,
14(6), 1626–1634.
Caratzoulas, S., Davis, M.E., Gorte, R.J., Gounder, R., Lobo, R.F., Nikolakis, V.,
Sandler, S.I., Snyder, M.A., Tsapatsis, M., Vlachos, D.G. (2014) “Challenges of
and insights into acid-catalyzed transformations of sugars,” The Journal of
Physical Chemistry C, 118(40), 22815–22833.
Caratzoulas, S., Vlachos, D.G. (2011) “Converting fructose to 5hydroxymethylfurfural: a quantum mechanics molecular mechanics study of the
mechanism and energetics,” Carbohydrate research, 346(5), 664–672.
Chang, C., Xu, G., Jiang, X. (2012) “Production of ethyl levulinate by direct
conversion of wheat straw in ethanol media,” Bioresource Technology, 121, 93–
99.
Che, P., Lu, F., Zhang, J., Huang, Y., Nie, X., Gao, J., Xu, J. (2012) “Catalytic
selective etherification of hydroxyl groups in 5-hydroxymethylfurfural over H 4
SiW 12 O 40/MCM-41 nanospheres for liquid fuel production,” Bioresource
Technology, 119, 433–436.
Chheda, J.N., Huber, G.W., Dumesic, J.A. (2007) “Liquid-phase catalytic processing
of biomass-derived oxygenated hydrocarbons to fuels and chemicals,”
Angewandte Chemie International Edition, 46(38), 7164–7183.
59
Chapter 2
Choudhary, V., Burnett, R.I., Vlachos, D.G., Sandler, S.I. (2012) “Dehydration of
glucose to 5-(hydroxymethyl) furfural and anhydroglucose: thermodynamic
insights,” The Journal of Physical Chemistry C, 116(8), 5116–5120.
Choudhary, V., Mushrif, S.H., Ho, C., Anderko, A., Nikolakis, V., Marinkovic, N.S.,
Frenkel, A.I., Sandler, S.I., Vlachos, D.G. (2013) “Insights into the interplay of
Lewis and Brønsted acid catalysts in glucose and fructose conversion to 5(hydroxymethyl) furfural and levulinic acid in aqueous media,” Journal of the
American Chemical Society, 135(10), 3997–4006.
Choudhary, V., Pinar, A.B., Lobo, R.F., Vlachos, D.G., Sandler, S.I. (2013)
“Comparison of homogeneous and heterogeneous catalysts for glucose-to-fructose
isomerization in aqueous media,” ChemSusChem, 6(12), 2369–2376.
Christensen, E., Williams, A., Paul, S., Burton, S., McCormick, R.L. (2011a)
“Properties and performance of levulinate esters as diesel blend components,”
Energy & Fuels, 25(11), 5422–5428.
Christensen, E., Yanowitz, J., Ratcliff, M., McCormick, R.L. (2011b) “Renewable
oxygenate blending effects on gasoline properties,” Energy & Fuels, 25(10),
4723–4733.
Dautzenberg, F.; Gruter, G. J. M. Method for the synthesis of 5-alkoxymethyl
furfural ethers and their use. Patent 8, 133, 289 B2, May, 2012. 2012.
Despax, S., Estrine, B., Hoffmann, N., Le Bras, J., Marinkovic, S., Muzart, J. (2013)
“Isomerization of D-glucose into D-fructose with a heterogeneous catalyst in
organic solvents,” Catalysis Communications, 39, 35–38.
60
Chapter 2
Fagan, P., Korovessi, E., Manzer, L., Mehta, R., Thomas, S. (2003) “Levulinic and
formic acid esters, as gasoline and diesel fuel oxygenate additives, prepared from
Biomass by Hydrolysis and esterification with olefins,” Patent WO2003085071.
Fan, C., Guan, H., Zhang, H., Wang, J., Wang, S., Wang, X. (2011) “Conversion of
fructose and glucose into 5-hydroxymethylfurfural catalyzed by a solid
heteropolyacid salt,” Biomass and Bioenergy, 35(7), 2659–2665.
Fengel, D., Wegener, G. (1983) Wood: Chemistry, Ultrastructure, Reactions, Walter
de Gruyter.
Fernandes, D., Rocha, A., Mai, E., Mota, C.J., Teixeira da Silva, V. (2012)
“Levulinic acid esterification with ethanol to ethyl levulinate production over
solid acid catalysts,” Applied Catalysis A: General, 425, 199–204.
Garves, K. (1988) “Acid catalyzed degradation of cellulose in alcohols,” Journal of
wood Chemistry and Technology, 8(1), 121–134.
Girisuta, B., Janssen, L., Heeres, H. (2006) “Green chemicals: A kinetic study on the
conversion of glucose to levulinic acid,” Chemical Engineering Research and
Design, 84(5), 339–349.
Grisel, R., van der Waal, J., de Jong, E., Huijgen, W. (2014) “Acid catalysed
alcoholysis of wheat straw: Towards second generation furan-derivatives,”
Catalysis Today, 223, 3–10.
Hayes, D.J., Fitzpatrick, S., Hayes, M.H., Ross, J.R. (2006) “The Biofine processProduction of levulinic acid, furfural, and formic acid from lignocellulosic
feedstocks,” Biorefineries-Industrial Processes and Product, 1, 139–164.
Hendriks, A., Zeeman, G. (2009) “Pretreatments to enhance the digestibility of
lignocellulosic biomass,” Bioresource Technology, 100(1), 10–18.
61
Chapter 2
Hu, X., Song, Y., Wu, L., Gholizadeh, M., Li, C.-Z. (2013) “One-pot synthesis of
levulinic acid/ester from C5 carbohydrates in a methanol medium,” ACS
Sustainable Chemistry & Engineering, 1(12), 1593–1599.
Hu, X., Wu, L., Wang, Y., Song, Y., Mourant, D., Gunawan, R., Gholizadeh, M., Li,
C.-Z. (2013) “Acid-catalyzed conversion of mono-and poly-sugars into platform
chemicals: Effects of molecular structure of sugar substrate,” Bioresource
Technology, 133, 469–474.
Joshi, H., Moser, B.R., Toler, J., Smith, W.F., Walker, T. (2011) “Ethyl levulinate: A
potential bio-based diluent for biodiesel which improves cold flow properties,”
Biomass and Bioenergy, 35(7), 3262–3266.
Joshi, S.S., Zodge, A.D., Pandare, K.V., Kulkarni, B.D. (2014) “Efficient conversion
of cellulose to levulinic acid by hydrothermal treatment using Zirconium dioxide
as a recyclable solid acid catalyst,” Industrial & Engineering Chemistry Research,
53(49), 18796–18805.
Khusnutdinov, R., Baiguzina, A., Smirnov, A., Mukminov, R., Dzhemilev, U.
(2007) “Furfuryl alcohol in synthesis of levulinic acid esters and difurylmethane
with Fe and Rh complexes,” Russian Journal of Applied Chemistry, 80(10), 1687–
1690.
Kimura, H., Nakahara, M., Matubayasi, N. (2011) “In situ kinetic study on
hydrothermal transformation of d-glucose into 5-hydroxymethylfurfural through
d-fructose with 13C NMR,” The Journal of Physical Chemistry A, 115(48),
14013–14021.
62
Chapter 2
Kimura, H., Nakahara, M., Matubayasi, N. (2013) “Solvent effect on pathways and
mechanisms for D-fructose conversion to 5-hydroxymethyl-2-furaldehyde: in situ
13C NMR study,” The Journal of Physical Chemistry A, 117(10), 2102–2113.
Koivisto, E., Ladommatos, N., Gold, M. (2015) “Compression ignition and exhaust
gasemissions of fuel molecules which can be produced from lignocellulosic
biomass: levulinates, valeric esters, and ketones,” Energy & Fuels, 29(9), 5875–
5884.
Kraus, G.A., Guney, T. (2012) “A direct synthesis of 5-alkoxymethylfurfural ethers
from fructose via sulfonic acid-functionalized ionic liquids,” Green Chemistry,
14(6), 1593–1596.
Kuo, C.-H., Poyraz, A.S., Jin, L., Meng, Y., Pahalagedara, L., Chen, S.-Y., Kriz,
D.A., Guild, C., Gudz, A., Suib, S.L. (2014) “Heterogeneous acidic TiO 2
nanoparticles for efficient conversion of biomass derived carbohydrates,” Green
Chemistry, 16(2), 785–791.
Lange, J.-P., van de Graaf, W.D., Haan, R.J. (2009) “Conversion of furfuryl alcohol
into ethyl levulinate using solid acid catalysts,” ChemSusChem, 2(5), 437–441.
Lanzafame, P., Temi, D., Perathoner, S., Centi, G., Macario, A., Aloise, A.,
Giordano, G. (2011) “Etherification of 5-hydroxymethyl-2-furfural (HMF) with
ethanol to biodiesel components using mesoporous solid acidic catalysts,”
Catalysis Today, 175(1), 435–441.
Lew, C.M., Rajabbeigi, N., Tsapatsis, M. (2012) “One-pot synthesis of 5(ethoxymethyl) furfural from glucose using Sn-BEA and Amberlyst catalysts,”
Industrial & Engineering Chemistry Research, 51(14), 5364–5366.
63
Chapter 2
Li, H., Saravanamurugan, S., Yang, S., Riisager, A. (2015) “Direct transformation of
carbohydrates to the biofuel 5-ethoxymethylfurfural by solid acid catalysts,”
Green Chemistry.
Liu, B., Zhang, Z., Huang, K., Fang, Z. (2013) “Efficient conversion of
carbohydrates into 5-ethoxymethylfurfural in ethanol catalyzed by AlCl 3,” Fuel,
113, 625–631.
Liu, J., Tang, Y., Fua, X. (2015) “Efficient conversion of carbohydratesto
ethoxymethylfurfural and levulinic acid ethyl ester under the catalysis of
recyclable DMSO/Brønsted acids,” Starch-Stärke.
Liu, R., Chen, J., Huang, X., Chen, L., Ma, L., Li, X. (2013) “Conversion of fructose
into 5-hydroxymethylfurfural and alkyl levulinates catalyzed by sulfonic acidfunctionalized carbon materials,” Green Chemistry, 15(10), 2895–2903.
Maldonado, G.M.G., Assary, R.S., Dumesic, J.A., Curtiss, L.A. (2012) “Acidcatalyzed conversion of furfuryl alcohol to ethyl levulinate in liquid ethanol,”
Energy & Environmental Science, 5(10), 8990–8997.
Mascal, M., Nikitin, E.B. (2008) “Direct, high-yield conversion of cellulose into
biofuel,” Angewandte Chemie, 120(41), 8042–8044.
Moliner, M., Román-Leshkov, Y., Davis, M.E. (2010) “Tin-containing zeolites are
highly active catalysts for the isomerization of glucose in water,” Proceedings of
the National Academy of Sciences, 107(14), 6164–6168.
Neves, P., Lima, S., Pillinger, M., Rocha, S.M., Rocha, J., Valente, A.A. (2013)
“Conversion of furfuryl alcohol to ethyl levulinate using porous aluminosilicate
acid catalysts,” Catalysis Today, 218, 76–84.
64
Chapter 2
Nikolla, E., Román-Leshkov, Y., Moliner, M., Davis, M.E. (2011) “‘One-pot’
synthesis of 5-(hydroxymethyl) furfural from carbohydrates using tin-beta
zeolite,” ACS Catalysis, 1(4), 408–410.
NIST (2016) 'NIST Standard Reference Database Number 69' [online], available:
http://webbook.nist.gov/chemistry/ [acessed 22 March 2016].
Olson, E.S., Kjelden, M.R., Schlag, A.J., Sharma, R.K. (2001) “Levulinate esters
from biomass wastes.”
Pagan-Torres, Y.J., Wang, T., Gallo, J.M.R., Shanks, B.H., Dumesic, J.A. (2012)
“Production of 5-hydroxymethylfurfural from glucose using a combination of
Lewis and Brønsted acid catalysts in water in a biphasic reactor with an
alkylphenol solvent,” ACS Catalysis, 2(6), 930–934.
Pasquale, G., Vázquez, P., Romanelli, G., Baronetti, G. (2012) “Catalytic upgrading
of levulinic acid to ethyl levulinate using reusable silica-included Wells-Dawson
heteropolyacid as catalyst,” Catalysis Communications, 18, 115–120.
Peng, L., Lin, L., Li, H. (2012) “Extremely low sulfuric acid catalyst system for
synthesis of methyl levulinate from glucose,” Industrial Crops and Products, 40,
136–144.
Peng, L., Lin, L., Zhang, J., Shi, J., Liu, S. (2011) “Solid acid catalyzed glucose
conversion to ethyl levulinate,” Applied Catalysis A: General, 397(1), 259–265.
Qi, L., Mui, Y.F., Lo, S.W., Lui, M.Y., Akien, G.R., Horváth, I.T. (2014) “Catalytic
conversion of fructose, glucose, and sucrose to 5-(hydroxymethyl) furfural and
levulinic and formic acids in gamma-valerolactone as a green solvent,” ACS
Catalysis, 4(5), 1470–1477.
65
Chapter 2
Sah, P.P., Ma, S.-Y. (1930) “Levulinic acid and its esters,” Journal of the American
Chemical Society, 52(12), 4880–4883.
Saravanamurugan, S., Nguyen Van Buu, O., Riisager, A. (2011) “Conversion of
mono-and disaccharides to ethyl levulinate and ethyl pyranoside with sulfonic
acid-functionalized ionic liquids,” ChemSusChem, 4(6), 723–726.
Saravanamurugan, S., Paniagua, M., Melero, J.A., Riisager, A. (2013) “Efficient
isomerization of glucose tofructose over zeolites in consecutive reactions in
alcohol and aqueous media,” Journal of the American Chemical Society, 135(14),
5246–5249.
Shimizu, K., Uozumi, R., Satsuma, A. (2009) “Enhanced production of
hydroxymethylfurfural from fructose with solid acid catalysts by simple water
removal methods,” Catalysis Communications, 10(14), 1849–1853.
Swift, T.D., Bagia, C., Choudhary, V., Peklaris, G., Nikolakis, V., Vlachos, D.G.
(2013) “Kinetics of homogeneous brønsted acid catalyzed fructose dehydration
and 5-hydroxymethyl furfural rehydration: A combined experimental and
computational study,” ACS Catalysis, 4(1), 259–267.
Tarabanko, V., Chernyak, M.Y., Aralova, S., Kuznetsov, B. (2002) “Kinetics of
levulinic acid formation from carbohydrates at moderate temperatures,” Reaction
Kinetics and Catalysis Letters, 75(1), 117–126.
Torres, A., Tsapatsis, M., Daoutidis, P. (2012) “Biomass to chemicals: Design of an
extractive-reaction process for the production of 5-hydroxymethylfurfural,”
Computers & Chemical Engineering, 42, 130–137.
66
Chapter 2
Tucker, M.H., Alamillo, R., Crisci, A.J., Gonzalez, G.M., Scott, S.L., Dumesic, J.A.
(2013) “Sustainable solvent systems for use in tandem carbohydrate dehydration
hydrogenation,” ACS Sustainable Chemistry & Engineering, 1(5), 554–560.
Wang, H., Deng, T., Wang, Y., Cui, X., Qi, Y., Mu, X., Hou, X., Zhu, Y. (2013)
“Graphene oxide as a facile acid catalyst for the one-pot conversion of
carbohydrates into 5-ethoxymethylfurfural,” Green Chemistry, 15(9), 2379–2383.
Wang, H., Deng, T., Wang, Y., Qi, Y., Hou, X., Zhu, Y. (2013) “Efficient catalytic
system for the conversion of fructose into 5-ethoxymethylfurfural,” Bioresource
Technology, 136, 394–400.
Wang, Z., Lei, T., Liu, L., Zhu, J., He, X., Li, Z. (2012) “Performance investigations
of a diesel engine using ethyl levulinate-diesel blends,” BioResources, 7(4), 5972–
5982.
Weingarten, R., Cho, J., Xing, R., Conner, W.C., Huber, G.W. (2012) “Kinetics and
reaction engineering of levulinic acid production from aqueous glucose
solutions,” ChemSusChem, 5(7), 1280–1290.
Windom, B.C., Lovestead, T.M., Mascal, M., Nikitin, E.B., Bruno, T.J. (2011)
“Advanced distillation curve analysis on ethyl levulinate as a diesel fuel
oxygenate and a hybrid biodiesel fuel,” Energy & Fuels, 25(4), 1878–1890.
Whistler, R.L., others (1964) “Methods in carbohydrate chemistry. Volume 4.,”
Methods in carbohydrate chemistry. Volume 4.
Xu, G.-Z., Chang, C., Zhu, W.-N., Li, B., Ma, X.-J., Du, F.-G. (2013) “A
comparative study on direct production of ethyl levulinate from glucose in ethanol
media catalysed by different acid catalysts,” Chemical Papers, 67(11), 1355–
1363.
67
Chapter 2
Yan, K., Wu, G., Wen, J., Chen, A. (2013) “One-step synthesis of mesoporous H 4
SiW 12 O 40-SiO 2 catalysts for the production of methyl and ethyl levulinate
biodiesel,” Catalysis Communications, 34, 58–63.
Yang, G., Pidko, E.A., Hensen, E.J. (2012) “Mechanism of brønsted acid-catalyzed
conversion of carbohydrates,” Journal of Catalysis, 295, 122–132.
Yang, L., Tsilomelekis, G., Caratzoulas, S., Vlachos, D.G. (2015) “Mechanism of
brønsted acid-catalyzed glucose dehydration,” ChemSusChem.
Yuan, Z., Zhang, Z., Zheng, J., Lin, J. (2015) “Efficient synthesis of promising
liquid fuels 5-ethoxymethylfurfural from carbohydrates,” Fuel, 150, 236–242.
Zhu, W., Chang, C., Ma, C., Du, F. (2014) “Kinetics of Glucose Ethanolysis
catalyzed by extremely low sulfuric acid in ethanol medium,” Chinese Journal of
Chemical Engineering, 22(2), 238–242.
68
Chapter 3
Chapter 3: Materials and Methods
To carry out the aims and objectives of this Thesis as outlined in Table 3.1 the
materials and methods section is broken into three core components:
1. The careful design of experiments in order to capture species concentrations
between reactants and products as a function of time, temperature and acid
concentration.
2. Employ best available analytical methods to quantify and qualify all reactant
species concentrations and pertinent variables in the reaction system
3. To utilise this information to decipher the validity of mechanistic
propositions and to derive kinetic parameters for all key reaction steps.
3. ddd
69
Chapter 3
Component
1. Experimental
Key features
Carefully designed experiments
Reflux reactions
Isobaric experiments with ethanol/acid/ hexose
carbohydrates conducted at condensed phase
conditions (Chapter 5)
Multiphase reactions
Biphasic experiments with ethanol/acid/ hexose
sugars and water/acid/systems (Chapter 4, 6 &7)
2. Analytical
Measuring concentrations
Gas chromatography
For the determination of analytes that can be
vaporised without destruction
Gas chromatography-mass
spectrometry
For the identification of analytes that cannot be
matched with known standards
Ion chromatography
For the quantification of carbohydrates and
carbohydrate derivatives
pH measurements
To determine the [H+] in the reaction media
(Chapter 5)
3. Computational
methods
Testing mechanistic propositions and deriving
kinetic parameters
Mass conserved
determination of rate
constants
Employed to test the validity of mechanistic
propositions
Determination of kinetic
parameters
Deriving activation energies and pre-exponential
factors for key reaction steps
Table 3.1 Summary of materials and methods employed in the construction of this
Thesis. A more detailed description is outlined below:
70
Chapter 3
3.1. Experimental procedure
In this Thesis two main experimental reactor setups are employed for the acid
catalysed transformations of lignocellulosic hexose carbohydrates: reflux reactions
and high pressure reactions.
3.1.1. Experimental configuration for experiments conducted under reflux
conditions
Chapter 5 employs a reflux reaction configuration where reactions are performed in a
20 cm3 spherical reactor at a prescribed temperature at atmospheric pressure (See
Figure 3.1). The reactor is heated by an external oil bath. The reaction temperature is
independently controlled and monitored using a thermocouple array (Stuart™ SCT1
temperature controller) and an in-situ magnetic propeller ensuring that the reaction
mixture is well mixed and homogeneous. Atmospheric pressure is regulated by
fitting the main reactor exit with an open-ended condensing unit (~ 277 K), thus
allowing the reaction to be at reflux. Typically the desired reaction temperature is
reached after 16 minutes from ambient temperature and the temperature is stable to
±1 K.
Figure 3.1 Reflux apparatus for condensed phase reaction synthesis (Chapter 5).
71
Chapter 3
3.1.2. Experimental configuration for high pressure experiments
In Chapters 4, 6 and 7 all experiments are carried out in glass pressure tubes (25.4
mm O.D. x10.2 cm), comprising of polytetraflouroethylene plugs with tetraflouro Orings for pressure sealing up to 1.03 MPa. A 5.0 ml aqueous solution containing the
desired concentration of reactant and catalyst is placed into the tubes. After being
sealed, each tube is placed for a defined period of time in an oil bath set at the
desired reaction temperature. When the reaction time is completed, each tube is
removed from the oil bath and immersed in a cold water bath to quench the reaction.
Samples are then prepared for analysis.
Figure 3.2 Apparatus for high pressure synthesis reactions.
3.2. Detection of analytes using ion chromatography
A high performance chromatography system (ICS-3000 Dionex) is used for the
detection and quantification of products obtained through the primary conversion of
lignocellulosic materials in the aqueous phase as illustrated in Figure 3.3. Two
different configurations are utilised for the chromatographic analyses carried out in
this Thesis and are shown in Figure 3.1.Technical information related to the
72
Chapter 3
equipment and columns presented in this section has been provided by Dionex
(Dionex 2004). The system comprised of the following components

Autosampler (As-50): Polypropylene vials containing 1.5 mL are placed
inside a tray in the device. After programing of the analysis conditions, an
automated arm with a sampling needle takes a defined amount of sample
(usually 10µl) and delivers it to the injection valve. The auto sampler also
includes a source of eluent
for deionising and degasing water for flushing
the sampling system and injection valve after the sample had been injected.

Dual Pump: Comprises of two double-headed pumps feeding the six-port
loading valve prior to the guard and analytical columns, and feeding the
three-way valve prior to the detectors, respectively. The primary head in the
pump passes eluent at the configured flow rate to the secondary pump head.
On the other hand, the secondary pump works as a reservoir to deliver the
eluent constantly to the system during the refill stroke in the primary head.
Figure 3.3 The ICS-3000 Dionex system at the University of Limerick.
73
Chapter 3
(a) Configuration employed for the detection of monosaccharides as adopted from
(Dionex 2004).
(b) Configuration employed for the detection of furanic compounds and organic
acids as adopted from (Dionex 2004).
Figure 3.4 Schematic diagram of the chromatographic system for the analysis of
monosaccharides and furanic compounds: 1) Sampling needle, 2) Sample syringe,
3) Vial tray, 4) Flush reservoir, 5) Injection valve, 6) Six-port loading valve, 7)
Guard column, 8) Analytical column, 9) Three way-valve, 10) Electrochemical
detector, 11) Pre-column pump, 12) Post-column pump, diode array detector.
74
Chapter 3
This detector component of the system contains a six-port loading valve, guard and
analytical columns, a three-way mixing valve, and an electrochemical detector. A
loop (10 µL) in the loading valve allowed the mixing of the sample with the eluent
and carries it to the guard and analytical columns. The temperature inside this
module is controlled through the PC programme of the system using Chromeleon®
software depending on the column and separation being conducted.
3.2.1. Electrochemical detector
An electrochemical detector is employed for the detection of carbohydrates and
compounds with carbohydrate related functional groups. It operates based on the
electrical current that is released by the oxidation of analytes at an anode electrode.
The electrochemical detector consists of an amperometric or flow-through cell that
contains a counter electrode, a gold working electrode and a reference electrode
(AgCl). The detection follows a pulsed amperometric mode in which different
potentials are applied to the cell. During the first potential, the analyte is oxidised on
the gold electrode by applying a certain electrochemical potential (0.1 V). The
charge in coulombs is measured by integrating the cell current over a definite time
during this first period. During the second period, a high negative potential (-2.0 V)
is applied for a short period of time in order to flush out any analyte oxidation
products followed by a higher potential ( 0.6 V), that oxidises the surface of the gold
electrode (Dionex 2003). After this a lower potential is applied for reducing the gold
surface and reconditioning the electrode for the next cycle. Figure 3.5 shows the
pulsed amperometric mode followed during each cycle in the electrochemical
analysis used for the detection and quantification of carbohydrates.
75
Chapter 3
2
Measurement
Reconditioning
Potinetial (V)
1
Intregration
0
-1
-2
0.0
0.1
0.2
0.3
0.4
0.5
Time / Seconds
Figure 3.5 Amperometric pulse for electrochemical detection of monosaccharides.
3.2.2. Diode array detector
A diode array detector is employed for the detection of furan compounds and organic
acids. This analytical device operates based on the spectrophotometric properties of
the analystes and comprises of a tungsten lamp (visible to near infrared range) and a
deuterium lamp (ultraviolet range) providing an operational wavelength range
between 190 and 800 nm. During the analysis, the eluent containing the sample will
flow through the cell irradiated by the light sources. The amount of analyte can be
determined by comparing the change in the intensity to that of the mobile phase.
3.2.3. Detection of carbohydrates and carbohydrate derived derivatives
For the seperation of D-fructose, D-galactose D-glucose, D-mannose D-xylose 5hydroxymethylfurfural,
ethyl
glucosides,
various
hexose
acid
catalysed
transformations species and ethylated carbohydrates as elaborated in Chapter 5, a
Dionex Carbopac PA1 column is employed. Using this analysis system (an anion
exchange guard (4-50 mm) and analytical column (4-250 mm) are connected in
76
Chapter 3
series as is illustrated in Figure 3.4. The Dionex Carbopac PA1 coulumn is made of a
10mm polystyrene / divinylbenzene substrate agglomerated into 350 nm particles of
latex and functionalised with quaternary amines (5%). The surface functionalisation
of the stationary phase in this column leads to the interaction of anions of the
analytes in the aqueous phase (neutral and basic pH) through ion exchange.
Hydroxyl groups found in carbohydrates are ionisable and as a consequence can be
irreversibly attached to the quaternary ammonium ions (-N+R3) of the column
(Dionex 2003). This column is heated to 291 K, and the analytes are eluted
isocratically using deionised water (18.2 M.cm) at a flow rate of 1.1 mL/min, and
detected using the electrochemical detector as mentioned earlier. The column is reconditioned using a gradient flow of 0.4 mol/L NaOH and 0.24 mol/L CH3COONa
for 3 minutes after each analysis.
Standard sugars solutions with a known concentration of the particular carbohydrates
of interest are used for calibration of the system. Exemplar chromatographs
employing ion chromatography in this Thesis are illustrated in Figure 3.6.
77
Chapter 3
600
Detector Response
500
400
300
200
100
(a)
Ethyl fructosides
(unknowns #1 & 2)
1.61 and 1.81 mins
unknown #3 & #4
2.11 and 2.41 mins
5-hydroxymethylfurfural
3.4 mins D-fructose 12.50 mins
Fructofuranose
D-mannose 10.4 mins
(unknown #5)
D-glucose 7.8 mins
15.27 mins
0
0
2
4
6
8
10
12
14
16
18
Time / Minutes
(b)
Unknown
#3 2.41
mins
400 Ethyl
glucoside
300 1.61 mins
Unknown #4 2.11 mins
Detector Response
500
D-glucose
Trace fructofuranose
D-fructose 12.8 mins
11.2 mins
200
100
0
0
2
4
6
8
10
12
14
16
18
Time / Minutes
(c)
5-HMF
35
Detector Response
30
D-Galactose
D-Mannose D-Fructose
25
20
15
D-Erythrose
10
5
Dihydroxyacetone
0
0
2
4
6
D- Glucose
8
10
12
14
Time / Minutes
Figure 3.6 Exemplar chromatographs employing the CarboPac PA1 column. a)
Exemplar chromatographs generated for the acid catalysed transformations of Dfructose in an ethanol media (Chapter 5&6). b) Exemplar chromatogram of mixed
carbohydrates in aqueous media (Chapter 4). c) Exemplar chromatogram of the acid
catalysed transformations of D-glucose in ethanol media (Chapter 7).
78
Chapter 3
3.2.4. Detection of furanic compounds and organic acids
For the determination of furanics and organic acids, chromatographic separation is
achieved using a Dionex Acclaim Organic acid guard (5 m, 410 mm) and analytical
(5 m, 4 250 mm) columns connected in series following the configuration shown in
Figure 3.4(b). The stationary phase in this column is a reverse-phase silica material
(17% carbon content) which due to its surface functionalisation, facilitates the
elution of hydrophilic aliphatic and aromatic organic acids. Isocratic separation is
carried out using 100 mM Na2SO4 at a pH of 2.65 (0.55 mL of methanesulfonic acid
per every 1 L of solution) at a flow rate of 0.8 mL/min with a temperature of 303 K
in the DC module. The eluted analytes are monitored using the DAD (Diode Array
Detector)-3000RS at different wavelengths, i.e. 210 nm for organic acids and 280
nm for 5-hydroxymethylfurfural and furfural. The concentration of each compound
in the liquid phase is determined using calibration curves obtained by analysing
standard solutions with known concentrations. Due to the higher affinity of the
stationary phase with furfural and 5-hydroxymethylfurfural, the total time required
for the analysis of each sample is 30 minutes. More information regarding the
detector response of pertinent furans and organic acids is available in Chapter 4.
Figure 3.4 provides a description of typical chromatographs obtained from the
detection of organic acids and furanic compounds as detected in this Thesis at 210
and 280 nm employing the above described technique.
3.3. Detecting analytes using gas chromatography
Gas chromatography (GC, Agilent Technologies 7820 A GC system) is employed to
detect compounds that can be vaporised without decomposition. A gas
chromatograph (GC) is an analytical instrument that separates components of a
79
Chapter 3
mixture in a sample (Gordon 1990) When employing gas chromatography a sample
is injected into a sampling port becoming vaporised and enters a gas stream which
transports the sample into a separation tube known as the “column”. Helium or
nitrogen is employed as the carrier gas. Depending on the column used the sample is
then separated into its various sub components by the particular column involved. A
flame ionisation detector is then employed to detect and quantify the amount of an
individual analyte present in the sample. Figure 3.7 illustrates the GC device used at
the University of Limerick. A detailed description of the major components of the
system is described below.
Figure 3.7 The Agilent 7820 A GC system at the University of Limerick.
3.3.1. Sample preparation
For GC analysis, a known mass (50 ± 5 mg ) of analyte is extracted from the
reaction media into 0.4 g of room temperature acetone and 0.8 g of 0.16 mg/g noctanol in acetone, this is preceded by the neutralisation of any remaining acid by the
addition of 50 mg of NaHCO3. This dilution and cooling procedure ensures that the
chemical reaction is effectively quenched. This sample is then filtered through 13
mm thick, 2 µm pore size syringe filters (Acrodisc) to remove any insoluble humic
80
Chapter 3
substances that may have been formed, and 1µl of the resulting solution is injected
into the sample inlet port of the GC.
3.3.2. Sample injection
A solvent is chosen that adequately dissolves the analytes of interest so that they can
be analysed. Typically 1-10µl of the sample of interest is injected, using an injection
syringe. The sample is injected in the inlet port of the GC. The syringe needle passes
through a thick rubber disc called a septum which re-seals itself again when the
syringe is pulled out. The sample inlet needs to be heated to a temperature that
ensures that each component of the sample is vapourised and is carried into the
column in gaseous form by the carrier gas typically nitrogen or helium.
3.3.3. Column employed
The column employed in Gas Chromatography is of critical importance in terms of
achieving adequate separation of the desired analyte (Grob and Barry 2004). When
configuring a separation method it is important to know the boiling point of the
analytes in a reaction sample. The lower the boiling point is the shorter the retention
time usually is because the compound will spent more time in the mobile gas phase.
This is why low boiling point solvents (acetone, ethyl acetate) are used as solvents to
dissolve the sample. Typically the column oven is configured to increase in
temperature at fixed intervals exploiting the different boiling points of solutes in a
specific sample. Polarity can also be exploited as a separation mechanism and it is
important to consider the polarity of the analyte versus the polarity of the column. In
general if the polarity phase of the analyte and the stationary phase of the column are
similar the retention time increases because the analyte interacts stronger with the
stationary phase. As a result, polar compounds have longer retention times on polar
81
Chapter 3
stationary phases and shorter retention times on non-polar columns at the same
temperature. For the detection of ethanol, diethyl ether, 5-ethoxymethylfurfural and
ethyl levulinate a polar phase, Restek™ crossbond polyethylene (30m, 0.25mm,
0.25µm, 30m) is employed. For the identification of compounds that could not be
matched to analytical standards, the GC is connected to a mass spectrometer (as
elaborated later) where a non-polar HP-5MS column (30 m, 0.25 mm ID, 0.25 µm)
is employed. For the aforementioned substances the injection port is maintained at
523 K, a temperature sufficiently high to ensure the full vaporisation of the expected
reaction components. A temperature program of 313 K increasing to 493 K at a rate
of 20 K per minute, remaining isothermal at 493 K for 5 minutes is found to achieve
adequate separation of these species. See Figure 3.8 for a sample GC chromatogram
for a representative ethanolysis test mixture.
8000
7000
Solvent
Ethanol
Detector Response
6000
5000
4000
Diethyl
Ether
3000
5-Ethoxymethylfurfural
2000
Ethyl Levulinate
Octanol
1000
0
0
2
4
6
8
10
12
Time / Minutes
Figure 3.8 Exemplar chromatogram illustrating species separation that are present in
a typical ethanolysis reaction mixture.
82
Chapter 3
3.3.4. Flame ionisation detector (FID)
In this Thesis flame ionisation detection (FID) is the major type of detection
employed. A FID normally consists of a hydrogen (H2)/air flame and a collector
plate. The carrier gas containing the sample post column passes through the flame,
which breaks down organic molecules and results in formation of CHO+ ions
(Agilient 2005). The ions produced are gathered on an electrode and they produce an
electrical signal. The current across this collector is thus proportional to the rate of
ionisation which in turn depends upon the concentration of a particular analyte in the
sample gas. Figure 3.9 depicts an illustration of a conventional FID detector. A
limitation of the FID detector is that it is limited to species containing carbon +
hydrogen and cannot detect species such as H2, O2 and NH3.
Figure 3.9 Schematic of a flame ionisation detector (FID) as extracted from
(Agilient 2011).
83
Chapter 3
3.3.5. Quantification of sample species
For the quantification of an individual analyte, species are identified by matching
retention-times to known standards, and quantified by calibration of detector
response to known concentrations (using n-octanol as internal standard). See Figure
3.10 for a sample FID response to diethyl ether, 5-ethoxymethylfurfural and ethyl
levulinate.
[Area Analyte] / [Area Internal Standard]
3.5
Diethyl Ether
Ethyl Levulinate
5-Ethoxymethylfurfural
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
1
2
3
4
5
6
7
[Analyte] / [Internal Standard]
Figure 3.10 FID detector response for ethanolysis mixture species of interest.
3.3.6. Gas chromatography-mass spectrometry
When the composition of a particular ethanolysis or other reaction sample is
unknown mass spectrometry is employed using an Agilient 5975C MSD (GC-MS),
which uses a HP-5MS column (30 m, 0.25 mm ID, 0.25 µm).
84
Chapter 3
Figure 3.11 The gas chromatography mass spectrometer at the University of
Limerick.
It is also employed to verify the identity of a particular peak at a fixed retention time
that cannot be matched by retention time to known compounds.
In this Thesis numerous separation programs are employed to identify specific
analytes and a complete description of each separation method used while compiling
this Thesis is unnecessary, however the technique for doing so is described in
Section 3.3.3
Mass spectrometry is a sensitive technique used to detect, identify and quantitate
molecules based on their mass to charge (m/z) (Hites 1997).
The three main
components of a mass spectrometer include an ion source, a mass analyser and a
digital detector.
Figure 3.12 Features of a mass spectrometer.
Firstly the sample is ionised, resulting in the molecule becoming charged allowing
the mass spectrometer to accelerate the ions throughout the remainder of the system.
85
Chapter 3
The ions subsequently encounter electrical and / or magnetic fields from the mass
analyser which deflect the paths of individual ions based on their mass to charge
ratio.
The Agilient 5975CMSD at the University of Limerick is equipped with a quadruple
detector. Combined electric potentials on the quadrople rods are set to pass only a
selected mass to charge ratio. All other ions that do not have a stable trajectory
through the quadruple mass analyser collide with the quadruple rods (Agilient 2009).
Figure 3.13 illustrates a schematic with the main features of a quadruple mass
spectrometer.
The entire process is performed under an extreme vacuum (10-6 to 10-8 torr) to
remove contaminating non-sample ions, which can collide with sample ions and alter
their paths or produce non-specific reaction products.
Figure 3.13 Schematic of a quadrupole mass spectrometer detector that is a feature
of the Agilient 5975C MSD at the University of Limerick extracted from (Gates et al
2016).
86
Chapter 3
The mass spectrometer is connected to a computer with Agilient software containing
the NIST database of mass spectra for most known chemical compounds.
155
30000
HO
109
25000
O
O
O
Intensity
20000
5-Hydroxymethylfurfural
15000
diacetal
97
M/Z 201
10000
5000
201
0
50
100
150
200
M/Z
Figure 3.14 Mass spectrum of 5-hdroxymethylfurfural diacetal.
Once the sample has been put through the GC-MS system, the spectrum can be
analysed so that each species in the sample can be identified based on the mass/
charge ratio and relative abundance of each ion. The mass spectrum obtained is then
scanned through the NIST database to identify the individual compound. Figure 3.14
presents a sample of a mass spectrum generated in this Thesis for a previously
unknown compound in an ethanolysis reaction mixture.
3.4. Estimation of hydrogen cations in solution
To estimate the amount of hydrogen cations [H+] in solution, the pH of the reaction
samples is determined (Chapter 5). For this purpose an Orion pH Ag/AgCl glass
electrode fitted to a VWR Symphony SB70P pH meter is employed. For pH
measurements, 0.4 g samples taken from the reaction media are diluted with 10 g of
deionised water. The pH meter is calibrated against buffer solutions (VWR
32032.291) of known [H+]. The pH of samples is variable with reaction time and
87
Chapter 3
condition but is always in the range of 1.7-2.4. The pH measurements are converted
to [H+].
As the pH measurements are acquired following dilution and at room temperature it
is necessary to account for the change in equilibrium that occurs .To evaluate the
effect of this the following procedure is employed. First, the measured [H+] is used
to calculate the concentration of sulphuric acid (SA) by solving a system of
equations shown below (3.3).
C H  C SO2  K a C HSO  0
4
4
C SA, 0  C HSO  C SO2  0
4
4
(3.3)
2C SA, 0  C HSO  C H   0
4
Then the H2SO4 concentration is multiplied by the dilution factor and thus converted
to the original [H2SO4] (ie the amount weighted out), is used to calculate [H+] at
reaction temperature by solving the same system of equations. H2SO4 is assumed to
disassociate fully to H+ and the bisulphate ion (Que et al 2011). Therefore, the
restrictions for the system of equations (3.3) originate from the equilibrium state of
the bisulphate ion and mass balances of sulphur and hydrogen: where CSA,0 is the
original sulphuric acid concentration (mol/L). The temperature dependence of the
dissociation constant, Ka, for the bisulphate ion is as is derived by Que et al.:
ln K a  5.393  1733.1/ T
(3.4)
where T is temperature in Kelvin.
Despite the small differences between dissociation in water and ethanol, the
procedure gave a good correlation between the calculation and real solutions. For
example the calculated H2SO4 concentrations used in Chapter 5 are determined to be
2, 6 and 12% lower than the prepared solutions of 0.098, 0.230 and 0.32 mol/L,
88
Chapter 3
respectively. Thus it can be concluded that the above described method is adequate
for estimating [H+] where robust thermodynamic information for H2SO4 dissociation
in ethanol is unavailable. More detailed information regarding this method and
H2SO4 dissociation in ethanol is discussed in Chapter 5.
3.5. Quantification of black insoluble materials (humins)
The formation of black insoluble materials known as humins are said to be rate
limiting in terms of effectively hydrolysing hexose sugars. In this Thesis humins are
estimated at the completion of each reaction, when the reaction mixture is filtered
through glass fibre paper (Whatman, grade GF/B 2.7 µm). The filter paper is
subsequently washed with ethanol and placed in an oven for 24 hours at 378 K. The
mass of the material remaining on the paper is determined by difference and referred
to as “humins”.
3.6. Computational methods employed
3.6.1. Rates, rate constants and reaction kinetics
When separate chemical compounds are present under particular conditions, reaction
kinetic describes the rates by which reactions take place and the rates of each
individual reaction (Seoud and Abdallah 2010). The most important kinetic
parameter is the rate constant. It describes the velocity of a reaction and the
concentration of chemical reactants (Green et al 2008). The rate constant can be
defined as the rate of concentration of a substance involved in the reaction with a
plus or minus sign depending on whether the chemical species is a reaction or a
product. The rate constant is the constant of proportionality “k” in the rate equation
as expressed by equation 3.5
89
Chapter 3
Rate = k×[CA]a×[CB]b
(3.5)
Where the rate is defined in mol cm-3 s-1, k is the rate constant CA and CB are
concentrations of A and B in mol/L. a and b describe the order of reaction with
respect to A and B respectively. Accurately determining the value of the rate
constant is of critical importance to ascertain the rate of any reaction applying its rate
equation. It has to noted that the rate constant is sensitive to environmental factors
and catalyst type.
The relationship between the rate constant “k” and temperature is defined using the
Arrhenius equation:
k  Ae
-
Ea
RT
(3.6)
where A is the frequency factor (s-1) although the units are dependent on reaction
order. Ea is the activation energy in kJ/mol, R the gas constant and T is the absolute
temperature in K.
3.6.2. Reaction mechanism
The reaction mechanism describes the process in which the reactants are converted
to the desired products through one or more intermediate stages and consecutive
reactions (Green et al 2008). Deriving plausible reaction mechanisms for the acid
catalysed transformations of hexose carbohydrates permeates throughout this Thesis
and is particularly pertinent to Chapters 4 and 5. The exact reaction mechanism is
often unknown which is the limiting step in configuring accurate reaction kinetics.
Typically the dynamics of the elementary reactions involved in the reaction
mechanism can be mathematically expressed by defining a set of coupled differential
equations that can be integrated simultaneously in order to derive the reactants
90
Chapter 3
concentrations for the entire system. However in most cases this cannot be achieved
by pen and paper and rate constants are typically derived by employing optimisation
techniques using powerful numerical tools such as Matalab or Fortran (Seoud and
Abdallah 2010).
3.6.3. Methods for estimating rate constants
In this Thesis rate constants are estimated with the aid of a Matlab computer
program. This is achieved using the optimization function ‘‘Fminsearch’’ on
Matblab. Fminsearch attempts to find a minimum of a scalar function of several
variables, starting at an initial estimate (Mathworks 2016a). This is generally
referred to as unconstrained nonlinear optimisation. Its aim is find a solution to a
problem specified by:
Min f (x)
(3.7)
Where f (x) is a function that returns a scalar, therefore the experimental data under
interrogation must be organised to comply with the objective function. When using
this method for estimating rate constants, optimization is the third step employed.
Firstly the experimental data has to generated, and secondly its needs to be
deciphered in terms of what is to be optimised. The following approach describes the
method that is employed to estimate rate constants in Chapter 5.
1. A Matlab program is designed with an objective file (ode 45 which is a
differential equation solver), a file with the postulated reaction mechanism,
and a code file integrating the two involving the reaction data. Note the
reaction data is organised in matrix form, and the dimensions of the
presentation of the data and matlab code must be consistent.
91
Chapter 3
2. A reaction mechanism for the system is postulated and defined. This is the
main variable in the method and one can expect it to be altered on numerous
occasions to find the best solution.
3. Initial guesses of the rate constants are defined along with simulation start,
end and interval times.
4. The simulation results are compared with experimental results at each
measured point. The difference between experimental and calculated values
are summed and squared to form an objective function F: The summation
starts from the initial time and ends at the final time.
F= ∑ (exp.conc –calc.conc.)2
(3.8)
5. Then the defined reaction rates are varied in all directions. For example
increasing k1 and k2 decreasing k1 and increasing k2 etc. For each case
simulated concentration profiles and new values for the objective function are
calculated. The rate constants corresponding to a minimum function are
stored and considered improved rate constants for final or next iteration.
6. The iteration proceeds until the absolute difference between two successive
functions is less than a predefined tolerance. These rate constant values are
then stored.
7. Various versions of the reaction mechanism are simulated until an optimum
solution is derived. This method employed is summarised in Figure 3.15.
For the determination of kinetic parameters in Chapters 6 and 7 a similar
minimisation technique is conducted using the lsqcurvefit function on Matlab based
on the Levenberg–Marquardt algorithm. This is also based on nonlinear curvefitting (data-fitting) problems in the least-squares sense. Here a slightly different
92
Chapter 3
approach is applied, the main difference being that the activation energy (Ea, kJ/mol)
and pre-exponential factor (A, min-1), Ea are the term being minimised using a
modified Arrhenius equation of the form:
ki  Ae

Ea  1
1 
 

R  T Tmean 
(3.9)
Where T is the temperature (K), and Tmean is the mean temperature in which the
experiments under interrogation are conducted at.
Figure 3.15 Flow chart of the method for estimation of rate constants and
mechanism evaluation employed in Chapter 5.
3.6.4. Evaluation of kinetic parameters and kinetic model operation
To monitor the accuracy of the proposed kinetic parameters a variety of approaches
are applied, particularly in Chapters 6 and 7.
Firstly a sensitivity analysis is
93
Chapter 3
conducted for each kinetic parameter derived, to ensure that each solution is derived
at a local minimum. If the solution has not reached a local minimum then the
solution derived is not optimum for that particular data set. The sensitivity analysis is
also useful for calculating the uncertainty of kinetic parameters. The broader the
peak of the curve in the sensitivity analysis, the greater the uncertainty of a specific
parameter. As well as this the estimated uncertainty is numerically calculated (±) at
95% confidence intervals.
Pre-exponential Value 5-Hydroxymethylfurfural - 5-Ethoxymethylfurfural
Objective Function Value
Objective Function Value
Estimated Value
1.2
0.6
0.0
0.0
0.1
0.2
Pre-exponential Factor
Figure 3.16 Example of sensitivity analysis conducted for the estimation of the preexponential factor in this Thesis.
The degree to which the kinetic model replicates the experimental data is monitored
by using the normalized root mean square error (R2), calculating the mechanism
fidelity index (Chapter 5), and by correlation matrices and residuals. Once the
optimum parameters are derived, they are substituted into the kinetic model to
predict time resolved species concentrations. The kinetic model predictions are then
validated against experimental data that is not used for model training to test the
degree to which the model can be extrapolated to other test conditions.
94
Chapter 3
3.7. References
Agilient 2011. Flame ionisation detector [online] http://www.chemagilent.com/contents.php?id=1001675 [accessed 11 January 2016].
Agilient 2009. Operation manual, Agilient 7000series triple quad GC/MS: Operation
manual.
Agilient 2005. Techical Note. A guide to interpreting detector specifications for gas
chromatography.
Dionex 2004. Technical Note 20. Analysis of carbohydrates by high performance
anion-exchange chromatography with pulsed amperometric detection.
Dionex 2003. Application note 92: The determination of sugars in molasses by high
performance anion exchange by pulsed amperometric detection.
Gates, P (2016) ‘‘Gas Chromatography mass spectrometry (GC/MS) ’’ available
[online] http://www.bris.ac.uk/nerclsmsf/techniques/gcms.html [accessed 11
January 2016].
Gordon, M.H. (1990) “Principles of gas chromatography,” in Principles and
Applications of Gas Chromatography in Food Analysis, Springer, 11–58.
Green, D.W., others (2008) Perry’s Chemical Engineers’ Handbook, McGraw-hill
New York.
Grob, R.L., Barry, E.F. (2004) Modern Practice of Gas Chromatography, John
Wiley & Sons.
Hites, R.A. (1997) “Gas chromatography mass spectrometry,” Handbook of
instrumental techniques for analytical chemistry, 609–626.
95
Chapter 3
Mathworks 2016a, fminsearch [online]
http://uk.mathworks.com/help/matlab/ref/fminsearch.html [accessed 11 January
2011].
Que, H., Song, Y., Chen, C.-C. (2011) “Thermodynamic modeling of the sulfuric
acid-water- sulfur trioxide system with the symmetric electrolyte NRTL model,”
Journal of Chemical & Engineering Data, 56(4), 963–977.
Sigma Aldrich 2016, Sigma Aldrich Hydranal Reagents [online]
http://www.sigmaaldrich.com/analytical-chromatography/titration/hydranal.html
[accessed 11 January 2016].
Seoud, A.-L.A., Abdallah, L.A. (2010) “Two optimization methods to determine the
rate constants of a complex chemical reaction using FORTRAN and MATLAB,”
American Journal of Applied Sciences, 7(4), 509.
96
Chapter 4
Chapter 4: Non-Stoichiometric Formation of
Formic and Levulinic Acids from the
Hydrolysis of Biomass Derived Hexose
Carbohydrates
This Chapter sets the basis for the mechanistic understanding of the acid catalysed
transformation of hexose carbohydrates in aqueous systems. It focuses on the
formation of formic and levulinic acids from hexose sugars, placing a particular
emphasis on the formic-levulinic acid as a diagnostic for interrogating the hexose
carbohydrate acid hydrolysis reaction mechanism.
It has been published in RSC Advances:
4. dfgdf
Flannelly T., Lopes M., Kupiainen L., Dooley, S. and Leahy J.J., 2015. NonStoichiometric Formation of Formic and Levulinic Acids from the Hydrolysis of
Biomass Derived Hexose Carbohydrates. RSC Advances. 2016, 6, 5797-5804.
DOI: 10.1039/C5RA25172A
97
Chapter 4
4.1. Abstract
This study challenges the assumption often postulated in the literature regarding the
stoichiometric formation of formic and levulinic acids from the acid hydrolysis of
hexose carbohydrates. Acid hydrolysis experiments are conducted with 2.5 wt%
H2SO4 in aqueous media with a series of reactants relevant to the hydrolysis systems
of
hexoses;
D-fructose,
D-galactose,
D-glucose,
D-mannose,
5-
hydroxymethylfurfural, D-erythrose, levulinic acid, furfuryl alcohol, furfural,
dihydroxyacetone, glyceraldehyde, pyruvaldehyde and formic acid at 423 K. We
show that the hydrolysis of 5-hydroxymethylfurfural, which is the main intermediate
between hexose carbohydrates and levulinic acid does result in the stoichiometric
formation of formic and levulinic acids. However, in all cases with hexose
carbohydrates as reactant, formic acid is observed in excess fractions to levulinic
acid, implying the common assumption inaccurate. At steady-state conversions of
the reactant, the formic and levulinic acid ratio for D-fructose, D-glucose, D-mannose
and D-galactose is shown to be 1.08 ±0.04, 1.15 ±0.05, 1.20 ±0.10 and 1.19 ±0.04
respectively. Combining this work and pertinent literature suggests there are at least
four potential pathways depending on reaction condition responsible for the excess
formic acid; through furfuryl alcohol and furfural formation and through the
transformation of D-erythrose and pyruvaldehyde
98
Chapter 4
4.2. Introduction
Levulinic acid is a bio-based platform chemical formed by the treatment of hexose
carbohydrates from lignocellulosic biomass, and is a precursor for the production of
potential future fuels and chemicals. Levulinic acid possesses the versatile ketone
and carboxylic acid functional groups, which has led to the US Department of
Energy recognising it as one of the top ten most attractive value-added chemicals
obtainable from biomass (Werpy et al 2004; Bozell and Petersen 2010). Such
valuable chemicals include γ-valerolactone (Braden et al 2011) and levulinate esters,
(Pasquale et al 2012) among other appealing chemicals such as angelica lactone,
diphenolic acid, and δ-amino levulinic acid (Bozell et al 2000). Acid hydrolysis is
presently the prevalent approach for levulinic acid generation from cellulosic
materials. Thus there is considerable work being conducted to comprehend the
mechanistic details, in order to maximise yields of levulinic acid formation from
lignocellulosic derived cellulose and hemicelluloses (Dussan et al 2013; Girisuta et
al 2013). Typical processes for levulinic acid formation employ high temperatures
(423-453 K), (purportedly) and various acids as catalyst, where yields of up to 70
mol% have been achieved with 1-5 wt% sulphuric acid (Rackemann and Doherty
2011).
Less attention is given to the formic acid that is produced as a reaction (by-) product
with levulinic acid. Formic acid is a valuable product in its own right and can be
used as a commodity in the chemical and textiles industry, as a catalyst, a hydrogen
carrier and a road salting component (Mukherjee et al 2015). In particular, the
capability of formic acid as a hydrogen carrier is appealing; therefore attempts to
optimize this reaction are being aggressively pursued by employing a variety of
homogenous and heterogeneous catalysts (Zacharska et al 2015). The original
99
Chapter 4
Biofine process (Fitzpatrick 1990) suggested that formic and levulinic acids are
formed stoichiometrically from lignocellulosic biomass and the potential formic acid
formed as side products is largely ignored (See Figure 4.1).
Figure 4.1 The historical understanding of the stoichiometric formation of formic
and
levulinic
acids,
from
cellulosic
and
hemicellulosic
derived
hexose
carbohydrates.
This historical assumption still prevails in the literature with many reports of the
stoichiometric formation of formic and levulinic acids from cellulose and cellulosic
derived hexoses (Joshi et al 2014; Shen and Wyman 2012). Concentration ratios of
unity are frequently stated, in some cases it is unclear whether the formic acid
concentrations reported are measured, or merely assumed following the historical
appraisal. It is worth noting that formic acid can undergo decomposition at high
temperatures leading to uncertainties about the exact amount of formic acid formed.
This is apparent in the work of Zhang et al. (Zhang et al 2015) where the authors
state that formic acid degraded to H2O and CO2 at 453 K, leading to levulinic acid
100
Chapter 4
concentrations in excess of formic acid concentrations (a sub-unity ratio). Recently
there has been other reports of non-equimolar ratios of formic and levulinic acids
from the hydrolysis of hexose starting materials at steady-state. For example in 2013
Swift et al.(Swift et al 2013) claimed to report for the first time of the formation of
non-stoichiometric ratios of formic to levulinic acids from the dehydration of Dfructose. Other recent studies have also reported non stoichiometric ratios using Dfructose, D-glucose and cellulose as reactants, with both Qi et al. and Kumar et al.
reporting ratios in excess of 1. Whilst the formation and consumption of formic acid
from hexose hydrolysis is poorly understood, it is clear that formic and levulinic
acids are formed stoichiometrically from 5-hydroxymethylfurfural as reactant
(Girisuta et al 2006; Swift et al 2013). This common assumption and apparent
confusion may be due to not treating the hydrolysis of hexoses and the subsequent
hydrolysis of 5-hydroxymethylfurfural as separate identities. A library of formic-tolevulinic acid ratios reported in the literature from various reactants is summarized in
Table 4.1.
101
Chapter 4
Reactant
Temperature
(K)
FA/ LA
Ratio
Reference
5-HMF
371-454
1
Girisuta et al.
5-HMF
343-423
H2O, HCl
1
Swift et al.
D-Fructose
403
H2SO4
GVL solvent
>1
Qi et al.
D-Fructose
403
H2SO4
GVL solvent
>1
Qi et al.
D-Fructose
371–454
HCl
>1
Swift et al
D-Fructose
513
H3PO4
>1
Asghari et al.
D-Fructose
403
H2SO4
GVL solvent
>1
Qi et al.
D-Glucose
363
HCl and
ZnBr2
1.46
Kumar et al.
D-Glucose
453
Maleic acid,
AlCl3
<1
Zhang et al.
D-Glucose
413
HCl
1.10
Yang et al.
D-Glucose
443
Mineral acids
1.20
Rackemann
and Doherty
Cellulose
453
ZnO2
>1
Joshi et al.
Cellulose
473 – 433
HCl
1
Shen and
Wyman
Catalyst
H2SO4
Table 4.1 Literature overview of formic/levulinic acid ratios reported using 5hydroxymethylfurfural and hexoses as reactant. Note water is the medium unless
stated.
102
Chapter 4
Clearly, formic acid production results from a complicated chemical mechanism.
However, information on all of the discrete pathways responsible for the nonequimolarity to levulinic acid is as of yet unclear. However, there is common
agreement that the excess in formic acid is likely to be formed either from direct
hexose decomposition or indirectly through hexose consumption of intermediate
species during hexose hydrolysis (Salak Asghari and Yoshida 2006; Rackemann and
Doherty 2012). For example Qi et al. when using an isotopic labelling approach to
decipher the D-glucose/D-fructose dehydration pathways for the formation of
levulinic acid, observed non isotopically labelled formic acid originating from both
D-glucose
and D-fructose (Qi et al 2014). Asghari and Yoshida hypothesized when
observing non-equimolar ratios of formic and levulinic acid that, formic acid and
other organic acids were directly produced from the decomposition of D-fructose
(Salak Asghari and Yoshida 2006). Formic acid has also been reported as a
degradation product of well-known hexose decomposition products such as
dihydroxyacetone, glyceraldehyde and pyruvaldehyde at high temperatures (573 K)
in subcritical water (Kishida et al 2006; Srokol et al 2004). Joshi et al. using a
zirconium dioxide catalyst suggested that the excess in formic acid is formed from a
D-glucose starting material through the formation of 1,6-D-anhydroglucose which
then subsequently undergoes hydrolysis forming furfural, formic acid and hydrogen
stoichiometrically (Joshi et al 2014). Moreover, of recent significance is the study
conducted by Yang et al. who used computational density functional theory to
elaborate a “micro-kinetic” model for the glucose/Bronsted acid aqueous system.
They infer that the excess formic acid appears at high temperatures and originates
from aldol condensation chemistry involving the carbon “6” atom of D-glucose
which results in the direct formation of furfuryl alcohol and in the concurrent release
103
Chapter 4
of formic acid (Yang et al 2015). Another possible source of formic acid is through
the intermediate of D-erythrose which has been reported both from D-fructose
(Peterson et al 2008) and D-glucose (Matsumura et al 2006) as primary reactants.
Despite all of this information, there is still an apparent lack of clarity on the
supposed stoichiometric formation of formic and levulinic acids from acid
hydrolysis of hexose. For example Kumar et al. recently stated ‘‘theoretically,
equimolar amounts of levulinic and formic acids were expected from the conversion
of D-glucose through the intermediate 5-hydroxymethylfurfural’’ (Kumar et al
2015). To this effect, Victor et al. commented‘‘ In principle the ratio (wt/wt%) of
levulinic and formic acids in the product hydrolyzate should be 2.5 as only one
molecule of formic acid is formed per each glucose molecule that is converted to
levulinic acid’’(Victor et al 2014).
In light of all of the above it is necessary to intentionally challenge the common
assumption that formic and levulinic acids are formed stoichiometrically from
hexose carbohydrate starting materials with scientific rigour for the first time. To do
so we perform experimental characterisation of the mechanism of acid hydrolysis
using 2.5 wt% H2SO4 in water for a series of reactants relevant to the hydrolysis
systems of hexoses extending to; D-fructose, D-galactose, D-glucose, D-mannose, 5hydroxymethylfurfural, D-erythrose, levulinic acid, furfuryl alcohol, furfural,
pyruvaldehyde, dihydroxyacetone, glyceraldehyde and formic acid at 423 K.
104
Chapter 4
4.3. Experiment
4.3.1. Materials
D-fructopyranose (CAS 57-48-7, 99% purity), α/β-D-glucopyranose (CAS 5099-7, 99% purity) α/β-D-mannopyranose, (3458-28-4, 99% purity)
D-
galactopyranose (CAS 59-23-4 99% purity), sulphuric acid (H2SO4, 95-97%
purity), D-glyceraldehyde (CAS 56-82-6 90% purity), pyruvaldehyde (CAS
78-98-8 40% purity), lactic acid (CAS 50-21-5 85% purity), acetic acid ( CAS
64-19-7 99% purity) 5-hydroxymethylfurfural (CAS 67-47-0, 99% purity)
furfural (CAS 98-08-1, 98% purity) furfuryl alcohol (CAS 98-00-0 97.5%
purity) and levulinic acid (CAS 59-23-4 97% purity) are each obtained from
Sigma
Aldrich
Ireland
and
used
without
further
purification.
Dihydroxyacetone (CAS 96-26-4 97% purity) and D-erythrose (CAS 533-493)
are purchased from Carbosynth UK and used without further purification.
4.3.2. Experiment configuration
Experiments are carried out with D-fructose, D-galactose, D-glucose, D-mannose as
reactants for determining the formic/levulinic acid ratio. The experiments are
executed at the prescribed temperature using H2SO4 to catalyse the system and
samples are taken at fixed intervals of 30 min, 1h, 2h, 4h 6h, 8h and 10h. A control
reaction is conducted with 5-hydroxymethylfurfural for the purposes of the
validation of experimental results. Additional experiments are carried out using the
same reaction conditions with levulinic acid, furfuryl
alcohol, furfural,
dihydroxyacetone, glyceraldehyde, pyruvaldehdye, and formic acid for the
determination of the origins of non equimolar amounts of formic acid. A reactant
105
Chapter 4
concentration of 0.3 mol/L is chosen in order to replicate typical acid hydrolysis
reports in the literature. A test temperature of 423 K is selected in order to ensure
that no formic acid decomposition occurs once formed. All experiments are carried
out in glass pressure tubes (25.4 mm O.D. x10.2 cm), comprising of
polytetraflouroethylene plugs and flouroelastomer with tetraflouro O-rings for
pressure sealing up to 1.03 MPa. A 5.0 ml aqueous solution containing the desired
concentration of reactant and H2SO4 (2 wt%) is placed into the tubes. After being
sealed, each tube is placed for a defined period of time in an oil bath set at the
desired reaction temperature. When the reaction time is completed, each tube is
removed from the oil bath and immersed in a cold water bath to quench the reaction.
Samples are then prepared for analysis.
Identification and quantification of D-fructose, D-galactose, D-glucose, D-mannose 5hydroxymethylfurfural, and dihydroxyacetone is carried with a ion chromatography
system (IC) system (Dionex Corp., Sunnydale, CA) equipped with a pulsed
amperometric detector (AS, 10 µL sample loop, Dionex Corp., Sunnydale, CA).
Analysis is performed at 291 K by isocratic elution with deionised water (18.2
MΩ.cm at a flow rate of 1.1 ml/min) using a Dionex CarboPac PA1 carbohydrate
column. The column is reconditioned using a mixture of 0.4 mol/L sodium
hydroxide and 0.2 4mol/L sodium acetate after each analysis.
For the determination of levulinic acid, furfuryl alcohol, glyceraldehyde,
pyruvaldehyde, lactic acid and acetic acid, chromatographic separation is achieved
using a Dionex Acclaim® Organic Acid column (5 µm, 4.6×25 mm) coupled with a
guard column cartridge (5 µm, 4.6×10 mm). Isocratic separation is carried out using
100 mM Na2SO4 at a pH of 2.65 (0.55 ml of methanesulfonic acid per every 1 L of
solution) with a flow rate of 0.8 mL/min employing a temperature of 303 K for
106
Chapter 4
separation. A DAD-3000RS is employed at a wavelength of 210 nm for the detection
of analytes. For both types of analysis described, species are identified by matching
retention times to known standards, and quantified by calibration of detector
response to known concentrations. Detector responses for compounds of interest and
chromatograms for both analytical methods are illustrated in Figures 4.2-4.5.
350
D-Glucose k= 282
D-Galactose k=245
D-Mannose k=228
D-Fructose k= 218
Dihydroxyacetone k=58
300
Detector Response
250
200
150
100
50
0
0.0
0.2
0.4
0.6
0.8
1.0
Concentration, mg/ml
Figure 4.2 Detector response for carbohydrates and their derivatives using the
Carbopac™ PA1 column.
Figure 4.3 Detector responses for compounds of interest using the Dionex Acclaim®
Organic Acid column.
107
Chapter 4
16
14
Formic Acid
Detector Response
12
10
Levulinic Acid
8
5-HMF
Lactic Acid
Acetic Acid
6
4
Furfural
2
0
Glyceraldehyde
-2
5
10
15
Time, Minutes
Figure 4.4 Typical chromatogram from the acid catalysed degradation of hexose
sugars using an Acclaim® Organic Acid column.
5-HMF
35
D-Mannose
D-Fructose
D- Galactose
Detector Response
30
25
20
15
D-Erythrose
10
5
Dihydroxyacetone
0
0
2
4
6
D- Glucose
8
10
12
14
Time, Minutes
Figure 4.5 Typical chromatogram from the acid catalysed degradation of hexose
sugar using a PA1 Dionex CarboPac PA1 carbohydrate column.
4.3.3. Calculations and uncertainty analysis
Molar yields, selectivity and the amount of excess formic acid are calculated by the
implementation of the following equations:
108
Chapter 4
C
C
0
(Y),t
Molar Yield (Y) =
×100
C
(X)0
C
C
(Z)t
(Z)0
Selectivit y (Z) =
×100
C
C
0
(X),t
(1)
(2)
Where:
C(Y),t is the molar concentration of the product of interest at a time (t).
C(Y),0 is the molar concentration of the product of interest at time zero.
C(X),0 is the molar concentration of the reactant at time zero.
C(Z),t is the molar concentration of the product of interest at a time (t).
C(Z),0 is the molar concentration of the product of interest at time zero.
C(X),t is the molar concentration of the reactant at a time (t).
The term “excess formic acid” is defined by the equation 3:
C
Excess Formic Acid =
C
(FA),T,
(FA),5-HMF
×100
C
(FA),5-HMF
(3)
C(FA),T is the total molar concentration of formic acid detected.
C(FA),5-HMF is the molar concentration of formic acid detected from the reaction of 5hydroxymethylfurfural.
This is assumed to be the same is the detected
concentrations of levulinic acid.
Where possible all experiments are repeated in duplicate and experimental
uncertainties are denoted on all data sets reported.
109
Chapter 4
4.4. Results and discussion
4.4.1. Hexose conversion
The kinetic timescale as well as the mechanistic detail of the dehydration reactions is
found to vary with each hexose tested. Figure 4.6 illustrates this behaviour in terms
of reactant conversion and their tendency to form 5-hydroxymethylfurfural, formic
acid and levulinic acid at 423 K catalysed by 2.5 wt% H2SO4. The rate of primary
reactant conversion (Figure 4.6 (a)) is in the order of 5-hydroxymethylfurfural > Dfructose > D-mannose > D-galactose > D-glucose. This is in line with the findings of
Baugh and McCarty who found the hexose carbohydrate consumption rate to order
as; D-mannose > D-galactose > D-glucose (Baugh and McCarty 1988).
4.4.2. Levulinic acid yields
The conversion rates of D-mannose, D-glucose and D-galactose are relatively similar
(Figure 4.6(a)), but importantly their mechanistic propensity to form 5hydroxymethylfurfural differs. This can be attributed to the steric configuration of
their hydroxyl groups which results in significant differences in selectivities to their
acid hydrolysis products, such as 5-hydroxymethylfurfural and levulinic acid (Hu et
al 2013). Selectivity to 5-hydroxymethylfurfural from the model hexoses is in the
order of; D-fructose > D-glucose > D-mannose > D-galactose (Figure 4.6(b)),
subsequently yields of levulinic acid from all model hexoses follow the same
patterns as with 5-hydroxymethylfurfural formation (Figure 4.6(c)). This is
consistent with the trend observed by Hu et al. who when investigating carbohydrate
structure on yields of levulinic acids found them to be in the order of D-fructose >
D-glucose > D-galactose (Hu et al 2013).
110
Chapter 4
It is evident that both selectivity to 5-hydroxymethylfurfural and yields of levulinic
acid from D-fructose (70 mol%) are higher than those from the other hexose
carbohydrates. This can be attributed to the fact that D-fructopyranose can readily
tautomerise to fructofuranoses (Angyal 1969) structures which are more
thermodynamically favoured towards protonation and subsequently dehydrate to 5hydroxymethylfurfural than the equivalent pyranose structures that are present in Dglucose, D-mannose and D-galactose.
Figure
4.6
Experimental
data
for;
(a)
Reactant
conversions,
(b)
5-
hydroxymethylfurfural (5-HMF) yields, (c) Formic acid yields, (d) Levulinic acid
yields. All experiments use 2.5 wt% H2SO4 in water at 423 K with a reactant loading
of 0.3 mol/L.
111
Chapter 4
Levulinic acid yields from D-galactose (32 mol%) are the lowest from all the model
hexoses. D-galactose is a large molecule, and due to steric hindrances is more likely
to decompose to unwanted by-products such as dihydroxyacetone than the other
model compounds. Dihydroxyacetone is detected in significantly higher quantities
for D-galactose than from any of the other hexose carbohydrates (Figure 4.11), which
at least offers a partial explanation for the poor yields of levulinic acid achieved. As
expected levulinic acid yields are significantly higher from 5-hydroxymethylfurfural
(83 mol%) than from any of the other hexose carbohydrates. Thus it is clearly
evident why achieving high selectivities to 5-hydroxymethylfurfural from hexose
carbohydrates is seen as the key step for a sustainable bio-refining industry. Perhaps
somewhat unsurprisingly, concentrations of formic acid are found to be higher than
levulinic acid for all model hexose compounds, Figure 4.6(c-d).
4.4.3. Ratio of formic to levulinic acid
Figure 4.7 depict that formic and levulinic acids are stable under the
conditions employed in this study, and thus are appropriate in order to
accurately ascertain the formic to levulinic acid ratio.
Figure 4.7 Stability of formic acid in the presence of 2.5 wt% H2SO4 at 423 K.
Note the variability of conversion measurement is within experimental uncertainty.
112
Chapter 4
Control reactions conducted using 5-hydroxymethylfurfural as reactant result in
equimolar formic and levulinic acid as was observed by the work of others
(Girisunta et al 2006; Swift et al 2013). In contrast, the ratio of formic to levulinic
acids for all model hexose carbohydrates is found to be >1 (Figure 4.8). Figure 4.6
demonstrates that there is a decrease in the formic/levulinic acid ratios observed with
respect to time, particularly in relation to D-mannose, D-glucose and D-galactose.
This can be attributed to the fact that formic acid is found to form from the
aforementioned hexoses at a faster rate than from the hydration of 5hydroxymethylfurfural, the presumed dominant pathway. Consequently, with time,
as 5-hydroxymethylfurfural is formed and consumed, the high formic/levulinic acid
ratio observed at early stages of reaction decreases.
Figure 4.8 Formic/levulinic acid ratios per time considering each model compound
as reactant.
When the model reactions have proceeded to steady-state, the formic/levulinic acid
ratios for D-fructose, D-glucose, D-mannose and D-galactose are found to be 1.08
±0.04, 1.15 ±0.05, 1.20 ±0.10 and 1.19 ±0.04 respectively. The formic/levulinic acid
ratios observed for D-fructose are in line with the values observed by Swift et al.
113
Chapter 4
(Swift et al 2013) and Qi et al. (Qi et al 2014) They both observed formic/levulinic
acid ratios of just over 1. The ratios observed for D-glucose are lower than that
observed by Kumar et al. (1.54) however, as all values quoted employ different
reaction systems (temperatures, catalysts), precise comparisons are not appropriate
(Kumar et al 2015). To the best of the author’s knowledge little information
regarding formic/levulinic acid ratios for D-mannose and D-galactose is available in
the literature. However Swift et al. (Swift et al 2015) recently stated that D-mannose
undergoes significant losses to formic acid particularly at low temperatures whilst
undergoing acid hydrolysis. From experiment D-fructose is found to have the lowest
formic/levulinic acid ratio of all the hexose carbohydrates. A trend cannot be
articulated for D-galactose, D-glucose, D-mannose as the reported values are within
the uncertainty estimates for each steady-state measurement.
4.4.4. Pathways to formic acid
The excess in formic/levulinic acid
ratios observed for each model hexose
carbohydrate deems it necessary to investigate formic acid formation from known
hexose intermediates in the system not derived from 5-hydroxymethylfurfural. The
acid catalysed transformation of hexose carbohydrates is a complicated process, with
numerous parallel reactions occurring. To this effect, Fusaro et al. (Fusaro et al
2015) categorised the acid catalysed transformation of carbohydrates into the
following reaction types: isomerisations, dehydrations, fragmentations and
condensations. Typically isomerisation reactions can occur between hexose
carbohydrates, especially between D-fructose, D-mannose and D-glucose while
tautomerisation can also occur between furanose and pyranose structures of the same
compounds. The main dehydration products include 5-hydroxymethylfurfural,
114
Chapter 4
anhydrohexoses
and
numerous
precursors
to
5-hydroxymethylfurfural.
Fragmentation products include furfural, dihydroxyacetone, glyceraldehyde and any
products arising from their subsequent degradation, whilst humins represent the main
condensation products. Thus there are numerous potential pathways for excess
formic acid formation
Figure 4.9 shows a simple summary of the behaviour of hexose carbohydrates in
aqueous acidic media from the literature and also describes the origin of any
potential sources of excess formic acid that have been postulated and are worthy of
investigation. At this time a complete census of all potential sources of excess formic
acid is overly ambitious, particularly for D-mannose and D-galactose which are less
well studied than D-fructose and D-glucose. From consulting the literature it appears
excess formic acid formation can be categorised into 2 distinctive groups;
1. Excess formic acid from direct hexose transformations.
2. Excess formic acid from hexose derived intermediates.
4.4.5. Excess formic acid from direct hexose transformations
Furfuryl alcohol
Perhaps the most pertinent reported pathway for excess formic acid has been recently
reported by Yang et al. in which they suggested the formation of excess formic acid
from the direct dehydration of hexose carbohydrates to furfuryl alcohol, liberating 2
moles of H2O.
They proposed that the formic acid is formed by the aldol
condensation that originates from retro-aldo chemistry in aqueous media involving
the carbon 6 atom of D-glucose, subsequently liberating furfuryl alcohol. In our tests
conducted with D-galactose, D-glucose, D-fructose and D-mannose furfuryl alcohol is
115
Chapter 4
not detected as a stable identity. Consequently under the test conditions of our study
furfuryl alcohol is found to be very unstable, with almost 100% conversion after 5
minutes at 453 K catalysed by 1.5wt% H2SO4 (See Figure 4.10(a)).Therefore
furfuryl alcohol detection as a stable intermediate is difficult. The excess formic acid
arises as the selectivity of furfuryl alcohol hydration to levulinic acid is significantly
lower than from 5-hydroxymethylfurfural (See Figure 4.10(b)).
Taking D-glucose as an example, using the formic/levulinic acid ratio of 1.15 and
levulinic acid yields of 46% measured in our investigation, the proposal of Yang et
al. is exercised. Using their suggested pathway and our data, results in a selectivity
of 8.2% for furfuryl alcohol formation from D-glucose. This is assuming that the
described furfuryl alcohol pathway is solely responsible for the excess formic acid.
This illustrates how relatively low selectivity of hexose sugars to furfuryl alcohol can
have a significant effect on the formic/levulinic acid ratio.
Furfural
The investigation of the formation of furfural as a possible source of formic acid is
relevant, as recently Joshi et al. have postulated that formic acid can be formed as a
by-product in the formation of furfural though the intermediate of 1,6 anhydro-Dglucose liberating equimolar amounts of H2O, H2 and furfural (Joshi et al 2014).
Their study was conducted at 423 K, using cellulose as the primary reactant
catalysed by a zirconium dioxide catalyst and found significant amounts of furfural
(up to 10 mol%).
116
Chapter 4
In this study trace amounts of furfural (0.01 mol%) are detected in the order of Dfructose > D-mannose > D-glucose and D-galactose (See Figure 4.11).
117
Chapter 4
1
2
Figure 4.9 Possible pathways for formic acid formation that have been reported in the literature.
118
Chapter 4
Hexose
Carbohydrates
-H2O H+
+H2O H+
O
O
O
+ HO
Furfural
OH
OH
Anhydrohexose
+
HO
O
+
Formic Acid
H2
(b)
Interestingly, the highest formic/levulinic acid ratio was reported by Kumar et al.
(1.54) who also employed a Lewis acid, using metal bromides coupled with HCl to
catalyse the reaction (Kumar et al 2015) It has also been postulated that excess
formic acid is derived from the degradation of furfural (Mielenz et al 2009). No
formic acid is detected from the acid catalysed degradation of furfural in our
investigation.
Figure 4.10 (a) Stability of model compounds at 453 K catalysed by 2.5 wt% H2SO4
b) selectivities of 5-hydroxymethylfurfural and furfuryl alcohol to levulinic acid at
453 K catalysed by 2.5 wt% H2SO4. Note only formic acid is detected from Derythrose.
4.4.6. Excess formic acid from hexose derived intermediates.
D-Erythrose and Glycolaldehyde
Equimolar D-erythrose and D-glycolaldehyde have both been reported from the
decomposition of D-glucose (Matsumura et al 2006) and D-fructose (Mӧller et al
119
Chapter 4
2012,) Peterson et al. (Peterson et al 2008) have reported D-erythrose to thermally
degrade to form acetic and formic acids between 573-623 K.
D-erythrose is detected in our investigation from D-glucose, and D-fructose in trace
amounts. No D-erythrose is detected from D-mannose or D-galactose. As Figure
4.10(a) depicts D-erythrose is shown to be very reactive under the conditions of this
study, and degrades to give formic acid yields of 18 mol% at 100% conversion with
acetic acid the other main product detected. Therefore it is likely that D-erythrose is a
contributor to the excess formic acid formed from D-glucose under conventional
hexose acid hydrolysis systems. It is also probable that D-glycolaldehyde, which has
been postulated to form stoichiometrically with D-erythrose (Mӧller et al 2012) can
undergo transformations to form formic acid. For instance Kishida et al. achieved
formic acid yields of 13 mol% at 573 K catalysed by NaOH (Kishida et al 2006).
Potential formic acid formation from D-glycolaldehyde isn’t investigated in our
study due to the difficulty in obtaining it as reactant.
Dihydroxyacetone, Glyceraldehyde and Pyruvaldehyde
Dihydroxyacetone is the most abundant hexose decomposition product detected in
this investigation in the form of D-galactose > D-mannose > D-glucose > D-fructose
and it has been suggested that its acid catalysed transformations can form formic
acid through the pyruvaldehyde intermediate (See Figure 4.11).
120
Chapter 4
Gao et al. report a competing pathway for the degradation of pyruvaldehyde (Gao et
al 2013) forming equimolar formic acid and acetaldehyde at temperatures between
443-483 K using NaOH as the catalyst. Kishida et al. (Kishida et al 2006) detected
formic yields of 5.3 mol% from glyceraldehyde at 473 K under alkali conditions.
In our study dihydroxyacetone and glyceraldehyde are found to be reactive (Figure
4.10(a)) and are converted quite quickly to pyruvaldehyde which is observed to
degrade at a slower rate to lactic and acetic acids. No formic acid is detected for any
of
the
reactions
conducted
with
dihydroxyacetone
glyceraldehyde
and
pyruvaldehyde as reactants. Our investigation suggests, that the formic acid detected
by others can be explained by the high temperatures employed and from the use
alkali conditions to catalyse the cellulose/hexose transformations.
4.4.7. Outlook
This investigation highlights that the formation of formic acid from hexose
carbohydrates is a complex process with several potential pathways
contributing, other than through the hydration of 5-hydroxymethylfurfural.
Therefore it is clear the formic and levulinic acid are not formed
stoichiometrically from hexose carbohydrates and need to be treated as
separate identities when making predictions of formic acid concentrations for
the hydrolysis of cellulose and hemicellulosic derived hexose carbohydrates.
121
Chapter 4
As Figure 4.10(a) illustrates the potential formic acid forming intermediates
are unstable for the test conditions of this study. To ascertain the exact
numerical contribution of each potential formic acid pathway, dedicated
mechanistic experiments and sophisticated kinetic modelling would need to be
performed on each specific reactant at a range of conditions, which represents
an avenue for future study.
Figure 4.11 Molar yields of dihydroxyacetone (a), the relative abundance of acetic
acid (b) and the relative abundance of furfural (c) detected for all model hexose
carbohydrates.
122
Chapter 4
4.5. Conclusion
The formic to levulinic acid concentration ratio for the direct hydrolysis of hexose
carbohydrates is found to be >1. This deems the common assumption that formic and
levulinic acids are formed stoichiometrically from the acid catalysed hydrolysis of
hexose carbohydrates to be inaccurate. At steady-state, formic/levulinic ratios for Dfructose, D-galactose, D-glucose and D-mannose are found to be 1.08 ±0.04, 1.15
±0.05, 1.20 ±0.10 and 1.19 ±0.04. Combining this work and pertinent literature
suggests there are at least four potential pathways depending on reaction condition
responsible for the excess formic acid, through furfuryl alcohol, furfural formation
and through the transformation of pyruvaldehyde and D-erythrose.
4.6. References
Aida, T.M., Tajima, K., Watanabe, M., Saito, Y., Kuroda, K., Nonaka, T., Hattori,
H., Smith, R.L., Arai, K. (2007) “Reactions of d-fructose in water at temperatures
up to 400° C and pressures up to 100MPa,” The Journal of Supercritical Fluids,
42(1), 110–119.
Angyal, S.J. (1969) “The composition and conformation of sugars in solution,”
Angewandte Chemie International Edition in English, 8(3), 157–166.
Baugh, K.D., McCarty, P.L. (1988) “Thermochemical pretreatment of lignocellulose
to enhance methane fermentation: I. Monosaccharide and furfurals hydrothermal
decomposition and product formation rates,” Biotechnology and Bioengineering,
31(1), 50–61.
Bozell, J.J., Moens, L., Elliott, D., Wang, Y., Neuenscwander, G., Fitzpatrick, S.,
Bilski, R., Jarnefeld, J. (2000) “Production of levulinic acid and use as a platform
123
Chapter 4
chemical for derived products,” Resources, Conservation and Recycling, 28(3),
227–239.
Bozell, J.J., Petersen, G.R. (2010) “Technology development for the production of
biobased products from biorefinery carbohydrates—the US Department of
Energy’s ‘top 10’ revisited,” Green Chemistry, 12(4), 539–554.
Braden, D.J., Henao, C.A., Heltzel, J., Maravelias, C.C., Dumesic, J.A. (2011)
“Production of liquid hydrocarbon fuels by catalytic conversion of biomassderived levulinic acid,” Green Chemistry, 13(7), 1755–1765.
Dussan, K., Girisuta, B., Haverty, D., Leahy, J., Hayes, M. (2013) “Kinetics of
levulinic acid and furfural production from Miscanthus x giganteus,” Bioresource
Technology, 149, 216–224.
Fitzpatrick, S.W. (1990) “Lignocellulose degradation to furfural and levulinic acid.”
Fusaro, M.B., Chagnault, V., Postel, D. (2015) “Reactivity of d-fructose and dxylose in acidic media in homogeneous phases,” Carbohydrate Research, 409, 9–
19.
Gao, P., Li, G., Yang, F., Lv, X.-N., Fan, H., Meng, L., Yu, X.-Q. (2013)
“Preparation of lactic acid, formic acid and acetic acid from cotton cellulose by
the alkaline pre-treatment and hydrothermal degradation,” Industrial Crops and
Products, 48, 61–67.
Girisuta, B., Dussan, K., Haverty, D., Leahy, J., Hayes, M. (2013) “A kinetic study
of acid catalysed hydrolysis of sugar cane bagasse to levulinic acid,” Chemical
Engineering Journal, 217, 61–70.
124
Chapter 4
Girisuta, B., Janssen, L., Heeres, H. (2006) “A kinetic study on the decomposition of
5-hydroxymethylfurfural into levulinic acid,” Green Chemistry, 8(8), 701–709.
Hu, X., Wu, L., Wang, Y., Song, Y., Mourant, D., Gunawan, R., Gholizadeh, M., Li,
C.-Z. (2013) “Acid-catalyzed conversion of mono-and poly-sugars into platform
chemicals: Effects of molecular structure of sugar substrate,” Bioresource
Technology, 133, 469–474.
Joshi, S.S., Zodge, A.D., Pandare, K.V., Kulkarni, B.D. (2014) “Efficient conversion
of cellulose to levulinic acid by hydrothermal treatment using zirconium dioxide
as a recyclable solid acid catalyst,” Industrial & Engineering Chemistry Research,
53(49), 18796–18805.
Kishida, H., Jin, F., Yan, X., Moriya, T., Enomoto, H. (2006) “Formation of lactic
acid from glycolaldehyde by alkaline hydrothermal reaction,” Carbohydrate
Research, 341(15), 2619–2623.
Kumar, V.B., Pulidindi, I.N., Gedanken, A. (2015) “Synergistic catalytic effect of
the ZnBr 2-HCl system for levulinic acid production using microwave
irradiation,” RSC Advances, 5(15), 11043–11048.
Matsumura, Y., Yanachi, S., Yoshida, T. (2006) “Glucose decomposition kinetics in
water at 25 MPa in the temperature range of 448-673 K,” Industrial &
engineering chemistry research, 45(6), 1875–1879.
Mielenz, J.R., Klasson, K.T., Adney, W.S., McMillan, J.D. (2009) Biotechnology for
fuels and chemicals: The twenty-eighth symposium., Springer Science & Business
Media.
125
Chapter 4
Mukherjee, A., Dumont, M.-J., Raghavan, V. (2015) “Review: Sustainable
production of hydroxymethylfurfural and levulinic acid: Challenges and
opportunities,” Biomass and Bioenergy, 72, 143–183.
Mӧller, M., Harnisch, F., Schrӧder, U. (2012) “Microwave-assisted hydrothermal
degradation of fructose and glucose in subcritical water,” Biomass and Bioenergy,
39, 389–398.
Pasquale, G., Vázquez, P., Romanelli, G., Baronetti, G. (2012) “Catalytic upgrading
of levulinic acid to ethyl levulinate using reusable silica-included Wells-Dawson
heteropolyacid as catalyst,” Catalysis Communications, 18, 115–120.
Peterson, A.A., Vogel, F., Lachance, R.P., Frӧling, M., Antal Jr, M.J., Tester, J.W.
(2008) “Thermochemical biofuel production in hydrothermal media: a review of
sub-and supercritical water technologies,” Energy & Environmental Science, 1(1),
32–65.
Qi, L., Mui, Y.F., Lo, S.W., Lui, M.Y., Akien, G.R., Horváth, I.T. (2014) “Catalytic
Conversion of Fructose, Glucose, and Sucrose to 5-(Hydroxymethyl) furfural and
Levulinic and Formic Acids in gamma-valerolactone as a green solvent,” Acs
catalysis, 4(5), 1470–1477.
Rackemann, D.W., Doherty, W.O. (2011) “The conversion of lignocellulosics to
levulinic acid,” Biofuels, Bioproducts and Biorefining, 5(2), 198–214.
Rackemann, D.W., Doherty, W.O. (2012) “A review on the production of levulinic
acid and furanics from sugars,” in 34th Australian Society of Sugar Cane
Technologists Conference.
126
Chapter 4
Salak Asghari, F., Yoshida, H. (2006) “Acid-catalyzed production of 5hydroxymethyl furfural from D-fructose in subcritical water,” Industrial &
engineering chemistry research, 45(7), 2163–2173.
Shen, J., Wyman, C.E. (2012) “Hydrochloric acid-catalyzed levulinic acid formation
from cellulose: data and kinetic model to maximize yields,” AIChE Journal,
58(1), 236–246.
Srokol, Z., Bouche, A.-G., van Estrik, A., Strik, R.C., Maschmeyer, T., Peters, J.A.
(2004) “Hydrothermal upgrading of biomass to biofuel; studies on some
monosaccharide model compounds,” Carbohydrate Research, 339(10), 1717–
1726.
Swift, T.D., Bagia, C., Choudhary, V., Peklaris, G., Nikolakis, V., Vlachos, D.G.
(2013) “Kinetics of Homogeneous Brønsted Acid Catalyzed Fructose Dehydration
and 5-Hydroxymethyl Furfural Rehydration: A Combined Experimental and
Computational Study,” ACS Catalysis, 4(1), 259–267.
Swift, T.D., Nguyen, H., Anderko, A., Nikolakis, V., Vlachos, D.G. (2015) “Tandem
Lewis/Brønsted homogeneous acid catalysis: conversion of glucose to 5hydoxymethylfurfural in an aqueous chromium (iii) chloride and hydrochloric
acid solution,” Green Chemistry, 17(10), 4725–4735.
Victor, A., Pulidindi, I.N., Gedanken, A. (2014) “Levulinic acid production from
Cicer arietinum, cotton, Pinus radiata and sugarcane bagasse,” RSC Advances,
4(84), 44706–44711.
Werpy, T., Petersen, G., Aden, A., Bozell, J., Holladay, J., White, J., Manheim, A.,
Eliot, D., Lasure, L., Jones, S. (2004) “Top value added chemicals from biomass.
127
Chapter 4
Volume 1-Results of screening for potential candidates from sugars and synthesis
gas.”
Yang, L., Tsilomelekis, G., Caratzoulas, S., Vlachos, D.G. (2015) “Mechanism of
Brønsted Acid-Catalyzed Glucose Dehydration,” ChemSusChem.
Zacharska, M., Podyacheva, O.Y., Kibis, L.S., Boronin, A.I., Senkovskiy, B.V.,
Gerasimov, E.Y., Taran, O.P., Ayusheev, A.B., Parmon, V.N., Leahy, J., others
(2015) “Ruthenium Clusters on Carbon Nanofibers for Formic Acid
Decomposition: Effect of Doping the Support with Nitrogen,” ChemCatChem,
7(18), 2910–2917.
Zhang, X., Hewetson, B.B., Mosier, N.S. (2015) “Kinetics of Maleic Acid and
Aluminum Chloride Catalyzed Dehydration and Degradation of Glucose,” Energy
& Fuels.
128
Chapter 5
Chapter 5: Reaction Pathway Analysis of Ethyl
Levulinate and 5-Ethoxymethylfurfural from
D-Fructose Acid Hydrolysis in Ethanol
Chapter 4 shows that there are clearly a number of competitive pathways for the acid
catalysed transformation of hexose sugar in aqueous systems. We presume the same
scenario exists in an ethanol media, with the premise that there should be a more
diverse range of molecules with ethylated functionalities. Building on the
mechanistic understanding achieved in Chapter 4, this Chapter explores the reaction
mechanism for the ethanolysis of D-fructose to ethyl levulinate and 5ethoxymethylfurfurfural and utilises kinetic modelling to test the validity of
mechanistic propositions.
5.
sdfsdf
This work has been presented at the ACS 2015 Spring National Meeting in Denver
Colorado:
Flannelly T., Dooley S., Leahy, J.J. (2015) ‘‘Mechanism and kinetics of advantaged
biofuels synthesis from D-fructose’’ Paper No 20, ACS Spring National Meeting,
Denver Colorado, March 22-26, 2015.
Flannelly T., Dooley S., Leahy, J.J. (2015) “Reaction Pathway Analysis of Ethyl
levulinate and 5-Ethoxymethylfurfural from D-Fructose Acid Hydrolysis in Ethanol,”
Energy & Fuels. 29((11), 7554-7565.
DOI: 10.1021/acs.energyfuels.5b01481
129
Chapter 5
5.1.
Abstract
This study utilises numerical modelling to provide a mechanistic discussion of the
synthesis
of
the
advanced
biofuel
candidates,
ethyl
levulinate
and
5-
ethoxymethylfurfural, from α/β-D-fructopyranose (D-fructose) in a condensed phase
homogeneous ethanol system at 351 K catalysed by hydrogen cations. A mechanistic
comprehension is pursued by detailed measurements of reactant, intermediate and
product species temporal evolutions, as a function of H2SO4 (0.09, 0.22, 0.32 mol/L)
and (0.14, 0.29, 0.43 mol/L) concentration, also considering the addition of water to
the ethanol media (0, 12, 24 mass% water in ethanol). D-fructose, 5hydroxymethylfurfural, 5-ethoxymethylfurfural, ethyl levulinate, and several other
intermediate species are quantified as major species fractions at 45-85% of the initial
D-fructose mass. To inform the mechanistic discussion mass-conserved chemically
authentic kinetic models and empirical rate constants are derived each assuming a
first order relationship to the hydrogen cation concentration. The optimal synthesised
fractions of ethyl levulinate and 5-ethoxymethylfurfural considered as fuel
components, achieve a mass yield of 63% with respect to the fructose mass and a
volumetric energy valorisation (HCombustion, kcal/mL) of 215% with respect to the
ethanol consumed, indicating the viability of the synthesis.
130
Chapter 5
5.2. Introduction
Presently there is a growing effort to find renewable and sustainable alternatives to
petroleum derived bulk chemicals and fuels. The catalytic conversion of biomass
derived cellulose and hemicellulose to platform chemicals has been widely
recognised as an opportunity to develop a carbohydrate based chemical industry
(Serrano-Ruiz et al 2011; Yan et al 2014) Lignocellulosic derived hexose and
pentose sugars have potential as a sustainable alternative to the carbohydrates
derived from edible crop matter. The United States Department of Energy have
identified promising renewable chemical building blocks that may be produced from
such biomass derived sugars (Werpy et al 2004). Ethyl levulinate and 5ethoxymethylfurfural, Figure 5.1, are two such promising furan derived chemicals
for potential use as transportation fuels.
Figure 5.1 Molecular structures of ethyl levulinate and 5-ethoxymethylfurfural.
Ethyl levulinate has received a significant amount of attention purporting its
potential use as a fuel (Wang et al 2012; Chia and Dumesic 2011). Chemical systems
reporting its synthesis include; ethylation of levulinic acid, furfuryl alcohol and 5chloromethylfurfural (Maldondo 2012; Mascal and Nikitin 2010; Zhang et al 2011;
Fernandes et al 2012; Maldonado et al 2012; Pasquale et al 2012; Yan et al 2013).
Ethyl levulinate production from biomasses have been reported (Garves 1988; Chang
131
Chapter 5
et al 2012; Grisel et al 2014) however, yields have been modest at 40-50%. There
have also been considerable difficulties in its synthesis from glucose and cellulose
with 44.8 mol% the highest yield reported to date using glucose as a starting material
in an ethanol/H2SO4 system (Zhu et al 2014).
5-ethoxymethylfurfural has received more limited suggestions as a fuel component
(Wang et al 2013) despite its high volumetric density of 1.099 g/mL (298 K). Gruter
and Dautzenberg (Dautzenberg and Gruter 2012) suggest an enthalpy of formation of
120.1 kcal/mol, corresponding to an enthalpy of combustion of 7.87 kcal/g (see
Table 5.1). These terms correspond to a volumetric energy density of 8.66 kcal/mL
(36.24 MJ/L at 298 K), thus being advantageous over those of other oxygenated fuel
components, such as ethanol (7.11 kcal/mL), and ethyl levulinate (7.53 kcal/mL).
Indeed the value is comparable to petroleum derived diesel and gasoline (8.53
kcal/mL for the gasoline primary reference fuel, iso-octane). Mascal and Nikitin
(Mascal
and
Nikitin
2008)
used
5-chloromethylfurfural
to
convert
5-
hydroxymethylfurfural to 5-ethoxymethylfurfural. Other reports use aluminium
chloride as a catalyst to transform glucose to 5-ethoxymethylfurfural in an
ethanol/water medium (Yang et al 2012) and also from using a mixture of Sn-Beta
and Amberlyst catalysts (Lew et al 2012).
Fuel
∆ Hcombustion kcal/g
∆ Hcombustion kcal/mL
Ethanol
9.75
7.63
5-Ethoxymethylfurfural
7.87
8.66
Ethyl Levulinate
7.41
7.53
Table 5.1. Enthalpies of combustion for proposed transportation fuel components.
132
Chapter 5
Like ethyl levulinate, the production of 5-ethoxymethylfurfural from glucose or
other cellulosic sugars presents challenges. The desired etherification of glucose is
suppressed by side reactions of various polymerisations and acetalizations producing
the recalcitrant humic substances (Alam et al 2012; Yanget al 2012). 40-50% of
lignocellulosic biomass is made-up of cellulosic glucose polymers, accounting for
the largest proportion of hexoses that may be obtained from biomass (Climent et al
2014). As such glucose-like hexose sugars are cheaper and more readily available
than pentose sugars.
It is well understood that the steric and electronic configurations of the hydroxyl
groups of sugars, significantly affect yields of esterification products from
monosaccharides (Hu et al 2013). As a consequence, fructose is much easier to
convert into furan related products than glucose. The desired isomerisation from
glucose is known to be facilitated by an aqueous/organic media where the
equilibrium population have high proportions of labile α and β‐fructofuranose
structures (Angyal 1969) There is a general consensus that in order for glucose to be
efficiently converted into furanic derivatives, it must initially tautomerize into
fructose species (Saravanamurugan et al 2011; Climent et al 2014). There are several
recent reports describing reaction conditions that promote glucose isomerisation to
fructose in chemical media (Choudharyet al 2013; Despax et al 2013; BermejoDeval et al 2014). For example, Despax et al. (Despax et al 2013) reported on the
use of heterogeneous catalysts in organic solvent mixtures showing ~ 68%
conversion of glucose to fructose. The successful isomerisation of glucose to
fructose in an alcohol medium by the deliberate formation of methyl fructoside
species as intermediates between glucose conversions to fructose is of particular
potential
significance
for
the
production
of
ethyl
levulinate
and
5133
Chapter 5
ethoxymethylfurfural. Saravanamurugan et al. (Saravanamurugan et al 2013)
achieved 55% conversion from glucose to fructose by a one hour reaction in
methanol at 393 K using an H-USY zeolite. This suggests that an alcohol may be
employed as both a solvent and alkylating agent simultaneously, whilst also
facilitating the required isomerization of glucose to fructose. In this way, ethyl
levulinate and 5-ethoxymethylfurfural may be synthesised directly, rather than
relying on the intermediary ethylation of the levulinic acid produced in an aqueous
system.
In addition to this sugar inter-conversion, the subsequent mechanism of fructose
consumption is the obvious further limiting step in achieving viable yields of furanic
derivatives (Zhao et al 2007). Given this prevailing position in the literature an
improved mechanistic understanding of ethyl levulinate and 5-ethoxymethylfurfural
synthesis from fructose, as well as from glucose is sought. There are recent reports of
kinetic studies conducted on the purported H+ (hydrogen cation) homogeneously
catalysed dehydration of D-fructose to 5-hydroxymethylfurfural
in water
(Caratzoulas and Vlachos 2011; Assary et al 2012; Swift et al 2013). However, little
is known of the analogous D-fructose dehydration in the presence of ethanol and H+.
Plausible reaction pathways have been suggested,(Saravanamurugan et al 2011;
Balakrishnan et al 2012) but no quantitative kinetic data for 5-ethoxymethylfurfural
and ethyl levulinate synthesis have been reported.
We pursue an improved mechanistic comprehension employing a hierarchical
modelling approach that studies one sugar sub-mechanism at a time. In this context,
we study the bottle-neck α/-D-fructopyranose (D-fructose) sub-mechanism initially,
which once understood would allow the more complex D-glucose and cellulose sub
models to be developed in a hierarchical manner. In order to develop realistic
134
Chapter 5
reaction kinetics, it is necessarily to limit the modelling complexity. To do so, we
choose a homogeneous catalytic system of α/-D-fructose/H2SO4/ethanol, thereby
minimising the mass transfer complexities of multiphase heterogeneous reactions.
The aim is to establish the main mechanistic relationships between reactant,
intermediate and intended product species such as to inform mechanistic discussion
and to test the validity of a viable reaction mechanism by the derivation of empirical
rate constants. By so doing the viability of preferentially producing one proposed
fuel component over the other, or to what extent this is achievable may be
determined in a rigorous scientific manner.
5.3. Experimental configuration
5.3.1. Materials
Ethanol, normal-octanol, acetone, (99% purity), α/β-D-fructopyranose (CAS 57-487, 99% purity), α/β-D-glucopyranose (CAS 50-99-7, 99% purity) α/β-Dmannopyranose, (3458-28-4, 99% purity), hence forth “D-fructose” “D-glucose” and
“D-mannose”
respectively,
sulphuric
acid
(H2SO4,
95-97%
purity),
5-
hydroxymethylfurfural (CAS 67-47-0, 99% purity) furfural (CAS 98-08-1, 98 %
purity),and ethyl levulinate (CAS 539-88-8, 97% purity) are each obtained from
Sigma Aldrich Ireland. Ethyl-α-D-glucopyranoside (CAS 34625-23-5, 98% purity) is
obtained from Carbosyth Ltd. UK, and 5-ethoxymethylfurfural (CAS 1917-65-3, 9697% purity) is purchased from Akos Organics Gmbh, Germany.
5.3.2. Experimental
Reactions are performed in a 20 cm3 spherical reactor at isothermal conditions of
351 ±1 K at atmospheric pressure. The reactor is heated by an external oil bath. The
135
Chapter 5
reaction temperature is independently controlled and monitored by a thermocouple
array (Stuart™ SCT1 temperature controller) and an in-situ magnetic propeller
ensures that the reaction mixture (D-fructose/H2SO4/ethanol) is well mixed and
homogeneous. Atmospheric pressure is regulated by fitting the main reactor exit with
an open-ended condensing unit (~ 277 K), thus allowing the reaction to be at reflux.
For the test conditions reported here, Table 5.2 a heating time of 16 minutes is
required for the reacting mixture to be heated from ambient to the prescribed
reaction temperature of 351 ±1 K. Reaction conditions are selected (Table 5.2) to
parameterise the influence of [H2SO4] and [D-fructose] on the reaction mechanism,
whilst also considering three scenarios of ethanol/water as reaction media. Reaction
progress is monitored by removing and analysing a 50 mg sample of the bulk
reaction (0.104 g of D-fructose in 15.78 g of ethanol) every hour for 480 minutes,
resulting in a small cumulative perturbation to the overall system mass. Control tests
at the most severe conditions of Table 5.2, replacing D-fructose with ethyl levulinate,
show ethyl levulinate degradation to be within the estimated experimental
uncertainty of the measurement, indicating it as a stable end-product. Control
reactions are also performed substituting 5-hydroxymethylfurfural and 5ethoxymethylfurfural as starting materials for the purposes of identifying the origins
of various intermediate species, as elaborated later.
136
Chapter 5
Test #
Sulphuric acid
concentration,
mol/L
concentration, mol/L
Ethanol/water mass
ratio
1
0.09
0.29
100/0
2
0.22
0.29
100/0
3
0.32
0.29
100/0
4
0.11
0.15
100/0
5
0.34
0.43
100/0
6
0.22
0.29
88/12
7
0.22
0.29
76/24
D-Fructose
Table 5.2 Experimental variables for 8 hour reflux reactions at 351 K.
5.3.3. Analytical methods
The concentrations of ethyl levulinate and 5-ethoxymethylfurfural, are analysed by
gas chromatography (GC, Agilient Technologies 7820 A GC system) fitted with a
Restek Stabilwax capillary column (30 m, 0.25 mm ID, 0.25 µm), employing
hydrogen carrier gas and a flame ionisation detector. Species are identified by
matching retention-times to known standards, and quantified by calibration of
detector response to known concentrations (using n-octanol as internal standard).
The injection port is maintained at 523 K, a temperature sufficiently high to ensure
the full vaporisation of the expected reaction components. A temperature program of
40 K increasing to 493 K at a rate of 20 K per minute, remaining isothermal at 493 K
for 5 minutes is found to achieve adequate separation of these species from the
ethanol/water media. GC-MS analysis is also employed for the identification of
sample species using an Agilient 5975C MSD, which uses a HP-5MS column (30 m,
0.25 mm ID, 0.25 µm) otherwise employing the same variables as for GC–FID
137
Chapter 5
analysis. For GC analysis, a known mass (50 ± 5 mg ) of analyte is extracted from
the reaction media into 0.4 g of room temperature acetone and 0.8 g of 0.16 mg/g noctanol in acetone, this is followed by the neutralisation of any remaining acid by the
addition of 50 mg of NaHCO3. This dilution and cooling procedure ensures that the
chemical reaction is effectively quenched. This sample is then filtered through 13
mm thick, 2 µm pore size syringe filters (Acrodisc) to remove any insoluble humic
substances that may have been formed, and 1µl of the resulting solution is injected
into the sample inlet port of the GC.
Identification and quantification of D-fructose, D-glucose, 5-hydroxymethylfurfural
and the various sugar-type derivatives is performed on an ion exchange liquid
chromatography system (IC) system (Dionex Corp., Sunnydale, CA) equipped with a
pulsed amperometric detector (AS, 10 µL sample loop, Dionex Corp., Sunnydale,
CA). Analysis is performed at 291 K by isocratic elution with deionised water (18.2
MΩ.cm at a flow rate of 1.1 mL/min) using a Dionex CarboPac PA1 carbohydrate
column. The column is reconditioned using a mixture of 0.4 mol/L sodium
hydroxide and 0.24 mol/L sodium acetate after each analysis. A 25 mg portion of the
sampled reaction media is diluted with 1.0 g of deionised water. As before, 50 mg of
NaHCO3 is added to neutralise any acid present. This sample is filtered as described
above before being analysed. D-fructose, D-glucose, and 5-hydroxymethylfurfural
concentrations are determined by detector calibration to mass prepared standard
solutions.
In all experiments the quantity of “humins” formed is very small, and so are only
determined at the completion of each reaction, when the reaction mixture is filtered
through glass fibre paper (Whatman, grade GF/B 2.7 µm). The filter paper is
subsequently washed with ethanol and placed in an oven for 24 hours at 378 K. The
138
Chapter 5
mass of the material remaining on the paper is determined by difference and referred
to as “humins”.
The pH of the reaction samples is also determined in order to measure hydrogen
cation concentrations [H+]. An Orion pH Ag/AgCl glass electrode fitted to a VWR
Symphony SB70P pH meter is employed. For pH measurements, 0.4 g samples
taken from the reaction media are diluted with 10 g of deionised water. The pH
meter is calibrated against buffer solutions (VWR 32032.291) of known [H+]. The
pH of samples is variable with reaction time and condition but is always in the range
of 1.7-2.4.
5.3.4. Measurement uncertainties
A reproducibility and repeatability study of Test #1 shows the overall experiment-toexperiment variability to be ± 12%, which is comparable to the majority of the
uncertainty estimates below. For GC analysis, experimental measurement
uncertainties are; ethyl levulinate (± 9.6%), 5-ethoxymethylfurfural (± 8.2%), Dfructose (± 5.6%) and 5-hydroxymethylfurfural (± 10.1%). Uncertainties in reported
[H+] are generally ± 4.5%. In addition to the several identified chemical species
discussed in Section 5.4.2.1 five discrete components separated and detected by IC
analysis cannot be identified by retention time matching to expected sugar
derivatives for which analytical standards are available. Figure 5.2 marks these
detections at 1.66, 1.81, 2.11, and 2.41 and 15.27 minutes for a representative
chromatogram. These species are termed “unknown # 1-5” for the purposes of
discussion. By testing and elimination, it is determined that these detections are not
due to the following compounds; ethyl levulinate, 5-ethoxymethylfurfural, levulinic
acid, formic acid, dihydroxyacetone, ethyl formate, ethyl α/β-D-glucopyranoside,
139
Chapter 5
H2SO4, furfural, or any species that result from a series of dummy reactions
comprising;
H2SO4/ethanol,
and
each
of
5-hydroxymethylfurfural,
5-
ethoxymethylfurfural, and ethyl levulinate at 351 K for 480 minutes at the most
extreme reaction conditions listed in Table 5.2. In this way, it is determined that the
unknown species originate from the reaction of D-fructose. As they account for a
considerable amount of the total ion chromatograph signal (see Table 5.3), their
identity is worthy of some speculation.
300
ethyl fructosides
(unknowns #1 & 2)
1.61 and 1.81 mins
Detector Response
250
unknown #3 & #4
2.11 and 2.41 mins
5-hydroxymethylfurfural
D-fructose 12.50 mins
3.4 mins
200
150
D-mannose
100
D-glucose
10.4 mins
(unknown #5)
15.27 mins
7.8 mins
50
0
0
2
4
6
8
10
12
14
16
18
Time / Minutes
Figure 5.2 An exemplar ion chromatogram typical of those obtained for Tests #1-7.
140
Chapter 5
Gas chromatography
(mol/L)
Ion chromatography
(mol/L)
Ion chromatography
(% of total detectable area)
Mass % of
stoichiometr
ic reaction
of D-fructose
with ethanol
accounted
for
Time
(min)
Ethyl
levulinate
5EMF
D-Fructose
5-HMF
Ethyl fructosides
(Unknowns #1-2)
Unknown #3
Unknown #4
Unknown #5
60
0.001
0.016
0.022
0.044
0.044
3.1
7.8
18.5
120
0.005
0.042
0.019
0.051
0.029
3.4
7.1
10.1
51.5
180
0.010
0.066
0.018
0.053
0.026
3.6
6.4
7.2
60.5
240
0.015
0.082
0.017
0.055
0.024
3.3
5.0
4.5
66.9
300
0.018
0.098
0.015
0.051
0.022
3.2
4.8
2.8
71.6
360
0.022
0.118
0.014
0.049
0.020
3.2
3.4
2.5
77.6
420
0.024
0.124
0.015
0.050
0.018
3.0
3.7
1.6
80.7
480
0.028
0.132
0.015
0.049
0.018
2.9
3.0
1.3
83.9
44.8
Table 5.3 Exemplar gas chromatography and ion chromatography data for Test #1 (0.29 mol/L D-fructose, 0.22 mol/L H2SO4, 351 K, in
100% ethanol). This data is also representative of Tests #2-7. Note abbreviations 5-hydroxymethylfurfural (5-HMF) and 5ethoxymethylfurfural (5-EMF).
141
Chapter 5
To separate carbohydrates, the PA1 Carbopac column exploits their weakly acidic
nature. At high pH values (supplied by the sodium hydroxide mobile phase) the
carbohydrates are partially ionised and can be separated by the anion exchange
mechanisms embedded on the column. More acidic carbohydrates bind more
strongly to the column and are retained for longer times. Table 5.4 demonstrates the
correlation of sugar pKa to retention time for a series of standard carbohydrates
tested.
Retention Time
Carbohydrate
pKa
methyl glucoside
13.71
1.61
D-galactose
12.39
6.7
D-glucose
12.28
8.2
D-xylose
12.15
10.23
D-mannose
12.08
11.53
D-fructose
12.03
12.52
Table 5.4.
(Minutes)
Carbohydrate pKa verses retention time at the ion exchange
chromatography conditions of this study.
5.3.5. Identification of unknown species
Of the five unknowns marked in Figure 5.2 (1.61, 1.81, 2.11, 2.41 and 15.27 mins),
it may thus be concluded that #1-4 are of higher pKa than D-fructose or D-glucose.
By analogy to the glucose/methanol/acid studies of Saravanamurugan et
al.(Saravanamurugan and Riisager 2013), who provide evidence of the formation of
various methylated pyranosides and furanosides as intermediate species; it is
speculated that the species that are eluted before 2 minutes are ethyl
fructopyranoside or fructofuranoside species. It is clear from the temporal evolution
of these identities that they are intermediates in the formation of the desired fuel
142
Chapter 5
components. Only Ethyl α-D-glucopyranoside (CAS 34625-23-5) and ethyl β-Dglucopyranoside analytical standards are presently available (Carbosynth Ltd.) and
show very similar retention times of 1.68 min and 1.72 minutes respectively under
the conditions of separation. This behaviour is consistent with the elutions at 1.61
and 1.81 minutes being similar such as C8H16O6 ethyl pyranosides or analogous ethyl
furanoside isomers (for which standards are not available).
300
Detector Response Units
250
200
150
D-galactose
Note 1 Response factor for estimating
D-xylose
k= 294
k=286
D-mannose k=262
D-glucose k=252
D-fructose k=218
concentration of unknown #3, k= 262  30
ethyl glucopyranoside
k=128
5-hydroxymethylfurfural k=106.8
dihydroxyacetone
k=78
100
50
Note 2. Response factor for estimating
concentration of unknowns #1& 2, k=104 25
0
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Concentration mg/ml
Figure 5.3 The pulsed ampermetric detector response for carbohydrate standards.
As only very small quantities of D-glucose are detected in any of the experimental
tests, it suggests that these entities are more likely fructose derived ethyl
fructofuranoside and/or fructopyranoside, C8H16O6 isomers. Henceforth these
substances will be termed “ethyl fructosides”.
As ethyl fructoside analytical
standards are unavailable, these moieties are quantified by standard preparations of
ethyl α-D-glucopyranoside and ethyl-β-D-glucopyranoside, which show exactly
equivalent response factors on the pulsed amperometry detector (Figure 5.3).
Correlation of pKa to retention time infers that unknown #5 (15.27 mins) is more
acidic in nature than D-fructose (12.5 mins) and is thus likely be a stable sugar
143
Chapter 5
intermediate. We speculate that unknown #5 is a C6 sugar species intermediate
between D-fructose and 5-hydroxymethylfurfural. A detailed discussion supporting
this is provided below.
Unknown #3 and #4 - Hexose Derivatives
The pulsed amperometry detector operates by noting the electrical current generated
from analyte oxidation at the surface of a gold electrode. It employs a low detection
voltage (0.023 MV) such that only the easily oxidizable functionalities that
characterise carbohydrates are detectable at sensitivities an order of magnitude
greater than for other classes of analyte. As a result, it can be inferred with a strong
degree of confidence that unknowns #3 and 4 (2.11 and 2.41 mins) are either
carbohydrates, or similar derivatives with carbohydrate functionalities, eluting very
close to known sugar decomposition products such as dihydroxy acetone (2.78
mins), and are thus also likely hexose derivatives. Unknowns #3 and #4 are not
considered in the mechanistic analysis.
Unknown #5
Angyal et al. (Angyal 1969) proposed that at 293 K, D-fructose in water equilibrates
to β-D-fructopyranose (68.2%), α-D-fructofuranose (22.4%), β-D-fructofuranose
(6.2%) and α-D-fructopyranose (2.7%). Many other distributions of these species in
aqueous systems and accounting mechanisms have been postulated in relation to the
synthesis of 5-hydroxymethylfurfural (Caratzoulas and Vlachos 2011; Assary et al.
2012; Zhang and Weitz 2012). There is a general consensus amongst the studies
conducted in water (Kimura et al 2011; Akien et al 2012; Assary et al 2012; Swift et
al 2013) that α/β-D-fructofuranose is the dominant species (with the open chain form
of the sugar also receiving consideration). The recent computational works of Assary
144
Chapter 5
et al. (Assary et al 2012) provide a somewhat independent perspective in relation to
the tautomeric behaviour of fructose in water, also concluding that α/β-Dfructofuranose is the thermodynamically favoured isomer.
However, there is lack of a similar understanding regarding the tautomeric
behaviours of D-fructose in non-aqueous media. In the only study conducted to date
in relation to the tautomeric equilibrium of fructose in an alcohol solvent, Kimura et
al. (Kimura et al 2013) report on the distribution of fructose tautomers in methanol
solvent at 343-423 K without any catalyst. With an iso-topic labelling approach, they
found that D-fructose in methanol resulted in various anhydro-α/β-D-fructofuranoses
in addition to the four furanose and pyranose isomers proposed by Angyal et al. The
authors note that these anhydride species are not present in water at equivalent
conditions.
Feng et al. conducted an investigation to determine the proton affinities and acid
dissociation constants of sugars, calculating theoretical pKa values of various
fructofuranoses and fructopyanose isomers in water from theory (Feng et al 2013).
Their calculations suggest that fructofuranoses are of higher pKa than
fructopyranoses, thus suggesting that unknown #5 is not a fructofuranose structure.
However owing to the large deviation between theoretical and experimental pKa
values, as noted by these authors, this is by no means a conclusive position.
In light of the literature discussion above, suggesting that fructofuranose structures
are the main tautomers responsible for 5-hydroxymethylfurfural formation in
fructose/water systems, it is assumed for kinetic modelling that α/β-D-fructopyranose
must undergo isomerisation to α/β-D-fructofuranose (and/or anhydro fructofuranose
structures) before any 5-hydroxymethylfurfural is formed. For Mechanism #4
145
Chapter 5
unknown #5 (intermediate 1) is modelled to be a fructofuranose structure, the
likelihood of which is informed by the kinetic model which is discussed in the text.
5.3.6. Quantification of unidentified species
As these molecular identities are unknown, their concentrations are estimated
considering the average detector response of all tested carbohydrates and
carbohydrate-derived standards; see Figure 5.3 for complete details.
Detector
response values (k) are obtained for the calibration of mass prepared standards of
known carbohydrates and oxygenated hydrocarbons at a series of known
concentrations in water. Figure 5.3 shows the detector response factor for each
carbohydrate. Note that all five tested carbohydrate species show similar response
factors, and are each distinct from those of the oxygenated hydrocarbons. This data
is used to estimate the detector response factors for the unknown species. Unknowns
#1 and #2 elute on the void of the column and are speculated to be ethyl fructoside
species and are thus quantified according to the calibration of detector response to
the ethyl glucoside standard. Note a small peak derived from both H2SO4 and ethanol
is eluted at a retention time of 1.61 minutes (same retention time as ethyl glucoside
species), however control reactions with H2SO4/ethanol show that this peak area
(detector response) is small in comparison to the peak area generated at 1.61 minutes
by the reaction with H2SO4/D-fructose/ethanol. This is incorporated into the
quantification of unknown #1 by subtracting areas from the pulsed amperometry
detector generated from control reactions with H2SO4/ethanol only. Unknown #5 is
speculated to be a carbohydrate fructofuranose (or anhydro fructofuranose) species
for kinetic modelling purposes. The detector response factors of all known
carbohydrate species fall in the range (218-294, Figure 5.3), the similar
146
Chapter 5
fructofuranose carbohydrate is thus quantified by assuming an average carbohydrate
response factor of 262.2. The corresponding quantitative uncertainties are estimated
as; unknowns #1 and #2 (ethyl fructosides, ±5%), unknowns #3 and #4 (±2%) and
unknown #5, ±12%.
5.4. Results and discussion
5.4.1. Experimental observations and reaction mechanism
Figure 5.4 depicts a proposed reaction mechanism for the formation of ethyl
levulinate and 5-ethoxymethylfurfural from D-fructose that is derived from both the
experimental measurements reported in this study and those conducted by others.
Figures 5.5-5.11 show the temporal evolution of the major species for representative
ethanol and ethanol/water conditions.
5.4.2. Trace species, not considered in mechanistic analysis
Trace
amounts
of
D-mannose,
D-glucose,
dihydroxyacetone,
and
5.5’(oxybis(methylene)bis-2-furfural are observed in each test condition #1-7.
Levulinic acid and furfural are also present at <1% and <3% of the initial D-fructose
mass respectively Very small quantities of dark insoluble humic products are also
formed but never exceed >1.9% of the initial D-fructose mass. Humins formation
from hexose sugars, such as D-fructose, is viewed as a rate limiting reaction for the
production of high yields of biofuels (Alam et al 2012) in entirely aqueous systems.
It is noteworthy that in the ethanol systems of this study, the quantities of humins are
grossly lower than those of comparative aqueous systems. Table 5.5 shows mass
fractions of species detected but are not considered in the mechanistic analysis.
147
Chapter 5
As all the species listed above are present only in very minor quantities, they offer no
significant numerical constraint and are thus not considered in the mechanistic
analysis.
Figure 5.4 Suggested reaction mechanism for the hydrolysis of D-fructose in
condensed phase ethanol/H2SO4.* Depicts species detected with heterogeneous
catalysis.
5.4.3. Major species, considered in mechanistic analysis
1. D-fructose
148
Chapter 5
Time/ Minutes
Species
60
120
180
240
300
0.98
1.28
1.34
1.51
Unknown 3 0.36
5.10
4.34
4.32
3.56
Unknown 4 5.80
1
3.87
2.50
1.99
0.59
Unknown 5 5.85
1.96
1.98
1.50
0.99
0.96
Furfural
Humins
1.41
1.44
1.28
1.25
Unknown 3 0.99
3.28
1.38
1.39
1.28
Unknown 4 4.07
2
2.58
1.87
0.91
0.37
Unknown 5 3.64
n/a
n/a
1.32
0.46
0.24
Furfural
Humins
1.40
1.14
0.83
0.61
Unknown 3 0.77
3.07
2.36
1.22
2.64
0.67
Unknown 4
3
0.53
0.10
0.03
0.03
Unknown 5 1.89
0.12
0.10
n/a
0.12
0.12
Furfural
Humins
Unknown 3
1.4
1.36
1.48
1.00
0.86
Unknown 4 4.32
3.16
1.50
1.08
1.22
4
Unknown 5
4.1
2.50
1.70
1.00
0.32
n/a
n/a
n/a
0.25
0.25
Furfural
Humins
Unknown 3 0.92
1.46
1.39
1.20
1.27
Unknown 4 4.23
3.24
1.24
1.32
1.14
5
Unknown 5 3.25
2.44
1.83
0.90
0.35
1.45
0.62
0.59
0.06
0.12
Furfural
Humins
0.60
0.75
0.95
0.95
Unknown 3 0.38
4.07
4.25
4.59
3.95
Unknown 4 3.92
6
2.24
3.02
2.12
2.30
Unknown 5 3.47
2.93
1.50
1.00
1.6
1.15
Furfural
Humins
n/a
0.130
0.22
0.37
0.43
Unknown 3
1.08
1.25
1.62
2.98
Unknown 4 0.93
7
1.24
1.634
2.13
2.31
Unknown 5 1.44
n/a
n/a
n/a
n/a
n/a
Furfural
Humins
Table 5.5 Mass fractions of species detected but not considered
Test #
360
420
1.32
2.9
0.53
0.67
1.26
2.71
0.61
0.39
1.20
0.88
0.20
n/a
0.94
0.79
0.17
n/a
0.45
0.00
0.01
0.12
0.32
0.00
0.00
0.12
1.12
0.92
0.22
n/a
0.94
0.76
0.22
n/a
1.25
0.81
0.18
n/a
0.87
0.75
0.16
n/a
1.03
3.90
2.03
1.08
0.91
3.68
1.99
0.85
0.58
2.89
2.28
n/a
0.66
2.87
2.63
n/a
480
1.25
2.52
0.51
0.29
1.22
0.65
0.51
0.16
n/a
1.42
0.26
0.00
0.00
0.12
1.89
0.54
0.46
0.14
n/a
1.54
0.66
0.54
0.16
n/a
1.76
1.05
3.4
1.83
0.81
0.53
0.68
3.01
2.74
n/a
0.23
in the mechanistic
analysis (with the exception of unknown #5) as the detected quantities are not
appreciable,( <6 mass % in the worst case). Data reported per time, per test and
presented as a mass % of the initial D-fructose used.
149
Chapter 5
The starting reactant, D-fructose, is a mixture of α/β-D-fructopyranose. It is
converted to the intended products through the series of intermediate species, and
chemical reaction as is discussed in this section.
Amongst studies conducted in water (Caratzoulas and Vlachos 2011; Assary et al
2012; Kimura et al 2013) there is a general consensus that α/β-D-fructofuranose is
the dominant intermediate species responsible for 5-hydroxymethylfurfural
formation (the open chain form of the sugar also receives consideration).The recent
computational works of Assary et al.(Assary et al 2012) provide a somewhat
independent perspective in relation to the tautomeric behaviour of fructose in water,
also concluding that α/β-D-fructofuranose is the thermodynamically favoured
isomer. In light of the literature discussion above, suggesting that fructofuranose
structures are the main tautomer responsible for 5-hydroxymethylfurfural formation
in fructose/water systems, for the purposes of kinetic modelling it is assumed that
α/β-D-fructopyranose isomerises to α/β-D-fructofuranose or the anhydro form which
is labelled as “intermediate species”.
As is discussed later, comparison of kinetic model calculations to experimental data
quantitatively assesses this assumption.
2. 5-hydroxymethylfurfural
Significant quantities of 5-hydroxymethylfurfural (~ 20% of the initial D-fructose
mass) are detected in each of the experimental tests, in both ethanol and
ethanol/water
media.
As
would
be
expected,
larger
fractions
of
5-
150
Chapter 5
hydroxymethylfurfural are observed in the more aqueous media and the complex
chemistry responsible is the subject of extensive on-going research. (Caratzoulas and
Vlachos 2011; Zhang and Weitz 2012; Kimura et al 2013; Swift et al 2013). The
model is intentionally configured to objectively consider the formation of 5hydroxymethylfurfural from two sources. Firstly, the direct dehydration of the
fructopyranose starting material:
and secondly, by the dehydration of the suggested fructofuranose or anhydro form
species (Unknown # 5/intermediate species):
3. Alkyl fructosides
Alkyl fructosides have been widely described as an ethylation product of D-fructose
(Liu et al 2012; Démolis et al 2014). For example, the formation of methyl
fructosides was recently reported by Saravanamurugan et al.(Saravanamurugan et al
2013), apparently identifying β-D-fructopyranoside, β-D-methyl fructofuranoside and
α-D-methyl fructofuranoside, when using H-USY zeolites at 393 K. Liu et al. (Liu et
al 2012) also note alkyl fructoside formation, when working in ethanol media, ethyl
fructofuranosides and ethyl fructopyranosides were observed as dehydration
products of D-fructose using NaCl as a catalyst at 373 K. Alkyl fructosides were also
detected as a product of D-fructose in a recent study conducted by Tucker et al.
(Tucker et al 2013). In the system studied here, ethyl fructosides are detected at the
151
Chapter 5
earliest reaction times sampled, apparently as D-fructose solvates with ethanol. Two
distinct peaks, at 1.61 and 1.81 minutes, are apparent in the ion-exchange
chromatogram (Figure 5.2). Thus, it is likely that “ethyl fructosides” comprise both
ethyl fructopyranosides and ethyl fructofuranosides (208.21 g/mole). The isomers
are not distinguished in the mechanistic analysis, where the ratio of
pyranosides/furanosides is not partitioned. Ethyl fructoside species are assumed to be
formed by the following empirical reaction:
As discussed later, in order to reproduce the experimental data, the model shows that
the reverse reaction must be considered as a significant process:
4. 5-hydroxymethylfurfural
The ethylation of 5-hydroxymethylfurfural to 5-ethoxymethylfurfural is the most
accepted pathway accounting for the consumption of 5-hydroxymethylfurfural in an
ethanol medium (Saravanamurugan et al 2011; Climent et al 2014). It is noted that
Balakrishnan et al.(Balakrishnan et al 2012) and Hu et al. (Hu et al 2013) observed
5-hydroxymethylfurfural diethylacetal when working with solid acid catalysts,
concluding that this species is an intermediate in the formation of 5ethoxymethylfurfural. 5-hydroxymethylfurfural is considered to react with one
molecule of ethanol to form one molecule of 5-ethoxymethylfurfural:
152
Chapter 5
Figures 5.5(a)-5.11(a) show that the ethyl fructoside concentration is transient.
Following the suggestions of Demolis et al. (Démolis et al 2014) 5ethoxymethylfurfural is also considered to be produced by the dehydration of the
ethyl fructoside species:
5. Ethyl levulinate
The model considers ethyl levulinate to form stoichiometrically from the hydration
of 5-ethoxymethylfurfural consistent with the prevalent position in the literature
(Peng et al 2011; Saravanamurugan et al 2011; Démolis et al 2014).
Here, it is worth noting that Balakrishnan et al. (Balakrishnan et al 2012) report the
acetalisation of 5-ethoxymethylfurfural to 5-ethoxymethylfurfural diethylacetal at
348 K using Amberlyst-15) to be competitive with the intended hydration to ethyl
levulinate. Though we do not preclude the formation of 5-ethoxymethylfurfural
diethylacetal, as the species is not detected, it is not considered in the mechanistic
analysis. Later, when discussing the model calculations, we show that this
imposition results in the model calculated ethyl levulinate fractions to sometimes
vary by approximately a factor of two relative to experiment when the 5ethoxymethylfurfural concentration is very accurately reproduced, indicating a
deficiency in this particular mechanistic proposition suggesting that there is another
pathway accounting for ethyl levulinate formation. Figure 5.6 provides a glossary of
structures as discussed in this Section.
153
Chapter 5
Species
Structural Formula
Comments
Starting
α/β-D-fructopyranose
material
used
in
experiments. Referred to as
D-
fructose”.
The most prominent fructose
α/β-D-fructofuranose
tautomer
responsible
for
5-
hydroxymethylfurfural
formation.
α/β-D-glucopyranose
Referred to in the text as “Dglucose”.
One of two isomers referred to
α/β-D-ethyl fructofuranoside
in text as “ethyl fructosides”
together
with
α/β-D-ethyl
pyranoside below.
One of two isomers referred in
α/β-D-ethyl fructopyranoside
text
as
together
“ethyl
with
fructosides”
α/β-
D-ethyl
furanoside above.
An examplar anhydro-fructose
species as reported by Kimura et
1,6-anhydro-α-D-fructofuranose
al. in methanol solution. A
purported stable intermediate
between
D-fructose and 5-
hydroxymethylfurfural.
5-hydroxymethylfurfural
diethylacetal
The
5-hydroxymethylfurfural
diethylacetal structure reported
by Balakrishnan et al. 2012.
The
5-ethoxymethylfurfural
diethylacetal structure reported
5-ethoxymethylfurfural
diethylacetal
by Balakrishnan et al. 2012 and
Hu et al. 2012.
Table 5.6 Glossary of carbohydrate structures and notes on terminology used in text.
154
Chapter 5
0.20
1.4
Concentration, mol/L
Concentration, mol/L
1.6
intermiediate species
ethyl fructosides
5-hydroxymethylfurfural
5-ethoxymethylfurfural
ethyl levulinate
0.25
0.15
0.10
140
mass accounted for
formic acid
water
ethanol consumed
(b)
120
100
1.2
80
1.0
60
0.8
0.6
40
0.4
20
0.05
0.2
0.00
0
% mass of stoichiometric reaction
of D-fructose with ethanol
1.8
(a)
D-fructopyranose
0.30
0.0
0
100
200
300
400
0
500
100
200
300
-20
500
400
Time/ Minutes
Time / Minutes
Figure 5.5 Species fractions for Test #1, D-fructose/H2SO4 (0.29/0.09 mol/L) at 351
K, in 100% ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines).
0.20
140
mass accounted for
formic acid
water
ethanol consumed
1.6
1.4
0.15
0.10
0.05
(b)
120
100
1.2
80
1.0
60
0.8
0.6
40
0.4
20
0.2
0.00
0
% mass of stoichiometric reaction
of D-fructose with ethanol
Concentration, mol/L
0.25
1.8
(a)
Concentration, mol/L
D-fructopyranose
intermediate species
ethyl fructosides
5-hydroxymethyl furfural
5-ethoxymethyl furfural
ethyl levulinate
0.30
0.0
0
100
200
300
400
500
Time / Minutes
0
100
200
300
400
-20
500
Time/ Minutes
Figure 5.6 Species fractions for Test #2, D-fructose/H2SO4 (0.29/0.22 mol/L) at 351
K, in 100% ethanol. Mechanism #4.
Experimental values (symbols), model
calculations (lines).
155
Chapter 5
1.6
1.4
Concentration, mol/L
0.20
0.15
0.10
140
mass accounted for
formic acid
water
ethanol consumed
(b)
120
100
1.2
80
1.0
60
0.8
0.6
40
0.4
20
0.05
0.2
0.00
0
% mass of stoichiometric reaction
of D-fructose with ethanol
intermediate species
ethyl fructosides
5-hydroxymethylfurfural
5-ethoxymethylfurfural
ethyl levulinate
0.25
Concentration, mol/L
1.8
(a)
D-fructopyranose
0.30
0.0
0
100
200
300
400
0
500
100
200
300
-20
500
400
Time/ Minutes
Time / Minutes
Figure 5.7 Species fractions for Test #3, D-fructose/H2SO4 (0.29/0.32 mol/L) at 351
K, in 100% ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines).
(b)
100
0.08
0.05
0.03
80
60
0.4
40
0.2
20
0
0.00
% mass of stoichiometric reaction
of D-fructose with ethanol
120
0.6
Concentration / M
0.10
140
mass accounted for
formic acid
water
ethanol consumed
0.8
intermediate species
ethyl fructosides
5-hydroxymethylfurfural
5-ethxoy methyllfurfural
ethyl levulinate
0.13
Concentration, mol/L
(a)
D-fructopyranose
0.15
0.0
0
100
200
300
Time / Minutes
400
500
0
100
200
300
400
-20
500
Time/ Minutes
Figure 5.8 Species fractions for Test #4, D-fructose/H2SO4 (0.15/0.11 mol/L) at 351
K, in 100% ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines).
156
Chapter 5
0.22
0.15
0.07
SI6(b)
120
100
80
1.5
60
1.0
40
20
0.5
0
0.00
% mass of stoichiometric reaction
of D-fructose with ethanol
2.0
Concentration, mol/L
0.30
140
mass accounted for
formic acid
water
ethanol consumed
2.5
intermediate species
ethyl fructosides
5-hydroxymethylfurfural
5-ethoxymethylfurfural
ethyl levulinate
0.38
Concentration, mol/L
SI6(a)
D-fructopyranose
0.45
0.0
0
100
200
300
400
500
0
100
Time / Minutes
200
300
-20
500
400
Time/ Minutes
% mass of stoichiometric reaction
of d-fructose with ethanol
Figure 5.9 Species fractions for Test #5, D-fructose/H2SO4 (0.43/0.34 mol/L) at 351
K, in 100 % ethanol. Mechanism #4. Experimental values (symbols), model
calculations (lines).
1.6
1.4
Concentration, mol/L
0.20
0.15
0.10
140
mass accounted for
formic acid
water
ethanol consumed
(b)
120
100
1.2
80
1.0
60
0.8
0.6
40
0.4
20
0.05
0.2
0.00
0
% mass of stoichiometric reaction
of d-fructose with ethanol
intermediate species
ethyl fructosides
5-hydroxymethylfurfural
5-ethoxy methylfurfural
ethyl levulinate
0.25
Concentration mol/L
1.8
a)
D-fructopyranose
0.30
0.0
0
100
200
300
400
Time / Minutes
500
0
100
200
300
400
-20
500
Time/ Minutes
Figure 5.10 Species fractions for Test #6, D-fructose/H2SO4 (0.29/0.22 mol/L) at
351 K, in 88/12 mass % ethanol/water. Mechanism #4. Experimental values
(symbols), model calculations (lines). Note, water concentrations shown are
additional to the 12% water media.
157
Chapter 5
1.6
1.4
Concentration, mol/L
0.30
0.25
0.20
0.15
0.10
400
60
0.8
0.6
40
0.4
20
0.0
300
100
80
0.00
200
120
1.0
0.2
100
5(b)
1.2
0.05
0
140
mass accounted for
formic acid
water
ethanol consumed
500
0
0
100
Time / Minutes
200
300
400
% mass of stoichiometric reaction
of D-fructose with ethanol
intermediate species
ethyl fructosides
5-hydroxymethylfurfural
5-ethoxymethylfurfural
ethyl levulinate
0.35
Concentration, mol/L
1.8
5(a)
D-fructopyranose
0.40
-20
500
Time/ Minutes
Figure 5.11 Species fractions for Test #7, D-fructose/H2SO4 (0.29/0.22 mol/L) at
351 K, in a 76/24 ratio by mass % ethanol/water ratio. Mechanism #4. Experimental
values (symbols) model calculations (lines).
5.4.4. Hydrogen cation concentrations
The reaction rate accelerating influence of hydrogen cations is embodied in the
kinetic model by the imposition of an empirically determined hydrogen cation
concentration, [H+]. H2SO4 dissociation in 100% water is well defined by Sippola
(Sippola 2012) and described by the following reactions:
In aqueous solution R1 is fast and the acid fully dissociates (as H3O+ fully
dissociates to H+ + H2O), however HSO4- does not completely dissociate (pKa 1.92)
(Sippola 2012). No robust kinetic or thermodynamic information is available for
H2SO4 dissociation in an ethanol medium. Consequently, the [H+] measurements are
evaluated using the reaction equilibrium constant parameters derived by Que et al.
(Que et al 2011) for H2SO4 dissociation in water. The relations of Que et al. describe
the equilibrium of R1 and R2 as a function of temperature and concentration and are
implemented in a numerical model through Matlab with a balance of each atomic
158
Chapter 5
species. By this procedure, any change in equilibrium of R1 and R2 due to changes
in temperature and concentration when samples are diluted with water and cooled to
298 K during the analytical quenching process are accounted for.
Using these equilibrium constants, a 0.22 mol/L H2SO4 solution in water at 351 K
dissociates to 0.36 mol/L [H+], rather than 0.44 mol/L [H+] as may be commonly
assumed. To provide more meaningful representations, the [H+] determined
experimentally are presented in Figure 5.12 by normalisation to the Que et al.
equilibrium [H+] for the equivalent water systems. This presentation assumes that the
relations derived by Que et al. for aqueous systems are also valid for ethanol
systems.
From Figure 5.12, in the entirely aqueous medium, the measured [H +] remains
consistent with the dissociation values of Que et al. for the duration of the test. Note
that the initial maximum observed experimentally in aqueous systems is closely, but
not exactly consistent, to the suggested equilibrium value. In ethanol media, two
features are noted. Firstly the peak [H+] dissociated is ~70-90% of that of the
aqueous system. This is to be expected due to the decrease in dielectric constants
between water and ethanol (Bell 1973). More importantly, for all conditions studied
it is evident that the [H+] is a dynamic function of time, showing an initial rapid
decrease in the first 30 minutes of reaction followed by a further more gradual
decrease until an equilibrium is reached.
159
Measured hydrogen cation concentration normalised to
the theoretical H2SO4 dissociation at 351K in water
Chapter 5
100
80
60
40
20
0
0.09 M H2SO4 (100/0)
0.22 M H2SO4 (100/0)
0.32 M H2SO4 (100/0)
0.22 M H2SO4 (76/24)
0.22 M H2SO4 (88/12)
0.22 M H2SO4 (0/100)
100
200
300
400
500
Time / Minutes
Figure 5.12 Hydrogen cation (H+) behaviour at reaction conditions of Table 5.2
where brackets donate ethanol/water mass ratio. [H2SO4] dissociation equilibrium is
calculated by the relations of Que et al.
GC-MS analysis shows that this degradation in [H+] is associated with the formation
of ethyl hydrogen sulphate, presumably corresponding to the reaction of a protonated
ethanol molecule with the HSO4- ion resulting from R1 above:
Formic acid is expected to be present as an intermediate species. Figures 3-5 and
SI3-SI6 estimate the maximum [formic acid] generated by the reaction system as
0.08 mol/L, compared to [H2SO4] of 0.09 mol/L, 0.22 mol/L, 0.32 mol/L. However,
as formic acid dissociation is weak (pKa 3.75), its contribution to the total [H+] is
insignificant (~1%) in comparison to that of H2SO4. Therefore its dissociation is not
explicitly included in the model, but is implicit in the global [H+] population via the
pH measurements. The following equations are used in the model to embody the
effect of [H+]:
160
Chapter 5
Equations of the form
-x
-x
y = y0 +A1 ×(1-eT1 )+A2 ×(1-eT2 ) are utilised to accurately
incorporate the empirical measurements of [H+] to the kinetic model in a time
dependent manner for the 100% ethanol tests.
-x
⁄
2.1E-3
Test #1 : y = 1.1E-01 - 3.6E-02 ×
-x⁄
1.1E01 )
1-e
(
Test #2 : y = 3.3E-01 - 4.7E-02 × (1-e
- 2.3E-02 × (1-e
)
-x⁄
4.1 )
Test #3 : y = 4.4E-01 - 1.3E-01 × (1-e
-x⁄
7.9E01 )
- 1.1E-01 × (1-e
-x⁄
3.0E2 )
-x
⁄
4.3E-01
- 7.9E-02 ×
1-e
(
)
As [H+] behavior is not as complicated in water, more simple relations are sufficient
to accurately incorporate the empirical measurements of [H+] to the kinetic model:
Test #6: (90 ethanol/ 20 water): y=-0.0002(x)+0.2874
Test#7: (80ethanol/water):
5.4.5.
y=-0.0002(x)+0.3134
Kinetic model
Presently, a detailed and fundamental comprehension of the reaction kinetics
involved in the conversion of sugars to simpler oxygenated or hydrocarbon
molecules is beyond the state-of-the-art (Swift et al 2013). Many efforts are pursuing
this goal in aqueous systems by a variety of experimental, theoretical and modelling
approaches (Kimura et al 2011, Assary and Curtiss 2012; Assary et al 2012; Swift et
al 2013). In those systems, as here, the initial information requiring definition is of
the chemical reaction pathways through which the major portion of the chemical flux
is converted. This information is obtainable through the coupling of experimentally
determined species mass fractions, through a series of conservation equations, into a
161
Chapter 5
chemical model that describes a reaction mechanism with numerical rigour. A
relatively simple reaction mechanism is formulated to provide a basic
comprehension of the system’s chemistry (see Figure 5.13) both from considering
experimental observations from this study and from consulting pertinent
computational and experimental evidence from the literature. In the modelling, a
conservation of D-fructose mass is imposed, constraining the modelling analysis such
that experimental datasets are objective tests with which to evaluate the veracity of
the chemical reaction mechanism proposed.
By evaluating this model against reliable time-resolved experimental data, the
chemical timescales involved in each inter-conversion (empirical rate constants) may
also be estimated. Figure 5.13 presents the reaction mechanism(s) proposed for
testing against the experimental datasets through kinetic modelling. To achieve this,
a set of ordinary differential equations are defined, each assuming a pseudo-first
order dependence to the empirically determined [H+]. Each differential equation is of
the form:
d [ D-fructose]
dt
= - (k1 × [H+ ] × [D-fructose])
which considers D-fructose as an example, where kx designates the rate constant
derived by the model for each reaction, always considering the entire data set of
Table 5.2.
162
Chapter 5
Figure 5.13 Reaction mechanism derived from experimental observations and
kinetic modelling. k7 and k8 are derived from Mechanism #3 for Tests #6-7 where
water is added to the reaction media.
Four discreet chemical mechanisms comprised of the reactions and rate constants
designated by “kx” in Figure 5.13 are iteratively considered. Each mechanism
includes
k1-k3 describing
the
sequential
reaction of
D-fructose,
to
5-
hydroxymethylfurfural, to 5-ethoxymethylfurfural, to ethyl levulinate. Further
reactions are additionally considered to more completely describe the reacting flux:
Mechanism #1: k1-k3.
Mechanism #2: k1-k3, k4, k6.
163
Chapter 5
Mechanism #3: k1-k3, k4, k5.
Mechanism #4: k1-k3, k4, k5, k7, k8, where, k1-5 derived from Mechanism #3
are given to the model as constraints to derive and k7-8.
5.4.6. Computational methods
The rate constants associated with each reaction are determined by the minimization
of errors between the experimental measurements and the kinetic model calculated
species fractions for each set of ordinary differential equations, via multiple linear
regressions. The fminsearch error minimization protocol based on the Nelder-Mead
optimization method (Lagarias et al 1998) is implemented using Matlab to determine
the numerical rate constant term kx that allows for best reproduction of the data. The
following assumptions and rules are applied in formulating the model and are
important when analysing the model calculations to the chemical reality of the
system:
1. Reaction conditions are homogeneous, isobaric and isothermal.
2. The reaction proceeds according to the mechanisms defined in Figure 5.13.
3. The mass owing to the stoichiometric reaction of D-fructose with ethanol is
conserved.
4. Reaction rates are pseudo-first order with respect to the concentrations of
hydrogen cation and sugar or sugar-derived reactants. i.e. the individual
reaction rates are independent of the ethanol/water concentration.
The calculated species fractions and corresponding reaction rate constants are
presented in Figures 5.5(a)-5.11(a) (as lines) and in Tables 5.7-5.10. The degree to
which the model calculations reproduce the experimental data is determined using
164
Chapter 5
the normalised root mean square error (R2), Tables 5.7-5.10. A mechanism fidelity
index is also considered to provide an objective evaluation of the overall fidelity of
the chemical mechanism to the chemical reality, it is defined as:
D − 𝑓𝑟𝑢𝑐𝑡𝑜𝑠𝑒
𝑚𝑎𝑠𝑠 % 𝑐𝑜𝑛𝑠𝑖𝑑𝑒𝑟𝑒𝑑 𝑥 𝑅 2
The differential equations formulated for each mechanism are displayed below:
Mechanism #1
1.
2.
d [D-fructose]
= - (k1 × [H+ ] × [D-fructose])
dt
d [5-hydroxymethylfurfural]
=
dt
(k1 × [H+ ] × [D-fructose]) – (k2 × [H+ ] × [5-hydroxymethylfurfural])
3.
d [5- ethoxy methyl furfural]
dt
=
([k2 × [H+ ] × [5-hydroxymethylfurfural]) - (k3 × [H+ ] × [5-ethoxymethylfurfural])
4.
d [ ethyl levulinate]
dt
= (k3 × [H+ ] × [5-ethoxymethylfurfural])
Mechanism #2
5.
6.
7.
d [ D-fructose]
= - (k1 × [H+ ] × [D-fructose]) + (k4 × [H+ ] × [D-fructose])
dt
d [ ethyl fructosides]
dt
= (k4 × [H+ ] × [D-fructose]) - (k6 × [H+ ] × [ethyl fructosides])
d [ 5-hydroxymethylfurfural]
=
dt
(k1 × [H+ ] × [D-fructose]) - (k2 × [H+ ] × [5-hydroxymethylfurfural])
8.
d [ 5-ethoxymethylfurfural]
dt
+
(k2 × [H ] × [5-hydroxymethylfurfural]) + (k6 × [H+ ] × [ethyl fructosides]) )
- (k3 × [H+ ] × [5-ethoxymethylfurfural])
9.
d [ ethyl levulinate]
dt
= (k3 × [H+ ] × [5-ethoxymethylfurfural])
165
Chapter 5
Mechanism #3
10.
d [ D-fructose]
=
dt
-(((k1 ×[H+ ] × [D-fructose]) + (k4 × [H+ ] × [D-fructose])) + (k5 × [H+ ] × [ethyl fructosides])
11.
12.
d [ ethyl fructosides]
dt
= (k4 × [H+ ] × [D-fructose]) - (k5 × [H+ ] × [ethyl fructosides])
d [ 5-hydroxymethylfurfural]
=
dt
(k1 × [H+ ] × [D-fructose]) - ((k2 × [H+ ] × [5-hydroxymethylfurfural]))
13.
d [5- ethoxy methyl furfural]
=
dt
(k2 × [H+ ] × [5-hydroxymethylfurfural]) - (k3 × [H+ ] × [5-ethoxymethylfurfural])
14.
d [ ethyl levulinate]
dt
= (k3 × [H+ ] × [5-ethoxymethylfurfural])
Mechanism #4
15.
d [ D-fructose]
=
dt
((k1 ×[H+ ] × [D-fructose]) + (k4 × [H+ ] × [D-fructose])+(k9 × [H+ ] × [D-fructose]))
+ (k5 × [H+ ] × [ethyl fructosides])
16.
17.
18.
d [ intermediate species]
= (k7 × [H+ ] × [D-fructose]) - (k8 × [H+ ] × [ intermediate-species])
dt
d [ ethyl fructosides]
dt
= (k4 × [H+ ] × [D-fructose]) - (k5 × [H+ ]× [ethyl fructosides])
d [5-hydroxymethylfurfural]
=
dt
((k1 × [H+ ] × [D-fructose]) + (k8 × [H+ ] × [intermediate species])) ((k2 × [H+ ] × [5-hydroxymethylfurfural]))
19.
d [5- ethoxy methyl furfural]
dt
=
(k2 × [H+ ] × [5-hydroxymethylfurfural]) - (k3 × [H+ ] × [5-ethoxymethylfurfural])
20.
d [ ethyl levulinate]
dt
= ((k3 × [H+ ] × [5-ethoxymethylfurfural]))
166
Chapter 5
5.4.7. Chemical reaction mechanism and kinetics discussion
Reaction mechanisms #1-4 are tested against the experimental data and their validity
judged by the mechanism fidelity index (see Table 5.7-5.10). Mechanism #1 returns
an average fidelity index of 0.37 across tests #1-5 (see Table 5.7), and as low 0.12
for Test #1. As this mechanism considers only D-fructose, 5-hydroxymethylfurfural,
5-ethoxymethylfurfural and ethyl levulinate, the low fidelity indices indicate the
poor quality of experimental data reproduction. The omission of the intermediate
species and the speculated ethyl fructoside species in particular, invalidates this
mechanism. The poor fidelity index of this simple mechanism highlights the
complexity of the conversion processes involved.
In addition to these species, Mechanisms #2 and #3 both also consider the
importance of ethyl fructoside species in deciphering the mechanism, their formation
is
considered via k4 (Figure
5.13).
Their consumption to
form
5-
ethoxymethylfurfural is considered in Mechanism #2 by the inclusion of k6, whereas
in Mechanism #3, it is considered by allowing the reversible dehydration of ethyl
fructosides to the D-fructose starting material through k5. Across the ethanol media
experiments, Mechanism #2 returns an average fidelity index of 0.52 compared to
0.57 for Mechanism #3 (Tables 5.8 and 5.9). As the same amount of mass is
considered in both mechanisms, Mechanism #3 is thus a more accurate
representation of the chemical reality.
In addition to the species of Mechanism #3, Mechanism #4 considers intermediate
species where the rate constants derived in Mechanism #3 are used as constraints to
accurately evaluate k7-k8. Flood et al. (Flood et al 1996) report a value of 1.5E-3 s-1
for the tautomerisation of fructopyranose to fructofuranose in a 90/10 v/v%
ethanol/water solution at 351 K without [H+]. For the same transformation in 100%
167
Chapter 5
ethanol, the model of this study shows rate constants of 7.5 E-4, which is 50%
slower than reported by Flood et al. in the absence of H+. This suggests that
unknown #5 is not a fructofuranose species and that the purported pyranose/furanose
isomerisation may be too quick to be isolated by the methods employed in this study.
Never the less it’s inclusion in the model as an intermediate between fructopyranose
and 5-hydroxymethylfurfural is important in obtaining a high fidelity model.
Mechanism #4 returns the best fidelity index of those tested, averaging to 0.62 for
Tests #1-5 (Table 5.10) and is thus the focus of discussion.
Table 5.10 shows the reaction rate constants deduced by the kinetic modelling
analysis employing Mechanism #4 for each experimental condition. The degree to
which each rate constants derived from each test condition is consistent provides an
indication of the accuracy of the particular estimation in the context of the reaction
mechanism analysed. Note that the rate constants derived from tests with 100%
ethanol (#1-5) are closely consistent, always within a-factor-of-two (with exception
of k3, a-factor-of-four) in the worst case, and often agree much more closely. It is
noticeable that the rate constants for the dehydrations are approximately an order of
magnitude lower in the aqueous systems than in entirely ethanol systems. For
example the addition of 12 mass % water to the solvent media depresses the reaction
rates by approximately an order of magnitude, the rate being further suppressed by a
further order of magnitude by the addition of 24 mass % water. Note that when in
aqueous media, the reversible reaction of D-fructose to ethyl fructosides favours the
formation of D-fructose (Tests #6-7), but when in an entirely ethanol medium the
equilibrium is adjusted toward ethyl fructoside formation. Thus the addition of water
pushes the equilibrium from ethyl fructoside (k4) to D-fructose (k5).
168
Chapter 5
The model over estimates ethyl levulinate formation by approximately a factor of
two as can be seen in Figures 5.5-5.11. The mathematical error minimisation
functions prioritise the reproduction of 5-ethoxymethylfurfural in preference to ethyl
levulinate. This is so, as the 5-ethoxymethylfurfural data is more constraining to the
model overall, as it bares closer connection to the other species measured, than do
the ethyl levulinate measurements. As the D-fructose mass is conserved by the
model, but experimentally not all D-fructose mass was associated with discrete
chemical entities, over estimations are to be expected.
Table 5.10 suggests that the slowest reaction in the system is the hydration of 5ethoxymethylfurfural to ethyl levulinate. This may present reaction engineering
opportunities to push the reaction towards 5-ethoxymethylfurfural production rather
than towards ethyl levulinate using dehydrating agents. That the model accurately
estimates ethyl levulinate concentrations while overestimating those of 5ethoxymethylfurfural indicates a deficiency in the postulated mechanism, suggesting
that 5-ethoxymethylfurfural is not the sole intermediate responsible for ethyl
levulinate formation. Hu et al. (Hu et al 2013) provide a possible alternative route
suggesting that 5-hydroxymethylfurfural diethylacetal degrades to levulinic acid
which subsequently esterifies by reaction with ethanol to ethyl levulinate. As already
stated, trace quantities of levulinic acid are detected in the reaction system;
examinations of this pathway present an avenue for further study. The mass
conserved modelling approach has developed baseline kinetic parameters and
highlighted mechanism deficiencies and species intermediates to be considered in
future model development.
.
169
Chapter 5
Rate constants with Mechanism #1
Test #
[H2SO4]
(mol/L)
[D-Fructose]
(mol/L)
Ethanol/Water
Mass Ratio
k1 (s-1)
k2 (s-1)
k3 (s-1)
R2
1
0.09
0.29
100
1.1 E-3
1.0 E-3
4.3 E-4
0.20
2
3
4
5
0.22
0.32
0.11
0.34
100
100
100
100
7.6 E-4
6.8 E-4
7.8 E-4
7.6 E-4
8.1 E-4
7.5 E-4
6.2 E-4
6.3 E-4
7.5 E-4
7.5 E-4
1.6 E-4
1.5 E-4
1.7 E-4
1.8 E-4
2.2 E-4
0.67
0.61
0.51
0.50
0.50
0.29
0.29
0.15
0.43
Mean of Tests #1-5
Mechanism
Fidelity
Index
0.12
0.52
0.46
0.39
0.35
0.37
Table 5.7 Rate constants derived from the modelling of Mechanism #1. Rate Constants (k) identification numbers are as described in
Figure 5.13.
Rate constants with Mechanism #2
Test #
[H2SO4]
(mol/L)
1
2
3
4
5
0.09
0.22
0.32
0.11
0.34
[DFructose]
(mol/L)
0.29
0.29
0.29
0.15
0.43
Mean of Tests #1-5
Ethanol/Water
Mass Ratio
k1 (s-1)
k4 (s-1)
k6( s-1)
k2( s-1)
k3 (s-1)
R2
Mechanism
Fidelity
Index
100
100
100
100
100
7.3E-4
6.1E-4
3.1E-4
4.4E-4
1.0E-3
6.2E-4
9.1E-4
2.2E-3
4.3E-4
6.0E-4
1.5E-3
1.1E-3
1.9E-3
3.3E-3
2.0E-3
9.1E-4
5.0E-4
1.7E-3
4.1E-4
2.8E-5
3.3E-4
2.8E-4
2.4E-4
2.6E-4
4.0E-4
8.0E-5
1.3E-4
1.4E-4
1.4E-4
1.8E-4
0.41
0.79
0.69
0.80
0.69
0.68
0.29
0.66
0.53
0.60
0.53
0.52
Table 5.8 Rate constants derived from the modelling of Mechanism #2. Rate Constants (k) identification numbers are as described in
Figure 5.13.
170
Chapter 5
Rate constants with Mechanism #3
Test #
[H2SO4]
(mol/L)
[D -Fructose]
(mol/L)
Ethanol/Water
Ratio
k1 (s-1)
k4 (s-1)
k5 (s-1)
k2 (s-1)
k3(s-1)
R2
Mechanis
m
Fidelity
Index
1
2
3
4
5
0.09
0.22
0.32
0.11
0.34
0.29
0.29
0.29
0.15
0.43
100
100
100
100
100
2.0E-3
1.7E-3
1.4E-3
1.2E-3
1.8E-3
2.2E-3
3.0E-3
2.2E-3
2.8E-3
2.7E-3
3.0E-3
1.6E-3
1.0E-3
2.5E-3
1.1E-3
1.3E-3
8.0E-4
5.5E-4
8.0E-4
7.8E-4
4.5E-4
1.3E-4
1.3E-4
1.5E-4
1.4E-4
0.65
0.83
0.72
0.82
0.74
0.46
0.69
0.55
0.62
0.52
1.6E-3
2.6E-3
1.8E-3
8.5E-4
2.0E-4
0.75
0.57
1.5E-4
5.0E-5
1.3E-4
1.4E-4
1.0E-4
8.7E-4
8.5E-5
1.3E-5
3.0E-5
N/A
0.31
0.81
0.25
0.62
Mean of Tests #1-5
6
7
0.22
0.22
0.29
0.29
88/12
76/24
Table 5.9 Rate constants derived from the modelling of Mechanism #3. Rate Constants (k) identification numbers are as described in
Figure 5.13.
171
Chapter 5
Table 5.10 Reaction rate constants derived from modelling of experimental data with Mechanism #4. Chemical reactions are assigned
to each rate constants (kx) in Figure 5.13 .The yield of 5-ethoxymethylfurfural (EMF) and ethyl levulinate (EL) refers to their combined
mass
at
480
minutes
relative
to
the
mass
of
stoichiometric
reaction
of
D-fructose
with
ethanol.
172
Chapter 5
The importance of the pH measurements to the analysis is to be noted. Neglecting
the transient nature of [H+] in the model, or assuming that its value corresponds to
the fully dissociated, twice [H2SO4], as is common, or assuming it remains otherwise
constant, may not result in accurate kinetic or mechanistic conclusions. To test this
hypothesis, the kinetic model is exercised with Mechanism #4 using the rate
constants deduced from experiment (Table 5.10) as input taking Test #2 as an
example reaction condition Assuming the [H+] is equal to the equilibrium
distributions derived by Que et al.(Que et al 2011) (0.36 mol/L compared to 0.44
mol/L for full dissociation) as a constant rather than a transient term predicts an [5ethoxymethylfurfural] of 0.10 mol/L after 480 minutes compared to 0.16 mol/L
when the transient nature of [H+] is incorporated. This is of integral significance
when compared to the actual experimental value of 0.16 mol/L. It is likely that this
behaviour will also be pertinent to acid hydrolysis in other organic solvents.
The imposition of conservation of D-fructose mass allows the model to make
quantitative time resolved predictions of species that are not measured by experiment
but are defined in the kinetic model, such as ethanol, water and formic acid. These
calculations are displayed in Figures 5.5-5.11 show that the concentrations of ethanol
consumed in the reaction are in the range of 0.1-0.3 mol/L, approximately equivalent
to the concentrations of the D-fructose starting material. It is also interesting to note
that according to this analysis the model estimates water to be formed at a ratio of
approximately 3-4 moles per mole of sugar reactant. As discussed below, significant
quantities of water are shown to retard the overall rate of reaction, but the model
shows the extent of incipient water production not to produce such a significant
effect.
173
Chapter 5
5.4.8. Yield analysis
The suggested reaction mechanism (#4) indicates the global chemical equation
below to account for the formation of 5-ethoxymethylfurfural and ethyl levulinate:
As ethanol is in excess, the yield is reported with respect to the mass % sum of the Dfructose and ethanol stoichiometric reaction to the target molecules of 5ethoxymethylfurfural and ethyl levulinate. For the optimal condition investigated,
Test #3 after 480 minutes, 63% of this mass is valorised to 5-ethoxymethylfurfural
and ethyl levulinate. Though the focus of this study is mechanistic rather than yield
optimisation, comparison with the data obtained by Balakrishnan et al.
(Balakrishnan et al 2012) is relevant. Using 10 mole % silica embedded H2SO4 at
383 K for 24 hours, molar yields of 69% 5-ethoxymethylfurfural and 17% ethyl
levulinate are reported. Assuming the global equations above are also valid for the
Balakrishnan et al. study, the corresponding mass yield is a comparable 59%.
However, it should be noted that the acid to D-fructose molar ratio of Balakrishnan et
al. is ~ 1:10, whereas this study employs the more aggressive 1:1, but with lower
temperatures and pressures and shorter time scales.
5.4.9. Volumetric energy density
The practicality of the synthesis of 5-ethoxymethylfurfural and ethyl levulinate, as
transportation fuels, from ethanol is evaluated by considering the energy valorisation
achieved through reaction with the lignocellulose derived sugar. Note that for ground
transportation fuels the evaluation of energy valorisation is most meaningful on a
174
Chapter 5
volumetric basis. In this regard, the higher mass per volume density of 5ethoxymethylfurfural is of key importance. Consulting Table 1, the enthalpy of
combustion of ethanol is 7.63 kcal/mL, that of 5-ethoxymethylfurfural 8.66 kcal/mL
and ethyl levulinate 7.53 kcal/mL. From this analysis the optimal synthesis achieved
in this study is considered by analysing Figures 5.5-5.11 for the condition and
residence time at which the highest volumetric energy density occurs, considering 5ethoxymethylfurfural and ethyl levulinate as viable fuel components. This occurs for
Test #2 (Figure 5.6(a)) after 420 minutes, producing a 0.16/0.05 mol/L mixture of 5ethoxymethylfurfural/ethyl levulinate, where Figure 5.6(b) shows an ethanol
consumption of 0.26 mol/L. Again, using the model derived R5 and R6; 15.2 mLs
of ethanol is converted to 22.4 mLs 5-ethoxymethylfurfural and 7.2 mLs of ethyl
levulinate. This represents an energy valorisation of an additional ~215% with
respect to the energy density of the ethanol consumed. It is also worth noting that
this value would increase further if other intermediates were to be considered as fuel
components, e.g. 5-hydroxymethylfurfural.
5.5. Conclusions
The advanced biofuel candidates, ethyl levulinate and 5-ethoxymethylfurfural, are
synthesised from α/β-D-fructopyranose (D-fructose) in a condensed phase
homogeneous
sulphuric
acid
ethanol
system
at
351
K. D-fructose,
5-
hydroxymethylfurfural, 5-ethoxymethylfurfural, ethyl levulinate and intermediate
products, are quantified as major species fractions, summing to 45-85% of the initial
fructose mass, with furfural, D-glucose and D-mannose quantified as minor species
fractions, always summing to <5% of the initial fructose mass.
175
Chapter 5
Pursuing a mechanistic comprehension, a kinetic model analysis shows that the
fructopyranose starting material is unlikely to undergo direct transformation to 5hydroxymethylfurfural indicating at least one stable reaction intermediate is required
for 5-hydroxymethylfurfural production. The model further proposes that
experimentally observed ethyl fructoside species are preferentially converted back to
the D-fructose starting material rather than undergoing further hydrolysis to 5ethoxymethylfurfural,
which
is
produced
by
the
ethylation
of
5-
hydroxymethylfurfural. The modelling analysis identifies the hydration of 5ethoxymethylfurfural to ethyl levulinate as the slowest reaction in the system and
that this is not the sole pathway responsible for ethyl levulinate formation. The
presence of water in the solvent medium is shown to aggressively retard the overall
rate of reaction with respect to an entirely ethanol medium. Importantly, in ethanol
media, the hydrogen cation concentration is shown to be a dynamic function of time
and not catalytically recycled. Incorporating this behaviour in the kinetic modelling
analysis is demonstrated to be crucial to deriving meaningful kinetic and mechanistic
parameters.
The optimal condition studied shows the synthesised fractions of ethyl levulinate and
5-ethoxymethylfurfural, considered as fuel components, to achieve a volumetric
energy valorisation (HCombustion) of 215% with respect to the ethanol consumed by
reaction.
5.6. References
Akien, G.R., Qi, L., Horváth, I.T. (2012) “Molecular mapping of the acid catalysed
dehydration of fructose,” Chemical Communications, 48(47), 5850–5852.
176
Chapter 5
Alam, M.I., De, S., Dutta, S., Saha, B. (2012) “Solid-acid and ionic-liquid catalyzed
one-pot transformation of biorenewable substrates into a platform chemical and a
promising biofuel,” RSC Advances, 2(17), 6890–6896.
Angyal, S.J. (1969) “The composition and conformation of sugars in solution,”
Angewandte Chemie International Edition in English, 8(3), 157–166.
Assary, R.S., Curtiss, L.A. (2012) “Comparison of sugar molecule decomposition
through glucose and fructose: A high-level quantum chemical study,” Energy &
Fuels, 26(2), 1344–1352.
Assary, R.S., Kim, T., Low, J.J., Greeley, J., Curtiss, L.A. (2012) “Glucose and
fructose to platform chemicals: understanding the thermodynamic landscapes of
acid-catalysed reactions using high-level ab initio methods,” Physical Chemistry
Chemical Physics, 14(48), 16603–16611.
Balakrishnan, M., Sacia, E.R., Bell, A.T. (2012) “Etherification and reductive
etherification of 5-(hydroxymethyl) furfural: 5-(alkoxymethyl) furfurals and 2, 5bis (alkoxymethyl) furans as potential bio-diesel candidates,” Green Chemistry,
14(6), 1626–1634.
R. P. Bell, The Proton in Chemistry, Chapman and Hall London, 1973, vol. 1.
Bermejo-Deval, R., Orazov, M., Gounder, R., Hwang, S.-J., Davis, M.E. (2014)
“Active sites in Sn-beta for glucose isomerization to fructose and epimerization to
mannose,” ACS Catalysis, 4(7), 2288–2297.
Caratzoulas, S., Vlachos, D.G. (2011) “Converting fructose to 5hydroxymethylfurfural: a quantum mechanics/molecular mechanics study of the
mechanism and energetics,” Carbohydrate Research, 346(5), 664–672.
177
Chapter 5
Chang, C., Xu, G., Jiang, X. (2012) “Production of ethyl levulinate by direct
conversion of wheat straw in ethanol media,” Bioresource Technology, 121, 93–
99.
Chia, M., Dumesic, J.A. (2011) “Liquid-phase catalytic transfer hydrogenation and
cyclization of levulinic acid and its esters to gamma-valerolactone over metal
oxide catalysts,” Chemical Communications, 47(44), 12233–12235.
Choudhary, V., Mushrif, S.H., Ho, C., Anderko, A., Nikolakis, V., Marinkovic, N.S.,
Frenkel, A.I., Sandler, S.I., Vlachos, D.G. (2013) “Insights into the interplay of
Lewis and Brønsted acid catalysts in glucose and fructose conversion to 5(hydroxymethyl) furfural and levulinic acid in aqueous media,” Journal of the
American Chemical Society, 135(10), 3997–4006.
Choudhary, V., Pinar, A.B., Lobo, R.F., Vlachos, D.G., Sandler, S.I. (2013)
“Comparison of homogeneous and heterogeneous ctalysts for glucose-tofructose
isomerization in aqueous media,” ChemSusChem, 6(12), 2369–2376.
Climent, M.J., Corma, A., Iborra, S. (2014) “Conversion of biomass platform
molecules into fuel additives and liquid hydrocarbon fuels,” Green Chemistry,
16(2), 516–547.
Dautzenberg, F.; Gruter, G. J. M. Method for the synthesis of 5-alkoxymethyl
furfural ethers and their use. Patent 8, 133, 289 B2, May, 2012.
Démolis, A., Essayem, N., Rataboul, F. (2014) “Synthesis and applications of alkyl
levulinates,” ACS Sustainable Chemistry & Engineering.
Despax, S., Estrine, B., Hoffmann, N., Le Bras, J., Marinkovic, S., Muzart, J. (2013)
“Isomerization of d-glucose into d-fructose with a heterogeneous catalyst in
organic solvents,” Catalysis Communications, 39, 35–38.
178
Chapter 5
Feng, S., Bagia, C., Mpourmpakis, G. (2013) “Determination of proton affinities and
acidity constants of sugars,” The Journal of Physical Chemistry A, 117(24),
5211–5219.
Fernandes, D., Rocha, A., Mai, E., Mota, C.J., Teixeira da Silva, V. (2012)
“Levulinic acid esterification with ethanol to ethyl levulinate production over
solid acid catalysts,” Applied Catalysis A: General, 425, 199–204.
Flood, A.E., Johns, M.R., White, E.T. (1996) “Mutarotation of D-fructose in
aqueous-ethanolic solutions and its influence on crystallisation,” Carbohydrate
research, 288, 45–56.
Garves, K. (1988) “Acid catalyzed degradation of cellulose in alcohols,” Journal of
wood chemistry and technology, 8(1), 121–134.
Grisel, R., van der Waal, J., de Jong, E., Huijgen, W. (2014) “Acid catalysed
alcoholysis of wheat straw: Towards second generation furan-derivatives,”
Catalysis Today, 223, 3–10.
Hu, X., Wu, L., Wang, Y., Song, Y., Mourant, D., Gunawan, R., Gholizadeh, M., Li,
C.-Z. (2013) “Acid-catalyzed conversion of mono-and poly-sugars into platform
chemicals: Effects of molecular structure of sugar substrate,” Bioresource
Technology, 133, 469–474.
Kimura, H., Nakahara, M., Matubayasi, N. (2011) “In situ kinetic study on
hydrothermal transformation of d-glucose into 5-hydroxymethylfurfural through
d-fructose with 13C NMR,” The Journal of Physical Chemistry A, 115(48),
14013–14021.
179
Chapter 5
Kimura, H., Nakahara, M., Matubayasi, N. (2013) “Solvent effect on pathways and
mechanisms for D-fructose conversion to 5-hydroxymethyl-2-furaldehyde: in situ
13C NMR study,” The Journal of Physical Chemistry A, 117(10), 2102–2113.
Lagarias, J.C., Reeds, J.A., Wright, M.H., Wright, P.E. (1998) “Convergence
properties of the Nelder-Mead simplex method in low dimensions,” SIAM Journal
on Optimization, 9(1), 112–147.
Lew, C.M., Rajabbeigi, N., Tsapatsis, M. (2012) “One-pot synthesis of 5(ethoxymethyl) furfural from glucose using Sn-BEA and Amberlyst catalysts,”
Industrial & Engineering Chemistry Research, 51(14), 5364–5366.
Liu, J., Tang, Y., Wu, K., Bi, C., Cui, Q. (2012) “Conversion of fructose into 5hydroxymethylfurfural (HMF) and its derivatives promoted by inorganic salt in
alcohol,” Carbohydrate Research, 350, 20–24.
Maldonado, G.M.G., Assary, R.S., Dumesic, J.A., Curtiss, L.A. (2012) “Acidcatalyzed conversion of furfuryl alcohol to ethyl levulinate in liquid ethanol,”
Energy & Environmental Science, 5(10), 8990–8997.
Mascal, M., Nikitin, E.B. (2008) “Direct, high-yield conversion of cellulose into
biofuel,” Angewandte Chemie, 120(41), 8042–8044.
Mascal, M., Nikitin, E.B. (2010) “High-yield conversion of plant biomass into the
key value-added feedstocks 5-(hydroxymethyl) furfural, levulinic acid, and
levulinic esters via 5-(chloromethyl) furfural,” Green Chemistry, 12(3), 370–373.
Pasquale, G., Vázquez, P., Romanelli, G., Baronetti, G. (2012) “Catalytic upgrading
of levulinic acid to ethyl levulinate using reusable silica-included Wells-Dawson
heteropolyacid as catalyst,” Catalysis Communications, 18, 115–120.
180
Chapter 5
Peng, L., Lin, L., Zhang, J., Shi, J., Liu, S. (2011) “Solid acid catalyzed glucose
conversion to ethyl levulinate,” Applied Catalysis A: General, 397(1), 259–265.
Que, H., Song, Y., Chen, C.-C. (2011) “Thermodynamic modeling of the sulfuric
acid- water- sulfur trioxide system with the symmetric electrolyte NRTL model,”
Journal of Chemical & Engineering Data, 56(4), 963–977.
Saravanamurugan, S., Nguyen Van Buu, O., Riisager, A. (2011) “Conversion of
mono-and disaccharides to ethyl levulinate and ethyl pyranoside with sulfonic
acid-functionalized ionic liquids,” ChemSusChem, 4(6), 723–726.
Saravanamurugan, S., Paniagua, M., Melero, J.A., Riisager, A. (2013) “Efficient
isomerization of glucose tofructose over zeolites in consecutivereactions in
alcohol and aqueous media,” Journal of the American Chemical Society, 135(14),
5246–5249.
Saravanamurugan, S., Riisager, A. (2013) “Zeolite catalyzed transformation of
carbohydrates to alkyl levulinates,” ChemCatChem, 5(7), 1754–1757.
Serrano-Ruiz, J.C., Luque, R., Sepúlveda-Escribano, A. (2011) “Transformations of
biomass-derived platform molecules: from high added-value chemicals to fuels
via aqueous-phase processing,” Chemical Society Reviews, 40(11), 5266–5281.
Sippola, H. (2012) “Critical evaluation of the 2nd dissociation constants for aqueous
sulfuric acid,” Thermochimica Acta, 532, 65–77.
Swift, T.D., Bagia, C., Choudhary, V., Peklaris, G., Nikolakis, V., Vlachos, D.G.
(2013) “Kinetics of homogeneous brønsted acid catalyzed fructose dehydration
and 5-hydroxymethyl furfural rehydration: A combined experimental and
computational study,” ACS Catalysis, 4(1), 259–267.
181
Chapter 5
Tucker, M.H., Alamillo, R., Crisci, A.J., Gonzalez, G.M., Scott, S.L., Dumesic, J.A.
(2013) “Sustainable solvent systems for use in tandem carbohydrate dehydration
hydrogenation,” ACS Sustainable Chemistry & Engineering, 1(5), 554–560.
Wang, H., Deng, T., Wang, Y., Qi, Y., Hou, X., Zhu, Y. (2013) “Efficient catalytic
system for the conversion of fructose into 5-ethoxymethylfurfural,” Bioresource
Technology, 136, 394–400.
Wang, Z., Lei, T., Liu, L., Zhu, J., He, X., Li, Z. (2012) “Performance investigations
of a diesel engine using ethyl levulinate-diesel blends,” BioResources, 7(4),
5972–5982.
Werpy, T., Petersen, G., Aden, A., Bozell, J., Holladay, J., White, J., Manheim, A.,
Eliot, D., Lasure, L., Jones, S. (2004) “Top value added chemicals from biomass.
Volume 1-Results of screening for potential candidates from sugars and synthesis
gas.”
Yan, K., Wu, G., Lafleur, T., Jarvis, C. (2014) “Production, properties and catalytic
hydrogenation of furfural to fuel additives and value-added chemicals,”
Renewable and Sustainable Energy Reviews, 38, 663–676.
Yan, K., Wu, G., Wen, J., Chen, A. (2013) “One-step synthesis of mesoporous H 4
SiW 12 O 40-SiO 2 catalysts for the production of methyl and ethyl levulinate
biodiesel,” Catalysis Communications, 34, 58–63.
Yang, G., Pidko, E.A., Hensen, E.J. (2012) “Mechanism of brønsted acid-catalyzed
conversion of carbohydrates,” Journal of Catalysis, 295, 122–132.
Yang, Y., Hu, C., Abu-Omar, M.M. (2012) “Conversion of glucose into furans in the
presence of AlCl3 in an ethanol-water solvent system,” Bioresource Technology,
116, 190–194.
182
Chapter 5
Zhang, J., Weitz, E. (2012) “An in situ NMR study of the mechanism for the
catalytic conversion of fructose to 5-hydroxymethylfurfural and then to levulinic
acid using 13C labeled d-fructose,” ACS Catalysis, 2(6), 1211–1218.
Zhang, Z., Dong, K., Zhao, Z.K. (2011) “Efficient conversion of furfuryl alcohol
into alkyl levulinates catalyzed by an organic-inorganic hybrid solid acid
catalyst,” ChemSusChem, 4(1), 112–118.
Zhao, H., Holladay, J.E., Brown, H., Zhang, Z.C. (2007) “Metal chlorides in ionic
liquid solvents convert sugars to 5-hydroxymethylfurfural,” Science, 316(5831),
1597–1600.
Zhu, W., Chang, C., Ma, C., Du, F. (2014) “Kinetics of glucose ethanolysis
catalyzed by extremely low sulfuric acid in ethanol medium,” Chinese Journal of
Chemical Engineering, 22(2), 238–242.
183
Chapter 6
Chapter 6: A Low Temperature Kinetic Model
for the Ethanolysis of D-Fructose to 5Ethoxymethylfurfural in a One-Pot-Synthesis
In this Chapter a kinetic and mechanistic study is conducted on the one-pot
ethanolysis of D-fructose to 5-ethoxymethylfurfural. Whilst Chapter 5 focuses, on
the ethanolysis mechanism between D-fructose and 5-hydroxymethylfurfural, this
chapter is concerned with deriving kinetic parameters that accurately predict
concentrations of 5-ethoxymethylfurfural as a function of time, concentration and
temperature.
The results of this Chapter have been presented at a 30 minute seminar at the 67th
Irish Universities Chemistry Research Colloquium at NUI Maynooth:
Flannelly T., Howard M., Dooley, S., Leahy, J.J. (2015) ‘‘Mechanism and Kinetics
of Advantaged Biofuels Synthesis from D-Fructose’’ 67th Irish University Chemistry
Research Colloquium, 25th June 2015, NUI Maynooth.
6. Chapter
184
Chapter 6
6.1. Abstract
This study conducts an extensive experimental and kinetic study of the ethanolysis
of D-fructose to 5-ethoxymethylfurfural at relatively low temperatures (331-351 K)
catalysed by H2SO4 (0.035-0.13 mol/L) in a one-pot synthesis. The temporal
evolution of the main reaction species observed from 151 individual experiments are
utilised to test a number of incremental propositions of the reaction mechanism
where pseudo first order kinetics are employed with respect to H2SO4. The proposed
model (R2 = 0.965) considers 6 individual reaction pathways and suggests that high
H2SO4 concentrations at low temperatures over long reaction periods is preferable
for maximising yields of 5-ethoxymethylfurfural obtainable in a plug flow reactor
configuration. Model extrapolation suggests that 353 K is the optimum temperature
for the production of high yields of 5-ethoxymethylfurfural from D-fructose after 480
minutes. The rate limiting step is found to be the conversion of a D-fructose
intermediate to 5-hydroxymethylfurfural which requires an activation energy of 75
kJ/mol. Importantly the activation energies for each individual reaction pathway are
significantly lower than the literature values reported in aqueous systems, indicating
the viability of the ethanol solvent system.
185
Chapter 6
6.2. Introduction
Environmental, legislative and security imperatives mean the world is increasingly
looking beyond the conventional finite fossil and geological resources for its energy
and raw materials. Lignocellulosic and hemicellulose plant matter present a major
source of renewable carbon and can provide sustainable alternatives to finite fossil
fuels if they can be converted to useful biofuels (Werpy et al 2004; Kobayashi and
Fukuoka 2013; Climent et al 2014). One such strategy that is currently being
employed for producing viable liquid transportation fuels is the dehydration of
carbohydrates in an alcohol solvent. The alcohol acts as the reaction mediating
solvent, but it also allows additional chemical functionalization of the carbohydrate,
making the reaction products much more suitable as liquid transportation fuels than
is otherwise the case, notably improving the energy density (Maldonado et al 2012).
It has also been shown that the rate of carbohydrate reaction is accelerated by
approximately an order of magnitude in ethanol relative to water as solvent
(Flannelly et al 2015).
The hydrolysis of hexoses in alcohol media also produces a lot less of the
troublesome insoluble humic materials than in aqueous systems (Maldonado et al
2012). One may thus expect both the energy requirements and the eventual yield of
the transforming plant matter carbon and hydrogen into useful molecules to be
positively influenced with the use of ethanol as solvent. We select ethanol as the
alcohol solvent of choice as it is both easily produced by anaerobic and
thermochemical methods (Badger 2002), but its fuel properties, (low energy density,
high volatility and high octane number) restrict its usefulness as a blend-stock with
petroleum derived fuels (Hansen et al 2005). Therefore its reaction with hexose
valorises both the hexose carbohydrates and the ethanol simultaneously.
186
Chapter 6
A variety of potential drop-in liquid transportation fuel components such as
levulinate esters (Christensen et al 2011; Windom et al 2011) and alkyl formates can
be synthesised by the dehydration of carbohydrates in alcohol solvents. Here, we
suppose the high volumetric density of alkoxy alkyl furfurals as a class of molecules
that show promise as fuels in terms of volumetric energy density. For example, the
volumetric energy density of 5-ethoxymethylfurfural ∆Hcombustion kcal/ml is 44%
greater (30.3 MJ/L) than that of ethanol and it is only 5% lower than conventional
gasoline (31.9 MJ/L) (Saha et al 2015). A high O/C would also indicate a strong
propensity to reduce soot formation, which when considered with its high volumetric
energy density (Dautzenberg and Gruter 2012) make it a strongly advantaged fuel
additive.
There have been several literature reports of methods for the production of 5ethoxymethylfurfural in the literature (Dautzenberg and Gruter 2012; Balakrishnan et
al 2012) Perhaps the most common is from the etherification of 5hydroxymethylfurfural. For instance Balakrishnan et al. (Balakrishnan et al 2012)
using a variety of both homogenous and heterogeneous catalysts achieved 5ethoxymethylfurfural yields of 81 mol% and 57 mol% respectively with H2SO4
and DowexSWX8 at 348 K with 5 mol% catalyst for after 24 hours. Che et al. (Che
et al 2012) achieved selectivities of 84.1% of 5-ethoxymethylfurfural from the
etherification of 5-hydroxymethylfurfural using a H4SIW12O40/MCM-41 catalyst at
363 K for 2 hours. Mascal and Nikitin developed an alternative method of 5hydroxymethylfurfural etherification in an ethanol medium where the hydroxyl
group of 5-hydroxymethylfurfural is first replaced with a chlorine atom to form 5chloromethylfurfural (Mascal and Nikitin 2008). Subsequent substitution of the
187
Chapter 6
chlorine atom forms 5-ethoxymethylfurfural. Of more significance however is the
reporting of the one-pot ethanolysis of hexose sugars to 5-ethoxymethylfurfural
especially from D-fructose. A one-pot synthesis approach will significantly reduce
separation steps for products produced in intermediate steps and hence the overall
production costs. Using this one-pot synthesis approach Liu et al. (Liu et al 2013)
achieved yields of 71 mol% 5-ethoxymethylfurfural using 1mmol AlCl3 at 373 K,
whereas Wang et al. (Wang et al 2013) produced 5-ethoxymethylfurfural with the
heterogeneous catalyst H3PW12O40. In addition to all this, ionic liquids (Kraus and
Guney 2012) have also been employed with some success.
Recently there have been some reports of the formation of 5-ethoxymethylfurfural
from D-glucose starting material (Yang et al 2012; Lew et al 2012). For example
Lew et al. achieved yields of 31 mol% from D-glucose using a Sn-Beta catalyst at
363 K. However the efficient dehydration of D-glucose to 5-ethoxymethylfurfural in
ethanol remains challenging, representing an obstacle in the synthetic process
(Climent et al 2014). Facilitating glucose-fructose isomerisation is regarded as a
solution to overcoming these challenges for the formation of high yields of biofuel
components from lignocellulosic derived carbohydrates. Fructose in solution is
thermodynamically easier to protonate than glucose, and consequently is easier to
convert in high yields to furan and furan related compounds (Assary and Curtiss
2012). Thus it anticipated that fructose will be the end product in any efficient
mechanistic scheme for furan formation; therefore it is necessary to evaluate the
kinetic parameters associated with the formation of 5-ethoxymethylfurfural from Dfructose.
To date little kinetic information is available regarding the acid catalysed conversion
of hexoses in non-aqueous systems such as ethanol. For example, despite numerous
188
Chapter 6
studies on the ethanolysis of hexoses (Balakrishnan et al 2012; Peng et al 2011), only
four relevant kinetic studies have been reported to date (Flannelly et al 2015; Zhu et
al 2014; Sacia et al 2014). Both Peng et al. (Peng et al 2012) and Zhu et al. (Zhu et al
2014) produced kinetic models for the production of ethyl levulinate from D-glucose
considering only two chemical reactions with associated rate constants, one
pertaining to ethyl levulinate formation and one to describe all of the other reactions
occurring in the system in a lumped manner. Sacia et al. (Sacia et al 2014) conducted
a
kinetic
study on
the
etherification
of
5-hydoxymethylfurfural
to
5-
ethoxymethylfurfural and subsequently to their acetal analogues over Amberlyst 15.
However there have been no comprehensive kinetic studies conducted linking the
hexose sugar to the intended product in an integrated hierarchical manner in an
alcohol solvent. Thus there is a need to break such systems into incremental stages
linking reactant to product in order to determine more realistic kinetic parameters
pursuant to a more adequate comprehension of the potential industrially important
platforms. The first step in doing so was conducted in a recent study (Flannelly et al
2015) in which we performed a reaction pathway analysis study of the acid
hydrolysis of D-fructose to ethyl levulinate, and 5-ethoxymethylfurfural at 351 K
catalysed by hydrogen cations in ethanol (ethanolysis). A more advanced framework
for conducting detailed kinetic studies in non-aqueous solvents was demonstrated. In
particular the need to investigate the entire variety of chemical species formed by the
reaction, and not only the intended synthetic targets was stressed. It was determined
that it is likely that D-fructose undergoes a reversible reaction with an ethyl
fructoside
species
at
a
rate
much
faster
than
the
formation
of
5-
hydroxymethylfurfural. The addition of water to the solvent media at 10 vol%, and
20 vol% was shown to alter both the predominant chemical mechanism and
189
Chapter 6
consequently the overall rate of reaction (by one and two orders of magnitude
respectively). Phenomenological rate constants for each discrete pathway were
estimated but as the focus of the study was reaction pathway analysis no temperature
dependent kinetic parameters were pursued.
Motivated by the above, we aim to develop a temperature dependent understanding
of the ethanolysis of D-fructose to 5-ethoxymethylfurfural in a homogenous Dfructose/ethanol/H2SO4 catalytic system by deriving kinetic parameters that capture
most of the essential pathways. Relatively low temperature conditions are chosen as
it is envisioned that any process that maximises D-glucose/D-fructose isomerisation
will be conducted at mild temperatures (>393 K)(Saravanamurugan et al 2013),
hence the basic understanding of the reaction kinetics at such temperatures is
essential.
Additionally it has been reported that the hydration of 5-
ethoxymethylfurfural to ethyl levulinate at 353 K is rather slow (Liu et al 2013;
Flannelly et al 2015). As a result of this it envisioned that low temperatures are
preferable for optimization of 5-ethoxymethylfurfural yields. To derive accurate
kinetic parameters we analyse the chemical mechanism hierarchically, breaking it
into three separate subsystems, building the reaction mechanism in an incremental
manner, thus limiting the uncertainty. Under otherwise equivalent conditions, the
ethanolysis sub-model of each major intermediate is examined one-at-a-time in the
order
of;
5-ethoxymethylfurfural,
5-hydroxymethylfurfural,
D-fructose,
and
subsequently integrating the individually obtained parameters to form an overall
semi-detailed kinetic model for the D-fructose system. It is anticipated that the
kinetic parameters developed will help provide a framework for the more complex
D-glucose and cellulose sub models that more realistically reflect actual biomass
feed-stocks, whilst also allowing for more the incorporation of more fundamental
190
Chapter 6
quantities that are obtainable from quantum mechanical methods as our
comprehension develops.
6.3. Experimental
6.3.1. Experimental configuration
All experiments, are carried out in glass pressure tubes (25.4 mm O.D. x10.2 cm),
comprising of PTFE plugs and FETFE O-rings for pressure sealing up to 1.03 MPa.
A 5-mL aqueous solution containing the desired amount of the particular reactant
and H2SO4 is placed in the tube resulting in a pressurised reaction. After being
sealed, each tube is placed for a definite period of time in an oil bath at the desired
reaction temperature. When the reaction time is completed, each tube is removed
from the oil bath and immersed in a cold water bath to quench the reaction. The
reaction temperature is independently controlled and monitored by a thermocouple
array (Stuart™ SCT1 temperature controller) and an in-situ magnetic propeller
ensures that the reaction mixture (reactant/H2SO4/ethanol) is homogeneous. For the
test conditions reported here, Table 6.1, a heating time of 4 minutes is required for
the reacting mixture to be heated from ambient to the prescribed reaction
temperature (±1 K). Reaction progress is monitored by removing and analysing a 50
mg sample of the bulk reaction at fixed intervals.
191
Chapter 6
Reactant
[Reactant]
(Mol/L)
[H2SO4]
(Mol/L)
Temperature
(K)
Sampling
Intervals
(Mins)
5-EMF
0.29
0.13
351
15, 30 60, 120,
240, 360, 480
5-EMF
0.29
0.05, 0.1, 0.2
373, 383, 393
30, 60, 120,
180, 240
5-HMF
0.45
0.035, 0.068,
0.130
331, 341, 351
15, 30 60, 120,
240, 360, 480
D-Fructose
0.29
0.035, 0.068,
0.130
331, 341, 351
15, 30 60, 120,
240, 360, 480
Table 6.1 Experimental variables for low temperature ethanolysis at 331-351 K
catalysed
by
H2SO4.
Abbreviations:
5-ethoxymethylfurfural
(5-EMF),
5-
hydroxymethylfurfural (5-HMF).
Reactions are conducted with D-fructose and 5-hydroxymethylfurfural at 3
temperature ranges (331 K, 341 K, 351 K) whilst higher temperature experiments are
conducted with ethyl 5-ethoxymethylfurfural for reasons to be elaborated later (373
K, 383 K and 393 K) all in the presence of H2SO4 (See Table 6.1).
6.3.2. Materials and methods
Ethanol, normal-octanol, acetone, (99 % purity), D-fructose (CAS 57-48-7, 99%
purity), sulphuric acid (H2SO4, 95-97% purity), 5-hydroxymethylfurfural (CAS 6747-0, 99 % purity) 5-ethoxymethylfurfural (CAS 98-08-1, 98 % purity), and ethyl
levulinate (CAS 539-88-8, 97 % purity) are each obtained from Sigma Aldrich
Ireland. 5-ethoxymethylfurfural (CAS 1917-65-3, 96-97 % purity) is purchased from
Akos Organics Gmbh, Germany.
192
Chapter 6
6.3.3. Analytical procedure
The concentrations of ethyl levulinate and 5-ethoxymethylfurfural, are analysed by
gas chromatography (GC, Agilient Technologies 7820 A GC system) fitted with a
Restek Stabilwax capillary column (30 m, 0.25 mm ID, 0.25 µm), employing
hydrogen carrier gas and a flame ionisation detector. Species are identified by
matching retention-times to known standards, and quantified by calibration of
detector response to known concentrations (using n-octanol as internal standard).
The injection port is maintained at 523 K, a temperature sufficiently high to ensure
the full vaporisation of the expected reaction components. A temperature program of
313 K increasing to 493 K at a rate of 20 K per minute, remaining isothermal at 293
K for 5 minutes is found to achieve adequate separation of these species from the
ethanol/water media GC-MS analysis is also employed for the identification of
sample species using an Agilient 5975C MSD, which uses a HP-5MS column (30 m,
0.25 mm ID, 0.25 µm) otherwise employing the same variables as for GC–FID
analysis. For GC analysis, a known mass (50 ± 5 mg ) of analyte is extracted from
the reaction media into 0.4 g of room temperature acetone and 0.8 g of 0.16 mg/g noctanol in acetone, this is followed by the neutralisation of any remaining acid by the
addition of 50 mg of NaHCO3. This dilution and cooling procedure ensures that the
chemical reaction is effectively quenched. This sample is then filtered through 13
mm thick, 2 µm pore size syringe filters (Acrodisc) to remove any insoluble humic
substances that may have been formed, and 1µl of the resulting solution is injected
into the sample inlet port of the GC.
Identification and quantification of D-fructose, 5-hydroxymethylfurfural and the
various sugar-type derivatives is performed on an ion exchange liquid
chromatography system (IC) system (Dionex Corp., Sunnydale, CA) equipped with a
193
Chapter 6
pulsed amperometric detector (AS, 10 µL sample loop, Dionex Corp., Sunnydale,
CA). Analysis is performed at 291 K by isocratic elution with deionised water (18.2
MΩ.cm at a flow rate of 1.1 ml/min) using a Dionex CarboPac PA1 carbohydrate
column. The column is reconditioned using a mixture of 0.4 mol/L sodium
hydroxide and 0.24 mol/L sodium acetate after each analysis. A 25 mg portion of the
sampled reaction media is diluted with 1.0 g of deionised water. As before, 50 mg of
NaHCO3 is added to neutralise any acid present. This sample is filtered as described
above before being analysed, D-fructose and 5-hydroxymethylfurfural concentrations
are determined by detector calibration to mass prepared standard solutions.
6.3.4. Incorporating the effect of H2SO4 on reaction rates
A variety of approaches have been used to incorporate the effect of catalyst on
reaction rates. For example Marcotullio et al. (Marcotullio et al 2009) estimated the
activity of hydrogen cations by using an electrolyte Non-Random Two-Liquid
(eNRTL) model for aqueous H2SO4 systems. Girisuta et al. (Girisuta et al 2006)
estimated their kinetics by a first order dependence to the prescribed concentration of
H2SO4, as did Zhu et al. (Zhu et al 2014) in the only kinetic study of sugar
hydrolysis conducted in an ethanol medium. Perhaps, the most realistic treatment to
date has been the imposition of rate laws considering the equilibrium concentration
of hydrogen cation following dissociation of H2SO4 at the prescribed reaction
temperature (Kupiainen et al 2012). However, the temperature dependence of H2SO4
dissociation in ethanol is poorly understood and therefore presently infeasible to
model. Additionally we have previously reported that H2SO4 may react with ethanol,
thus resulting in the consumption of hydrogen cations (Flannelly et al 2015).
194
Chapter 6
Therefore to conserve the usability of the kinetic model, the initial [H2SO4] is used to
embody the effect of catalyst on reaction rates.
6.3.5. Estimation of kinetic parameters
The rate constant of each reaction pathway is assumed to follow an Arrhenius
relationship. The activation energy (Ea, kJ/mol) and pre-exponential factor (A, min1
), is estimated assuming a ‘‘pseudo first order’’ relationship with respect to each
reactant and [H2SO4]. As ethanol is present in excess concentration it is not
considered as a rate limiting factor and thus not embodied in the kinetic analysis. It is
assumed that the concentration of ethanol does not change significantly and thus, has
no effect on the reaction rates as at the conditions employed minimal ethanol
conversion to diethyl ether is detected. To determine the relationship between
reaction rates and temperature a modified Arrhenius equation is employed of the
form:
ki  Ae

Ea  1
1 
 

R  T Tmean 
(6.1)
Where T is the temperature (K), Tmean is the mean temperature in which the
experiments under interrogation are conducted at and R is the universal gas constant.
A and Ea are estimated by minimizing the difference between experimental and
model computed species mole concentrations using the Levenberg–Marquardt
algorithm available within the ‘lsqcurvefit’ function on Matlab. Ordinary differential
equations (odes) are integrated numerically by the algorithm solver ‘ode15s’ on
Matlab. Reaction rates are written of the form:
d [5 - EMF]
 (k 1[ H 2 SO4 ]  [5  EMF])
dt
(6.2)
195
Chapter 6
Rate equations for the optimum mechanism are available in the Supporting
Information. In our previous study a mechanism fidelity index was employed to test
the mechanistic validity of the individual reaction pathways.(Flannelly et al 2015) In
this investigation where the aim is to obtain functional engineering parameters in a
hierarchical approach, the quality of fit of the proposed model is determined by using
the normalized root mean square error (R2). Additionally the accuracy of the
proposed mechanistic preposition is monitored by correlation matrices, residuals,
and by conducting sensitivity analysis showing the objective function versus the
derived parameter values.
6.3.6. Kinetic modeling
The formation of 5-ethoxymethylfurfural from D-fructose is a multistep complex
chemical process. Figure 6.1 depicts a general mechanism, derived from the
experimental data discussed below, for the ethanolysis of D-fructose to 5ethoxymethylfurfural in ethanol catalysed by H2SO4 in a homogenous systems. With
present comprehension it is not feasible to consider every species listed in Figure 6.1
in a numerical model.
196
Chapter 6
Figure 6.1 Suggested reaction mechanism for the ethanolysis of D-fructose in
homogenous H2SO4 catalysed conditions.
197
Chapter 6
Figure 6.2 The seven mechanisms considered for kinetic modelling. The kinetic
parameters derived for the 5-ethoxymethylfurfural (Mechanism 2) and 5hydroxymethylfurfural (Mechanism 4) are used as constraints to evaluate the
parameters for the D-fructose mechanism.
Therefore the development of a skeletal model that captures the major
mechanistic features of the overall synthetic process presents a valuable
intermediate goal pursuant to a fuller comprehension.
198
Chapter 6
6.4. Results and discussion
6.4.1. Experimental observations and kinetic modelling
6.4.1.1.
5-Ethoxymethylfurfural experimental observations
The objective of the kinetic model is to accurately predict the concentrations of 5ethoxymethylfurfural as a function of temperature, time and [H2SO4]. In order to do
so it is necessary to investigate the conditions responsible for its consumption. The
main pathway that has been reported responsible for its depletion in an acidic ethanol
media is its hydration to ethyl levulinate (Hu and Li 2011) However, experiment
shows at the temperatures conducted with D-fructose (331-351 K), little 5ethoxymethylfurfural consumption is observed. For example at 351 K no hydration
of 5-ethoxymethylfurfural to ethyl levulinate is observed, with only <20%
conversion of the initial [5-ethoxymethylfurfural] reactant at reaction times of 480
temperatures greater than 373 K significant 5-ethoxymethylfurfural degradation is
observed (See Figure 6.8(b)). For example at 383 K, 54% conversion with 0.2 mol/L
H2SO4 is observed after 240 minutes with 81% of the initial starting mass recovered
considering only 5-ethoxymethylfurfural and ethyl levulinate.
199
Chapter 6
0.1 Mol/L H2SO4 / 0.29 Mol/L 5-Ethoxymethylfurfural
5-Ethoxymethylfurfural
Ethyl levulinate
1.0
373 K
383 K
393 K
Mol Fraction
0.8
0.6
0.4
0.2
0.0
0
50
100
150
200
250
Time / Minutes
Figure 6.3 Temperature dependent behaviour of 5-ethoxymethylfurfural hydration to
ethyl levulinate.
The acetalisation of 5-ethoxymethylfurfural to 5-ethoxymethylfurfural diacetal has
been reported in the literature using heterogeneous solid acid catalysts (Chen et al
2014) however the diacetal analogue is not detected is our investigation.
1.8
1.5
ln (Co/C)
1.2
0.9
0.6
0.3
373 K
383 K
393 K
0.0
0
50
100
150
200
250
Time, Minutes
Figure 6.4 Reaction order for the consumption of 5-ethoxymethylfurfural (H2SO4/5ethoxymethylfurfural 0.035/0.29 Mol/L). Note the overall reaction conforms to first
order kinetics.
200
Chapter 6
6.4.1.2.
Guided
5-Ethoxymethylfurfural sub-model
by
these
observations
we
consider
the
hydration
of
5-
ethoxymethylfurfural to ethyl levulinate (R1);
As previously mentioned this is not a 100% selective process, therefore the
model is configured to consider the formation of by-products from 5ethoxymethylfurfural (R2).
The simple combination of R1 and R2 (Mechanism 1) does not result in a
satisfactory description of the experimental measurements as the [ethyl levulinate]
cannot be accounted for. At 383 K and 393 K it is evident that the selectivity of the
reaction to towards ethyl levulinate formation increases with temperature (See Figure
6.3). It is likely a complicated reaction that cannot be influenced by a simple first
order dependence on H2SO4. As it a hydration reaction is likely that the [water] has a
significant effect on reaction rates. To quantify the exact effect of this is beyond the
scope of this study and represents an avenue for future study. To account for this,
Mechanism 2 considers just R2 which describes the degradation of 5ethoxymethylfurfural to side-products which includes ethyl levulinate. Exercising
this approach is found to be adequate predict [5-ethoxymethylfurfural] consumption
(R2=0.87) (See Figure 6.5)
201
Chapter 6
0.35
2
R =0.87
5-Ethoxymethylfurfural
Predicted Value, Mol/L
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Experimental Value, Mol/L
Figure 6.5 A parity plot, illustrating the difference between experimental
measurements and model predictions for the 5-ethoxymethylfurfural sub-model
(Mechanism #2).
6.4.1.3.
5-Hydroxymethylfurfural experimental observations
From Figures 6.8(c)-(d), 5-ethoxymethylfurfural is the major product observed
from the acid catalyzed ethanolysis of 5-hydroxymethylfurfural at low
temperatures
(331-351
K).
Peak
experimental
yields
of
5-
ethoxymethylfurfural of 60 mol% are detected with ethyl levulinate being the
other reaction product (17 mol%) using 0.13 mol/L H2SO4 at 351 K. 5Hydroxymethylfurfural, it is observed to undergo dimerization to 5’(oxybis(methylene)bis-furfural almost instantaneously as it dissolves in ethanol.
Conversely it is minor pathway, as molar balances considering 5hydroxymethylfurfural, 5-ethoxymethylfurfural and ethyl levulinate only, are
typically 75-95 mol% of the 5-hydroxymethylfurfural reactant (See Figure
6.8(c)-(d)). It is excluded from the kinetic model as quantification is not
possible due to the lack of available analytical standards.
202
Chapter 6
1.5
ln Co/C
1.2
0.9
0.6
0.3
331 K
341 K
351 K
0.0
0
100
200
300
400
500
Time, Minutes
Figure 6.6 Reaction order for the consumption of 5-hydroxymethylfurfural (5hydroxymethylfurfural/H2SO4/0.035/0.29 Mol/L).
6.4.1.4.
5-Hydroxymethylfurfural sub-model.
From Figures 6.8(c)-(d), 5-ethoxymethylfurfural is the major product observed
from the acid catalyzed ethanolysis of 5-hydroxymethylfurfural at low
temperatures
(331-351
K).
Peak
experimental
yields
of
5-
ethoxymethylfurfural of 71 mol% are detected with ethyl levulinate being the
other reaction product (22 mol%) using 0.13 mol/L H2SO4 at 351 K. 5Hydroxymethylfurfural, it is observed to undergo dimerization to 5’(oxybis(methylene)bis-furfural almost instantaneously as it dissolves in ethanol.
Conversely it is minor pathway, as molar balances considering 5hydroxymethylfurfural, 5-ethoxymethylfurfural and ethyl levulinate only, are
typically 85-95 mol% of the 5-hydroxymethylfurfural reactant (See Figure
6.8(c)-(d)). It is excluded from the kinetic model as quantification is not
possible due to the lack of available analytical standards.
203
Chapter 6
For kinetic modeling Mechanism 3 considers the ethanolysis of 5hydroxymethylfurfural to 5-ethoxymethylfurfural (R3) in addition to R1 and
R2 (See Figure 6.2, for each discrete mechanism considered):
As expected from the results of the experiments conducted with 5ethoxymethylfurfural, Mechanism 3 results in an unsatisfactory outcome as it
cannot reproduce the observed [ethyl levulinate] at 331-351 K. Indeed
calculated quantities are twice that of experiment. This confirms that the ethyl
levulinate produced (from fructose) under such conditions (331-351 K) is not
formed from 5-ethoxymethylfurfural sub-mechanism. This finding is in
agreement with the finding of our previous study. Therefore it is necessary to
consider an additional pathway to develop a high fidelity model. Experimental
observations suggest that ethyl levulinate is formed relatively fast when the
concentrations of 5-hydroxymethylfurfural are at their highest. The rate of
ethyl levulinate formation decreases once the 5-hydroxymethylfurfural has
been consumed. A similar trend is observed by Che et al.(Che et al 2012) Two
such plausible pathways that could possibly explain this behaviour are the
hydration of 5-hydroxymethylfurfural to levulinic acid and the pathway
suggested by Hu et al. (Hu and Li 2011) who postulated that 5hydroxymethylfurfural diacetal
degrades to levulinic acid which in turn
undergoes esterification to ethyl levulinate. Levulinic acid is not detected as an
intermediate at significant quantities in our study, however a control reaction
conducted with levulinic acid at 351 K with 0.1 mol/L H2SO4, results in >90%
204
Chapter 6
conversion of levulinic acid after 30 minutes. This study is not able to
ascertain the exact route responsible, and the exact stoichiometry of the
proposed pathway remains unclear. In order to conserve the overall
complexity of the system ethyl levulinate is modelled to come directly from 5hydroxymethylfurfural as intermediate.
0.5
0.4
Predicted Value, Mol/L
2
R =0.9912
Ethyl Levulinate
5-Hydroxymethylfurfural
5-Ethoxymethylfurfural
0.3
0.2
0.1
0.0
0.0
0.1
0.2
0.3
0.4
0.5
Experimental Value, Mol/L
Figure 6.7 A parity plot, illustrating the difference between experimental
measurements and model predictions for the 5-hydroxymethylfurfural sub-model
(Mechanism #4).
Incorporating R4 (Mechanism 4) results in the accurate prediction of all 3 reaction
species (R2 = 0.991). The kinetic parameters that are derived in this subsection are
used as constraints to evaluate the parameters for the D-fructose mechanism (See
Supporting Information for more details).
205
Chapter 6
6.4.1.5.
D-Fructose experimental observations
Ethanolysis experiments conducted with D-fructose results in the formation of 5hydroxymethylfurfural, 5-ethoxymethylfurfural and ethyl levulinate as major
fractions. In contrast to the atom balances of the 5-hydroxymethylfurfural sub
mechanism,
considering
D-fructose,
5-hydroxymethylfurfural,
5-
ethoxymethylfurfural and ethyl levulinate with D-fructose as reactant the atom
balances are as low as 30 mol% (See Figure 6.8(e)-(f)). In a previous study we have
shown that there is a further variety of intermediates between D-fructose and 5hydroxymethylfurfural, with ethyl fructosides being postulated as the dominant
intermediate. Peak 5-ethoxymethylfurfural yields of 50 mol% are observed from Dfructose. Combined peak yields of 5-ethoxymethylfurfural and ethyl levulinate yields
of 60 mol% are detected, which is
considerably less than from 5-
hydroxymethylfurfural (71 mol%). This can be attributed to the formation of byproducts such as furfural, dihydroxyacetone and humins which are also detected
here; their formation is discussed in a previous study (Flannelly et al 2015).
6.4.1.6.
D-Fructose sub-model
The reaction mechanism accounting for the conversion of D-fructose to 5hydroxymethylfurfural is complex, and is currently the subject of extensive
research in aqueous systems. Firstly R5 is considered in the model to describe
the dehydration of D-fructose to 5-hydroxymethylfurfural in addition to the
pathways listed in Mechanism 5.
206
Chapter 6
Only considering R5 is not reasonable in the pursuit of deriving a high fidelity
model as there are a variety of intermediates that have been postulated
between
D-fructose
and
5-hydroxymethylfurfural
including
various
fructofuranose species in aqueous systems (Kimura et al 2013) The variety of
intermediates increases with the employment of non-aqueous solvents, such as
ethanol. For instance anhydrofructose intermediate structures have been
reported in solvents such as gamma-valerolactone (Qi et al 2014) and methanol
as well as alkyl fructosides (Liu et al 2012) in alcohol solvents. Tucker et al.
(Tucker et al 2013) detected difructose anhydrides and ethyl fructosides in an
ethanol solvent. Currently a mechanistic understanding of each individual
pathway is lacking, limiting the detailed description of a higher hierarchical
kinetic mode for fructose ethanolysis. To simplify this, the model considers an
intermediate formed directly from D-fructose as a surrogate for the ethyl
fructosides, difructose anyhydride and other species.
Two different pathways are considered to describe the reaction behaviour of the Dfructose intermediate. Firstly Mechanism 5 considers R7 which describes the
dehydration of the D-fructose-intermediate to 5-hydroxymethylfurfural.
207
Chapter 6
However, when exercising this approach, the model could not accurately
predict the behaviour of 5-hydroxymethylfurfural formation. The polynomial
nature of the [D-fructose] profiles (See Figure 6.8(e)-(f), and See Supporting
Information), suggests that there is equilibrium between the D-fructose
intermediate and D-fructose, which tallies with the work of Trucker et al. and is line
with the results of our previous study. Therefore Mechanism 6 considers a reversible
equilibrium reaction with of D-fructose (R8) omitting R7.
Figure 6.1 illustrates that there is a diversity of by-products produced that don’t
contribute to the formation of the intended reaction products. The formation of byproducts is embodied in the mechanistic proposition by employing R9 (Mechanism
6).
Utilising this approach results in an unsatisfactory outcome, as the model calculated
[5-hydroxymethylfurfural] are higher than observed by experiment.
208
Chapter 6
Figure 6.8 A sample comparison of the main reaction species fractions between
experimental measurements (symbols) and model calculations (lines) catalysed by
H2SO4 (0.035-0.13 mol/L H2SO4) with a starting material of (a b) 5ethoxymethylfurfural (5-EMF) (c d) 5-hydroxymethylfurfural (5-HMF) (e g) Dfructose. Note 5-ethoxymethylfurfural hydration to ethyl levulinate could not be
fitted in the kinetic model (a b).
Mechanism 7 omits R9 and is otherwise the same as Mechanism 6, in any case the
lost mass is accounted for through pathways R6 and R8. Employing this approach
209
Chapter 6
results in a satisfactory outcome (R2 =0.965) and accurately reproduces the
experimental
measurements
of
D-fructose,
5-hydroxymethylfurfural,
5-
ethoxymethylfurfural and ethyl levulinate at 331-351 K. Therefore Mechanism 7 is
deemed the most accurate mechanistic proposition from the experiments conducted
in this investigation.
6.4.2. Implications of the kinetic model
The derived kinetic parameters for the optimum mechanism (Mechanism #7) are
listed in Table 6.2, whilst Figure 6.9 portrays a simple representation of the accuracy
of the proposed model within the tested conditions (331-351 K).
It is notable that the estimated activation energies for each reaction pathway
considered in the ethanolysis system are significantly lower than in conventional
aqueous systems. Dallas Swift et al. (Swift et al 2013) suggest the activation energy
for the reaction of the main of D-fructose intermediate to 5-hydroxymethylfurfural to
be 115 kJ/mol whilst Asghari and Yoshida (Salak Asghari and Yoshida 2006)
reported an activation energy of 160 kJ/mol for the same pathway, all be it their
experiments were conducted at higher temperatures (T >483 K). This compares to an
activation energy of 75 kJ/mol obtained in our investigation for the dehydration of Dfructose to 5-hydroxymethylfurfural. The estimated activation energy required to
hydrate 5-hydroxymethylfurfural to ethyl levulinate is found to be 59 kJ/mol, and is
also significantly lower than reports for its hydration in aqueous systems (86
kJ/mol). This is pertinent as it illustrates that the ethanolysis system, represent an
energetically more efficient avenue to valorize the valuable synthesized 5hydroxymethylfurfural than in standard aqueous systems.
210
Chapter 6
Reaction
Ea
(kJ mol-1)
A
(min-1)
k (351 K)
(s-1)
6
71 ±2
3.42 ±0.65
1.17E-01
8
60 ±17
0.81 ±0.17
1.77E-03
5
82 ±9
0.15 ±0.012
5.92E-03
3
55 ±6
0.06 ±0.002
1.77E-03
2*
86 ±13
0.04 ±0.003
4.68E-05
4
59 ±7
0.013 ±0.002
4.14E-04
Table 6.2 Estimated activation energies (Ea), Pre-exponential factors (A) with
corresponding rate constants at 351 K. Error margins correspond to 95% confidence
intervals. See Figure 6.2, Mechanism 7 for reaction numbers.* Calculated at a mean
temperature of 383 K.
It is notable that the estimated activation energies for each reaction pathway
considered in the ethanolysis system are significantly lower than in conventional
aqueous systems. Dallas Swift et al. (Swift et al 2013) suggest the activation energy
for the reaction of the main of D-fructose intermediate to 5-hydroxymethylfurfural to
be 115 kJ/mol whilst Asghari and Yoshida (Salak Asghari and Yoshida 2006)
reported an activation energy of 160 kJ/mol for the same pathway, all be it their
experiments were conducted at higher temperatures (T >483 K). This compares to an
activation energy of 75 kJ/mol obtained in our investigation for the dehydration of Dfructose to 5-hydroxymethylfurfural. The estimated activation energy required to
hydrate 5-hydroxymethylfurfural to ethyl levulinate is found to be 59 kJ/mol, and is
also significantly lower than reports for its hydration in aqueous systems (86
kJ/mol). This is pertinent as it illustrates that the ethanolysis system, represent an
211
Chapter 6
energetically more efficient avenue to valorize the valuable synthesized 5hydroxymethylfurfural than in standard aqueous systems.
Figure 6.9 A parity plot, illustrating the difference between experimental
measurements and model predictions, Mechanism 6.
To examine the yields of 5-ethoxymethylfurfural obtainable the kinetic model is
exercised in a plug reaction configuration. The model suggests yields of up to 60
mol% and 72 mol% are obtainable when configured to maximize its formation both
from D-fructose and 5-hydroxymethylfurfural respectively, compared to that of 50
and 60 mol% detected by experiment. [H2SO4] is found to positively affect yields of
5-ethoxymethylfurfural (See Figure 6.10), however, the effect of temperature is more
complicated, due to the increased reactivity of 5-ethoxymethylfurfural with
temperature. Thus, there is fine balance between yields of 5-ethoxymethylfurfural
and production of unwanted by-products, such as ethyl levulinate. The model
suggests that high [H2SO4] and low temperatures employed over long reaction times,
are the most suitable for optimising yields of 5-ethoxymethylfurfural. This is the
case, as the activation energy for 5-ethoxymethylfurfural consumption (R2) is
212
Chapter 6
significantly higher than from its formation (R4), 86 kJ/mol compared to 55 kJ/mol.
For example at 351 K the ratio between the rate constant responsible for its
formation and that of its consumption is 17/1. The model predicts that this ratio
decreases to 9/1 at 383 K. Optimum temperatures for yields of 5ethxoymethylfurfual, using the highest H2SO4 employed in this study (0.13 mol/L),
is 353 K and 338 K from D-fructose and 5-hydroxymethylfurfural respectively (See
Figure 6.10). Other studies suggest that high temperatures are optimum for
maximising yields of 5-ethoxymethylfurfural from D-fructose in a one pot synthesis
(Yuan et al 2015) For instance Yuan et al. (Yuan et al 2015) using a Fe3O4
composite as catalyst found yields of 5-ethoxymethylfurfural to increase with
temperature and it appears under such conditions that 5-ethoxymethylfurfural is a lot
more stable than in the presence of H2SO4. It has also been suggested that high
temperatures
should
increase
the
selectivity
of
D-fructose
to
5-
hydroxymethylfurfural (Liu et al 2015) in a plug flow reactor in an ethanol system
which is in agreement which our model. However it is 5-ethoxymethylfurfural’s
propensity to undergo degradation at elevated temperatures (>373 K) in the presence
of H2SO4 (>373 K) mean that that higher temperatures suggested in such studies
aren’t feasible in our system.
The uncertainty for the kinetic parameters derived for the reversible reaction
between D-fructose and the D-fructose intermediate (R6 and R8) is high. This is so,
as the model suggests that there isn’t a strong dependence on temperature, in the
range of conditions employed in this investigation, which is quite narrow (See
Supporting Information). Additionally the mechanistic understanding here is weak;
no robust measurements are obtainable in order supply the model with more
reasonable constraints. Thus a more thorough detailed study is required, for a more
213
Chapter 6
comprehensive mechanistic insight and is an avenue for further study. Finally our
experiments suggest that >25% of the D-fructose mass is lost between D-fructose and
5-hydroxymethylfurfural, it is therefore acknowledged that using a less aggressive
catalyst
than
H2SO4 would
potentially
result
in
higher
yields
of
5-
ethoxymethylfurfural. For example Liu et al. (Liu et al 2013) achieved yields of 5ethoxymethylfurfural of 71.2 mol% using AlCl3 from D-fructose. Screening catalyst
performance for maximum selectivity of 5-ethoxymethylfurfural formation is beyond
the scope of this work but it is anticipated that the methodologies and insights will
prove valuable for further kinetic studies for the Bronsted acid catalyzed
transformations of D-fructose and indeed D-glucose in alcohol solvents to
oxygenated hydrocarbons.
214
Chapter 6
Figure 6.10 Three Dimensional (a c) and contour plots (b d) depicting the maximum
yields of 5-ethoxymethylfurfural obtainable after 480 minutes, from 5hydroxymethylfurfural (a b) reactants and D-fructose (c d) as reactants with the range
of [H2SO4] employed in this study. Note the optimum temperatures for 5ethoxymethylfurfural production are found to be 353 K and 343 K from D-fructose
and 5-hydroxymethylfurfural as reactants.
215
Chapter 6
Pre-exponential Value Intermediate - Fructose
0.5
Objective Function Value
Estimated Value
Objective Function Value
Estimated Value
Objective Function Value
Objective Function Value
Pre-exponential Value Fructose - Intermediate
1.5
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
0.4
0.3
0.2
0.1
0.0
0
10
20
0.0
0.5
Pre-exponential Factor
1.0
1.5
2.0
2.5
3.0
Pre-exponential Factor
Pre-exponential Value Fructose - 5-Hydroxymethylfurfural
Pre-exponential Value 5-Hydroxymethylfurfural - 5-Ethoxymethylfurfural
0.9
Objective Function Value
Estimated Value
0.16
Objective Function Value
Estimated Value
0.8
Objective Function Value
Objective Function Value
0.18
0.14
0.12
0.10
0.08
0.06
0.7
0.6
0.5
0.4
0.3
0.2
0.04
0.1
0.02
0.0
0.0
0.2
0.4
0.6
0.8
0.0
0.1
Pre-exponential Factor
0.3
Pre-exponential Factor 5-Ethoxymethylfurfural - X
Pre-exponential Value 5-Hydroxymethylfurfural - Ethyl Levulinate
0.3
0.9
Objective Function Value
Estimated Value
Object Function
Estimated Value
0.8
Object Function Value
Objective Function Value
0.2
Pre-exponential Factor
0.2
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.1
0.0
0.00
0.02
0.04
Pre-exponential Factor
0.06
0.08
0.0
0.1
0.2
0.3
0.4
Pre-exponential Factor
Figure 6.11 Sensitivity analysis for the derived pre-exponential factors as reported in
Table 6.2.
216
Chapter 6
Activation Energy D-Fructose - Intermediate
Objective Function Value
Estimated Value
Objective Function Value
Estimated Value
0.06
Objective Function Value
0.055
Objective Function Value
Activation Energy Intermediate - D-Fructose
0.050
0.045
0.040
0.05
0.04
0.035
0.030
0.03
40000
50000
60000
70000
80000
90000
100000
0
20000
Activation Energy / J/mol
40000
60000
80000
100000
Activation Energy / J/mol
Activation Energy 5-Hydroxymethylfurfural - 5-Ethoxymethylfurfural
Activation Energy D-Fructose - 5-Hydroxymethylfurfural
0.24
Objective Function Value
Estimated Value
0.22
Objective Function Value
Objective Function Value
Objective Function Value
Estimated Value
0.12
0.06
0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.00
0
50000
100000
150000
0
200000
Activation Energy 5-Hydroxymethylfurfural - Ethyl Levulinate
60000
80000
100000
Objective Function
Parameter
0.050
Objective Function Value
Objective Function Value
40000
Activation Energy 5-Ethoxymethylfurfural - X
Objective Function Value
Estimated Value
0.14
20000
Activation Energy / J/mol
Activation Energy / J/mol
0.12
0.10
0.045
0.040
0.08
0
50000
100000
Activation Energy / J/mol
150000
92000
96000
100000
Activation Energy / J/mol
Figure 6.12 Sensitivity analysis for the activation energies as reported in Table 6.2.
217
Chapter 6
6.5. Conclusions
This work presents an extensive hierarchical experimental kinetic study on the
ethanolysis of D-fructose to 5-ethoxymethylfurfural at low temperatures (331-351 K)
catalysed by H2SO4 (0.035-0.13 mol/L) in a one-pot system. Under such conditions,
the
model
adequately
describes
the
concentrations
of
D-fructose,
5-
hydroxymethylfurfural, and 5-ethoxymethylfurfural as a function of time, H2SO4
concentration and temperature. The optimised model considers a reaction mechanism
of six discrete chemical reactions including one reversible reaction, of D-fructose to a
D-fructose intermediate. The reaction pathway accounting for the formation of 5ethoxymethylfurfural through the etherification of 5-hydroxymethylfurfural is found
to be 17 times faster than the pathway responsible for its consumption at 351 K, with
the relationship between formation and consumption, found be largely dependent on
temperature. Extrapolation of the model indicates that 353 K is the optimum
temperature for 5-ethoxymethylfurfural formation from D-fructose within the range
of conditions investigated in this study.
6.6. References
Assary, R.S., Curtiss, L.A. (2012) “Comparison of sugar molecule decomposition
through glucose and fructose: A high-level quantum chemical study,” Energy &
Fuels, 26(2), 1344–1352.
Badger, P. (2002) “Ethanol from cellulose: A general review,” Trends in new crops
and new uses, 17–21.
Balakrishnan, M., Sacia, E.R., Bell, A.T. (2012) “Etherification and reductive
etherification of 5-(hydroxymethyl) furfural: 5-(alkoxymethyl) furfurals and 2, 5-
218
Chapter 6
bis (alkoxymethyl) furans as potential bio-diesel candidates,” Green Chemistry,
14(6), 1626–1634.
Che, P., Lu, F., Zhang, J., Huang, Y., Nie, X., Gao, J., Xu, J. (2012) “Catalytic
selective etherification of hydroxyl groups in 5-hydroxymethylfurfural over H 4
SiW 12 O 40/MCM-41 nanospheres for liquid fuel production,” Bioresource
Technology, 119, 433–436.
Chen,
J.,
Zhao,
G.,
Chen,
L.
(2014)
“Efficient
production
of
5-
hydroxymethylfurfural and alkyl levulinate from biomass carbohydrate using
ionic liquid-based polyoxometalate salts,” RSC Advances, 4(8), 4194–4202.
Christensen, E., Williams, A., Paul, S., Burton, S., McCormick, R.L. (2011)
“Properties and performance of levulinate esters as diesel blend components,”
Energy & Fuels, 25(11), 5422–5428.
Climent, M.J., Corma, A., Iborra, S. (2014) “Conversion of biomass platform
molecules into fuel additives and liquid hydrocarbon fuels,” Green Chemistry,
16(2), 516–547.
Dautzenberg, F.; Gruter, G. J. M. Method for the synthesis of 5-alkoxymethyl
furfural ethers and their use. Patent 8, 133, 289 B2, May, 2012.
Flannelly, T., Dooley, S., Leahy, J.J. (2015) “Reaction Pathway Analysis of Ethyl
levulinate and 5-Ethoxymethylfurfural from D-Fructose Acid Hydrolysis in
Ethanol,” Energy & Fuels.
Girisuta, B., Janssen, L., Heeres, H. (2006) “A kinetic study on the decomposition of
5-hydroxymethylfurfural into levulinic acid,” Green Chemistry, 8(8), 701–709.
219
Chapter 6
Hansen, A.C., Zhang, Q., Lyne, P.W. (2005) “Ethanol-diesel fuel blends--a review,”
Bioresource Technology, 96(3), 277–285.
Hu, X., Li, C.-Z. (2011) “Levulinic esters from the acid-catalysed reactions of sugars
and alcohols as part of a bio-refinery,” Green Chemistry, 13(7), 1676–1679.
Kimura, H., Nakahara, M., Matubayasi, N. (2013) “Solvent effect on pathways and
mechanisms for D-fructose conversion to 5-hydroxymethyl-2-furaldehyde: in situ
13C NMR study,” The Journal of Physical Chemistry A, 117(10), 2102–2113.
Kobayashi, H., Fukuoka, A. (2013) “Synthesis and utilisation of sugar compounds
derived from lignocellulosic biomass,” Green Chemistry, 15(7), 1740–1763.
Kraus, G.A., Guney, T. (2012) “A direct synthesis of 5-alkoxymethylfurfural ethers
from fructose via sulfonic acid-functionalized ionic liquids,” Green Chemistry,
14(6), 1593–1596.
Kupiainen, L., Ahola, J., Tanskanen, J. (2012) “Distinct effect of formic and sulfuric
acids on cellulose hydrolysis at high temperature,” Industrial & Engineering
Chemistry Research, 51(8), 3295–3300.
Lew, C.M., Rajabbeigi, N., Tsapatsis, M. (2012) “One-pot synthesis of 5(ethoxymethyl) furfural from glucose using Sn-BEA and Amberlyst catalysts,”
Industrial & Engineering Chemistry Research, 51(14), 5364–5366.
Liu, B., Zhang, Z., Huang, K., Fang, Z. (2013) “Efficient conversion of
carbohydrates into 5-ethoxymethylfurfural in ethanol catalyzed by AlCl3,” Fuel,
113, 625–631.
220
Chapter 6
Liu, J., Tang, Y., Fua, X. (2015) “Efficient conversion of carbohydratesto
ethoxymethylfurfural and levulinic acid ethyl ester under the catalysis of
recyclable DMSO/Brønsted acids,” Starch-Stärke.
Liu, J., Tang, Y., Wu, K., Bi, C., Cui, Q. (2012) “Conversion of fructose into 5hydroxymethylfurfural (HMF) and its derivatives promoted by inorganic salt in
alcohol,” Carbohydrate Research, 350, 20–24.
Liu, R., Chen, J., Huang, X., Chen, L., Ma, L., Li, X. (2013) “Conversion of fructose
into 5-hydroxymethylfurfural and alkyl levulinates catalyzed by sulfonic acidfunctionalized carbon materials,” Green Chemistry, 15(10), 2895–2903.
Maldonado, G.M.G., Assary, R.S., Dumesic, J.A., Curtiss, L.A. (2012) “Acidcatalyzed conversion of furfuryl alcohol to ethyl levulinate in liquid ethanol,”
Energy & Environmental Science, 5(10), 8990–8997.
Marcotullio, G., Cardoso, M.A.T., De Jong, W., Verkooijen, A.H. (2009)
“Bioenergy II: furfural destruction kinetics during sulphuric acid-catalyzed
production from biomass,” International Journal of Chemical Reactor
Engineering, 7(1).
Mascal, M., Nikitin, E.B. (2008) “Direct, High-Yield Conversion of Cellulose into
Biofuel,” Angewandte Chemie, 120(41), 8042–8044.
Peng, L., Lin, L., Li, H. (2012) “Extremely low sulfuric acid catalyst system for
synthesis of methyl levulinate from glucose,” Industrial Crops and Products, 40,
136–144.
Peng, L., Lin, L., Zhang, J., Shi, J., Liu, S. (2011) “Solid acid catalyzed glucose
conversion to ethyl levulinate,” Applied Catalysis A: General, 397(1), 259–265.
221
Chapter 6
Qi, L., Mui, Y.F., Lo, S.W., Lui, M.Y., Akien, G.R., Horváth, I.T. (2014) “Catalytic
Conversion of Fructose, Glucose, and Sucrose to 5-(Hydroxymethyl) furfural and
Levulinic and Formic Acids in gamma-Valerolactone As a Green Solvent,” ACS
Catalysis, 4(5), 1470–1477.
Sacia, E.R., Balakrishnan, M., Bell, A.T. (2014) “Biomass conversion to diesel via
the etherification of furanyl alcohols catalyzed by Amberlyst-15,” Journal of
Catalysis, 313, 70–79.
Saha, B., Bohre, A., Dutta, S., Abu-Omar, M.M. (2015) “Upgrading Furfurals to
Drop-in Biofuels: An Overview,” ACS Sustainable Chemistry & Engineering.
Salak Asghari, F., Yoshida, H. (2006) “Acid-catalyzed production of 5hydroxymethyl furfural from D-fructose in subcritical water,” Industrial &
Engineering Chemistry Research, 45(7), 2163–2173.
Saravanamurugan, S., Paniagua, M., Melero, J.A., Riisager, A. (2013) “Efficient
Isomerization of Glucose to Fructose over Zeolites in Consecutive Reactions in
Alcohol and Aqueous Media,” Journal of the American Chemical Society,
135(14), 5246–5249.
Swift, T.D., Bagia, C., Choudhary, V., Peklaris, G., Nikolakis, V., Vlachos, D.G.
(2013) “Kinetics of Homogeneous Brønsted Acid Catalyzed Fructose Dehydration
and 5-Hydroxymethyl Furfural Rehydration: A Combined Experimental and
Computational Study,” ACS Catalysis, 4(1), 259–267.
Tucker, M.H., Alamillo, R., Crisci, A.J., Gonzalez, G.M., Scott, S.L., Dumesic, J.A.
(2013) “Sustainable solvent systems for use in tandem carbohydrate dehydration
hydrogenation,” ACS Sustainable Chemistry & Engineering, 1(5), 554–560.
222
Chapter 6
Wang, H., Deng, T., Wang, Y., Qi, Y., Hou, X., Zhu, Y. (2013) “Efficient catalytic
system for the conversion of fructose into 5-ethoxymethylfurfural,” Bioresource
Technology, 136, 394–400.
Werpy, T., Petersen, G., Aden, A., Bozell, J., Holladay, J., White, J., Manheim, A.,
Eliot, D., Lasure, L., Jones, S. (2004) “Top value added chemicals from biomass.
Volume 1-Results of screening for potential candidates from sugars and synthesis
gas.”
Windom, B.C., Lovestead, T.M., Mascal, M., Nikitin, E.B., Bruno, T.J. (2011)
“Advanced distillation curve analysis on ethyl levulinate as a diesel fuel
oxygenate and a hybrid biodiesel fuel,” Energy & Fuels, 25(4), 1878–1890.
Yang, Y., Hu, C., Abu-Omar, M.M. (2012) “Conversion of glucose into furans in the
presence of AlCl3 in an ethanol-water solvent system,” Bioresource Technology,
116, 190–194.
Yuan, Z., Zhang, Z., Zheng, J., Lin, J. (2015) “Efficient synthesis of promising
liquid fuels 5-ethoxymethylfurfural from carbohydrates,” Fuel, 150, 236–242.
Zhu, W., Chang, C., Ma, C., Du, F. (2014) “Kinetics of Glucose Ethanolysis
Catalyzed by Extremely Low Sulfuric Acid in Ethanol Medium,” Chinese Journal
of
Chemical
Engineering,
22(2),
238–242.
223
Chapter 7
Chapter 7: Synthesis of “Diesel” or “Gasoline”
Oxygenated Fuel Components from Hexose
Carbohydrates by Reaction with Ethanol
This Chapter describes a mechanistic study on the Bronsted acid catalysed
ethanolysis of D-glucose to ethyl levulinate. In addition to this, particular attention is
given to the temperature dependent formation of diethyl ether from ethanol and the
range of potential fuel components that can be produced in ethanolysis systems is
explored. Finally it is demonstrated how the mechanistic knowledge obtained can be
leveraged to produce a drop-in oxygenated fuel additive with fuel properties
mimicking conventional diesel or gasoline fuels.
This work has been presented at the 2015 European Union Cost Initiative titled
‘‘Smart Energy Carriers as Liquid Transportation fuels ’’ on the 25th August in
Thessaloniki Greece
Flannelly T., Howard M., Dooley S., Leahy, J.J. (2015) ‘‘Synthesis of “Diesel” or
“Gasoline” Oxygenated Fuel Components from Hexose Carbohydrates by Reaction
with Ethanol ’’ Cost Action CM1404 ‘‘Smart Energy Carriers as Liquid
Transportation Fuels’’ 27th August 2015, Thessaloniki, Greece.
7. dd
3333
224
Chapter 7
7.1. Abstract
A preliminary mechanistic investigation is conducted on the reaction pathways
responsible for the ethanolysis of D-glucose to ethyl levulinate (351-423 K) catalysed
by H2SO4 (0.015-0.075 mol/L). Under condensed phase conditions (351 K), no ethyl
levulinate is formed, with ethyl glucosides and anhydroglucose structures the main
products identified. Significant amounts of ethyl levulinate are formed only above
423 K (0.015 mol/L H2SO4) and peak steady-state ethyl levulinate yields of 49 mol%
are detected at 453 K (0.075 mol/L H2SO4) compared to yields of 63 mol% achieved
from D-fructose under equivalent conditions. The mechanisms of the ethanolysis of
D-glucose and D-fructose to ethyl levulinate, are found to be different. The main
pathway for the D-glucose mechanism goes through an ethyl glucoside intermediate,
which then dehydrates once liberating formic acid. Significantly the main reaction
flux for ethyl levulinate formation does not advance through any furan intermediates
as it does in the D-fructose ethanolysis system. This finding is in conflict with the
literature. A simple kinetic model is derived to describe the acid catalysed
transformation of ethanol to diethyl ether. It is found only to be a significant pathway
above 373 K having an activation energy of 78 kJ/mol.
The fuel properties of the range of fuel molecules produced in ethanolysis systems
are presented. It is shown that by integrating mechanistic understanding, kinetic
parameters and fuel properties of the synthesized molecules, drop-in tailor-made
fuels can be synthesized in a highly flexible “one-pot” process designed for specific
‘‘diesel or gasoline’’ purposes.
225
Chapter 7
7.2. Introduction
As a result of economic, environmental and legislative imperatives, the need to
develop secure and sustainable alternatives to fossil derived liquid transportation
fuels is urgent (Werpy et al 2004). As of January 2016, the price of a barrel of crude
oil is exceptionally low at $30 (NASDAQ 2016). Despite this, concerns about future
supply and the rapidly evolving consequences of increased CO2 in the atmosphere,
means the necessity for alternative sources to petroleum derived fuel is as great as
ever (Demirbas 2009). Conversion of biomass/lignocellulosic plant materials is
regarded as one of the most promising alternatives to fossil fuels for the production
of biofuels (Climent et al 2014; Stӧcker 2008; Lynd et al 1991). Ethanol is by far the
biggest bio-refining product, with an average of 935,000 barrels produced per year in
the US in 2015 (NREL 2015). However, ethanol is a poor fuel, having a low
∆HCombustion compared to conventional petrol or gasoline, absorbing water easily, is
corrosive and is difficult to transport due to its high volatility (NREL 2015).
There have been considerable efforts invested in recent years in pursuit of upgrading
the unlocked lignocellulosic sugars to fuels other than ethanol. Promising fuel
additives include 2-methyl furan (Thewes et al 2011) and methyl tetrahydrofuran
(Bond et al 2010). Other potential promising fuels conversion pathways include, the
reforming of hexoses to alkanes, conversion to valeric esters, or butane based fuels
(Serrano-Ruiz and Dumesic 2011) and dimethyl ether (Semelsberger et al 2006).
Various other fuel molecules have also been suggested and the predominant
approach in scholarly literature appears to invest considerable intellectual time and
resources on understanding their combustion properties and performance. Most of
the existing fuel production scenarios involve multiple processing steps with several
separations required. After extensive combustion related tests and engine trials, they
226
Chapter 7
may very well be plausible fuel component candidates but there synthesis viability
remains unknown. Often the economic and engineering concerns regarding the
synthesis of these fuel candidates from lignocellulosic materials mean the
plausibility of them being manufactured at scale is unlikely (Balan 2014). In this
research Chapter an alternative philosophy is followed as is depicted in Figure 7.1,
which compares our approach with the conventional methodology alleviating a lot of
concerns in synthesising the above mentioned molecules. A one-pot hydrolysis
means the costly processes of product separation and purification is avoided. This
approach also means that extensive fuel performance tests can be conducted on a
“realistic” mixture of fuel components rather than one individual candidate. Here
two practical feed-stocks, ethanol and lignocellulosic hexoses both of which are
abundant and relatively inexpensive, can both be valorised to produce a mixture of
fuel components in a one-pot system, producing a drop-in fuel.
Figure 7.1 A potentially more practical approach for the synthesis of liquid
transportation fuels.
227
Chapter 7
Once the composition is known and the synthesis process understood, it then may be
more prudent to conduct tests on combustion properties and performance.
The ethanol/hexose system employed in this investigation results in the formation of
ethyl levulinate and 5-ethoxymethylfurfural both of which show considerable
promise as fuel blend additives due to their high volumetric energy (Christensen et al
2011; Dautzenberg and Gruter 2012). In addition to these preferred products,
valorisation with ethanol results in a diverse set of other possible fuel components
(e.g. ethyl formate) that have varied energy densities, engine usabilities and fuel
properties with respect to ethanol. The synthetic route consists of two competing
parallel processes, the acid catalysed transformations of hexoses, and the acid
catalysed transformation of ethanol mainly to diethyl ether which is less well studied
in the literature. For example, diethyl ether is potentially a useful fuel component as
it has a cetane number of 150, compared to that of ethanol (12) and indeed diesel
(54) (Murphy et al 2004). Thus, integrating the ethanol and hexose systems offers an
opportunity to produce fuel additives of adjustable properties to allow for blending
with diesel or gasolines fuels in a drop-in manner.
The variables that can be used as free parameters to manipulate the fuel properties
are ethanol which at temperatures above 373 K can be converted to diethyl ether and
the lignocellulosic derived hexoses D-fructose and D-glucose. Chapters 5 and 6
discuss in detail the mechanisms responsible for the reaction of ethanol with Dfructose catalysed by hydrogen cations, with 5-ethoxymethylfurfural the main
reaction product at lower temperatures (>373 K), however the mechanism for the
transformation D-glucose in ethanol has yet to be derived. When the mechanism is
fully understood, it can also be used as parameter in which to manipulate the
composition of fuel properties of the resultant ethanolysis mixture.
228
Chapter 7
To date only two pertinent kinetic studies have been conducted, regarding D-glucose
ethanolysis. Zhu et al. formulated a simple model with only two reaction pathways
considered, modelling D-glucose directly to ethyl levulinate and side products (Zhu
et al 2014). Using this approach they obtained activation energies of 70 and 120
kJ/mol for the two pathways respectively, whilst in an analogous study in the
equivalent methanol system by Peng et al. they found activation energies of 104 and
123.3 kJ/mol (Peng et al 2012). As well as this, robust mechanistic information
regarding the formation of ethyl levulinate from D-glucose is scarce. In one of the
most relevant studies in this regard Hu et al. found that ethyl glucosides were the
main product of D-glucose ethanolysis other than ethyl levulinate (Hu et al 2011a).
The other most applicable study was also conducted by Hu et al. Here they
conducted an investigation on reaction pathways during the esterification of Dglucose in methanol catalysed by Amberlyst 70 (Hu et al 2011b). They also found 5methoxymethylfurfural
and 5-(hydroxymethyl)-2-(dimethoxymethyl) furan as
significant intermediates along with methyl levulinate and methyl glucosides. There
have been also surprisingly few studies conducted on the acid catalysed
transformation of ethanol to diethyl ether, which is almost as important as the
parallel pathways valorising the hexose sugars. It has been reported that the
employment of USY zeolites in conjunction with H2SO4, can inhibit diethyl ether
formation (Xu et al 2013), however few kinetic parameters are available regarding
the temperature dependent transformation behaviour of ethanol during conventional
ethanolysis processes.
This study focuses on three main core components:
1. An investigation of the Bronsted acid catalysed mechanism for the
ethanolysis of D-glucose to ethyl levulinate in order to compare the D229
Chapter 7
glucose/D-fructose ethanolysis mechanisms, species distribution, and yields
achievable for the pertinent reaction products.
2. Investigating the temperature dependent transformation of ethanol to diethyl
ether.
3. Demonstrating how the mechanistic information obtained from the Dfructose/D-glucose/ethanol, ethanolysis system can be leveraged to
manipulate the molecular diversity of the resulting fuel mixture to produce
fuel components of adjustable properties to allow for blending with diesel or
gasoline in a drop-in manner. This information is sought by utilising simple
fuel properties such as ∆HCombustion, research octane number and cetane
number to mimic diesel and gasoline fuel properties.
7.3. Experimental
7.3.1. Materials
Ethanol, diethyl ether (CAS 60-29-7 >99% purity) normal-octanol, acetone, (99%
purity), α/β-D-fructopyranose (CAS 57-48-7, 99% purity), β-D-glucopyranose (CAS
50-99-7, 99% purity)
hence forth “D-fructose” and “D-glucose” respectively,
sulphuric acid (H2SO4, 95-97% purity), 5-hydroxymethylfurfural (CAS 67-47-0, 99
% purity) and ethyl levulinate (CAS 539-88-8, 97% purity) are each obtained from
Sigma Aldrich Ireland. Ethyl-β-D-glucopyranoside (CAS 34625-23-5, 98% purity) is
obtained from Carbosyth Ltd. UK, and 5-ethoxymethylfurfural (CAS 1917-65-3, 9697% purity) is purchased from Akos Organics Gmbh, Germany.
230
Chapter 7
7.3.2. Experimental configuration
All ethanolysis experiments, are carried out in glass pressure tubes (25.4 mm
O.D. x10.2 cm), comprising of PTFE plugs and FETFE O-rings for pressure
sealing up to 1.03 MPa. A 5-mL aqueous solution containing the desired amount of
the particular reactant and H2SO4 is placed in the tube. After being sealed, each tube
is placed for a definite period of time in an oil bath at the desired reaction
temperature as outlined in Table 7.1. When the reaction time is completed, each tube
is removed from the oil bath and immersed in a cold water bath to quench the
reaction. The reaction temperature is independently controlled and monitored by a
thermocouple array (Stuart™ SCT1 temperature controller) and an in-situ magnetic
propeller ensures that the reaction mixture (reactant/H2SO4/ethanol) is homogeneous.
For the test conditions reported here, Table 7.1, a heating time of 4 minutes is
required for the reacting mixture to be heated from ambient to the prescribed
reaction temperature (±1 K). Reaction progress is monitored by removing and
analysing a 50 mg sample of the bulk reaction at fixed intervals. Ethanolysis
reactions conducted at above 351 K result in a biphasic system consisting of a liquid
and gas phase.
7.3.3. Analytical methods
The
concentrations
of
ethanol,
diethyl
ethoxymethylfurfural, are analysed by
ether,
ethyl
levulinate
and
5-
gas chromatography (GC, Agilient
Technologies 7820 A GC system) fitted with a Restek Stabilwax capillary column
(30 m, 0.25 mm ID, 0.25 µm), employing hydrogen as the carrier gas and a flame
ionisation detector. Species are identified by matching retention-times to known
standards, and quantified by calibration of detector response to known
231
Chapter 7
concentrations (using n-octanol as internal standard). The injection port is
maintained at 523 K, a temperature sufficiently high to ensure the full vaporisation
of the expected reaction components. A temperature program of 313 K increasing to
493 K at a rate of 20 K per minute, remaining isothermal at 493 K for 5 minutes is
found to achieve adequate separation of these species from the ethanol/water media.
GC-MS analysis is also employed for the identification of sample species using an
Agilient 5975C MSD, which uses a HP-5MS column (30 m, 0.25 mm ID, 0.25 µm)
otherwise employing the same variables as for GC–FID analysis. For GC analysis, a
known mass (50 ± 5 mg ) of analyte is extracted from the reaction media into 0.4 g
of room temperature acetone and 0.8 g of 0.16 mg/g n-octanol in acetone, this is
followed by the neutralisation of any remaining acid by the addition of 50 mg of
NaHCO3. This dilution and cooling procedure ensures that the chemical reaction is
effectively quenched. This sample is then filtered through 13 mm thick, 2 µm pore
size syringe filters (Acrodisc) to remove any insoluble humic substances that may
have been formed, and 1µl of the resulting solution is injected into the sample inlet
port of the GC.
Identification and quantification of D-glucose, D-fructose, ethyl glucosides, 5hydroxymethylfurfural and the various sugar-type derivatives is performed on an ion
exchange liquid chromatography system (IC) system (Dionex Corp., Sunnydale, CA)
equipped with a pulsed amperometric detector (AS, 10 µL sample loop, Dionex
Corp., Sunnydale, CA). Analysis is performed at 291 K by isocratic elution with
deionised water (18.2 MΩ.cm at a flow rate of 1.1 ml/min) using a Dionex CarboPac
PA1 carbohydrate column. The column is reconditioned using a mixture of 0.4
mol/L sodium hydroxide and 0.24 mol/L sodium acetate after each analysis.
232
Chapter 7
Reactant
[Reactant]
(Mol/L)
[H2SO4]
(Mol/L)
Temperature
(K)
Sampling
Intervals
(Mins)
D-Fructose
0.29
0.1
351
10,20,30
D-Fructose
0.115
0.04
438
10,20,30
D-Glucose
0.29
0.1
351
10,20,30, 480
D-Glucose
0.145
0.015, 0.04,
0.075
423
D-Glucose
0.145
0.015, 0.04,
0.075
438
D-Glucose
0.145
0.015, 0.04,
0.075
453
Ethyl-βGlucosides
0.145
0.075
453
15, 30, 60,
120, 240, 480,
930*
15, 30, 60,
120, 240, 480,
930*
15, 30, 60,
120, 240, 480,
930
15, 30, 60,
120, 240, 480
Table 7.1 Table of experimental conditions.* depicts experiments carried on for
longer periods until reactions have proceeded to steady-state.
A 25 mg portion of the sampled reaction media is diluted with 1.0 g of deionised
water. As before, 50 mg of NaHCO3 is added to neutralise any acid present This
sample is filtered as described above before being analysed, D-glucose, D-fructose
and 5-hydroxymethylfurfural concentrations are determined by detector calibration
to mass prepared standard solutions.
7.3.4. Kinetic modelling
To estimate the kinetic parameters associated with the dehydration of ethanol to
diethyl ether, it is assumed that the rates follow an Arrhenius relationship. The
activation energy (Ea, kJ/mol) and pre-exponential factor (A, min-1), is estimated
assuming a ‘‘pseudo first order’’ relationship with respect to each reactant and
233
Chapter 7
[H2SO4]. To determine the relationship between reaction rates and temperature a
modified Arrhenius equation is employed of the form:
ki  Ae

Ea  1
1 
 

R  T Tmean 
(7.1)
Where T is the temperature (K), Tmean is the mean temperature in which the
experiments under interrogation are conducted at and R is the universal gas constant.
A and Ea are estimated by minimizing the difference between experimental and
model computed species mole concentrations using the Levenberg–Marquardt
algorithm available within the ‘lsqcurvefit’ function on Matlab. An ordinary
differential equations is integrated numerically by the algorithm solver ‘ode15s’ on
Matlab.
7.4. Results and discussion
7.4.1. Reaction mechanism for Bronsted acid catalysed ethanolysis of D-glucose
Ethanolysis experiments conducted with D-glucose at 351 K, results in no 5hydroxymethylfurfural, 5-ethoxymethylfurfural or indeed ethyl levulinate formation.
This is in comparison to the fructose mechanism where it has been shown that 5ethoxymethylfurfural and ethyl levulinate are formed through the intermediate 5hydroxymethylfurfural and can be produced from D-fructose under condensed phase
ethanolysis conditions (Flannelly et al 2015). GC-MS analysis suggests that main
reaction products at 351 K consist of ethyl glucosides and various anhydroglucose
species. Further screening experiments conducted at 373 K, 393 K and 413 K still
results in little formation of furans or ethyl levulinate, with ethyl glucosides forming
a stable reaction product. It is only above 423 K that any considerable amounts of
ethyl
levulinate
are
formed.
Significantly,
only
trace
amounts
of
5234
Chapter 7
ethoxymethylfurfural and no 5-hydroxymethylfurfural are detected for the
ethanolysis of D-glucose at 423 K. This is in contrast to the ethanolysis mechanism
for D-fructose, where substantial amounts of 5-ethoxymethylfurfural (up to 34
mol%) and 5-hydroxymethylfurfural ( up to 20 mol%) are detected along with ethyl
levulinate (See Figure 7.2(b)). This information is noteworthy as it provides
compelling evidence that the mechanism for the Bronsted acid catalysed ethanolysis
formation of ethyl levulinate between D-fructose and D-glucose are different. This is
in conflict with the formation of levulinic acid in conventional aqueous hydrolysis
systems where the prevailing literature on mechanisms for hexose transformations
suggests that both D-glucose and D-fructose dehydrate to 5-hydroxymethylfurfural as
an intermediate to levulinic acid (Choudhary et al 2012; Swift et al 2013).
438 K, 0.115 Mol/L D-Fructose, 0.04 Mol/L H 2SO4
Concentration, Mol/L
0.125
125
100
0.100
75
0.075
0.050
50
0.025
25
0
0.000
0
100
200
300
400
Time / Minutes
500
600
D-Fructose
Ethyl Levulinate
5-Hydroxymethylfurfural
5-Ethoxymethylfurfural
0.150
700
0.125
Concentration, Mol/L
0.150
150
Mass balance / Ethanol
converted to diethyl ether
(a)
Mass balance
D-Glucose
Ethanol conversion
Ethyl Levulinate
to
diethyl
ether
5-Hydroxymethylfurfural
5-Ethoxymethylfurfural
Ethyl Glucoside
Mass balance
(b)
Ethanol conversion
to diethyl ether
150
125
100
0.100
75
0.075
0.050
50
0.025
25
Mass balance / Ethanol
converted to diethyl ether
438 K, 0.115 Mol/L D-Glucose, 0.04 Mol/L H 2SO4
0
0.000
0
50
100
150
200
250
Time / Minutes
Figure 7.2 The difference between the (a) D-glucose and (b) D-fructose Bronsted
acid mechanism at 438 K catalysed by 0.04 mol/L H2SO4 in ethanol. Note no 5hydroxymethylfurfural is produced in the D-glucose ethanolysis system.
It is also significant, as it has been postulated that in order to successfully valorise
the hexose carbohydrate, D-fructose and D-glucose isomerisation must occur.
235
Chapter 7
As the D-fructose/D-glucose ethanolysis mechanisms are different it is necessary to
ascertain the reaction pathway to account for the formation of ethyl levulinate from
ethyl glucosides. The most plausible route is that ethyl glucoside dehydrates once to
form ethyl levulinate liberating one mole of formic acid. It is also plausible that ethyl
glucosides can dehydrate three times to form 5-ethoxymethylfurfural, which
subsequently hydrates with two moles of water to form ethyl levulinate. There are no
other products detected by GC-MS analysis from the control reactions conducted
with ethyl glucosides, indicating that there may be no other stable detectable
intermediates between itself and ethyl levulinate. However there have been other
pathways suggested in the literature employing heterogeneous catalysis. For example
Hu et al. observed ethyl glucosides to degrade to 5-hydroxymethylfurfural diacetal
using Amberlyst 15 to catalyse the reaction (Hu and Li 2011). They subsequently
observed the acetal structure to degrade via an unknown pathway to levulinic acid;
however the diacetal structure is not detected on the GC-MS analysis. It can be
concluded that the main Bronsted acid catalysed ethanolysis transformation pathway
for D-glucose does not go through any furan intermediates to form ethyl levulinate.
Given that ethyl levulinate formation is found to occur only above 423 K,
experiments are conducted between 423 K, 438 K, 453 K to see the maximum yields
of ethyl levulinate achievable and if it is concurrent with the temperature dependant
degradation of ethanol to diethyl ether. At steady-state the maximum ethyl levulinate
yields detected are 49 mol% ±3.9 at 453 K employing 0.07 mol/L H2SO4 (See Figure
7.3(e)). Not surprisingly, higher yields can be achieved from D-fructose, as at
steady-state maximum yields of 63 mol% ethyl levulinate are detected (See Figure
7.2(b). It appears the Bronsted acid D-fructose ethanolysis system can be configured
to produce higher yields of ethyl levulinate than that of 5-ethoxymethylfurfural.
236
Chapter 7
Under condensed phase ethanolysis conditions peak 5-ethoxymethylfurfural yields of
58 mol% can be achieved when the reaction is configured to produce 5ethoxymethylfurfural over ethyl levulinate (See Chapter 4 and 5). This is promising
as it shows that up to 28 mol% higher yields of the desirable fuel components can be
synthesised from D-fructose which to an extent justifies the considerable amount of
effort being conducted on facilitating D-glucose/D-fructose isomerisation in the
literature.
237
Chapter 7
Figure 7.3 Concentrations as a function per time for the ethanolysis of D-glucose to ethyl
levulinate, as for select conditions of Table 7.1 Note each reaction is presented with
different timescales to ensure the glucose to ethyl levulinate reaction has reached steadystate.
238
Chapter 7
The highest detected yields of 49 mol% from D-glucose, poses the question
regarding the fate of the other 51 mol% and the potential use of the resulting
products as fuel components.
To explore this further, Figure 7.4 portrays an overall schematic of all the plausible
reactions pathways in a conventional ethanolysis system derived from the work
conducted in this Chapter, this Thesis, and also through consultation of pertinent
literature. In comparison to the formation of furans and their derivatives, the
pathways responsible for the acid catalysed transformation of hexoses to other
products are less well defined. In other alcoholysis studies employing similar
conditions (H2SO4, 373-453 K), alkyl acetates (Hu et al 2011; Chang et al 2012),
alkyl pyruvates (Hu et al 2011) and alkyl formates (in excess molar ratios to alkyl
levulinate) have been reported. Despite this, no information purporting to the
mechanisms of their formation has been provided. In Chapter 4, in aqueous systems
it was shown that, hexoses such as D-glucose can undergo transformations to Derythrose, forming equimolar amounts of glycoladehyde. D-erythrose is unstable at
423 K and was found to decompose to form acetic and formic acids.
It is assumed that the ethyl acetate formed in ethanolysis systems is formed through
this pathway from the esterification of the acetic acid (See Figure 7.4). In Chapter 4,
in aqueous systems D-glucose is also observed to degrade to form dihydroxyacetone,
which in turn dehydrates to form pyruvaldehyde. Presumably the ethyl pyruvate is
formed from the esterification of pyruvaldehyde. Pyruvaldehyde can also undergo
transformations to lactic acid, which partakes in esterification to form ethyl lactate.
All are potential fuel components and a discussion pertaining to their fuel properties
is provided in Section 7.4.3.
239
Chapter 7
7.4.2. Conversion of ethanol to diethyl ether
During the composition of this Thesis, ethanol has been shown to be consumed
through reaction with H2SO4 to form ethyl hydrogen sulphate, via reaction with the
hexose derived derivatives (See Chapter 5) and more importantly as observed in this
Chapter to form diethyl ether which is a promising fuel component in its own right.
For example it has long been known that it is a cold start additive for engines, is an
excellent compression ignition fuel with ∆HCombustion greater than ethanol and has a
cetane number > 125 (US DOE) (See Table 7.3). Surprisingly, kinetic information
regarding it is not available in the literature. Under condensed phase conditions
(<371 K) (Chapter 5 and 6) relatively little diethyl ether is formed. Even at 423 K
using 0.015 mol/L H2SO4 only 14 mass % of ethanol is converted to diethyl ether
after 16 hours (See Figure 7.3 (a)).
240
Chapter 7
Figure 7.4 Detailed mechanism for the Bronsted acid catalysed ethanolysis of Dglucose. Note all exciting fuel components produced are highlighted in green boxes.
1-3 describes the main reaction flux for the formation of ethyl levulinate. Each reach
reaction assumes a first order relationship with [H+]. The detection of ethyl acetate,
ethyl pyruvate and ethyl lactate is not pursued in this preliminary study but have
been reported in other Bronsted acid ethanolysis systems.
241
Chapter 7
In pursuit of the formulation of a fuel with adjustable properties it is important to
determine the temperature dependent behaviour of the formation of ethanol to
diethyl ether. To do this, a simple kinetic model is formulated using the conditions
outlined in Table 7.1. Diethyl ether formation profiles can be obtained from
consulting Figure 7.3(a)-7.3(e).
The model considers the following reaction:
Using this approach it is evident that the formation of diethyl ether follows an
Arrhenius relationship as the proposed model can accurately replicate the
experimental data (See Figure 7.5).
Reaction
Ea (kJ/mol)
A (min-1)
1
78.9
0.0203
Table 7.2 Derived kinetic parameters for the derived for the reaction of ethanol
Whilst using the kinetic parameters as depicted in Table 7.2, is it possible to predict
the diethyl ether formation pattern as a function of temperature. Using these
parameters, the model predicts that at 403 K, 9 mol% of the ethanol is converted to
diethyl ether after 16 hours while using
0.015 mol/L H2SO4. Conducting the
reaction for the same time period at 373 K, it is predicted that 0.65 mol% of the
ethanol is converted to diethyl ether.
242
Chapter 7
18
16
Ethanol
Diethyl Ether
2
R = 0.9945
Model Prediction (Mol/L)
14
12
10
8
6
4
2
0
0
2
4
6
8
10
12
14
16
18
Experimental Value (Mol/L)
Figure 7.5 A parity plot illustrating the accuracy of the proposed model for the
formation of diethyl ether from ethanol (423-453 K).
7.4.3. Manipulating the reaction mechanism to produce fuels of tailor-able
properties, be it ‘‘diesel’’ or ‘‘gasoline’’.
Table 7.3 provides a library of fuel properties for each fuel molecule produced in the
ethanolysis system including, ∆HCombustion , research octane number, cetane number,
and the H/C ratio. However to demonstrate how the hexose ethanolysis system can
be leveraged to produce fuels with properties mimicking diesel or gasoline, five of
the
most
crucial
molecules
are
chosen;
ethanol,
diethyl
ether,
5-
ethoxymethylfurfural, ethyl levulinate and ethyl formate. In this Thesis, temperature
resolved concentrations of these species have been accumulated from both
condensed phase and biphasic conditions. The fuel properties considered are cetane
number, research octane number and ∆HCombustion. The cetane number is an inverse
function of a fuels ignition delay and describes the time period between the start of
injection and the first identifiable pressure increase during the combustion of the fuel
243
Chapter 7
Chemical Name
Chemical
Formula
Molecular
Weight
(g/mol)
Density
(g/cm3)
Boiling
Point
Melting
Point
Flash
Point
H/C
ratio
∆H°C
(kJ/mol)
∆H°C
(kJ/ml)
Vapour
Pressure
(kPa)
C6H12O6
180.16
1.54
800.15 K
423.15 K
559.82 K
2:1
2805
23.98
N/A
N/A
C8H10O3
154.16
1.099
508.15 K
N/A
>383.15 K
5:4
4074.7
29.04
N/A
48
29
C7H12O3
144.17
1.016
487.15 K
240.35 K
N/A
12:7
3523
24.83
0.03
5
110
C6H6O3
126.11
1.29
387.15 K –
389.15 K
303.15 K –
307.15 K
352.15 K
1:1
2781
28.44
0.005
N/A
N/A
C5H10O3
118.13
1.03
427.15 K
247.15 K
319.15 K
2:1
N/A
N/A
0.3
Unknown
N/A
C5H8O3
116.12
1.045
417.15 K
215.15 K
319.15 K
8:5
N/A
N/A
0.5
Unknown
Unknown
C4H8O2
88.11
0.902
350.25 K
198.15 K
270.16 K
2:1
2238
22.91
9.73
Unknown.
116
(C2H5)2O
74.12
0.7134
307.75 K
157.15 K
233.15 K
5:2
2726
26.24
56.3
150
Unknown
C3H6O2
74.08
0.917
327.15 K
193.15 K
254.15 K
2:1
1643
20.35
26.12
8.5.
109
C2H5OH
46.07
0.789
351.15 K
159.15 K
287.15 K
3:1
1367
23.41
5.95
11
108
Cetane
Number
Research
Octane
Number
N/A
Table 7.3 Fuel properties of all pertinent fuel molecules present produced in the ethanolysis system in this Thesis, obtained from Sigma
Aldrich (Sigma Aldrich 2015). Cetane numbers are obtained from Murphy (Murphy et al 2004) and calculated research octane numbers
are estimated using the correlations of Hass and Dryer (Hass and Dryer 2013).
244
Chapter 7
It is an appropriate metric for the determination of the merits of a diesel fuel
additive.
Road diesel fuels typically have cetane numbers of ~54 (European
Commission 2009). The research octane number is the inverse of the cetane number
and is an appropriate metric to determine the merits of the usage of a gasoline fuel.
Typical gasoline has a minimum research octane number of 95 (European
Commission 2009). Taking the D-glucose ethanolysis system as an example Figure
7.6 illustrates that by blending different vol% of the main fuel components produced;
diethyl ether, ethanol and ethyl levulinate, a fuel with the same centane and octane
number as gasoline and diesel can beformulated.
Figure 7.6 Manipulating species mole fractions of ethanolysis products to produce a
fuel mixture to mimic the centane number of conventional diesel (a) and gasoline
(b), as is highlighted by a white line on both figures. Note the white line corresponds
to the centane number of diesel (54) (a) and the research octane number (94) of
gasoline (b).
A detailed mechanistic and kinetic understanding of the system unlocks the
possibility to tailor the reaction system for the production of specific fuel mixtures
with desired combustion properties. To illustrate this, the theoretical combined fuel
245
Chapter 7
properties of the D-fructose synthetic system conducted under condensed phase
conditions (See Chapter 5) are compared to the species fractions of the synthesised
fuel components as presented in this investigation. Energy dense fuel mixtures with
cetane numbers in the range of ~6 – 44 can be produced,see Figure 7.7.
Figure 7.7 A comparison of the cetane numbers and energy densities of the flux of
fuel components produced per time between condensed phase and biphasic reaction
conditions. Note it is assumed that all the water has been removed. Condensed phase
conditions D-fructose/H2SO4 0.29/0.1 mol/L, biphasic conditions D-fructose/H2SO4
0.115/0.04 mol/L.
This diversity present is as a result of the variety of fuel species synthesized in the
system and the interconversion of their mass fractions as the reaction proceeds.
Quenching the reaction at a defined time, results in specified fuel properties. In the
biphasic D-fructose ethanolysis synthesis, ethyl levulinate is the main hexose derived
energy product with substantial amounts of diethyl ether being formed. This
produces a lower energy density fuel with a high cetane number making it a suitable
diesel blend component.
In the condensed phase ethanolysis synthesis, 5-
246
Chapter 7
ethoxymethylfurfural is the most abundant hexose derived energy product, but under
such conditions no diethyl ether is produced. Therefore, the resulting fuel blend is of
a low cetane number but is of notably higher volumetric energy density, indicating it
may be appropriate as a gasoline additive (See Figure 7.8). The energy of
valorisation with respect to the ethanol consumed in the condensed phase and
biphasic system are 270% and 207% respectively, whilst the energy density of the
resultant fuel mixtures are 30.56 MJ/L (condensed phase system) and 27.38 MJ/L
(biphasic system), which are comparable to that of conventional gasoline.
Figure 7.8 The chemical mechanism and fuel chemistry of a D-fructose/ethanol
synthetic fuel system interpreted as gasoline or diesel fuel additives.
7.4.4. Outlook, limitations and future work
This Chapter presents a preliminary investigation of how the molecular diversity of
the ethanolysis system can be leveraged to produce a one-pot synthetic mixture of
fuel molecules with custom made fuel properties. However significant further work
must be conducted to build on what has been presented in order to further develop
the scientific ideas and premise as is presented in this Chapter:
247
Chapter 7
1. This Chapter or Thesis does not experimentally deal with the issue of hexose
loading. Ideally the minimum ratio of ethanol/hexose is sought in order to
maximise the energy valorisation of the system. It is also necessary to
ascertain what effect the hexose/ethanol loading has on the systems
mechanism and reaction kinetics.
2. It is necessary to elucidate the effect of the formation of water on the fuel
properties and reaction mechanism of the overall system.
3. It is imperative to develop a functional kinetic model for the ethanolysis of Dglucose to ethyl levulinate, similar to the model developed for D-fructose in
Chapter 6, so that it can be integrated with the D-fructose system in a
hierarchical manner.
4. Experimentally time resolved species concentrations of ethyl pyruvate, ethyl
formate, ethyl acetate and ethyl lactate need to be derived.
5. Fuel performance properties such as solubility in petroleum fuels need to be
established on the mixture of molecular species produced in the ethanolysis
systems.
6. Finally the premise for the overall ethanolysis process is predicated on
successful D-glucose isomerisation by employing a Bronsted/Lewis tandem
system, of which is still at an early stage of development in the scientific
community.
7.5. Conclusion
A preliminary mechanistic investigation is conducted on the reaction mechanism
responsible for the ethanolysis of D-glucose to ethyl levulinate (351-423 K) catalysed
248
Chapter 7
by H2SO4 (0.015-0.075 mol/L). Under condensed phase conditions (351 K), no ethyl
levulinate is formed with ethyl glucosides and anhydroglucose structures the main
products detected. Only significant amounts of ethyl levulinate are formed above 423
K (0.015 mol/L H2SO4) and peak ethyl levulinate yields of 49 mol% are detected at
453 K (0.075 mol/L H2SO4), compared to 63 mol% achieved from D-fructose. The
mechanisms for the ethanolysis of D-glucose and D-fructose to ethyl levulinate are
found to be different. The main pathway for the D-glucose mechanism goes through
an ethyl glucoside intermediate, which then dehydrates once liberating formic acid.
Significantly the main reaction flux for ethyl levulinate does not progress through
any furan intermediates as is the D-fructose system invalidating the common
assumption in the literature. A mechanistic description is provided for the formation
of other potential fuel components such as ethyl lactate, ethyl acetate and ethyl
pyruvate.
A simple kinetic model is derived to describe the acid catalysed transformation of
ethanol to diethyl ether. It is found only to be a significant pathway above 373 K and
is found to have an activation energy of 78 kJ/mol. A library of fuel properties
produced for the range of fuel molecules produced in the ethanolysis system is
presented. Finally it is illustrated that by integrating mechanistic understanding and
fuel properties of the synthesized molecules, drop-in tailor-made fuel additives can
be synthesized in a highly flexible “one-pot” process designed for specific purposes.
7.5 References
Balan, V. (2014) “Current challenges in commercially producing biofuels from
lignocellulosic biomass,” ISRN Biotechnology, 2014.
249
Chapter 7
Bond, J.Q., Alonso, D.M., Wang, D., West, R.M., Dumesic, J.A. (2010) “Integrated
catalytic conversion of gamma-valerolactone to liquid alkenes for transportation
fuels,” Science, 327(5969), 1110–1114.
Chang, C., Jiang, X.X., Zhang, T., Li, B. (2012) “Effect of reaction parameters on
the production of ethyl levulinate from glucose in ethanol,” Advanced Materials
Research, 512, 388–391.
Chheda, J.N., Román-Leshkov, Y., Dumesic, J.A. (2007) “Production of 5hydroxymethylfurfural and furfural by dehydration of biomass-derived mono-and
poly-saccharides,” Green Chemistry, 9(4), 342–350.
Choudhary, V., Burnett, R.I., Vlachos, D.G., Sandler, S.I. (2012) “Dehydration of
glucose to 5-(hydroxymethyl) furfural and anhydroglucose: thermodynamic
insights,” The Journal of Physical Chemistry C, 116(8), 5116–5120.
Christensen, E., Williams, A., Paul, S., Burton, S., McCormick, R.L. (2011)
“Properties and performance of levulinate esters as diesel blend components,”
Energy & Fuels, 25(11), 5422–5428.
Climent, M.J., Corma, A., Iborra, S. (2014) “Conversion of biomass platform
molecules into fuel additives and liquid hydrocarbon fuels,” Green Chemistry,
16(2), 516–547.
Demirbas, A. (2009) “Political, economic and environmental impacts of biofuels: A
review,” Applied Energy, 86, S108–S117.
European Commission (2009) “Directive 2009/30/EC of the European Parliament
and of the Council of 23 April 2009 amending Directive 98/70/EC as regards the
specification of petrol, diesel and gas-oil and introducing a mechanism to monitor
and reduce greenhouse gas emissions and amending Council Directive
250
Chapter 7
1999/32/EC as regards the specification of fuel used by inland waterway vessels
and repealing Directive 93/12,” EEC.
Flannelly, T., Dooley, S., Leahy, J.J. (2015) “Reaction pathway analysis of ethyl
levulinate and 5-ethoxymethylfurfural from D-fructose acid hydrolysis in
ethanol,” Energy & Fuels.
Hu, X., Li, C.-Z. (2011a) “Levulinic esters from the acid-catalysed reactions of
sugars and alcohols as part of a bio-refinery,” Green Chem., 13(7), 1676–1679.
Hu, X., Lievens, C., Larcher, A., Li, C.-Z. (2011b) “Reaction pathways of glucose
during esterification: Effects of reaction parameters on the formation of humin
type polymers,” Bioresource Technology, 102(21), 10104–10113.
Lynd, L.R., Cushman, J.H., Nichols, R.J., Wyman, C.E., others (1991) “Fuel ethanol
from cellulosic biomass.,” Science(Washington), 251(4999), 1318–1323.
NREL 2015 'Ethanol Blended Fuels',[online], available:
http://www.nrel.gov/education/pdfs/educational_resources/high_school/teachers_
guide_ethanol.pdf [acessed 11 October 2015].
Murphy, M.J., Taylor, J.D., McCormick, R.L. (2004) Compendium of experimental
cetane number data, National Renewable Energy Laboratory Golden, CO.
Peng, L., Lin, L., Li, H. (2012) “Extremely low sulfuric acid catalyst system for
synthesis of methyl levulinate from glucose,” Industrial Crops and Products, 40,
136–144.
Semelsberger, T.A., Borup, R.L., Greene, H.L. (2006) “Dimethyl ether (DME) as an
alternative fuel,” Journal of Power Sources, 156(2), 497–511.
251
Chapter 7
Serrano-Ruiz, J.C., Dumesic, J.A. (2011) “Catalytic routes for the conversion of
biomass into liquid hydrocarbon transportation fuels,” Energy & Environmental
Science, 4(1), 83–99.
Sigma Aldrich 2015 'Product Search' https://www.sigmaaldrich.com/ireland.html
[accessed 21 November 2015].
Stӧcker, M. (2008) “Biofuels and biomass-to-liquid fuels in the biorefinery:
Catalytic conversion of lignocellulosic biomass using porous materials,”
Angewandte Chemie International Edition, 47(48), 9200–9211.
Swift, T.D., Bagia, C., Choudhary, V., Peklaris, G., Nikolakis, V., Vlachos, D.G.
(2013) “Kinetics of homogeneous brønsted acid catalyzed fructose dehydration
and 5-hydroxymethyl furfural rehydration: A combined experimental and
computational study,” ACS Catalysis, 4(1), 259–267.
Haas F.M., Dryer F.L. (2013) “Prediction of biofuel ignition quality using a
DCN↔RON interconversion tool” Fall Technical Meeting of the Eastern States
Section of the Combustion Institute Hosted by Clemson University, Clemson, SC.
Oct 13-16 2013.
Thewes, M., Muether, M., Pischinger, S., Budde, M., Brunn, A., Sehr, A., Adomeit,
P., Klankermayer, J. (2011) “Analysis of the impact of 2-methylfuran on mixture
formation and combustion in a direct-injection spark-ignition engine,” Energy &
Fuels, 25(12), 5549–5561.
Werpy, T., Petersen, G., Aden, A., Bozell, J., Holladay, J., White, J., Manheim, A.,
Eliot, D., Lasure, L., Jones, S. (2004) “Top value added chemicals from biomass.
Volume 1-Results of screening for potential candidates from sugars and synthesis
gas.”
252
Chapter 7
Xu, G.-Z., Chang, C., Zhu, W.-N., Li, B., Ma, X.-J., Du, F.-G. (2013) “A
comparative study on direct production of ethyl levulinate from glucose in ethanol
media catalysed by different acid catalysts,” Chemical Papers, 67(11), 1355–
1363.
Zhu, W., Chang, C., Ma, C., Du, F. (2014) “Kinetics of glucose ethanolysis
catalyzed by extremely low sulfuric acid in ethanol medium,” Chinese Journal of
Chemical Engineering, 22(2), 238–242.
253
Chapter 8
Chapter 8: Conclusions
8. concluision
254
Chapter 8
8.1. Conclusions
Presently the price of oil has never been lower this century at $37 a barrel
(02/03/2016) (NASDAQ 2016), so the need for short term liquid fuel transportation
components may be considered less urgent than at the outset of this Thesis, when a
barrel of oil was priced at ~$110 (NASDAQ 2012). However this may not remain
the case for long as historically the price of oil is prone to boom and bust cycles, but
judged on a 10 year cycle remains on an upward price trend (Baumeister and Kilian
2016). More importantly the recent climatic effects of increased CO2 in the
atmosphere means the need for carbon neutral transportation fuel is as crucial as
ever. This Thesis contributes to the development of substitutes to fossil derived fuels
by proposing the ethanolysis system as an alternative means for valorising
lignocellulosic hexose sugars in comparison to conventional aqueous systems. The
provision of reaction mechanisms and kinetics in especially non-aqueous systems
represents a bottleneck for a long term sustainable industry valorising the
lignocellulosic hexose through acid hydrolysis. A greater mechanistic understanding
is provided by performing detailed reaction pathways analysis.
It is determined that the structural confirmation of the hexose carbohydrates
contributes greatly to yields of levulinic acid achievable. Levulinic acid yields from
hexose carbohydrates at steady-state are found to be in the order of; D-fructose > Dglucose > D-mannose > D-galactose. The approach of pursing a mechanistic
understanding for each hexose transformation pathway in the system allowed it to be
determined that formic and levulinic acid are not formed stoichiometrically from the
acid hydrolysis of hexose carbohydrates as is the common assumption in the
literature. At steady-state conversions of the reactant, the formic/levulinic acid for D-
255
Chapter 8
fructose, D-glucose, D-mannose and D-galactose is shown to be 1.08 ±0.04, 1.15
±0.05, 1.20 ±0.10 and 1.19±0.04 respectively. Combining this work and pertinent
literature suggests there are at least four potential pathways depending on reaction
condition responsible for the excess formic acid, through furfuryl alchol, furfural
formation and through the transformation of pyruvaldehyde and D-erythrose. The
mechanistic understanding gained whilst ultilising the formic/levulinic ratio as a
diagnostic will prove valuable in closing the mass balances for hydrolysis systems.
Moving to the more complicated ethanolysis hexose mechanisms, the transportation
biofuel components 5-ethoxymethylfurfural and ethyl levulinate can be synthesised
in
a
homogeneous
sulphuric
acid
ethanolysis
system.
D-fructose,
5-
hydroxymethylfurfural, 5-ethoxymethylfurfural are quantified as major species
fractions, summing to 45-85% of the initial fructose mass, with furfural, D-glucose
and D-mannose quantified as minor species fractions, always summing to <5% of the
initial fructose mass. The mechanism fidelity index is introduced as a metric to
evaluate and describe the understanding of the synthetic systems chemistry and is
successfully employed to test the validity of mechanistic propositions. By utilising
numeric kinetic modelling the analysis shows that the fructopyranose starting
material is unlikely to undergo direct transformation to 5-hydroxymethylfurfural
indicating that at least one stable reaction intermediate is required for 5hydroxymethylfurfural production. The model further proposes that experimentally
observed ethyl fructoside species are preferentially converted back to the D-fructose
starting
material
ethoxymethylfurfural,
rather
which
than
is
undergoing
produced
further
by
the
hydrolysis
to
5-
ethylation
of
5-
hydroxymethylfurfural. The modelling analysis identifies the hydration of 5ethoxymethylfurfural to ethyl levulinate as the slowest reaction in the system and
256
Chapter 8
that this is not the sole pathway responsible for ethyl levulinate formation. The
presence of water in the solvent media is shown to aggressively retard the overall
rate of reaction with respect to an entirely ethanol medium. For example the addition
of 10% v/v water to the ethanol media is shown to slow reaction rates by an order of
magnitude. This is important as the model shows that there is a significant amount of
in-situ water formation produced from the transformation of D-fructose to the
intended biofuel components. This will be of critical kinetic consequence when
hydrolysis is conducted on more concentrated solutions in ethanol and indeed in
other non-aqueous solvents. Importantly, in ethanol media, the hydrogen cation
concentration is shown to be a dynamic function of time and not catalytically
recycled. Incorporating this behaviour in the kinetic modelling analysis is
demonstrated to be crucial to deriving meaningful kinetic and mechanistic
parameters. This will prove important when formulating kinetic models for hexose
transformations in other solvents such as γ-valerolactone.
It is demonstrated that when conducting temperature dependent kinetic modelling in
a system with multiple competing reactions, the system has to be broken into
individual sub-components. The kinetic model is then formulated by building each
of these sub-components in a hierarchical manner. This method should be exercised
when formulating kinetic models on all such systems, as it generates enough
constraints to derive realistic kinetic parameters. Using this approach the D-fructose
ethanolysis sub-system in broken into D-fructose, 5-hydroxymethylfurfural and 5ethoxymethylfurfural subcomponents to derive a kinetic model for the ethanolysis of
D-fructose to 5-ethoxymethylfurfural. Condensed phase conditions (331-351 K)
catalysed by H2SO4 (0.035-0.13 mol/L) in a one-pot system favour the
formation of 5-ethoxymethylfurfural as opposed to ethyl levulinate. This is so,
257
Chapter 8
as
the
reaction
pathway
accounting
for
the
formation
of
5-
ethoxymethylfurfural through the etherification of 5-hydroxymethylfurfural is
found to be seventeen times faster than the pathway responsible for its
consumption at 351 K, with the relationship between formation and
consumption, found to be largely dependent on temperature. Extrapolation of
the model indicates that 353 K is the optimum temperature for 5ethoxymethylfurfural formation from D-fructose within the range of conditions
investigated in this study.
It appears that the strategy of basing kinetic models on the D-fructose mechanism in
the first step in a hierarchical plan linking the hexose sugar to cellulose to biofuel
component seems justified.
This is so, as the yields of the hexose derived
preferential fuel components are shown to be ~20% higher from D-fructose than
from D-glucose. Unexpectedly, the mechanisms for the ethanolysis of D-glucose and
D-fructose to ethyl levulinate are found to be different. The main reaction flux from
D-glucose to ethyl levulinate progresses through no furanic intermediates. Therefore
to maintain the mechanistic philosophy as outlined in this Thesis it will be necessary
to employ an in tandem Lewis/Bronsted acid system to facilitate glucose to fructose
isomerisation to maximise yields of ethyl levulinate with lignocellulosic glucose as
reactant. The diversity of potential fuel components produced in the one-pot
ethanolysis system means that by integrating mechanistic understanding and fuel
properties of the synthesized molecules, drop-in tailor-made fuel additives can be
synthesized in a highly flexible “one-pot” process designed for specific purposes.
258
Chapter 8
8.2. Recommendations for future work
To further build on the ethanolysis system as presented in this Thesis, future work
should be focused on the following main core components.
8.2.1. Utilisation of more sophisticated analytical techniques
This Thesis employs basic analytical techniques such as gas and high pressure liquid
chromatography to quantify species concentrations as a function of time; however
this is not sufficient to detect all species in the system, and in particular unstable
intermediates. Mass balances as low as 30% are detected for the D-fructose
ethanolysis system as is described in Chapter 5. In order to gain a more
comprehensive mechanistic understanding of the system’s chemistry it in necessary
to pursue analytical techniques that will help account for every carbon, hydrogen,
and oxygen atom in the system. The most vital analytical challenge in this regard is
to detect the range of unstable intermediates between the D-fructose and 5hydroxymethylfurfural submechanism. In-situ NMR represents the state-of-the art
analytical technique for the detection of such species. Its employment in an ethanol
system as described by Kimura et al. in aqueous systems should considerably aid the
detection of hexose like species, particularly at early stages of reaction (Kimura et al
2011).
8.2.2. Incorporating
the
effect
of
water
concentration
and
reactant
concentration as parameters in the kinetic model.
Chapter 5 illustrates how water in the ethanol media significantly retards reaction
rates and how its presence can alter the reaction mechanism. Considerable amounts
of water are formed from the dehydration of ethanol to diethyl ether. One molecule
259
Chapter 8
of hexose carbohydrate produces at least four molecules of water in the case of the
formation of 5-ethoxymethylfurfural. Considering both of these factors, reactant
concentration will have a significant effect on the mechanism and kinetics of the
system. This Thesis does not consider reactant concentration as a kinetic parameter
owing to the initial complexity and unknown nature of the reaction mechanisms in
the ethanolysis process. However, as there is now a degree of certainty established in
relation to the reaction mechanism it is envisioned that future work will consider
both the effect of water and reactant concentration as distinct kinetic parameters.
8.2.3. Employing
computational
chemistry
to
determine
the
validity
mechanistic propositions
Computational chemistry techniques can be employed to further test and inform the
validity of experimentally derived mechanistic propositions. The equilibrium and the
transition state structures of pertinent molecules can be evaluated using gradientcorrected density functional theory (DFT) with the Becke three-parameter exchange
functional and the Lee-Yang-Parr correlation functional (B3LYP) level of theory as
employed by Yang et al. (Yang et al 2015). This theoretical platform can be used to
validate experimental postulations and contribute to a more fundamental
understanding of the thermodynamics of the systems by computing potential energy
surfaces (PES). This information can supplement the existing kinetic parameters to
formulate a more fundamental high fidelity kinetic model. For example Yang et al.
excerised PES to postulate a pathway for formic acid formation from D-glucose
through a fufuryl alcohol intermediate, that couldn’t be articulated fully by
experiment.
260
Chapter 8
8.2.4. Formulating
kinetic
models
incorporating
glucose
to
fructose
isomerisation
The next step in the development of a hierarchical kinetic model linking D-fructose
to cellulose is developing reaction pathways in the model facilitating D-glucose/ Dfructose
isomerisation
in
an
ethanol
solvent.
Typically,
glucose/fructose
isomerisation is facilitated by employing in tandem Lewis/Bronsted acid systems.
Using this approach Swift et al. developed a micro kinetic model to describe this
reaction in aqueous systems (Swift et al 2015). Deriving a similar model with
ethanol as solvent would greatly improve the viability of the ethanolysis system.
8.2.5. Conducting fuel property analysis on ethanolysis fuel products
Chapter 7 identifies the range of molecules that can be produced in the ethanolysis
synthesis and calculates some basic fuel properties. Future work should focus on
deriving experimentally derived fuel properties of the various designed ethanolysis
mixtures. Particular attention should be given to properties such as; ignition delay
time, research octane numbers, centane number and miscibility and stability of the
mixtures when blended with typical gasoline and diesel.
8.3. References
Baumeister, C., Kilian, L. (2016) “DP11035 forty years of oil pricefluctuations: why
the price of oil may still surprise us.”
Kimura, H., Nakahara, M., Matubayasi, N. (2011) “In situ kinetic study on
hydrothermal transformation of d-glucose into 5-hydroxymethylfurfural through
D-fructose with 13C NMR,” The Journal of Physical Chemistry A, 115(48),
14013–14021.
261
Chapter 8
NASDAQ (2016) 'Crude -oil', [online], available:
http://www.nasdaq.com/markets/crude-oil.aspx [acessed 03 March 2016].
NASDAQ (2012) 'Crude -oil', [online], available:
http://www.nasdaq.com/markets/crude-oil.aspx [acessed 07 December 2012].
Swift, T.D., Nguyen, H., Anderko, A., Nikolakis, V., Vlachos, D.G. (2015) “Tandem
Lewis/Brønsted homogeneous acid catalysis: conversion of glucose to 5hydoxymethylfurfural in an aqueous chromium (iii) chloride and hydrochloric
acid solution,” Green Chemistry, 17(10), 4725–4735.
Yang, L., Tsilomelekis, G., Caratzoulas, S., Vlachos, D.G. (2015) “Mechanism of
Brønsted Acid-Catalyzed Glucose Dehydration,” ChemSusChem.
262