 

I 5.53.,

1A....

:1,

 

 

2.2.. 3.x. $5.3

.. h».

”3.?“ , .. .3
.. ... ._ «Jaw.

ari . . 3

ix.‘ W.’.W—'h‘t¢”ﬁ
.0..-
an.» .ﬁiﬂ .
ﬁﬁaﬁmgmﬁm
3m. .3

:1.
, 13.x.

... L!
._.{Q .I.
25.4.4.0”.

 

 

LIBRARIES
MICHIGAN STATE UNIVERSITY
EAST LANSING, MICH 48824-1048

This is to certify that the
dissertation entitled

CATALYTIC PROCESSING OF MONOSACCHARIDES AND
SUGAR ALCOHOLS TO VALUE-ADDED PRODUCTS

presented by

Michael William Shafer

has been accepted towards fulﬁllment
of the requirements for the

Chemical Engineering and

PhD- degree in Material Science

 

 

Major Pfofessor's Signature

12/21/04

Date

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m

 

 

CATALYTIC PROCESSING OF MONOSACCHARIDES
AND SUGAR ALCOHOLS TO VALUE-ADDED
PRODUCTS

By
Michael William Shafer

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Chemical Engineering and Material Science

2004

ABSTRACT

CATALYTIC PROCESSING OF MONOSACCHARIDES AND SUGAR
ALCOHOLS TO VALUE-ADDED PRODUCTS

By

Michael William Shafer

The conversion of renewable, bio-based sugars and sugar alcohols to desired C3
polyols, propylene glycol (PG) and glycerol (G0), is the subject of this dissertation.
Catalytic hydrogenolysis of sugar alcohols can give high yields of PG and GO (> 55%);
the goal of this work is to achieve even better yields by gaining a more thorough
understanding of the underlying mechanism of the reaction.

Through adsorption studies of sorbitol on ruthenium metal and related H—D
exchange experiments involving sorbitol hydrogenolysis, we conclude that catalytic
hydrogenolysis is the combination of three chemistries: dehydrogenation of the sugar
alcohol, alkaline degradation of the unsaturated sugar analog, and hydrogenation of
cleaved aldehydic intermediates from the sugar to form polyols. 13C NMR analysis of
sorbitol hydrogenolysis in deuterated water showed incorporation of deuterium at C1 of
sorbitol, which only can occur with dehydrogenation. Addition of soluble base to
sorbitol in the presence of catalyst or to glucose in solution produced signiﬁcant amounts
of lactic acid and other organic acids, indicative of alkaline degradation of sugars via
retro-aldol condensation described previously in the literature.

Alkaline degradation of monosaccharides (glucose, fructose, etc.) enables
cleavage of sugar feedstock at low temperature with high selectivity toward C3 products.

In addition, intermediates from alkaline degradation can hydrogenate to form propylene

glycol and glycerol. Used simultaneously, the combination of chemistries is a novel
approach to convert sugar feedstock to PG and GO. Observation and characterization of
alkaline degradation with hydrogenation is the major contribution of this research.

Combining alkaline degradation of sugar and hydrogenation cleaves sugar
feedstock and produces PG and G0 at temperatures and pressures much lower than those
required for hydrogenolysis (40°C, 100 psi vs. 200°C, 1200 psi). The C3 yield ﬁ'om
alkaline degradation with hydrogenation was comparable to that of hydrogenolysis (54%
vs. 55%) but at much milder conditions. Analysis of the product distribution also showed
potential for further improvement. In some instances, over 80% of the feedstock that
underwent alkaline degradation formed desired C3 products. However, a signiﬁcant
portion of the feedstock hydrogenated to form sugar alcohols, signiﬁcantly reducing C3
yield. Specialized catalysts were tested to selectively hydrogenate C3 intermediates to
reduce formation of sorbitol and mannitol and ﬁrrther improve PG and GO yields.

A kinetic model of the alkaline degradation with hydrogenation reaction was
developed to mathematically describe the reaction system and provide a prognostic tool
for product distributions based on feedstock composition. Concentration versus time data
from ten experiments was used to ﬁt a parametric model that accounted for isomerization
of sugar feedstock, formation of intermediates ﬁ'om alkaline degradation, hydrogenation
of feedstock and intermediates, and adsorption and desorption of products.

Methane production was also explored using carbohydrate feedstock. Glucose,
sorbitol, and glycerol readily converted to methane with >90% yield using a 5% Ru/C
catalyst at 200°C and 1200 psi H2. The process may prove economically viable if raw

materials are produced on site instead of purchased from an outside source.

To Elizabeth

Foryourpaﬁence
For your laughter
Foryourspnﬁ
For your love

For you

iv

ACKNOWLEDGMENTS

I would ﬁrst like to thank Dr. Dennis Miller for his guidance and support as my
advisor through my Ph.D. studies. I would also like to thank Drs. James Jackson,
Lawrence Drzal, Bruce Dale, and Todd Werpy for serving on my graduate committee. A
sincere thank you to group members Frank Jere, Lars Peereboom, Dalila Kovacs, Simona
Marincean, Navin Asthana, Atul Dhale, Zhigang Zhang, Dushyant Shekhawat, and
Shubham Chopade for discussion and camaraderie. I also want to share my appreciation
of Lauren Hendry and Brian Hogle for their assistance with running experiments.

Finally, a word of thanks to the support staff in the ofﬁce of Chemical Engineering and

Material Science. Thank you all!

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... xii

LIST OF FIGURES ................................................................................... xv

LIST OF SYMBOLS AND ABBREVIATIONS ............................................. xix
Chapter 1 - Introduction ......................................................................... 1
1.1 Background .................................................................................. 1
1.2 Signiﬁcance .................................................................................. 3
1.3 Literature Review ......................................................................... 5
1.3.1 Hydrogenation ..................................................................... 5
1.3.2 Hydrogenolysis .................................................................... 8

1.3.3 Alkaline degradation ............................................................ 14

1.4 Summary .................................................................................... 16

Chapter 2 - Experimental Methods ........................................................ 18

2.1 Hydrogenolysis .............................................................................. 18

2.1.1 Reactor set-up ...................................................................... 18

2.1.2 Materials for reaction ............................................................ 18

2.1.3 Operating procedures ........................................................... 19

2.1.4 Analysis .......................................................................... 20

2.2 Adsorption studies .......................................................................... 20

2.2.1 Reactor set-up ...................................................................... 20

2.2.2 Materials for Reaction ........................................................... 21

2.2.3 Operating procedures ............................................................ 21

vi

2.2.4 Analysis ................................................................................ 22

2.3 Alkaline degradation ....................................................................... 23
2.3.1 Reactor set-up ....................................................................... 23
2.3.2 Materials for reaction ............................................................. 23
2.3.3 Operating procedures ............................................................ 24
2.3.4 Analysis ........................................................................... 24

2.4 Alkaline degradation with hydrogenation ........................................ 25
2.4.1 Reactor set-up ....................................................................... 25
2.4.2 Materials for reaction ............................................................. 25
2.4.3 Operating procedures ............................................................ 26
2.4.4 Analysis ................................................................................. 27

2.5 Selective hydrogenation of intermediates ....................................... 27
2.5.1 Reactor set-up ....................................................................... 27
2.5.2 Materials for reaction ............................................................. 28
2.5.3 Operating procedures ............................................................ 29
2.5.4 Analysis ................................................................................. 29

2.6 Conversion of carbohydrates to methane via hydrogenolysis ........ 30
2.6.1 Reactor set-up ....................................................................... 30
2.6.2 Materials for reaction ............................................................. 30
2.6.3 Operating procedures ............................................................ 30
2.6.4 Analysis ................................................................................. 31

2.7 Catalyst characterization ................................................................ 32
2.7.1 Physisorption ......................................................................... 32

vii

2.7.2 Chemisorption ....................................................................... 33

 

Chapter 3 - Preliminary sorbitol hydrogenolysis studies ..................... 36

3.1 Hydrogenolysis experiments ........................................................... 36

3.1.1 Overview ........................................................................... 36

3.1.2 Mesh size study ................................................................ 37

3.1.3 Continuous base addition ..................................................... 42

3.2 Adsorption studies ........................................................................... 43

3.2.1 Overview ................................................................................ 43

3.2.2 Sorbitol adsorption experiments ........................................... 44

3.2.3 Results and Discussion ...................................................... 47

3.2.4 Summary ........................................................................... 54
Chapter 4 - Metal and base-catalyzed C-C cleavage of sugar alcohol and

monosaccharide feedstock 56

4.1 Metal-catalyzed alkaline degradation ........................................... 56

4.1.1 Control experiments .............................................................. 56

4.1.2 Effect of base loading on product distribution ..................... 57

4.1.3 Effect of catalyst metal on product distribution ..................... 62

4.1.4 Effect of temperature ............................................................. 64

4.1.5 Effect of feedstock ................................................................ 65

4.1.6 Summary ........................................................................... 67

4.2 Solution-based alkaline degradation ........................................... 68

4.2.1 Effect of temperature - alkaline degradation .......................... 69

4.2.2 Effect of feedstock sugar — alkaline degradation ................... 7O

viii

4.2.3 Effect of initial sugar concentration - alkaline degradation 70

4.2.4 Summary ........................................................................... 72

Chapter 5 - Alkaline degradation with hydrogenation .......................... 74
5.1 Overview ...................................................................................... 74
5.2 Parametric study: alkaline degradation with hydrogenation .......... 74

5.2.1 Effect of catalyst ................................................................ 76
5.2.2 Effect of temperature ............................................................. 77
5.2.3 Effect of hydrogen pressure .................................................. 79
5.2.4 Effect of catalyst loading ..................................................... 83
5.2.5 Effect of initial base concentration ......................................... 86
5.2.6 Effect of initial sugar concentration ........................................ 88
5.2.7 Effect of feedstock sugar ...................................................... 89
5.3 Summary and signiﬁcance of results ........................................... 90

Chapter 6 - A kinetic model for simultaneous alkaline degradation and

hydrogenation ...................................................................................... 94
6.1 Overview ...................................................................................... 94
6.2 Mass transfer calculations ......................................................... 94

6.2.1 Catalyst suspension .............................................................. 94
6.2.2 Three phase mass transfer .................................................... 95
6.2.3 Reaction rate and mass transfer ........................................... 97
6.2.4 Gas-liquid mass transfer coefﬁcient ....................................... 98
6.2.5 Liquid-solid mass transfer coefﬁcient ................................ 101
6.2.6 lntraparticle mass transfer ...................................................... 102
6.2.7 Results from mass transfer calculations ................................ 103

ix

6.3 The kinetic model ........................................................................... 104

6.3.1 Hydrogenation ....................................................................... 105
6.3.2 Alkaline degradation .............................................................. 113
6.3.3 Combining the equations ...................................................... 116
6.3.4 Assumptions of the model ...................................................... 118
6.4 Calculation of concentration versus time ......................................... 118
6.5 Results ...................................................................................... 120
6.5.1 lsomerization equilibria ...................................................... 125
6.5.2 Hydrogenation rate constants ........................................... 127
6.5.3 Adsorption equilibrium constants ........................................... 128
6.5.4 Alkaline degradation reactions ........................................... 129
Chapter 7 - Continued development of alkaline degradation with

hydrogenation and related work ........................................... 130

7.1 Modiﬁed catalyst for selective C3 hydrogenation ............................. 130
7.1.1 Selective 03 hydrogenation in the presence of glucose ........ 130

7.1.2 Alkaline degradation with hydrogenation using Pt catalysts 131

7.2 Methane production from carbohydrate feedstock .......................... 134
7.2.1 Sorbitol to methane ............................................................... 134

7.2.2 Glucose to methane .............................................................. 135

7.2.3 Glycerol to methane .............................................................. 136

7.3 Summary ...................................................................................... 137
Chapter 8 - Conclusions and Future Work ........................................... 140
8.1 Summary ...................................................................................... 140
8.2 Conclusions .................................................................................... 141

8.2.1 Adsorption studies ................................................................. 141

8.2.2 Alkaline degradation with hydrogenation ................................ 142
8.2.3 03 selective hydrogenation .................................................... 143
8.2.4 Kinetic model of alkaline degradation with hydrogenation ...... 144
8.2.5 Conversion of carbohydrates to methane .............................. 145

8.3 Process economics of alkaline degradation with hydrogenation 146

8.4 Future work ...................................................................................... 149

8.4.1 Catalyst screening for alkaline degradation with hydrogenation 149

8.4.2 Hydrogenolysis of carbohydrates to methane ..................... 149
8.4.3 Catalyst characterization ...................................................... 150
Appendix A - Rate equations for the kinetic model ................................ 151

Appendix B - Visual Basic coding used for kinetic model calculations 156

Appendix C - Comparison of kinetic model to experimental data for alkaline

degradation with hydrogenation .............................................................. 167
Appendix D - Master list of experiments ................................................ 195
References ................................................................................................. 207

xi

LIST OF TABLES

Table 1-1: Hydrogenolysis patents ........................................................................... 13

Table 1-2: Summary of prior art for lactic acid production via alkaline degradation 15

Table 2-1: Order of charging the reactants for experiments ........................................... 29
Table 3-1: Product distribution of trickle-bed experiments ............................................ 38
Table 3-2: Product distribution of continuous base addition experiments ....................... 43

Table 3-3: Liquid and gas phase products of sorbitol adsorption at 120°C (Exp. 163)
Conditions: 120°C, 100 psig He, 10.0g Ru sponge, 50 mg sorbitol, 160 mL solution 50

Table 3-4: Calculations for Em of sorbitol dehydrogenation [ -ln (k) vs. 1 /T ]

Conditions: 0.5 g/L, 10.0g ruthenium sponge, 100 psi helium ........................................ 52
Table 4-1: C3 product distribution - effect of initial base concentration ....................... 61
Table 4—2: C1 and C2 product distribution — effect of initial base concentration ............. 62
Table 4-3: C3 product distribution at 6 hours — eﬂ‘ect of catalyst metal .......................... 63
Table 44: C2 product distribution at 6 hours — eﬁ‘ect of catalyst metal .......................... 64
Table 4-5: C3 product distribution - eﬂ‘ect of reaction temperature ................................ 64
Table 445: C2 product distribution — effect of reaction temperature ................................ 65

Table 4-7: Product distribution from alkaline degradation of xylitol (Experiment 79) 66

Table 4—8: Product distribution from alkaline degradation of glycerol ............................ 67
Table 4—9: C2 product distribution — alkaline degradation of glycerol ............................. 67
Table 4-10: Product distribution — alkaline degradation of glucose ................................ 69
Table 4-11: C3 yield of high and low concentrations of xylose and glucose .................... 72
Table 4-12: pK. values for glucose and fructose at varying concentrations .................... 72

Table 5-1: Overall C3 production and reaction comparison — Eﬂ‘ect of catalyst metal
Conditions - 40°C, 100 psig, 0.18 M fructose, 0.85 M KOH, 0.1g catalyst ...................... 76

Table 5-2: C3 product distribution of Ru/C and Pd/C experiments
Conditions - 40°C, 100 psig, 0.18 M fructose, 0.85 M KOH, 0.1g catalyst ...................... 77

xii

Table 5-3: C3 and C6 production and reaction comparison - Effect of temperature
Conditions - 100 psig, 0.18 M fructose, 0.85 M KOH, 0. lg 5% Ru/C catalyst ................. 78

Table 5-4: C3 product distribution - Eﬂ‘ect of temperature
Conditions - 100 psig, 0.18 M ﬁ'uctose, 0.85 M KOH, 0. lg 5% Ru/C eatalyst ................. 79

Table 5-5: Selectivity of alkaline degradation with hydrogenation — H2 pressure eﬂ‘ect
Conditions - 40°C, 0.18 M fructose, 0.85 M KOH, 0. lg 5% Ru/C catalyst ...................... 79

Table 5-6: Selectivity of alkaline degradation with hydrogenation - H2 pressure eﬁ‘ect
Conditions - 40°C, 0.18 M fructose, 0.85 M KOH, 0. lg 5% Ru/C catalyst ...................... 80

Table 5—7: Comparison of catalyst loading for alkaline degradation with hydrogenation
Conditions — 40°C, 0.18 M fructose, 0.85 M KOH, 100 psig H2, 5% Ru/C eatalyst ......... 84

Table 5—8: Selectivity of EG, PG, GO, and LA - effect of catalyst loading
Conditions - 40°C, 0.18 M fructose, 0.85 M KOH, 100 psig H2, 5% Ru/C eatalyst ......... 86

Table 5-9: C3 and C6 selectivity of reaction with varying base concentration
Conditions - 40°C, 0.18 M fructose, 100 psig H2, 0. lg 5% Ru/C catalyst ........................ 87

Table 5-10: Selectivity of EG, PG, GO, and LA - effect of initial base concentration
Conditions — 40°C, 0.18 M fructose, 100 psig H2, 0. lg 5% Ru/C eatalyst ........................ 87

Table 5-11: C3 and C5 selectivity of reaction with varying sugar concentration
Conditions - 40°C, 0.85 M KOH, 100 psig H2, 0. lg 5% Ru/C catalyst ........................ 88

Table 5-12: C3 and C6 selectivity product distribution - effect of feedstock sugar
Conditions — 40°C, 0.18 M sugar, 0.85 M KOH, 100 psig H2, 0. lg 5% Ru/C catalyst ..... 89

Table 5-13: Selectivity of EG, PG, GO, and LA - effect of feedstock sugar
Conditions — 40°C, 0.18 M sugar, 0.85 M KOH, 100 psig H2, 0. lg 5% Ru/C catalyst ..... 90

Table 5-14: Comparison of Experiments 144 and 153 .................................................. 93

Table 6-1: Mass transfer calculations for Experiment 140 .............................................. 103
Table 6-2: Abbreviations used in kinetic model ............................................................. 105
Table 6-3: Experimental conditions used for kinetic model - 40°C, 0.85 M KOH ......... 119
Table 64: Best ﬁt parametric values ................................................................................ 124
Table 6-5: Sugar isomerization equilibria values (kfmm / km) ................................... 125
Table 6-6: Equilibrium constants, reaction AG° (kJ/mol), and total AG° (kl/moi) .......... 126
Table 6-7: Comparing equilibrium values with hydroxide concentrations at 78°C ........ 127
Table 6-8: Parameters for adsorption/desorption pathways ............................................ 129

xiii

Table 7-1: Comparison of Ru/C and Pt(S)/C eatalysts
Conditions - 40°C, 100 psig, 0.18 M fructose, 0.85 M KOH, 0. lg catalyst (dry basis) ..... 133

Table 7-2: Yield of methane and ethane from hydrogenolysis of sorbitol

Conditions — 1200 psig total pressure, 1.0g catalyst, 0.55 M sorbitol ............................... 135
Table 7-3: Yield of methane and ethane from hydrogenolysis of glucose

Conditions — 1200 psig total pressure, 1.0g catalyst, 0.55 M glucose ............................... 136
Table 7-4: Yield of methane and ethane from hydrogenolysis of glycerol

Conditions - 1200 psig total pressure, 1.0g catalyst, 1.1 M glycerol ................................. 136
Table 7-5: Estimated price of methane production with bio-based feedstock ................. 138
Table 8-1: Raw material cost for Experiment 152 (Basis: 1 lb of fructose) .................... 147
Table 8-2: Raw material cost for alternate feedstock (Basis: 1 lb of glucose) ................ 148
Table A-1: Abbreviations used for rate constants and kinetic model .............................. 152

Table A-2: Description of rate constants and abbreviations used in the kinetic model 153

xiv

Figure 1-1:

Figure 1-2:

Figure 1-3:
Figure 2-1:
Figure 3-1:
Figure 3-2:

Figure 3-3:

Figure 3-4:
Figure 4-1:
Figure 4-2:

Figure 4-3:

Figure 5-1:

Figure 5-2:

Figure 5-3:

Figure 6-1:

Figure 6-2:

Figure 6-3:

LIST OF FIGURES

Key molecules formed during reaction ....................................................... 2

Sugars and sugar alcohols of interest as feedstock and reaction

intermediates ............................................................................................. 3
Pathways of sorbitol hydrogenolysis ........................................................... 9
Schematic for continuous trickle-bed reactor .............................................. 19
Comparison of (a) fully wetted and (b) partially wetted catalyst ................ 39
13C NMR study of sorbitol over Ru/C - evidence of dehydrogenation ....... 45
Concentration of sorbitol and total liquid phase products vs. Time

Sorbitol adsorption experiments .................................................................. 48
Determination of activation energy of sorbitol dehydrogenation ................ 53
pH and conversion vs. time - low base ........................................................ 58
pH and conversion vs. time - excess base .................................................... 58
Total sugar concentration from mannose, glucose, and fructose

feedstock vs. Time Conditions - 22°C, 0.002 M sugar, 1.0 M NaOH ......... 71

Concentration of sorbitol (SO) and mannitol (M0) vs. Time - Hydrogen eﬂ‘ect
Conditions - 40°C, 0.18 M fructose, 0.85 M KOH, 0. lg 5% Ru/C catalyst 81

Concentration of propylene glycol (PG) and glycerol (G0) vs. Time — Hydrogen
eﬂ‘ect Conditions — 40°C, 0.18 M fructose, 0.85 M KOH,

0.1g 5% Ru/C catalyst .................................................................................. 82
Concentration of sorbitol (SO) and mannitol (M0) vs. Time —

Effect of catalyst loading Conditions - 40°C, 0.18 M ﬁ'uctose, 0.85 M KOH,
100 psig H2, 5% Ru/C catalyst ..................................................................... 85
Mass transfer in a three phase catalytic reaction .......................................... 96
Derived result for fractional coverage of vacant sites of type 1 (SW) for
adsorption of hydrogenation products (Equations 47-50) ............................... 111
Reaction pathways of proposed alkaline degradation with hydrogenation model

k”: Rate constant for reactant (A) to product (B) .................................... 117

XV

Figure 6-4:

Figure 6-5:

Figure 6-6:

Figure 7-1:

Figure 7-2:

Figure 01:

Figure C-2:

Figure C-3:

Figure C-4:

Figure C-5:

Figure C-6:

Figure C-7:

Figure C-8:

Comparison of experimental data and kinetic model - Experiment 146
3.17% fructose, 5.3% KOH, 40°C, 100 psig Hz, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time .................... 121

Comparison of experimental data and kinetic model - Experiment 146
3.17% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ........... 122

Comparison of experimental data and kinetic model - Experiment 146
3.17% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Condensation products, fomtic acid, EG, and glycolic acid vs. Time ........ 123

Selective hydrogenation of glyceraldehyde with glucose
Feedstock: 0.2 wt% glyceraldehyde, 0.2 wt% glucose

Conditions: 40°C, 200 psig H2, 0. lg - 3% Pt(S)/C .................................... 132
Selective hydrogenation of glyceraldehyde with glucose

After four hours of reaction

Conditions: 40°C, 200 psig H2, 0. lg - 3% Pt(S)/C .................................... 132

Comparison of experimental data and kinetic model - Exp 115
3.16% ﬁ'uctose, 5.3% KOH, 40°C, 100 psig H2, 0.05 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 168

Comparison of experimental data and kinetic model - Exp 115
3.16% ﬁ'uctose, 5.3% KOH, 40°C, 100 psig H2, 0.05 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 169

Comparison of experimental data and kinetic model - Exp 115
2.0% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.1 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 170

Comparison of experimental data and kinetic model - Exp 119
1.00% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 171

Comparison of experimental data and kinetic model - Exp 119
1.00% ﬁ'uctose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 172

Comparison of experimental data and kinetic model - Exp 119
1.00% fructose, 5.3% KOH, 40°C, 100 psig [-12, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 173

Comparison of experimental data and kinetic model - Exp 140
3.17% fructose, 5.3% KOH, 40°C, 200 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 174

Comparison of experimental data and kinetic model - Exp 140

3.17% fructose, 5.3% KOH, 40°C, 200 psig H2, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 175

xvi

Figure 09:

Figure C-10:

Figure C-1 1:

Figure C-12:

Figure C-13:

Figure C-l4:

Figure C-15:

Figure 016:

Figure C-l7:

Figure C-l8:

Figure C-19:

Figure C-20:

Comparison of experimental data and kinetic model - Exp 140
3.17% fructose, 5.3% KOH, 40°C, 200 psig H2, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 176

Comparison of experimental data and kinetic model - Exp 141
3.17% glucose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 177

Comparison of experimental data and kinetic model - Exp 141
3.17% glucose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 178

Comparison of experimental data and kinetic model - Exp 141
3.17% glucose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 179

Comparison of experimental data and kinetic model - Exp 142
3.17% mannose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 180

Comparison of experimental data and kinetic model - Exp 142
3.17% mannose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 181

Comparison of experimental data and kinetic model - Exp 142
3.17% mannose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 182

Comparison of experimental data and kinetic model - Exp 144
2.0% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 183

Comparison of experimental data and kinetic model - Exp 144
2.0% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 184

Comparison of experimental data and kinetic model - Exp 144
2.0% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 185

Comparison of experimental data and kinetic model - Exp 147
3.17% fructose, 5.3% KOH, 40°C, 25 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 186

Comparison of experimental data and kinetic model - Exp 147

3.17% ﬁ'uctose, 5.3% KOH, 40°C, 25 psig H2, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 187

xvii

Figure 021:

Figure C-22:

Figure 023:

Figure 024:

Figure C-25:

Figure C-26:

Figure 027:

Comparison of experimental data and kinetic model - Exp 147
3.17% ﬁ‘uctose, 5.3% KOH, 40°C, 25 psig HZ, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 188

Comparison of experimental data and kinetic model - Exp 149
3.17% fructose, 5.3% KOH, 40°C, 50 psig H2, 0.10 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 189

Comparison of experimental data and kinetic model - Exp 149
3.17% ﬁ'uctose, 5.3% KOH, 40°C, 50 psig 1-12, 0.10 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 190

Comparison of experimental data and kinetic model - Exp 149
3.17% ﬁ'uctose, 5.3% KOH, 40°C, 50 psig H2, 0.10 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 191

Comparison of experimental data and kinetic model - Exp 152
3.17% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.20 grams 5 wt% Ru/C
Fructose, glucose, mannose, sorbitol, and mannitol vs. Time ................. 192

Comparison of experimental data and kinetic model - Exp 152
3.17% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.20 grams 5 wt% Ru/C
Lactic acid, glycerol, propylene glycol, and glyceric acid vs. Time ....... 193

Comparison of experimental data and kinetic model - Exp 152

3.17% fructose, 5.3% KOH, 40°C, 100 psig H2, 0.20 grams 5 wt% Ru/C
Condensation products, formic acid, EG, and glycolic acid vs. Time ..... 194

xviii

LIST OF SYMBOLS AND ABBREVIATIONS

Engﬁsh:

a

Aarca
ADS

gas-liquid interfacial area per unit volume of reactor [cm2 / cm3]
ratio of catalyst surface area to volume of ﬂuid [m2 / m3]
Effective area of 1 metal atom [m2 / atom of metal]

adsorption

AWE.“ = Atomic weight of the supported metal [g metal / atom of metal]

CA, 1,
CA, s
CH2, 1.
CH2, s
Cir-s2
Cs
Cso
CV2
CD
CZ

concentration of liquid phase reactants [mol / L]

concentration of liquid phase reactants in the catalyst [mol / L]
concentration of hydrogen in the liquid phase [mol / L]
concentration of hydrogen in the catalyst [mol / L]
concentration of type 2 sites occupied by monatomic hydrogen
species surface concentration [mol / L]

Concentration of sorbitol [mol / liter]

concentration of vacant type 2 sites

condensation products

Cannizzaro reaction products

% Disp = Metal dispersion - fraction of metal available for adsorbate interaction

di

dP
d'r
DA
De
DES
.138mt
EG
F
FA

8

impeller diameter [m]

Particle diameter [m]

reactor diameter [m]

diﬂ’usion rate of species A [m / sz]
e21). = effective diffusivity [m2 / s]

desorption

= Activation energy [J / mol ]

ethylene glycol
fructose
formic acid

Gravitational constant [9.8066 m/52]

xix

kg- m

 

gc = Gravitational proportionality factor [1 szN ]
G = glucose

U d3 2
Ga = Galileo number of the liquid phase = L PpLg

[dimensionless]
HiO‘SBy

GA = glyceraldehyde
GA2 = glycolaldehyde
GLYA = glyceric acid

GLYA2 = glycolic acid

G0 = glycerol

H = Henry’s Law constant [atm / (mole solute / mole solution)]
H2, H2 = hydrogen

k = 1“ order rate constant [ hr '1]

kwws = rate constant for hydrogen adsorption

km)“ = rate constant for hydrogen desorption

k1, = liquid ﬁlm mass transfer coeﬂicient [m / 5]
km = gas-liquid mass transfer [1 /m - s]

k0 = pre-exponential factor [ hr '1]

k, = liquid-solid mass transfer coefficient [m / s]

km a = liquid-solid mass transfer of liquid phase reactants [1 / m - s]

ks,rr2 a = liquid-solid mass transfer [1 lm - s]

K = equilibrium constant - kADs / kpgs

Kg = equilibrium constant for hydrogen (kiMDs / knees)

L = ratio of catalyst surface area per unit volume ﬂuid [m2 / m3]
L = d,, / 6 = characteristic length of the catalyst [m]

LA = lactic acid

M = mannose

MA = Metal surface area [m2 / g sample]

M0 = mannitol
N = stirring speed [3"]
Nmin = minimum stirring speed for catalyst suspension [rotations / sec]

Pm
PA
PG

I'm-ADS

4'80

Ref

Si, SI
82, 82

ST
SFcalc
Sh
SO

W

pressure

partial pressure of solute (112) in the gas phase
hydrogen pressure

pyruvaldehyde

propylene glycol

reaction rate [mol / L - 3]

rate of hydrogen adsorption

dC
7‘?)- = Rate ofsorbitol dehydrogenation [3101/ liter / hf]
Idealgasconstant[J/mol/K]

reaction rate
observed reaction rate of the species [kmol / m3 - s]

. . ULpLdP . .
Reynolds number of the liquid phase = — [drmensronless]
11 r. (1 " 8 B )

catalytic site — type 1

catalytic site — type 2

External surface area of the catalyst

surface tension [N / m]

Atoms of metal per molecule of adsorbed gas
Sherwood number [dimensionless]

sorbitol

time

Temperature [ K ]

gas velocity [m / s]

Superﬁcial liquid velocity [m/s]

y-int. of linear best ﬁt for adsorbed volume per gram vs. P [cm3 gas/g metal]
volume of the solution [cm3]

Volume of the catalyst

catalyst loading [grams catalyst per 100 grams solution]

Weight % = Weight ﬁ'action of catalyst that is metal [dimensionless]

X

mole fraction of solute (H2) in liquid phase

xxi

Z = Length of the catalyst bed [m]
Greek:
B = constant determined by impeller geometry
AP = Pressure drop through the catalyst [N/mz]
e = support porosity [0.6 for the catalyst]
83 = Packed-bed void per reactor volume (bed porosity) [dimensionless]
neg = Wetting efﬁciency [dimensionless]
rid) 2 = observable modulus
m = Liquid viscosity [___kg ]
rn - s
pi, = Density of the liquid phase [kg/m3]
pp = particle density [kg / m3]

(D, = (6 / dp) / (Sp / VP) = Sphericity of the catalyst

xxii

Chapter 1 - Introduction

1.1 Background

The Miller group at Michigan State University focuses on the evolving
technology of renewable resource reﬁning. We seek to produce commodity chemicals
from readily available bio-based feedstock. Recent group members studied itaconic acid
production], lactic acid hydrogenation to propylene glycolz, stereoretentive amino alcohol
synthesis3, and the processing of sugar alcohols to value—added polyols4, which provides
the thrust of this research.

Previous work focused mainly on hydrogenolysis as the chemistry of choice to
produce C3 polyols, such as glycerol and propylene glycol, from sorbitol and xylitol
feedstocks. Using a continuous trickle-bed reactor, Chopade et a1.4 produced these C3
products with better selectivity and yield than previously reported. The study optimized
reactor conditions to achieve these improved results. It did not include, however, insight
into the mechanisms of the reaction or reasoning on why yields improved with speciﬁc
changes to the reaction environment. The research presented delves into the underlying
chemistries of hydrogenolysis yielding a deeper understanding of the reaction and
alternatives for processing sugars at milder conditions to desired products.

As an introduction to the molecules of interest, Figure 1-1 illustrates desired
products of the reactions as well as key intermediates necessary for elevated C3 yields.

Figure 1-2 provides illustrations of the monosaccharides and sugar alcohols used as

 

' Hogle B. et a1., Ind. Eng. Chem. Res., 2001, 41, 2069-2073
2 Zhang 2. et al., Appl. Cat. A: Gen, 2001, 219, 89-98

3 Jere. F.T., Org. Left, 2003, 5, 527-530

4 Chopade, s. eta1.,U.S. Patent 6,291,725, 2001

 

OH

HO /i\/OH
OH HO

glycerol propylene glycol

lactic0 acid

 

 

 

Desired 03 products

 

1%, \/U\

ruval deh de
glyceraldehyde py y

 

 

 

 

Key 0;, intermediates

 

A

O HO\/iLOl-l
formic acid

HO
glycolic acid glyceric acido

 

 

 

Organic acids formed during reaction

 

OH OH

  
 

OH

HO\/\
OH HO HO

eth lene 1 col OH
y g y threitol crythritol

~-|IIIIIOH

 

 

 

Other polyols formed during reaction

Figure 1-1: Key molecules formed during reaction

feedstock and that form during these reactions. These molecules will be referred to

during the course of this dissertation and are provided here as a reference.

H

    

O OH HO HO OH HO

t\“‘o

HO O___'

HO OH
fructose mannose glucose

H HO HO OH HO

HO HO

      

HO OH HO OH
psi cose mannitol sorbitol

Figure 1-2: Sugars and sugar alcohols of interest as feedstock and reaction intermediates

1.2 Signiﬁcance

Finding alternatives to petroleum-derived feedstocks for commodity chemical
production is the main driver of this research. Petroleum is the carbon source of choice
for virtually all organic chemicalss. Crude oil supplies, however, appear to be waning in
light of increasing global demand. At current usage and production rates, it is estimated
that known reserves of petroleum will be depleted in approximately 40 years,6 although

6.7.8

an exact timetable is debatable. As a result, the US. government is supporting

initiatives to decrease oil dependence9, which includes this work.

 

5 The American Chemical Society, Technology Vision 2020: The (1.5". Chemical Industry, July 2001

6 British Petroleum, BP Statistical Review of World Energy, June 2003

7 British Petroleum, BP Statistical Review of World Energy, June 2002

8 British Petroleum, BP Statistical Review of World Energy, June 2001

9 Biomass Technical Advisory Committee, Vision for Bioenergy and Biobased Products in the United
States, October 2002

Another interest in this research is the economic potential of converting simple
sugars into value-added polyols. Propylene glycol and glycerol sell for $0.65 and $0.75
per pound10 respectively while sugars are available at $0.06 - $0.12 per pound depending
on the degree of reﬁnement“. Annual US. demand for propylene glycol and glycerol is
approximately one billion pounds12 and 500 million pounds13 respectively, leaving
sizeable markets available. If this technology is successful, it has the advantage of an
“environmentally friendly” label as well as a more stable source of carbon.

The speciﬁc thrust of this work is to continue development of sugar reﬁning
technologies and improve them to produce high yields of C3 products (i.e. >75% of
carbon fed to desired products). The research reported here shows potential improvement
on existing technologies as well as an important shift in focus on the reaction
environment. Literature regarding hydrogenolysis focuses on C3 product selectivity and
yield instead of the speciﬁc mechanistic information necessary to understand the
underlying chemistries. This study demonstrates, on the other hand, that hydrogenolysis
is actually the combination of three chemistries: dehydrogenation, alkaline degradation,
and hydrogenation. With this in mind, we deﬁne the effects of these chemistries and

produce an improved picture of how the reactants interact.

 

‘0 Chemical Market Reporter, March 15, 2004, 265

” http://ers.usda.gov/brieﬁng/sugar/Data/datahtm, viewed on August 27, 2004. last update July 13, 2004
'2 Chemical Market Reporter, September 24, 2001, 260

'3 Chemical Market Reporter, December 17, 2001, 260

1 .3 Literature Review
1 .3.1 Hydrogenation

Hydrogenation is the addition of gaseous hydrogen to saturate a double bond (e. g.
C = O or C = C).14 Reactions typically require the use of catalyst with a supported metal,
such as nickel or platinum, at elevated temperatures (3 00-3 50°C) and hydrogen pressures
(500-3000 psi) for the addition to occur. Hydrogenation is used extensively in petroleum
processing to remove sulﬁlr and nitrogen from oil (hydrotreating), to form lower boiling
point compounds from long carbon chain hydrocarbons by cleaving the molecule
(hydrocracking), and a number of other reactions. 15 Of particular interest to this work are
liquid phase hydrogenations to value-added products from biomass-derived materials.

Bowden and Adkins16 described ester hydrogenation using a copper chromate
catalyst at 250°C and 150-200 atm of H2 pressure. Although using mainly phenyl esters
as reactants, the authors reported hydrogenation (more accurately, hydrogenolysis) of
butyl lactate to 1,2-propylene glycol (PG) with 81% yield. Among the substrates studied,
1,2 PG example was unique, compared to the other hydrogenation products, in that it did
not maintain optical activity of the feedstock. More recently, Jere et al.3 found that the
loss of chirality was an effect of elevated temperature and not a direct result of reducing
the carboxylic acid to an alcohol in their work with alanine, a similarly or-substituted

chiral compound.

 

‘4 Vollhardt, K. Peter C. and Schore, N.E., Organic Chemistry, 2nd Edition, p. 250, W.H. Freeman and
Company, New York, 1995

‘5 Kroschwitz, J.-, Howe-Grant, M., editors, Kirk-0thmer Encyclopedia ofchemical technology, 4th Edition.
18, pp. 446-447, Wiley, New York, 1992

‘6 Bowden, 13.; Adkins, H., J. Am. Chem. Soo, 1934, 56, 689-691

Adkins and Pavlicl7 continued work studying ester hydrogenation over Raney
nickel in non-aqueous solvent, namely dry ethanol, noting lower temperatures of 25-7 5°C
and similar pressures (150-200 atm) for hydrogenation of amino acid esters. Another
paper on ester hydrogenation18 showed that high loadings of catalyst beneﬁt product
yields by lowering the reaction temperature and thus, limiting side product formation.
With this work, near quantitative yields of 1,2 PG from ethyl lactate as well as 1,3 diols
from malonic acid derivatives were observed.

Camahan et al.19 demonstrated the reducing ability of ruthenium (as Ru02 and
Ru/C) at lower temperatures than previous catalysts. Succinic, glycolic, adipic, acetic,
and oxalic acid were hydrogenated to their corresponding glycols with yields of 47-80%
at temperatures of 140-160°C and H2 pressures as high as 1000 atm. While these
reactions ran at temperatures comparable to those described by Adkins, ruthenium
allowed for direct hydrogenation of carboxylic acids instead of their esters. Zhang et a1.2
extended this use of ruthenium to the direct hydrogenation of lactic acid to 1,2 PG, in
batch-wise and continuous fashion, with yields in excess of 90% at reduced temperature
(130°C) and pressure (70-140 atm).

A number of papers cover the hydrogenation of sugars to sugar alcohols, which is
of immediate interest to the research at hand. The Kirk-Othmer handbook20 describes
sorbitol production through batch or continuous hydrogenation of glucose using nickel or

Raney nickel as a catalyst. Wisniak et al.21’22'23 described xylose, glucose, and fructose

 

'7 Adkins, H.; Pavlic, A.A., J. Am. Chem. Soc, 1947, 69, 3039-3041

‘8 Adkins, 11,-, Billica, H.R., J. Am. Chem. Soc., 1948, 70, 3121-3125

‘9 Camahan, J.E. et a1., J. Am. Chem. Soc., 1955, 77, 3766-3768

2° Kroschwitz, 1.; Howe-Grant, M., editors, Kirk-Othmer Encyclopedia ofchemical technology, 4‘h Edition,
23, pp. 105-106, Wiley, New York, 1992

2' Wisniak, J.et a1.,1nd. Eng. Chem. Prod. Res. Develop, 1974, 13 (1), 75-79

22 Wisniak, J.; Hershkowitz, M.; Stein, s.,1nd. Eng. Chem. Prod. Res Develop, 1974, 13 (4), 232-236

hydrogenation over ruthenium, platinum, Raney nickel, rhodium, and palladium for
comparison of activity as well as kinetic studies. Papers by Brahme and Doraiswamy“,
Verma and Gehlaway”, and van Gorp et a1.26 gave similar accounts of reaction
conditions and catalysts utilized. Cerino et a1.27 explored the use of promoters to improve
the activity of Raney nickel and combat the common problem of leached metal in the
product stream. Guo et al.28 used a Cr promoter on a Ru-B amorphous catalyst in order
to improve activity as well.

Hegedi'ls et al.29 attempted selective hydrogenation of fructose to mannitol using a
supported copper catalyst and the addition of sodium borate to the reaction. The borate
was thought to interact with ﬁ'uctose to form the B-furanose form of the sugar which is
more selective toward mannitol. Yields for mannitol were as high as 80% with the
drawback of extended reaction times due to inhibition from the borate ions. Heinen et
a1.30 pursued selective hydrogenation to mannitol as well by promoting Pt/C and Pd/C
catalysts with tin for a mannitol yield of 63%.

Mikkola et al.31 discussed the effects of solvent properties on xylose
hydrogenation. Of particular interest was the balance of organic and aqueous solvent.

Water was favorable due to high sugar solubility but organic solutions provided much

 

3’ Wisniak, 1.; Simon, R.; Ind. Eng. Chem. Prod. Res. Dev.. 1979, 18 (1), 50-57

2“ Brahme, P.H.; Doraiswamy, L.K.; Ind. Eng. Chem, Process Des. Dev., 1976, 15 (1), 130-137

25 Verma, R.; Gehlawat, 1.14; J. Chem. Tech. Biotechnol, 1989, 46, 295-301

26 van Gorp, K. et a1., Cat. Today, 1999,52, 349-361

27 Cerino, P.J.', Fleche, G.; Gallezot, P.; Salome, J.P., Heterogeneous Catalysis and Fine Chemicals 1]:
proceedings of the 2nd international symposium, Poitiers, October 2-5, 1990, pp. 231-236, Guisnet M. et
a1. (editors), Elsevier, Amsterdam; New York, 1991

2" Guo, H.; Li, H.; Xu, Y.; Wang, M., Mater. Lett, 2002,57, 392-398

29 Hegediis, M.; Gobolos, S.; Margitfalvi, J .L., Heterogenous Catalysis and Fine Chemicals III:
proceedings of the 3rd international symposium, Poitiers, April 5-8, 1993, pp. 187-194, Guisnet, M. et a1.
(editors), Elsevier, Amsterdam; New York, 1993

3° Heinen, A.W.; Peters, J.A.; van Bekkum, H., Carbohyd. Res, 2000, 328, 449-457

3‘ Mikkola, J.-P. et al., J. Chem. Tech. and Biotech, 2001, 76, 90-100

higher hydrogen solubility allowing for higher hydrogenation rates. The authors offered
an example with an 80:20 water/ethanol blend leading to an approximate 3-fold increase
in hydrogenation activity with respect to aqueous media. Further, they noted that
solutions above 30% ethanol led to precipitation of sugar at the desired loadings.

A ﬁnal paper of note from Kusserow et a1.32 describes hydrogenation of glucose
using nickel and ruthenium dispersed in molecular sieves. The yield of sorbitol was
unimpressive at < 10%, but smaller molecules, such as acrolein, hydrogenate easily. This
preference toward smaller, linear molecules may prove useful for hydrogenation of
alkaline degradation products (see Section 1.3.3) to glycerol and PG without the

competitive formation of sorbitol and mannitol.

1.3.2 Hydrogenolysis

The term hydrogenolysis was coined by Ellis33 as an extension of other similar
reactions, such as hydrolysis, where a bond is broken and then hydrogen is added to the
fragments. As an example, Figure 1-3 illustrates sorbitol hydrogenolysis with potential
products depending on where the carbon chain is broken. Literature results for
hydrogenolysis and alkaline degradation are reported in terms of selectivity and yield.
Selectivity is the number of moles of product formed per mole of reactant converted
while yield is the number of moles of product per mole of reactant fed to the reaction.
All desired products are C3 compounds (propylene glycol, glycerol, lactic acid). This
gives a theoretical maximum for selectivity and yield of 2.0 for C6 feedstock (glucose,
sorbitol, etc.) and 1.0 for C5 feedstock (xylose, xylitol, etc) based on the carbon available

for the reaction.

 

32 Kusserow, 13.; Schimpf, s; Claus, P., Adv. Synth. Catal, 2003, 345, 289-299
33 Ellis, C., Hydrogenation of Organic Substances, 3rd Ed, D. Van Nostrand Company, New York, 1930

max—one??? EELS .«o $33353 an; PEER

._oo>._o
._oo>._o nommoﬁo
mzm3>zpm mzma>aoma
:0 IO N: IO
:6 ,/\ f\
/ iv
o: :o mo

4. a. e\

mi .U I «U N:
._oEmmom mouse

 

 

 

 

 

 

 

 

._O._._m 1.5mm. :ofo 40......_>x
a :35
no no a no a
$05 + m o: E no m «E :6 ill I
4 a o: liv £0 + a Li o:
©m+ =0 Hﬁ: mUINU no I «DI—U IOii I
:35 :oamo :oﬁo

Connor and Adkins34 conducted hydrogenolysis experiments using a Cu/CrOx
catalyst at 250°C and 2600 psig H2 on over 70 compounds. These runs included glycerol
and mannitol showing 85% and 24% selectivity to PG respectively. Natta et a1.35 with a
Cu/CrOx catalyst used glucose in non-aqueous solvents, namely methanol and ethanol, to
form PG with yields as high as 71% at 270°C and 2800 psig H2. Clark36 focused on
glycerol production using sorbitol as the starting material in aqueous media. The author
reported selective carbon chain cleavage at C3—C4 and demonstrated yields of 40%, 17%,
and 16% of glycerol, PG, and EG, respectively, using a Ni/Kieselguhr catalyst at 215°C
and 2000 psig. Van Ling et al.37 compared C3 product distribution of sugar
hydrogenolysis and alkaline degradation of sugars (alkaline degradation — see Section
1.3.3). The authors observed similar selectivity to lactic acid, for alkaline degradation,
and PG/glycerol, for hydrogenolysis, which conﬁrmed that both reactions use similar

1.38 compared hydrogenolysis of sorbitol

pathways to product formation. Sohounloue eta
at different temperatures and base concentrations (180 — 240°C, pH = 8.5 — 12.5) in
aqueous media. The authors reported that elevated Ca(OH)2 loading doubled the reaction
rate, higher temperatures under basic conditions (pH = 12.5) increased C3 selectivity,
Ru/Si02 showed ten times the activity of Raney nickel, and an activation energy for

sorbitol disappearance of 13-16 kcal/mol. Montassier et al.39 reported 86% selectivity to

PG from glycerol using a Raney copper in neutral conditions and a 15/85 split to PG and

 

3“ Connor, R.; Adkins, H., J. Am. Chem. Soc., 1932,54, 4678-4690

35 Natta, G.; Rigamonti, R.', Beati, E., Berichte D. Chem. Gesellschaﬂ, 1943, 768, 641-656
3" Clark, 1., Ind. Eng. Chem, 1958, 50(8), 1125-1126

3" van Ling, G.; Vlugter, 1., J. oprp. Chem, 1969, 19, 4345

’8 Sohounloue, D. et al., React. Kin. Catal. Lett., 1983, 22 (3-4), 391-397

3° Montassier, C. et al., Bull. Soc. Chim, 1989, 126(2), 148-155

10

1.40 demonstrated the use of Ru/C as a

lactic acid at pH = 14, respectively. Muller eta
hydrogenolysis catalyst. Mannitol feedstock formed glycerol and propylene glycol with
56% yield while fructose yielded 48% to the two products.

Montassier et a1.41 explored the use of a sulfur-modiﬁed Ru/C to improve PG
selectivity with glycerol hydrogenolysis. Glycerol to PG showed 79% selectivity
compared to 91% observed with Raney copper. Xylitol and sorbitol are also mentioned
as hydrogenolysis feedstocks. The authors observe C2-C3 cleavage of xylitol with 86%
selectivity and 64% selectivity to PG and glycerol with sorbitol.

A number of patents also report conversion of sugar alcohols to C3 products.
Table 1-1 summarizes the product distribution from these runs as well as reaction
conditions. Larcher42 reports a yield of 44 wt% PG from a two step process of glucose
hydrogenation and subsequent hydrogenolysis with a Ni/CrOx catalyst at 275°C and 3000
psig H2. Rothrock43 continued this work with similar conditions and added CaCO3 to
increase the pH to 10 yielding 28% glycerol and 37% EG/PG. Conradin et a1.“45 ‘46 used
a Ni/kieselguhr catalyst to explore the effect of higher hydrogen pressures and elevated
pH. Hydrogenolysis of cane sugar (sucrose) yielded 32% glycerol, 15% EG, and 21%
PG. Kasehagen47 demonstrated 60% selectivity toward glycerol using Ni/Fe/Cu

48,49

supported on diatomaceous earth in a continuous reactor design. Sirkar and Chao et

 

‘0 Muller, P.; Rimmelin, P.; Hinderrnann, J .: Kieffer, R; Kiennemann, A.; Carré, J., Heterogeneous
Catalysis and Fine Chemicals 1], Guisnet M. et a1. (Editors), Elsevier Science Publishers B.V., Amsterdam,
1991

4‘ Montassier, C. et al., J. ofMol. Catal., 1991, 70(1). 99-110

‘2 Larchar, A., us. Patent 1,963,997, 1934

‘3 Rothrock, 11., US. Patent 2,004,135, 1935

4“ Conradin, F. et al, US. Patent 2,852,570, 1958

‘5 Conradin, F. et al, US. Patent 2,965,679, 1960

4" Conradin, F. et al, US. Patent 3,030,429. 1962

‘7 Kasehagen, L., US. Patent 3,396,199, 1968

‘8 Sirkar, US. Patent 4,338,472, 1982

ll

al.50 used a Ni/SiO2 catalyst and CaO promoter for glycerol selectivity of 70 wt%. The
catalyst however deactivated over time requiring replacement or additional reduction.
Arena51 observed 69% selectivity to glycerol and PG from sorbitol reacted with Ru/SiOx
impregnated with Ti and BaO added as a base promoter. Tanikella52 produced PG and
EG with 40% and 35% selectivity, respectively, using xylitol as a feedstock in a non-
aqueous solvent. Dubeck et al.53’54 modiﬁed a Ru/C catalyst with sulfur to increase PG
selectivity to 65%. These studies report the highest PG selectivity of the literature
reviewed. Andrews and Klaeren55 introduced the use of a soluble catalyst, H2Ru(PPh)4,
for hydrogenolysis of fructose and demonstrated selectivity to glycerol and EG at 38%
and 8% respectively. Gubitosa et 81.5 6‘5 7’5 8’5 9‘60 focus on using a trickle-bed reactor and
developing catalyst. The authors discuss higher selectivity to C3 compounds with higher
base/sorbitol ratio and demonstrate selectivity to PG — 53%, glycerol — 2%, and lactic
acid — 4% and total C3 selectivity of 59%. Chopade et a1.4 with use of a trickle-bed
reactor demonstrate C2-C3 bond cleavage of xylitol with 85% selectivity producing PG
and EG with 70% and 68% selectivity respectively. In addition, the authors
demonstrated similar C3 selectivity (58%) as Gubitosa et a1.54 using sorbitol as a

feedstock.

 

‘9 Sirkar, A., US. Patent 4,380,678, 1983

5° Chao, 1.; Hulbers, D., US. Patent 4,366,332, 1982

5‘ Aretct, 3., US. Patent 4,401,823, 1983

52 Tanikella, M, us. Patent 4,404,411, 1983

53 Dubeck, M.; Knapp, (3., US. Patent 4,430,253, 1984
5“ Dubeck, M.; Knapp, 6., us. Patent 4,476,331, 1984
55 Andrews, M.; Klaeren, S., US. Patent 5,026,927, 1991
56 Gubitosa, (3.; Casale, 0, US. Patent 5,326,912, 1994
57 Gubitosa, (3.; Casale, C., US. Patent 5,354,914, 1994
5* Gubitosa, (3.; Casale, C., 11.5. Patent 5,403,805, 1995
59 Gubitosa, G.; Casale, C., vs. Patent 5,543,379, 1996
5° Gubitosa, (3.; Casale, C.. US. Patent 5,600,028, 1997

12

Table 1-1: Hydrogenolysis Patents

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

U.S.
Authors T P . .
Patent . Catalyst Base . Signiﬁcance
Assn nee °C 51
(Year) ( g 1 1 (p )
First patent for
1.963,997 Larchar . hydrogenolysis of
(1934) (DuPont) NI'CIO" None 275 3000 sugars and sugar
alcohols
Improvement of
2,004,135 Rothrock . CaCO3 3300~ .
(1935) (DuPont) Ni-CrOx pH = 12 250 4400 :Aasrgher With use of
. High 11» showed
2,965,679 Conradin et a]. . . C30 1500- . . . .‘
(1960) (Inventa A.G.) NVK‘es‘i'guh’ pH = 11 220 11700 “nubl‘mg we“ °“
reaction
2,868.847 Boyers Ru as hydrogenation
(195 9) (Engelhard) Ru/C None 130 1000 catalyst
Kasehagen Ni/Fe/Cu on . 0
3’219966’39 (Allied diatomaceous 3:013 210 2000 :l‘eclgrglf 42 “M
Chemicals) earth 1" y
. Yield of 70%
4’313985532 (11:31:; ) Ni/Kieselguhr (31%??? 230 gig: glycerol, but catalyst
’ ' p deactivated Quickly
69% selectivity to
4,496,780 Arena . 8210
(1985) (UOP. Inc.) Ru / Tl-A1303 pH N l 3 180 3200 glycerol and PG. Ru
as catalyst.
Use of non-aq solvent
. . and heavy base
4:399:31 afﬁx?) Ni / Si-Alox an“ Monk 275 4000 concentrations (20
mol%). Focused on
xylitol feedstock.
o . .
4.430.253 Dubeck Ru/C with Cao 250 2400- fﬁ’gﬁzlsicst‘e‘gyif PG
(1984) (Ethyl Corp.) Na2S-9H20 pH ~ 14 3300 literature.
5 026 927 Andrews et a1. Homogenous
’( 1 99’ 1) (Brookhaven H2Ru(PPh3)4 KOH 100 300 catalyst. Much
Nat. Labs) milder conditions.
Claims 60% yield of
5,210,335 Schuster et 31. PG. Closer to 42%
(1993) (B ASF) Co/Cu/MnO, None 250 4000 with loss of gases
taken into account.
Gubitosa et a1. Continuous ﬁxed bed
5’310939’2? (Montecatini Ru/C 21,3911, 225 2200 reactor. C3
Tech.) p selectivity at 59%
85% C2-C3 cleavage
selectivity with
6'32901335 {magi}? a" Ru/C 39*}, 230 1200 xylitol. 58% C3
9 selectivity with
sorbitol.

 

 

 

1 .3.3 Alkaline degradation

Lobry de Bruyn and Alberda van Ekensteinm’62 ﬁrst noted the isomerization of
sugars in their classic papers from the 18908. They stated that mannose, glucose, and
fructose interconvert while in basic media. NefS3 furthered this notion by also noting the
presence of enediol intermediates and numerous products formed from base-catalyzed
cleaving. Evans et a1.64 added to the list of sugars involved with isomerization and
observe a signiﬁcant presence of pyruvaldehyde as an intermediate, which converts easily
to lactic acid in alkaline media as noted by Denis“. Shaffer and F riedemann66 continued
mechanistic work by mapping out Nef s suggestions as well as focusing on lactic acid
production. They noted that higher concentrations of base promote lactic acid formation
while lower concentrations favor dihydroxyacetone. Two patents also focused on lactic
acid production directly from sugars. Braun67 noted 72-75% yield of lactic acid
recovered as a zinc salt from cane sugar (sucrose) while Locké8 observed 67% yield.
Yield to lactic acid is quite good for these methods, but the processes are not amenable to
scale-up. Lock suggests dropwise addition of sugar to boiling caustic while Braun uses
temperatures in the range of 210-220°C both of which are unfavorable. Table 1-2

summarizes the conditions and yields for the technologies explored.

 

6' de Bmyn, C. A L., Rec]. Trav. Chem. Pays-Bas,1895,14, 156-165

62 de Bruyn C. A. L.; van Ekenstein, w. A., Recl. Trav. Chem. Pays-Bas, 1895, 14. 203-216
53 Nef, 1., Ann, 1907, 357, 294-312

6“ Evans, w. L.; Edgar, R. 11.; Hoff, G. P., J. Am. Chem. Soc, 1926, 48, 2665-2677

65 Denis, w. Am. Chem. J., 1907, 38, 561-594

‘56 Shaﬂer, P. A; Friedemann, T. 13.. J. ofBiol. Chem, 1930.86. 345-375

67 Braun, 6., US. Patent 2,024,565, 1935

63 Lock, R 11., us. Patent 2,382,889, 1945

14

 

 

 

 

 

 

 

 

 

 

Table 1-2: Summary of prior art for lactic acid production via alkaline degradation
8:12: 32:: 1281 all “if“
0.55 M 1 5.0 M 2 37 + 65* 100 18+7 days 50 66
2.0 M 1 6.0 M 2 37 + 65* 100 18+7 days 40 66
1.2 M 2 2.65 M 1’ 210-220 100 ~2 hours 72-75 67
0.75 M 3 8.0 M a 100 ~100 4 hours 67 68
1.8 M 4 0.45 M 2 225 61 2 hours 17 69**
1.8 M 4 0.45 M 2 225 8 2 hours 3 69**
2.29 M 2 2.6 M ° 251-260 100 2.5 hours 47 70
2.29 M 2 2.6 M ° 234-238 100 7.5 hours 47 70
2.29 M 2 2.8 M ° 234-239 100 7.5 hours 44 70

 

 

 

 

 

 

 

 

 

‘ glucose - 2 cane sugar - 3 sucrose - 4 sorbitol

: NaOH - 1’ Ca(OH)2 - ° CaO

* Reaction time was 18 days at 37°C and then 7 days at 65°C
** Reactions used 5% Ru/C to cleave sorbitol under nitrogen

In general, larger base to sugar ratios improved lactic acid yield. Lower
temperatures (3 7°C) slowed the alkaline degradation pathways signiﬁcantly without the
beneﬁt of improved yield to lactic acid. Temperatures of 230-260°C converted the sugars
much more readily, but only at the same yields of 40-50%. High base to sugar ratios
combined with relatively moderate temperatures (100 — 210°C) led to the highest yields
of around 70%.

Work continued with improved mechanisms and calculated rate constants.
MacLaurin and Green71 presented a simple model for isomerization and degradation of
glucose, ﬁuctose, and mannose to acidic products. They used ﬁrst order equations with
respect to sugar concentration to calculate rate constants. Their results showed that

fructose degraded to products an order of magnitude faster than glucose and mannose. de

 

69 Tronconi, E. et al., Chem. Eng. sci, 1992, 47 (9-11), 2451-2456
2" Montgomery R.; Ranca, R A.,1nd. Eng. Chem, 1953, 45(5), 1136-1147
2' MacLaurin, D. 1.; Green, 1. w., Can. J. Chem, 1969,47, 3947-3955

15

Wit et a1.72 estimated the kinetics of enediol formation using a UV cuvette and noted
intermediate evolution versus time. They also suggested the following mechanism for
alkaline degradation:
sugar + OH' H sugar anion H enediol anion H p-elimination H degradation products

For added completeness, de Bruijn et a1. introduced a series of papers”‘74'75‘76‘77’78
looking at multiple aspects of alkaline degradation. One paper inspected the effect of
reaction parameters on the product distribution“. They outline the following results:

- C3 selectivity improved and >C6 acid formation reduced with higher

hydroxide concentrations and lower sugar concentrations

I Ca2+ promoted lactic acid pathways

I Temperature had no effect on product distribution from 50-90°C
Another paper73 added psicose as an important sugar intermediate within the degradation
mechanism and included kinetic rates of isomerization and cleavage reactions. An
additional paper75 summarized several chemistries occurring in the alkaline environment
and describes their contribution to the overall product distribution. The same from Delft

University later published a comprehensive review of alkaline degradation79 summarizing

their work as well as providing ample background on the technology.

1 .4 Summary
Yields to C3 products from biomass-derived feedstock are currently at about 60%.
The literature shows that hydrogenolysis of sorbitol has the best prospects of becoming

commercially viable, based on C3 yields and possibilities for continuous processes. High

 

22 de Wit, G.; Kieboom, A. P. G.; van Bekkum, 11., Carbohyd. Res, 1979, 74, 157-175

22 de Bruijn, 1. M.; Kieboom, A. P. G.; van Bekkum, 11., Recl. Trav. Chim. Pays-Bas, 1986, 105, 176-183
74 de Bruijn, J. M., et al., Int. Sugar Journal, 1987, 89 (1066), 194-210

25 de Bruijn, 1. M; Kieboom, A. P. G.; van Bekkum, 11., Recl. Trav. Chim. Pays-Bas, 1987, 106. 3543
2“ dc Bruijn, 1. M.; Kieboom, A. P. G.; van Bekkum, J. Carbohyd. Chem, 1986, 5 (4), 561-569

22 de Bruijn, 1. M.; Kieboom, A. P. G.; van Bekkurn, Starch, 1987, 39 (1). 23-28

78 de Bruijn, J. M.; Touwslager, F .; Kieboom, A. P. G.; van Bekkurn, 11., Starch, 1987, 39 (2), 49-52

29 de Bruijn, 1. M.; Kieboom, A. P. G.; van Bekkum, Sug. Tech. Rev., 1986, 13, 21-52

16

process temperatures, however, contribute to lower C3 selectivity and yield due to the
formation of gases. In addition, high process pressures (1000 psi and higher), while not
prohibitive, are less than desirable for scale-up.

For this technology to be successful, the overall yield to propylene glycol,
glycerol, and lactic acid needs to be around 75% in order to be competitive with current
petroleum-based production methods. Hydrogenolysis and alkaline degradation alone are
not sufﬁcient to reach this goal. Understanding the reaction pathways that promote C3
product formation for each chemistry, however, provides the potential for elevating C3

yields to the desired levels.

17

Chapter 2 - Experimental Methods

2.1 Hydrogenolysis
2.1.1 Reactor set-up

Initial experiments with hydrogenolysis used a continuous trickle-bed reactor.
The reactor consisted of a 2.5 cm inner diameter stainless steel tube packed with catalyst
supported with a stainless steel screen. The column was heated using a PID controller
(Omega) connected to heating tape wrapped around the reactor. A thermocouple placed
in an axial thermowell provided temperature data to the controller. Liquid was fed to the
column using an HPLC pump (Bio-Rad) and gas via mass ﬂow controllers (Porter). To
avoid channeling of the liquid feed, a 10 cm long distributor consisting of 2 mm glass
beads was added at the top of the column. A back-pressure regulator (Tescom)
maintained reaction pressure at the desired set point while mass flow controllers (MF C s)
regulated the ﬂow of hydrogen to the system. Rotameters and a buret with soap ﬁlm
measured the volumetric ﬂow rate and allowed calibration of the MFCs. Liquid samples
were collected via gas-liquid separation tanks to maintain continuous ﬂow. Post-reactor
vents also allowed for sampling of the gas for hydrocarbon production. Figure 2-1

provides a reactor schematic.

2.1.2 Materials for reaction

Sorbitol (99+% Aldrich), potassium hydroxide (85% Aldrich), and ultra-high
purity hydrogen (99.999% AGA, BOC) were the main reactants for hydrogenolysis.
Feedstock was prepared using HPLC grade water (J .T. Baker). Ultra-high purity helium

(99.999% AGA, BOC) was used to purge the reactor and compressed air (21% O2,

18

balance N2 - AGA) was used to passivate the catalyst at the end of the reaction. All

components were used without ﬁirther puriﬁcation.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mass flow controller Safety relief valve
H
2 +16 @1 7
*— Liquid distributor
Inert (He) T
i Reactor
Feed {”1 !
Vent * ,4, a1 IJ'i
Rotameters To ma 55
Back pressure
regulator ® spectrometer
‘ 1‘ ﬁrﬁ 4- V
Er M J Gas-liquid
separator/
Mass sampling tanks
spectrometer

 

 

 

Figure 2-1: Schematic for continuous trickle-bed reactor

2.1.3 Operating procedures

Once the column was packed with the proper catalyst and sealed, the system was
pressurized with helium to 1000 psig to detect leaks. When the pressure drop was less
than 10 psig over the course of an hour, the reactor was vented and then pressurized again
to 300 psig with hydrogen to begin pro-reduction of the catalyst. The column was heated
to 150°C and maintained at that temperature for at least two hours. Once complete, the
temperature was raised to the desired set point (typically ZOO-230°C), the column was

pressurized with hydrogen (1000-1200 psig), and feedstock (1.5 M sorbitol and 0.18 M

19

KOH) was then added to the reactor at a rate of 0.3-0.4 mL/min to begin the reaction.
Samples were taken periodically to observe the extent of reaction. Once the product
distribution remained essentially the same from sample to sample (i.e. steady-state),
temperature and/or pressure was adjusted for another set of conditions or the reactor was
stopped. Upon completion of the experiment, the catalyst bed was ﬂushed with water

and passivated with oxygen.

2.1.4 Analysis

Liquid phase products were analyzed using high performance liquid
chromatography (HPLC). Samples were passed through a 0.22 micron ﬁlter (Millipore)
and then diluted to a 1 wt% basis (i.e. 0.1g of 25 wt% sorbitol solution diluted to 2.5
grams of solution) using 5 mM (0.005 M) sulfuric acid. An internal standard, 2-propanol,
was added and the sample was injected into an Aminex I-IPX-87H column (Bio-Rad) at a
temperature of 50°C and a mobile phase of 5 mM sulfuric acid ﬂowing at 0.45 mL/min.
Liquid phase products were detected using a refractive index detector (Refractomonitor
1V, Thermo-Finnigan) and an ultra-violet detector (UV 3000, Thermo-Finnigan). The
ultraviolet wavelength was set at 210 nm to detect unsaturated compounds, namely
carboxylic acids. Data were compiled and analyzed using TSP-1000 soﬁware (Thermo-

Finnigan).

2.2 Adsorption studies
2.2.1 Reactor set-up
Adsorption studies were carried out in a 300 mL batch reactor (Parr Instruments)

with internal stirrer and a controller (Wadlow) that maintains temperature within 1°C and

20

agitation within 5 rpm. The reaction vessel is constructed of 3 16 stainless steel with an
upper temperature limit of 350°C and a maximum allowable working pressure (MAWP)
of 3000 psi. To remove samples periodically without disrupting the reaction, a sampling
loop was attached. Reactions were agitated with a four-blade marine type impeller and

baffles were added to enhance mixing.

2.2.2 Materials for reaction

Ruthenium sponge (~100 mesh, 99.9%, Aldrich) was the metal catalyst used for
adsorption studies. Sorbitol (99+%, Aldrich) was the substrate of interest and potassium
hydroxide (85+%, Aldrich) was used as well to study effects of pH. Solutions were
prepared with HPLC grade water (J .T. Baker). The ruthenium was pre-reduced with
UHP hydrogen (99.999%, AGA/BOC) and the reactor was pressurized with UHP helium
(99.999%, AGA/BOC) during the adsorption experiment to provide an inert atmosphere.

All compounds were used without ﬁirther puriﬁcation.

2.2.3 Operating procedures
2.2.3.1 Pre-reduction of catalyst

The ruthenium sponge (typically 10.0 grams) was pre-reduced before each
reaction in order to activate the metal surface. Sponge was placed in the reactor,
subsequently sealed, and then purged six times with hydrogen gas to remove air from the
reaction. Once purged, the reactor was heated to 150°C and pressurized to 200 psig for at
least one hour. After this time, the reactor was cooled and then purged with helium to

prepare for the reaction.

21

2.2.3.2 Charging of the feedstock

Feedstock was prepared using sorbitol and HPLC grade water to form 100 grams
of solution. Once prepared, helium was bubbled through the liquid to remove dissolved
gases. The purged solution was then added to a clean charging vessel and purged again
to remove air present in the vessel. The vessel was then connected to the reactor,
pressurized, and opened to the reactor to transfer the feedstock solution. Using the
charging vessel prevented the addition of oxygen to the reaction, which could deactivate

the active catalytic sites on the metal.

2.2.3.3 Sampling

Liquid samples were removed during the reaction via a sample loop. A valve
connected to a dip tube within the reactor was opened to place the sample loop in line. A
three-way valve was opened to send the liquid sample into the holding area of the loop
and closed. A third valve to the atmosphere was opened for sample collection. This
process was repeated until 2-5 grams of material was collected. Afterward, the sample
loop was blown out with air to remove excess liquid and catalyst by opening the three-

way valve to a supply of compressed air.

2.2.4 Analysis
Liquid phase products were analyzed in the same matter as those described in
Section 2.1.4. However, samples were not diluted due to low concentrations of organic

products.

22

2.3 Alkaline degradation
2.3.1 Reactor set-up

Two types of alkaline degradation reactions were run: metal-catalyzed and base-
catalyzed. The metal-catalyzed runs used ruthenium sponge to initiate carbon-carbon
cleavage of saturated polyols, such as sorbitol and xylitol, to products. Base-catalyzed
experiments used soluble base, mainly potassium hydroxide, to cleave carbon-carbon
bonds in monosaccharides. Both were run in the batch reactor system as described in

section 2.2.1.

2.3.2 Materials for reaction
2.3.2.1 Metal-catalyzed degradations

Metal-catalyzed degradation runs used ruthenium sponge (-100 mesh, 99.9%,
Aldrich) for reactions. Sorbitol (99+%, Aldrich), xylitol (98%, Aldrich), and glycerol
(99+% Aldrich) were used in these reactions. Potassium hydroxide (85+%, Aldrich) was
the sole soluble base used in these studies. Solutions were prepared with HPLC grade
water (J .T. Baker). Ultra high purity hydrogen (99.999%, AGA/BOC) was used to
reduce the ruthenium metal and UHP helium (99.999%, AGA/BOC) served as the

reaction gas phase. All compounds were used without further puriﬁcation.

2.3.2.2 Base-catalyzed degradations
Glucose (99+%, Gibco BRL), fructose (99+%, Aldrich), mannose (99%, Aldrich),
and xylose (99%, Aldrich) were the sugar substrates used for reactions with potassium

hydroxide (85+%, Aldrich) as the soluble base. Solutions were prepared with HPLC

23

grade water (J .T. Baker); UHP helium (99.999%, AGA/BOC) served as the reaction gas

phase. All compounds were used without further puriﬁcation.

2.3.3 Operating procedures
2.3.3.1 Pre-reduction of catalyst
The ruthenium sponge used in these reactions was reduced in the same manner as

described in Section 2.2.3.1.

2.3.3.2 Charging of the feedstock

Feedstock was prepared by weighing out reactant sugar and potassium hydroxide
separately. HPLC grade water was then added to dissolve the sugar and potassium
hydroxide separately. The two solutions were then mixed together and water was added
to make 100 grams of solution. Once prepared, helium was bubbled through the liquid to

remove dissolved gases and charged in the same manner as described in Section 2.2.3.2.

2.3.3.3 Sampling

Liquid samples were removed in the same manner as described in Section 2.2.3.3.

2.3.4 Analysis

Samples removed from the reactor were diluted to a 1 wt% basis of the reactant
sugar with I-H’LC grade water or dilute sulfuric acid to lower the pH of the solution and
reduce continued degradation reactions in the sample. Samples were analyzed within 45
minutes of removal from the reactor for true product distributions. Feedstocks with
initial sugar concentrations of less than 1 wt% were not diluted. Product distribution was

then determined using HPLC as previously described.

24

Product formation was also conﬁrmed by using gas chromatography-mass
spectroscopy (GC-MS). Samples were sent over to the MSU Mass Spectroscopy Facility
for analysis. The polyols were dried and trimethylsilated using
Bis(trimethylsilyl)triﬂuoroacetamide (BSTFA) and 1% trimethylchlorosilane (TMCS) as
silylating agents with pyridine to remove water. Each product has a distinctive spectrum

allowing conﬁrmation of known products and identiﬁcation of unknown compounds.

2.4 Alkaline degradation with hydrogenation
2.4.1 Reactor set-up

This set of reactions combined the chemistries of alkaline degradation and
hydrogenation using a slightly modiﬁed version of the batch reactor system described in
Section 2.2.1. The sample loop used for these experiments was shortened and had a
larger tube diameter as well (1/4” from 1/8”). This allowed for shorter sampling times to

minimize the time before analysis of the sample.

2.4.2 Materials for reaction

Several catalysts were used to promote hydrogenation during the reactions. These
consisted of 5 wt% Pt/C (Aldrich), 5 wt% Pd/C (Aldrich), and 5 wt% Ru/C (Aldrich).
Glucose (99+%, Gibco BRL), fructose (99+%, Aldrich), mannose (99%, Aldrich), and
xylose (99%, Aldrich) were the sugar substrates used for reactions with potassium
hydroxide (85+%, Aldrich) as the sole soluble base. Solutions were prepared with HPLC
grade water (J .T. Baker). Ultra high purity hydrogen (99.999%, AGA/BOC) was used to
reduce the catalysts and as a reactant gas. All compounds were used without further

puriﬁcation.

25

2.4.3 Operating procedures
2.4.3.1 Pre-reduction of catalyst

Of the catalysts used in these experiments, only the 5 wt% Ru/C required pre-
reduction. Platinum and palladium return to the zero valence level readily when in the
presence of hydrogen, while ruthenium requires high temperature and the presence of
hydrogen for reduction. Ruthenium on carbon was placed in the reactor and then sealed.
The system was purged six times with hydrogen, pressurized to 200 psig, and then heated
to 230°C overnight (12-16 hours). The reactor was then vented and prepared for addition

of the reactants.

2.4.3.2 Charging of the feedstock

To address reactivity of the sugar when base is present, the charging procedure
was modiﬁed. The feedstock was charged in two parts to avoid conversion of the sugar
before being added to the reactor. Potassium hydroxide solution was prepared and placed
in the charging vessel. Hydrogen was used to pressurize the charging vessel from the
bottom to allow hydrogen to bubble through the solution removing other dissolved gases.
The vessel was then vented and the process was repeated twice more. Potassium
hydroxide was charged to the reactor and heated to 5-7°C above the desired reaction
temperature. The sugar feedstock was then fed to the charging vessel, purged in the same
manner as potassium hydroxide, and charged to the reactor. The reactor was pressurized

with hydrogen and the reaction medium was agitated to commence the run.

26

2.4.3.3 Sampling

Before each sample, the sample loop was washed with one to three grams of the
reactor contents to remove any residual material from previous samples and ensure that
fresh material from the reaction was analyzed. Aﬁer washing, samples were taken in the

manner described in Section 2.2.3.3.

2.4.4 Analysis
Samples removed ﬁom the reactor were ﬁltered and diluted to a 1 wt% basis of
the reactant sugar with 25 mM sulfuric acid to lower the pH of the sample and reduce

sample degradation. This was done within ten minutes of drawing the sample. 5000 11L

of the diluted material was measured out via pipetter and then 10 uL of 2-propanol added
as an internal standard. The prepared sample was then injected to the HPLC column
using an auto-injector (LDC/Milton Roy) within 45 minutes of when the sample was ﬁrst
drawn. Dilution and analysis of the samples was done quickly to avoid ﬁiither
degradation of the sample from base-catalyzed reactions. Refractive index and ultra-
violet data were stored by computer for analysis using Star Chromatography software

(Varian).

2.5 Selective hydrogenation of intermediates
2.5.1 Reactor set-up

These hydrogenations used the batch reactor system as described in Section 2.4.1.

27

2.5.2 Materials for reaction
2.5.2.1 C3 versus C3 selective hydrogenation

Two catalysts, 5 wt% Pt(Ge)/C (Engelhard Italia) and 3 wt% Pt(S)/C (Engelhard
Italia), were used for these experiments. Glyceraldehyde dimer (99%, Aldrich) and
glucose (99+%, Gibco BRL) were utilized for this study. Solutions were prepared with
HPLC grade water (J .T. Baker). Ultra high purity hydrogen (99.999%, AGA/BOC) was

used for reactions. All compounds were used without further puriﬁcation.

2.5.2.2 Product inhibition of glyceraldehyde hydrogenation

The two platinum catalysts were used again for these experiments.
Glyceraldehyde dimer (99%, Aldrich) was used as the primary reactant. Saturated
products studied for inhibition were ethylene glycol (99%, Spectrum), pr0pylene glycol
(99.5%, Jade Scientiﬁc), glycerol (99+%, Aldrich), sorbitol (99%, Aldrich), and mannitol
(99%, Aldrich). Solutions were prepared with HPLC grade water (J .T. Baker). Ultra
high purity hydrogen (99.999%, AGA/BOC) was used for reactions. All compounds

were used without further puriﬁcation.

2.5.2.3 Alkaline degradation with hydrogenation
Materials used were the same as those of the previous alkaline
degradation/hydrogenation studies (Section 2.4.2) with the exception of using the Pt(S)/C

and Pt(Ge)/C catalysts.

28

2.5.3 Operating procedures
2.5.3.1 Pre-reduction of catalyst

Prior to reaction, the platinum catalysts were activated by purging the reactor six
times with hydrogen. In addition, the catalyst was stirred with water overnight to rewet

surfaces that may have dried out during storage to maintain proper activity.

2.5.3.2 Charging of the feedstock

Feedstock was charged in two parts to minimize time for the reactants to reach the
desired temperature. The ﬁrst charge consisted of solution prepared with non-reacting
compounds. Table 2-1 below shows the three types of experiments and the order in

which materials were charged.

Table 2-1: Order of charging the reactants for experiments

 

 

 

 

Experiment 1St charge (100 grams) 2'“I charge (100 grams)
C3 selective hydrogenation Water Glyceraldehyde plus sugar
Product inhibition Inhibitor (Sorbitol, PG, etc.) Glyceraldehyde
Alkaline degradation/ KOH Reac tan t sugar
hydrogenation

 

 

 

 

 

2.5.3.3 Sampling

Samples were drawn in the same manner as that described in Section 2.4.3.3.

2.5.4 Analysis
2.5.4.1 C3 versus Ce selective hydrogenation
Samples were passed through a 0.22 micron ﬁlter (Millipore) to remove catalyst

from the solution. Samples were not diluted due to relatively low concentrations of

29

glyceraldehyde or other compounds in the reaction (0.2 - 0.8 wt%). HPLC analysis was

carried out as previously described in Section 2.4.4.

2.5.4.2 Product inhibition of glyceraldehyde hydrogenation

Analysis was performed as described in Section 2.5.4.1.

2.5.4.3 Alkaline degradation with hydrogenation

Analysis was performed as described in Section 2.4.4.

2.6 Conversion of carbohydrates to methane via hydrogenolysis
2.6.1 Reactor set-up
The batch reactor system described in Section 2.4.1 was used for these

experiments.

2.6.2 Materials for reaction

Catalysts used for this set of reactions were 5 wt% Ru/C (PMC) and 3% Ru/TiO2
(Degussa). Glucose (99+%, Gibco BRL), sorbitol (99+%, Aldrich), and glycerol (99%,
Aldrich) were used for feedstock. Solutions were prepared with HPLC grade water (J.T.
Baker). Ultra high purity hydrogen (99.999%, AGA/BOC) was used to reduce the

catalysts and as a reactant gas. All compounds were used without further puriﬁcation.

2.6.3 Operating procedures
2.6.3.1 Pre-reduction of catalyst
Before each experiment, catalyst was loaded into the reactor and sealed. The

headspace was then ﬂushed with hydrogen six times to remove air. Both types of

30

catalysts were pre-reduced overnight at 23 0°C under 300 psig of hydrogen and then

cooled to room temperature. Feedstock was charged to the reactor after this point.

2.6.3.2 Charging of the feedstock

Feedstock was loaded into the charging vessel, purged three times with hydrogen,
and then sent to the reactor. The headspace was vented to minimize the amount of
hydrogen in the reactor and prevent premature conversion. The reactor was then heated
to the desired temperature with mild agitation (~3 00 rpm). Once at temperature, the

reactor was pressurized with hydrogen and agitated at 1000 rpm to begin the reaction.

2.6.3.3 Sampling

Gas samples were removed from the headspace and collected with a 7x7 TedlarTM
gas bag (DuPont) before the liquid sample was taken. The gas bag was ﬁlled once with
reactor gas, emptied, and ﬁlled again to remove residual gas from the previous sample.

Liquid samples were drawn in the same manner as those described in Section 2.4.3.3.

2.6.4 Analysis

Gas samples were analyzed using a Varian 3300 gas chromatograph using a
Carbosieve SH column (Supleco). Temperatures for the injector and the thermal
conductivity detector (TCD) were 250°C and 225°C respectively. The temperature
program for the column was 35°C for two minutes followed by a 30°C / min ramp to
225°C. This temperature was held for 15 minutes and then the column was cooled for the
next sample. Ultra high purity helium was the carrier gas and data were collected using

Star Chromatography software (Varian).

31

Liquid samples were ﬁltered to remove catalyst from the solution. The sample
was then diluted with HPLC water to a 1 wt% basis of the reactant. HPLC analysis, as

previously described in Section 2.4.4, was used for product identiﬁcation.

2.7 Catalyst characterization

Catalysts used in the reactions of interest were characterized using a number of
standard methods in order to compare them to other available catalysts and allow for
proper replication of experiments. Mercury porosimetry80 was used to determine pore
volume of each catalyst as well as the pore size distribution throughout the particles.
Surface area, as determined by the Brunauer-Emmett-Teller (BET) method“, was also
used as a comparative result. Chemisorption of carbon monoxide and hydrogen was used
to calculate metal dispersion on the catalysts and served as the ﬁnal method of

characterization used in these studies.

2.7.1 Physisorption

Physisorption measurements, namely the BET method, provide a standard
comparison of catalyst surface area between materials. It is also important when
manufacturing catalysts to produce consistent supports and dispersion of metal to ensure

nearly identical catalytic activity.

2.7.1.1 Equipment
BET surface area measurements were conducted using the Accelerated Surface
Area and Porosimetry System or ASAP 2010 (Micromeritics). Software from

Micromeritics was used to operate the equipment and analyze the catalyst.

 

#0 Gregg, S.J.; Sing, K.S.W., Adsorption, Surface Area and Porosity, 2nd Edition, 1982, 173-190, New York
3‘ Brtuiauer, S.; Emmett, P.H.; Teller, E., J. Am. Chem. Soc, 1938, 60(2), 309-319

32

2.7.1.2 Materials

Liquid nitrogen was used to lower the analysis temperature to the required level
of 77 K. Ultra-high purity nitrogen (99.999%, AGA/BOC) was used for the nitrogen
physisorption experiments. The catalysts from the previous sections were used for

analysis.

2.7.1.3 Operating procedure

Samples were ﬁrst prepared by degassing the material in a pre-weighed glass
sample bulb to remove water and other potential contaminants. The sample was then
reweighed to determine the actual amount of catalyst present in the sample bulb. Next,
the sample was placed in the analysis port and evacuated. The sample bulb was then
placed in a Dewar ﬂask of liquid nitrogen and further evacuated until equilibrium was
reached. Nitrogen was then dosed in incrementally and the amount of gas physisorbed
was recorded. The data were analyzed to determine the surface area of catalyst as

described by the BET method.79

2.7.2 Chemisorption

Chemisorption serves as another method for characterizing catalysts. Adsorbing
gases, such as hydrogen and carbon monoxide, on to activated metal sites provides
information about the fraction of active metal that is available for reaction. This
measurement, known as dispersion, is another important comparison of catalysts in that
the more metal that is available for reaction, the higher the activity. Keeping this
parameter consistent is important in catalyst manufacturing and understanding evolution

of catalyst behavior from run to run.

33

2.7.2.1 Equipment
Chemisorption measurements were conducted using the ASAP 2010 and

corresponding software (Micromeritics).

2.7.2.2 Materials

Gases used for chemisorption analysis were UHP hydrogen (99.999%
AGA/BOC) and carbon monoxide (99.5% BOC). UHP helium (99.999%, AGA/BOC)
was used to purge the samples before analysis. The catalysts from the previous sections

were used for analysis.

2.7.2.3 Operating procedure

Samples were ﬁrst prepared by degassing the material in a pre—weighed glass
sample tube to remove water and other potential contaminants. The sample was then
reweighed to determine the actual amount of catalyst present. Once complete, the sample
tube was placed in the analysis port of the ASAP 2010 system. The sample tube was then
placed under vacuum and heated to remove adsorbed material on the catalyst. Next, the
catalyst was reduced under hydrogen to activate the metal sites. Then, the sample tube
was cooled and evacuated again. At this point, analysis commenced with dosing of the
analysis gas and subsequent measurement of the adsorbed gas. Dispersion was calculated

using the following equation:

 

 

 

 

 

( N
100% x 100% VXSF
o ' .___ calc
ADISP m3 Weight % 1
c ( )
22414
mole \ AWmetal /

34

Where:
% Disp = Metal dispersion — fraction of metal available for adsorbate interaction
V = y-int. of linear best ﬁt for adsorbed volume per gram vs. P [cm3 gas/g metal]
$leC = Atoms of metal per molecule of adsorbed gas [atom of metal/mole of gas]
Weight % = Weight fraction of catalyst that is metal [dimensionless]

AWmmr = Atomic weight of the supported metal [g metal / atom of metal]

Metal surface area was calculated using equation 2-2:

 

 

6.023 x 1023 a—to-m—S
MSA : (”“13“Ole V X SFcalc X Aarea (2)
22414
mole

Where:
MSA = Metal surface area [m2 / g sample]
V = y-int. of linear best ﬁt for adsorbed volume per gram vs. P [cm3 gas/g metal]
8lec = Atoms of metal per molecule of adsorbed gas [atom of metal/mole of gas]

Ama = Effective area of 1 metal atom [m2 / atom of metal]

Both calculations were completed by the Micromeritics soﬁware and described in the

user manual.82

 

82

ASAP 2010 Chemi: Accelerated Surface Area and Porosimetry System Operator’s Manual,
Micromeritics, September 1996, Appendix C

35

Chapter 3 — Preliminary sorbitol hydrogenolysis studies

The focus of these initial experiments was to become familiar with the
hydrogenolysis reaction and develop a mechanistic picture by observing adsorption of
sorbitol on ruthenium metal. Previous hydrogenolysis studies at MSU utilized a
continuous trickle-bed reactor to conduct the majority of experiments. Subsequently, this
apparatus was used to observe the reaction and to mimic similar conditions of this work.
In addition, batch reactions of sorbitol hydrogenolysis were used to observe the
progression of the reaction over time. Adsorption studies were an extension of this work
to determine how sorbitol interacts with the catalytic surface. Of interest was the

geometry of adsorption and the amount of adsorption occurring on the catalyst.

3.1 Hydrogenolysis experiments
3.1.1 Overview
Previous work by Chopade et al.4, involved hydrogenolysis of sorbitol and xylitol
to desired C2 and C3 products: ethylene glycol (EG), propylene glycol (PG), glycerol
(GO), and lactic acid (LA). initial experiments involved getting familiar with the
technology of sugar alcohol conversion. Ideally, one mole of sorbitol fed to the reactor
would yield two moles of a C3 product and this is the basis of comparison for each
experiment. Results are reported in terms of selectivity, with units of mole of product per
mole of reactant converted, and yield, with units of mole of product per mole of reactant
fed. For both selectivity and yield, the maximum value is 2.0 with respect to C3 products.
Experiments were conducted in both a batch reactor and continuous trickle-bed
reactor. Initial sorbitol loadings were 25 wt% (1.5 M) with 1 wt% (0.17 M) potassium

hydroxide added as a base promoter. Temperatures varied from ZOO-210°C with

36

hydrogen pressures at 1200 psig for trickle bed reactions and 1350 psig for the batch
reactor. Reactions were catalyzed with 1.0 grams of 2.5 wt% ruthenium on carbon
(PNNL).

Typical product distributions showed selectivity to PG, GO, LA, and EG at 0.57,
0.40, 0.07, and 0.29, respectively, with total C3 selectivity at 1.04 (52% of quantitative).
Conversion varied from 70-95% depending on factors such as temperature and amount of
catalyst. These results demonstrate the technology and serve as the base line for the

remaining studies.

3.1.2 Mesh size study

Another set of reactions was used to observe the effect of catalyst particles sizes
in trickle-bed reactions. Since each individual particle is wetted in the trickle ﬂow
regime, changing the size of the solid particle could have an effect on the mass transfer of
reactants to the catalytic surface and, thus, affect overall activity. Al-Dahhan and
Dudukovic'83 noted the use of ﬁnes in trickle bed reactors to increase the wetting of
catalyst and maintain consistent packing for reactions as well.

Two sizes of catalyst were used for these runs: coarse particles of 16-30 mesh
(93 7-500 microns) and intermediate particles of 50-100 mesh (300-150 microns).
Catalyst was crushed using a mortar and pestle and then sieved using A.S.T.M. standard
sieves (W.S. Tyler Inc). For reactions, the trickle-bed reactor was loaded with 25 grams
of pro-reduced ruthenium on carbon catalyst. The catalyst was reduced under 300 psig of
hydrogen at 150°C for two hours and then wetted with feedstock solution. Feedstock

consisted of 25 wt% sorbitol (1.5 M) and 1.0 wt% potassium hydroxide (0.17 M).

 

83 Al-Dahhan, M.H.; Dudukovié, M.P., A1ChE Journal, 1996,42 (9), 2594-2604

37

Reactions were carried out at 200°C and 1200 psig H2 with a liquid ﬂow rate of 0.31
mL/min and hydrogen ﬂow at 50 cm3/min (STP) to maintain a 5:1 ratio of hydrogen to
sorbitol molar ﬂow. The reaction continued until HPLC analysis showed identical or
near-identical product distributions of consecutive samples, indicating that steady-state
had been achieved.

Product distributions between the three runs were nearly identical after each
reaction reached steady-state (i.e. consistent product distribution, conversion, etc.
observed in consecutive samples). Table 3-1 shows selectivity for ethylene glycol,
propylene glycol, glycerol, lactic acid, and total C3 products. Conversion and catalyst

loading are also provided for comparison.

Table 3-1: Product distribution of trickle-bed experiments

 

 

 

 

 

a... P133731" 52231:; :3. 2?. s53. sit. c...
27 16-30 26.4g 0.57 0.40 0.07 1.04 93%
14 16-30 25.3g 0.51 0.50 0.04 1.04 89%
19 50-100 25.0g 0.52 0.50 0.05 1.07 75%

 

 

 

 

 

 

 

 

 

Selectivity to C3 products was nearly identical between the three experiments.
Propylene glycol shows a slightly higher selectivity in the case of 16-30 mesh catalyst.
This, however, is an effect of glycerol formed from sorbitol ﬁlrther converting to
propylene glycol. A repetition of the 16-30 mesh experiment showed similar C3 product
distribution as the 50-100 mesh case, conﬁrming the similarities.

One difference between the mesh sizes was the conversion of sorbitol. With 50-
100 mesh catalyst sorbitol conversion was signiﬁcantly lower than that of the 16-30 mesh

catalyst. This reduction in conversion may be due to mass transfer effects from the

38

change in catalyst size. To validate this, complete wetting of the catalyst had to be
veriﬁed. Al-Dahhan and Dudukovic84 developed an equation for wetting efficiency in
trickle-bed reactors that estimated the fraction of wetted catalyst given reaction
conditions. When wetting eﬁiciencies tend toward zero, gas phase reactants diffuse
directly into the catalyst, effectively bypassing the liquid phase. This increases the
catalytic activity signiﬁcantly, with respect to fully wetted catalyst, as mass transfer of
hydrogen is no longer limited by the solubility of gas in the liquid phase. Figure 3-1

illustrates this difference.

    
  

Liquid . . Liquid
quurd L1qu1d
Gas
1 Gas
Catalyst ‘
Particle

Catalyst

Particle ’

Stagna’ .
Liquid

Stagnant
Liquid

 

Flowing Flowing zghye
Liquid Liquid
(a) (b)

Figure 3-1: Comparison of (a) ﬁllly wetted and (b) partially wetted catalyst

Equation 1 was used for computation of the wetting efficiency:
1/9
[AP/Z]
1/3 '1' 7‘ ”
ULpLdP ] PLg
11L(1"813) 9191388133, (1)
2 3
11L(I ‘ 8B)

no, =1.104[

 

8" Al-Dahhan, M.H.; Dudukovic, M.P.. Chem. Eng. Sci, 1995. 50 (15), 2377—2389

39

Where:

TlCE
UL

m
0P

PL

813

REL

N

go

Gar,

Wetting efficiency [dimensionless]
Superﬁcial liquid velocity [m/s]
Density of the liquid phase [kg/m3]

Particle diameter [m]
Liquid viscosity {—k—g—l
in - s

Packed-bed void per reactor volume (bed porosity) [dimensionless]

ULPLdP

Reynolds number of the liquid phase = [dimensionless]
11 L (1 " 3 B )

Pressure drop through the catalyst [N/mz]
Length of the catalyst bed [in]
Gravitational constant [9.8066 m/sz]

 

Gravitational proportionality factor [1 kg-m ]
s -N
. . . ULdiPig . .
Galileo number of the liqu1d phase = —2(1—)3 [dimens1onless]
11L — t3B

The calculated wetting efficiency was 0.13 and 0.14 for the 16-30 and 50-100

mesh sizes respectively, which showed that the catalyst was not completely wetted. This

assumed that the equation was valid outside of the ranges for the liquid phase Reynolds

number (ReL) established by the Dudukovic paper (0.1 to 10). A pressure drop (AP) of 1

and 11 kPa across the catalyst bed (16-30 mesh and 50-100 mesh, respectively) was

estimated using the Kozeny-Carman equation85 (Equation 2), which is applicable for ﬂow

through packed beds for Reynolds numbers of less than 1.0.

 

85 McCabe, W.L. et al., Unit Operations of Chemical Engineering, 5‘h Edition, p. 153, McGraw Hill Inc,

New York, 1993

40

AP_ 150UL 111. (l—cB)2
Z g.¢§d§ 33B ‘22

 

 

Where:
(I), = (6 / dp) / (Sp / Vp) = Sphericity of the catalyst
S1, = External surface area of the catalyst

Vp = Volume of the catalyst

A bed porosity (an) of 0.36 was estimated for both mesh sizes using a table from
McCabe et al.86 based on loose packing of spheres. Sphericity of the catalyst was
estimated to be 0.7 based on a table with estimates of Sphericity for different materials in
McCabe et al.87

The wetting efﬁciency is most dependent on the Reynolds number of the liquid
phase. For experiments, a liquid volumetric ﬂow rate of 0.32 mL/min was used
corresponding to a Reynolds number of approximately 3 x 103. This, expectedly, gave a
reduced wetting efficiency. The parameter of interest in this set of studies, the particle

diameter (dp), was used to calculate both the Reynolds number and Galileo number to

determine neg.

With a large pressure drop (i.e. [AP / Z] / pLg >> 1), not; varies with dp'w.
However, with the low liquid ﬂow rate, the pressure drop was relatively small, which
made the effect of a three-fold decrease in particle diameter (750 microns to 250 microns)

only a 7% increase of net; (from 0.13 to 0.14) instead of a potential 28%. Essentially, no

further wetting of the catalyst was observed. Thus, the change in particle diameter was

 

86 McCabe, W.L. et al., Unit Operations of Chemical Engineering. 50‘ Edition, p. 155. McGraw Hill Inc.
New York, 1993
“-2 Ibid. p. 928

41

not sufﬁcient to reduce the mass transfer of hydrogen to the catalytic surface and could
not account for the drop in conversion.

With mass transfer effects ruled out, the difference in conversion between the two
particle sizes must have been due to some other factor, such as reduced residence time
due to channeling of the liquid phase or inconsistent reduction of the ruthenium catalyst
between reactions. Crushing of the catalyst by mortar and pestle may have closed off
pores in the crushed 50-100 mesh catalyst, as well. This, in turn, would seal off active
metal from the reaction. Regardless of the drop in conversion, the product distribution

was not signiﬁcantly affected by the change in particle size of the catalyst.

3.1.3 Continuous base addition

This set of experiments used the batch reactor system to observe the effects of
adding base over the course of sorbitol hydrogenolysis. Each reaction used potassium
hydroxide as a base “promoter” to improve C3 selectivity, as was consistent with prior
work 4. Over the course of the experiment, pH levels dropped due to formation of
organic acids, such as lactic acid. Adding base continuously provided a method for
maintaining the pH as a strategy of increasing C3 selecitivity.

Four experiments were performed comparing the continuous feeding of water and
0.5 wt% potassium hydroxide solution at rates of 0.1 mL/min and temperatures of 200°C
and 210°C. All reactions were carried out under 1350 psig H2 and with catalyst loadings
of 1.0 gram (dry basis) of 5 wt% ruthenium on carbon (PMC). Feedstock was 25 wt%
sorbitol (1.5 M) and 0.5 wt% KOH (0.08 M). Table 3-2 displays the product

distributions from these runs.

42

Table 3-2: Product distribution of continuous base addition experiments

 

 

 

 

 

Continuous Tem P C
EXP feed (°C)p Seci. SGecl). SeAI. $931. Conv.
13 Water 200 0.15 0.28 0.08 0.51 67%
15 0.5% KOH 200 0.21 0.18 0.05 0.44 61%
8 Water 210 0.15 0.26 0.08 0.49 96%
9 0.5% KOH 210 0.18 0.23 0.06 0.47 99%

 

 

 

 

 

 

 

 

 

 

Results show no improvement in C3 selectivity with base addition and, in fact,
appear to have slightly lower preference toward C3 materials. The product distribution
also shows an apparent preference toward PG for the base addition cases. This is an
effect of glycerol dehydrogenating to propylene glycol via a base-catalyzed mechanism.
With the maintained pH from base addition, this process was promoted as demonstrated

by the increased PG selectivity and decrease in glycerol selectivity.

3.2 Adsorption studies
3.2.1 Overview

In the interest of gaining mechanistic information, a series of adsorption studies
was designed to observe interactions between sorbitol and ruthenium metal under an inert
atmosphere. A dilute solution of sorbitol was stirred with activated ruthenium metal at
constant temperature and sampled as the reaction progressed. A decrease in sorbitol
concentration from the bulk liquid phase was assumed to be adsorption. The amount of
sorbitol removed from the bulk phase would then be compared to the measured
dispersion of the catalyst to compare the number of active sites to the substrate adsorbed.
And ﬁom this, an estimate of the adsorbed area of sorbitol on the metal surface and

relative orientation of the sugar alcohol with ruthenium.

43

We corroborated evidence of sorbitol dehydrogenation using a deuterium labeling
experiment and 13 C NMR. Sorbitol was placed in deuterated water (D20) and reacted
with ruthenium catalyst at 230°C under 1200 psig of deuterium gas. As shown in Figure
3-2, incorporation of deuterium at carbon 1 (C1) was demonstrated by induced splitting
in the NMR spectrum. This could only happen with dehydrogenation occurring at C1
followed by hydrogenation of the unsaturated intermediate with deuterium. The result
also suggested an end-on adsorption of sorbitol on to ruthenium metal prior to
dehydrogenation. This result is contrary to a model presented by Muller et al.38 which

suggests interaction of three hydroxyl groups at the active metal site.

3.2.2 Sorbitol Adsorption Experiments

Ruthenium metal sponge was used for adsorption experiments to provide a
sufﬁcient amount of metal surface area without a supporting material. Supports, such as
activated carbon can absorb as much as 0.3 grams of substrate per gram. This amount of
absorption would make any change in sorbitol concentration due to adsorption on the
active metal statistically insigniﬁcant. Metal particle diameters were approximately 100-
150 microns and the catalyst had an observed BET surface area of 0.5 mZ/gram. For
comparison, ten grams of ruthenium sponge has approximately the same metal surface
area as one gram of 5 wt% Ru/C catalyst. Solid spheres of the same diameter have an
effective surface area of approximately 5 x 104 mz/gram, so the sponge has substantial

porosity.

44

cosmcomocgcou Lo 8:830 1 03m 55 6:98 .6 Loam ”=22 0oF "N-.." 2:9".

 

DO

 

IO

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5% mm B 6.6 8 cm No em. .8 we. 0K: MK 4K 6K mK om
2%.. ._.CLL LCrL _ L .: . L .. L: ___... ___. L .TrEr :_ Z. . L .‘r _ . .Lﬁﬂfth
a; 3.11s. $§f§$3€s§§ Trims"? . EX}? “3.5%.. Li... {a 5.12.4? _zﬁrfiga}
_.._ _5 ___. .L l
__ L... ,.
oo no I o :5 a f
j _ 5:98
. 3 use; 4 he
on so 98 89 .068
new 1022 .06 ~15. com
8 no :9. a: 323.6... so.
5&8 No S on ma. op No 8 we we 0K NK 3K 6K R 8
(tittigfwicieéaf $11411!) _ _{gstrirrrtirr s. .21. 11.5....1ig \ 5.1.3. \. fastidiirxllrlzi
:_ _ ___ L...
_ A A.
xo :0 I
3 3:28
xooﬁmuoou
6: man 98 89 .062
1022 .046 ~12 8m
:0 IO :9. OK; ._2_Eow $2

6 a eoaaoeeoa a as, 3.58

 

 

use due so

 

 

 

_oeeom

 

 

45

 

 

101

The catalyst was further characterized using hydrogen and carbon monoxide
chemisorption experiments available on the ASAP 2010. Ruthenium metal was reduced
at high temperature to remove any strongly bound materials on the surface and then
exposed to dosed amounts of hydrogen or carbon monoxide. Based on the amount of gas
adsorbed to the metal, the number of active sites was calculated and reported as
dispersion. Dispersion is the fraction of metal atoms that adsorb the analysis gas over the
total metal atoms available. For the ruthenium sponge, dispersion was found to be 0.06%
with a metal surface area of approximately 0.3 m2 per gram.

With the dispersion known, an estimate of potential sorbitol adsorption was made
per gram of sponge. One gram of ruthenium (M.W. 101.06) is roughly 0.01 moles. The
amount of this metal active in H2 chemisorption at 308 K was 0.06% or 6 x 106 moles.
With one molecule of sorbitol adsorbed to each active site, approximately 1 mg of
substrate would bind to one gram of sponge.

Another estimate of sorbitol adsorption was based on the BET surface area
measurement. From the NMR data, sorbitol adsorption appeared to be end-on. Thus, the
adsorbed area of the substrate would be relatively small. An estimate of 30 A2, or
roughly double the adsorbed area of one molecule of nitrogen, was used for each
molecule of sorbitol. Thus, covering the 0.5 mz/gram metal surface area with one
monolayer would require 0.5 mg of sorbitol per gram.

Each run consisted of 10.0 grams of ruthenium sponge and 50 mg of sorbitol in
100 mL of water, agitated under helium at 100 psig in a Parr batch reactor. The two
estimated values of adsorption per gram of sponge represented a range of 5 to 10 mg of

sorbitol. This expected amount of adsorption would lower the concentration of sorbitol

46

in the liquid phase as measured by HPLC. A larger decrease in concentration would
suggest a smaller adsorption area, multi-layer adsorption, a reaction occurring on the
metal surface, or a combination of the three. A smaller drop in concentration would
indicate no reaction and an effective adsorption area of over 30 A2. In addition, the
activation energy of sorbitol dehydrogenation was determined using a set of ﬁve
experiments using identical loadings of feedstock (50 mg sorbitol) and catalyst (10.0

grams ruthenium sponge) at different temperatures (100 to 150°C).

3.2.3 Results and Discussion

Experiments showed two surprising results: a larger than expected decrease in
sorbitol concentration and the presence of chemical species other than sorbitol in the
liquid phase. Apparently, the interaction between ruthenium and sorbitol is not limited to
adsorption. A number of liquid-phase products formed that were consistent with polyols
and anhydro-products from HPLC analysis. In addition, carbon monoxide and carbon
dioxide were produced in the vapor phase, which demonstrated C-C cleavage of sorbitol.

Temperature was also important due to the increased formation of gas as the
temperature increased. At 40°C, no discernable amount (<2%) of sorbitol converted or
adsorbed to the ruthenium surface. As temperatures increased, liquid phase products,
such as mannitol, xylitol, threitol, and glycerol, began to form during the reaction.
Carbon balances calculated from HPLC showed a gradual decrease in the amount of
carbon in the liquid phase of the reaction, which suggested that material was being
adsorbed by the ruthenium metal. Figure 3-3 illustrates the change in sorbitol and total
liquid phase product concentrations versus time at 40, 120, and 150°C (Experiments 101,

102, and 100 respectively).

47

352335 83983 _oxﬁom
DEC. .m> $030K 3an 25v: 3898 Eaten .Ho cosmbaoocoo ”mun oaswrm

 

 

 

 

 

 

 

:5 we:
w r _ . _ od
0002320. 4 to
23.2. .38 :o .. 4..
0°02 - .398 :o .. .....m N o
0009 $20. .. .
23g. .38. La .. « $3.-.-
liilltlllili .- Wifff. %! y Md
ooomzoggsmiin .....u.wm.n+f.::: %
.. /
23g 38le W iiiiiiiiiiii .. .. i i i

 

 

Ooov - _ouﬁaowlT

 

 

 

48

(116) uouenuaouoo

 

 

 

pm

Results ﬁ'om these runs showed an apparent increase in adsorption of organic
material to the metal surface. At 120°C, ~7 mg of material, or 0.7 mg/g sponge, was
removed from the liquid phase, while at 150°C, 23 mg was removed (2.3 mg/g). The
quantity of “adsorbed” material rose with temperature, which did not make sense, and the
amount removed from solution at 150°C was well above prior estimates of 0.5 — 1.0 mg/g
sponge.

To address these inconsistencies, an extended run was conducted (Experiment
163) at 120°C with the same catalyst loading (10.0 grams of ruthenium sponge) and
pressure (100 psig He) of previous adsorption experiments with a similar feedstock of 50
mg of sorbitol in 160 mL of water. No liquid or gas phase samples were taken over the
course of the run to avoid the removal of material from the reaction. After 40 hours, the
reactor was cooled to room temperature and the gas and liquid samples were taken.

Analysis of the gas phase indicated the presence of carbon dioxide, methane, and
ethane in the headspace of the reactor. When the carbon balance included these gases, it
closed to nearly 100%, which showed that no detectable amount of material adsorbed to
the metal catalyst. HPLC identiﬁed a number of products in the liquid phase. The

product distribution from this experiment is found in Table 3-3.

49

 

 

0rd

Table 3-3: Liquid and gas phase products of sorbitol adsorption at 120°C (Exp. 163)
Conditions: 120°C, 100 psig He, 10.0g Ru sponge, 50 mg sorbitol, 160 mL solution

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

% of carbon
Products cl 0 su r e*
Sorbitol 52 :t 2
Xylitol 7 i 1
Mannitol 4 :t 1
Acetic acid 3 i 1
Threitol 2 i l
Glycerol **
Unknown *** 12 :t 2
Total liquid phase 80 i 6
Methane 12 i 2
Carbon dioxide 4 ﬂ: 1
Ethane **
Total gas phase 16 i 3

* Carbon balance is based on 19.8 mg of carbon from the sorbitol feedstock

** Detected, but not in signiﬁcant quantity

*** The concentration for unknown compounds was determined by using
an average response factor based on known compounds

The energy of activation for sorbitol dehydrogenation was calculated based on our

postulated reaction pathway of sorbitol dehydrogenation prior to further reaction. Thus,

any change in sorbitol concentration was the result of dehydrogenation. Using a ﬁrst

order rate equation:

-rso = k * Cso

dt

=_k‘Cso

(3)

(4)

Where:

 

 

dC
-rso = (1:0 = Rate of sorbitol dehydrogenation [mol / liter / hr]
k = 1St order rate constant [hr ‘1]
C30 = Concentration of sorbitol [mol / liter]
_ _ . . dcso
From Equation 4, the rate constant k can be estimated by calculating dt at the

onset of the reaction. Using concentration versus time data, the change of concentration

dt

 

was measured between the ﬁrst two data points and used to calculate . This result

was then divided by the initial sorbitol concentration of 0.5 g/L or 2.74 x 10'3 moles/ liter
for the value of k. The rate constant was then related to the activation energy of

dehydrogenation by Arrhenius’ law“:

-Eact

 

Where:
k = lSt order rate constant [ hr '1]
k0 = pre-exponential factor [hr '1]
-EM = Activation energy [ J / mol ]
R = Ideal gas constant [J / mol / K ]
T = Temperature [K ]

 

88 Levenspiel, 0., Chemical Reaction Engineering, 3ml Edition, p. 27, Wiley, New York, 1999

51

Transforming the equation:

 

_ : act
1n(k) RT + 1n(k0 ) (6)
Or:
_ E ac 1
- 1n(k) = —E_‘[-f] + 1n(ko) (7)

The rate constant k was calculated at temperatures of 100, 110, 120, 135, and
150°C. Plotting -ln(k) versus 1 / T yielded Figure 3-4, which was ﬁtted with a linear
trendline. The slope of the trendline represented Em / R, or the activation energy of
sorbitol dehydrogenation over the ideal gas constant. From this, an activation energy of

30 kJ / mol was calculated. Table 3-4 displays the calculations from these experiments.

Table 3-4: Calculations for Em of sorbitol dehydrogenation [ -ln (k) vs. 1 / T ]
Conditions: 0.5 g/L, 10.0g ruthenium sponge, 100 psi helium

 

 

 

 

 

 

 

 

Tem 1 IT x 103 2% " 10‘ '"ma' 05° x 103
EXP' (10p (K4) moles [macs ) " '" (k)
E ) liter -hr
liter -hr
109 373 2.68 3.4 2.74 2.07
103 383 2.61 2.6 2.74 2.35
102 393 2.54 7.1 2.74 1.35
111 408 2.45 11.0 2.74 0.92
100 423 2.36 7.7 2.74 1.27

 

 

 

 

 

 

 

 

52

CNN

cobwcowohgnow ﬁoHESm mo >905 cosﬂggm mo 5528:8800 61mm 9.59%

:gro. 3:

 

 

 

 

 

 

 

 

 

mod 8N mmd omd mvN ova mmN omN
_ p _ _ _ _ _ 0.0
md
w 1 2
m. ...
3 W:
x
(
. ow
mm o H mm
w 5s - 63 u >
mm

 

 

 

 

on

53

Work by Peereboom89 focused on an even more dilute system and showed that
sorbitol reacted with ruthenium at temperatures as low as 30°C and adsorbed to the
catalyst within thirty minutes of contact. After several hours, the amount of sorbitol
adsorbed was only 0.1 mg per gram of sponge, which corresponded to an adsorption area
of 90 A2. The amount of sorbitol adsorbed was signiﬁcantly lower than the initial
estimates of 0.5-1.0 mg per gram of sponge. Based on the low temperature of the
experiment, this result suggested that sorbitol adsorption has a small activation energy.
The small amount of adsorption also suggested 1) a larger than expected adsorption area
for sorbitol (3 times larger) with strong adsorption, or 2) sorbitol adsorbs weakly to the
metal surface and occupies only a small fraction of the sites. The large adsorbed area
would require sorbitol to occupy ﬁve to ten active sites on the metal surface in order to
prevent additional sorbitol molecules to interact. This would require multiple hydroxyl
groups to complex with the catalyst metal. H-D exchange on the carbon chain of sorbitol
would then be expected at multiple sites. From Figure 3-2 however, H-D exchange was
only observed at C 1. Kinetic modeling of glucose hydrogenation to sorbitol also reﬁites
the multiple adsorption site model due to excellent agreement with experimental data
when sorbitol adsorption is neglected.”22 Thus, weak adsorption of sorbitol is an

appropriate conclusion.

3.2.4 Summary
Overall, this set of results illustrates several key steps in the hydrogenolysis
process. First, through 13 C NMR work and initial HPLC analysis, dehydrogenation was

conﬁrmed as the ﬁrst step of hydrogenolysis. Second, ruthenium and sorbitol react to

 

89 Unpublished data from Miller group produced by L. Peereboom

54

form products regardless of the presence of hydrogen or low temperatures. Ruthenium
appeared to lower the activation energy sufficiently to enable dehydrogenation at low
temperature. With the addition of high hydrogen pressures, as used in hydrogenolysis,
sorbitol is favored over the dehydrogenated intermediate. Higher temperatures are,
therefore, necessary to overcome this preference to sorbitol. In addition, the higher
temperature allows for sufficient activity in the hydrogenolysis reaction to convert the
feedstock within a few hours.

Another important observation is that sorbitol does not remain adsorbed to metal
catalyst in appreciable amounts and thus, only weakly adsorbs to active catalyst sites.
This is an effect of a larger than expected adsorption area or low energy barriers for
adsorption and desorption of the sugar alcohol. The result is consistent with observations
of the carbon chain of sorbitol using 13C NMR and with kinetic modeling of glucose
hydrogenation to sorbitol reported in the literature.

Sorbitol-metal interaction was studied ﬁtrther by observing the effects of adding
base to imitate conditions of hydrogenolysis for a direct comparison. Chapter 4

summarizes these experiments.

55

Chapter 4 - Metal and base-catalyzed C-C cleavage of
sugar alcohol and monosaccharide feedstock

4.1 Metal-catalyzed alkaline degradation

Hydrogenolysis of sorbitol is enhanced by the addition of a base “promoter”,
typically sodium hydroxide or potassium hydroxide, to produce higher yields of desired
C3 materials: propylene glycol, glycerol, and lactic acid. To study this promoter effect,
additional adsorption experiments were conducted with base added to the feedstock.
Preliminary results with 0.6 wt% sorbitol (0.033 M) and 1.0 wt% KOH (0.17 M) at
150°C showed production of lactic acid, propylene glycol, glycerol, and ethylene glycol
without the addition of hydrogen (Experiment 56).

This result was explored further by using higher concentrations of sorbitol (32 g/L
or 0.18 M) to see if this trend in C3 production continued. In addition, different metal
catalysts, temperatures, and initial base concentrations were used to observe the
subsequent effect on product distributions. Xylitol and glycerol were also used as
feedstock to compare metal-substrate interactions and observe carbon-carbon cleavage
tendencies relative to sorbitol. These results are described in detail in the following

sections.

4.1.1 Control experiments

Control experiments were used to show that the combination of base and metal
was necessary to convert the feedstock. A 0.6 wt% sorbitol solution (0.03 M) with 1.0
wt% KOH (0.16 M) was heated to 150°C for two hours under helium. Less than 1% of
the sorbitol reacted, which indicated that base alone was not sufﬁcient to react with

sorbitol. Three additional experiments used initial sorbitol concentrations of 0. l, 0.2, and

56

0.6 wt% (0.0055, 0.011, 0.033 M) with 10.0 grams of ruthenium metal under helium.

The rates of reaction for the three runs were 0.032, 0.056, and 0.035 grams of sorbitol per
hour, respectively. While the rates were non-zero, each was small enough to demonstrate
that metal alone is not sufficient for rapid conversion of the substrate. Thus, the
combination of metal catalyst and base in the feedstock is needed for signiﬁcant

conversion of the sorbitol.

4.1.2 Effect of base loading on product distribution

Of the parameters studied, base concentration had the most dramatic effect on the
reaction. At high sorbitol concentrations, conversion of the feedstock halted without
sufﬁcient potassium hydroxide present. Figures 4-1 and 4-2 illustrate that high pH was
necessary for continued reaction. The pH of each sample was measured by an Accumet
950 pH meter (Fisher Scientiﬁc). One limitation of the pH meter was higher than
expected readings for pH levels > 13. For instance, the feedstock with 1.47 M KOH
should have had a pH of 14.2 instead of 14.9 as reported in Figure 4-2. The graph does
properly reﬂect, however, the effect of pH on the conversion of the reaction.
Experiments were conducted at 150°C under 100 psig helium with 10.0g ruthenium

sponge catalyst.

57

 

 

 

Experiment 63 Experiment 68

 

 

 

 

   

 

 

0.15 M KOH, 0.87 M sorbitol 1.47 M KOH, 0.90 M sorbitol
r+pH +Conv L—j—pH +Conv

6° 60

5° 50

4° 5, 40 8
" “a

30 g 30 E

20 3 20 §

10 10

O 0

0 120 240 360 480 O 120 240 360 480
Time (min) Time (min)

 

 

 

Figure 41-]: pH and conversion vs. time - low base Figure 4-2: pH and conversion vs. time - excess base

In Chapter 3, it was shown that sorbitol adsorbs to the metal surface and
dehydrogenates. In these reactions a hexose is thus formed, which is available in the
reaction medium to undergo base-catalyzed carbon-carbon cleavage reactions. With
sufficient base present, the alkaline degradation pathways continue, presumedly, until the
feedstock is exhausted (see Figure 4-2). However, at lower initial base concentrations,
organic acids that form neutralize the reaction medium effectively stopping the carbon-
carbon cleavage of the transient sugars formed from sorbitol dehydrogenation (Figure 4-
1). Quantitatively, 10% of 0.87 M sorbitol converted aﬁer 8 hours in Experiment 63.
Assuming two moles of organic acid form with each mole sorbitol reacted, ~0.18 M of
acids would form and neutralize the 0.15 M KOH in the feedstock. The pH level was
still above neutral at the end of the experiment, which is explained by the formation of
non-acidic products, such as propylene glycol and glycerol, which do not neutralize
KOH. When the transient sugars stopped reacting, sorbitol ceased to react as well. This

could be explained in two ways:

58

1) When sorbitol dehydrogenates to form a hexose, the hexose adsorbs to the catalyst
surface. With sufficient base, the sugar desorbs and reacts toward products. When
the base is exhausted, the glucose remains on the catalyst surface with higher afﬁnity

than the sorbitol, effectively shutting down the dehydrogenation reaction.

2) The dehydrogenation reaction is reversible between sorbitol and the sugar formed.
With alkaline degradation, the sugar continues to convert to products, which
promotes dehydrogenation. When the degradation stops, sorbitol and the glucose

analog reach an equilibrium, which halts any further conversion of sorbitol.

Of the two options, the ﬁrst is inconsistent with published results. Several
papers”22 that involve the hydrogenation of sugars to sugar alcohols note that the
reaction is between chemisorbed hydrogen and unadsorbed sugar based on kinetic
models. This suggested that the sugars produced from dehydrogenation do not compete
for the active sites used by sorbitol.

In addition to reduced activity in the reaction, the product distribution at lower
initial base concentration shifted toward making propylene glycol instead of lactic acid.
This was attributed to reduced rates of organic acid formation at lower pH”, which
allowed hydrogenation of unsaturated intermediates to occur. These intermediates, such
as glyceraldehyde and pyruvaldehyde, form via retro-aldol condensation and ﬁ1rther react
to form other intermediates, recondense, or rearrange to form organic acids. When
hydrogen and catalyst are present, they may also hydrogenate, which would account for

the production of glycerol and propylene glycol. However, in these reactions, only

helium was added to the reactor. Thus, for PG to form, hydrogen must have been

59

obtained from the sorbitol feedstock or from the residual hydrogen on the pre-reduced
catalyst.

Chemisorption experiments show that 0.07 cm3 (STP) of hydrogen adsorb per
gram of ruthenium sponge. This corresponds to 0.7 cm3 or 1045 moles of hydrogen that
would have been available to each reaction. Runs from this set of studies typically
produced 10'3 moles of propylene glycol and glycerol, which would more than exhaust
the supply of hydrogen on the catalyst. Thus, the hydrogen source must be from the
sorbitol feedstock.

Overall, the selectivity to C3 products did not change signiﬁcantly between
experiments, but the distribution of the C3 products did change at different base
concentrations. As discussed earlier, experiments with lower levels of base showed
higher selectivity toward propylene glycol while higher initial base concentrations in the
feedstock produced more potassium lactate. While the majority of C3 product was lactic
acid, there was a clear shift toward propylene glycol, as seen in Table 4-1. Results are
reported in terms of selectivity, which reﬂects the number of moles of product formed per
mole of sorbitol reacted. Propylene glycol (PG) and lactic acid (LA) are reported
separately. C3 selectivity is the combination of the C3 products propylene glycol, lactic

acid, glyceric acid, and glycerol.

60

Table 4-1: C3 product distribution — effect of initial base concentration

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Exp 33.22: $3.2. .53. $3. sit. :3. c... “32.?" if.)
58 0.03 M 0.15 M 0.16 0.53 0.68* 2.10 96% 97% 2.5
64 0.03 M 0.15 M 0.12 0.67 079* 1.73 97% 104% 3.0
62 0.18 M 0.15 M 0.16 0.65 081* 1.13 43% 95% 8.0
80 0.18 M 0.83 M 0.05 0.70 0.85 0.95 91% 87% 8.0
67 0.18 M 0.83 M 0.06 0.77 083* 1.15 99% 85% 7.0
63 0.87 M 0.16 M 0.25 0.62 087* 1.02 11% 101% 8.5
68 0.90 M 1.47 M 0.05 0.81 085* 0.76 62% 86% 8.3

 

 

* C3 sel. for these experiments did not include glycerol or glyceric acid due to lack of UV spectroscopy

For experiments 58, 62, 63, 64, 67, and 68, the ultra-violet response at 210 nm
was not recorded for the samples. Propylene glycol and lactic acid were observed using
the refractive index response. However, glycerol and glyceric acid were not identiﬁed
because the peaks could not be resolved. Fomiic acid and glycerol have very similar
retention times with the column conditions used for HPLC analysis. Likewise, the
responses for sorbitol and glyceric acid overlap, which made quantiﬁcation difﬁcult.
When used in combination with the UV response, the combination of glyceric acid and
sorbitol as well as formic acid and glycerol can be resolved. Experiment 80 used both
UV and RI responses to resolve all four components. Glycerol and glyceric acid had a
selectivity of 0.04 and 0.09 respectively. With the inclusion of glycerol, the apparent
concentration of formic acid decreased. Thus, the reported selectivity of formic acid for
reactions, with the exception of Experiment 80, is inﬂated.

Table 4-2 lists other minor products of the reactions. C2 products of ethylene
glycol (EG) and glycolic acid (GlyA2) are included along with formic and acetic acids

(FA, AA). Changes in base and sorbitol concentrations did not show a large change

61

between selectivity among the experiments. There appears to be a slight preference
toward ethylene glycol and glycolic acid with lower initial base concentrations with
respect to the sorbitol concentration. The effect, however, is not signiﬁcant enough to

report conclusively.

Table 4-2: C1 and C2 product distribution — effect of initial base concentration

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Exp Sorb. KOH EG GIyA2 AA FA
Cone. Conc. Sel. Sel. Sel. Sel.

58 0.03 M 0.15 M 0.04 0.22 — 2.10
64 0.03 M 0.15 M 0.06 0.21 — 1.73
62 0.18M 0.15M 0.12 0.17 — 1.13
80 0.18 M 0.83 M 0.05 0.12 0.10 0.95
67 0.18 M 0.83 M 0.03 0.14 — 1.15
63 0.87 M 0.16 M 0.08 0.22 — 1.02
68 0.90 M 1.47 M 0.04 0.13 — 0.76

 

 

Another effect of higher initial base concentrations was an increase in gas
products. The carbon balance steadily decreased when more potassium hydroxide was
added to the feedstock. This reﬂected the production of carbon dioxide and methane as
the reaction progressed. Quantitative results were not determined for these runs, but the
gas products CH4 and CO2 were consistent with gas samples from the adsorption

experiments reported in Chapter 3.

4.1.3. Effect of catalyst metal on product distribution

This group of experiments compared the product distribution of the alkaline
degradation reactions that used 10.0 grams of ruthenium sponge, 1.0 grams of ruthenium
on carbon, and 1.0 grams of palladium on carbon as catalysts. Both ruthenium catalysts

were pre-reduced to activate the metal (see Section 2.3.3 . 1) while palladium required no

62

pre-reduction. Feedstock concentrations of sorbitol and potassium hydroxide remained
consistent for each run at 0.18 M and 0.85 M respectively. Reaction temperature and
pressure were at 150°C and 100 psig of helium.

The three catalysts showed similar selectivity to C3 products, and more
speciﬁcally, toward lactic acid as shown in Table 4-3. Palladium had higher selectivity
and yield toward C3 products, but the activity of the catalyst, i.e. the rate of sorbitol
conversion, was not as high as the ruthenium catalysts. Both ruthenium catalysts also
formed signiﬁcantly more formic acid compared to palladium. Since gas formed with the
palladium catalyst (73% carbon balance) and alkaline degradation pathways are solution
based, the lower selectivity to formic acid with palladium was due to conversion of

formic acid precursors to gas.

Table 4-3: C3 product distribution at 6 hours — effect of catalyst metal

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EXP 9‘39"!“ .53. s33. 35.. 83.2“ s21. :3. CW- C33?"
80 Ru sponge 0.05 0.68 0.04 0.09 0.86 0.95 91% 87%
69 Ru/C 0.10 0.68 —— 0.06 0.84 0.79 80% 79%
83 Pd/C 0.09 0.72 0.05 0.12 0.98 0.10 71% 79%

 

 

The C2 product distribution (Table 4—4) between experiments showed no notable
change between catalyst metal. Ethylene glycol (EG), glycolic acid (GlyA2), and acetic
acid (AA) all form in small amounts. Production of ethylene glycol further demonstrates

the in situ production of hydrogen during the reaction.

63

Table 4-4: C2 product distribution at 6 hours — effect of catalyst metal

 

 

 

 

 

 

 

 

Exp Catalyst SE3 6:2? :3. Conv. egg?"
80 Ru sponge 0.05 0.12 0.09 91% 87%
69 Ru/C 0.07 0.05 — 80% 79%
83 Pd/C 0.09 0.02 0.11 72% 79%

 

 

 

 

 

4.1.4 Effect of temperature

Three reactions conducted at 150, 170, and 190°C were used to observe the effect
of temperature on product distribution. The initial concentration for sorbitol and KOH
remained at 0.18 M and 0.85 M, respectively, and the reaction gas was helium at 100

psig. Table 4-5 summarizes the ﬁndings from these runs.

Table 4-5: C3 product distribution — effect of reaction temperature

 

 

 

 

 

 

 

 

 

 

 

EXP- T381" .53. $3. sii. :3. CW- 6.8.3... (11.,
80 150 0.05 0.70 0.82 0.95 91% 87% 6.0
76 170 0.04 0.72 0.76 0.96 98% 70% 4.0
77 190 0.04 0.62 0.70 1.09 100% 71% 2.0

 

 

 

 

Selectivity to lactic acid and C3 products decreased slightly from 150 to 190°C,
while formic acid production remained essentially constant. Gas formation was
promoted with higher temperatures as evident from decreasing carbon balances. Further,
at 190°C, gas continued to form as the reaction progressed.

C2 products converted into gas as the reaction temperature increased, as seen in
Table 4-6. Ethylene glycol and glycolic acid diminished with higher temperatures while

acetic acid production remained constant.

64

Table 4-6: C2 product distribution — effect of reaction temperature

 

 

 

 

EXP T138” 33. 63?." 33. CW- C23?"
80 150 0.05 0.12 0.09 91% 87%
76 170 0.02 0.06 0.09 98% 69%
77 190 0.00 0.04 0.09 99% 69%

 

 

 

 

 

 

 

 

 

4.1.5 Effect of feedstock

In addition to sorbitol, xylitol and glycerol were used as feedstocks to observe
similar metal-polyol interactions already studied. Xylitol is of interest from earlier work
at MSU.4 Glycerol is of interest because of its production from bio-diesel synthesis. The
similar structure of xylitol, glycerol, and sorbitol (i.e. saturated polyols) allowed
comparisons of dehydrogenation and patterns in the carbon-carbon cleavage selectivity.
Initial concentrations of xylitol and glycerol were 0.18 M with a base concentration 0.85
M. Reactions were carried out at 150°C under 100 psig of helium.

Xylitol showed the same general type of product distribution as sorbitol
degradation. Lactic acid was the primary C3 product with smaller, but signiﬁcant
amounts of propylene glycol and glycerol observed as well. Formation of PG and
glycerol ﬁirther demonstrated the initial dehydrogenation step and degradation of the
subsequent sugar seen with sorbitol. F orrnic and glycolic acids also formed in signiﬁcant
amounts, which was consistent with alkaline degradation pathways. Table 4-7

summarizes the product distribution after six hours.

65

Table 4-7: Product distribution from alkaline degradation of xylitol (Experiment 79)

 

PG GO LA GIyA C3 EG GIyA2 FA Conv C t
Sel. Sel. Sel. Sel. Sel. Sel. Sel. Sel. ' Bal. (hr)

0.05 0.05 0.48 0.08 0.66 0.03 0.18 0.76 62% 92% 6.0
Maximum selectivity with xylitol for C3 products and formic acid (FA) is 1.0 and 5.0 respectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

Presumedly, C3 and C2 selectivity would be nearly the same with cleavage of C2
and C3 of the xylitol carbon chain. However, glycolic acid (GlyA2) and ethylene glycol
(EG) account for only one third of the total C3 produced. This discrepancy can be
explained, however, by the limited number of reactions available for the C2 aldehydic
intermediate glycolaldehyde. Aﬁer retro-aldol condensation of the C5 sugar to
glyceraldehyde (C3) and glycolaldehyde (C2), the C3 intermediate can undergo B-
elimination of hydrogen to pyruvaldehyde and further react to form lactic acid, which
happened readily as seen in Table 4-7. However, glycolaldehyde cannot undergo B-
elimination of hydrogen because it does not have a [3 carbon. Thus, the intermediate can
only hydrogenate to ethylene glycol, undergo a Cannizzaro reaction to form glycolic acid
and ethylene glycol, or recondense with other intermediates to react further. From the
low selectivity of glycolic acid and ethylene glycol, glycolaldehyde must recondense with
higher selectivity.

Two runs with glycerol compared different proportions of glycerol to potassium
hydroxide and the effect on product distribution (see Table 4-8). Both reactions
converted glycerol to lactic acid with the highest selectivity. The lower concentration
run, however, showed signiﬁcantly higher selectivity toward lactic acid, propylene
glycol, and glyceric acid (0.66 versus 0.39). Further, selectivity to formic acid was much

lower at the reduced base concentration (0.60 versus 1.03). The lower carbon balance for

66

 

the higher glycerol and potassium hydroxide loadings also showed increased gas

production.

Table 4—8: Product distribution from alkaline degradation of glycerol

 

 

 

GO KOH PG LA C3 FA Carbon t
EXP Conc. Cone. Sel. Sel. Sel. Sel. Conv. Bal. (hr)
72 0.065 M 0.15 M 0.10 0.44 0.66 0.60 81% 98% 6.0
78 0.18 M 0.85 M 0.06 0.30 0.38 1.03 62% 82% 6.0

 

 

 

 

 

 

 

 

 

 

 

 

Maximum selectivity with glycerol for C3 products and formic acid (FA) is 1.0 and 3.0 respectively.

The C2 product distribution (Table 4-9) also showed carbon-carbon cleaving of
glycerol in addition to C3 products. Ethylene glycol selectivity remained constant while

selectivity to glycolic acid dropped from Experiment 72 to 78.

Table 4-9: C2 product distribution — alkaline degradation of glycerol

 

 

 

GO KOH EG GIyA2 Carbon
EXP Conc. Conc. Sel. Sel. Conv. Bal.
72 0.065 M 0.15 M 0.03 0.14 81% 98%
78 0.18 M 0.85 M 0.03 0.04 82% 82%

 

 

 

 

 

 

 

 

 

4.1 .6 Summary

For alkaline degradation of sugar alcohols to occur, both base and an activated
metal catalyst are necessary to see signiﬁcant conversion of the feedstock.
Dehydrogenation of a primary hydroxyl group over the metal catalyst is the ﬁrst step of
the reaction as seen from product distributions and gas formation. Dehydrogenation and
subsequent degradation of the sugar explains the formation of intermediates during the
reaction as well as the presence of saturated polyols formed via hydrogenation (e. g. PG,

GO, and EG). These could not form without ﬁ'ee hydrogen in the system or adsorbed

67

hydrogen on the catalyst. Chemisorbed hydrogen from the reduction step of the catalyst
only accounts for 1045 moles of hydrogen while the hydrogenated products need
~10'3 moles of H2 in order to form. Thus, H2 is produced from the feedstock.

The reaction temperatures and pressures were signiﬁcantly reduced with respect
to hydrogenolysis while maintaining similar selectivity to C3 products, albeit with lower
feedstock concentrations. The addition of base to the reaction allowed for continued
conversion without the need for higher temperatures. To bring this reaction ﬂirther, the
sugar alcohol feedstocks were replaced with monosaccharides. Since dehydrogenation is
the ﬁrst step of alkaline degradation of sorbitol, it seemed prudent to start with a sugar
instead to remove the dehydrogenation step and directly react to products. Section 4.2

describes the efforts taken in this direction.

4.2 Solution-based alkaline degradation

Base-catalyzed carbon-carbon cleavage of monosaccharides, otherwise known as
alkaline degradation, has been well researched with several papers covering mechanisms,
product distributions, and the various chemistries associated with the reactions.“77 Since
the literature is well-developed, only a few experiments were necessary. These runs
served to conﬁrm literature results, increase familiarity with the chemistry, and improve
analytical techniques to identify and quantify products. The process variables of
temperature, feedstock sugar, and initial sugar concentration in the feedstock were varied
to observe transient behavior of the reactants and products. No metal catalyst was used

for these experiments, as none was needed to cleave the sugars.

68

4.2.1 Effect of temperature - alkaline degradation

Experiments in which the temperature was varied demonstrated how readily
alkaline degradation proceeds. Runs at 70°C and 150°C completed in 30 minutes or less
and gave nearly identical product compositions. When the temperature was reduced to
40°C, the reaction slowed sufficiently to observe changes in concentrations of reactants
and products over time. Typical runs at 40°C lasted six hours with intermittent sampling
every 30 to 60 minutes. Table 4-10 outlines the major products of glucose degradation at
40°C. Glucose and potassium hydroxide concentrations of the feedstock were 0.18 M

and 0.85 M respectively.

Table 4-10: Product distribution — alkaline degradation of glucose

 

 

 

 

 

 

 

 

 

 

 

 

 

lsom LA GlyA Cs GlyA2 AA FA C
EXP Sel. Sel. Sel. Sel. Sel. Sel. Sel. C°‘“" Bal. ”h"
90 0.19 0.90 0.09 0.99 0.07 0.04 0.28 87% 97% 6.0

 

 

Consistent with results reported in literature, lactic acid was the primary C3
product from the degradation reaction with a minor amount of glyceric acid. Unlike the
sorbitol degradation reactions, no glycerol or propylene glycol was produced. No metal
catalyst was present to hydrogenate intermediates and no hydrogen was evolved from
dehydrogenation of sorbitol, which explains this result. Yields of lactic acid reportedly
increase with further decreases in the process temperature.77 However, the reaction times
increase a signiﬁcant amount with little gain in lactic acid selectivity. At 22°C, complete
conversion of glucose and ﬁuctose feedstock takes over two days, as shown in Figure 4-

3, while even lower temperatures require several weeks.64

69

4.2.2 Effect of feedstock sugar - alkaline degradation

Reducing the temperature ﬁirther to 22°C provided insight on the reactivity of
glucose, fructose, and mannose. Plotting the combined concentration of these sugars
versus time over the course of the reaction showed that fructose was more reactive than
glucose, which was much more reactive than mannose. Figure 4-3 illustrates this point.
By using the total concentration of the three sugars, the isomerization reactions were
ignored and only the conversion to degradation products accounted for the decrease in
sugar concentration.

The result was consistent with MacLaurin‘I’9 and illustrated how ﬁ'uctose cleaves
readily to the intermediates necessary for organic acid production. Glucose and mannose
did not convert as quickly, which reﬂected the need for each to isomerize to ﬁ'uctose

before products formed.

4.2.3 Effect of initial sugar concentration — alkaline degradation

A ﬁnal set of experiments involved varying the initial concentration of feedstock
sugar while maintaining a constant base loading. From de Bruijn et al.72, low
concentrations of sugar produce higher selectivity to lactic acid relative to higher sugar
loadings. Experimental results conﬁrmed this result with glucose and xylose as reported
in Table 4-11. The de Bruijn paper conjectured that the increase in C3 selectivity is due
to the lower ionization constant (pKa) for sugars at lower concentrations (see Table 4—12).
The lower pKa increased ionization of the sugars and enhanced isomerization rates, via an
enediol intermediate, relative to higher sugar concentration experiments. This increased
the residence time of fructose in the reaction medium. Retro-aldol condensation of

fructose then produced lactic acid with higher selectivity. Also, with less sugar, less

70

EOmZ 2 o; .Ewsm 2 wood .Oomm - £83980
08C. .9? 069.600.“ 882:.“ was 683% 688.88 89w 85.35588 Ema. :38. "We 0.53m

$505 wE_._.
a. a. 8 8 em 9 N, o o

 

 

 

 

 

$302”. :9 . . d: .. o...

 

1.7 3

omooam. 1. I /M

C- om

 

owogms. 11.1

 

I
w
3.. ON
4.
vi. 8
u
.,
Om

oé

 

 

 

 

 

(1/6) uonenuaouoo

71

organic acid was produced to neutralize the base. Since the retro-aldol condensation and
rearrangement reactions necessary for lactic acid production are base-catalyzed, the
reduced amount of neutralization allowed the reactions to go to completion without
competition from other side reactions, again enhancing the selectivity. The reactions
were conducted under helium at 150°C and 100 psig of total pressure with 0.85 M

potassium hydroxide in the feedstock.

Table 4-11: C3 yield of high and low concentrations of xylose and glucose

 

 

 

 

 

 

 

 

 

 

 

 

 

0
EXP Reactant C(o 111ch S11. (3:: :9le thedorgtiical (fir)
85 Xylose 0.01 0.90 0.00 0.90 90% 1.0
84 Xylose O. 18 0.57 0.05 0.62 62% 2.0
82 Glucose 0.01 1.31 0.22 1.53 76% 1.0
88 Glucose 0.18 0.76 0.24 1.00 62% 1.5

 

 

Table 4-12: pKa values for glucose and fructose at varying concentrations

 

Conc. (M) Glucose Fructose

 

 

 

 

 

 

 

0.01 12.7 12.5
0.50 13.8
1.14 14.0 14.2

 

4.2.4 Summary

Alkaline degradation provides a method to cleave sugars to C3 products with high
selectivity at relatively low temperatures. Fructose showed the highest rate of conversion
of the sugars tested with glucose at a somewhat lower activity and mannose at a much
lower level. Reduced concentrations of initial sugar feedstock increased the selectivity to

lactic acid, which was consistent with the literature.

72

The main product of alkaline degradation is lactic acid instead of the more
desirable products, propylene glycol and glycerol. In order to shift this product
distribution, hydrogenation is necessary. Conceptually, the main intermediates from the
degradation pathways, glyceraldehyde and pyruvaldehyde, could be hydrogenated as they
form to yield glycerol and propylene glycol, respectively. This act of “capturing” the
aldehydic intermediates would shift the C3 product distribution away from lactic acid and
toward the desired polyols without the high temperatures required for hydrogenolysis.

Chapter 5 addresses this strategy.

73

Chapter 5 - Alkaline degradation with hydrogenation

5.1 - Overview

The combination of hydrogenation and alkaline degradation represents the main
work of the research presented. Chapters 3 and 4 parsed sorbitol hydrogenolysis into
distinct chemical reactions that occur simultaneously: dehydrogenation of sorbitol to a
hexose complex, base-catalyzed carbon-carbon cleavage of the sugar into aldehydic
intermediates, and subsequent hydrogenation of these intermediates into saturated
products. Using those results, experiments were conducted to eliminate the ﬁrst step of
the process by starting with a sugar instead of a sugar alcohol and then optimize the
remaining two reactions to produce propylene glycol, glycerol, and lactic acid.

Previous work by de Bruijn et al.77 outlines pathways for glyceraldehyde and
pyruvaldehyde production from the sugar cleavage reactions. Simple hydrogenation of
these aldehydes produces glycerol and propylene glycol directly. The strategy of this
work was to promote hydrogenation of the intermediates as they formed to increase
yields of glycerol and propylene glycol, reduce organic acid production (namely lactic
acid), and do all of this at lower temperatures and hydrogen pressures relative to

hydrogenolysis. The results are summarized below.

5.2 - Parametric study: alkaline degradation with hydrogenation

A parametric study was used to observe a number of variables over a range of
values. Hydrogen pressure, type of catalyst, catalyst loading, concentration of base,
sugar loading, and temperature were the process variables adjusted to see corresponding
changes in product distribution. Results from each section are presented in tabular form

with the following descriptions:

74

Conv:

C3 sel:

Max C3:

% of Max:

C3 yield:

Fraction of starting material hydrogenated to a sugar alcohol.
Fraction of feedstock sugar that isomerized to a different C5 sugar.

Fraction of feedstock that formed non-C5 products via alkaline
degradation.

The fraction of feedstock sugar that hydrogenated, isomerized, or
degraded. I.e. Conv. = Hyd. + 180. + Deg

Selectivity of the reaction to form the C3 products glycerol, propylene
glycol, lactic acid, and glyceric acid. That is, the number of moles of C3
products formed with respect to the number of moles of feedstock sugar
that reacted. For example, if 0.1 moles of glucose reacted and formed
0.16 moles of C3 products, C3 sel. would be equal to 0.16 / 0.1 or 1.6
moles of C3 product per mole of glucose reacted. The maximum value of
C3 sel. is 2.0, based on C6 feedstock.

The maximum amount of C3 product that could form based on the amount
of converted feedstock sugar. Max C3 = 2*Deg, which reﬂects the
potential formation of two moles of C3 products from one mole of C6
sugar.

C3 sel. divided by Max C3. This result relates the amount of C3 products
that were produced to the maximum amount of C3 products that could
have been produced based on the fraction of feedstock sugar that
degraded.

The amount of C3 products formed with respect to the amount of feedstock
charged to the reactor. C3 yield has units of mole product formed per
mole of feedstock fed.

Standard conditions for each reaction were 40°C at 100 psig of hydrogen with an

agitation rate of 1000 rpm. Five percent ruthenium on carbon was the standard catalyst

loaded at 0.1 grams, dry basis, which was pre-reduced at 23 0°C overnight at 200 psig of

hydrogen. The feedstock for each reaction was 100 grams of total solution consisting of

3.17 wt% (0.18 M) ﬁ'uctose and 5.3% (0.85 M) potassium hydroxide. The high level of

potassium hydroxide was necessary to promote the desired base-catalyzed reactions of

fructose (e. g. retro-aldol condensation) with improved yields, with respect to lower base

75

 

 

ml

loadings. Since ﬁ'uctose could produce two equivalents of organic acid, super-
stoichiometric loadings of base was necessary to maintain high pH levels throughout the
reaction. In addition, higher pH levels have been shown to increase the yield of C3
products.77

Variations of hydrogen pressure, initial base concentration, feedstock sugar, and
other parameters were observed to determine the effect on product distribution. The

studies are summarized below.

5.2.1 — Effect of catalyst

Three traditional hydrogenation catalysts were used to investigate their use in the
combined reaction. 5 wt% Ru/C (Aldrich), 5 wt% Pt/C (Aldrich), and 5 wt% Pd/C
(Aldrich) were selected for the study. Loading was 0.1 grams (dry basis) for each
catalyst, which reacted with 100 grams of solution containing 0.18 M fructose and
0.85 M KOH. Table 5-1 outlines the overall C3 production for each metal after 6 hours

of reaction and compares the extent of alkaline degradation to hydrogenation.

Table 5-1: Overall C3 production and reaction comparison — Effect of catalyst metal

Conditions - 40°C, 100 psig, 0.18 M fructose, 0.85 M KOH, 0.1g catalyst

 

0
Exp Metal Conv. Hyd. lso. Deg. 3:: Max 42:: yiiid

146 Ru/C 95% 0.15 0.11 0.69 1.00 1.38 72 0.94
138 Pt/C 96% 0.23 0.10 0.62 0.87 1.24 70 0.84
139 Pd/C 94% 0.16 0.12 0.67 1.02 1.34 76 0.96

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

The ﬁnal product distributions were remarkably similar between the three active
metals used in the experiments. Platinum showed a higher preference toward sugar

hydrogenation than palladium and ruthenium, which reduced the overall C3 yield for the

76

metal. Ruthenium and palladium, however, looked essentially identical. Further

examination of the C3 product distribution in Table 5-2 resolved the decision for the

choice of metal.

Table 5-2: C3 product distribution of Ru/C and Pd/C experiments

Conditions - 40°C, 100 psig, 0.18 M ﬁ’uctose, 0.85 M KOH, 0.1g catalyst

 

 

 

PG GO LA GlyA C3 C3
EXP Catalyst Sel. Sel. Sel. Sel. sel. yield
146 Ru/C 0.25 0.19 0.51 0.04 1.00 0.94
139 Pd/C 0.15 0.16 0.67 0.04 1.02 0.96

 

 

 

 

 

 

 

 

 

 

Table 5-2 shows that the ruthenium catalyst promoted the hydrogenation of
intermediates toward propylene glycol and glycerol with higher selectivity than
palladium. The polyol products are preferred over organic acids because of higher sale
value and because acids form potassium salts under these reaction conditions and must be
precipitated out of solution and sufﬁciently acidiﬁed to form the free acid for further
processing. This series of separations would add signiﬁcant cost to the process. Thus,

ruthenium on carbon was chosen for the remainder of the parametric study.

5.2.2 — Effect of temperature

19-24 69. 77

Papers regarding hydrogenation and alkaline degradation show signiﬁcant
changes in the individual reaction rates and product yields when temperature changes.
For simultaneous alkaline degradation and hydrogenation, it was not known if the
reactions would have corresponding changes in conversion or if one would dominate the

conversion. Table 5-3 shows product distributions from three runs at 30, 40, and 50°C

77

for comparison. Reaction times were 6 hours, 6 hours, and 3 hours for 30, 40, and 50°C

respectively.

Table 5-3: C3 and C6 production and reaction comparison — Effect of temperature

Conditions - 100 psig, 0.18 M ﬁuctose, 0.85 M KOH, 0.1g 5% Ru/C catalyst

 

 

 

 

 

Exp T333? Conv. Hyd. Iso. Deg. s21. MC? :25: yiiid
143 30 78% 0.43 0.14 0.22 0.30 0.44 68 0.23
146 40 95% 0.15 0.11 0.69 1.00 1.38 72 0.94
153 50 99% 0.08 0.02 0.89 1.10 - 1.78 62 1.09

 

 

 

 

 

 

 

 

 

 

 

The rate of alkaline degradation was enhanced as reaction temperatures increased
from 30 to 50°C with respect to hydrogenation. The sugar cleavage reactions consisted
of 89% of the conversion at 50°C while it was only 22% at 30°C. Raising the
temperature provided a means to limit sugar hydrogenation and potentially improve C3
yields by promoting the degradation reactions. However, the conversion of the ﬁuctose
to C3 products was not as efficient at 50°C, as seen by the ‘% of max’ column. Only 62%
of the degraded sugar was converted to C3 products compared to 70% at 30 and 40°C. At
50°C, pathways toward glyceraldehyde were promoted (i.e. degradation), as well as
condensation reactions toward undesired products, which caused the ‘% of max’ to drop.

Table 5-4 provides a comparison of C3 product distributions with different
temperatures. Though it increased conversion, the increase in temperature also promoted

lactic acid formation over the desired polyols, propylene glycol and glycerol.

78

Table 5-4: C3 product distribution — Effect of temperature
Conditions - 100 psig, 0.18 M ﬁ'uctose, 0.85 M KOH, 0.1g 5% Ru/C catalyst

 

 

 

 

 

 

 

 

 

 

 

Exp Temp PG GO LA C3 C3
( C) Sel. Sel. Sel. sel. yield
143 30 0.03 0.22 0.05 0.30 0.24
146 40 0.25 0.19 0.51 1.00 0.94
153 50 0.24 0.09 0.70 1.10 1.09

 

 

5.2.3 - Effect of hydrogen pressure

Another factor affecting the competition between alkaline degradation and

hydrogenation was the amount of hydrogen present over the course of the experiment.

Alkaline degradation of the feedstock sugar occurs regardless of hydrogen pressure

because it is a solution-based chemistry. However, when higher concentrations of

hydrogen are present, hydrogenation becomes more prominent and affects the product

distribution. A range of ﬁve hydrogen pressures, from 0 to 200 psig, was used to observe

the effects on C3 and C6 product selectivity at 40°C.

Table 5-5: Selectivity of alkaline degradation with hydrogenation — H2 pressure effect
Conditions — 40°C, 0.18 M ﬁuctose, 0.85 M KOH, 0.1g 5% Ru/C catalyst

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Total P* C3 Max % of C3
Exp ( psi 9) Conv. Hyd. lso. Deg. s el. 03 max yield
147 25 93% 0.13 0.13 0.66 0.93 1.32 70 0.87
149 50 93% 0.14 0.08 0.71 0.95 1.42 67 0.88
146 100 95% 0.15 0.11 0.69 0.99 1.38 72 0.94
140 200 96% 0.37 0.07 0.52 0.81 1.04 78 0.78

 

* Reaction gas was essentially hydrogen since the vapor pressure of water at 40°C is 1.1 psi.

The experiments yielded somewhat surprising results regarding hydrogenation

activity. Hydrogen pressures of 25 to 100 psig showed little change in the amount of

79

 

sorbitol and mannitol formed. However, raising the pressure to 200 psig more than
doubled the selectivity and yield to sugar alcohols. At the lower pressures,
hydrogenation appeared to be 0th order with respect to hydrogen, while at 200 psig, the
rise in sugar alcohol production coincided with the rise in pressure. Figure 5-1 and Table
5-5 illustrate these results. This effect is partially attributed to uncertainties with
concentration measurements (i 3%) but is not readily explained.

Distributions of the cleaved hydrogenation products of ethylene glycol, propylene
glycol, and glycerol show a similar effect with the lower pressure experiments versus the
200 psig experiment, as seen in Table 5-6. Concentration versus time data of propylene
glycol and glycerol also conﬁrms the sequence of intermediate formation in the alkaline
degradation. From de Bruijn et al.77, two moles of glyceraldehyde form from fructose via
the retro-aldol condensation. Glyceraldehyde then undergoes a B-elimination of water to
form pyruvaldehyde. Pyruvaldehyde then hydrogenates to propylene glycol or forms
lactic acid via a base-catalyzed isomerization. Thus, glycerol, from glyceraldehyde,
should form before propylene glycol forms from pyruvaldehyde. This effect is illustrated

in Figure 5-2 at 50 psig of hydrogen pressure.

Table 5-6: Selectivity of alkaline degradation with hydrogenation -— H2 pressure effect
Conditions — 40°C, 0.18 M fructose, 0.85 M KOH, 0.1g 5% Ru/C catalyst

 

Total P EG PG GO LA

 

 

Exp (psig) sel. Sel. Sel. Sel.
147 25 0.03 0.19 0.10 0.56
149 50 0.04 0.21 0.13 0.53

 

146 100 0.05 0.25 0.19 0.51
140 200 0.05 0.18 0.33 0.30

 

 

 

 

 

 

 

 

 

80

3&8 03m .8 9 .o .mom 2 8.0 .3888 2 w 2 .o .009. I 88680

38mm cowoaczm - 08E. .m> A033 688$ 28 A08 LSEBm mo 88.5.8080 ”—-m 9.sz

9808-02.41
988705....5
9.8 8.02.191
98 mm-o_2|.l
98 828-9-
98878.3...
98 8.8.10:

98 mm - omlol

3.5 «5:.

 

 

 

 

 

 

 

 

 

 

 

 

................. U
m. 1. 1 .w w. \a 8
\.%\ 0.0
- . -18. .
w. ....... w ....... .m. 1111111 a. ....... my o n

 

 

Qw

(116) uonenuaouoo

81

oom

Hmbﬁmo 03m o\om 9 .o FEOM 2 3.0 .3803 2 w H .o 609‘ I £85680
Soho cowoaam - 08E. .m> AOOV 3.6wa 98 A03 Bobw 0:23on we 858E880 "N6 9.sz

2:5 9.5.
com own 09 ow? oo o

 

— p - f — 0.0

\.

 

“'3.
o

 

0.
‘—

 

 

(1/6) uonenuaouoo

 

\. \\

 

 

W\ 9mg om . 00 IT 9mg om . On. lml

L0.
0!

 

 

o.m

82

Further, from Table 5-6 and Figure 5-2, more PG was produced than glycerol by
the end of the reaction. This suggests that as more glyceraldehyde becomes available, the
rate of conversion to pyruvaldehyde is faster than the hydrogenation of glyceraldehyde to
glycerol. Thus, more pyruvaldehyde becomes available for hydrogenation to PG or
rearrangement to lactic acid. But, as hydrogen pressure is increased, this difference in
activity decreases. More glycerol is formed, which limits the extent of conversion of the
glyceraldehyde-pyruvaldehyde pathway and decreases the amount of pyruvaldehyde
available to form PG. This effect is evident from the crossover between PG and GO
selectivity in Table 5-6 with increasing pressure and greater selectivity for glycerol

relative to PG at a total pressure of 200 psig.

5.2.4 - Effect of catalyst loading

Catalyst loading was another parameter adjusted to observe the hydrogenation—
alkaline degradation competition. Additional amounts of catalyst should lead to
increased hydrogenation activity as more catalytic sites are present for materials to react.
The purpose of this set of runs was to maximize C3 yield by changing the amount of
catalyst at constant temperature (40°C), pressure (100 psig total), initial fructose
concentration (0.18 M), and initial KOH concentration (0.85 M). Four experiments, with
loadings of 0.0 to 0.20 grams of catalyst, were used. Table 5-7 displays the overall

product distributions from this set.

83

Table 5-7: Comparison of catalyst loading for alkaline degradation with hydrogenation
Conditions — 40°C, 0.18 M fructose, 0.85 M KOH, 100 psig H2, 5% Ru/C catalyst

 

 

 

 

 

 

Catalyst C3 Max % of C3
EXP loading Conv. Hyd. lso. Deg. sel. C3 max yield
90* 0.00g 87% 0.00 0.16 0.71 0.90 1.42 65 0.78
115 0.05g 94% 0.14 0.10 0.70 0.97 1.40 69 0.91
146 0.10g 95% 0.15 0.11 0.69 0.99 1.38 72 0.94
152 0.20g 96% 0.25 0.08 0.63 0.89 1.26 71 0.85

 

 

 

 

 

 

 

 

 

 

 

* Feedstock for experiment 90 was 0.18 M glucose instead of fructose

Higher catalyst loading showed higher activity toward hydrogenation, as
expected, but the effect was not as dramatic as that of the rise in hydrogen pressure. This
suggests that the surface concentration of hydrogen is more important than the number of
active sites available for hydrogenation

This set of experiments showed 0th order behavior with respect to catalyst loading,
similar to the hydrogen effect, as illustrated in Figure 5-3. At low catalyst loadings (0.05
and 0.10 gram) virtually no change was observed in sorbitol and mannitol production. At
the 0.20 gram loading, a signiﬁcant increase in sugar alcohol production was observed.

Table 5-8 provides the C3 selectivity data from these runs. Changes in C3
selectivity were also similar to those observed in the hydrogen effect studies. Preference
toward glycerol grew as the amount of catalyst increased. Acid formation was also
inhibited with the increase in active metal available indicating that the extent of

hydrogenation of C3 intermediates increases with increased catalyst loading.

84

80.0 - 02 Iol

83.8 :q:

groomimn

80.0 - owlol

“@880 95m o\om am mag 09 ,3va 2 3.0 .8803 2 w H .o doc? I 80:36:00
@682 8&880 mo 80am - 085. .m> AOEV 3:888 98 AOmV EHESW mo 8855800 "ﬂm 95mg

8:5 2::

own oom ovm ow _. om _.

h

 

 

 

 

- od

 

 

 

oe

- 0N

 

 

 

 

3 8
('1/6) uogtenuaouoo

0.0

 

 

ow

85

Table 5-8: Selectivity of EG, PG, GO, and LA - effect of catalyst loading
Conditions — 40°C, 0.18 M fructose, 0.85 M KOH, 100 psig H2, 5% Ru/C catalyst

 

 

 

 

 

 

 

 

Exp Catalyst EG PG GO LA

loading sel. Sel. Sel. Sel.
90* 0.00g 0.00 0.00 0.00 0.90
115 0.05g 0.04 0.16 0.16 0.65
146 0.10g 0.05 0.25 0.19 0.51
152 0.20g 0.06 0.29 0.33 0.24

 

 

 

 

 

* Feedstock for experiment 90 was 0.18 M glucose instead of fructose

5.2.5 — Effect of initial base concentration

One potential problem with the alkaline degradation concept is the need for
excessive amounts of base. Potassium hydroxide is relatively inexpensive, but the
addition of another raw material cuts into potential proﬁt margins. In addition,
downstream neutralization would be necessary to extract acidic products. Corrosion of
the metal reactor becomes an issue as well at the high pH levels required.

High pH levels are important, however, in order to maintain selectivity to C3
products at low temperature and hydrogen pressure. The retro-aldol condensation
reaction is base-catalyzed and has the potential for high selectivity to C3 products (> 75%
of theoretical) with monosaccharide feedstock. Thus, the thrust of this set of experiments
was to determine how the product distribution changes with different loadings of
potassium hydroxide. Table 5-9 outlines the overall C3 and C6 selectivity.

Hydrogenation activity rose dramatically as the amount of base in the feedstock
decreased. Selectivity to the sugar alcohols climbed from 0.15 to 0.79 mole of sugar
alcohol produced per mole of fructose reacted when the base concentration was cut by

80%. The amount of base was not sufficient to favor cleavage of the sugar over

86

hydrogenation. However, the 0.38 M loading demonstrated that feedstock that does

undergo degradation forms C3 materials with essentially the same preference as the

higher loading, as seen in the ‘% of max’ column in Table 5—8. Thus, at lower base

loadings, higher C3 yields may be possible, but with 0.1 grams of catalyst and 100 psig of

H2 present, hydrogenation was too fast to realize them.

Table 5-9: C3 and C6 selectivity of reaction with varying base concentration

Conditions — 40°C, 0.18 M fructose, 100 psig H2, 0.1g 5% Ru/C catalyst

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Exp :3": Conv. Hyd. lso. Deg. s21. Max 32:: yiceld
146 0.85 M 95% 0.15 0.11 0.69 0.99 1.38 72 0.94
116 0.38 M 92% 0.50 0.11 0.30 0.41 0.60 68 0.38
117 0.17 M 97% 0.79 0.05 0.13 0.12 0.26 46 0.12

 

 

Table 5-10 lists the C3 product distribution from the series of experiments at
different base loadings. The results show that alkaline degradation slows down

signiﬁcantly with reduced levels of base. This is consistent with observations by de

172

Bruijn et a and was conﬁrmed by the lack of propylene glycol and lactic acid formed

at the 0.17 M loading.

Table 5-10: Selectivity of EG, PG, GO, and LA — effect of initial base concentration
Conditions — 40°C, 0.18 M fructose, 100 psig H2, 0.1g 5% Ru/C catalyst

 

 

 

 

 

Exp KOH EG PG GO LA

Conc. sel. Sel. Sel. Sel.
146 0.85 M 0.05 0.25 0.19 0.51
116 0.38 M 0.04 0.09 0.27 0.04
117 0.17 M 0.01 0.01 0.10 0.00

 

 

 

 

 

 

 

87

5.2.6 - Effect of initial sugar concentration

Results from literature77 and earlier experiments with alkaline degradation (see
Section 4.2) showed that low sugar concentrations led to high C3 selectivity. Higher C3
selectivity meant promoted degradation pathways and potential to realize higher C3 yields
when hydrogenation was added. Three levels of sugar concentration were used to see if
higher yields would be realized in the presence of hydrogen and catalyst. Table 5-11

displays the overall selectivity from the reaction.

Table 5-11: C3 and C6 selectivity of reaction with varying sugar concentration
Conditions — 40°C, 0.85 M KOH, 100 psig H2, 0.1g 5% Ru/C catalyst

 

 

 

 

lnit.
Exp gugar Conv. Hyd. lso. Deg. 52‘; Max 1:12;: ygjd
one.
119 0.06 M 97% 0.57 0.04 0.37 0.50 0.74 68 0.37
144 0.12 M 96% 0.22 0.09 0.64 1.09 1.28 85 1.05
146 o. 18 M 95% o. 15 0.11 0.69 0.99 1.38 72 0.94

 

 

 

 

 

 

 

 

 

 

 

 

Decreasing the sugar concentration from 0.18 M to 0.12 M showed an
improvement in C3 selectivity. However, further reduction to the 0.06 M loading
reversed the trend with a reduced efﬁciency to C3 products. The improvement gained by
lowering the sugar content was offset by the increase in hydrogenation activity to the
sugar alcohols. With a lower proportion of feedstock to catalytic sites, the hydrogenation
reaction was promoted and decreased the potential C3 yield.

One other interesting observation was the constant amount of sugar alcohol
produced between the three concentrations. Hydrogenation selectivity of 0.57, 0.22, and
0.15 for the feedstock concentrations of 0.06 M, 0.12 M, and 0.18 M, respectively,

yielded 0.0034, 0.0026, and 0.0024 moles of sorbitol and mannitol combined. This

88

suggested 0‘11 order hydrogenation kinetics with respect to sugar concentration. Thus,
with higher sugar loading, more of the feedstock should undergo degradation pathways to
potentially form C3 products. However, too much sugar led to higher levels of organic
acid formed, which lowered the solution pH and reduced C3 selectivity via retro-aldol

condensation.

5.2.7 - Effect of feedstock sugar

The ﬁnal group of experiments involved the speciﬁc sugar used as feedstock for
the reaction. Alkaline degradation of fructose, glucose, and mannose showed that
fructose had the highest activity of the three. Table 5-12 compares the product

distributions from the three sugars.

Table 5-12: C3 and C6 selectivity product distribution — effect of feedstock sugar
Conditions — 40°C, 0.18 M sugar, 0.85 M KOH, 100 psig H2, 0.1g 5% Ru/C catalyst

 

Feed C3 Max %of C3
sugar sel. C; max yield

146 Fructose 95% 0.15 0.11 0.69 0.99 1.38 72 0.94

141 Glucose 90% 0.28 0.10 0.52 0.76 1.04 73 0.68
142 Mannose 75% 0.21 0.18 0.36 0.59 0.72 82 0.44

Exp Conv. Hyd. Iso. Deg.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

In the presence of hydrogenation catalyst, fructose showed the highest rate of
conversion to products. Fructose also showed the highest activity toward degradation
suggesting that it was the main intermediate leading into the sugar cleavage reactions.
Mannose demonstrated a surprisingly high proportion of degraded material converting to
C3 products. However, the slower rate of reaction to C3 products (i.e. ‘Deg.’) in

comparison with fructose made it less appealing. The relatively high selectivity to

89

mannitol (hydrogenation product) also reduced the C3 selectivity and yield signiﬁcantly.

Table 5-13 compares the C3 product distribution from the three feedstock sugars.

Table 5-13: Selectivity of EG, PG, GO, and LA — effect of feedstock sugar
Conditions — 40°C, 0.18 M sugar, 0.85 M KOH, 100 psig H2, 0.1g 5% Ru/C catalyst

 

Feedstock EG PG GO LA
sugar sel. Sel. Sel. Sel.

146 Fructose 0.05 0.25 0.19 0.51

141 Glucose 0.06 0.21 0.30 0.25
142 Mannose 0.05 0.15 0.20 0.24.

Exp

 

 

 

 

 

 

 

 

 

 

 

Both glucose and mannose showed higher selectivity to glycerol than propylene
glycol. Fructose, on the other hand, had higher selectivity to propylene glycol than
glycerol. In addition, selectivity to lactic acid was signiﬁcantly higher for fructose. This
suggested that glucose and mannose form glyceraldehyde at a slower rate than fructose,
presumably by isomerizing to fructose before undergoing the retro-aldol condensation to
glyceraldehyde. The regulated reaction had the beneﬁt of producing a higher fraction of
C3 products in polyol form (67% for glucose and 59% for mannose) versus the direct
route with fructose (46% of C3’s to PG and GO). But, the extended time to form the
necessary intermediates allowed more time for hydrogenation of the sugar feedstock, thus

lowering their overall attractiveness as feedstocks.

5.3 — Summary and significance of results
The parametric studies described in this chapter serve to better deﬁne the reaction
environment of the alkaline degradation — hydrogenation reaction. Through these

experiments, the concept of “capturing” active C3 aldehydic intermediates to form the

90

desired products of propylene glycol and glycerol was proven and carried out at
temperatures and pressures signiﬁcantly lower than usual hydrogenolysis conditions
(40°C and 100 psig H2 with sugar feedstock versus 230°C and 1200 psig H2 with sugar
alcohol feedstock).

Each of the parameters studied played an important role in developing the product
distribution at the end of the reaction. Temperature showed the largest effect. Alkaline
degradation pathways were enhanced a signiﬁcant amount with just a 10°C increase in
the process temperature. The loading of potassium hydroxide had a large effect as well.
Superstoichiometn'c amounts of base with respect to the sugar feedstock were necessary
to maintain good selectivity to C3 products and to continue alkaline degradation
chemistry. Without sufﬁciently high pH, the degradation reactions shut down and
allowed feedstock hydrogenation to take over.

Hydrogen pressure and catalyst loading were also crucial in determining product
distributions. Arr overload of either diminished C3 yields signiﬁcantly by hydrogenating
the feedstock sugar to its corresponding sugar alcohol. These two studies also conﬁrmed
suggested pathways for degradation chemistry simply by observing the concentration
versus time data. Taking the pathways already developed, product distributions were
fairly easy to explain, even as the reactions progressed.

The study that involved the different feedstock sugars also highlighted a potential
strategy for ﬁrrther C3 yield improvement. Preference to C3 products is based mainly on
the conversion of fructose into glyceraldehyde. From that intermediate, several different
reactions can occur, diminishing selectivity to desired products. However, if

glyceraldehyde is hydrogenated immediately, those pathways are eliminated. Glucose

91

and mannose demonstrated this concept by maintaining high efﬁciency to C3 products
with a preference toward glycerol and propylene glycol. Hydrogenation of the sugar
must be avoided, however, to have desirable C3 yields.

Experiment 144 (40°C, 100 psig H2, 0.12 M ﬁuctose, 0.85 M KOH, 6 hour
reaction time) showed the most promising results from this group of studies. Of the
reacted feedstock that followed the alkaline degradation pathways, 85% was converted to
the desired C3 products of propylene glycol, glycerol, lactic acid, and glyceric acid (‘% of
max). From Table 5-14, 64% of the fructose feedstock underwent alkaline degradation
(‘Deg’). Potentially, this could produce 1.28 moles of desired C3 per mole of ﬁuctose
reacted (‘Max C3’). A selectivity of 1.09 moles of C3 per mole of fructose reacted was
observed (‘C3 sel.’), which led to a ‘% of max’ of 85%. Experiment 152 (50°C, 100 psig
H2, 0.18 M fructose, 0.85 M KOH, 6 hour reaction time) has a higher C3 yield than
Experiment 144, however, the efficiency to C3 products is much lower (62% vs. 85%).

In addition, Experiment 144 has the added potential of recycling to further improve the
overall C3 yield. The hydrogenation products of sorbitol and mannitol can be converted
to glycerol and propylene glycol with yields of ~70% of theoretical. This would increase
the overall C3 yield by 0.3 (70% * 0.22 moles sugar alcohol per mole fructose * 2 moles
of C3 per mole of C6) to 1.35 moles of C3 per mole of fructose fed. Further, the
isomerized sugars can further react to form the same desired C3 products to raise the ﬁnal

yield to 1.4 moles C3 per mole of fructose fed, or 70% of theoretical.

92

Table 5-14: Comparison of Experiments 144 and 153

 

 

 

 

 

 

 

 

 

 

 

 

 

lnit.
03 Max %of C3
Exp gugar Conv. Hyd. lso. Deg. sel. C3 max yield
one.
144 0.12M 96% 0.22 0.09 0.64 1.09 1.28 85 1.05
153 0.18M 99% 0.08 0.02 0.89 1.10 1.78 62 1.09

 

 

The overall C3 yield of the process was diminished due to the use of a
conventional catalyst that indiscriminately hydrogenated both sugar feedstock and
degradation intermediates. Thus, sugar feedstock that could have formed the desired C3
products instead, was converted into mannitol or sorbitol. A catalyst that preferentially
hydrogenates glyceraldehyde and pyruvaldehyde instead of feedstock sugar would be of
great beneﬁt to this system. Candidate catalysts for this selective hydrogenation are
discussed in Chapter 7.

In addition to observing the effects of various parameters on the overall product
distribution of the reaction, a kinetic model of the reaction system was developed.

Derivation and comparison of the model to experimental results is discussed in Chapter 6.

93

Chapter 6 — A kinetic model for simultaneous alkaline
degradation and hydrogenation

6.1 - Overview

This chapter outlines efforts to describe the alkaline degradation — hydrogenation
reaction rates with a mathematical model. Several papers describe pathways,
mechanisms, and kinetics for sugar hydrogenation”23 and for alkaline degradation69'76.
The work presented here seeks to combine the two reactions into a general model. The
next sections summarize mass transfer calculations of reactant transport through the
reaction, describe the derivation of the model, outline assumptions made to simplify the
network of reactions, discuss the method used for calculating concentration versus time,

and compare the ﬁnal model and data collected.

6.2 - Mass transfer calculations

Mass transfer rates of the reactants were calculated to ensure that the derived
kinetic model was based on surface rates of reaction in the catalyst instead of mass
transport of reactants between phases. Several correlations from the literature were used
to calculate minimum stirring rates for catalyst suspension and mass transfer coefﬁcients

for these experiments and are described below.

6.2.1 — Catalyst suspension
To ensure proper mixing of catalyst within reactor, it was necessary to make sure
that the catalyst was completely suspended in the solution as the reaction progressed.

This maximized the amount mass transfer area of the catalyst for transport of hydrogen

94

and unsaturated intermediates to the metal sites. Zwietering90 proposed a correlation for

minimum stirring speed necessary to completely suspend catalyst within a bafﬂed vessel:

— B'dp «3.130454% _pL)o.4s -w°'l3

 

Nmin p355 . dpss (1)
Where:

Nmin = minimum stirring speed for catalyst suspension [rotations / sec]

[3 = constant determined by impeller geometry

w‘ = catalyst loading [grams catalyst per 100 grams solution]

clp = particle diameter [m]

m = liquid viscosity [kg / m°s]

g = gravitation constant [9.8066 m / s2]

pp = particle density [kg / m3]

p1, = liquid density [kg / m3]

d3 = impeller diameter [m]

Reactions used 0.1 grams of catalyst per 100 mL of solution, which required a
minimum stirring speed of approximately 400 rpm. An agitation rate of 1000 rpm was

used to ensure complete suspension of the catalyst.

6.2.2 - Three phase mass transfer
For hydrogenation reactions to occur within this system, unsaturated reactants
from the liquid phase and hydrogen from the gas phase must undergo the following steps

to reach the active metal sites in the catalyst in order to react:

 

9° Zwietering, T. H., Chem. Eng. Sel., 1958, 8, 244

95

1. Transport of hydrogen from the gas phase to the gas-liquid interface

2. Transport of hydrogen from the gas-liquid interface to the bulk liquid

3. Transport of hydrogen and unsaturated reactants (e. g. glucose, fructose,
glyceraldehyde) ﬁ'om the liquid phase to the catalyst surface through the solid-
liquid interface

4. lntraparticle diﬂusion of hydrogen and reactants through the catalyst pores
5. Adsorption of reactants on the active sites of the catalyst

6. Surface reaction (e. g. glucose + H2 —> sorbitol)

7. Desorption of products

Figure 6-1 illustrates the phases of the reaction and relative concentration of

reactants from the gas and liquid phases.

 

 
 
 

 

 

 

 

 

1» E E .9 m _E_ 0
8 E E g 8 ~= E
.C (I) 'O = .C :2 Q.
0' W '5 x a 3
A g (D .g 3 2' g
o -J m E 8
8
/
c / .-.-:-:«:.: :-:-:~ -----
.9 .-:3:
E . .
E L1qu1d ’
8 Phase .....
S reactants ...
0
\
H2 gas \
phase \
t

 

 

 

 

 

 

 

 

Radial position

Figure 6-1: Mass transfer in a three phase catalytic reaction

96

6.2.3 - Reaction rate and mass transfer

Hydrogen and the liquid phase reactants pass through interfaces between phases
to react. The rate of transfer of the compounds between phases can affect the overall
reaction rate depending on the relative rates of mass transfer for each reactant. For

calculations, liquid phase reactants are assumed to have the characteristics of fructose.

Gas-liquid transfer:

-Ro-H2 ' L = kLa (CH2, G - CH2, L) (2)

Liquid-solid transfer:

'RG-HZ ' L = ksm a (CH2, L - CH2, 8) (3)

‘RGA ' L = ks,Aa (CA, L - CA, s) (4)
Where:

-RG = observed reaction rate of the species [kmol / m3 - s]

L = ratio of catalyst surface area per unit volume ﬂuid [m2 / m3]

kLa = gas-liquid mass transfer [1 / m- s]

ksm a = liquid-solid mass transfer [1 /m - 5]

km a = liquid-solid mass transfer of liquid phase reactants [1 / m - s]
C1121, = concentration of hydrogen in the liquid phase [mol / L]

CA, L = concentration of liquid phase reactants [mol / L]

C1123 = concentration of hydrogen in the catalyst [mol / L]

CA, s = concentration of liquid phase reactants in the catalyst [mol / L]

97

In these equations, the driving force for mass transfer is the difference in
concentration between phases, which can be measured directly or calculated (e.g.
hydrogen concentration in the liquid phase with Henry’s Law). The mass transfer
coefﬁcients from the expressions can be determined experimentally or calculated based
on correlations in the literature. With values for reactant concentration and mass transfer
coefficients, the observed reaction rate may be calculated.

In addition, the maximum amount of mass transfer for each reactant may be

determined by setting the driving force equal to the higher of the two concentrations:

Maximum G-L mass transfer rate = kLa (C1320 — CH2, L) (5)
Maximum L-S mass transfer rate = ks,A a (CA, L) (6)
ksm a (CH2, L) (7)

The maximum rate of mass transfer for each reactant is compared to the observed
reaction rate. Ifthe observed reaction rate is signiﬁcantly less than the maximum rate of

mass transfer then the reaction is not mass transfer limited.

6.2.4 — Gas-liquid mass transfer coefﬁcient

For these reactions, the gas phase comprised of hydrogen and water vapor.
However, with a process temperature of 40°C, the partial pressure of water was only 1.1
psi. This accounted for less than 3% of the gas phase at the lowest reaction pressure of
the experiments considered (Experiment 147, 25 psig) and was subsequently assumed to
be hydrogen for calculation purposes.

To estimate the mass transfer of hydrogen from the gas phase to the liquid phase,

concentrations of hydrogen in both phases and the value for the mass transfer coefﬁcient

98

were needed. The gas phase concentration was calculated using the ideal gas law and the

liquid phase concentration was calculated using Henry’s Law”:

P = Hx (8)
Where:

P = partial pressure of solute (H2) in the gas phase

H = Henry’s Law constant [atm / (mole solute / mole solution)]

x = mole ﬁ'action of solute (H2) in liquid phase

Henry’s law constant for hydrogen at 40°C is 75,100 atm / (mole H2 / mole
water)”, which corresponds to a solubility of 1.1 x 10'5 moles / cm’. For calculation
Purposes, the solubility of hydrogen was assumed to be the same as that of water.

The gas-liquid mass transfer coefﬁcient (kLa) was calculated using correlations
ﬁ‘Om Yagi and Yoshida93 and Bern et al.94 The correlation ﬁom Yagi and Yoshida was

Used to adequately predict CO2 adsorption for a glycerol-water system at 30°C and is as

fCDllows:

Sh =£ﬁ=006[m]lﬁ [Siz—Nz—JO-w [ll—LTJTﬂTﬁYEﬂLJMZ (9)
DA 11L g PLDA ST “6

Where:
Sh = Sherwood number [dimensionless]
kLa = gas-liquid mass transfer [1 / m - s]

 

 

91 Felder, RM; Rousseau, RW., Elementary Principles of Chemical Processes, 2ad Ed., p. 248, Wiley,
gew York, 1986
Y in RH; Green, D.W., Perry ’5 Chemical Engineers’ Handbook, 7‘h Ed, p. 2-126, McGraw-Hill, New
0 1997
93
,4 Yagi, H.; Yoshida, E, Ind. Eng. Chem. Proc. Des. Dev., 1975, 14,488
Bern, L.; Lidefelt, 1.0.; Schoon, N.H., J. Am. Chem. See, 1976, 53, 463

99

.9-
ll

impeller diameter [m]

D1, = diffusion rate of species A [m / 32]

N = stirring speed [8"]

pL = liquid density [kg / m3]

112 = liquid viscosity [kg / m°s]

g = gravitational constant [9.8066 m / s2]
us = gas velocity [m / 5]

ST = surface tension [N / m]

The correlation developed by Bern (Equation 9) was based on a slurry reactor
used to hydrogenate oil in a mass transfer controlled regime and sufficiently predicted

gas-liquid mass transfer for 30 and 500 liter reaction vessels.

0.011.N1.16 «1:979 -u0'32

 

kLa = VI13.0521 g (10)
kL = liquid ﬁlm mass transfer coefficient [m / s]

a = gas-liquid interfacial area per unit volume of reactor [cm2 / cm’]
d1 = impeller diameter [cm]

N = stirring speed [3"]

ug = gas velocity [cm / s]

pL = liquid density [g / cm’]

VL = volume of the solution [cm’]

100

6.2.5 - Liquid-solid mass transfer coefﬁcient
Transfer of hydrogen and the liquid phase reactants from the bulk liquid phase to
the catalyst surface is necessary for the reaction to occur. The liquid-solid mass transfer

coefﬁcient, ks, is determined for all reactants using the Boon-Long equation.95

0.283 0.173 -0.011 _
ksdp O 046 andpd-FN pigd: w'VL dT 0019[ ”L )0461
= . ‘ — —— — —— 11
DA 1’'L #12. 9: dp PLDA ( )

Where:

 

 

k5 = liquid-solid mass transfer coefﬁcient [m / 3]
(1p = particle diameter [m]

DA = diffusion rate of species A [m2/ s]

dT = reactor diameter [m]

N = stirring speed [3"]

Lu = liquid viscosity [kg / m's]

pl, = liquid density [kg / m3]

g = gravitational constant [9.8066 m/ 32]

w’ = catalyst loading [kg / 100 kg solution]

VL = liquid volume [m3]

The mass transfer area for the liquid-solid interface is calculated using properties

of the catalyst and the conditions of the experiment:

6 f
a: WPL

d (12)

ppp

 

9’ Boon-Long, S.; Laguerie, C.; Couderc, J.P., Chem. Eng. Sci., 1978, 33, 813

101

Where:

ratio of catalyst surface area to volume of ﬂuid [m2 / m3]

a =
w’ = catalyst loading [g catalyst / 100g solution]
pL = liquid density [kg / m3]

dp = particle diameter [m]

pp = particle density [kg / m3]

With values for k. and a, the liquid-solid mass transfer rate may be determined.

6.2.6 — lntraparticle mass transfer
The ﬁnal step for hydrogenation is the diffusion of the reactants through the pores
of the catalyst to the active metal sites. The Weiss-Prater’6 criterion was used to compare

the rate of mass transfer within the catalyst and the reaction rate:

 

2

71(1)2 = 5:: (13)
Where:

11(1) 2 = observable modulus

r = reaction rate [mol / L ° 3]

L = dp / 6 = characteristic length of the catalyst [m]

C5 = species surface concentration [mol / L]

D. = 82D. = effective diffusivity [m2 / s]

8 = support porosity [0.6 for the catalyst]

 

96 Weiss, P.B.; Prater, C.D., Adv. Catal., 1954, 6, 143

102

The observable modulus is the ratio of the observed reaction rate to the effective
diffusion rate in the catalyst particle. Ifn<I>2 < 1, then the rate of mass transport is greater

than the reaction rate and the reaction is not mass transfer limited. Otherwise, the

reaction rate is limited by the diffusion of reactant in the catalyst.

6.2.7 — Results from mass transfer calculations

The maximum mass transfer rates for hydrogen and the liquid phase reactants
(assumed to be fructose) were calculated for the experiment with the highest rate of
hydrogenation (Experiment 140 — 40°C, 200 psig H2, 0.18 M fructose, 0.1g 5 wt% Ru/C

catalyst). Table 6-1 reports these results.

Table 6-1: Mass transfer calculations for Experiment 140

 

 

 

 

 

 

 

 

 

ECluation Description Ratenl‘::i'7i:’n§.:f5131
kLa . C112 xaétggmG-L mass transfer 6.7 x 103
““1.“
kLa - Cm 81111380115013? "anger 7.5 x 10’5
..
RA. MAX 13:12:?Eicntlgmum reaction 6.6 x 104
11¢ 2 Observed modulus for H2 95 x 10.5
ml) 2 33:55:31 modulus for 5.2 x 104

 

 

 

 

 

 

103

Mass transfer of hydrogen from the vapor phase to the liquid phase was not rate
limiting as the rate of transfer was two orders of magnitude faster than the maximum rate
of liquid-solid mass transfer. Further, the mass transfer of hydrogen and ﬁuctose was
larger than the maximum observed reaction rate. However, this difference was not large
enough to deﬁnitively state that transport of the reactants was not rate-limiting. As the
reaction progressed, however, the rate of hydrogen consumption dropped signiﬁcantly,
which removed potential mass transport limitations. Finally, the observable modulus for
both hydrogen and fructose show that the reactants diffuse readily through the catalyst
pores (i.e. 111112 << 1). Thus, the surface reaction was rate limiting and not the mass

transport of reactants.

6.3 - The kinetic model

The difﬁculty of modeling this system lies in the number of different reactions
that occur simultaneously. The alkaline degradation and hydrogenation reactions were
ﬁrst examined separately and then combined to form the complete model. Each section
summarizes the assumptions made for these reactions.

Nomenclature for rate constants will describe the direction of the reaction. For
instance, kw: is the rate constant for the isomerization of mannose to ﬁuctose. ksoADs
represents the rate constant for sorbitol adsorption. Kso is the equilibrium constant for
sorbitol adsorption. Concentrations of liquid phase compounds are described with
brackets. For example, glucose concentration is represented by [G]. Catalytic site
concentrations are designated with C and a subscript, e. g. C1132 describes the
concentration of type 2 catalytic sites occupied by hydrogen. Table 6-2 lists the

abbreviations for the compounds of interest.

104

Table 6-2: Abbreviations used in kinetic model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abbreviation Description
G Glucose
M Mannose
F Fructose
SO Sorbitol
MO Mannitol
GA Glyceraldehyde
PA Pyruvaldehyde
GA2 Glycolaldehyde
EG Ethylene glycol
PG Propylene glycol
GO Glycerol
LA Lactic acid
FA Formic acid
GLYA Glyceric acid
GLYA2 Glycolic acid
CD Condensation products
CZ Cannizzaro reaction
H2, H2 Hydrogen
S1, S1 Catalytic site — type 1
S2, 82 Catalytic site — type 2
R Reaction rate
P Pressure
t Time
k Rate constant
ADS Adsorption
DES Desorption
K Equilibrium constant —
kADS / kDES
6.3.1 - Hydrogenation

The addition of supported-metal catalyst and hydrogen add the element of
hydrogenation to the alkaline degradation pathways. Sorbitol, mannitol, glycerol,

propylene glycol, and ethylene glycol are the main products from hydrogenation.

105

Hydrogenation was modeled using a two-site Langmuir-Hinshelwood model”. On type
1 sites, reactants adsorb and react with hydrogen adsorbed on adjacent type 2 sites.
Hydrogen dissociates to adsorb atomically to the type 2 site, consistent with work from
Neuroclé’8 and sugar hydrogenation work from Wisniak et al.23. The derivation for the

adsorption of hydrogen follows:
H2+2 S: S ZH‘Sz (14)

This equation represents the reversible dissociative adsorption of hydrogen on to type 2
sites (S2) from the model. Diatomic hydrogen adsorbs on to two separate sites and then

desorbs to form the diatomic form again. In equation form:
2 2
rHZ-ADS = 1kH—DEs 'CHsz —kH—ADS 'PH2 'Cvz (15)

2
C H-S2

H

 

rHZ-ADS =kH-ADS[ -PH2 'Cilz] (16)

Where:
1'112-A1)s = rate of hydrogen adsorption
kmms = rate constant for hydrogen desorption
€11.32 = concentration of type 2 sites occupied by monatomic hydrogen

kH-ADs = rate constant for hydrogen adsorption

Pm = hydrogen pressure
CV2 = concentration of vacant type 2 sites
K; = equilibrium constant for hydrogen (kmms / kH-DES)

 

97 Fogler H.S., Elements of Chemical Reaction Engineering, 2"d Ed., p.256, Prentice-Hall, New York, 1992
9“ Pallasana, v., Neurock, M., J. Catal., 2002, 209, 289-305

106

Assuming that hydrogen adsorption and desorption are rapid, mums = 0, yielding:

 

C2.
0 = kH—ADS[ 121:2 "Pm 'Cilz] (17)
C121-$2 = KH 'PHZ 'Cilz (13)

CH-SZ =Cv2 \lKH 'PHZ (19)

Using a total site balance for type 2 sites (C12):
Crz = CV2 + CH-SZ (20)

CV2 =CT2 -CH-82 (21)
CH-S2 =(Cr2 ' CH-S2)°\lKH “P112 (22)
C12 \lKH '_PH2

C, = _
H32 1+JKH-PH2

(23)

Or in terms of fractional coverage, 9:

 

(9 = \lKHPI-IZ
H'SZ Hm (24)

Thus, the fraction of type 2 sites occupied by hydrogen was described by the hydrogen
pressure and equilibrium constant for hydrogen in this system.

The hydrogenation products of sorbitol, mannitol, glycerol, propylene glycol, and
ethylene glycol are formed from unsaturated intermediates of the reaction. Glucose
forms sorbitol, mannose forms mannitol, and ﬁuctose forms a combination of sorbitol
and mannitol when hydrogenated. Glycerol is formed from the reduction of the

intermediate glyceraldehyde, propylene glycol from pyruvaldehyde, and ethylene glycol

107

from glycolaldehyde. Each of these aldehydes is formed via alkaline degradation
pathways.

Several sugar hydrogenation papersn'25 give experimental evidence that sugar
hydrogenation proceeds via a simple mechanism: Unadsorbed sugar reacts with
disassociately chemisorbed hydrogen (i.e. monatomic) to form an adsorbed molecule of
sugar alcohol, which desorbs to complete the reaction.

To simplify the model, the aldehydic intermediates glyceraldehyde,
pyruvaldehyde, and glycolaldehyde have been assumed to follow the same mechanism.
Thus, once formed in solution, the aldehydes react with chemisorbed hydrogen to
produce their corresponding polyol and desorb to their respective product. Further, each
saturated product is taken to be an end product. Thus, no hydrogenolysis occurs to
complicate the reaction model. This is a fair assumption, given the low temperature
(40°C) considered for the work. Work by Peereboom86 in our lab has shown
dehydrogenation of saturated polyols occurring at temperatures at or below 40°C, but the
rate of reaction was so low that it can be ignored in contrast to the conditions used in
these reactions. Previous hydrogenolysis work required temperatures at approximately
200°C to cause signiﬁcant polyol hydrogenolysis. With these assumptions in mind, the

hydrogenation reactions are as follows:

G + 2 H-s2 + s. -+ 803, + 2 S2 (25)
M + 2 1132 + 31 ~ Mo-sl + 2 $2 (26)
r + 2 H62 + 81 -. 508, + 2 s2 (27)
F + 2 H-s2 + 81 -» 114081 + 2 s2 (28)

108

GA + 2 H'Sz + $1 -’ GO'S] + 2 82 (29)
PA + 2 H62 + $1 —’ PG‘S] + 2 $2 (30)
GA2+2H-S2+S1-'EG-S1+2S2 (31)

The products desorb reversibly to complete the description:

3051 s so + s. (32)
MG-sl r. MO + s, (33)
Go-sl 2 G0 + s. (34)
PGsl 2:. PG + s. (35)
EGSI s EG+ s1 (36)

Rate equations for each of the adsorbed intermediates are as follows:

SO ~S
rsosr = kG-SO[G]'C12-l-S2CV1 + kSO-ADS [1501' CV1 " —[ K 1]] + kF-SO [F1 Cfrszcvr (37)
so

MO- 8
- [ 1]] + kr-moiFl‘ Cirszcvr (3 8)

 

rMO sr = kM—MolMl' Cirszcvr + kMO—ADS[[M(1° Cvr MO

1”Gosr = kGA-GO [GAl' Cir-szcvr + kGO—ADS [M01 Cv1 — kGO-DES ‘160 '31] (39)
rPG-Sl = kPA-PG [PAl'Cil-szcvr + kPG—ADS [PGl'Cvr _kPG-DES ' [PG ' S1] (40)
rraGsr = k GA2—EG[GA2]C12~1~S2C V1 + 1‘ EG-ADS [EGl' C V1 ‘ 1‘ EG—DES ' [EG '31] (41)
Using the psuedo steady state hypothesis (PS SH)”,each of these equations is set to zero.
This assumed that the rates of adsorption intermediate formation and product desorption

were fast relative the overall reaction rate. Thus, the concentration of each intermediate

can be solved for in terms of measurable product concentrations.

 

99 Fogler H.S., Elements of Chemical Reaction Engineering, 2"d Ed., p.341, Prentice-Hall, New York, 1992

109

C2, C
[SO '51]= (kG-SO [G]+ kF-SO [F])—k—H‘§Z—v—l+ Kso [501' CV1 (42)

SO-DES
C2, C
[M0 ' Sr 1: (kM-MO [M]+ kF—MO [FD ‘kﬂ'z—EV—l + KMO [M01 Cv1 (43)
MO-D s
caecvr
[Go'sil= kGA-GolGAl'W'l" KGO[GO]-Cv1 (44)
Caecv.
[PG-s.1=k..-..1PA1-;-—+Karmic. (.5)
PG—DES
C2, c
[EG‘SI]=kGA2—EG [GAzl'Iis'z‘i+KEG [EGl'Cw (46)
EG-DES

Using a total site balance for type 1 sites, fractional coverage was derived and the

results are listed on Figure 6-2.

The rate equations for desorbed products are as follows:

130 = —kSO_ADS [so]-CVl +kSO_DES -[SO-S1] (51)
me = -kMo-ADs [MOI-CV. +kMO—DES -[M0 '81] (52)
r60 = —kGO_ADS [Mo]-CVl +kGO_DES -[Go-sl] (53)
rpe = —kpe-ADs [PGl'Cw +kPG-DEs -[PG ~81] (54)
re. =—kee-ADs[EGl-Cv. weaves 1139's.] (551

110

89:. 82388
32608 888863 mo 8389.8 Sm C>®v _ 25 mo 88 “88> .«o owegoo 888.3 com :38.— BZSQ an.» charm

Tom: 88 +781; Log 8M +33 62M :8. 8m + _ u <
H 825?

 

 

 

 

 

ﬂ mmDIOmv— mmalgv— meIOOv— mMQIO—Za— meIOmv—

- , , - 8.x
_N<£6m1~<oo_ + _<m_81§x + —<Eoo1<oe_ + 75021; $212.28. + 7:81.:— $2 813—w N0 + <

 

HS
L G

890mg 85128. 85186. $6.026. madma—
, - - w - . ,w 8.: S L
Wed 8129. + 7x A; 81s; + _<Eoo1<oo_ + 3021.:— +w>aoz12x + £81.:— LEomex NO + < O 1 L0

 

 

 

ﬁmérvmu .._1 ﬁmﬁvmu 1T ~m.80 1T ~m.OEU .+. ~m.OmU 1T ~>U H ~50

111

Substituting in expressions for the adsorbed species, the ﬁnal equations become:

rso = (kG-so[G]+kp-so[F])-®%.sz®w (56)
we = (kM-Mo[M]+kF-Mo[F])-®é.sz®w (57)
r60 = kGA_GO [GM-(931829“ (58)
rm = kpA-pG[PA]-®%1.sz®v1 (59)
rag = kGAz-EG [GA2]-®ﬁ.sz®v. I (60)

No reversible reaction was included in the model (i.e. dehydrogenation) due to the
low temperature of the reaction system. While dehydrogenation may occur at 40°C,

hydrogenation is greatly favored.

Each rate equation is ﬁrst order with respect to the reactant. Since the reaction is

in a batch system, the change in concentration versus time is as such:

 

 

 

 

 

dlitO] = (kc—so [G]+kF—so [1=])-®§1.82(v9(,1 (61)
(ills/Ito] = (k M—MO [M]+ kF_MO [F]) 831.326)“ (62)
d[:‘rtO] = kGA_GO [GA]°@ir.s2@v1 (63)
dlth] = kPA—PG [pA].@§I_82®V1 (64)
d[:5tG] = kGAz—EG [GA21'9irs29w (65)

112

These equations were taken into account with the alkaline degradation equations
and combined to form the overall model. The next section discusses the derivation of the

alkaline degradation pathways.

6.3.2. - Alkaline degradation

To begin, the feedstock sugars (glucose, fructose, and mannose) isomerize to form
the other sugars, via the Lobry de Bruyn — Alberda von Ekenstein pathwaysfnﬁ2 For
simplicity and to remain consistent with similar papers, isomerization reactions are ﬁrst

order with respect to the sugars. Their equations take this form:

rp = '(km + kF-M)[F] + kM-FlMl + koF[G] (65)
to = '(kG-F + kc-M)[G] + kM—Gl-M] + kF-G[F] (67)
rM = -(kM-F + kMONM] + kF-MlF] + ko-M[G] (63)

In addition to isomerization, ﬁ'uctose undergoes alkaline degradation to two moles of
glyceraldehyde, primarily, via base-catalyzed retro-aldol condensation. At
superstoichiometric loadings of base with respect to the feedstock sugar, retro-aldol
condensation is the preferred conversion pathway.

From kinetic modeling, de Bruijn et al.73 show that ﬁ'uctose and psicose account
for 80-90% of the acid products formed from the degradation reactions. Since psicose is
formed ﬁ'om fructose, it was eliminated from the model and acids formed from psicose
were assumed to come from fructose. Glucose and mannose accounted for the remaining
10-20% of acid products, of which glucose is responsible for nearly all of the acidic
material. Thus, the model included only isomerization and hydrogenation reactions for

mannose and included a degradation pathway for glucose to intermediates. Since glucose

113

is an aldose, it forms a C2 and C4 aldehyde during the retro-aldol reaction. The aldehydes
formed have short lifetimes at the high pH of reaction and undergo further conversions.

The potential reactions for each are discussed below.

6.3.2.1 -Aldehydic intermediates: Glycolaldehyde

Glycolaldehyde, the C2 intermediate formed from glucose, can undergo a
Cannizzaro reaction100 to form one mole of glycolic acid and one mole of ethylene
glycol. It can also be hydrogenated to form ethylene glycol. The C4 intermediate can
form threaldehyde or erythraldehyde since it has a chiral center. Again, these
intermediates can undergo the Cannizzaro reaction to form the corresponding carboxylic
acid and sugar alcohol (threonic acid and threitol or erythonic acid and erythritol) or
simply hydrogenate to the sugar alcohol. In the overall model, only the C2 compounds
were included for consideration to simplify the model and because there is more interest
in ethylene glycol than in threitol or erythritol. In addition, the reactions would have
added parameters to the model without contributing very much to the overall picture.
The reactions yielded the following equation for the kinetic model:

d[GA2]

dt :kG-GAZ[G]-kGA2-G [GAZJZ -kGA2-CZ[GA2]2 -kGA2-EG[GA2].®%I-82®Vl (69)

 

Glucose to glycolaldehyde is reversible, but requires a mole each of
glycolaldehyde and threaldehyde (a C4 intermediate). However, since the C4 compounds
were not accounted for in the model, another equivalent of [GA2] was included to
estimate the concentration of threaldehyde to produce the [GA2]2 term to estimate the

condensation of glycolaldehyde and threaldehyde to glucose. Glycerol, as noted in

 

'00 Cannizzaro, S.,Ann., 1853, 88, 129

114

section 6.2. 1, was assumed to be an end-product and thus, not react ﬁsrther. Glyceric acid

was considered an end product as well once it formed the salt potassium glycerate.

6.3.2.2 - Aldehydic intermediates: Glyceraldehyde
The other intermediate formed directly from feedstock sugar, glyceraldehyde, played
an important role in the model. The C3 intermediate could:

- Disproportionate to one mole of glyceric acid and glycerol, via the Cannizzaro
reaction

- Undergo hydrogenation to form glycerol

- Undergo B-elimination of water to form pyruvaldehyde

— Condense to by-products

In equation form:

d[GA]
dt

 

= kG-GA [F] — (kGA-F + kGA-CZ) [GA]2 - (kGA-PA+ kGA-CD + kGA-Go ' 9124.329“ )[GA] (70)

Similar to glucose, fructose to glyceraldehyde was considered reversible. The
second order term reﬂects two moles of glyceraldehyde necessary to form one mole of
ﬁuctose. The other reactions, as listed above, were considered one-way with no
backward reaction occurring, consistent with hydrogenation products remaining stable

and salt formation being irreversible under alkaline conditions.

6.3.2.3 - Aldehydic intermediates: Pyruvaldehyde

Pyruvaldehyde was the other intermediate of interest in the kinetic model. Formed
via the B-elimination of hydrogen from glyceraldehyde, the intermediate is the precursor

for a number of pathways:

- Hydrogenates to form propylene glycol

- Undergoes benzylic acid rearrangement to form lactic acid

115

- Dicarbonyl cleavage to form formic acetic acids

- Condenses to by—products

With these reactions, the rate equation became:

d[PA]
dt

= kGA—PA [GA]' (1‘ PA-LA + k PA—FA + k PA-CD + kPA-PG ' 911-829w )[PA] (71)
Propylene glycol production required two moles of H2, which accounts for the 4th
power term for (93.32. Formic and lactic acid (in salt form), condensation by-products,

and propylene glycol were assumed to all form irreversibly.

6.3.3 - Combining the equations

With the isomerization and hydrogenation reactions accounted for, the derived
equations must be combined to completely describe the system. The sugar isomerization,
degradation, and hydrogenation equations occur simultaneously and must be modeled
appropriately. Figure 6-3 gives an illustration of the modeled pathways.

In all, 16 differential equations with 40 parameters describe the reaction
environment. These equations (Appendix A) were solved simultaneously using a 4th
order Runge-Kutta numerical method as described by Camahan et al. 101 in Microsoﬁ

102

Excel using Visual Basic (see Appendix B for program code). Initial conditions were

set by the makeup of the feedstock and conditions of the reaction.

 

‘°' Camahan, 13.; Luther, H. ; Wilkes, 1., Applied Numerical Methods, Wiley, New York, 1969
'02 Modiﬁed version of program in Jere, F .T., PhD. Dissertation, Michigan State University, 2003

116

coaomo. 835550 I No
«.8305 cozmmcoucoo I no
28 e__8>_m I «<30
Eon utmozo I <50

Eon 29:2 I <n_

Hum 282 I 5

_Eeo>_m I 00

Born 2226 I 0m
_oo>_m 05.395 I on.
0352283 I «<0
oquEgEE I «E
euzogeeezm I <0
8:52: I 0.2

_ogEom I ow

30:52. I s.

088:... I n.

38:6 I 0

Am: 6:on 8 2v E832 8m E828 83H Hm.<u_

_ovoE 53983.6»: 55 5592on acne—R vamoaoa me @3353 couosom um-» 95»:—

 

 

 

 

862x
00...?!
(I— (>40 00
we ... \J/
m V 0e...
Omaﬂu‘uxl {21136203001
by?
k . a
<n_% e %o
G
y We. k see?
W o.~<ov_
_>_ Iv. N<0 Iv Om.

 

117

6.3.4 — Assumptions of the model
A number of assumptions for this kinetic model have already been discussed.
They are summarized here:

- Reactions are ﬁrst order with respect to the reactant

- Organic acids formed are in salt form and do not react further

- The drop in hydroxide concentration from the formation of acid products does not
affect the rates of alkaline degradation pathways

- Glycerol, propylene glycol, and ethylene glycol are stable under reaction
conditions and do not react further

- Sorbitol and mannitol do not dehydrogenate. This was based on the reaction rates
at the process temperature of 40°C. A degradation term was added to account for
apparent conversion of the sugar alcohols toward the end of experiments.

- The pathway of glyceraldehyde to pyruvaldehyde is not reversible. It was
assumed that the conversion of pyruvaldehyde to PG would be rapid enough to
not allow a reverse reaction.

- Ruthenium metal has two separate sites: one for hydrogen adsorption, another for
the organic substrates

- All substrates that hydrogenate are in solution as reactants and complete the

reaction adsorbed to the catalyst site
6.4 - Calculation of concentration versus time

In all, ten experiments were considered in the ﬁtting of the kinetic model. These
experiments represent the standard conditions of 40°C, 100 psig H2 pressure, 0.1 grams
of catalyst loading, and 3.17 wt% ﬁ'uctose feedstock, and variations based on hydrogen
pressure, sugar loading, feedstock sugar used and catalyst loading. Each reaction had the
same loading of potassium hydroxide, 5.3 wt%, agitation rates (1000 rpm), and catalyst
reduction conditions (see Table 6-2). Using a 4th order Runge-Kutta numerical method?1

the derived equations previously discussed were solved simultaneously at discrete times

118

and then compared to the collected data. One hundred eighty-six time steps were used
for the calculations to reduce the amount of rounding error between iterations and set

time steps at two minutes exactly for comparison to concentration versus time data.

Table 6-3: Experimental conditions used for kinetic model - 40°C, 0.85 M KOH

 

 

 

 

 

 

 

 

 

 

 

Experiment “3332?“ 7.332;" Lfféilésfg, ”’ $2.?”
115 Fructose 3.17 0.05 100
119 Fructose 1.00 0.10 100
140 Fructose 3.17 0.10 200
141 Glucose 3.17 0.10 100
142 Mannose 3.17 0.10 100
144 Fructose 2.01 0.10 100
146 Fructose 3.17 0.10 100
147 Fructose 3.17 0.10 25
149 Fructose 3.17 0.10 50
152 Fructose 3.17 0.20 100

 

 

 

 

 

 

 

Concentration versus time data were collected for thirteen compounds in each
experiment, giving a total of 1274 data points. These data were compared to the
calculated value from the model at the nearest time point over ten experiments. The
difference was then squared and combined with the other residuals to form a total error
used to compare how well the model ﬁt the data.

To reduce the total error and improve the model, parameters were varied in
random walk fashion by a set percentage (e. g. 10%). The new value for the parameter
was then used to calculate new concentrations for the kinetic model. Once complete, the
total error ﬁ'om the 10 experiments was calculated and compared to the best result. Ifthe

error was reduced, the new value for the parameter was kept and the program proceeded

119

to the next step using the reﬁned parametric set. If not, the previous set of parameters
was restored and then the next random walk was performed.

When no improvement was found, the random number generated, and thus the
walk taken, was recorded into an array. For each new walk, the random number was
compared to those used since the last improvement. Any matches in the array led to the
generation of another number since the random walk had already been taken. Ifall of the
steps were used without ﬁrrther improvement, the program notiﬁed the user and halted
computation. Any reduction in the total error wiped out the values in the array and
allowed any step to occur aﬁer that point. For 40 parameters, this bit of coding reduced
the amount of random walks needed by over 80% and ensured that each permutation had

been exhausted.

6.5 - Results

Table 6-3 lists the values of the 40 parameters that yielded the lowest total sum of
squares error from the 10 experiments. The exact values have been truncated for easier
presentation. Conﬁdence intervals were not determined due to the large number of
parameters in the set of equations. As described by Belohlav et a1. 103, models with a
large number of parameters tend to have interdependence between each rate constant
creating large conﬁdence intervals that serve no useful purpose and, thus, were not
included.

The kinetic model shows good agreement with the data collected ﬁ'om the ten
experiments included. Trends in sugar isomerization, hydrogenation, and degradation

were matched very well as seen in Figures 6-4, 6-5, and 6-6 for Experiment 146.

 

"’3 Belohlav, z.; Zamostny, P.; Kluson, P.; Volf, 1., Can. J. ofChem. Eng, 1997, 75, 73 5-742

120

Eozii

see - 0.2
E - ow
see - 0m
5 - _2

3%-:

5.0.1:

3%- 0
Ed.

3%-“.

OOOOOOOOOOOO

d

0

08¢. .m> “0:588 93 433.38 68:58 682% 6883..“

03m .sz W mean as em med 2: .008 .mom some .8803 {En
oz “ﬁancee . ﬂeece 82.2 as see 3555?. .8 8&8an "I. as»:

A952: 25...

 

 

 

0.
o
r

 

0.
to
F

 

 

 

(116) uoneuuaouoo

o 0.
l0 0
N N

 

 

.../A
o. .
O
00

 

 

9mm

121

08E. .9 Boa ouoobw Ea 483w 0:2»qu 4809mm“ .38 owes
03m .er n mean as em men 2: .98 .mom :2 .3303 as»;
o: “ﬁgmxm - ESE 33.2 c5 88 358.5%”. mo 838800 "We 0.59%

.9505 25...

 

 

 

 

 

 

 

o w v m N _. o
P _ _ r p p 0.0
E 5.6 x x x x x x laminate
I ..ln' 1.. IIII'U Dre.\ N e
. .................................... U. ....... - \ OF
.u. .........
Bmo-<>..o x ... n. ................... ... m\
U r. ............. q ¢ 1‘ ON
D ........................... \\ O
- ....................................... 4 4 0\ 0
E On. 4 4 \I\ U
Isle om o
llllll a
38.8 n IIIIIII o \ W
IIIliulnlrtl. J
IIIII o ov m...
H_hloo.lll o m-
U
£3.00 4 m
h
5.431

 

O O
l‘ (O

 

 

 

0d

122

06:. .m> B8 omoobw c8 .0m .38 8:8,“ .8889 5.88200

98 8.3 n ea c; e: men 2: .008 .mox $8 .3208 $28.
0: E0895 - 805 82.2 Ea 3% 385.5%... me 88958 "we 2:»...—

3505 05:.

E-~<Eo.ii n n

 

I.III
'IIII
IIIIII
lllllllll'
Illll

 

EMU - N<>40 X

o .
o
E - om ........... o
b m...
3% - em a \
8
E - E . I I I \
ma
3% - E < \
E- 8| \ o m
8

Smu-00 0 \

 

 

 

 

 

 

 

 

 

oé

123

(116) uonenuaouoo

Appendix C contains the graphs that compare the experimental data to the ﬁtted

model. Some difﬁculties did arise in ﬁtting sorbitol and mannitol data from experiments

with varying hydrogen pressure and catalyst loading. Pressures from 25-100 psig

(Experiments 147, 149, and 146) showed very similar amounts of sugar alcohol

production. An increase to 200 psig showed a much larger response, which caused

signiﬁcant errors early on in the calculations. Further details of the speciﬁc portions of

the model, namely the sugar isomerization reactions, hydrogenation, and alkaline

degradation reactions, are continued in the sections below.

Table 6-4: Best ﬁt parametric values

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Constant Value Units Constant Value Units
km 0.598 111' '1 kmm 1.2 x 1014 hr '1 gm, '5
k0... 0.668 hr '1 km.“ 6.6 hr '1
km 0.089 hr '1 1mm 1.8 hr '1
194.}: 0.197 hr .1 “(GA2-E0 1.2 X 108 hr -1 gmetal -3
km. 0.022 111' '1 lama 0.0006 L mol “ III"
km 0.059 hr '1 kww 0.00008 hr '1
167-30 8.9 x 107 hr '1 gm. '3 kmADs 21.8 —
kF-MO 5.2 X 107 hr -1 grnetal .3 ksoqms 2.5 X 106 mol I...1 111' '1
kc..so 6.1 x 107 hr '1 gm... '3 moms 116 —

kM-MO 7.8 X 106 hr .1 gmetal -3 kMO-DES 2430 mol L-1 hl’ -1
km... 0.52 In '1 Imam 4.5 —
kg” 3.9 L mol ‘1 hr ‘ kooogs 799 mol L'1 hr '1
ream; 0.486 hr ‘1 kmws 154 —
kCAM 12.5 L mol '1 hr ‘ kmpgs 1.3 x 1010 mol L" hr '1
km. 0.024 hr '1 kmws 1483 —
kGAcz 4.5 L mol '1 hr 1 kmms 15.5 mol L'1 hr '1
km...) 1.8 x 109 III '1 gm... '3 Kn 8.4 psi ‘1
kg...” 8.4 hr '1 ksow 0.073 hr '1
ream) 0.007 hr '1 mow 0.0005 hr ‘1
rep..-” 0.0001 hr '1 kmm 0.018 . hr ‘1

 

 

6.5.1 - lsomerization equilibria

lsomerization equilibrium terms were compared to literature values to verify the
validity of the determined rate constants, or at least demonstrate consistency with
previous research. Sugar isomerization reactions have been previously
characterized69’7l’lM’los’m, which allowed for reasonable benchmarks. Table 6-4

summarizes the equilibrium constants from the kinetic model and literature values.

Table 6-5: Sugar isomerization equilibria values (kfm / knee-so)

 

 

 

 

 

 

 

 

 

 

 

 

Tem
KG-M KG-F KIA-F (ec)p PH Ref.
0.39 1.12 2.21 40 13.8 from Table 6-8
0.52 — 1.94 0.91 — 0.99 1.67 - 2.04 22 14 MacLaurin”
0.42 1.09 2.67 78 13 de Bruijn et al.79
0.29 0.85 — 0.89 3.00 25 7.4 Tewari94

 

Ranges for the equilibrium results are based on uncertainty ranges provided by the
appropriate references. For the results of MacLaurin and Green“, 1000 equilibrium
constants were calculated based on a Gaussian distribution of the error for each
parameter. The conﬁdence interval was then found to be the range of the 2.5th percentile
to the 97.5“ percentile for ~95% conﬁdence. Equilibrium constants for Tewari94 were
calculated using the standard Gibbs energy change of reaction, or AG°. With this value
for each reaction (glucose to mannose, etc.), the equilibrium constant for the reaction was

calculated by relating K and AG°1°72

 

‘04 Tewari, Y.B.; Goldberg, RN., Biophys. Chem, 1986, 24, 291-294

“’5 Camacho-Rubia, F. et al., Can. J. ofChem. Eng, 1995, 73, 935-940

'06 Franks, F., Pure andAppl. Chem, 1987, 59 (9), 1189-1202

'07 Smith, J.M.; Van Ness, H.C.; Abbott, MM., Introduction to Chemical Engineering Thermodynamics,
5th Ed, 1). 567, The McGraw-Hill Companies, New York, 1996

125

— RT In K = AG° (72)

-AG°
K =
exp[ RT J (73)

Equations 59 and 60 also veriﬁed the validity of the isomerization equilibrium

 

constants determined from the kinetic model ﬁtting. Each isomerization reaction has a
value for AG°. When placed in series, i.e. glucose to ﬁ'uctose to mannose to glucose, the
total AG° equals zero because the initial and end states areidentical. Table 6-5 the

equilibrium constants, calculated AG° for each isomerization, and the total AG°.

Table 6-6: Equilibrium constants, reaction AG° (kJ/mol), and total AG° (kJ/mol)

 

 

 

 

 

 

 

 

 

 

 

 

...... .... 2:. I‘I‘i. ‘Z’éf'

This research 0.39 2.21 0.89 2.3 -1.9 0.3 0.7
MacLaurin” 1.0 1.83 1.05 0.0 -1.5 -0.1 -1.6
de Bruijn et M.” 0.42 2.67 0.92 2.1 -24 0.2 -0.1
Tewarig“ 0.29 3.00 1.15 3.0 -2.7 -03 0.0

 

 

 

The kinetic model showed a total AG° of reaction of 0.7 kJ/mol based on
calculations from the ﬁtted rate constants. This was a signiﬁcant discrepancy from the
expected value of 0.0 kJ/mol. To improve this aspect of the model, an additional
equation is needed to calculate total AG° for the system and adjust rate constants in such
awaythatA °=0.

High pH levels also affect K values, as illustrated by de Bruijn.77 Table 6-6
shows equilibrium constants for the isomerization reactions over a range of hydroxide

concentrations as calculated in the paper.

126

 

 

 

 

 

Table 6-7: Comparing equilibrium values with hydroxide concentrations at 78°C
OH' concentration (M)
0.001 0.0035 0.010 0.035 0.100
KGM 0.39 0.38 0.39 0.38 0.42
Ker 1.3 1.2 1.3 1.2 1.1
KIM: 3.3 3.3 3.3 3.3 2.7

 

 

 

 

 

 

 

 

The pH effect was more pronounced at the higher base concentrations, which was
of interest with the reactions used for the kinetic model. It was assumed that reductions
in base concentration due to acid formation/neutralization would not affect the rate of
reaction. The relatively large change in all three equilibrium constants between 0.035 M

and 0.100 M suggested that this hypothesis was incorrect.

6.5.2 — Hydrogenation rate constants
Parameters that described the rate of hydrogenation of various aldehydes were
numerically large, which reﬂected the inclusion of catalyst weight, in grams of metal, in

the rate equations. These amounts were typically 0.005 grams and were cubed to include
the fractional coverage terms, (9,382 and @v, , ﬁom each hydrogenation term. The

relative rates of hydrogenation either reﬂect trends observed in literature or those
implicitly included with the model, as is described below.

The rate constants for hydrogenation to glycerol, propylene glycol, and ethylene
glycol were all an order of magnitude larger than the hydrogenation rates for the sugars.
This reﬂects the high reactivity of glyceraldehyde, pyruvaldehyde, and glycolaldehyde
and the assumption that these aldehydic intermediates had minimal concentrations over

the course of the experiment. The maximum concentrations for glyceraldehyde,

127

glycolaldehyde and pyruvaldehyde over the course of all ten experiments were 0.77, 0.95,
and 0.69 g/L, respectively. Aﬁer six hours of reaction, these concentrations were
typically less than 0.1 g/L.

Relative rates of sorbitol and mannitol production were fairly different from
published results. From Table 6-3, ﬁ'uctose shows a 9:5 preference toward sorbitol
production over mannitol during hydrogenation. Wisniak and Simon21 reported a slight
preference to mannitol during experiments. However, their results did not include
temperatures below 85°C. In addition, the experiments for the Wisniak paper were run at

neutral pH instead of alkaline conditions, which makes comparison difﬁcult.

6.5.3 - Adsorption equilibrium constants

The model also determined rates of adsorption and desorption for the
hydrogenation products. In each case, desorption was preferred by many orders of
magnitude, which was apparent from the amounts of these products in the liquid phase
and indicated little product afﬁnity for the catalytic surface. Relative adsorption rates
between the ﬁve hydrogenation products show little difference in binding ability of each
substrate. Further, the large values for each equilibrium constant suggested little or no
product inhibition. Table 6-8 has the values for adsorption, desorption, and equilibrium

for each hydrogenation product.

128

Table 6-8: Parameters for adsorption/desorption pathways

 

 

 

 

 

 

 

 

Product kilos knee K
so 21.8 2.5 x106 8.7 x 10'6
MO 116 2430 0.048
G0 4.5 799 0.006
PG 153 1.3x10'° 1.1x10’8
EG 1483 15.5 95.4
H2 8.4

 

 

 

 

 

The small value of K for four of the compounds (sorbitol, mannitol, glycerol, and
propylene glycol) suggests that these terms can be removed from the kinetic modeling.

This would further simplify calculations without losing accuracy in the model.

6.5.4 - Alkaline degradation reactions
Unfortunately, not much comparison was available for rates of reaction for the
degradation portion of the model. Previous work merely described sugars going to acidic

69, 73

products as opposed to speciﬁc products as done in this work. Overall, the acidic

products observed in the experiments were included in the kinetic model and accounted

for sufﬁciently.

129

Chapter 7 — Continued development of alkaline
degradation with hydrogenation and related work

This chapter outlines efforts to further improve C3 yields of the alkaline
degradation hydrogenation reaction and describes other projects related to this work that

were conducted to enrich the overall effort.

7.1 - Modiﬁed catalyst for selective C; hydrogenation

The alkaline degradation with hydrogenation reaction involves a competition
between carbon-carbon cleavage of the sugar feedstock to C3 intermediates and
hydrogenation of the sugar to the corresponding sugar alcohol. With the typical
ruthenium catalysts, unsaturated compounds are hydrogenated indiscriminately, which
lowers potential C3 yields of the reaction signiﬁcantly. If a catalyst could hydrogenate
the intermediates glyceraldehyde and pyruvaldehyde without reacting with the sugar

feedstock, C3 yields would rise signiﬁcantly.

7.1.1 — Selective C3 hydrogenation in the presence of glucose

Two modiﬁed platinum catalysts, a bimetallic platinum-germanium catalyst and a
sulﬁded platinum catalyst, were obtained ﬁom Paciﬁc Northwest National Laboratories
(PNNL). Both catalysts demonstrated selective hydrogenation of aldehydes in the
presence of sugars and were candidates for the desired selective hydrogenation of
glyceraldehyde and pyruvaldehyde.

Reactions used 0.1 grams of catalyst (dry basis) to hydrogenate a solution of 0.4
wt% (0.044 M) glyceraldehyde dimer with 0.8 wt% glucose (0.044 M) added to the
feedstock. Hydrogenations were carried out at 40 to 60°C and under 100 to 200 psig of

hydrogen at neutral pH. Figures 7-1 and 7-2 show the selective hydrogenation of

130

glyceraldehyde from the feedstock (Figure 7-1) to a sample after four hours of reaction
(Figure 7-2). Sixty percent of the glyceraldehyde converted to glycerol while less than
2% of the feedstock glucose hydrogenated, which demonstrated the desired effect of
selective hydrogenation. Control experiments with glucose with Pt(S)/C (Exp 161) and
fructose with Pt(S)/C (Exp 164) under hydrogen showed a similar low amount of
hydrogenation aﬁer four hours of reaction (< 2% conversion for both experiments).
Ideally, this catalyst can be used in the alkaline degradation with hydrogenation
reaction to capture aldehydic intermediates (i.e. glyceraldehyde and pyruvaldehyde) to
form the saturated polyols glycerol and propylene glycol. This would beneﬁt the reaction
by reducing the amount of organic acid formation, which would maintain a high pH level
to promote selectivity to C3 products via retro-aldol condensation. More importantly,
sorbitol and mannitol production would be avoided, thus allowing more of the feedstock
sugar to undergo carbon-carbon cleavage to C3 intermediates and further enhance C3

yield.

7.1.2 - Alkaline degradation with hydrogenation using Pt catalysts
Using the sulfided platinum catalyst, alkaline degradation with hydrogenation
reactions were carried out for comparison to ruthenium on carbon. The reaction was
carried out at 40°C with a hydrogen pressure of 100 psig and a catalyst loading of 0.2
grams (dry basis). Experiments used fructose and base concentrations of 0. 18 M and

0.85 M, respectively. Table 7-1 compares the product distribution of the two runs.

131

055 can - 9d a: 93 8m .99 “82:28 0:95 can - 9.0 a: 93 com .98 ”22:28

 

 

 

 

 

 

 

 

 

 

 

 

 

5:32 .6 £30: 52 .2? $826 £5, «6 63528829 eats Nd ”x0233“.
$83.0 5E 88:6 5;,
03202883 .6 coamcmmegc 982$ nu-» 8:9". 833.833 ho coumcomegz 932$ 34. 952...
.5 .2_ or m. 2. 3. a: or m. 3-
...:II , Ilslwlw . p... Ll i l. - s ll 0 llqlmﬁ-
-.. a B
..A
m -3
I . O
nlu 2 i 9
.A
m m .M L...“
w m 1% m
I w m
m m n
m ... w m
u- n I mg. ..A m
.A m D.
a s ...
a L:
- 3:
a
u 2:

 

 

 

 

 

132

 

 

Table 7-1: Comparison of Ru/C and Pt(S)/C catalysts
Conditions - 40°C, 100 psig, 0.18 M fructose, 0.85 M KOH, 0. lg catalyst (dry basis)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C3 Max % of C3
Exp Catalyst Conv. Hyd. lso. Deg. Sel. C: max yield
146 5% Ru/C 95% 0.15 0.11 0.69 1.00 1.38 72 0.94
167 3% Pt(S)/C 96% 0.21 0.08 0.67 0.82 1.34 62 0.79
168 5% Pt(Ge)/C 97% 0.37 0.07 0.53 0.70 1.06 66 0.68

 

 

The sulﬁded platinum showed no apparent improved selectivity for C3
hydrogenation. Mannitol and sorbitol formed readily and the ﬁ'action of degraded
material that formed C3 products was the same as the ruthenium on carbon catalyst. The
change in reactivity of the sugars ﬁom neutral conditions to basic conditions was an
effect of ionizing the sugar. At neutral conditions, all but a negligible amount of sugar is
in cyclic form108 while glyceraldehyde is linear in nature and is able to adsorb to the
metal. Glucose and fructose hydrogenate in the liquid phase, as opposed to an adsorbed

State19-22

, which does not allow the sugar to compete with glyceraldehyde for the open
sites.

When ionized under high pH conditions, sugars assume an open form to
isomerize and enolize.109 In this Open form, the sugar can adsorb to active sites and is no
longer strongly affected by other adsorbed species. Thus, sorbitol and mannitol formed

readily, which decreased the C3 yield due to the loss of feedstock to hydrogenation

products in a similar fashion as Ru/C catalyst.

 

”8 Ellis, A.V.; Wilson, MA, J. Org. Chem, 2002, 67, 8469-8474
‘°9 deWit, G. ;Kieboom, AP.G.; van Bekkum, H., Carbohyd Res., 1979, 74, 157-175

133

7.2 — Methane production from carbohydrate feedstock

Catalyst screening by Chopade et al.4 for sorbitol hydrogenolysis revealed that
certain catalysts converted sorbitol into methane with high selectivity. Initially, the
reaction was not of interest due to the loss of potential yield to desired products.
However, in light of increased interest in hydrogen production, the reaction was revisited.

Hydrogen is produced mainly by steam reformation. Natural gas and steam are
passed over a nickel-based catalyst at temperatures of 800-870°C and pressures of 300-
400 psig to form three moles of hydrogen from one mole of water and methane.110 Thus,
methane to hydrogen is a well-developed process, albeit energy intensive. With a bio-
based source of methane, hydrogen could be produced directly from renewable resources
in lieu of petroleum feedstocks.

With the previous results of methane production, we investigated the use of
sorbitol and glucose as feedstock, relating it directly to previous work. We also included

glycerol due to its increasing grth as a side product from bio-diesel production.

7.2.1 — Sorbitol to methane

For these reactions, 10 wt% sorbitol (0.55 M) solution was reacted at ZOO-210°C
under 1000 psi of hydrogen pressure. Five weight percent (5 wt%) ruthenium on carbon
and 3 wt% ruthenium on titania were the two catalysts used in this study to compare the
effect of support. The catalyst loading was 1.0 grams (dry basis) for each. Reaction
times were typically four to six hours, or until liquid phase products were no longer
detectable by HPLC. Table 7-2 shows the reaction conditions along with selectivity and

yield to product gases.

 

“° Kroschwitz, 1.; Howe-Grant, M, editors, Kirk-0thmer Encyclopedia ofchemical technology, 4th Edition,
13, p. 852, Wiley, New York, 1992

134

Table 7-2: Yield of methane and ethane ﬁom hydrogenolysis of sorbitol
Conditions - 1200 psig total pressure, 1.0g catalyst, 0.55 M sorbitol

 

 

 

Temp. Hz pressure CH4 CzHo Carbon
EXP “my“ °c (psi) Yield Yield balance
L7 Ru/ TiOz 210 1000 5.7 0.5 112%
L13 Ru/C 200 1000 7.1 0.3 136%

 

 

 

 

 

 

 

 

 

Yields of methane and ethane have units of moles of product per mole of reactant
converted. For these experiments, the maximum value for yield is 6.0, corresponding the
six carbons in sorbitol and one carbon in methane. Carbon balances were above 100%
closure due to errors associated with dilution of the reactor during repressurization after
sampling and quantitative analysis of gas chromatography.

Both carbon and titania supports show high yield to methane and readily convert
sorbitol. Analysis with gas chromatography showed only methane and ethane produced
in the reaction. No appreciable difference between the titania and carbon support was
noted with respect to methane selectivity. Thus, carbon support would be preferable to

titania due to lower costs per gram of material.

7.2.2 - Glucose to methane

Given the success of sorbitol to methane, glucose to methane was the logical
progression for the reaction. Ideally, glucose would hydrogenate to sorbitol and then
form methane with similar yields while reducing the feedstock cost ﬁ'om $0.40 to $0.07
per pound. Table 7-3 outlines the results of these runs. For this reaction, glucose was
hydrogenated to sorbitol at 120°C and 600 psig of hydrogen and then heated to 200°C

and pressurized to 1200 psig to begin hydrogenolysis.

135

Table 7-3: Yield of methane and ethane from hydrogenolysis of glucose

Conditions — 1200 psig total pressure, 1.0g catalyst, 0.55 M glucose

 

 

 

 

 

 

 

 

 

Temp Hz pressure CH4 02H; Carbon
EXP Catalyst °C (psi) Yield Yield balance
L1 1 Ru/T 103 120/200 600/ 1 000 5. 1 0.4 99%

 

Glucose shows the same selectivity to methane as sorbitol, although the reaction

rate was slower. After six hours of reaction, about 2% of the original carbon ﬁ'om the

feedstock remained in the liquid phase compared to none with the sorbitol cracking

experiment. Cracking of the liquid phase products continued beyond six hours as after 24

hours, the liquid phase products were completely reacted. High methane yield and

complete conversion of the feedstock makes glucose a viable option as a feedstock for

this process.

7.2.3 — Glycerol to methane

Another set of cracking reactions was conducted to convert glycerol to methane.

Ruthenium supported on titania was used as catalyst for these reactions. Table 7-4

outlines the results of these runs.

Table 7-4: Yield of methane and ethane from hydrogenolysis of glycerol

Conditions - 1200 psig total pressure, 1.0g catalyst, 1.1 M glycerol

 

 

 

 

 

 

 

 

 

 

 

Tea "2 1:5?“ 22.1.: 3521.1:
L10 Ru/Ti02 200 1000 3.2 0.3 118%
L12 Ru/TiOz 190 1000 2.8 0.2 103%
L14 Ru/TiOz 200 1000 3.2 0.3 117%

 

 

136

 

Glycerol showed high methane yields in the same manner as sorbitol and glucose.
An interested observation was also made while analyzing the liquid phase products over
time. As glycerol cracks to form methane and ethane, ethylene glycol, propylene glycol,
ethanol, and l-propanol form as the only intermediates of the reaction. This simple
product set could make this system especially interesting for study of the mechanisms of
polyol cracking. Such an analysis, however, was beyond the scope of this particular

study.

7.3 — Summary

Two modiﬁed platinum catalysts, Pt(Ge)/C and Pt(S)/C, demonstrated selective
hydrogenation of the C3 intermediate glyceraldehyde while in the presence of glucose at
neutral pH. This reaction showed potential for improving C3 yields of alkaline
degradation with hydrogenation of fructose and glucose by reducing the amount of
sorbitol and mannitol produced from hydrogenation of the feedstock. However, these
catalysts had higher activity for sugar alcohol production than Ru/C. The addition of
base to the reaction enabled the platinum catalysts to hydrogenate the sugars, presumably,
due to unsaturated, acyclic intermediates that formed during isomerization (e. g. enediols).

Selective hydrogenation of C3 intermediates still holds promise, however. Size
exclusion catalysts have yet to be explored as a method for hydrogenating the C3
intermediates while ignoring the C5 feedstock sugars. With a proper pore size
distribution, glyceraldehyde and pyruvaldehyde could diffuse readily into the catalyst to
form glycerol and propylene glycol while glucose, ﬁ'uctose, and mannose remain in
solution to undergo alkaline degradation. Catalysts with high hydrogenation activity

would be ideal to maximize production of PG and GO while minimizing the residence

137

time of C3 intermediates, which would reduce the amount of organic acid produced from
rearrangement reactions of the intermediates.

While selective hydrogenation of C3 intermediates was not very successful,
experiments that converted carbohydrates to methane showed interesting results. Using
sorbitol, glucose, and glycerol as feedstock, between 80 and 90% of the feedstock carbon
converted into methane while the rest converted to ethane when using a ruthenium
catalyst. Complete conversion of the feedstock was usually observed within six hours at
temperatures and pressures comparable to sorbitol hydrogenolysis (200°C, 1200 psi).
With glycerol as feedstock, only four liquid phase intermediates appeared over the course
of the reaction: ethylene glycol, propylene glycol, l-propanol, and ethanol. Further
study of this reaction may lead to insight on the hydrogenolysis pathways of glycerol to
PG and EG.

Economically, the production of methane from sorbitol (SO), glucose (G), or
glycerol (GO) feedstock is currently unfavorable. Table 7-5 outlines raw material costs

for feedstock, hydrogen, and an estimate of overall price of methane production.

Table 7-5: Estimated price of methane production with bio-based feedstock

 

Feed H2 RM Total

Feed Equation cost“ cost‘ cost“ cost”

 

so €611.40. + 11 H2 —+ 6 CH. + 6 H20 $0.038 ‘° $0.004 $0.042 $0.084
G €613.20. + 12 H; —> 6 CH. + 6 H20 $0.005 “ $0.004 $0.009 $0.018

 

 

GO €311.03 + 5 H2 —» 3 CH. + 3 H20 $0.053 '° $0.004 $0.057 $0.114

 

 

 

 

 

 

 

 

* Cost basis is per mole of methane produced
** Total cost is calculated by estimating raw material (RM) costs to be 50% of the total process costs

138

 

 

 

 

 

press

to be

Mill

C011

Natural gas prices in August, 2004 were $6.25 per 1000 scf 1” (l 113 at 1 atm
pressure and 60°F) or roughly $0.005 per mole of methane. While glucose is the closest
to being competitive with pricing, it is still three times as large as the current price of
natural gas. Thus, only a sizable increase in methane/natural gas pricing would promote

conversion of glucose to methane on a large scale.

 

1” U.S. Natural Gas Prices, bttp://tonto.eia.doe.gov/dnav/ng/ng_pri_sum_dcu_nuS_m.htm. viewed
November 15, 2004, last updated October 29, 2004

139

Chapter 8 - Conclusions and Future Work

To conclude this dissertation, a summary of and conclusions drawn from the
research is provided to give a roadmap of the work completed. In addition, a brief look at
process economics is taken to compare current prices of propylene glycol and glycerol to
the process costs involved with alkaline degradation with hydrogenation. Finally, a
section on future work draws up directions this work provides for ﬁlrther process

improvement and alternate directions the research may take.

8.1 — Summary

This research started as a continuation of prior group work to convert sorbitol into
propylene glycol and glycerol via hydrogenolysis. While the conditions were optimized
to maximize the amount of PG and GO produced, little was known about the mechanisms
involved with sorbitol conversion and cleavage. To begin investigating the mechanisms,
sorbitol adsorption experiments were used to observe the interaction between sorbitol and
the metal catalyst. Analysis showed two surprising results: 1) Sorbitol dehydrogenates
before cleaving into products and 2) The addition of soluble base to these adsorption
experiments increases the rate of conversion and raises the yield to C3 products with
respect to runs with just sorbitol in the feedstock. Searching the literature searching led
to the well-known pathways of base-catalyzed degradation of sugar. Alkaline
degradation of sugar, however, leads to the production of organic acids, mainly lactic
acid, instead of the desired polyols propylene glycol and glycerol. To address this
problem, a hydrogenation catalyst was added to the reaction in order to hydrogenate

intermediates as they form and capture them as desired C3 products instead of organic

140

acid salts. Experiments that varied reaction parameters mapped out the product
distribution and the effects of each variable on the product distribution. To ﬁirther
improve C3 yield and selectivity, other catalysts were used as alternatives to ruthenium on
carbon to preferentially hydrogenate C3 intermediates over the sugar feedstock. In
addition, a kinetic model was developed to mathematically describe the reaction and to
determine the relative rates of the pathways that promote PG and GO production. The
model also provides prognostic value by predicting product distributions given feedstock
composition, catalyst loading, and hydrogen pressure. Other work presented was an
extension of the main work for other potential applications such as conversion of

carbohydrates to methane.

8.2 - Conclusions
8.2.1 - Adsorption studies

Adsorption work with sorbitol and ruthenium metal in combination with 13C
NMR analysis of sorbitol hydrogenolysis showed dehydrogenation of the sugar alcohol
as the primary step of hydrogenolysis. Additionally, < C5 acid products, such as lactic
acid, formed when soluble base (e. g. potassium hydroxide) was added to the feedstock in
the adsorption experiments. Through the combination of catalyzed dehydrogenation of
sorbitol and the alkaline degradation of the subsequent sugar analog, C3 products of
interest formed. From this, we conclude that sorbitol hydrogenolysis proceeds through
the same combination of dehydrogenation and alkaline degradation to products. The
negligible amount of organic acids (< 2%) in the product distribution of sorbitol
hydrogenolysis experiments is consistent with this conclusion given the hydrogenation

activity of ruthenium catalyst and high hydrogen pressure present. Unsaturated

141

intermediates hydrogenate immediately upon formation to form propylene glycol and

glycerol.

8.2.2 — Alkaline degradation with hydrogenation

The combination of alkaline degradation of monosaccharides with the
simultaneous hydrogenation of the reaction intermediates presents a novel process for the
production of propylene glycol and glycerol at reduced temperature and pressure with
respect to sorbitol hydrogenolysis (40-50°C, 100 psig H3 vs. 230°C, 1200 psig H2). The
highest C3 yield observed was 1.09 moles C3 product per mole of fructose fed (55% of
theoretical) with process conditions 50°C, 100 psig H2, 0.18 M fructose, 0.85 M KOH,
and 0.1g of 5% Ru/C (Experiment 153). This is comparable to yields reported by
Chopade et al.4 Further, the reaction shows the potential for C3 product yields as high as
85% of theoretical. At conditions of 40°C, 100 psig H2, 0.12 M ﬁ'uctose, 0.85 M KOH,
0.1g of 5% Ru/C (Experiment 144), 85% of the feedstock sugar that converted to < C5
products formed C3 products of interest. This result is a signiﬁcant improvement from
current technologies. However, the overall C3 yield of the experiment, 53% of
theoretical, reﬂects the formation of sorbitol and mannitol ﬁom hydrogenation of the
ﬁ'uctose feedstock. Thus, there is potential for high C3 yield, but it has not yet to be
realized.

Some process concerns require ﬁlrther attention in future work. The high
feedstock concentration of soluble base would essentially double the raw material cost
and present potential corrosion problems for process equipment. Dilute concentrations of
the feedstock sugar with respect to sorbitol hydrogenolysis (3.2 wt% vs. 25 wt%) lessen

the potential throughput of this process as well. Finally, a proper catalyst must be found

142

to reach the potential C3 yield of 85%. Traditional ruthenium catalysts hydrogenate
without regard for the nature of the unsaturated species. Without preferential C3
hydrogenation over C6 hydrogenation, sugar alcohols form readily, which lessens the C3

yield of the process.

8.2.3 — C3 selective hydrogenation

Work with modiﬁed platinum catalysts from PNNL demonstrated selective C3
hydrogenation. At neutral conditions, glyceraldehyde hydrogenated to glycerol in the
presence of glucose or fructose with little conversion of the sugars (< 2%). However, at
the conditions of the experiments (pH > 14), sugars hydrogenated readily in addition to
the aldehydic intermediates that formed during alkaline degradation. Thus, the
hydrogenation of sugars was an effect of the base addition. Presumably, hydrogenation
occurred during isomerization of the sugar with the formation of an acyclic, enediol
intermediate and subsequent C=C bond, which reacted with the active metal and
hydrogen to saturate the bond.

Modifying the active metal on the catalyst was not sufficient to selectively
hydrogenate the C3 intermediates in alkaline conditions. A catalyst that excludes
reactants based on size would be very USCfUl. The larger sugar molecules would remain
in the bulk liquid phase to degrade into C3 intermediates while the smaller intermediates
would diffuse into the catalyst to hydrogenate and form propylene glycol and glycerol.
In addition, high hydrogenation activity would be necessary to minimize the amount of

organic acid formation.

143

8.2.4 — Kinetic model of alkaline degradation with hydrogenation

The kinetic model of the alkaline degradation with hydrogenation reaction
presented in Chapter 6 provides a starting point for continued mathematical description of
this reaction. Concentration versus time data and the calculated concentrations from the
ﬁtted model show general trends of sugar isomerization, conversion, and proper
distribution of products. However, some of the quirks of the reactions are not accounted
for from the model. For instance, in Experiment 140, sorbitol and mannitol formed rather
quickly with respect to the other products and then leveled off for the remainder of the
reaction. The model reaches the same concentration eventually, but not with the same
proﬁle.

From the ﬁtted rate constants, the model can be simpliﬁed by removing the
equilibrium constant terms for sorbitol, mannitol, propylene glycol, and glycerol (Kso,
KMO, KpG, Koo) from the rate equations. These constants were essentially zero and did
not contribute signiﬁcantly to the overall model. Physically, this means that as the
products form, they desorb readily and do not readsorb to the catalyst surface. This is
consistent with kinetic analyses with sugar hydrogenation.” 22’ 23

To further improve the model, it may be necessary to include hydroxide
concentration when calculating sugar isomerization and conversion rates. From de Bruijn
et al.73 the rate of isomerization changes at different pH levels. Unfortunately, this may
be true for nearly all rate constants as alkaline degradation chemistry is dependent on
base concentration. Adding base dependent terms to each rate constant associated with

alkaline degradation would nearly double the number of parameters used, making the

144

model unwieldy. However, limiting the base dependent terms to just the isomerization
terms may be sufficient for a better ﬁt to the data.

For a more robust model, higher concentrations of initial feedstock will need to be
included. Ideally, the sugar reactants will reach concentrations in excess of 250 grams
per liter to increase production of C3 products. Neutralization of the reaction media will
certainly play a larger role with potentially more organic acid that can form ﬁ'om alkaline
degradation. The higher sugar loading will also affect hydrogenation rates, which must
be taken into account. Further, determining intermediate concentrations (i.e.
glyceraldehyde, pyruvaldehyde) over the course of the reaction will be important as well.
Iflarge amounts of intermediate form at the higher sugar loadings relative to the current

feedstock concentrations, the rate constants will need to be adjusted accordingly.

8.2.5 - Conversion of carbohydrates to methane

Hydrogenolysis of carbohydrate feedstock (sorbitol, glucose, and glycerol) at
neutral conditions produced methane with yields in excess of 85% of quantitative with
ethane as the lone by-product. The only raw materials of the reaction were hydrogen and
the carbohydrate, which made the reaction very simple. Quantitative analysis of the gas
products was complicated by dilution of the reactor headspace to maintain hydrogen
pressure and continue the reaction. Combined with quantitative uncertainties involved
with gas chromatography, carbon balances in some cases were in excess of 130%. With
only two gas phase products, however, the proportions of each could be easily
determined.

Of the three feedstocks, glucose was the most economically favorable at $0.07 per

pound versus $0.40 and $0.80 per pound for sorbitol and glycerol respectively. Despite

145

the lower price of glucose, production of methane from the sugar would still cost roughly
three times more than market price of methane. In a bioreﬁnery setting, however, the
process may become more viable due to a reduction in cost for glucose feedstock.
Instead of purchasing the sugar, it could be produced directly at the facility and used in
the hydrogenolysis process. This could conceivably reduce the price to $0.03 a pound or
less and cut the methane production cost by one-third. This option may be particularly
attractive in areas where natural gas and methane are not readily available or relatively

expensive. Methane production for 1116] cells may be another option worth consideration.

8.3 - Process economics of alkaline degradation with hydrogenation

Ideally, with the right process conditions and catalyst, alkaline degradation and
hydrogenation of sugar will be an economically viable way to produce propylene glycol
and glycerol. This section outlines raw material costs for the process and shows
estimates of total production cost for PG and glycerol. For comparison, these values
were calculated for both current yields realized and potential yields ﬁ'om experimental
work.

Raw material costs of the process include the feedstock sugar, soluble base, and
hydrogen. For the majority of experiments presented, fructose served as feedstock and
potassium hydroxide as the soluble base. Prices for these and hydrogen are listed in
Table 8-1. Experiment 152 showed the best yield for propylene glycol and glycerol with
feedstock of 3.2 wt% fructose and 5.3 wt% potassium hydroxide. KOH was added in
ﬂake form at ~85%, which brought the actual amount of base added to 4.5 wt%. Thus,

for every pound of fructose added, 1.4 pounds of KOH was added to the reaction.

146

Hydrogenation products accounted for 0.013 pounds of H2 per pound fructose fed. Table

8-1 outlines the raw material costs for the experiment.

Table 8-1: Raw material cost for Experiment 152 (Basis: 1 lb of fructose)

 

 

 

 

 

 

 

 

 

 

Raw Amount of raw Cost of RM Feedstock price
Material material (lb) (Sllb) (Sllb fructose)
Fructose 1.0 0.11 0.11

Potassium

hydroxide 1.4 0.33 0.46
Hydrogen 0.013 0.50 0.0065
Total Cost

per lb 0.58
ﬁ'uctose

 

From this feedstock, propylene glycol and glycerol were produced with yields of
0.28 and 0.32 moles product per mole ﬁ'uctose fed, respectively. This corresponded to
yields of 0. 12 lbs PG and 0.16 lbs glycerol per pound of fructose fed. Raw material costs
for the two products separately were $4.80 per lb PG and $3.60 per lb glycerol produced.
With PG and glycerol combined, the raw material cost was $2.10 per pound. Overall
process costs would be roughly twice the raw material cost leading to a total of $4.00 per
pound of product. Market prices for PG and glycerol are $0.65 and $0.80 per pound,
respectively. Clearly, yields were not sufficient for a reasonable selling price.

With different feedstock and signiﬁcantly higher yields to PG and glycerol, the
economics change considerably. In Experiment 144, 85% of the degraded material
formed C3 products. Assuming a proper catalyst (i.e. selectively hydrogenated C3
intermediates) was found and all of the C3 products were propylene glycol and glycerol,

hydrogenation products would account for 2.4 moles of H2 per mole of glucose fed, or

147

0.027 lb H2 per lb glucose. Table 8-2 shows the raw material costs associated with the

alternate feedstock.

Table 8-2: Raw material cost for alternate feedstock (Basis: 1 lb of glucose)

 

 

 

 

 

 

Raw Amount of raw Cost of RM Feedstock price
Material material (lb) (Sllb) (Sllb fructose)
Glucose 1 .0 0.07 0.07
Sodium
hydroxide 1.0 0.065 0.065
Hydrogen 0.027 0.50 0.014
Total Cost '
per lb 0.15
ﬁuctose

 

 

 

 

 

With the two substitutions of fructose to glucose and KOH to NaOH, the raw
material cost was cut by nearly 75%. Further, with 80% of theoretical yield to propylene
glycol and glycerol, production costs drop signiﬁcantly as well. With an assumed
distribution of 0.8 moles of product per mole of glucose fed for both PG and GO (1:1),
0.34 pounds of PG and 0.41 pounds of GO would be produced per pound of glucose fed.
Separately, the cost of each would be $0.44 per pound PG and $0.37 per pound of
glycerol. Combined, the production cost would be only $0.20 per pound. Even with raw
material costs at a conservative 50% of the total process expense, PG and GO could be
manufactured for $0.40 per pound. Thus, the potential exists for this process to be

competitive with current methods of PG and glycerol manufacturing.

148

8.4 - Future Work
8.4.1 - Catalyst screening for alkaline degradation with hydrogenation

In order for the alkaline degradation with hydrogenation reaction to be successful,
a catalyst needs to be found to selectively hydrogenate C3 intermediates as they form to
produce the desired products of propylene glycol. Presumably, a catalyst with particular
pore sizes would allow the smaller C3 molecules to enter the freely while excluding the
C6 feedstock molecules. Said catalyst would also have metal dispersed within the pores
to hydrogenate the intermediates as they enter the catalyst and allow the products to
difﬁlse to the bulk liquid phase. While molecular sieves have been already used in
similar size-selective reactions,32 the catalysts used in that work are unstable at high pH.
Other supports, such as carbon, are stable at high pH but are difficult to produce with -
precise pore size distributions necessary for the reaction. As catalyst technologies

improve, this option will be more viable and worth further study.

8.4.2 — Hydrogenolysis of carbohydrates to methane

Conversion of sorbitol, glucose, and glycerol showed high yields to methane
(>85% of theoretical) with only the carbohydrates and hydrogen as raw materials,
standard catalyst (5 wt% Ru/C), and typical hydrogenolysis conditions (200°C, 1200
psig). Continued work should focus on ﬁrrther improving the methane yield to at or near
100% by changing reaction temperature and pressure and observing the effects on the gas
phase products.

Further economic analysis of glucose to methane is warranted as well. While

converting purchased glucose to methane does not show economic potential, sugars

149

produced on-site would provide a suitable feedstock at a more favorable raw material
cost. The offset in price may be sufficient for process considerations.

Conversion of glycerol to methane is also of ﬁrrther interest due to the small
number of intermediates (propylene glycol, ethylene glycol, ethanol, and l-propanol)
produced during the reaction. This would simplify mechanistic work of glycerol
cracking to methane and, more importantly, to propylene glycol. With the increased
production of biodiesel, glycerol as a by-product becomes a potential feedstock for

ﬁrrther processing.

8.4.3 - Catalyst characterization

Ruthenium on carbon catalyst was used as the primary catalyst for most of the
reactions presented in this work and in prior group work with sorbitol hydrogenolysis.4
Prior to experiments, the catalyst was pre-reduced to activate the metal and promote
catalytic activity. While the pre-reduction conditions remained consistent for each set of
studies, the effect of different pre-reduction conditions, however, was not examined.
Different oxidation states of the dispersed ruthenium metal may have an effect on the
overall product distribution or the rate of reaction. In addition, pretreatment of the
catalyst with alkaline solution may have an effect due to interaction with the cations
present. Surface techniques, such as x-ray photoelectron spectroscopy (XPS), may
provide insight on changes to the catalyst due to different pre-reduction conditions and
the effects of alkali metals. Temperature programmed desorption (TPD) would also
provide information on what temperatures and lengths of time would be suitable for pre-

reduction to yield consistent catalytic activity.

150

APPENDIX A

Rate equations for the kinetic model

151

Sixteen equations describe the kinetic model derived for the alkaline degradation —

hydrogenation system. Abbreviations and descriptions of rate constants are provided in

Tables A-1 and A-2.

Table A-1: Abbreviations used for rate constants and kinetic model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abbreviation Description
G Glucose
M Mannose
F Fructose
SO Sorbitol
MO Mannitol
GA Glyceraldehyde
PA Pyruvaldehyde
GA2 Glycolaldehyde
EG Ethylene glycol
PG Propylene glycol
GO Glycerol
LA Lactic acid
FA Formic acid
GLYA Glyceric acid
GLYA2 Glycolic acid
CD Condensation products
CZ Cannizzaro reaction
H2 Hydrogen
S1 Catalytic site — type 1
S2 Catalytic site - type 2
Gus: Fractional coverage .of
hydrogen on type 2 Sites
Ow Fraction of vacant type 1 sites
r Reaction rate
t Time
[ ] Concentration

 

 

152

Table A-2: Description of rate constants and abbreviations used in the kinetic model

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00:39am Description
kn; lsomerization of fructose to glucose
km: lsomerization of glucose to fructose
kp-M lsomerization of fructose to mannose
kw: lsomerization of mannose to fructose
kGM lsomerization of glucose to mannose
kM.G lsomerization of mannose to glucose
kmo Hydrogenation of ﬁ'uctose to sorbitol
kp-Mo Hydrogenation of fructose to mannitol
kaso Hydrogenation of glucose to sorbitol
kMMo Hydrogenation of mannose to mannitol
kFoA Retro-aldol condensation of fructose to glyceraldehyde
kc“: Aldol condensation of glyceraldehyde to fructose
keg” Retro-aldol condensation of glucose to glycolaldehyde
keno Aldol condensation of glycolaldehyde to glucose
kg-“ Degradation of fructose to formic acid
known Degradation of fructose to glycolic acid
kGAcz Cannizzaro reaction of glyceraldehyde to glyceric acid and glycerol
karma Hydrogenation of glyceraldehyde to glycerol
kGA-PA B-elimination of glyceraldehyde to pyruvaldehyde
km“) Condensation of glyceraldehyde to by—products
kPA—FA a-dicarbonyl cleavage of pyruvaldehyde to formic acid
kpA-pG Hydrogenation of pyruvaldehyde to propylene glycol
kpA.LA Benzylic acid rearrangement of pyruvaldehyde to lactic acid
kpAcn Condensation of pyruvaldehyde to by-products
kGA2-EG Hydrogenation of glycolaldehyde to ethylene glycol
kGA2.cz Cannizzaro reaction of glycolaldehyde to glycolic acid and ethylene glycol
kcmco Condensation of glycolaldehyde to by—products
K30 Adsorption equilibrium constant of sorbitol
KMo Adsorption equilibrium constant of mannitol
K00 Adsorption equilibrium constant of glycerol
Kpo Adsorption equilibrium constant of propylene glycol
KEG Adsorption equilibrium constant of ethylene glycol
Kn Adsorption equilibrium constant of hydrogen

 

 

 

153

~>®Nm.m®._ﬂzu_ OEIEJIHA— Elmu—mi—HMVH EIQv—mi—HEH—AKIEX... OIEv—vl "ﬂ“

N_H<0Hml<mvx+ ~>®Nm.~w®.HWHAOEI.n—¥+ Omlmv—V law/@mlzxiTHOHmlOv—mlﬁmﬂx<Olm¥+ Elmx+ Olmxvl NEW

NHQOH OIN<OX + _>®Nm.wm©.~mv_ OmtOv—lmzﬂ Olzv— +7; Olmv— 1.1—“mu”: 201.0% + 210v— + "_lOv—vl "Hm“

N

N—QOH ”~01ng + _>®Nm.—.N~®._HN<OH OMIQOJH «6

6m?

 

a
—> Nm.m . gl<nm "I
e No Ta 2 8%

N

L<£lNorll¢qe~l+_>®Nm.m®.—<&ooi<oo_u :6

.068

Hozﬂ OUIOEv—l —>®Nm.ﬂ®.A Tm”— OEIu—v— +_H2H_0212v—v "_Hlmvuwmmm
a

HOW”— QUIOmX l ~>®Nm+N~®A T; Omlmx+ﬂmvﬂ OmIOv—v "a

154

8
_02 Boss +_Om_ oolomximeﬁ no- Que—+7}; colic—Leg codex njlmu e

38-..... E -. . ..
N NU N<Ox E>AOHU

 

N i a
N_<o_ No-68. 1 Tie?

 

THU— <mln~¥ +_H<n¢ (mlxamv— N qua—H U

l :6
To»; .3 <3— "Ev:

Soamiﬁ. 2-56.1?0. <el<ex+_§:8-<&_+ E43: 3-:le u _mm?

at
N _N<O_ 512067 _>®Nm.m_®._m<£ 29. l_o_ 96-00—ng 8.297 n 20 e

:6
seamiéﬁ 8-61.; are: Lézeéei NoleeviTexoolei idea. i To?

155

APPENDIX B

Visual Basic coding used for kinetic model calculations

156

The numerical method coded was adapted from Applied Numerical Methods by B.
Camahan, H. Luther, and J. Wilkes, published in 1969 by Wiley. Mathematical notation

is consistent with Camahan et al.

Sub RungeKutta4()

Dim rowvar(1 To 100) 'Row location of dependent variable
Dim y(1 To 100) As Double 'Array for dependent variable values
Dim k1(30) As Double 'kl, k2, k3, and k4 are the

Dim k2(30) As Double 'coefficients for the 4“ order Runge-
Dim k3(30) As Double 'Kutta method

Dim k4(30) As Double

Dim tempy(100) As Double 'Holds temporary values of y( )

Dim expnames(1 To 30) As Variant 'Used for labeling Excel worksheets
Dim cao(19) As Double 'cao-oso hold initial values of

Dim cbo(19) As Double 'dependent variables

Dim cao(19) As Double

Dim cdo(19) As Double

Dim ceo(19) As Double

Dim cfo(19) As Double

Dim cgo(19) As Double

Dim cbo(19) As Double

Dim cio(19) As Double

Dim cjo(19) As Double

Dim cko(19) As Double

Dim clo(19) As Double

Dim cmo(19) As Double

Dim cno(19) As Double

Dim coo(19).As Double

Dim cpo(19) As Double

Dim cqo(19) As Double

Dim cro(19) As Double

Dim cso(19) As Double

Dim.repchk(100) As Double 'Array to keep track of random.walks
Dim parameters<50l As Double 'Constants used in differential eqns

Dim rowform(l To 100) As Single 'Array to note rows with dependent vars

walks = Worksheets("Set Up").Cells(19, 10).Value
'Number of random.walks to take before stopping the program

runs = Worksheets("Set Up").Cells(13, 10).Value
'NMmber of experiments to solve differential equations for

summary_col = 1
summary_row = 0

'Position to start recording parameters on the ‘Summary' sheet

jump=0
'Used to move from.row to row when storing experiment data

157

For i 1 To runs
expnames(i) = Worksheets("Experiments").Cells(1 + jump, 1).Value
cao(i) = Worksheets("Experiments").Cells(3 + jump, 2).Value
cbo(i) = Worksheets("Experiments").Cells(4 + jump, 2).Value
ceo(i) = Worksheets("Experiments").Cells(5 + jump, 2).Value
cdo(i) = Worksheets("Experiments").Cells(6 + jump, 2).Value
ceo(i) = Worksheets("Experiments").Cells(7 + jump, 2).Value
cfo(i) = Worksheets("Experiments").Cells(8 + jump, 2).Value
cgo(i) = Worksheets("Experiments").Cells(9 + jump, 2).Value
cbo(i) = Worksheets("Experiments").Cells(10 + jump, 2).Value
cio(i) = Worksheets("Experiments").Cells(11 + jump, 2).Value
cjo(i) = Worksheets("Experiments").Cells(12 + jump, 2).Value
cko(i) = Worksheets("Experiments").Cells(13 + jump, 2).Value
clo(i) = Worksheets("Experiments").Cells(14 + jump, 2).Value
cmo(i) = Worksheets("Experiments").Cells(15 + jump, 2).Value
cno(i) = Worksheets("Experiments").Cells(16 + jump, 2).Value
coo(i) = Worksheets("Experiments").Cells(17 + jump, 2).Value
cpo(i) = Worksheets("Experiments").Cells(18 + jump, 2).Value
cqo(i) = Worksheets("Experiments").Cells(19 + jump, 2).Value
cro(i) = Worksheets("Experiments").Cells(20 + jump, 2).Value
jump = jump + 39
Next 1

'The for-next loop collects the initial boundary conditions, i.e. the

'feedstock concentrations of the reactions, and stores them.in the
'corresponding arrays of caot ) to cro( ). more initial conditions
'are possible, but separate arrays must be included for the program
'to run properly

'START OF COMPUTATION LOOP
For current_step = 0 To walks

Worksheets("Set Up").Range("jl7").Value
'Displays current step on worksheet

current_step

For pmer = 1 To 50

parameters(pmer) = Worksheets(“Set Up").Cells(26 + pmer, 8).Value
Next pmer

'Stores current parameters into array ‘parameters'
check = 0

'flag to make sure a random.step has not been taken yet. If checkul,

'then the program.continues. Otherwise, the loop repeats

Do While check = 0
Randomize
randoml = Int(40 * Rnd + 27)

randomZ Int(2 * Rnd)

repchk(no_imp) randoml * 10 + random2
'Generates a randem.number representing a random.walk and stores the
'result in the array ‘repchk'

158

If no_imp = 0 Then
GoTo 10

End If
'If the random.walk is the first since an improvement, then there is
'no need to check for a repeat

For i = 0 To no_imp - 1

If repchk(no_imp) = repchk(i) Then
check = 0
i = 10000

End If

If i = no_imp - 1 Then
check = 1

End If

Next 1

'The for-next loop compares the current walk to the recorded walks.
'If there is a match, then another random.number is generated. ‘With
'no match, check-l and.the program continues

Loop

10 Worksheets("Set Up").Range("j24").Value = no_imp
'Displays number of steps taken since an improvement was observed

If random2 = 1 Then
Worksheets("Set Up").Cells(randoml, 8).Value = _
Worksheets("Set Up").Cells(randoml, 8).Value * 1.1
Else
Worksheets("Set Up").Cells(randoml, 8).Value _
Worksheets("Set Up").Cells(randoml, 8).Value * 0.9
End If
'Changes the chosen parameter up or down a set percentage (10% in
'this case)

Worksheets("Set Up").Cells(randoml, 2).Value = _
Worksheets("Set Up").Cells(randoml, 8).Value
'Replaces the altered.parameter to be used for calculations

159

For reaction = 1 To runs
Worksheets("Set Up").Range("jl6").Value = reaction
'Displays the number of the experiment being calculated

'I of dependent variables that match the differential eqn

j
jj = 0 'I of differential equations

Worksheets("Set Up").Range("b9").Value = cao(reaction)

Worksheets("Set Up").Range("b10").Value = cbo(reaction)
Worksheets("Set Up").Range("b11").Value = ceo(reaction)
Worksheets("Set Up").Range("b12").Value = cdo(reaction)
Worksheets("Set Up").Range("b13").Value = ceo(reaction)
Worksheets("Set Up").Range("b14").Value = cfo(reaction)
Worksheets("Set Up").Range("b15").Value = cqo(reaction)
Worksheets("Set Up").Range("bl6").Value = cbo(reaction)
Worksheets("Set Up").Range("bl7").Value = cio(reaction)
Worksheets("Set Up").Range("b18").Value = cjo(reaction)
Worksheets("Set Up").Range("bl9").Value = cko(reaction)
Worksheets("Set Up").Range("b20").Value = clo(reaction)
Worksheets("Set Up").Range("b21").Value = cmo(reaction)
Worksheets("Set Up").Range("b22").Value = cno(reaction)
Worksheets("Set Up").Range("b23").Value = coo(reaction)
Worksheets("Set Up").Range("b24").Value = cpo(reaction)
Worksheets("Set Up").Range("b25").Value = cqo(reaction)
Worksheets("Set Up").Range("b26").Value = cro(reaction)

'Displays initial boundary conditions of differential equation

Worksheets(expnames(reaction) & " Output").Cells(2, l) =
Worksheets("Set Up").Range("indepvar").Value
'Displays the independent variable on the Output Sheet
Worksheets(expnames(reaction) & " Output").Cells(B, l) =
Worksheets("Set Up").Range("initialindep").Value

'Places initial value of independent variable on the Output Sheet
rowl = ActiveWorkbook.Names("startingpt").RefersToRange.Row + 1
columnl = ActiveWorkbook.Names("startingpt").RefersToRange.Column

'Used as reference point for locating variable names

columnvar = columnl + 1
'Column number of all the variables (column B)

160

For i = 0 To 100

If Worksheets("Set Up").Cells(rowl + i, column1)) Like "d(*" Then
jj = jj + 1
'Cells with d(*)/dt are counted as differential equations

If (jj > 20) Then
notice = MsgBox("You have more than 20 Diff Eqns. The program can
only solve 20.", vbOKOnly, "Error")
End
End If
'If there are more than 20 diff eqns, then end the program“ More
'equations are possible, but this version is limited to 20.

rowdeq(jj) = rowl + i
'8tores the row number of the differential equation in ‘rowdeq'

temp = Worksheets("Set Up").Cells(rowl + i, columnl).Value
temp Right(temp, Len(temp) - 2)
temp Left(temp, InStr(temp, ")"))
temp = Left(temp, Len(temp) - 1)
’Finds the dependent variable of the differential equation

For k = 0 To 100

If (Not IsEmpty(Worksheets("Set Up").Cells(rowl + k, column1)))
And (Worksheets("Set Up").Cells(rowl + k, column1))
Like temp Then
j = j + 1
'If the dependent variable of the differential equation matches
'a variable listed on the set up page, Q of dependent variables
'increments by l

rowvar(j) = rowl + k
'Stores row number of dependent variable in ‘rowvar'

z(j) = Worksheets("Set Up").Cells(rowl + k, columnl + 1)
'Stores initial value of jth dependent variable

Worksheets(expnames(reaction) & " Output").Cells(3, 1 + j) = _
Worksheets("Set Up").Cells(rowl + k, columnl + 1)
'Displays initial value of dependent variable on the output
'sheet of the corresponding experiment

Worksheets(expnames(reaction) & " Output").Cells(2, 1 + j) = _
Worksheets("Set Up").Cells(rowl + k, columnl).Value
'Displays name of dependent variable on the output sheet of
'the corresponding experiment
End If
Next k
End If

Next i

161

If (jj <> j) Then
notice = MsgBox("The program cannot continue because differential
equations were not found for every dependent variable.", vbOKOnly,

"Error")
End

End If
'If the O of differential equations does not match the number of

'dependent variables, then the program.ends with an error

Worksheets(expnames(reaction) & " Output").Range("A1") =
expnames(reaction) & " Output"

Worksheets(expnames(reaction) & "
Output").Range("Al").HorizontalAlignment = leeft
Worksheets(expnames(reaction) & " Output").Range(WAl").Font.Bold = True
Worksheets(expnames(reaction) & " Output").Range("Al").Font.Italic =
True

Worksheets(expnames(reaction) & " Output").Range(WA1").Font.Size = 12
Worksheets(expnames(reaction) & "
Output").Range("2:2").HorizontalAlignment = xlCenter
Worksheets(expnames(reaction) & " Output").Range("2:2").Font.Bold =

True
'Formats experiment output sheet for results of calculations

Stoprow = 3 + Worksheets("Set Up").Range("numsteps")
'Stops iteration desired.amount of time steps have been taken

h = Worksheets("Set Up").Range("stepsize")
'Step size of time. Consistent with Carnahan

CurrentRow = 3
'Starting point on Output sheet

initial = Worksheets("Set Up").Range("initialindep").Value
'rirst time point of the calculation

162

”MAIN'CALCULAIIGN LOOP

Do While CurrentRow < Stoprow
t = Worksheets(expnames(reaction)

columnl).Value
Worksheets("Set Up").Range("initialindep").Value = t

& " Output").Cells(CurrentRow,

For i = 1 To j

y(i) = Worksheets(expnames(reaction) & " Output").Cells(CurrentRow,
columnl + i).Value
Worksheets("Set Up").Cells(rowvar(i), columnvar) = y(i)
tempY(i) = y(i)
Next i

'ror-next loop takes values of dependent variables at time t and
'stores them in y(). Then, copies those values to the ‘Set Up' sheet
'so that derivatives can be calculated from.formulas entered by the
'user. Then, tempyl) stores initial y values to be used in Runge-

'Kutta coefficient determination

For k = 1 To j
k1(k) = Worksheets("Set Up").Cells(rowdeq(k), columnvar).Value

tempy(k) = y(k) + h / 2 * k1(k)

Worksheets("Set Up").Cells(rowvar(k), columnvar) = tempy(k)
k2(k) = Worksheets("Set Up").Cells(rowdeq(k), columnvar).Value
Next k

'Calculates the coefficients kl and k2

For k = 1 To j
tempy(k) = y(k) + h / 2 * k2(k)

Worksheets("Set Up").Cells(rowvar(k), columnvar) = tempy(k)
k3(k) = Worksheets("Set Up").Cells(rowdeq(k), columnvar).Value
Next k

'Calculates the coefficient k3

For k = 1 To j
tempy(k) = y(k) + h * k3(k)

Worksheets("Set Up").Cells(rowvar(k), columnvar) = tempy(k)
k4(k) = Worksheets("Set Up").Cells(rowdeq(k), columnvar).Value
Next k

'Calculates the coefficient k4

'For each coefficient calculation, concentration and derivative
“values must be refreshed to ensure proper calculation

163

For k = 1 To j
If (Not IsNumeric(Worksheets("Set Up").Cells(rowdeq(k),
columnvar).Value)) Then
msg = "There was a problem with the calculation.
msg = msg & "At least one derivative"
msg msg & " has become non-numeric due to an error."
msg = msg & " You will need to reinitialize all dependent
variables after"
msg = msg & " you find the error. Be sure to reinitialize the"
msg = msg & " independent variable also."
notice = MsgBox(msg, vbOKOnly, "Runtime Error")
End
End If
Next k
'If any of the differential equations becomes non-numeric, the

'program.is stopped

CurrentRow = CurrentRow + l
'Increment the row number to allow proper placement of calculated
'value of dependent variable

Worksheets(expnames(reaction) & " Output").CellleurrentRow,
columnl).Value = t + h
Worksheets("Set Up").Range("initialindep").Value = t + h
'Prints out next time step on Output sheet and.updates the time on
'the Set Up sheet

For i = 1 To j
Worksheets(expnames(reaction) & " Output").Cells(CurrentRow, columnl
+ i) = y(i) + h / 6 * (kl(i) + 2 * k2(i) + 2 * k3(i) + k4(i))

Next i
'Predicts next value of dependent variable using the Range-Kutta
'algorithm.from.the calculated coefficients and displays the result
'on the Output sheet

Loop
'Repeats the loop until the last time step has been taken for the
'experiment

PAdds final values of dependent variables to Runge Kutta Sheet
For i = 1 To j

y(i) = Worksheets(expnames(reaction) & " Output").Cells(Stoprow,
columnl + i).Value
Worksheets("Set Up").Cells(rowvar(i), columnvar) = y(i)

Next 1

'Adds the final values of the dependent variables to the Output sheet
'and the Set Up sheet

Worksheets("Set Up").Range("initialindep") = initial
Worksheets("Set Up").Activate
'Resets the independent variable to its initial value

Next reaction
'Continues on to the next reaction for calculation

164

'SECTION TO COMPARE CALCULATED RESULTS WITH EXPERIMENTAL DATA.AND TO
'DETERMINE Il‘ RANDOM m IMPROVED noon. OR NOT

jump3 = 1

'Place holder for finding variables on Experiments worksheet
For reaction3 = 1 To runs
p=l

'3 of the data point from.the experimental data

For q = 0 To 1000
'Larger qmax would.be necessary for more than 1000 steps in the
'process. For the work done with the model, 1000 was more than

'sufficient.

If IsEmpty(Worksheets("Experiments").Cells(21 + p + jump3, 1).Value)
Then
Exit For

End If
'If there are no more experimental points, exit the for-next loop

If Worksheets(expnames(reaction3) & " Output").Cells(3 + q, 1).Value
>= Worksheets("Experiments").Cells(21 + p + jump3, 1).Value Then

For i = 1 To j
Worksheets("Experiments").Cells(21 + p + jump3, 20 + i).Value = _
Worksheets(expnames(reaction3) & " Output").Cells(3 + q, i +

1).Value
Next i

p = p + 1

End If
'If the independent variable on the output sheet is >- the
'independent variable from the Experiments sheet, then print the
'dependent variable results from.Output on to the Experiments
'sheets for comparison. Increment p to look for the p+1th data

'point.
Next q

jump3 = jump3 + 39
VMbves place holder from.one experiment to the next

Next reaction3
'Repeat for all reactions to be calculated

Worksheets("Set Up").Activate

If parameters(4l) < (Worksheets("Experiments").Cells(9, l6).Value +

0.01) Then
For pmer = 1 To 40
Worksheets("Set Up").Cells(26 + pmer, 8).Value = parameters(pmer)

Next pmer

no_imp = no_imp + l

165

If no_imp = 80 Then
notice = MsgBox("All random walks have been taken without
improvement.", vbOKOnly, "Error")
End

End If

End If
'If the total error from.the latest iteration is greater than the
'best case, then reset the parameters to the best case and try
'another random.walk. However, if all of the walks have been taken,
'then the program.ends.

If parameters(41) > (Worksheets("Experiments").Cells(9, 16).Value +
0.01) Then

no_imp = 0
Worksheets("Set Up").Cells(67, 8).Value =
Worksheets("Experiments").Cells(9, 16).Value

For k = 0 To 39
Worksheets("Summary").Cells(2 + k + summary_row, summary_col).Value
= Worksheets("Set Up").Cells(27 + k, 1).Value
Worksheets("Summary").Cells(2 + k + summary_row, summary_col +
1).Value = Worksheets("Set Up").Cells(27 + k, 2).Value

Next k

Worksheets("Summary").Cells(43 + summary_row, summary_col).Value =
"Grand Average"

Worksheets("Summary").Cells(43 + summary_row, summary_col + 1).Value
= Worksheets("Experiments").Cells(9, 16).Value

summary_row = summary_row + 50
If summary_row > 65000 Then
summary_row = 0
summary_col = summary_col + 3
End If
End If
'If the total error from.the current iteration is less than the best
'case by more than 0.01 then the new set of parameters is copied to
'the Set Up sheet as the next starting point for random walks. The
'new, lowest total error replaces the old result on the Set Up sheet.
'Eurther, the new’parameters are recorded on the Summary sheet with
'the total error associated with it. If there are over 1300
'improvememts, then the parameters will be reported on the Summary
'sheet 3 columns over to avoid an error.

Worksheets("Set Up").Activate

Next current_step
'Repeat the procedure until the number of desired random walks are
'taken or until all random.walks have been taken without improvement

End Sub

166

APPENDIX C

Comparison of kinetic model to experimental data for alkaline

degradation with hydrogenation

167

5-0.2.3.:
$8.02 0
sowii
38-8 x
5-2 ............
38-5. u
5-0.1}.
88-0 4
all

Emmi“. 0

0:5. .m> 8:588 was. 483.5% 68:58 .8853 .omoaoEm

03 $3 u mean 8.0 on was 2: .008 £on .32 emerge as: .m
w: as - loeoa 883 as see 35588 do 82858 "to 9...»...—

AmSo£ oE_._.

 

 

 

 

 

 

/ .
0.0?

 

O
O
N

 

/
o e
IO
N

/v
0.0m

 

 

 

 

0.0m

(“116) uogeriuaouoo

168

redial...

Sac - (>40 X

e - on. ............

38.9. n

E-OO.III

38-06 4

E - (4 III.

38-5 6

Sum. .m> Boa ouoobw one .39an ocoaooa ..oBobm .28 0.525
03 are n mega 3o or was 2: .99. .mom some .8633 as .m
n: as - _oeoe erode e5 as 385.88 to eosaaaoo ad 25»;

A2305 oE_._.

 

 

 

 

 

 

 

o o v m N _. o
L p _ e ............................................ w ............... I 0.0
.................................................................... i-e.“
x x m ....... Imlnlmlittmte\\
IPI lllllllllllllll IDI lllllll 41-“4‘ .
q |¢ IHlll ON
4 IWI lllllll ..I 0
ll 0
o 3 m
0
a
U
15..
co m
n"
O
U
0 ...}
o.w m
o h

 

 

oNF

169

08E. .m> Boa oncobm Ea .Um do“ ego.“ £2605 couamcoccoo

03 are m mesa 3 or med 2: .008 .mom son .8683 :3
8: ea. - 388 88.2 e5 38 3888?... no 88888 "no as»:

$505 one...

 

 

 

 

 

 

 

e e e m
4 4 4 4 ..i
x x xx ..........
eras/50.1... .....H.H....u.......u ....... d- ...... m thrive...
D ‘‘‘‘‘‘‘‘‘‘‘
sou - «536 x
. m.
E - 0w ............ o u
o 3 ..w
38 - em a W
o m
Eidill e o . m.
- 3 u
38 - E 4 ﬁle
in
E - DUI .
om
Sac - 00 o

 

 

 

ON

170

5.02.1.1
9.3.02 o
sowii
38-0w x
5-2 ............
3%-: u
5-0.1::
386 <
eful

8%-”. o

05E. .m> BEES: 98 49388. 68588 .8853 .omoaoem

03 are 8 mean 2 .o or med 2: .008 .mox son .8363 .88;
o: as - 388 8er as see 388.898 do 88.5.38 "to as»...—

Amaaoé oE_._.

 

 

 

 

 

 

O
o
u
o
a
u
1?
co m
D
0
U
\Il
cs Bi
1
(
0.2

 

 

 

 

o.N_.

171

5-5.40-1...

Soc - (>16 x

5 - on. ............

58-05 a

E-OO.III

Sac - 00 4

E - <._ I

38.5. 6

0:5. .m> 28 2.82pm p8 Aoobw 08235 Aobobw .28 2584

cam .85 w 858 S .o .8 med 2: .008 .mox :8 .8885 .585
o: as - 385 88.5 5... see 3858.8 .8 858558 5.0 .5»...—

3505 0:5.

 

 

 

 

 

 

 

 

 

 

 

 

C)

 

 

o. v. <51 0. co. o. v. cs 0.
1- 1- ‘- 1- O O O O Q
(116) uogenuaouoo

°Q
,_

O.
N

172

E- N<>._0..I.I

Soc - N<>40 X

5 - om ............

Soc - 0w D

E-<n_.lll

38-5 d

E- 001111

58-8 6

229. .m> 28 2.82m 28 .0m .28 2:58 .82.qu 5255280
08 83 m 855» o; .8 med 2: .008 .mox son .8385 88.2
o: as - 865 525 ea. 88 585.588 re 8.8980 so 9...»...—

AmSoé oE_._.

 

 

 

 

 

 

 

 

 

 

o n v m
1* _ . .
N D
DI ................ D llllllll .4 ..IlII..|
\ ed
0 o o
o o o
\ O O m o
o...
O
O \
\ we

 

 

v...

(116) uonenuaouoo

173

E-OS_.I..I
88.02 0
Ecol...
98.8 x
5-2 ............
8.8-2 a
5-0.111
58-0 d
E-ul

Sand. 0

22H .m> 82588 28 4938.... 80588 885% .omoaoam

08 83 m 88 as ..m see as .008 .58 .88 .8205 x: ....
2: ea - 885 5er as 58 385.8%... 56 8.888 .5 2.85

$505 25...

 

 

 

 

 

 

 

0.
IO
N

 

 

 

0.0m

 

- 0.0m

(116) uogtenuaouoo

174

E-<>u_0..l.l

Emu - (>40 x

E - on. ............

9%-9. u

5-0.9::

33-00 4

E - <._ I

9%-5 0

0:5. .m> Boa 0.523 93 483w 283905 4089mm .28 38..—
03 53 m 85% ed am ya can .98 £8“ $3 .3303 $52
03 9Q - ~82: owe—HE 28 8% 35%“. mo nonnamﬁoo "ad 95»:—

AmSoé wE_._.

 

 

 

 

 

 

 

 

 

 

o.o

0. o. o. o. o. o.
(0 l0 '6' (0 N F
(1/6) uogmnuaouoo

0.
l‘

0.x

175

E- N<>-_0..I.I

Emu - N<>-_0 X

E - 0m ............

3%-8 a

E-<n_i-

Emu .. (n. 4

E - DO|

.320 - DO 0

«SE. .m> Boa ozoobw 2% .0m .28 2.58 .3038.” :ouamcowcoo

03m $3 w mean 26 am mam o8 .008 .mom :3 .8225 32.».
as ea - E55 0.50:: Ea 3% 3555ng mo 8.9.38 5.0 2:5

3505 «EC.

 

 

o o v m N F o
W X W _ .. .......- ..:.-.. .....- X . od
...................... x..-..........-..... x x m by.
q ......................................... q m. a ..... .H.\.\\
m Ln luff--.. ...... o no
DI

............ hail-Hi- --.........- \
I:- o 3

 

 

\ 0 m...

 

0 ca

 

 

\ o”.

O
3
(116) uouenuaouoo

 

 

 

oé

 

 

0v

176

5-0—zit.-
8mv-0_2 o
5-8;!-
3mu-0m x
5-2 ............
327.2 B
5-0.:-
saoé <
awful

3%-”. o

08F .m> “BEE 98 40:88 6858.: .36on .omogoam

03m .sz n ma 2 .o a: man 2: .008 .mom 3% 4.1.622MW .\E .m
E axm - ESE 80$ Ba 3% 355.5%... .8 8.338 55 as»:

$505 95...

 

 

 

 

 

 

 

 

 

 

1/ ed. 0
«I, m
[4]! m
I. .
on. u
A, n-
/ e
(I n
u. . O
a, com u
a )
, 6
x can W
’
a
f
” 0.8
’
4

 

 

 

0.0m

177

u_u—-<>-_0..I.I

San - (>40 X

E - on ............

Emu-0n_ a

5-00.:-

9360 4

E - <-_ l

3%-5 o

05.5. .m> Boa ouoobm v5 Acobm 0:23on 4089mm .28 3.83
03 :3 m 25% S .o am mg 2: .008 .mox :2 .0883 $25
3 ea. - 3on 0.582 as as @5589"... go 8.5958 ":6 as»:

$505 95...

 

 

 

 

 

x x x x- ......... \-
I ......................................... :D‘
............................... ﬂ-\\\\
.- Ill-“...I‘JDPII o
..I-t-W
Bull...
0
4
0 4
0
Q

 

 

 

 

 

 

 

06

0.0

o.®

ON

Qw

(116) uogwnuaouoo

178

E-N<>40..I.I

San - N<>._O X

E - om ............

3%-8 u

E-<n_.lll

Emu - (u. 4

E - DO-III

San - DO 0

08m. .m> Boa 0:093 98 .0m .28 28.8.“ £2605 couagvcoo

03 :3 n mas» o; am mug 2: .008 .mom :2 .382» $2 .m
3 ea - 32: 08:2 as as Eaves?» no 8.938 "2.0 95»:

3505 «EC.

 

 

. o.o

 

0.0

 

o...

 

m...

 

0N

 

0N

 

o.m

 

Om

 

 

oé

 

 

0v

(116) uogumuaouoo

179

562-..-
Emu-o_2 o
5-8;!-
3%-ow x
5-2 ............
3%-: a
5-0.:-
336 <
girl

3%-“. 0

«ES. .m> 6:538 25 ..ouBHOm 68:58 .3853 .umouoam

03m .on m was» 2 .o ..m ”a 2: .008 .mox xmw .8858 :5 .m
N: gxm - ESE $8.2 Ba 3% 388.593 go 8....538 “2.0 2:»:

$505 95...

 

 

 

 

III
II
I
l

 

 

 

 

 

 

 

0.0..

0.9.

0.
o
N

Q
If)
N

0.0m

(116) uonenuaouoo

180

oEF .m> Boa onoobw 98 Acobw 28335 488$» .38 303

93. .35 n 383m 26 am mg 8. .008 .mox .32 .3053... 3: .m
a: .5 - 385 03:2 8. 33 353.5930 83.3980 35 as»...

$505 95...

 

 

5-350;!-

3% - <50 x

 

E - an. ............

 

Emuén. n

 

E-OO.III

 

San - 00 4

 

E - (4 ll

 

33-5 0

 

 

 

 

ed

(116) uonenuaouoo

181

E- N<>-_O..I. I

Emu - N<>40 X

5 - om ............

58.8 a

5-57-

8.85. 4

E - OOIII-

9.8.00 0

0:5. .m> Eon omoobw Es .0m .28 3:58 .33on9 scram—86:00

03 .on m 855 o; am 55 2: .98 .:0x 3% .3855 3: .m
a: 5m - 585 882 55 55 3555ng .5 835558 “2.0 9...»:

$505 «EC.

 

 

 

 

 

 

 

 

 

0.0

‘0.
o

q
...

‘0.
,.

0.
N

ON

(116) uognmuaouoo

182

5-0.2.:-
Bmu-o_2 o
5-8;!-
Bmv-om x
5-5. ............
38.5. n.
5-0.:-
386 <
“_ﬁull

Numb-n. 0

25¢. .m> 6:558 95 48558 68558 683% 6883"“

055 .38 3 355m 2 .o ..m m8 2: .005 .mox 38 .3355 go.“
a; 5m - 555 582 55 35 555.5530 53.5558 35 3.55

3505 «EC.

 

 

 

 

 

 

 

 

 

o.o

90
o
o
u

0.9 m
u
n-
e
D
o

0.2 u
)
6
I
1
(

98

 

 

0.,-“N

183

E-(>40..l.l

Emu - (>40 X

5 - on ............

38-03 a

E-O0.III

Sac - 00 d

E - (4 I

58-5 0

one. 43> Boa ouoobw Ea 409nm 054.8on 4883» .28 384

03 .38 n 553 e .o ..m 38 2: .95 .mox .38 .3035 38
3: 5m - 385 882 55 35 555553 .5 535558 5.0 3.3....

3505 «EC.

 

 

 

 

 

 

 

 

 

 

<>

 

 

0.
N

0.
co

Q
<1-

oh

0.0

(116) uogumuaouoo

184

E- N<>.._0;I.I

Emu - N<>40 X

E - om ............

3%-8 u

again:

3%-5. q

E - DUI

San - DO 0

0:5. .m> Boa 239nm 98 .0m .28 "EEO.“ nmu‘oﬁvoa couagvcoo

03 :E n ma 2 .o “mm mi 2: .008 .mom :3 .8903 $2
3; axm - ESE 26:2 98 Saw Egggo mo :OmEQEoU “3.0 959%

3505 25...

 

 

 

 

o o v m N _. o
_ . _ . _ . ........... I a 0.0
d 4 < < .......-.-..< ........ -G ........ 4 EV
x x «-x .......... x x ... . ...\
....................................... II n loll ‘;\\ 0
. ............. m- . I I I . .U. ... ... ....t.
||||||||| mnII'Illd lllllll'lllllul o
n. ‘‘‘‘‘‘‘‘‘‘‘‘ .
II o o
0
o
0 o _.
o
o
o 0...
ed

 

 

 

 

mN

(116) uoneuuaouoo

185

y_u_-oz...:u
3mv.05. o
“_ﬁomii
3%-ow x
5-2 ............
38.5. n.
5.0:}.
38-0 4
“ad—III

3%-“. 0

25h .m> 8:588 98 40:88 6858.: 68on 0338$

93 $3 n 25% 2d am mg 2 .008 .mox $3 .0883 $5;
3 ﬁrm - 32: 30:2 Ea 3% 3855?” ‘8 soggaoo as as»:

3505 «EC.

 

 

 

 

 

0.0..

 

/ 0.2

 

/
O O
to O
N (\l

 

 

 

 

 

Qmm

('1/6) uommuaouoo

186

Eiﬁoii

Ewu - (>40 x

E - on ............

3%-? u

E-OO.III

Emu - 00 4

E - (4 III

33-3 0

0:5. .9 28 ouoobw Ea 483w 054305 4809mm 4:8 0483
03m $4.3 n ma 2 .o .Nm ma 2 .008 £84 $3 .8953 x: .m
E ea, - ESE 88.2 93 as 38%?» mo cogqaoo as 25:.

$505 «EC.

 

 

 

 

 

 

\ .
‘
\
\\\ \
\‘ 0
ii. on

 

 

 

 

 

'I.||l \0 O?
0.0
\\ 0
\ 0 OD
0.x.
0 o.»
o 3

 

 

 

 

0.9

187

(116) uommuaouoo

08E. .m> Boa 339nm 98 a8 .28 358m .326on :ouamcoccoo

03m .53 n 85m 26 am mi 8 .008 .mom :3 .0833 REM
E as - 3on 882 c5 3% Eﬁaﬁaxo no Engage :3 as»:

$505 25...

 

Ill
‘-II§IIOI\'|IOIIII|II|
....
I!|IOO\

 

...................................................... m a - .
BEE... ........... m..- m M m l.$<.i-\\nHH\m\c 0
5 . «<50. . u . .. an ....... ﬂ ....... H. ............ MHZE... L.\¢
«50-250 x nnnnnnnnnnn o o

 

5. 0m. ............ o o \

188

 

3%-8 n. \

 

0
E-<n_iuu \

mean .. (n— <

E- OO|

320-8 0

 

 

 

 

 

 

0N

QM

0.0

ad

(116) uonenuaouoo

5-0.2.!i
33.02 0
E-Om...:.l
3.20-8 x
5-2 ............
3%-: n
56...:
3%-0 q
“_ﬁull

3%-”. 0

0:5. .m> 40:58.: 98 40:88 “39:55 683% “888.5

03 :43 n mega ad .£ mum on .008 “mom $2 .8303 «Ea
9: ea - 40on 382 2a 38 355.5% mo 8.0.538 and as»;

$505 0.5...

 

 

 

 

 

0.0..

 

0.0..

 

 

odN

l
0.
to
N

 

 

0.0m

 

 

0.0m

(116) uogmnuaouoo

189

5-5.6.5:

Emu - (>40 X

E - on ............

Emu - 0n. D

5.00.111

Sac-00 4

E- (4 .III.

San-<4 0

05C- .m> Boa ouoobw 98 4033 2843on 40.53% .28 384
03 $5. ..- ...a 2d a: man on .008 .mom :2 .3325 x»; ,
9: ea - 4805 88: Ba 38 355.8ng .8 coma-meow and as»:

3505 «E:-

 

 

o w v m
X IIIIIII WW ....... ha I ...... 3| .......
Q 4 4 ...1.4. .................

 

 

 

 

 

 

 

 

 

 

 

o.m

06

0.0

0.0

OS

0d

ed

(116) uogmnuaouoo

190

25.4- .m> Boa omoobw 0:0 .0m .28 0:58 £00080 0000300000

03 .on n ma 2 .o am mug cm .008 .mom 3.1.0 .8325 :32
a: 90 - 4085 88.2 05 38 3558me 0o 8030600 and 2:5

$505 0E:-

 

 

o o v m N F o
........................................................... ma -
a ........ . ........... a. ............. m m B «1.1-mum.”
E-N<>._0.ii - 4 u.n<.!-.!...<ui- -I-u-n....-.11-
|||||||||||||| Illl‘lll'ulo
San-«<50 x 111111 o

 

a: I am ............ 0 0 \

191

 

3%- 0m 0 \

 

E..(n_.lll

\

 

Ban-E a \

(116) uonenuaouoo

 

ETD—0| \\

 

300.00 0 \

 

 

 

5-021;:
300.02 o
5.00;}:
300-0w x
5-5. ............
Emu-.2 a
5.0.1:
300.0 4
EJTIII
3%-“. o

005. .m> 40:58:“ 05 4338a 68:58 683% .omouoam

03 .03 m 8E» 80 am man 2: .008 £004 :2 .8380 :2 .m
an as - 4088 382 9a 3% 388593 mo 803800 "mu-0 2...»...—

Amhzocv 05:.

 

0.0

 

 

0.0

 

0.0..

 

0.0 _-

 

 

 

/
o 0.
IO 0
N N

 

 

 

 

0.0m

(116) uogmnuaouoo

192

E-(>40..l-l

Emu - (>40 X

E - on ............

San-0n. n

E-00.III

$00 - O0 4

E - (4 II

33-5 o

005. .m> 28 2.803» 05 409nm 084305 40$obw .38 0.0.03

03 :3 n mega 8.0 am man 2: .008 .mom :2 .3325 x: .m
«2 ea - 420% 882 Ba 3% 388505 MO 88.0500 ”8.0 2:»...—

3505 0.5...

 

 

 

 

 

 

 

 

o o v m N _- o
x x x x x x .1 ix..- x .0
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII ....v

E ..z

4 u

\ .... \
.. s
m . \
\
1 .. N
m ...- .
m. Vx‘d
s
m ans
\.
D \-.\ d
w \-\.
m \u... 4
\\ \ \H-\----<
II III.- ..I I!!! 1 .Q\ 1
II 1' [III II Illl 6111 111111
I|l lul | os<u uuuuuuuuuuuuuuuuuuuuu

4 ...................

 

 

0.0

o..-

0.
N

0-
co

0.
V

0.0

0.0

(116) uogmnuaouoo

193

E-N(>40.-I.I

Emu - N(>40 X

E - 0m ............

San-0m n.

Eéuiu:

Emu - (u. 4

E- QO|

Emu-00 o

025. .m> Boa oaoobm 05 .Om .200 0:000.“ £02005 2003:8200

03: 2:3 m mega :08 a: ma: 2: .008 .20: 2.2 .3325 2.2m
an: an: - E58 0.8%: 25 $8 388.805 0o 823300 "5.0 9...»:—

3505 05:.

 

 

 

 

 

 

 

 

 

 

 

 

m...

ON

ON

o.m

On

00

(116) uogmnuaouoo

194

APPENDIX D

Master list of experiments

195

 

230th2 some «a 855:. co .0.8N do:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_mNup .838. 22988. .2; 8:982 888 ..m 8.2 8:28 89 :0: o: e8 m 80.0 mm 8823
2380082 :08 8 meEE om .900: .mn .8 .8008 . _ . .
.mNuh .8858 23882.2 .2; 8:982 8&8 :0.F 00:28 000: IO: 0: E8 0 00 N mm 3323
23882.2 :08 E 8352.. 00 .903 .mn .om €308 . _ .
.8": .888. 22988: 2; 8:882 888 8.: 8:28 m8 :0: o: e8 m 00 m Nm S85
2888: 88 8 82:26 8 .082 .2. .8 .888 . . . .
.88 .838. 222852 52 8:982 828 2: 8:28 8.. :0: o: e8 a 8 N 5 2:28:
28.88. 88 a 8525 8 .082 .2. .8 .888 . . . .
...-Nu: .2852 222822 2; 82:88 898 2: 8:28 82 :0: o: e8 a 8 N 8 2:25
22288: 88 a ___E 8 .088 .8. .2: 8&8 . _ .
dun-r 83.8. 238888 5:: 8:982 828 8.. 8:28 RN :0: 0: E8 m om N mN 09329
23288: 88 a 22 8 0.98 .8. .2: .888 . . .
...-Nuh 8852 23288. 2; 88.88 828 8.: 8:28 82 :0: o: e8 a o 2 8 8328
22.288: 88 a 25 8 0.28 .8. .2: .888 . . _ .
.mNuh 88.8. 22238. 2; 8882 828 8.: 8:28 80.. :0: o: e8 m 00 m 0N 883
2328.8: 88 a 25 8 0.28 .2: .2: 8&8 . . . .
.8": .888: 83988: 5:: 8:982 828 00.F 8:28 moor IO: 0: E8 m 00 N mN 098‘:
22288: 88 a e: N .088 .8: .8. .888. . . . .
.mNup .8qu 222852 2; 82:82 828 2: 8:28 82 :0: o: e8 :0 m8 o ...N 28::
23988. 88 8 m2 N .088 .8: .8. .888 . . _ .
.mwuk .8858 28898: .2; 8:882 898 no; 8:28 009 IO: 0: E8 a mNo o NN 00:98
22288. 88 a e: N 0.08 .2: .2: .888 . . . .
.mmuk .9882 2328.8: 5:: 8882 828 no; 8:28 000.. IOx 0: E8 m mNo 0 ON 85:?
cozatommo 8:880 3:308: axm mama

 

 

196

 

CF:

 

 

 

 

 

 

 

 

 

 

.mNub .989... 238.392 5.? 8.8.8.2

 

 

 

 

 

8. .8. .8. .8 .... .8 .8 .8 .8. © 8.8 888 3. 8.... 58 8o. .19. 8.. .88 a Bed 8 .288
.98.". 88.8 .o 8:828: 8.82
9:8 8 855:.

SN .98. 8 ...E 8. .088 .8. n F .88 8:88 3. 8.... £8 8o. .19. 8.. .28 m 88 8 .288
:8 mEoEm. 2.8.qu8 5.; 8:8 8.8.8.2
928

8 8.8.5 o... .98. .... se 8. .088 .8. 888 am 88. 58 8o. .19. o: .88 m 88 mm .298
u ._. .889... 938892 5.; >88 88.8.2
088

... 88:: 8. .98. 8 ___... 8. .088 .8. 888 3. 8.... 58 8o. .19. o: .98 m 88 8 2.28
n ... .9899 0.2.898. 5.; 8:8 8:982

.88.... 8. u . .98. n F .88 8:982 888 3. 8.... £8 8o. .19. o: .98 m 88 m... .228

8.8:: 8. n . .98. u .. 288 8:982 8:88 3. 8.... £8 8o. .19. o: .98 a 88 3. .288

8.8:: 8. u . .98. u .. .88 8:282 888 3. 8.... 58 8o. .19. o: .88 a wood 8 .288
N1 2288 268. o. 0.8.. 8 888.

888 .98. 88 ... 8.. 8 .088 .28 888 am 88. £8 8o. .19. 8.8 .28 a omod o.. .2828
.8. .8. .8 u .. 8.98. .8. 58 8.8 .2
22288. 88 8 8.8:: 8 .088

88 .8. .8. .8 u . .888. 2.8.88. 888 am 8.... .18 8o. .19. 8.8 .88 o 82o 8 .288
:8 88 5.3 .89898 8.982

9.8 88 .8. .8. .8 .8". . . .
8.8. 88:. s 88.8 8 2:88 8288 .828 oz 8:28 8o. 19. o: e8 8 28 o 8 .288
95 5.09:0 one a mo SE . o . . . .
. E . s . 8 o 8. 8. 888 3m 8.... 8:28 8°. 19. o: e8 a N8 o 8 .828

 

 

197

 

0.82 ewa 5? 85.2.82

 

 

 

 

 

 

 

 

 

 

 

 

3 - . . .
.2; v No.08. .NI 98 8m. 8< .2; o 3.; WNW. 8; £8 82 :9. 98 m e8 8 m on 385
no.8? .898 8 .8“. .5 285%... .8. 20
2852 8 in .2: .8; .8 .8 .8 © 8.8 03m . . . .
c on m m on m
.oooﬂuh 88.8 8 88888 2:22 82; 3 - 84 ._ 8F :9. mm m e m m 8 33:3
5:. 8... .8» den .08 6 98w . . . . .
.983 ._2_eow 3 88858 3.9.? 093% 3. 8 2 £8 82 :9. 8a m 98 8 3 8 38:.
255.8.v.m.~.r©nemw . . . . .
.oooﬂﬂ ..288 .6 88888 3.9.? 85% am 8 2 £8 82 :9. 8..” m e8 8 m 8 58::
.56 2: .8; .8 .8 Q 88m . . . . .
.oooﬂﬂ ..e§a 3 88888 3.9.? 88% 3m 8 2 £8 88 :9. 8 F .288 8 o 8 38::
.se 8; .8 Q 88m . . . .
.98qu .858 .8 888.8“. 05.8.? 8228 oz 58 89 :9. 8 r e8 8 0 mm 382:
.55 o2 .09 .8 3 98w . . . . .
.o.8.ﬂ .358 8 888.3. 2:32 8808 am 8 2 £8 82 10v. 8 F e8 8 o .8 362:
.55 S... 58 .8... .82 @ 98w . . . . .
.98?» 8:28 8 88928.. 2:22 888 am 8 8 £8 89 :9. 8 8 £8 8 3 8 353.
5:. 8v .SN .82 6 98m . . . . .
.98?» .358 8 8888: 2:32 888 ..m 8 8 £8 82 :0x 8 r 88 mm m 8 88:2
.088
a 888. 8:on .55 our .8 Q 3.53 088.9. 3m 8.2 £8 89 .19. o3 .98 88° 5 5822
.obﬂﬂ ..828 8 8:828“. 05.8.2
55 3 .8 Q 88m . . . . .
.985 ..288 s 888.8 2.32 88% 8 8 8 £8 89 :9. 8 F 98 8 m 8 383.
EE
8; .09 .88 .8 .3 .8 .218 .2 Q 98m 8:28 3. 8.2 £8 89 .19. 8.? .28 a oood 8 588

0.03"... ..oﬁtom 3 5:23.33 ¢c=S=<

 

 

 

 

 

 

198

 

.8.... 8F .8 .8 .8 a 988

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8: 8.8.. 88 8.888888. .2888

 

 

 

 

.988»... .8828. 8 8.88888 85.9.2 88.888 82 £8 88 :9. 88.8 88.8. 888 .8 88:88
.8... 8 .8 .F. .8 © 8.88 08.. . . . .
c 08 o a o. 08 a
0.88"» 8.8.8 8 8.88888 85.9.2 8...; 8 - 8.F ._ o8 :9. 88 8 _ 8.8. N 8 88 8:88:88
.8.... 88 .8 .8 © 888 . . . .
.985 .3828 8 8.8888 85.9.: 8.28 oz 888 88 :9. 88 8 888:8 88 8 N8 8888
.ezo.v.~@aEam . . . .
8.87"» .8828 8 8.8888 85.9.: .828 82 £8 88 :9. 88 8 888:8 BF 8 F8 8888
8888.89.88 . . . . .
8.88"» 88.8 8 8.8888 85.5.: 8888... ..m 8 8 8.88 88 :9. 88 8 e8 8 8 8 88:3
.8... 8 .F. .~ 6 . . . . .
8.88 .o.8Fu.F 8...? 8 8.88288 8.8.2 8888 am 8 8 £8 88 :9. 88 8 88...? 88 N 8 «28:3
88:86.86..an . . . . .
.o.8F£ .288 8 8.8.8888 8.8.8.2 8888 ..m 8 8 £8 88 :9. 88 8 .238 a: F 88 88:3
82.8.8888»: 28388838
8.8 8.888 8.8 .82 .8... 8 .8 .N .F 6 8.88 8888 8. 8.8 8.88 88 :9. 88.8 .888 8.8 8 88:88
8.88"» .288 8 5.8888 85.9.2
.8888.8.~.F@8588 . . . . .
.985 88.3 8 8.888% 85.9.: 89.2.8 8. 8 8 8.88 88 :9. 88 8 e8 8 8 88 «8:88
88.88 8888 88 N£080 . a . . .
8 8.88 5.; .2888 8 8.8888 85.9.2 8888 am 8 8 8.88 88 £0.80 88 8 e8 8 8 8» 88888
.8... 8 ® 888 0.838 . . . . .
88.8 8 8.888888; 8.8888 :3 8888 3. 8 8 8.88 88 :9. 88 8 e8 8 8 E. 8828
.8... 8 .8 .F a. . . . . .
.8588 .985 .888. 8 8.8.8.88 85.8.2 89.2.8 3. 8 8 :88 88 :9. 88 8 88.8. 8 8 8 888.8
.8: w .o g Qan . . . . .
888$ .288 8 8.888888 85.9.2 888% a: 8 8 =88 88 :9. 8 F .988 8 8 8 88.8
.8... 8 8 8 N F a 888 .o.8F....F 8:88 3. 8.8 :88 88F .:9. 88.8 .888 8.8 E 8:88

 

 

199

 

80.880880 0.58:8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

88 8.88828: .8: 8.8 88 .08?» 0?. .8.; 8 - 8.8 8.88 88 :9. 88.8 8883.8 88.8 88 88:8
8.88288 8580.8 08: . . . . . .
88 8.888828: .8: 8.8 88 .08?» .8.; 8.8 8.8 888 88 8:080 88 8 :9. 88 8 8.8 88 8 8 888
80:82:80 05.2.8 . . . . . . .
88 8.888828: .8: 8.8 88 .083 08a. .88 8 8 F :88 88 «£080 88 8 :9. 888 8 828 88 8 08 8:88:88
8.8888... 8588.8 08: . . . . . .
98 8.888828: .8: 8.8 88 .0888 .8.; 8.8 8.8 :88 88 8:080 88 8 :9. 88:. 8 8.8 8. 8 88 88:88
8280880 2:858 - . . . . . . .
.88 8.888828: .8: 8.8.. 88 .08qu 0?. .88 8 8 8 £8 88 8:080 88 8 :9. 88:. 8 8:8 88 8 .88 88:88
8280880 258:8 - . _ . _ .
88 88288.8: .8: 8.8 8 .0888 08“. 8...; 8 8 F :88 88 :9. 888 8 88828 88 8 88 8:88:88
.cozauﬂ 00 859:8 - . . . . .
8.8 88888828: .8: 8.8 88 .0888 0?. 8...; 8 8 8 :88 88 :0x 88 8 8883.8 at 8 8 882.8
.:_E
888 .88 .88 .88 .88 .88 .88 a 88.8 .8: 08a. 8...; 8 - 8.. :88 88 :9. 88.8 8883.8 88.8 88 8888
8.3 88 .08qu 8888.8 8 8.888828:
.88. 888 .88 .88 .88 .88 .88 w 8:88 . . . .
.0888 .8828 88 8.88.288 858...... .8828 oz 8.88 88 :9. 88 8 88828 88 8 88 88:8
.8.... 88 .88 .88 a 8:88 . . . .
.0888 .8828 8 8.888888 85.9.2 888.8 oz 58 88 :9. 88 8 88828 88 8 88 88:8
.58 ca 3 9:5 . . . .
.0885 .8828 8 888288 85.9.: .8828 oz :88 88 :9. 88 8 88828 8: 8 88 8888
EE 8 @ 088m . . . .
.0888. 8888.5 8 8.888888 85.8.? 8228 82 £88 88 :9. 88 8 8828.8 at 8 88 8:88
0.888 :9. . a . _ . . .
EE . 9:8 . . . .
. 88 88 © 8 8828 82 8.88 88 :9. 88 8 8828. 88 o 88 88:88

.0888 .8828. 8 88.888888 2.80.2

 

 

 

 

 

 

200

 

2:2 8:88. 088 .8..

282$ 93.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

25 88888»: .NI 98 co: .9352

 

 

 

 

c an m . m . .30 E o .
.9.“ 28 88888»: .8. 98 2: .9959 9...; m 82o ._ oo: :02 m m 8 o 2 m 2: 8828
2:2 8882 082 .28 282$ . . . .
. : cm a m 30 E m
.8 23 8888.81 .8. 98 2: .ooovup 03m 2.; m 9.0 ._ oo: :9. o m 8 o 2 o 3: 88:5
.28 289$ . . . .
: ow m a $0 a m
.9.“ 25 88888: .~: 98 2: 6.8.2 93. 2? m 88 ._ oo: :9. m m 8 u. 2 m 2: 8:28
.88 289$ . . . .
c on m a one 2 m
9.8 98 88888»: a: 98 2: 6.8"» 93”. 2.; m 8.: ._ oo: :9. o m 8 u. 2 m 8: 8888
:8 EB 82 .o... 98 2: .082": 88% ..m 8.2 £8 moo: ..ogeom 2: o.om S: 8828
:8 ee 82 .0: 98 2: .982 888 3m 8.2 £8 ooo: __ogeom 2: o.om o: 8:28
=8 Se 82 .o: 98 2: .9828. 88% 3m 9:: £8 ooo: 923.8 2: 28 mo: 8:28
.28 289$ . . . .
c On a m mac 2 m
.8 23 88888»: .~_._ 98 2: .982 can. 2.; ... Bo ._ oo: :02 o m 8 o 2 m mo: 888
.28 282$ . . . .
c on m a one 2 m
.8 28 888889.. .m: 98 2: .982 oi 2.; ... Bo ._ oo: :9. m m 8 o 2 o B: 8:28
.8.. 6.28 93. . . . .
c on u a one 3.. m
.8 25 888889.. .~: 98 2: .99.": 2.; m 88... ._ oo: :9 m m 8 o 2 o co: 8928
62% . .
2 . o8 _ n Foam $0 2 m
88888 I «I _ 2: 0.9. 9 03m 22 m 8.: _ oo: 5 o t m mo: 82:5
88888 8:98 6.2% . . _ .
c 0.0. m m 9.0 E a
25 88888»: .8. 98 2: .993 98. 2.; m 8.: ._ oo: :09. 3” m U o 2 m vo: 88:8
:8 Se 82 .o: 98 2: 68:2 88% E 8.2 £8 ooo: 92.28 2: «.8 no: 88::
:8 EB 82 a: 98 2: 6.822 88% E 8.2 £8 moo: __ogeom 9: Nov mo: 8:2:
8:88 Se 82 .o: 98 2: .982 88% 3m 8.2 £8 ooo: ..ogeom 2: o.om 6: 8:8:
.28 Ea 82 a: 98 2: .082": 888 3". 8.2: 58 oo: .2588 E 22. oo: M288:
8:888 2:9...“ oi 2; m - 8.: £8 8o: .19. 8mm. 68.2 82 8 8:2:

 

 

201

 

828392.95 oz .832. N .33 902693

635929

 

 

 

 

 

 

 

 

 

 

 

 

 

.93 9:: 532383...

.N... 98. 09. .969"...

 

03m 2.; m o:.o

 

 

 

£8 8 28.8 2.: .N... 98 8: .98": 922:... £8 ooo: 82. cc 88:6 at...” mm: 8: :::
:8 28 o
.88 am 6.233%
92; :_ 88 .8 88 E 8882.8 2 8o 028?. £8 ooo: .82. 8 .8835 3:8 R: 88::
828828.. 8 .98... .2. 98 8: .98»... :8 83
8.88. 28.088
.28 8 8:38 .8 :8..." 8.28 .8828 920?. £8 ooo: ....ov. mood .8836 on; om: 8822
82888:. 80 .9. 98 2: 6.8": .8 98...
... 889.88 28.088
8.28 2:0 88822.. 2 So .8828 26?. .8 Ru... 58 ooo: .19. 83 88.8 3:8 mm: 888:
8 8888.82 oz .9. 98 8: .98":
. . 295$ o . . .
8.8828... N... 98 8: 0.8-: 98. 2E ... 2... 58 oo: 8.2. o: 8202: om o vm: 88.98
28 28828: 289$ o . . . .
8888.8: .8. 98 8: .98": 23. 9...; m 2... 58 oo: .2252 o o : 822:: o2 n mm: 8898
28 22828: 289$ . . . . .
8.8829... .9. 98 8: .98": 23. 2.; m 2... £8 .8: .828 8 : 8202: 9: o 8: 88:8
. . 289$ . . .
8.8829,... «I 98 8: 0.8": 23. 2.; m 2... £8 82 88 8 8202: o2 o :2 88:8
8. 888m. .8.. 205$ . o . . o .
...... 98 8.8822... .8. 98 8: .92.": 23. 2.; m 2.8 £8 ooo: :9. o m 828.... 2 m om: 88:8
.98 289$ o . o . . a o .
.8 8.: 8888.8... .8. 98 8: .98": 23. 9.9, m 2... 58 oo: :9. o m 820 : o : 2: 828
8:. 289$ . . . .
c a E o
...... 93 8.8822: .9. 98 8: .98»: 23. 2.; m 9.8 ..8 8o: ...9. o m 820 . 2 o 2: 888
.28 289$ o . . . .
E a
9.... 93 88:829.: .8 98 8: .98": 03.... 8;... 2.8 £8 8: ...9. oo: 820 o 2 m t: 888
.ao - . a a
u 289$ £8 8o: ...9. 8 N 8225 8: o 2: 8:88

 

 

202

 

289$

 

 

 

 

 

 

 

 

 

 

 

 

. . a : . . .
29.88. 088 u... 98 8m 98 0.8. .....3 ... 2... £8 oo: ...9. 98 m 8208. o»: m 8: 8.8.8
. . 289$ . . . .
.. a8 a
8882-..... 8 N... . 8: 98 8.... 9...; ... 2... £8 8: ...9. 28 m 828.. E: m 8: 8.8.3
. . 289$ . . . .
.8.. a mo .
29.88. 8 N... . 8: 98 2... ......s ... 2... £8 ooo: :9. 9.8 m 822:. on: m mm: 8.28
. . 289$ . . . .
. o 8 c o ..c a
29.88. 988 w... _ on 98 0.8. 2.; ... 2... ._ m moo: :9. m2. .... 8o... . t m 8: 8.2.8
. . 289$ . . . .
. c 0» o a
29.082 998 N... 98 8 98 0.8. 9......» 2... ._ oo: ...9. mm m 8208. an: m 8: 83:8
2...; u... ... 828. 2.; «8.... 898... 2.8.8.8. .
c a o > >
2 29... 8. 9.28902 a: 98 8m .98 9.8.... ..8 88... A 9. .8 \o: 2. 5928 6 on o mm: 8.28:
28:08 3 2:2.» 2.88 no.2... 9.2.895.
.8 2... .22. 89988 8.828 8.2.: 9.8.8 ..8 88... ....o... ooo: .19. 988 .8836 8.: .8: 8.: :N:
.... 29828 98.22 a... 98 98 .98
8:88.. N... 326. 3 an: :3 6.2.3ch . . . .
2828 .952 222 .2. 98 8: .98": 9.2... .8 88... £8 98: :9. 8... m 8.8:... at m 8: 8.8a:
.oo .9. .8. .2 0.2 8.. 0m 2.8.088 .. .8 ooo: ....9. 33 .8835 at... an: 88a:
228 8292 92.2.. a... 98 com .988. 2.8.8 ...... 82o .
.>..oo 8cm .8.. 8 Eco Ow :o 822.28% oz 6.2.35: . . . . .
.88... 8.8:... .... oo .9. 98 8.. .98": 2.2... .8 98... moo: 82. 8 88.8 on o 2.... 8m 0 :2 8.8::
8.2228 9.8.. .8. ...... .39.. .8 2.8.8.8. . . .
.oo 2 29.88. 890 .2. 98 8: .98": 9.8... .8 a... woo: 82. 8 828.288 wow o on: 8.2::
.82 ... .... .8; 28 88.088
...2 .28 .8 80882 ... .9... 8... .8 .02 226...". ooo: .82. 8 635928.... 8.; 8 98.: mm: 8.2::
8.. 0n. ... 29.885 a... 98 8: .98": ...... a8...

 

 

 

 

 

 

203

 

8882.28 8 .~_._ 98 8m .98

6.8388

£8 88 .6923 88 838.988 98

v9.

#90 (no

 

 

 

 

 

 

 

 

 

 

 

 

 

 

26:8 888
8882 988 .N: 98 2: .98 wﬁmwwc £8 32 .19. 83 6882.. at.” 8.. 8a:8
.8882 988 .m: 98 8? .93 manomﬁumvo c._om 89 .19. 83 688.5 at...” «9 2288
. . 6.8.8.8 a . a . . 9 Eu .
8882.28 8 «I 98 8m 98 . 58 com on. 3 o mo 5288 _ v o E? 3328
2g: 88 c
. . 68.8.8 8 . 3a. . 9 3a.
8882.28 8 «I 98 8m 98 . 58 com .28 _ v c on 5228 . c o 2.: 228
0:98 98 o
. . . 205$ . . . .
8882 98m «x 98 8 92. 23... 9.? m 28 £8 89 :9. 98 m 8225 at m 9: 888
. . 9.8.088 a . z > a a .
8882.28 8 N... 98 8. 98 28?. 8nd 53 com on 5288 _ v 0 9: 3:28
. . . 289$ . . . o .
8882 988 «I 98 mm 99. 23. 3.3 m 28 £8 89 :9. man m 8225 t m t; 288
. . . 289$ . . . a .
88:8. 988 m... 98 8. 93 03m $8 ... 28 £8 89 :9. a8 m 0882.. t a mi .288
5882.28 2. .m: 98 8. .o 8 922.088 c .8 82 838.288 98 9: 33:8
. . . 28: .8 888 .
388 8288 55 .0 205$ . . . .
o a o o E a
9. 8:9: .8882 928 .m... 98 2: .93 23. 9.3 m 28 .... m 88 :9. 8 m m 6 h o N 3: 358
. . 289$ . . . a .
.3882 98m «1 98 8. 98 23.. 8; ... 2.8 £8 82 :9. 88 m 3225 t m «3 3393
. . . 305$ . . . .
8882 988 a: 98 8. 93 23. 8: m 28 c._om 83 :9. 98 m 8258 at m «3 '25:.
3%»; .829: £9.23 50¢. 89 .IOx 30m .3833 3.2m 5:. «2823

8858 .2 8 58m 9.. 98 2: .93

 

03m 9.? m 2.0

 

 

 

 

 

204

 

6.8.388

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8882.28 2. .9._ 98 8. .92. 2825 £8 83. .19. 82m. .8082. 8...... 8. 388
.8 88
. . 89.282 . . . a .
8882.28 2. 9. 98 8. 98 2.8.x. 88 £8 83. :9. 83 m 8202.. t .... 8. 3:28
88822.. 9. .9. 98 8m .0 8 82998 c .8 88 .888 8.3 8889288 8.3 8. 3398
. . . 2.8.8 98.3 .
. . 6.8288 . .
8882.28 8 9.. 98 8m 98 2.2: 88 £8 80. 88.8 8 o 8. 3898
. . €58.82 . .
8882.28 8 9.. 98 8w 98 2.2.9 88 c._8 83. 8202. 8 o 8. 3828
98m ... 3882.28 a: 98 8. .98. 39.8.0. 3... 8.2 £8 38. .828 8.8 8. 388
. . 6.2.28.2 . . . .
8882.28 2. 9.. 98 8. 92. 8 - 2.2.8 8.3 £8 88. :9. 3.3 m 8225 8. m 8. 3393
.2398 2 5.8998 6.2.0.95. . .
88v .8882 8 .u: 98 8m .98 .8 - 2.2.9 3.... .98 8o. 88:... 8 o .o. 3. 8. 3
. . €88.82 . a .
8882.8 2. 9.. 98 com 98 8 - 2.2.... 3.... £8 83. 8.38.28 .a 8 o 8. 38:3
. . €88.88 . > 9 a .
8882.28 2. 9. 98 8m 98 8 - 2.2.8 2.3 £8 38. 8 8228 _ 8 o 8. 33:3
8882.28 2. .9._ 98 8m .98 8.1mmmﬂmw8 .98 8... 8.89288 28 8. 3323
8882 938 .9. 98 8m .98 hﬁmwﬁ £8 88 .19. 83.8 .8202. 8.8 8. 388
. . AEWC—Ogmv a . a . . > > a o .
8882.28 8 9.. 98 8m 98 . £8 8m .828 m o 8 8228 _ v o 8. 3898
2.2: 38 o
8882.28 8 .N: 98 8m .98 68.082 £8 88 838.288 88 mm. 3898

0539 SN...

 

 

 

 

 

 

205

 

 

 

 

 

 

 

«I 98 8m. .98... «oﬁaﬂws. £8 ooo. ..288 8.8 «... 3898
N: 98 SN. .98 osmonmr... £8 ooo. .888 mod. n... 3898
9. 98 8m. .93." Swamp. 58 So. .288 83. «S 3:98
2 8820 - 9. 98 8~..%M¢w.ﬁcwhw_mmw «oammzwqu. £8 38. .888 8.8 .... 389 8
9. 98 8m. .98~ “OHM“..WMWS £8 ooo. ..288 moo. o... 38.8
9.. 98 28. .938 .8889 £8 ooo. .898 mod. ... 38.8

 

NOE—.1 *0 no...

 

 

 

 

 

206

10.

11.

12.

13.

14.

15.

l6.

17.

18.

REFERENCES

Hogle B. et al., Ind Eng. Chem. Res., 2001, 41, 2069-2073
Zhang, Z. et al., Appl. Cat. A: Gen, 2001, 219, 89-98

Jere, F.T., Org. Lett., 2003, 5, 527-530 .

Chopade, S. et al., US. Patent 6,291,725, 2001

The American Chemical Society, Technology Vision 2020: The US. Chemical
Industry, July 2001

British Petroleum, BP Statistical Review of World Energy, June 2003
British Petroleum, BP Statistical Review of World Energy, June 2002
British Petroleum, BP Statistical Review of World Energy, June 2001

Biomass Technical Advisory Committee, Vision for Bioenergy and Biobased
Products in the United States, October 2002

Chemical Market Reporter, March 15, 2004, 265

http://ers.usda.gov/brieﬁng/sugar/Data/datahtm, viewed on August 27, 2004, last
update July 13, 2004

Chemical Market Reporter, September 24, 2001, 260
Chemical Market Reporter, December 17, 2001, 260

Vollhardt, K. Peter C. and Schore, N.E., Organic Chemistry, 2“" Ed., p. 250, W.H.
Freeman and Company, New York, 1995

Kroschwitz, J .; Howe-Grant, M., editors, Kirk-OIhmer Encyclopedia of chemical
technology, 4“1 Ed., 13, pp. 446-447, Wiley, New York, 1992

Bowden, E.; Adkins, H., J. Am. Chem. Soc., 1934, 56, 689-691
Adkins, H.; Pavlic, AA, .1. Am. Chem. Soc., 1947, 69, 3039-3041

Adkins, H. ; Billica, H.R., J. Am. Chem. Soc., 1948, 70, 3121-3125

207

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Camahan, J.E. et al., J. Am. Chem. Soc., 1955, 77, 3766-3768

Kroschwitz, J .; Howe-Grant, M., editors, Kirk-0thmer Encyclopedia of chemical
technology, 4'” Ed., 23, pp. 105-106, Wiley, New York, 1992

Wisniak, J .et al., Ind Eng. Chem. Prod Res. Develop, 1974, 13 (1), 75-79

Wisniak, J.; Hershkowitz, M.; Stein, 8., Ind Eng. Chem. Prod Res Develop, 1974,
13 (4), 232-236

Wisniak, J.; Simon, R.; Ind Eng. Chem. Prod Res. Dev., 1979, 18(1), 50—57

Brahme, P.H.; Doraiswamy, L.K.; Ind Eng. Chem, Process Des. Dev., 1976, 15
(I), 130-137

Verma, R.; Gehlawat, J .K.; J. Chem. Tech. Biotechnol, 1989, 46, 295-301

van Gorp, K. et al., Cat. Today, 1999, 52, 349-361

Cerino, P.J.; F leche, G.; Gallezot, P.; Salome, J.P., Heterogeneous Catalysis and
Fine Chemicals II: proceedings of the 2nd international symposium, Poitiers,
October 2-5, 1990, pp. 231-236, Guisnet M. et al. (editors), Elsevier, Amsterdam;
New York, 1991

Guo, H.; Li, H.; Xu, Y.; Wang, M., Mater. Lett., 2002, 5 7, 392-398

Hegedus, M.; Gobolcs, S.; Margitfalvi, J.L., Heterogenous Catalysis and Fine
Chemicals III: proceedings of the 3rd international symposium, Poitiers, April 5-8,
1993, pp. 187-194, Guisnet, M. et al. (editors), Elsevier, Amsterdam; New York,
1993

Heinen, A.W.; Peters, J .A.; van Bekkum, H., Cwbohyd Res., 2000, 328, 449-457
Mikkola, J.-P. et al., J. Chem. Tech. and Biotech., 2001, 76, 90-100

Kusserow, B.; Schimpf, S.; Claus, P., Adv. Synth. Catal., 2003, 345, 289—299

Ellis, C., Hydrogenation of Organic Substances, 3‘“1 Ed, D. Van Nostrand
Company, New York, 1930

Connor, R.; Adkins, H., J. Am. Chem. Soc., 1932, 54, 4678-4690

Natta, G.; Rigamonti, R.; Beati, E., Berichte D. Chem. Gesellschaft, 1943, 768,
641-656

Clark, 1., Ind. Eng. Chem, 1958, 50 (8), 1125-1126

208

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

van Ling, G.; Vlugter, J., J. of App. Chem, 1969, 19, 43-45
Sohounloue, D. et al., React. Kin. Catal. Lett., 1983, 22 (3-4), 391-3 97
Montassier, C. et al., Bull. Soc. Chim., 1989, 126(2), 148-155

Muller, P.; Rimmelin, P.; Hindermann, J.; Kieffer, R.; Kiennemann, A. ; Carré, J .,
Heterogeneous Catalysis and Fine Chemicals II, Guisnet M. et al. (Editors),
Elsevier Science Publishers B.V., Amsterdam, 1991

Montassier, C. et al., J. ofMol. Catal., 1991, 70(1), 994 10

Larchar, A., US. Patent 1,963,997, 1934

Rothrock, H., US. Patent 2,004,135, 1935

Conradin, F. et al, US. Patent 2,852,570, 1958

Conradin, F. et al, US. Patent 2,965,679, 1960

Conradin, F. et al, US. Patent 3,030,429, 1962

Kasehagen, L., US. Patent 3,396,199, 1968

Sirkar, US. Patent 4,338,472, 1982

Sirkar, A., US. Patent 4,380,678, 1983

Chao, J.; Hulbers, D., US Patent 4,366,332, 1982

Arena, B., US. Patent 4,401,823, 1983

Tanikella, M., US. Patent 4,404,411, 1983

Dubeck, M.; Knapp, 6., US. Patent 4,430,253, 1984

Dubeck, M.; Knapp, G., US Patent 4,476,331, 1984

Andrews, M.; Klaeren, S., US Patent 5,026,927, 1991

Gubitosa, G.; Casale, C., US. Patent 5,326,912, 1994

Gubitosa, G.; Casale, C., US. Patent 5,354,914, 1994

Gubitosa, G.; Casale, C., US Patent 5,403,805, 1995

209

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

Gubitosa, G.; Casale, 0, US. Patent 5,543,379, 1996
Gubitosa, G.; Casale, C., US. Patent 5,600,028, 1997
de Bruyn, C. A. L., Recl. Trav. Chem. Pays-Bas, 1895, 14, 156-165

de Bruyn C. A. L.; van Ekenstein, W. A., Red. T rav. Chem. Pays-Bas, 1895, 14,
203-216

Nef, J., Ann., 1907, 35 7, 294-312

Evans, W. L.; Edgar, R. H.; Hoff: G. P., J. Am. Chem. Soc., 1926, 48, 2665-2677
Denis, W. Am. Chem. J., 1907, 38, 561-594

Shaffer, P. A., Friedemann, T. E., J. of Biol. Chem, 1930, 86, 345-375

Braun, 6., US. Patent 2,024,565, 1935

Lock, R. R, US. Patent 2,382,889, 1945

Tronconi, E. et al., Chem. Eng. Sci., 1992, 47 (9-11), 2451-2456

Montgomery R.; Ranca, R. A., Ind Eng. Chem, 1953, 45 (5), 1136-1147
MacLaurin, D. J.; Green, J. W., Can. J. Chem, 1969, 47, 3947-3955

de Wit, G.; Kieboom, A. P. G.; van Bekkum, H., Carbohyd Res., 1979, 74, 157-
175

de Bruijn, J. M.; Kieboom, A. P. G.; van Bekkum, H., Recl. Trav. Chim. Pays-Bas,
1986, 105, 176-183

de Bruijn, J. M., et al., Int. Sugar Journal, 1987, 89 (1066), 194-210

de Bruijn, J. M.; Kieboom, A. P. G.; van Bekkum, H., Recl. Trav. Chim. Pays-Bas,
1987, 106, 35-43

de Bruijn, J. M.; Kieboom, A. P. G.; van Bekkum, J. Carbohyd Chem, 1986, 5
(4), 561-569

de Bruijn, J. M.; Kieboom, A. P. G.; van Bekkum, Starch, 1987, 39(1), 23-28

de Bruijn, J. M.; Touwslager, F .; Kieboom, A. P. G.; van Bekkum, H., Starch,
1987, 39(2), 49-52

210

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.
94.
95.
96.

97.

98.

de Bruijn, J. M.; Kieboom, A. P. G.; van Bekkum, Sug. Tech. Rev, 1986, 13, 21-52

Gregg, S.J.; Sing, K.S.W., Adsorption, Surface Area and Porosity, 2“cl Ed, pp. 173-
190, New York, 1982
Brunauer, S.; Emmett, P.H.; Teller, E., J. Am. Chem. Soc, 1938, 60(2), 309-319

Micromeritics, ASAP 2010 Chemi: Accelerated Surface Area and Porosimetry
System Operator 's Manual, Appendix C, September 1996

Al-Dahhan, M.H.; Dudukovié, M.P., AIChE Journal, 1996, 42 (9), 2594-2604
Al-Dahhan, M.H.; Dudukovic, M.P., Chem. Eng. Sci., 1995, 50 (15), 2377-2389

McCabe, W.L. et al., Unit Operations of Chemical Engineering, 5“ Ed, p. 153,
McGraw Hill Inc., New York, 1993

McCabe, W.L. et al., Unit Operations of Chemical Engineering, 5th Ed, p. 155,
McGraw Hill Inc., New York, 1993

McCabe, W.L. et al., Unit Operations of Chemical Engineering, 5th Ed, p. 928
McGraw Hill Inc., New York, 1993

Levenspiel, 0., Chemical Reaction Engineering, 3"l Edition, p. 27, Wiley, New
York, 1999

Unpublished data from Miller group produced by L. Peereboom
Zwietering, T. H., Chem. Eng. Sci., 1958, 8, 244

Felder, R.M.; Rousseau, R.W., Elementary Principles of Chemical Processes, 2“d
Ed, p. 248, Wiley, New York, 1986

Perry, R.H.; Green, D.W., Perry ’s Chemical Engineers ’Handbook, 7‘h Ed, p. 2-
126, McGraw-Hill, New York, 1997

Yagi, H.; Yoshida, R, Ind Eng. Chem. Proc. Des. Dev., 1975, 14, 488
Bern, L.; Lidefelt, J .O.; Schoon, N.H., J. Am. Chem. Soc., 1976, 53, 463
Boon-Long, S.; Laguerie, C.; Couderc, J.P., Chem. Eng. Sci., 1978, 33, 813
Weiss, P.B.; Prater, C.D., Adv. Catal., 1954, 6, 143

F ogler H.S., Elements of Chemical Reaction Engineering, 2“d Ed, p.256, Prentice-
Hall, New York, 1992

Pallasana, V., Neurock, M., J. Catal., 2002, 209, 289-305

211

99.

100.

101.

102.

103.

104.
105.
106.

107.

108.
109.

110.

111.

Fogler H.S., Elements of Chemical Reaction Engineering, 2"‘1 Ed, p.341, Prentice-
Hall, New York, 1992

Cannizzaro, S., Ann., 1853, 88, 129

Camahan, B.; Luther, H. ; Wilkes, J ., Applied Numerical Methods, Wiley, New
York, 1969

Jere, F.T., Ph.D. Dissertation, Michigan State University, 2003

Belohlav, Z.; Zamostny, P.; Kluson, P.; Volf, J ., Can. J. of Chem. Eng, 1997, 75,
73 5-742

Tewari, Y.B.; Goldberg, R.N., Biophys. Chem, 1986, 24, 291-294
Camacho-Rubio, F. et al., Can. J. of Chem. Eng, 1995, 73, 935-940
Franks, F., Pure and Appl. Chem, 1987, 59 (9), 1189-1202

Smith, J .M.; Van Ness, H.C.; Abbott, M.M., Introduction to Chemical Engineering
Thermoaynamics, 5th Ed, p. 567, The McGraw-Hill Companies, New York, 1996

Ellis, A.V.; Wilson, M.A., J. Org. Chem, 2002, 67, 8469-8474
deWit, G. ;Kieboom, A.P.G.; van Bekkum, H., Carbohyd Res., 1979, 74, 157-175

Kroschwitz, J.; Howe-Grant, M., editors, Kirk-Othmer Encyclopedia of chemical
technology, 4'h Edition, 13, p. 852, Wiley, New Yorlg 1992

US. Natural Gas Prices,
http://tonto.eia.doe.gov/dnav/ng/ng_pri_sum_dcu_nus_m.htm, viewed November
15, 2004, last updated October 29, 2004

212

   

nrlmlllgllglljllmlimp