“7—.“

ur-

 

- on
«a
~ s
-vl.n- ,
u
A

a... If“
:v»'..."~'.“'.‘ i
a. .
. «r-

“‘53
...- ..,..
.
357-?”

 

...
‘ ‘FI‘A N
w.- - m
.0...
2 ~ .

.-—...-.. ..

 

 

"5":

 

 

1.4.}

1...» .

.5:
1

‘1

I

-
.‘
a

7"
‘9

 

#:wfl'r

"éé‘f‘hfi; '
WW3;

 

 

J K x.)

Ph.D.

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

PROPERTIES AND SEPARATIONS OF
PLANT-DERIVED CHEMICALS

presented by

DUNG THI VU

has been accepted towards fulfillment
of the requirements for the

degree in Chemical wheemg

 

 

Jazz/7A4;

Major Professor’s Signature

5/8/07

Date

MSU is an affinnative-action, equal-opportunity employer

 

.-.-.-.-.-.-.-.-.-.-.-.-.--.-.-.----.. -»-.-i------- -- --m—

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DaleDue.indd-p.1

PROPERTIES AND SEPARATIONS OF
PLANT-DERIVED CHEMICALS

By

Dung Thi Vu

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

2007

 

ABSTRACT
PROPERTIES AND SEPARATIONS OF PLANT-DERIVED CHEMICALS
By

Dung Thi Vu

Biorenewable processing is highly dependent on the feasibility of separations.
This work considers the development of fractionation principles and thermodynamic
properties useful for process designs in two main areas: fractionation of lipids; and
measurement/prediction of properties for upgrading of organic acids and their
derivatives.

Ricinolein can be separated from the other triglycerides in castor oil by adsorbing
it onto acidic adsorbents or by concentrating it in the effluent stream of fixed-beds using
non-polar adsorbents. Solvents play an important role in adsorption. Replacing hexane
with ethanol, the hydroxylated triglycerin, which is preferentially adsorbed by Florisil in
hexane, will be released in the effluent. The adsorbent capacity of Florisil to the total oil
also significantly changes from 17 wt. % in hexane to 6 wt. % in ethanol.

Extractions of plant lipids usually use solvent mixtures of chloroform-methanol.
However, chloroform is very toxic and a possible carcinogen in humans. A proposed
process developed in this work successfully extracts sulfoquinovosyl diacylglycerol
(SQDG) from alfalfa and isolates it from other plant lipids using non-chlorinated
solvents. Lipases are deactivated during extraction using isopropanol at 50-55 °C.
Hexane-methanol mixtures are used in extraction and subsequent liquid-liquid

partitioning to remove proteins and phospholipids. The yield of SQDG using the

 

proposed process is comparable to literature and the amount of solvent used is less than
30 vol. % of the value reported in literature.

For purposes of upgrading of lactic acid, a model is developed for the oligomer
distribution in aqueous solutions. The model is extended to multicomponent VLE in
mixtures involving lactic acid and ethyl lactate oligomers. A P-x-y apparatus is also
developed to measure VLE and VLLE at temperatures between 25 °C and 80 °C and
pressures down to 0.7 kPa. Data pass either the area or point-to-point thermodynamic
consistency test.

Finally, the Step Potential Equilibria and Dynamics (SPEAD) molecular
simulation method is adapted for predicting vapor pressures of oxygenated compounds.
A method is developed for optimization of five and nine variable functions. From this
study, parameters for the secondary —OH, cyclic -O-, and -COO- groups are made
available for use in SPEAD. Vapor pressures of esters containing up to 30 carbons are

predicted within 25 % of the experimental values using these parameters.

A.“

 

Copyright by
DUNG THI VU
2007

Dedicated to my Lord and God, Jesus and his Holy mother Mary, for giving me the

inspiration and stamina to complete this dissertation.

 

ACKNOWLEDGMENTS

I would like to thank Dr. Carl T. Lira for his guidance and support as my advisor
through my Ph.D. studies. I am especially grateful for his patience and the
encouragement he gave me to finish my dissertation. I also wish to thank Dr. Dennis
Miller, Dr. ChristOph Benning, and Dr. Mark Worden for serving on my committee.
Special thanks to Dr. Miller for his support and giving me a chance to work with his
research group on reactive distillation projects, to Dr. Christoph Benning for
collaboration on sulfolipid project, and to Dr. Richard Elliott at the University of Ohio at
Akron for collaboration and support on Step Potential Equilibria and Dynamics
molecular simulation.

I also would like to thank my family Vu (Vii), extended family, Chuck T. Wynn
(Huynh Thién Hung), and Vivian Gendron for their help, support and unconditional love.

I also want to extend my appreciation to Dr. R. Leep and T. Dietz in the
Department of Crop and Soil Sciences for providing alfalfa; to Dr. Navin Asthana and
Dr. Aspi Kolah from Dr. Miller’s Reactive Distillation Group, and Dr. Wayne Riekhof
from Dr. Benning’s sulfolipid project for their support. Thanks also go to undergraduate
students, Jesse Wright and Brett Burkhart for their assistance in phase equilibrium

experiments and the administration of the MSU Chemical Engineering Department.

vi

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................. x
LIST OF FIGURES ............................................................................................. xii
PART 1:
SEPARATIONS OF PLANT-DERIVED CHEMICALS .......................................... 1
Chapter 1 — Background on Ricinolein and SQDG .......................................... 2
1.1 — Introduction on Castor Oil Separation ................................................................... 2
1.2 — Introduction on Plant Sulfolipids Separation ......................................................... 3
Chapter 2 - Selective Adsorption of Ricinolein Studies .................................. 7
2.1 -— Overview ................................................................................................................ 7
2.2 — Adsorbent Selection .............................................................................................. 7
2.3 — Batch Adsorption ................................................................................................... 8
2.3.1 — Experimental Preparation and Analysis .......................................................... 9
2.3.2 — Results and Discussion ................................................................................... 9
2.4 — Fixed-bed Adsorption .......................................................................................... 11
2.4.1 — Fixed-bed Design .......................................................................................... 12
2.4.2 — Sample Analysis ............................................................................................ 14
2.4.3 — Results and Discussion ................................................................................. 19
Chapter 3 — Extraction and Purification of SQDG .......................................... 22
3.1 — Overview .............................................................................................................. 22
3.2 - Stage (1) — Extraction and Coarse Fractionation Studies .................................... 22
3.2.1 — The Hildebrand Solubility Parameter ........................................................... 23
3.2.2 — Solvent Selection .......................................................................................... 24
3.2.3 - Experimental Preparation .............................................................................. 25
3.2.4 — Extraction Procedure ..................................................................................... 26
3.2.5 — Sample Analysis ............................................................................................ 27
3.2.6 — Results and Discussion ................................................................................. 31
3.2.7 — Improved Extraction and Coarse Fractionation ............................................ 34
3.3 — Stage (2) — Purification Studies ........................................................................... 35
3.3.1 - Adsorption using Florisil .............................................................................. 35
3.3.2 — Ion-exchange Chromatography Using DEAE-cellulose ............................... 37
3.4 -— A proposed Extraction and Purification Scheme ................................................. 38

vii

 

PART 2:

PROPERTIES OF PLANT-DERIVED CHEMICALS ........................................... 42
Chapter 4 - Phase Equilibria and Vapor Pressure ......................................... 43
4.1 — Vapor-Liquid Equilibrium Overview .................................................................. 43
4.2 — Vapor-Liquid Equilibrium Apparatuses and Measurements ............................... 43
4.3 - VLE of Pure components and Vapor Pressure Predictions ................................. 44
4.3.1 — Estimation Using the Clausius-Clapeyron Equation ..................................... 45
4.3.1.1 -— Assuming that AHv/AZ is Constant ....................................................... 45
4.3.1.2 — Using AHv/AZ Temperature Dependence ............................................. 47
4.3.1.3 — The Korsten Correlation ........................................................................ 47
4.3.1.4 — Summary of Methods Using the Clausius-Clayperon Equation ............ 48

4.3.2 — Highlights of Prediction Methods using Group Contribution ....................... 49

4.4 — Background on Step Potential Equilibria And Dynamics Simulation ................. 51
4.4.1 — SPEAD and Discontinuous Molecular Dynamics Algorithm ....................... 52
4.4.2 — Thermodynamics Perturbation Theory ......................................................... 54
4.4.3 - Vapor-Liquid Equilibrium using SPEAD ..................................................... 56

4.5 — Objective and Scope of Research ........................................................................ 59
Chapter 5 - Lactic Acid Oligomers Distribution ............................................. 61
5.1 — Overview .............................................................................................................. 61
5.2 -— The Lactic Acid Oligomers Distribution Model — A Reprint of the Paper
“Oligomer Distribution in Concentrated Lactic Acid Solutions ” ................................. 63
Chapter 6 - Vapor-Liquid Equilibrium ............................................................. 84
6.1 — Overview .............................................................................................................. 84
6.2 — P-x-y Apparatus and VLE Measurements ........................................................... 84

6.2.1 — P-x-y Data of Ethyl Lactate Systems and (Ethanol + Water at 40 0C) — A
Reprint of the Paper “Vapor-Liquid Equilibria in the Systems Ethyl Lactate +

Ethanol and Ethyl Lactate + Water ” ........................................................................ 87
6.2.2 — P-x data of Citrate Systems ......................................................................... 100
6.2.2.1 - Triethyl Citrate + Water ....................................................................... 100
6.2.2.2 — Triethyl Citrate + Ethanol .................................................................... 102
6.2.3 — P—x Data of Diethyl Succinate System ........................................................ 104
6.3 — T-x-y apparatus and measurements ................................................................... 105
6.3.1 — T-x-y Data of Lactic Acid + Water System ................................................ 108
6.3.2 — T-x-y Data of Lactic acid + Ethanol + Ethyl lactate + Water System ........ 111
6.3.3 — Limitation of T-x-y Fischer Still ................................................................. 117
Chapter 7 — Vapor Pressure Prediction using SPEAD ................................. 119
7.1 - Overview ............................................................................................................ 119
7.2 - SPEAD Simulation Procedure ........................................................................... 119
7.2.1 -— Pair Interaction Sites of the Interested Compounds .................................... 120
7.2.2 - Approach of Optimizing Secondary -OH and -COO- Groups .................... 121
7.2.3 — Mathematical Methodology ........................................................................ 122
viii

7.3 — Results of Optimization of the 2nd -OH and -COO- Groups ............................ 126

7.4 — Prediction of Psat for Ethyl Lactate and Methyl Lactate ................................... 129
7.4.1 - Effect of Intramolecular H-bonds in Lactates ............................................. 129
7.4.2 — A Common Point for Ethyl Lactate Oligomers .......................................... 131
7.4.3 — A Validation for Psat Prediction of Ethyl Lactate Oligomers ...................... 133

7.5 — Prediction of Psalt for Acetals .............................................................................. 134

7.6 — Bias in SPEAD Simulation and Regression of A2 ............................................. 138

Chapter 8 — Conclusion and Recommendations .......................................... 140

8.1 — Separation of Ricinolein .................................................................................... 140

8.2 - Extraction and Purification of SQDG ................................................................ 141

8.3 — Measured and Predicted VLE and Vapor Pressure ............................................ 142

APPENDIX A: EXPERIMENTAL ADSORPTION DATA .................................. 147
APPENDIX B: FORTRAN PROGRAM ............................................................ 188
APPENDIX C: MISCELLANEOUS DOCUMENTATION .................................. 198
CI — Vapor Pressure Data (predicted and experimental) ................................. 199
C2 - Structure of Ethyl Lactate and its Oligomers .......................................... 209
C3 — The .m3d Files used in Psat Predictions .................................................... 210
LIST OF REFERENCES ................................................................................... 217

ix

 

Table 2—1

Table 2-2

Table 2-3

Table 2-4

Table 2-5

Table 3-1

Table 3-2

Table 3-3

Table 4-1

Table 4-2

Table 6-1

Table 6-2

Table 6-3

Table 6-4

Table 6-5

Table 6-6

Table 6-7

Table 7-1

Table 7-2

Table 7-3

Table 7-4

Table 7-5

LIST OF TABLES

Triglyceride Contents in Soy-based and Castor Oils ..................................... 8
Summary of Batch Adsorption Experiments in Ethanol ............................... 10
Summary of Fixed-bed Adsorption Experiments .......................................... 13
Sample of Fixed-bed Adsorption Data (Run # 22) ........................................ 17
Summary of F ixed-bed Adsorption Results .................................................. l9
Solubility Parameters of Related Solvents .................................................... 24
Solvent Systems Used in Extraction Studies ................................................. 25
Result from Extraction Studies ...................................................................... 32
Errors in Estimating Psat Using the Clausius—Clayperon Equation ................ 48
ASPEN Predicted Psat of Lactic Acid and di-Lactic Acid ............................. 50
P-x Data of Triethyl Citrate (1) + Water (2) at 25.0 °C and 60.0 °C ........... 101
P-x Data of Triethyl Citrate (1) + Ethanol (2) at 40.0 °C ............................ 103
Diethyl Succinate (1) + Ethanol (2) at 50.0 °C ............................................ 104
T -x-y Data of Lactic acid (1) + Water (2) at 103.33 kPa ............................. 108
ASPEN Parameters of the Extended Antoine Equation. ............................. 112
Parameters used in ASPEN Prediction ........................................................ 114
Measurement and Prediction of LAs + Ethanol + ELAs + Water Systems. 115

Optimization and Validation of —OH and —COO— Sites ............................. 127
Parameters used in SPEAD Calculated Psat ................................................. 129
Predicted Psat of Methyl Lactate and Ethyl Lactate ..................................... 130
Measured T—P and SPEAD Prediction for Ethyl Lactate ............................ 131
Vapor Pressure of Ethyl lactate Oligomers Mixtures .................................. 133

Table 7-6 Optimization and Validation of the Cyclic —0— Site ................................... 135
Table 7-7 Prediction of Psat for Acetals using SPEAD ................................................ 136

Table 7-8 SPEAD Predicted T-P-x-y of 4HMD (1) + SHMD (2) at 373.15 °K .......... 137

xi

 

Figure 1.1
Figure 1.2
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5
Figure 3.6
Figure 4.1
Figure 4.2
Figure 6.1
Figure 6.2
Figure 6.3
Figure 6.4
Figure 6.5
Figure 6.6

Figure 6.7

LIST OF FIGURES

Structure of Ricinoleic Acid ........................................................................... 2
Structure of Phosphatidylchloline .................................................................. 4
Fixed-Bed Design ......................................................................................... 12
Example of Gas Chromatogram of Fatty Acid Methyl Esters ..................... l6
Breakthrough Curves of Total Oil and Glycerides (Run # 22) .................... 18
Fixed-bed Adsorption Using Hexane and Florisil (Run # 28) ..................... 20
FAB-MS Chromatogram of SQDG Sample ................................................. 3O
Extraction and Evaluation Scheme in Extraction Studies ............................ 33
Adsorption Using Florisil in Purification Studies ........................................ 36
Ion-Exchange Using DEAE in Purification Studies ..................................... 37
A Proposed Extraction and Purification Scheme ........................................ 40
Result of Extraction and Purification of SQDG ........................................... 4]
Illustration of the Multi-step Potential ......................................................... 54
Trends of A 1 and A 2 and their Fitted Polynomials Function ......................... 58
Set up of the P-x-y apparatus ........................................................................ 85
Configuration of the Agitator and Liquid Sampling Valves ........................ 86
Vapor Sampling Valve at the Inject (left) and Load (right) Position ........... 86
Structure of Triethyl Citrate ....... i ................................................................ 100
P-x of Triethyl Citrate (1) + Water (2) at 25 °C and 60 °C ........................ 102
P-x of Triethyl citrate (1) + Ethanol (2) at 40 °C ....................................... 103
Structure of Diethyl Succinate .................................................................... 104

xii

 

Figure 6.8
Figure 6.9
Figure 6.10

Figure 6.11

Figure 6.12
Figure 7.1
Figure 7.2
Figure 7.3
Figure 7.4
Figure 7.5
Figure 7.6

Figure 7.7

P-x of Diethyl Succinate (1) + Ethanol (2) at 50 °C ................................... 105
Overview of Fischer Recirculating Still ..................................................... 106
Schematic of Fischer Recirculating Still .................................................. 107

T-x-y of Lactic acid (1) + Water (2) at 103.33 kPa as a Function of Monomer

Concentration ............................................................................................. 1 1 l

ASPEN Predicted Psat of Lactic, di-Lactic, and tri-Lactic Acids ............. 113
The Interaction Sites of Ethyl Lactate Oligomers and Acetals .................. 121
Group Indices in 2-Alkanols and Base Esters used in Optimization .......... 122
Illustration of Error in Prediction of Psat ..................................................... 124
Trend of Predicted Vapor Pressure of Ethyl Lactate Oligomers ................ 133
SPEAD Predicted Vapor Pressure of Acetals ............................................ 137
Predicted VLE of 5HMD (1) + 4HMD (2) Mixtures at 373 °K. ................ 138
Trend of A2 in Simulation of Ethyl Lactate Oligomers .............................. 139

xiii

 

PART 1:

SEPARATIONS OF PLANT-DERIVED CHEMICALS

..__ - .—‘-- _——

 

Chapter 1 — Background on Ricinolein and SQDG

1.1 - Introduction on Castor Oil Separation

Castor oil is obtainable from castor bean (Ricinus Communis), which is
extensively cultivated in India, Brazil and China [1]. At present, castor oil is the only
source of commercially hydroxylated fatty acid containing up to 90 wt.% of ricinoleic
acid, (12-hydroxy—9Z-octadecenoic acid), which can be easily obtained by hydrolysis of
the corresponding triglycerides [2].

Triglycerides in castor oil are mostly triricinolein and diricinolein, derived from
esterification of fatty acids with glycerol. In a typical castor oil, the fatty acids’ contents
are 85-90 wt.% of ricinoleic, 3-5 wt.% of linolenic, 2-5 wt.% of oleic, 1-2 wt.% of
stearic, 1-3 wt.% of palmitic acid, and refined vegetable oils generally have less than 2

wt.% of non-glyceride components [3, 4].

HO

Figure 1.1 Structure of Ricinoleic Acid

Castor oil can be a renewable source of non-petroleum chemical feedstocks. In
addition to its excellent emollient and lubricating properties, which have been utilized in
wetting and dispersing dyes, pigments, and fillers in textiles and inks, castor oil also is an
excellent plasticizer or a surfactant for a wide variety of natural and synthetic resins,

waxes, polymers and elastomers due to its highly polar hydroxyl groups. Ricinoleic acid

is widely used in urethane-polymer, electronics, food, pharmaceutical, perfumes and
cosmetic industries [1, 2, 5].

For specific applications, highly pure ricinoleic fatty acid content is desirable.
Also, research is ongoing to genetically modify plants to selectively express functional
lipids. However, purity is difficult to achieve. In general, diStillation is impractical for
separation of high molecular molecules such as triglycerides, due to their low vapor
pressure. Emulsions and foam generation due to the presence of fatty acids in plant oils
can also create difficulties in liquid-liquid extractions, and require large quantities of
solvents for a complete separation. This would result in high capital, operating costs and
difficulty in disposing or recovering large amount of organic solvents [6].

Solid-phase extraction can be an attractive alternative separation process for
triglycerides. Chapter 2 in this dissertation will discuss the separation of triricinolein
from non-hydroxylated glycerides in castor oil using adsorption. Basically, castor oil is
blended with solvent then loaded into a column containing adsorbents, which have
different affinity to hydroxylated and non-hydroxylated glycerides. The stronger affinity
to one type of these glycerides allows adsorbents to preferentially bind non-covalently
and retain the selected glyceride(s) in the column while the others flush through. Then,
the adsorbed glyceride(s) are recovered from washing adsorbents with an appropriate

solvent, and the adsorbents are regenerated for future use.

1.2 - Introduction on Plant Sulfolipids Separation
Glycerolipids, the major class of thylakoid membrane lipids, composed of
glyceroglycolipids and glycerophospholipids, are derived from glycerol [7]. The most

abundant glycerophospholipids, so-called phosphatides, in membranes are

phosphatidylglycerol (PG), phosphatidylcholine (PC), phosphatidylethanolamine (PE),
and phosphatidylinositol (PI). The main glyceroglycolipids are monogalactosyl-
diacylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), and sulfoquinovosyl-
diacylglycerol (SQDG) [8].

In this dissertation, the terms glycolipids and phospholipids are substituted for
glycerolipids and glycerophospholipids, which have both Cl and C2 hydroxyl groups of
the glycerol backbone esterified to the carboxyl groups of the two fatty acid chains. The
C3 hydroxyl group of the glycerol is esterified to phosphoric moiety in phospholipids,
and it attached to a sugar molecule in glycolipids. However, a glycolipid such as SQDG
is called sulfolipid if its molecule contains a sulfur atom directly bonded to a carbon as
C—SO3H [9, 10].

Figures 1.2 and 1.3 show the example structure of phospholipids and sulfolipids,
phoshatidylcholine and SQDG, respectively. The acyl R1 and R2 groups in these
molecules can be different in length and degree of saturation. For SQDG in particular,
RlCOO- is often palmitic (C 16:0) and R2COO- is linolenic (Cl8z3, A 9’12’15) [1 l-14]. The
IUPAC name of SQDG is 1,2-di-O-acyl-3-0—(6’-deoxy-6’-sulfo-a—D-glucopyranosyl)-

sn-glycerol, and PC is l,2-diacyl-sn-glycero-3-phosphorylcholine [15, 16].

/O

/N+\/\ fi CH2 /0—c<
o- l

HZC R1

1\ /

O—C\

\0

Figure 1.2 Structure of Phosphatidylchloline

CHZSO3H

CH—O 0
%
\C CH 0 c
/ \R
i w 9“ 2
H\:H—C{ H (I: R
2 1
| A. ./
OH —
\\0

Figure 1.3 Structure of Sulfoquinovosyl Diacylglycerol [17]

Phospholipids have been known for more than 100 years, but the existence of
MGDG and DGDG was unknown until 1956 [18], and SQDG was first recognized in
1959 by Benson [10]. Sulfolipids were late in being discovered, partly because of the
lack of specific and sensitive reagents for detection of sulfate and sulfonate [19]. Since
1960, not only because of the demand for understanding their role in membrane structure
and function, sulfolipids have also attracted considerable interest due to their excellent
surfactant properties and more recently due to their activity against the AIDS virus [20].

As shown, phospholipids and sulfolipids are amphiphillic molecules, having a
hydrophobic end (fatty acids) and a hydrophilic end (phosphoric acid or sugar). These
molecules are associated and bound to proteins in plant membranes from their
hydrophilic end. Thus, extractions of sulfolipids or phospholipids from plants require a
dehydrating solvent such as methanol to rupture the lipid-protein linkages. However, the
hydrophobic end does not allow it to be very soluble in this type of polar solvent. It is
necessary to include a less polar solvent such as petroleum ether, chloroform, or diethyl

ether. The most efficient solvents reported in literature for plant lipids extractions are

 

ethanol-diethyl ether (3:1 v/v) and methanol-chloroform (2:1 v/v), though they are toxic
and chloroform in particular is a possible carcinogen in humans [11, 21].

Sulfolipids have been found in all green higher plants, algae, mosses,
cyanobacteria, purple sulfur and non-sulfur bacteria [15, 22-26]. As mentioned, they are
potentially an anti-AIDS virus, but a larger testing program including tests on humans
with the AIDS virus cannot begin until sulfolipids can be obtained in much larger
quantities [20]. Abundant sources of sulfolipids are necessary, but developments of new
separation techniques using non-chlorinated solvents are also critical.

In this dissertation, a novel process to recover sulfolipids from plant extraction
using non-chlorinated and less toxic solvents will be presented in chapter 3. The key
source of interest is SQDG in alfalfa, planted on the Michigan State University campus.
The yield of SQDG using the proposed process is comparable to literature and the

amount of solvents used is less than 30 vol. % of the value reported in literature.

Chapter 2 - Selective Adsorption of Ricinolein Studies

2.1 — Overview

This chapter summarizes the adsorption studies to separate glycerides of ricinoleic
acid from non-hydroxylated fatty acid esters in castor oil. Batch adsorptions are used to
screen the adsorbent candidates, and the evaluation is based on the adsorption capacity of
adsorbents to the total oil. Once the potential adsorbent is identified, its selectivity of
hydroxylated triglycerides to the non-hydroxylated is firrther studied using fixed-bed
adsorption.

Since all glycerides in soybean oil also are the glycerides in castor oil, soybean oil
is blended with castor oil to vary the concentration of feed solutions in fixed bed
adsorptions. This reduces errors in sample analyses, due to a very small portion of
unsaturated fatty acid contents in castor oil, and provides fractionation studies over a
wider range of compositions than possible in natural castor oil.

The effect of solvents and adsorption capacities of three different types of non-
polar, acidic, and basic adsorbents are evaluated for hydroxylated and non-hydroxylated
glycerides. As expected, ricinolein can be selectively adsorbed onto acidic adsorbents
such as Florisil. In contrast, the non-hydroxylated glycerides are more selective to the
non-polar adsorbents. The separations significantly improve when a less-polar solvent is

used with Florisil, or a more polar solvent is used with a non—polar adsorbent.

2.2 - Adsorbent Selection
Fatty acids in soybean and castor oils can be categorized as non-hydroxylated (all

acids except for ricinoleic acid) and hydroxylated acid (ricinoleic acid) as shown in Table

2-1. Separation using adsorption is expected to become attractive when the desired
functional component is available in large concentration, with smaller amounts of other
components. An objective in the design of adsorption columns is to adsorb the minor
components. Because the ricinoleic acid content is high in castor oil, this study aims to
adsorb non-hydroxylated triglycerides onto adsorbents and liberate the hydroxylated in
the effluent in order to minimize the quantity of adsorbents. Thus, non-polar adsorbents
which should have high affinity to the non-hydroxylated glycerides are more preferable
candidates than the acidic or basic adsorbents. Economical adsorbents such as silica gels

and activated carbon are first selected and the cost of regeneration is also considered.

Table 2-1 Triglyceride Contents in Soy-based and Castor Oils [2]

*Values in parenthesis are measured in this work.

 

Composition of Fatty Acids (wt.%)

 

 

 

 

 

 

 

 

 

Plant oil non-hflroxylated jydroxylated
Palmitic Stearic Oleic Linoleic Linolenic Ricinoleic
C16:0 C18:0 C18:1 C18:2 C18:3 C18:1
Soybean 10.1 4.2 24.3 51.5 8.3 0.0
(10. 5) (4. 3) (25.1) (53.5) (6. 6) (0.0)
Castor 1.0 1.0 3.0 4.2 0.3 89.5
(1.2) (1.4) (3. 7) (5.0) (0.6) (88.1)
C16:0: Hexadecanoic acid Cl8:2: 9,12-0ctadecadienoic acid
C1820: Octadecenoic acid a—C18z3: 9,12,15-0ctadecatrienoic acid
C18: 1: 9-0ctadecanoic acid Ricinoleic: IZ-hydroxy-9Z-octadecenoic acid
2.3 - Batch Adsorption

Batch adsorptions of castor oil in ethanol solutions onto activated carbon, silica
gels and a polymeric resin were performed. Castor oil [8001-79-4] was purchased from

Sigma-Aldrich and the absolute ethanol of ACS/USP grade [64-17-5] from Pharmacia

was used as received. Silica gel [112926-00-8] of pore size 22 A was from Supelco and
150 A was from Analtech. Amberlite® XAD-2 nonionic polystyrene based resin of 90 A
[9060-05-3] was from Aldrich and activated carbon F-400 [7440-44-0] was from Calgon.

Activated carbon is a non-polar adsorbent, and silica gel is slightly acidic. They
both have been applied to the clean up and purification of a wide range of synthetic and
natural compounds. Amberlite® XAD-2 is also a non-polar adsorbent, commercially used
to remove antibiotics, organic nitrogen, grease and various aromatic compounds from

aqueous streams [27].

2.3.1 - Experimental Preparation and Analysis

Prior to use, adsorbents were washed using ethanol, and silica gels were pre-
heated at 200 °C to eliminate the moisture. A predetermined quantity of adsorbent was
added to an oil solution at room temperature. Equilibration was carried out overnight in a
closed container to make sure that saturation of each batch was reached. Oil solution
samples were filtered to remove any fine adsorbent particles before analyses were taken.
Since the boiling point of castor oil is significantly higher than solvents, which were used
to make the oil solutions, concentrations of total oil in samples were determined using the
gravimetric method, after completely evaporating off solvents. Differences in the initial

and final oil concentrations of the solutions were used to evaluate the adsorbent capacity.

2.3.2 - Results and Discussion
Results are summarized in Table 2-2. Adsorption capacity is defined as the
maximum amount (gram) of castor oil that can be adsorbed onto one gram of the selected

adsorbent. As shown, polymeric XAD-2 and silica gel at 22 A had a low capacity to

castor oil in ethanol solvent. No adsorption seemed to occur with silica gel at 150 A.
The discrepancy in adsorption capacity of the two silica gels could be due to l) the
evaporation of ethanol in sample analyses using the gravimetric method, or 2) loss of
solvent by evaporation during twelve-hour experiments or 3) the incomplete removal of
moisture from silica gel. Assuming silica gel at 150 A pore had the same adsorption
capacity as the 22 A pore type, it was verified that calculated concentrations of castor oil
in batches 1-3 did not change after adsorption if only 3 % ethanol was evaporated. On
the other hand, oil solutions using silica gel at 22 A were more dilute, evaporating 10 %

of ethanol only reduced half of the calculated adsorption capacity of this silica gel.

Table 2-2 Summary of Batch Adsorption Experiments in Ethanol

 

 

 

 

 

 

 

 

 

Adsorbent Batch Load Solvent qmax on
# g—adsorbent/ g-ethanol/ g-adsorbed oil/
Type Source Igiastor oil 1g-castor oil lgadsorbent
1 1.2 1.4
Silica gel, 2 2.3 5.5
pore ~150 A Analtech 3 23 9.6 0.000
4 5.6 12.5
Silica gel, Supelco 5 14.5 5.5
0. I :1: 0.0 2
pore ~ 22 A 6 9.9 12.5 0 3 O
Amberlite .
X AD-2 Aldnch 7 6.3 11.2 0.013 :t 0.002
8 3.3 8.9
Activated
Carbon Calgon 9 7.3 8.3 0.06 i 002
10 11.0 14.4
11 12.8 17.2

 

The evaporation of solvent could also be a reason causing the calculated capacity
of XAD-2 in batch experiments lower than that obtained from fixed-bed adsorptions
discussed below. It was observed that XAD-2 was significantly swollen, sticky, and

difficult to remove from oil samples in batch adsorption.

10

 

Activated carbon has the largest adsorption capacity to castor oil among the
selected adsorbents, but data are inconsistent. The inconsistency could be due to the
defects sites on this adsorbent, caused by the presence of heteroatoms such as sulfur,
chlorine, nitrogen, and metal oxides in the manufacturing process of carbon [28].

As discussed before, non-polar adsorbents were more favorable in this study. The
acidic adsorbent, silica gel, showed the same capacity to castor oil as the non-polar
polymeric XAD-2. But, packing and backwashing of silica gels at the experimental
condition were difficult due to the fine texture. Therefore, silica gels were not included

in fixed-bed adsorption studies.

2.4 - Fixed-bed Adsorption

In addition to the polymeric XAD-2 and activated carbon used in batch
adsorptions, Florisil® [1343-88-0, 16-30 mesh, standard grade], and three different types
of methylene-bridge divinylbenzene-styrene copolymer adsorbents were selected for
fixed-bed adsorption studies: (Dowex® Optipore SD-2 [374558-57-3], Dowex® M-43
[195215-44-2], and Dowex® L-493 [211502-88-4]).

Florisil®, an acidic adsorbent, is a co-precipitate of silica and magnesia (MgO3Si),
extensively used in the chromatographic analysis of lipids and pesticides [29-31].
Dowex® Optipore SD-2, a non-polar adsorbent, containing the functional group of
tertiary amine, has been used as an alternative to activated carbon, for decolorization,
taste and odor removals in the processing of corn syrups and high fructose corn syrups.
Dowex® M-43 is a weak-base exchange resin, used to remove mineral and organic acids
such as acetic, formic, propionic, benzoic, halogen, sulfuric and phosphoric, from the air

and solutions. Dowex® L-493 is a hydrophobic resin with non-catalytic activity,

11

 

considered a better option than carbon in removing natural organics from water. In
general, polymeric materials have a low swell of ~ 5%, allowing for easy vessel design,
can potentially be recovered for future use, and they do not require furnace regeneration.

More details of the Dowex® adsorbents are available on the Dow Chemical web site [32].

2.4.1 - Fixed-bed Design

A fixed-bed was constructed using stainless steel of 3/8” OD x 6” length, shown
in Figure 2.1. All feed tubes used in this design were 1/16” OD. Adsorbates in feed
solution were fed to the fixed-bed using a micropump® (ELDEX, A-30-S model). Table

2-3 provides a summary of fixed-bed run conditions.

Fixed-bed

       
    

Feed solution

, . '
a... :f z
A

3 MI cropump

Figure 2.1 Fixed-Bed Design

Table 2-3 Summary of Fixed-bed Adsorption Experiments

 

 

 

Feed Solution Adsorbent
Run Castor Soy Solvent Flowrate Name Type Amount Volume

# (wt.%) (wt.%) (ml/min) (g) (ml)
1 1.05 0 ethanol 0.18 Carbon non-polar 2.256 4.8
2 0.72 0 ethanol 0.30 Carbon non-polar 2.256 4.8
3 0.68 0 ethanol 0.14 Carbon non-polar 2.276 4.8
4 0.99 0 ethanol 0. 1 8 Carbon non-polar 2.276 4.8
5 1.01 0 ethanol 0.18 Carbon non-polar 2.275 4.8
6 1 0 ethanol 0. 16 XAD-2 non-polar 1 .930 3 .8
7 2.46 0 ethanol 0.16 XAD-2 non-polar 2.802 5 .5
8 0 3 .69 ethanol 0.18 SD-2 non-polar 2.833 5 .0
9 0.8 1.6 ethanol 0.17 SD-2 non-polar 2.901 4.9
10 1.99 0.41 ethanol 0.19 SD-2 non-polar 2.836 4.8
l l 0.38 1.99 ethanol 0.20 SD-2 non-polar 2.837 4.8
12 0 2.44 ethanol 0.20 SD-2 non-polar 2.826 4.5
13 0 2.42 ethanol 0.19 SD-2 non-polar 2.83 5 4.5
14 0 2 .42 ethanol 0.19 SD-2 non-polar 2.832 4.5
15 3 .76 0 ethanol 0.] 8 SD-2 non-polar 2.832 4.5
16 1.2 1.21 ethanol 0.19 SD-2 non-polar 2.834 4.5
17 2.4 0 ethanol 0.17 SD-2 non-polar 2.835 4.5
18 1.64 0.84 ethanol 0.19 L-493 non-polar 2.865 4.9
19 0.82 1.63 ethanol 0.18 L-493 non-polar 2.827 . 4.8
20 2.48 0 ethanol 0.19 L-493 non-polar 2.875 4.9
21 2.51 0 ethanol 0.19 M-43 basic 2.837 4.5
22 0.8 1.59 ethanol 0.19 Florisil acidic 2.369 4.7
23 0 3 .77 hexane 0.14 Florisil acidic 2.408 4.7
24 0 2.46 hexane 0.14 Florisil acidic 2.407 4.7
25 1.43 0 hexane 0.21 F lorisil acidic 2.427 4.7
26 1.29 1.3 hexane 0.18 Florisil acidic 2.413 4.7
27 1.66 0.85 hexane 0.18 Florisil acidic 2.468 4.8
28 0.8 1.68 hexane 0.17 Florisil acidic 2.462 4.8
29 2.44 0 methanol 0.20 SD-2 non-polar 2.835 5.0
30 2.38 0 propanol 0.20 SD-2 non-polar 2.838 5.0
31 1.63 0.84 propanol 0.16 SD-2 non-polar 2.856 5.0
32 0.81 1.63 propanol 0.19 SD-2 non-polar 2.835 5.0

13

 

2.4.2 - Sample Analysis

Bed Volume -— Before adsorbents were packed into the fixed-bed, the volume of
wetted adsorbents (Vads), the so-called one bed volume, was measured using either a
buret or a graduated cylinder. The liquid flow was measured in bed volumes (BedVol)

defined as Bed Vol = F * T / Vads where F is the actual flow rate (ml/min) of feed solution,

T is time (minutes) at which the sample is collected.

Bed loading procedure — Debris was removed and adsorbents were washed using
the elution solvent. Then, slurry of adsorbents was slowly poured into the column (up to
80 % of column’s volume) and backwashing was performed to remove air bubbles.

Total oil concentration — Fractions containing about 23 g of the effluent from
fixed-bed were collected over 15-20 minute periods. The gravimetric method was used
to determine the total oil content in samples by evaporating solvent. For example, in run
# 22, the initial oil concentration Co = 2.42 %, fraction 5 was collected between T = 65
min and T = 85 min; tared = 25.4993 g, weight of sample (solvent + oil + tared) =
28.6127 g. After evaporating off solvent, weight of sample (oil + tared) = 25.5696 g.

_ 0
(25.5696 25.4993) = 2.26 % , and C/C, = 2.26 A)
(28.6127 — 25.4993) 2.42 %

C = oil % = = 0.933

 

 

Glyceride concentrations and FAME method — After evaporating off the
solvents to determine total oil concentration, samples of the effluents were converted to
fatty acid methyl esters (FAME) to find out the contents of non-hydroxylated and
hydroxylated triglycerides. A method was developed and verified to be compatible and
more efficient than the standard esterification method (AOCS, Ce 2-66) [33]. The

standard method requires the use of boron trifluoride (BF3), which is extremely

l4

flammable and decomposes on exposure to moisture. The alternative method, developed
in this study, eliminated the use of nitrogen, glove box to store BF3, and the heat to
elevate the chemical reaction in converting triglycerides into fatty acids methyl esters.

A droplet of each sample was blended in about 1.5 ml of hexane. After dissolving
in hexane, the sample was mixed with 0.5 ml of 2M KOH in methanol solution and
shaken thoroughly for 30 seconds. Finally, 2 ml of aqueous saturated KCl was added to
separate the methyl esters into the hexane phase.

The analyses of FAMEs were performed using gas chromatography. The
chromatograph (GC) was a Perkin-Elmer model 8500 equipped with a FID detector and

an Econo-Cap EC-WAX capillary column (15m x 0.53 mm ID x 1.2 pm). The column

and standard triglycerides were purchased from Alltech Associate, Inc [34]. The GC
used helium gas at 5m1/min and 5.0 psig. At t = 0 min, oven temperature (Tom) = 65 °C;
t = 4.5-12.5 min, Toven = 200 °C, and t =15.8 -— 21.8 min, Tom = 250 °C.

Figure 2.2 is an example GC Chromatogram of a castor-soybean oil mixture.
Assuming the same response factor for all FAMEs, mass concentrations of the
hydroxylated and non-hydroxylated glycerides in oil samples were determined from the
corresponding GC peak areas. The calculated fatty acids contents in soy-based and castor
oils from GC analyses using this assumption are in excellent agreement with literature,
shown in Table 2-1. GC analyses were also done for castor-soy oil mixtures to compare
with the calculated CW and C20 using the weight method. Difference between the two

methods was about 7 %, which is acceptable for GC analyses.

15

17.13

Ricinoleic

Linoleic

 

.89

8

Palmitic
8 21 01e1c

6.09

 

L1nolenic

 

 

10 O9

 

 

, 19.60
END

 

 

 

 

 

86

i 7.91 Stearic
r
I

E a.

.J

Figure 2.2 Example of Gas Chromatogram of Fatty Acid Methyl Esters
(FAMEs were derivatized from castor-soy (2:1 wt%./wt%.) oil mixture)

16

Breakthrough behavior — Table 2-4 and Figure 2.3 are the examples of

adsorption data and breakthrough curves included in Appendix A. In table 2-4, “Fraction

denotes the sample collected in the experiment, b: time (minutes) after fixed-bed starts, C

: number of bed volumes, d : total oil concentration, c’ f’ g: concentrations of total oil (C),

ricinolein (C1), and non-hydroxylated component (C2) in the effluent compared to their

Initial Co , Cpo, C2,0.

castor oil.

Table 2-4 Sample of Fixed-bed Adsorption Data (Run # 22)

Calculations of Cs are based on 88.1 wt. % of ricinoleic acid in

 

 

aFraction ”Time °Bed Vol doiI % ecrco ‘cucm 9C2IC2,0

1 15 0.620 0.05% 0.021 - -

2 32 1.323 1.08% 0.448 0.465 0.432
3 48 1.984 1.99% 0.824 0.756 0.840
4 65 2.686 2.21% 0.915 0.937 0.889
5 85 3.513 2.26% 0.933 0.882 0.939
6 115 4.753 2.32% 0.960 - -

7 133 5.497 2.34% 0.969 1.129 0.880
8 151 6.241 2.35% 0.971 1.291 0.810
11 183 7.563 2.37% 0.978 - -

12 222 9.175 2.41% 0.996 1.242 0.868

 

 

Adsorption capacity — As defined in the previous section, adsorption capacity
(qmax) of an adsorbent is the maximum amount of the adsorbed component held-up by one

gram of that adsorbent. In fixed-bed studies, it was calculated as follows:

 

amount of adsorbed oil (g) (2 1)

qmax, 0'" = 1 gram of adsorbent

A‘ (2.2)

Amount of adsorbed oil = total load“
A] '1' A2

 

17

Total load = Vads * Bed Vol *oil %*solution density (2.3)

where Vads, and oil % are already defined above. Bed V0150, is the number of bed volumes
at saturation point. Because the oil concentration was low, solution density was assumed
to be the same as solvent density in calculating the total oil solution loaded into the fixed-
bed. A. and A2 are respectively proportional to the amounts of adsorbed and non-

adsorbed oil at saturation as shown in Figure 2.3; the Al /(Al + A2) value in Eqn 2.2 was

determined by integration using the paper weight method. Though the concentrations in
the bed were not uniform spatially at all times, this method provides a quantitative

method to differentiate between various experimental conditions.

 

 

 

 

 

 

1.4 -
. 0 """"""""""
1.2 - 3 o
O. .0
a; 1.0 ~
0
5; 0.3 4
$2
5 0.6 ~
g 0.4 4
0.2 .
0.0 ' I I
o 3 e 9 12

Bed vol um e

Figure 2.3 Breakthrough Curves of Total Oil and Glycerides (Run # 22)
0: total oil, 0: hydroxyl, A: non-hydroxy, using Florisil and ethanol.

Selectivity of adsorbents — Similar to the relative volatility which measures the
simplicity in separations by distillation, a separation factor described below is usually

used to determine the equilibrium selectivity in adsorption [30]:

18

a $2.
AB XB YA

(2.4)
where XA and YA are the mole fractions of component A in adsorbed and fluid phase at

equilibrium, respectively. For qualitative analyses in this work, Eqn 2.4 can be modified

as follows:

A A
aOH -non0H : L0H 2,non0H (2'5)
A2,0H A1,non0H

 

where 0H and nonOH respectively denote hydroxylated and non-hydroxylated
components, A’s and A2’s are the areas enclosed by breakthrough curves of the
corresponding glycerides, C/C0 =1 and Bed volume = BedVolW, as illustrated for total oil

in Figure 2.3.

2.4.3 - Results and Discussion
Results of fixed bed adsorptions are summarized in Table 2-5. There were five
different experiments performed using activated carbon, but data related to this adsorbent

were omitted due to their inconsistency as discussed in section 2.3.2.

Table 2-5 Summary of F ixed-bed Adsorption Results

 

 

 

Adsorbent
Name Type s°"’°"t qmax,otl “OH-nonOH
M-43 weak base ethanol 0.06 i: 0.03 0.78 i 0.10
L-493 non-polar ethanol 0.10 i 0.03 0.45 :I: 0.21
XAD-2 non-polar ethanol 0.10 i 0.02 0.71 d: 0.20
SD-2 non-polar propanol 0.08 :l: 0.01 0.59 :t 0.18
SD-2 non-polar ethanol 0.12 i 0.02 0.55 i 0.12
SD-2 non-polar methanol 0.13 :l: 0.02 0.54 :l: 0.20
F lorisil acidic ethanol 0.06 2+: 0.01 0.26 i 0.10
Florisil acidic hexane 0.17 i 0.02 6.04 i 0.99

 

l9

Non-polar adsorbents have larger adsorption capacities to the total oil (~ 10 wt.%)
than basic and acidic adsorbents (~ 6 wt. %). The deviation from multiple runs and
deviation in determination of A1, A2 using the weight method were combined in
calculating the accuracy of qmam. and 01’s. The values of acmmnog for all adsorbents are
less than 1.0 when alcohols were used as solvents, indicating that non-hydroxylated
glycerides were preferentially adsorbed in a fixed-bed. Except for Florisil with hexane,
the 01's of adsorbents with alcohols are very much the same, showing similar capability
for separating glycerides of castor oil. But, separations in alcohols were not very
efficient because 01's are relatively close to unity.

Effect of solvents — The effect of solvents is significant in adsorption. More oil
adsorbs to Florisil and the selectivity of this adsorbent is switched from retaining non-
hydroxylated to hydroxylated glycerides if a non-polar solvent such as hexane is

substituted for ethanol (Figures 2.3 and 2.4).

 

     

 

 

 

2.5
.3- 2.0 —
Q
N
0.. 1.5 -
O
F. A'A ..... A
O A A .....
2 1.0 ~ A A
0..
O
2 0.5 ~
0 .
0.0 ~ """ O‘I‘O”O”O”Tflocu I I I
0 3 6 9 12 15 18
Bed volume

Figure 2.4 Fixed-bed Adsorption Using Hexane and Florisil (Run # 28)
9: total oil, 0: hydroxyl, A: non-hydroxy.

20

A similar effect to the selectivity of adsorbent is also found in adsorption with
SD-2. The a0”.,,0,,oy value increases when solvent changes from methanol to ethanol

and propanol. However, the SD-2 capacity for total oil shifts to the opposite direction
compared to the Florisil. More oil adsorbed to Florisil in hexane than in ethanol, but less
oil absorbed to SD-2 in propanol than SD-2 in methanol (Table 2-5).

Adsorbents ’ regenerability - All adsorbents used in the fixed-bed adsorptions can
be fully recovered using methanol. In the pairs of runs 1 and 2, 3 and 4, l3 and 17, 14
and 15 (Table 2-3), the first run was performed with fresh adsorbent and the second run
used regenerated adsorbents. Results show that regenerated adsorbents provided the

same adsorption capacity and selectivity as the fresh adsorbents.

21

Chapter 3 - Extraction and Purification of SQDG

3.1 — Overview

Extraction of lipids is commonly done using solvent mixture of chloroform-
methanol. However, chloroform is very toxic and a possible carcinogen in humans [35].
Isolation of any desired lipids from a multicomponent system requires a series of unit
operations. In the recovery of high-value lipids, the cost is often determined by handling
large volumes of solvents, thus rapid reduction of volume is preferable.

This chapter presents the results from the two-stage process: (1) extraction and
coarse fractionation; and (2) purification of SQDG, using non-chlorinated solvents. The
objective of the coarse fractionation in this work was to isolate a glycolipid fraction that
could be purified in subsequent steps. Initially, studies were performed using only minor
variations from published methods. The initial results are presented to justify the
modifications that led to the optimized process.

A proposed method to obtain sulfolipids from alfalfa using non-chlorinated
solvents was developed and demonstrated. This method gave a compatible yield of
SQDG as it is reported in the literature [11], but used less solvents and chemicals.
Significant findings included the use of hot 2-propanol at 50-55 °C to deactivate lipase
degradation of lipids during stage (1) of the process. These findings eliminated the need
for other reagents used in the literature including dry ice or liquid nitrogen, organic acids,

inorganic salts, and Florisil adsorbent.

3.2 - Stage (1) — Extraction and Coarse Fractionation Studies

Solvent selection involves two primary properties: volatility and solvent power.

22

Following lipid extraction, evaporation is commonly used to concentrate the extract
before another separation such as crystallization or adsorption can be performed.
Evaporating off solvents, however, cannot be done at high temperature, due to the heat
sensitivity of the most biological molecules. As discussed in chapter 1, section 1.2, there
is no literature report successfully using a single and non-toxic extraction solvent, with a
low boiling point, and compatible with the amphiphilic behavior of plant lipids.
Literature extractions use methanol-chloroform (2:1 v/v) predominantly. The starting
point in this study was to evaluate the extraction of SQDG using relatively low boiling
point blended-methanol solvents which have similar total Hildebrand values as that of

chloroform-methanol mixture.

3.2.1 - The Hildebrand Solubility Parameter
The Hildebrand solubility parameter 8 [36, 37] is considered the foundation of

solution theory, defined as follows:

V

m

6=f=[AH-RT]”2 (3.1)

where c is the cohesion energy density, AH is heat of vaporization, R is gas constant, T is
temperature, and Vm is molar volume. The dimension of 8 is (pressure)”2.

As it relates to the cohesion energy density or heat of vaporization, the Hildebrand
parameter is strongly affected by molecular interaction forces. It can be described using

the Hansen [3 8] three-parameter approach:
52 = 53, +6}, +6; (3.2)
where Ed is the dispersion contribution, SP is the polar contribution, and BI, is the hydrogen

bonding contribution.

23

In application, the Hildebrand parameter of a solvent mixture can be theoretically
calculated by averaging the Hildebrand parameters of the individual solvents by volume

fraction [39]. For example, mixture of 32 vol.% acetone (8 = 20.3 MPam) in toluene (5 =
18.2 MPam) has 5 value = 0.32*20.3 + 0.68*18.2 = 19.0. This approach was used to

calculate the mixing ratios for the selected solvent systems.

3.2.2 - Solvent Selection

Toluene, ethyl acetate, and acetone were selected from aromatic hydrocarbons,
esters, and ketones for this study. The selection was based on the similarity of the
Hildebrand values of these molecules and chloroform’s, and also the relatively lower
boiling points of these solvents compared to most of their molecular family members.
Extraction in chloroform was used as a benchmark to evaluate the selected solvents, and
trichloroethylene was used to determine how extraction results changed if the selected

solvents were replaced by their homologous molecules.

Table 3-1 Solubility Parameters of Related Solvents [40]

 

 

 

 

Solvent 5 (MPa)“2 8.4141321)"2 8p(MPa)l/2 mantel)“2
Toluene 18.2 18.0 1.4 2.0

Ethyl acetate 18.6 15.8 5.3 7.2
Trichloroethylene 18.8 18.0 3.1 5.3
Chloroform 19.0 17.8 3.1 5.7
Acetone 20.3 15.5 10.4 7.0
Methanol 29.7 15.1 12.3 22.3
Isopropanol 23 .5 15 .8 6.1 16.4
n-Hexane 14.9 14.9 0.0 0.0

Water 47.9 15.5 16.0 42.3

 

24

Tables 3-1 and 3-2 list the solvent systems and their solubility parameters. The
mixing ratios were assigned in order to have the same theoretically calculated Hildebrand
value for all systems compared to that of chloroform-methanol (2:1 v/v). The use of

formic acid and acetic acid are discussed in section 3.2.6.

Table 3-2 Solvent Systems Used in Extraction Studies

 

Volume Ratio

 

 

 

 

 

 

 

System
(1) A?) (3)
methanol (1) + toluene (2) + formic acid (3) 1 2.09 0.100
1 1.87 0.000
methanol (1) + acetone (2) + formic acid (3) 1 1,87 0,050
1 1.87 0.100
methanol (1) + chloroform (2) + formic acid (3) 1 2'00 005°
1 2.00 0.100
1 2.04 0.000
methanol (1) + ethyl acetate (2) + acetic acid (3) 1 2.04 0.050
1 2.04 0.100
1 2.04 0.075
methanol (1) + ethyl acetate (2) + formic acid (3) 1 2.04 0.100
1 2.04 0.050
methanol (1) + trichloroethylelne (2) + formic acid (3) 1 2.02 0.100

 

3.2.3 - Experimental Preparation

Alfalfa — Alfalfa containing 64-68 wt.% of moisture was harvested in the second
week of October 2003, from Michigan State University campus fields. Both stems and
leaves were collected to represent a commercial harvest. At the time of collection, a flail
harvester chopped the alfalfa stems and leaves into pieces that were about 1-2 inches in
length. The chopped alfalfa was pretreated with isopropanol or kept frozen at -5 °C for
future use. The moisture of alfalfa was measured using a MB-200 Ohaus moisture

balance.

25

Chemicals — A SQDG [59-1230-7] standard was purchased from Larodan Fine
Chemicals Company. Acetone [67-64-1], chloroform [67-66-3], toluene [108-88-3],
ethyl actetate [141-78-6] and hexane [110-54-3] were from Burdick & Jackson Company.
Methanol [67-56-1] and trichloroethylene [79-01-6] were from Fisher Scientific. Glacial
acetic acid [64-19-7], ammonium hydroxide [1336-21-6], and potassium phosphate,
monobasic [7778-77-0, crystal] were from EM Science, Inc. Potassium chloride [7447-
40-7, crystal, 2 99.0% grade] was from Spectrum Quality Products, Inc. Ammonium
sulfate [7783-20-2, crystal] was from Columbus Chemical Industrial, Inc. Formic acid
[64-18-6, 88 wt.%], sulfuric acid [7664-93-9, 98%], water [7732-18-5, HPLC grade], and
thin layer chromatographic plates [“Baker” 81250] were from J .T. Baker, Inc. DEAE-
cellulose (diethylaminodiethyl cellulose) [9013-34-7, 100-200 um pore] was from
BioChemika. Myristic acid [544-63-8, crystal, 99-100% grade] and F lorisil® [1343-88-0,
16-30 mesh, standard grade] were from Sigma-Aldrich. All solvents and reagents were
used as received, if not otherwise specified.

T kin-layer chromatographic plates — Sulfuric acid 50 wt.% and Ot-naphthol 2.4

wt.% in ethanol-sulfuric acid solution (8:1 v/v) were prepared for sample analyses. The
plates were impregnated with 0.15 M ammonium sulfate and dried in air, then activated

at 120 °C for 60-90 minutes before use.

3.2.4 - Extraction Procedure

Extraction —Alfalfa was mixed and ground in dry-ice using a mortar and pestle.
Extractions were performed on a scale of either 5 g or 50 g basis of alfalfa including
leaves and stems. For a 5 g basis, the ground alfalfa was placed in a filter bag,

submerged into the selected solvent mixture and squeezed repeatedly to collect the

26

extract. For a 50 g basis, the ground alfalfa was blended with solvent mixture using a
stainless steel blender, and then the extract was filtered though a vacuum funnel. In both
cases, fresh solvent was added and the process was repeated until the alfalfa turned light
yellow. All the extract was combined after extraction, and carried to the phase
separation.

Coarse Fractionation — The inorganic substances and polar lipids being more
polar than glycerolipids were eliminated from the extract using either saturated aqueous
potassium phosphate, monobasic (KH2PO4, d = 1.146 g/ml, pH = 4.02) or potassium
chloride solution (KCL, pH = 6.35) [22, 41]. Each volume of the extract was allowed to
contact with 0.2 volume of the salt solution, then centrifuged in a benchtop centrifuge at
1500 rpm for 2-3 minutes to completely separate the organic phase from the solid and the
aqueous phase.

An alternative method was applied to the methanol + acetone extract, which will
be discussed in section 3.2.5. Following coarse fractionation, the extract (organic phase)
was then evaporated at room temperature and 10-20 ian of vacuum to dryness, and the

obtainable product called dry extract.

3.2.5 - Sample Analysis

To this point, dry extract contained not only glycolipids including SQDG, but also
phospholipids, chlorophyll, and other pigments. Extractions were evaluated for the
SQDG content in the dry extract using one-dimensional thin-layer chromatography
(TLC) and gas chromatography (GC). Procedures of these analytical analyses and
examples of calculations are described below.

TLC Analysis - To minimize the error in using microbalance due to a small

27

amount of dry-extract (~ 0.05 g) obtained from 5 g basis of alfalfa, all dry extract was re-
dissolved in the same solvent mixture (~ 2 ml) used in the extraction to become dry
extract solution, (excluding the organic acids).

Samples of dry extract solution and standard SQDG were placed on TLC plates in

series of 2-3 11L drops, using a micropipet (Gilson, P20 model). The drop size was kept

to be less than 3 mm in diameter, and TLC plates were dried under nitrogen between
drops. The remainder of the dry extract solution was re-evaporated to determine the
amount of dry extract placed on the TLC plates.

Several mobile phases were evaluated for TLC analyses using acetone-toluene-
water (91:38:6 v/v/v, 9123427 v/v/v, 91:30:8 v/v/v, 85:66:0 v/v/v, 75:41:] v/v/v). The
mixture acetone-toluene-water (9123427 v/v/v) provided the best resolution; the retention
factor (Rf) value for lipids of interest were PC < PE < PI < SQDG < DGDG < PG <
MGDG.

Glycolipids were visualized by alpha-naphthol and sulfuric acid, in which MGDG
and DGDG bands were dark-blue, and SQDG was pink [42]. To locate lipids, the TLC
plate was sprayed with sulfuric acid, heated up to 150 °C in a vacuum oven for five

minutes then sprayed with the a—naphthol solution. The reaction of SQDG with 01-

naphthol is irreversible; therefore this reagent would only be used for a qualitative
purpose. The color of the lipid bands varied with treatment temperatures and
concentration of sulfuric acid. The SQDG band was dark blue or black if the plate was
over-sprayed or heated too long.

Phospholipids and SQDG were also identified from light-brown bands when the

TLC plate was stained quickly with iodine vapor [22, 43]. This qualitative method of

28

SQDG was based on the reaction of iodine with unsaturated compounds. It was
reversible and there was no effect on the SQDG band, if the contact time was short. The
SQDG band was scraped off the TLC plates for further quantitative analysis.

GC Analysis — The collected SQDG band from the TLC plate was derivatized

into methyl ester of fatty acids (FAME) for GC analysis. The SQDG sample and 5 pg of

myristic acid, an internal standard, were placed in a glass tube, sealed with a Teflon cap
and allowed to react with one ml of 1.0 N HCl in methanol at 80 °C for 40 min. Then,
the mixture was cooled down, and the FAMEs were extracted into hexane, using 1 ml of
hexane and 1ml of sodium chloride 0.09 wt.%.

Gas chromatography was performed at the same condition as described in chapter
2. Results showed that the major FAMEs derivatized from SQDG of alfalfa were
palmitate (~ 48 wt.%) and linolenate (~ 42 wt.%), assuming the same GC response
factors for all methyl esters including myristate. The C 18:1 (~7 wt.%) and C 18:2 (~3
wt.%) were also found in gas chromatograms, but they were not well recognized in the

literature, therefore calculating yield of SQDG only included C 16:0 and C 18:3 contents.
Example of calculation — In the extraction using toluene-methanol (2.09: 1 v/v),

Alfalfa = 4.95 g, total solvent used = 24.7 ml, total dry extract = 0.043 g

TLC sample size = 0.006 g, amount of myristic acid used in GC sample = 5 pg
GC peak areas: C 14:0 = 578.4, C 16: 0 = 1264, C 18: 3 = 855.8

FAME ofC 16:0, Mw =256, FAME ofC 18:3, Mw = 278

If both R1 and R2 of Figure 1.3 are C 16:0, Mw of SQDG = 794

IfR1 is C 16:0 and R2 is C 18:3, Mw ofSQDG = 816

29

, 1264 * 0.043 _

 

 

Total C 16 : 0 from extract'on = 5 — 4
l “g 578.4 0.006 “g
Total C18 : 3 from extraction 2 Sjug * 855.8 * 0043 = 27.9ug
578.4 0.006

 

54 27.9 27.9
TotalS DG :05 *794 ——— +816* =125.8
Q “g (256 278 i 278 “g

= 125.8 = 25 411g SQDG

Yield .
4.95 1g alfalfa

FAB-MS Analysis — The TLC band which was identified as SQDG using iodine
vapor in the extraction with ethyl acetate was also analyzed by Fast Atom Bombardment

Tandem Mass Spectrometry (FAB-MS).

iffifiggiggggpup~fwwm~--

 

 

 

 

 

 

 

 

[R1 : 1.23 min Scan# : (1,8) g
100 Mode : FBB- Int. : 14.85 [
442512 1
815.21 ]
3a 4
28»—
j 831.21
‘8 813.18 IA h‘
jiii' 1W0 A 853.28
1 l" l I UNI A ’
2 A wW‘Aw’t Ii l iUiAACNAAAAAAJ l 1110‘ 11.111411Imliu’lxttxmumwwwmwmm 1
. , , , , , m .u., I , .2r_
see 818 82a 83a 848 858 868 872 888 898
1
m / 7_ i

 

Figure 3.1 FAB-MS Chromatogram of SQDG Sample

Using a similar condition as described by Gage et al. [44], the predominant

molecular species [M-H]_ Was found at m/z 815 (Figure 3.1). This molecular species fits

30

the structure of SQDG with palmitic (C 16:0) at sn-l and linolenic (C 18:3) at sn-2

positions as described in the literature for alfalfa [1 1-13].

3.2.6 — Results and Discussion

Results of extractions given in Table 3-3 clearly show that SQDG can be
extracted from alfalfa using any solvent systems listed in Table 3-2. The deviation in the
reported yield is relatively large, due to the error in weighing small amounts of dry
extracts and the incidental loss of SQDG during the evaporation process. Extractions
using acetone, trichloride ethylene or chloroform gave a similar yield of SQDG, which
was apparently higher than that obtained from toluene and ethyl acetate. TLC analyses
were used to verify that the variations in yield of SQDG were not due to the loss into the
aqueous extraction phase.

Figure 3.2 describes the extraction and evaluation scheme. TLC showed that

toluene, the least polar solvent (smallest 8p), extracted the largest quantity of non-polar

lipids and pigments compared to other solvents. However, a very small amount of non-
glycophospholipids (polar lipids) was extracted along with SQDG in acetone. All solvent
systems listed in Table 3-1 were miscible. Chloroform and trichlorolethylene have
higher liquid densities than water, forming a two-liquid-phase extract before salts were
added; the lower phase was organic containing the extractable lipids and the upper phase
was aqueous. Conversely, the non-chlorinated solvents produced only one liquid-phase
extract, and glycolipids migrated to the upper phase when salts were added. In addition,
neither potassium phosphate, monobasic nor potassium chloride solution was used for
acetone + methanol + formic acid systems as already mentioned. The extract was

completely miscible in salt solutions. This behavior of acetone is different from the other

31

solvents, resulting from the high contribution of 511 and 5p in the Hildebrand solubility

value. The highly polar carbonyl group in acetone can form H-bonds to glycolipids and
solvate to potassium and phosphate monobasic or chloride ions. As further evidence of

acetone’s polarity, neutral lipids extracted from boar testis are insoluble in acetone at

4 °c [45].

Table 3-3 Result from Extraction Studies

 

 

 

 

Alfalfa Solvent Dry SQDG analysis Yield

Solvent used extract C16:0 C18:3 SQDG pg SQDG
(g) (m1) (8) ( 119) ( 119) ( 1119) (Par 18 of alfalfa)

Toluene 4.95 24.7 0.043 54 28 126 25.4
Ethyl acetate 5.02 24.3 0.073 64 49 174 34.6
Trichloroethylene 4.99 24.2 0.062 157 95 3 86 77.4
Chloroform 5.01 36.0 0.065 166 137 463 92.4
Acetone 5.02 23.0 0.078 163 126 442 88.0

 

* The yield of SQDG in literature is ~ 3 % of the total extractable lipid [l 1]. This value is equivalent to 30-
150 1.1g SQDG per lg of alfalfa containing 64-68 wt.% of moisture assuming 1-5 % of alfalfa is extractable
lipids [12, 46].

It has been reported that naturally occurring sulfolipase can be activated by
grinding or partial degradation of plant tissue [47]. Degradation of plant tissues and/or
the activation of sulfolipase in grinding alfalfa were observed from the presence of the
extra bands below SQDG when acids were not used in the extraction step. More
degradation was seen in extraction using non-chlorinated solvents, especially in
extraction using ethyl acetate. The two-liquid phases formed in the extraction with
chloroform as described may help to isolate sulfolipid from water and impede hydrolysis
and lipase reactions. Increasing the use of dry ice or liquid nitrogen to keep the
temperature as low as possible did not eliminate degradation, but bands from degradation

products were not shown on the TLCs if formic acid or acetic acid (acid-solvent, 0.02: l

32

v/v) was added to solvents. As a result, formic acid was used for most studied cases.

Formic acid has a lower boiling point than acetic acid and was easier to remove from the

extract using evaporation.

penetrate the tissue and inactivate the phospholypase [48].

Solid waste F——‘

 

 

Frozen alfalfa

 

 

 

 

 

Extract with
solvent

 

 

 

    
    

If solvent is
acetone

 

Add sat. aqueous

 

Centrifuge

 

 

 

 

 

Aqueous and

KH2P04

 

 

 

 

 

l

 

 

Organic phase

 

 

solid phases

 

 

Evaporation

 

 

 

     
 

f solvent was
acetone

1% Residue water

 

 

Re-dissolve with
solvent

 

 

 

Calculate
the yield of .
dry extract

 

Dry extract

 

 

 

 

 

 

Evaporation
(second time)

 

 

 

 

Re-dissolve with
solvent for TLC
analysis

 

 

 

 

 

Identify SQDG band
and scrape off

 

 

 

 

 

Derivatization to
FAME

 

 

In addition, formic acid has been used as a fixative to

Solid waste
(waxes)

 

GC analysis
for SQDG
composition

/

 

Figure 3.2 Extraction and Evaluation Scheme in Extraction Studies

33

3.2.7 — Improved Extraction and Coarse Fractionation

Further studies were conducted aimed at either reducing the excessive non-polar
lipids, chlorophyll, and pigments in extraction; or efficiently removing them from the
extract in phase partitioning. First, extractions were performed using only methanol or
acetone. It was found that an insignificant amount of SQDG could be extracted in
acetone alone, and no SQDG was extracted using pure methanol. Therefore, extractions
were carried out using isopropanol, which has the Hansen hydrogen bonding parameter
between the parameters of methanol and acetone. At room temperature, the extract was
viscous but it contained SQDG. Assuming the viscosity of extract was due to the low
solubility of proteins and non-lipids in isopropanol, extractions were repeated using hot
isopropanol. Results showed no extra TLC bands related to degradation of lipids, and a
complete extraction could be obtained using hot isopropanol, followed by hexane-
methanol (3:5 v/v). No lipids from the residue of extraction with hexane-methanol could
further be extracted in chloroform-methanol (2:1 v/v). This indicated the completion of
the above extraction using hot isopropanol and hexane-methanol, while simultaneously

eliminating lipid degradation.

To improve the coarse fractionation, liquid-liquid partitioning to remove non-
polar lipids from the extract was evaluated. When hexane was added together with
K2HPO4 solution to the extract of alfalfa in acetone + methanol + formic acid ( 1: 2.04:
0.1 v/v/v), the extract became three phases: the rich pigments and chlorophyll phase was
in the top, all extractable lipids including SQDG were in the middle, and solids were in
the bottom. Similar results were obtained when replacing acetone with ethyl acetate and

hexane by pentane, but SQDG was present in both the top and middle phases.

34

In a preliminary study to explore options for coarse fractionation, adsorption of
the extracts using silica gel [112926-00-8, 150 A pore] and activated carbon F-400 [7440-
44-0] were also studied. Results showed that chlorophyll and pigments were poorly
adsorbed onto silica gel, but they could be removed from the extract using activated

carbon.

3.3 - Stage (2) - Purification Studies

The dry extract was redissolved in solvent for the purification stage of the
process. The objective of this work was to find a replacement for chloroform used by
Benson [11]. Since acetone-methanol successfully extracted SQDG from alfalfa, the
mixture was used as the base solvent in purification studies. First, SQDG, DGDG, and
MGDG were separated from pigments, non-polar lipids and phospholipids using Florisil.
Then, SQDG was isolated from DGDG and MGDG using DEAE-cellulose ion-
exchanger. These adsorption and ion-exchange processes were monitored by TLC as

described earlier. Studies were conducted using 5 g basis of alfalfa in extractions.

3.3.1 - Adsorption using Florisil

Experimental procedure — Similar to the procedure described in literature [11],
10 g of Florisil adsorbent (~12 ml in volume) was wetted in methanol, slurry packed into
a glass column (SO-ml burette, 1.1 cm ID x 64 cm), and washed with 100 ml of methanol
followed by 50 ml of acetone. Then, the dry extract obtained from a complete
evaporation was re-dissolved in acetone-methanol (2:1 v/v) and loaded to the prepared
F lorisil column. The liquid level in the column was always kept at 0.5 cm above the

Florisil and flow rate of solvents through the column was about 2 ml/min.

35

Results and Discussion — Hexane and acetone-methanol at different gradients
were used to elute lipids from the Florisil column. Hexane is a non-polar solvent;

selected for elution of chlorophyll and pigments, based on extraction studies showing that

toluene (low 8p) extracted more non-polar lipids than acetone and ethyl acetate. The

selection of acetone was based on the observation that phospholipids were sparingly
soluble in this solvent, and also because it gave a similar yield of SQDG compared to the
yield of extraction using chloroform.

Figure 3.3 illustrates the adsorption scheme giving the best separation of SQDG
using Florisil. The elution started with 20 ml of hexane, followed by 22 ml of acetone-
methanol (10:1 v/v) and 20 ml of acetone-methanol (2:1 v/v). As expected, the majority
of pigments were removed by hexane and the third fraction eluted from Florisil using

acetone-methanol (2:1 v/v), was almost free of non-polar lipids.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Wash F lorisil with (I) Elute with hexane
methanol
. . . Adsorb the dry (2) Elute with acetone-
Pack FlorIsII Into column ; ------------ P extract ""0 methanol (10.1 v/v)
Florisil '

 

 

 

 

 

Wash column with
methanol, then followed by (3)
acetone ‘

Elute with acetone-
methanol (2:1 v/v)

 

 

 

 

 

 

 

 

 

 

 

 

Replace acetone with
hexane

..........

 

 

 

 

 

/ Extract is free /
of non-polar /

lipids

 

Figure 3.3 Adsorption Using Florisil in Purification Studies

36

3.3.2 — Ion-exchange Chromatography Using DEAE-cellulose

Experimental procedure — 10 g of DEAE-cellulose was washed and packed into a
custom-made column (0.8 cm-ID x 40 cm) using 50 ml of methanol and 50 ml of acetone
as described for the Florisil column. Following acetone, 100 ml of glacial acetic acid was
used to wash and exchange the DEAE-cellulose to the acetate form. The column was
allowed to stand overnight before the glacial acetic acid was removed using 100 ml of
methanol and followed by 100 ml of acetone.

The extract, relatively free of non-polar lipids from F lorisil was loaded to the
pretreated DEAE column. Similar to the procedure used in Florisil adsorption, studies

were performed using different solvents to remove non-SQDG from the extract.

 

 

 

 

 

 

 

 

 

 

 

W ‘1 ' _
85h DEAE WI h (1) Elute with acetone- Majority of dyes was ‘.
methanol .___+ .-
methanol (2:1 v/v) removed
(2)

 

 

Pack DEAE into " """"""" ~._
column Dyes and lipids (less
Elute with acetone ————P polar than MGDG)

 

 

 

 

 

 

‘ were removed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Exchange the (3)
Wash column with . 7 extract (free or __ _. .
acetone i non polar lipids) .. ' 2
i with DEAE . . . Majority of MGDG .
Elute With gIZClaI acetlc ‘ and some DGDG
ac1 _ [were removed_l_,.--"
Wash column with
. . . (4)
glaCIal acetic ac1d
Elute with methanol ———$ more MGDG and

 

 

 

. :DGDG were removed .54
Replace glaCIal .. . .

acetic acid with :
methanol ( 5 )

 

 

 

 

 

 

 

 

 
 
 

   

a Elute with acetone-
Replace methanol i methanol-cone. NH4OH

with acetone ....... ~ (4: I 20.2 v/v/v)

 

SQDG and trace of )

Figure 3.4 Ion-Exchange Using DEAE in Purification Studies

 

 

 

 

 

 

37

 

Results and Discussion - The best scheme and result from ion-exchange DEAE-
cellulose column is shown in Figure 3.4. For 5 g basis of alfalfa, the first elution used
135 ml of acetone-methanol (2:1 v/v), each elution in the next three steps used 100 ml of
the selected solvent, and the last step used 52 ml of acetone-methanol-ammonium
hydroxide mixture.

The final effluent from the DEAE column contained mainly SQDG, but
ammonium acetate and a small amount of DGDG were also present. The ammonium
acetate can be removed from SQDG using the O’Brien [49] dialysis method. In this
study, it was eliminated with liquid-liquid partioning using hexane-methanol-water (623:1

v/v/v).

3.4 - A proposed Extraction and Purification Scheme

Results from the above studies verified that extraction of SQDG from alfalfa can
be done using non-chlorinated solvents. The degradation of extracts can be controlled
using hot isopropanol. Non-polar lipids can be removed from the extract using
adsorption with activated carbon or phase partitioning with hexane and K2HPO4.
Phospholipids can be adsorbed onto Florisil or eliminated simultaneously with proteins

and water in liquid-liquid partitioning with hexane-methanol-water mixture.

Adsorptions with Florisil or activated carbon require a complete removal of water.
To avoid the energy costs involved in the evaporation of water, liquid-liquid partitioning
should be used where possible. Therefore, an efficient extraction and purification of
SQDG can be done as described here:

I. Engme deactivation: Prior to grinding, alfalfa is allowed to contact and

partially extracted with preheated isopropanol at 50-55 °C for 5 min.

38

II. Extraction: The same extraction procedure described in section 3.2.4 should
be applied except that the solvent system described in that section is replaced with
hexane-methanol (3:5 v/v). No salt is needed, and the extract is filtered out to remove
solids and residues.

[11. Evaporation: Evaporating as described in section 3.2.4 to remove about 90
vol. % of liquid from the extract and obtain the wet-extract. The evaporation is complete
if the level of liquid in the evaporator remains the same for five minutes.

IV. Proteins and phospholipids removal: The wet-extract is blended with
hexane-methanol-water (6:3:1 v/v/v). First, only methanol and hexane are used assuming
all liquid remaining in the wet-extract is water. Enough hexane-methanol-water mixture
is added to completely dissolve and separate the wet-extract into two liquid phases.
Proteins and most of phospholipids are discarded from the bottom phase. The rich
hexane phase containing SQDG is collected and evaporated to dryness.

V. Exchange SQDG with DEAE-cellglose: The DEAE-cellulose is prepared as
described in section 3.3.2. The dry-extract from step IV is re-dissolved in acetone (2:1
v/v) and loaded to the DEAE column. Elution starts with acetone-methanol (1:1 v/v) to
remove chlorophyll, pigments and neutral lipids. Then solvent is changed to isopropanol-
acetone-methanol (3:2:5 v/v/v) to elute polar-non SQDG lipids.

VI. Recovery of SODG: SQDG retained in the column is eluted using
isopropanol-methanol-ammonium hydroxide (5:5:1 v/v/v). The trace of PI and
ammonium acetate is removed from SQDG using hexane-methanol-water (6:3:1 v/v/v).

Results and strategies of the proposed method are shown in Figures 3.5 and 3.6.

39

 

Alfalfa E r i n n rifl inP r

 

i 1. Enzyme deactivation
using 2-propanol at 50-55 °C

1. Enzyme deactivation . .
2. PartialextractIon

 

 

 

 

 

3. Grinding process
4. Complete extraction
using hexane-methanol (3:5 v/v)

5. Filtration to remove the
inorganic substances

s————- 11. Extraction

Solid Waste

 

 

 

 

 

III. Evaporation 6. Complete evaporation

 

 

 

 

 

7. Liquid-liquid partitioning

4————— IV. S DG fra t' t' using hexane-methanol-water
Q C Iona Ion (6:3:1 v/v/v)

. Proteins & phospholipid removal

 

 

 

Proteins & 8
phospholipids

9. Exchange SQDG with DEAE
10. Elution of chlorophyll and neutral lipids

 

 

 

 

 

V. Exchange SQDG using acetone-methanol
, . With DEAE ( 1:1 v/v)
A“ leIds except i l Elution of olar non sulfoli id
SQDG " p p

using 2-propanol-acetone-methanol
(3:2:5 v/v/v)

12. Elution of SQDG
using 2-propanol-methanol- ammonium
hydroxide 0.1 M in methanol
(5:5:1 v/v/v)
l3. Polishing the isolated SQDG
i remove the trace of phosphatidylinositol
using hexane-methanol-water

SQDG (6:3: I v/v/v)

 

 

VI. Recovery of SQDG

 

 

 

Figure 3.5 A Proposed Extraction and Purification Scheme

40

Plate 11Activation at 120 C for 1hr

No(NH4pSO4

MGDGV. _.
PG
DGDG 2, ..
SQDG
m
PE
PC

A B c o E F G H A

Figure 3.6 Result of Extraction and Purification of SQDG

A - Crude extract obtained by the published method using chlorinated solvents [11].

B — Crude extract obtained in this work prior to step IV described in the proposed

scheme.

C, D — Afler step IV to remove proteins and phospholipids. Path C is after one
partitioning, path D is after two partionings.

E, F - Samples from step 11, prior to step 12.
G - Final purified product after step 13.

H — Combination of by-products demonstrating no loss of SQDG.

41

PART 2:

PROPERTIES OF PLANT-DERIVED CHEMICALS

42

Chapter 4 — Phase Equilibria and Vapor Pressure

4.1 - Vapor-Liquid Equilibrium Overview

Phase equilibria, particularly the vapor-liquid equilibrium (VLE) are the
foundation for a variety of separation methods used in the chemical and petrochemical
industries. Separations including simple distillation, azeotropic distillation, reactive
distillation, and flash operation cannot be designed and operated efficiently without
knowing VLE of the related components in the mixture.

Since the 19703, many international projects aimed to provide good
thermodynamic properties including VLE of pure chemicals, mixtures, and polymers.
These projects were sponsored by the Society of Chemical Engineering and
Biotechnology (DECHEMA) in Germany, Physical Properties Data Service (PPDS) in
the United Kingdom, the Union of Japanese Scientists and Engineers (JUSE) in Japan,
and The Design Institute for Physical Property (DIPPR) in the United States [50].
Thermodynamic data for many generic chemicals have become available, but that still
cannot entirely satisfy the needs of good experimental VLE data for all chemical process
designs. Results from simulations of a distillation depend not only on the thermodynamic

model used, but also on the quality of the VLE data.

4.2 - Vapor-Liquid Equilibrium Apparatuses and Measurements

The VLE measurements can be performed at isothermal (P-x-y: Pressure-liquid
and/or vapor compositions) or isobaric condition (T-x-y: Temperature-liquid ancflor
vapor compositions) depending on the type of VLE apparatus and physical properties of

the studied substances. In general, isothermal VLE measurements are performed at low

43

 

temperature and more suitable for thermally labile and reactive substances. Also, the
possibility of liquid entrainment in the vapor phase is small and there is no concern of
superheating of liquid phase at the isothermal condition. In addition, the isothermal VLE
data are more easily executed and interpreted, because the measurable variables are
directly related to the basic equation of vapor liquid equilibrium: yixi f1 = (1),- yiP and fl is
constant across the composition range.

The VLE apparatus is either static or dynamic, based on how the mixture
equilibrated. If the apparatus provides a circulation to liquid, vapor or both phases, it is
called the dynamic type, otherwise it is static [51]. In a static apparatus, a fixed overall
composition of degassed components is volumetrically or gravimetrically added to a
vessel, and the mixture is agitated using an internal stirrer to accelerate the attainment of
equilibration. In a dynamic apparatus such as the Fischer still, the studied vapor and
liquid phases are disengaged, sampled and recirculated.

The VLE experimental method can be either non-analytical or synthetic. The
method is called non-analytical if the phase compositions are pro-determined from pure
components and mass calculations. In the analytical method, Gas Chromatograph (GC)
and High Pressure Liquid Chromatograph (HPLC) are conventional techniques used to
obtain the phase compositions. Complete details of the apparatus and techniques used in

the VLE measurements are referred to Hala et al. [52] and Abbot [53].

4.3 - VLE of Pure components and Vapor Pressure Predictions
The VLE of pure compound occurs at its vapor pressure. For many years,
scientists and engineers have been seeking reliable methods to estimate the vapor

pressure when experimental data are not available. Basically, two different approaches

44

 

have been used, either through derivation of the Clausius-Clapeyron equation or through
group contribution approaches. The group contribution methods are usually based on the
UNIFAC groups, and the methods of predicting vapor pressure using the Clausius-
Clapeyron equation commonly require the critical properties, the heat of vaporization
and/or the vapor pressure at some reference temperature.

Numerous equations and correlations for the estimation of vapor pressure are
reported in the literature [54]. Most of the estimation and correlation methods are only
accurate for specific classes of compounds in small ranges of temperature [55-57]. For
example, the correlation that has been reported by Chiou and Freed [55] is only valid for
aromatic hydrocarbons, organo halogens, aliphatic alcohols, aliphatic acids and
chlorinated phenols at 25 °C. The following equations in the next sections have been
reported to be the most suitable for use in the largest range of chemicals and vapor

pressures.

4.3.1 - Estimation Using the Clausius-Clapeyron Equation
In most cases, the basic equations are derived from integration of the Clausius-

Clapeyron equation as follows:

dlanp_ 4111:
C” -AZ RT2

 

(4.1)

Different expressions of vapor pressure (Pop) can be obtained from Eqn. 4.1,

depending on the consideration of heat vaporization (AH.) and compressibility (AZ).

4.3.1.1 — Assuming that AHv/AZ is Constant

Re-arranging Eqn. 4.1:

45

 

 

_ AH, 9’2
dln Pvp _RAZ [T2] (42)

By taking the integral, Eqn 4.2 can be presented in the simplest form as follows:

It

lanpzAl T

(4.3)

Using the Antoine equation in the general form that permits representation of
curvature in lnP vs. lfl":

32

 

 

 

In P,,, = A2 — T _ C2 (44)
At the normal boiling point, T = T b, Pvp= 1 atm, or lanp= 0, then,
82 AH b 2

A = and B = ___V___ T _ C

2 Tb—Cz 2 A21) Rsz( b 2)
Eqn 4.4 in a complete form is:

2
AH T —C
AZ), RTb L(Tb ”C2) (T-Czi

where P,p is in atm, and AZ, is assumed to be 0.97 [58]. The constant C2 is estimated,
using the Thomson’s rule [59] as shown below:
C2 =—l8+0.19Tb (4.6)
The heat of vaporization at boiling point AHvb is evaluated by Fishtine [60], using

the modified Kistiakovskii [61] equation as follows:

A2” = V, = KF(8.75+ Rln Tb) (4.7)

 

where K; = 1+ 211/100, and ,u is the dipole moment of molecule. Most of dipole
moments fall in the range of 0 to 5 Debye units. The methods of calculating ,u are

referred to Nelken and Birkett [62].

46

 

4.3.1.2 - Using AHleZ Temperature Dependence

The temperature dependence of AH , is considered using a modification of the

Watson [63] correlation, as described below:

(4.8)

1—T/T ”’
AHV=AHvb( C)

l-T b / T c
Eqn 4.8 contains the critical temperature (Tc). It has been reported that the ratio
T/Tb varies from 1.3 to 1.7 for most organic compounds. Therefore, an estimation of

Tc z 1.5T], is used, and the derivation of Clausius Clapeyron equation yields:

— 2m(3— 2(r/Tb))’""1n(7/T,,) (4.9)

 

ln z AH,,, _(3-2(T/T,,))’"
VP AZbRTb (T/Tb)

where m = 0.19 is recommended for all liquids.

4.3.1.3 — The Korsten Correlation

Korsten discovered that vapor pressures of homologous compounds converge at a
common point (01), and logarithm of vapor pressure (lnP) of any chemicals is linear to
(l/T)"3o [64]. For example, the common (01) for all the hydrocarbons is (Ta =1994.49 °K,
Pa =1867.68 bar). The temperature Ta is the upper limit temperature for vapor pressure,
which cannot be exceeded by any component [64].

Substituting (Pa T a) into Eqn (4.9) yields:

1 1
lnP=lnPa+B[—————-] (4.10)
T130 T0130

The constant B, SIOpe of vapor curve in Eqn 4.10, is a characteristic of each
component. It is a linear function to molecular weight M065 within each homologous

series. The calculation of B and its applications are described by Korsten [64].

47

4.3.1.4 - Summary of Methods Using the Clausius-Clayperon Equation

The average and maximum errors using the Antoine equation and the modified
Watson correlation are sumrriarized in Table 4-1, as shown below. These equations
require the values of heat of vaporization and normal boiling point temperature for
estimating vapor pressure. The analytical form of vapor pressure is more complex for the

temperature dependence of AHV, but the estimation in the range of 10 — 760 mmHg is as

accurate as that obtained from a method considering AHv/AZ as the constant. Both

methods predict pressure of gases and liquids limited to 10-760 mmHg with ~ 3% error.

Table 4-1 Errors in Estimating P“t Using the Clausius-Clayperon Equation [54]

 

 

 

Pressure AHVIAZ is Constant AHVIAZ is Temp. Dependent
Average Maximum Average Maximum
(mmHg)
Error (%) Error (%) Error (%) Error (%)
10 - 760 2.7 6.6 2.5 7.1
10'3 - 10 86 100 39 50
10'7 - 10'3 Method is not available 47 200

 

Estimations at low pressures are inaccurate with very large average errors. The
average error in estimating vapor pressure using the Clausius-Clapeyron equation can
exceed 35% when the boiling point temperature and heat of vaporization are known. If
boiling point temperature and heat of vaporization are estimated, the deviation average
can be higher than 80%. Error is expected to increase with complexity of molecular
structures and low vapor pressures. Reid et al. state that none of the vapor equations in
the literature are suitable for estimating the vapor pressure below 10 mmHg within a 10%

deviation from the experimental data [56].

48

 

Predicting vapor pressure using the Korsten equation requires reliable vapor data
of other members of the same homologous series to determine the common point, but
data of the homologous series are not always available. However, the linear correlation
of molecular weight M065 and vapor pressure, described by Korsten, can be used to
improve the prediction of vapor pressure from the other methods, and Eqn 4.10 can also

be used to validate the prediction.

4.3.2 - Highlights of Prediction Methods using Group Contribution

In contrast to molecular structure methods, there is very little literature available
for the group contribution methods, which have been commonly based on UNIFAC
groups. A typical example of this method is the work of Fredenslund et al. [65, 66].

Predicting vapor pressure based on the group contribution has proved to be a very
difficult challenge [67], usually dependent on first estimating critical properties. The
deviation has been noted to be very large in predicting vapor pressures using group
contribution methods, and it is expected to increase with molecular complexity. Bureau
et al. [68] performed the evaluation for seven esters and found deviations averaging near
80% using UNIFAC groups. Similarly, Asher et al. [69] reported deviations averaging
near 300% for 76 multi-functional oxygen-containing organic compounds at temperatures
of 290-320 °K. The large deviations in the oxygen-containing compounds indicate the
difficult challenges in predicting vapor pressure of esters and acids, which usually form
the intra and/or intermolecular hydrogen bonds between molecules.

The indirect approach using a group contribution such as the Constantinous-Gani
[70] or Joback [71] with the Clausius-Clapeyron equation also is reported with high

deviations in estimating vapor pressure. For example, Ashler et al. [69] reported

49

deviations averaging near 900% using Joback method and the modified Lee-Kesler
equation, for the same data set used in UNIF AC validation.

The ASPEN simulation software usually estimates parameters for the extended
Antoine vapor pressure equation through a group contribution, developed by Juan-Carlos
Mani of ASPEN [72]. The Mani method uses Riedel’s equation [73] to estimate normal
boiling point and critical temperatures while Gani’s method is used to obtain critical
pressures. The Gani [70] method has been reported to be applicable up to the critical
temperature, and more accurate than Joback, Lydersen and Ambrose methods [74]. The
Mani method accurately correlates vapor pressure curves if some experimental vapor
pressure data values are available [75]. However, ASPEN predicted that pressure curves
of lactic and di-lactic acids intersect below their critical temperatures; between 493 °K
and 503 °K (Table 2.) This vapor pressures contradict the Korsten observation that the
common point temperature cannot be exceeded by any component vapor pressure in a

homologous series at any pressure [64].

Table 4-2 ASPEN Predicted PM of Lactic Acid and dI-Lactic Acid

 

 

 

 

Temp Lactic acid Di-Iactic acid Temp Lactic acid Di-Iactic acid
(°K) (kPa) (kPa) (°K) (kPa) (kPa)
273 1.51E-05 0.0002 503 153 151
323 0.0073 0.0256 513 204 197
373 0.4350 0.8157 540 420 381
423 7.419 9.801 570 848 727
473 57.65 61.98 600 1581 1289
483 81.19 84.63 660* 4648* 3481
493 112.27 113.74 675* N/A 4369*

 

*estimated critical pressures and temperatures of corresponding components.

50

The advantage of using the contribution method is that it does not require any
experimental data. However, this method in particular is not reliable for multi-functional

oxygen containing compounds due to the large error in estimation.

4.4 - Background on Step Potential Equilibria And Dynamics Simulation

In the near future, a possible breakthrough in predicting thermophysical properties
of fluids may be realized using molecular simulation. This technology should provide
excellent efficiency and accuracy in prediction because molecules are examined at
molecular level and state conditions [76]. However, the application of molecular
simulation has not been greatly recognized in the industry due to the lack of reliable
comparison studies and validation of different methods. To emphasize the capability of
utilizing the molecular simulation in engineering correlation and process models, a
contest was held in September 2004 by Case Scientific and various experts from industry.
Modeling groups around the world were challenged to predict vapor pressure, heat of
vaporization, the Henry’s law constant, and heats of mixing using the molecular
simulation method. Elliott and his co-workers from University of Akron won the first
prize for prediction of vapor pressure and heat vaporization. The judges ruled that the
Elliott model, SPEAD, is 50 times faster than more conventional molecular simulations
[77].

SPEAD is acronym of the Step Potential Equilibria and Dynamics simulation.
SPEAD is based on the discontinuous molecular dynamics (DMD) simulation and the
thermodynamic perturbation theory (TPT). It is in development by Elliott et al. and is
being implemented by ChemStations, Inc. as a physical properties standard model in

chemical process simulation [78, 79].

51

4.4.1 — SPEAD and Discontinuous Molecular Dynamics Algorithm
In the discontinuous molecular dynamics (DMD) simulation, step potential refers
to the description of molecular interaction energies by a series of discrete steps. The

simplest form of a step potential is square-well potential, described below:

+oo, r<o
uij = —e, o< r<Ao (4-11)
0 r>lo

where up is potential function, 0 is collision diameter, a is well depth, and 71 is the
potential well width.

The square-well potential model has been one of the most commonly considered
for computer simulation and statistical mechanical methods, because of its mathematical
simplicity [80]. However, research has shown that the simple square-well (one-step)
potential model offers limited capacity to characterize the experimental data, which
measured physical properties; and the same square-well potential model cannot be
transferred to the homologous molecules. On the other hand, the Lennard-Jones [80]
potential is more favorable for transferability, and a discrete Lennard-Jones potential can
provide properties that are similar to the continuous potential for spheres [81].

To improve the transferability that the one-step square-well potential model
cannot provide, the SPEAD method proposes a multi-step united-atom potential for the
attractive potentials. This approach is based on the hypothesis that a more transferable
discontinuous potential model could be obtained by adding steps of diminishing depth.
The attractive potential is divided into an infinite series of wells, with each well width
small enough so the energy of each step can be treated as a constant. The SPEAD

currently has four step-wells, separated at r/o = 1.2, 1.5, 1.8, 2.0, shown in Figure 4.1.

52

SPEAD estimates vapor pressures of hydrocarbons including aromatic
hydrocarbons, the low molecular ethers and alcohols, and simple esters with error of less
than 10% of the experimental values [78]. In these evaluations, the reduced temperatures
are used in the range of 0.45 to 1.0. Results show that the DMD method coupling with
the application of TPT theory used in SPEAD, provided an accuracy superior to the
published studies using Lennard-J ones model for characterization of vapor pressure. The
most accurate prediction of vapor pressure using the transferable Lennard-Jones
alternative model has ~15% error of the experimental data, reported by Fuchs et al.
However, the Lennard-Jones study is limited in the component types and temperature
ranges [82, 83].

SPEAD is a functional group approach applied to molecular simulation.
Molecules are broken into a number of interaction sites for computing the dynamics in
SPEAD simulation. For example, n—pentane is broken into 2CH3 sites and 3CH2 sites.
Each of these united-atom sites is described by a spherical multi-step potential. The
interaction sites together form a bonded molecule. All sites are allowed to freely vibrate
within their bond wells, and multiple bond wells are used to constrain bond angles. For
example, n—pentane is formed by tethering the sites together with wells centered 0.154
nm for adjacent sites. Additional pseudo-bond wells are centered at 0.265 nm for sites
two bonds away, confining the C-C-C bond angle to ~110 degrees. To simulate cis-
butene, the additional pseudo well is added between the sites that are three bonds distant
[78]. Simulations are conducted using infinite wells to represent bonds and pseudo-
bonds, and hard cores to represent collisions. Each interaction behaves like independent

hard sphere between collisions and changes velocities instantly at collision time, and the

53

energy of interaction at each distance is described using the potential functions. The
NVE simulations (number of molecules, volume, and energy are constant) and Newton’s
law of motion is applied to molecular interactions.

Following the simulation, the attractive wells are superimposed using perturbation
theory described in section 4.4.2. Three parameters are assigned for each interaction site
type: the diameter, the depth of the inner well, and the depth of the outer well. The two

intermediate wells are interpolated from the inner and the outer wells [84].

 

 

 

 

 

 

 

 

 

 

 

 

1
0.5 1
0
4"
3
-0.5 4
-1 i
- - - - Lennard-Jones Potential
Step Potential
-1.5 I I
0.0 1.0 2-0
r/o

Figure 4.1 Illustration of the Multi-step Potential

4.4.2 - Thermodynamics Perturbation Theory
The thermodynamic perturbation theory is used in SPEAD to simultaneously

compute thermodynamics and transport properties of fluid. Mathematically, perturbation

54

method involves series expansions which are called asymptotic expansions in terms of a
small parameter [85]. This method is found to be very useful in solving the initial and
boundary problems when the analytical solution cannot be obtained. In general, an
asymptotic expansion has the following form:

y(x;£) = y0 (x) + E)», (x) + 8 2y2 (x) + + £"yn (x) + O”+1 (8) (4.12)
where 8 is a small parameter, and n is the order of the expansion. The well depth is very
small in the multi-steps attractive potential; therefore, the free Helmholtz energy can be
expressed as an asymptotic expansion of well depth or pair potential energy. This
becomes the key to the perturbation theory used in SPEAD. The application of TPT in
SPEAD is similar to the work of Baker-Henderson [86]. Basically, it considers that
strong repulsive forces play the primary roles in determining fluid structure. The
properties of the fluid, therefore, can be calculated by starting with the pure repulsive part
(the reference system) and later incorporating the attractive part as a perturbation [87].

Applying the second order of Eqn 4.12 to the molar Helmholtz free energy using

a multi-well potential yields:

A2

 

 

 

(A-At' ) A0 —A1 A
R; W = RT g +F+F+O3WIJW (4.13)
’. ’. W N,- m
where A1 =21212A’n158 1 >0 "1 (4.14)
2:23222I222:( (Niijlkn>0’<Nl-j >0<len>0) ul-jmulkn (415)
2 =- 2 .
2NkB

In Eqns 4.13 - 4.15, N is the number of particles, and k3 is Boltzmann’s constant =

1.38><10'23 1102 kg 8'2 K", A 0 is the Helmholtz energy of the repulsive (reference) fluid. Ag

55

is the Helmholtz energy of the ideal gas. A 1 and A2 represent the first and second-order of
perturbation terms, m means the m’h well, <N,,-,,,> is the ensemble average of the
interaction site pairs of sites of types i around sites of type j inside the m’h well, I

designates the number of site types, w is the number of wells, thm is obtained from the

reference fluid simulation, and ( )0 denotes an ensemble average of the reference fluid.

For a two-site molecule with four wells per site, there are 16 terms in A1 and 256 terms in
A2. The interaction between two sites or groups of type i and j in well m which is the

geometric mean of the group potentials um and um, defined as follows:

uijm 2 “11m XuJ-jm (4.16)

The pair group potential energy is the asymptotic parameter, and temperature is
specified in Eqn 4.13. Because the reference simulation has no attractive potential
energy, the results of the simulation are independent of temperatures; any T can be

selected to run the simulation.

4.4.3 - Vapor-Liquid Equilibrium using SPEAD
The thermodynamic correlation between the molar Helmholtz energy (A) and
compressibility factor (Z) at constant temperature and volume (in NVE simulation) is

described as follows:

(4.17)

(A _ A'g )Ty __ 32(2— l)dp
RT _ 0 p

where R is the gas constant, V and p are respective molar volrune and density of fluid at
pressure (P) and temperature (T).

By taking the derivative then applying the fundamental theorem of calculus [88],

56

Eqn 4.17 becomes:
_ L d _ .
2-1+RT[d—p(.4 A,g)T,V] (4.18)

SPEAD uses packing fraction (n) [89], which is defined as: n = W where

W

pm is the mass density, Vw is the mass volume, and M,,.. is the molecular weight. It should

be noted that 11— =-Q’-"—= _p_. Applying thermodynamic perturbation theory to the

dn dpm dp

Helmholtz free energy in Eqn 4.18 as it is described in Eqn 4.13 yields:

_ “'(AO—Aig) 611411 61142 1 1 1
Z—1+n( dn ]+ndn(% )+nd—n(— 2): Z°+Z'(T)+ZZ(F) (4.19)

where Z0=I+n [ff—LC!" 457)), Z1=n%% and Z2: “2(73—2— ). The following

77d"

polynomial correlations are used to interpolate simulated results for 20, A 1 and A2 in

 

SPEAD as described below:
2 3
Zozl+am+a2n3+a3n (4.20)
(i -n)
=b177+b2772 +b3n3 +b4n4 (421)
2 3 4
A2 = 6171+ 0271 + 0371 + 0471 (4.22)

 

1+500n4

The expression of 20 follows the format of Carnahan-Starling [90] equation. The
COEbff‘icients of Eqn 4.20 are independent of the well depths and are obtained from
regl‘ession of data provided from hard molecule simulations at 21 different packing
fractions (17). The average numbers for site interactions and intermolecular site distances,

PTOVided from the simulation data, are combined with the well depths to generate A 1 and

57

A2 at each simulation density using Eqns 4.14 and 4.15. To provide the continuous
functions that represent the DMD/TPT calculations, the coefficients in Eqns 4.21 and
4.22 are regressed. The polynomial functions A) and A2 are based on the trends of their
curves, shown in Figure 4.2. Equations 4.20 - 4.22 provide the smoothed functions for
differentiation of the Helmholtz energies at any density subsequently used to generate

PVT information.

 

 

      

    

 

 

 

 

 

 

 

 

 

 

0 1 . 0 -
-1 .1
-0.05 4 ‘
-2 - -
-3 -0.1 - \
.. -4‘ ._
~ N .. a.--
‘1: V —O.15 ~ , 1——————-~—- ‘I
-5 1 .--____..._...
-6 J O Ethane ‘ -0‘2 .1 C n-8utanc
U rI-Butane i A n—Hexane
-7 ‘ [An-Hexane -O.25 _ O n-Octane
-8 ~ 90895325.-
'9 = r I . -03 - _ . . . r
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6
’7 77

Figure 4.2 Trends of A, and A2 and their Fitted Polynomials Function [84]

Phase equilibrium criteria — At phase equilibrium of any temperature (T) and

pressure (P), the following constraints must be satisfied for pure fluids:
Tsat =TV =TL Psat ____ PV =PL Gsat =GV =GL (423)

where L and V respectively denote for liquid and vapor. The vapor pressure (PW) is

determined from:
Psat z zLRTsatpL = ZVRTsatpV (4.24)

The Helmholtz energy (A) and the Gibbs free energy (G) are related as follows:

58

(G—G,g)T,P (A- A, TV

RT = RT +Z—1—ln(Z) (4-25)

 

Combining with Eqn. 4.17, Eqn 4.25 becomes:

HELL—Zn —1)dn+Z l-nZL_ "fly/’1)“

+ 2” - ln 2’” (4.26)
0 77

Algorithm in Calculating Pm — Vapor pressure is calculated by the following
algorithm: (1) guess PM"; (2) use Eqn. 4.24 to calculate 17" and 77V; (3) use 27L and 17V and
Eqns. 4.19 - 4.22 to evaluate each side of the Eqn. 4.26; (4) return to step 1 with a new
guess of P"” until the equality of Eqn 4.26 is obtained.

To perform parameter optimization, additional adjustments of the well parameters
are included to change the values of A1 and A2 that are used in the vapor pressure

calculation, and the PW" calculations are performed repeatly to optimize the square-well

potentials for individual sites.

4.5 - Objective and Scope of Research

Research on esterification to produce plant-derived esters and bio-diesel products
using reactive distillation at Michigan State University has gained reputation and has
been recognized with patents [91]. At present, the esters of interest are triethyl citrate,
diethyl succinate, ethyl lactate. For biodiesel, acetals of glycerol are of interest. All of
these oxygen-bearing compounds are relatively complex and have low vapor pressures;
therefore, thermodynamic properties such as binary vapor-liquid equilibrium (VLE) data
of components involved in the esterification are either very limited or not accessible in
the existing literature. To develop accurate process simulation designs for reactive

distillation, MSU must have the reliable phase equilibrium data.

59

This part of the dissertation focuses on providing substantially reliable
thermodynamic properties for economical industrial process designs of reactive
distillation to produce the above plant-derived esters and bio-diesel products. The
measured and predicted VLE of systems involved in the esterification are presented in
chapters 5 and 6, and the SPEAD predicted vapor pressures of components which are not

available for direct measurements are in chapter 7.

6O

Chapter 5 — Lactic Acid Oligomers Distribution

5.1 - Overview
Lactic acid (2-hydroxy propanoic acid) contains both hydroxy and carboxylic

functional groups. Lactic acid molecules can react to form oligomers as follows [92]:

OH OH OH 0
OH OH - H20 0
H3C + H3C H3C OH
0 o 0 CH3

 

 

 

Lactic acid (LAl) Lactoyllactic acid (LA?)
OH OH O OH O
OH + 0 -H20 0
H3C H3C "-2 OH H3C "-1 OH
0 0 CH3 0 CH3
Polylactic acid (LAM) Polylactic acid (LAD)

As described in literature [92-95], this study verified that dilute lactic acid of less
than 20 wt.% contains only lactic acid monomer (LA 1), but oligomers exist in
equilibrium in concentrated aqueous lactic acid solutions. If pure crystalline lactic acid
is stored at room temperature, it spontaneously initiates intermolecular reactions to form
water and esters, and the equilibrium mixture is obtained after sufficient time, containing
6 wt.% of water, 47 wt.% of lactic acid monomer (LA1) and 47 wt.% of polylactic acids
of a mean degree of polymerization 2.8. Heating strongly accelerates the reaction

without affecting equilibria [96, 97].

Esterification with ethanol producing ethyl lactate requires concentrated lactic
acid solution to minimize the amount of water. At the temperature of the reactive

distillation, lactic acid oligomers (LAz, LA3,. . ., LA") and ethyl lactate oligomers (EZLA,

61

E3LA,. . ., EnLA) will co-exist with lactic acid monomer, ethanol, ethyl lactate, and water.

 

OH O OH
O -H O Q C H
H3C OH + HO—Csz -—3- H c 0/ 2 5
”'1 3 n-1
CH3 CH3
Poly lactic acid (LAn) Poly ethyl lactate (EnLA)

To characterize the lactic acid oligomers involved in the esterification .of ethyl
lactate, a thermodynamic model has been developed using the chemical theory. The
model accurately predicts oligomers distribution over the full range of lactic acid
concentration. The superficial total lactic acid concentration of any lactic acid solution is

reliably determined using HPLC or titratable acidity.

62

5.2 — The Lactic Acid Oligomers Distribution Model — A Reprint of the

Paper “Oligomer Distribution in Concentrated Lactic Acid Solutions”

Following is a copy of the paper entitled Oligomer Distribution in Concentrated
Lactic Acid Solutions by D. T. Vu, A K. Kolah, N. S. Asthana, L. Peereboom, C.T. Lira
and D. J. Miller, published in Phase Fluid Equilibria, 236 (2005) 125-135. The paper is
reformatted to enlarge the text in figures and tables so that all parts of the dissertation are

readable from microfilm.

63

Available online at www.5ciencedirect.com

       

24,.“ “5'3" ”‘1,
‘1} s; :51 h; .355“ if:
7‘ r‘; .137; 1,:7'9'.“ _
$235.““ ocuncn@omlc1-
$10323"? ;_. .
fl' 0"
., 77’
El SEVIER Flurd Phase Equrhbna 236 (2005) 125-135

wvm.clscvierxomr’locatcffluid

Oligomer distribution in concentrated lactic acid solutions

Dung T. Vu, Aspi K. Kolah, Navinchandra S. Asthana, Lars Peereboom,
Carl T. Lira*, Dennis J. Miller

Chemical Engineering and Materials Science. Michigan State University 2527 Engineering Building,
East Lansing (USA). 48824-1226

Received 21 April 2005; received in revised form 1 June 2005; accepted 3 June 2005
Available online 10 August 2005

Abstract

Lactic acid (2-hydroxypropanoic acid) is a significant platform chemical for the biorenewable
economy. Concentrated aqueous solutions of lactic acid (>30 wt.%) contain a distribution of
oligomers that arise via intermolecular esterification. As a result, the titratable acidity changes
non-linearly with acid concentration. In this work, the oligomer distribution of lactic acid is
characterized using GC, GC/MS, and HPLC to extend existing literature data, and titratable
acidity is measured via titration with NaOH. A thermodynamic model with a single parameter is
proposed that accurately represents oligomer distribution and titratable acidity over the fill range
of lactic acid concentrations.

© 2005 Elsevier B.V. All rights reserved.

Keywords: Lactic acid; Oligomerization; Chemical theory; Esterification; Alpha-hydroxy acid;
2-Hydroxypropionic acid

* Corresponding author. Tel.: +1 517 355 9731; fax: +1 517 432 1105.
E-mail address." lira@egr.msu.edu (C.T. Lira).

64

1. Introduction

In recent years, there is increasing emphasis on using biorenewable materials as substitutes for
petroleum-based feedstocks. This paradigm shift is attributable to rising crude oil prices and the
increasing desire to reduce dependence on petroleum. A major building block for the
biorenewable economy is lactic acid (2-hydroxypropionic acid), an (It-hydroxy acid containing
both a hydroxyl and carboxylic acid functional group. For an excellent review on lactic acid the
reader is referred to Holten [l]. Lactic acid was first isolated by the Swedish scientist Scheele in
1780 [2], and first produced commercially in 1881 [3]. Applications for lactic acid are found in
the food (additive and preservative), pharmaceutical, cosmetic, textile, and leather industries.
Lactic acid can be formed either via fermentation of carbohydrate monomers or via a chemical
route, but since about 1990 only the fermentation route is practiced commercially. The recent
completion of the NatureWorks lactic acid facility for poly-lactic acid production, with an annual
capacity of 140,000 metric tonnes of polylactic acid (PLA) [4], has greatly enhanced the stature
of lactic acid as a key biorenewable platform.

Polylactic acid [5] is a versatile thermoplastic polymer that has useful mechanical properties
including high strength and high modulus. Applications of PLA include household commodity
products, polymers used in food contact, biomedical materials like surgical sutures, absorbable
bone plates for internal bone fixation, artificial skin, tissue scaffolds, and controlled release drugs.
PLA is one of the few polymers whose structure and properties can be modified by polymerizing
a controlled composition of the L - and D-isomers to give high molecular weight amorphous or
crystalline polymers. PLA has a degradation time of 6 months to 2 years in the environment.
For more details on PLA the reader is referred to Garlotta [6].

Esters of lactic acid, formed via combination with alcohols like methanol and ethanol, are finding
increased use as environmentally benign solvents. Lactic acid esters are biodegradable, non-
toxic, and have excellent solvent proper-ties, which make them attractive candidates to replace
halogenated solvents for a wide spectrum of uses. Esterification of lactic acid with alcohol can
also be used as a highly efficient method for purification of lactic acid from fermentation broths,
especially when lactic acid is desired in concentrated solutions.

It has been observed experimentally that dilute (<20 wt.%) lactic acid solutions contain only
lactic acid monomer (LA) [7], an observation that has been verified in this paper. However,
many processes involving lactic acid, including polymerization and esterification, require
concentrated lactic acid solutions, and lactic acid in these solutions undergoes intermolecular self-
esterification to form higher oligomers. This oligomerization occurs to an increasing degree at
high acid concentration, low water concentration, and high temperature.

In oligomerization, two molecules of lactic acid first react to form a linear dimer, commonly
called lactoyllactic acid (LAZ), along with a mole of water.

HO 0 H00

\ II \ ll
2 CH3CHCOH CH3CHC—C‘) (I? + H20
CH3CHCOH

 

 

(1)
Lactic Acid (LAI) Lactoyllactic acid (LAz)

Lactic acid also forms a cyclic dimer noted as lactide, but this compound is known to be unstable
in water [1] and thus is not a concern in this work. Lactoyllactic acid (LAZ) can further esterify

65

with LA1 to form the trimer lactoyl-lactoyllactic acid (LA3); this process can further continue to
give higher chain intermolecular polyesters LA4, LA5 and so on.

 

 

HO 0 Box ii
HO\ C”) CH \CHii o o CH3CHC-(i O
CH3CHCOH + 3 \ H CH3CHC—O o + H20
CH3CHCOH \
CH3CHCOH
(2)

Lactoyl-lactoyllactic acid (LA3)

The inherent tendency of aqueous lactic acid to form intermolecular esters in solution poses a
formidable obstacle in the modeling of its liquid-phase behavior and vapor-liquid phase
equilibria. For design of reaction and separation processes involving concentrated lactic acid
solutions, a model to predict thermodynamic properties of these complex chemically reactive
mixtures is an indispensable tool. This paper presents such a model that requires only one
parameter to adequately represent lactic acid solution behavior over the full range of
concentration.

1.1. Definition of concentrations

Experimental work on quantifying concentrations of lactic acid oligomers in aqueous solution has
been previously reported by Montgomery [7], Ueda and Terashima [8], and Watson [9], but the
methods used in reporting these concentrations and the definitions of concentrations are not
always clearly presented. Therefore, we clearly define here the quantities used to describe the
concentration of lactic acid and its oligomers in solution.

1.1.1. Equivalent monomer lactic acid

In the literature, it has been found convenient to express the concentration of lactic acid oligomers
as a percent of equivalent monomer lactic acid on a water free basis. We abbreviate such a
description with the acronym %EMLAj. To illustrate the concept, consider a solution consisting
of 50 mol water, 9.20 mol LA], 0.343 mol LAZ, and 0.0128 mol LA3. Upon hydrolysis of the
oligomers, 9.20 + 2 x 0.343 + 3 X 0.0128 = 9.924 mol lactic acid monomer would be present. The
amount of water present would be 50 - 0.343 - 2 x 0.0128 = 49.63 mol H20. The lactic acid in
the original solution is reported as 9.20 / 9.924 = 92.7% EMLA LA], 2 X 0.343/9.924 = 6.9%
EMLA LA2, and 3 X 0.0128 / 9.924 = 0.38% EMLA LA3. Introducing the molecular weight of
water and oligomers, the solution has a total mass of 50 X 18.02 + 9.20 x 90.08 + 0.343 X 162.14
+ 0.0128 X 234.21 = 1788.3 g.

1.1.2. Superficial weight percent

The superficial weight percent of lactic acid is expressed as the weight of total monomer with the
corresponding water of hydrolysis divided by total solution weight. For the example above, the
superficial wt.% is (9.924 mol LA X 90.08 / 1788.3 = 0.500) 50.0 wt.% lactic acid, and (49.63 X
18.02 / 1788.3 = 0.500) 50.0 wt.% water. When lactic acid is purchased, the concentrations
expressed in wt.% should be interpreted as superficial wt.%. In this manuscript, we explicitly
label such concentrations superficial wt.% to avoid confusion. When solutions are very
concentrated, the superficial concentration of lactic acid can exceed 100 wt.%. The concept of
125 superficial wt.% lactic acid arises from the fact that 100 g of a polymer (C3H402)n upon
hydrolysis gives rise to 100 X 90.08 / 72.06 = 125 g of lactic acid, where 90.08 is the molecular
weight of lactic acid monomer, and 72.06 is the molecular weight of the ester repeat unit in the

66

 

polymer. When an aqueous solution has a lactic acid content exceeding 100 superficial wt.%, the
water of esterification (oligomerization) has been removed from the solution, and the solution is
thus characterized by a negative superficial wt.% of water.

1.1.3. True weight percent

True weight percent utilizes the mass of a particular sample and the total mass of the individual
species within the solution. Using the same example again, the true wt.% values are 46.3 true
wt.% LA, (9.20 x 90.08/ 1788.3 = 0.463), 3.1 true wt.% LA; (0.343 X 162.14/ 1788.3 = 0.031),
0.17 true wt.% LA; (0.0128 x 234.21 / 1788.3 = 0.0017), and 50.4 true wt.% H20 (50 X 18.02 /
1788.3 = 0.504).

2. Experimental

2.1. Chemicals

Analytical grade aqueous lactic acid solutions were used in experiments: 85 superficial wt.% was
purchased from J .T. Baker, Inc. and 50 superficial wt.% was purchased from Purac, Inc. HPLC
grade water was purchased from J .T. Baker, Inc. HPLC grade acetonitrile was purchased from
EMD Chemicals. An aqueous solution of 85 wt.% phosphoric acid was purchased from J. T.
Baker, Inc.

2. 2. Preparation of oligomer solutions

Solutions of lactic acid below 50 superficial wt.% were prepared by adding water to 50
superficial wt.% lactic acid, whereas solutions between 50 superficial wt.% and 85 superficial
wt.% were prepared by mixing the 50% and 85% solutions. After mixing, the solutions were
heated at 80 °C for 1 week to increase the rate of formation of various oligomers of lactic acid.
To concentrate lactic acid above 85 wt.%, water was removed from 85 wt.% lactic acid at 45
mmHg using a vacuum distillation apparatus. At that pressure, the boiling point temperature
started at 30 °C for 90 superficial wt.% solution and rose to 135 °C for solutions of 120 superficial
wt.%. Following evaporation, the solutions were equilibrated by refluxing at 100 °C for 30 h.

2. 3. Analytical methods
The composition of lactic acid and its oligomers in solution was characterized using a
combination of three analytical techniques.

2. 3. 1. Titration

The composition of dilute solutions containing less than 20 superficial wt.% lactic acid contains >
98% EMLA LA] and water [1]. Lactic acid solution containing less than 10 superficial wt.% of
lactic acid contains 99.6% EMLA LA] [1,10], and direct titration with standardized 0.1N NaOH
(Sigma-Aldrich) gave an accurate analysis of LA1 in solution.

For solutions containing more than 20 but less than 85 superficial wt.% lactic acid, the total free
acidity of the solution was determined from titration with standard 0.1N NaOH. 1n solutions
above 85 superficial wt.%, titration with 0.1N NaOH occurred with too little base to accurate
determine the endpoint. More reproducible results were found when using 0.01N NaOH. In
addition, titrating the lactic solution in ice yielded more reproducible results due to decreased
probability of hydrolysis. Ester bonds present in oligomers are susceptible to hydrolysis in the
presence of aqueous NaOH at room temperature. This could lead to inconsistencies in
determination of total acid content by titration; therefore the solution was titrated in ice to
minimize hydrolysis. After titration of free acidity, excess NaOH was added and the solution was
heated to about 80 °C to hydrolyze the oligomers to monomeric sodium lactate. Hydrolysis was

67

carried out for two hours for solutions below 100 superficial wt.% and for four hours for solutions
above 100 superficial wt.%. The quantity of unreacted NaOH was determined by back titration
of the resultant solution with standardized 0.1N H2804 solution (Sigma-Aldrich). For
concentrations where only monomer and dimmer exist, the quantity of LA. in solution was
calculated by the difference between NaOH consumed for neutralization of total acid and the
quantity of NaOH consumed for the hydrolysis of ester linkage present in oligomers [l 1,12].

2. 3. 2. GC analysis and GC/MS analysis

Water concentrations in lactic acid standard solutions were verified using a Varian 3600 gas
chromatograph (GC) equipped with a thermal conductivity detector (TCD). The GC column was
3.25 mm OD x 4m long and was packed with 80/ 100 mesh Porapak-Q. The oven temperature
was held constant at 413 K for 2 min, ramped at 20 °C / min to 493 K, and held at 493 K for 6
min. The injector temperature was maintained at 493 K and the TCD block temperature was held
at 523 K. Helium was used as the carrier gas. HPLC grade acetonitrile was used as an internal
standard. Qualitative analysis of LA. and its higher oligomers LAZ, LA3 LA4, etc. by GC—MS
was carried out on a JEOL AX-505H double-focusing mass spectrometer coupled to a Hewlett-
Packard 5890J gas chromatograph via a heated interface. GC separation employed a J&W DB-23
fused-silica capillary column (30 m length x 0.25 m ID. with a 0.25 pm film coating). Splitless
injection was used. Helium gas flow was maintained at 1 mL/min. The GC temperature program
was initiated at 323 K and was ramped at 10 °C / min to 533 K. MS conditions were as follows:
interface temperature 523 K, ion source temperature 523 K, electron energy 70 eV, and scan
frequency was 1 Hz over the m/z range of 45 - 750. Prior to its injection for analysis by GC—MS,
LA], LAz, LA3, and LA, were derivatized with TMS {Propanoic acid, 2-[(trimethylsilyl)oxy -
trimethyl silyl ester} to enhance their volatility.

2.3.3. HPLC analysis

The concentration of LA] and oligomers in concentrated lactic acid solutions were quantified
using a Hewlett Packard 1090 Liquid Chromatograph equipped with an auto sampler, gradient
flow pump, oven and a Hitachi-L400H UV detector set at 210 nm. Lactic acid samples below 85
superficial wt.% were analyzed using a mobile phase of water + acetonitrile in gradient

concentration at a flow rate 1 mL/min on a Novapak C18 column (3.9mm X 150 mm). Both

water and acetonitrile were acidified using 2 mL of 85% (w/v) phosphoric acid in 1 L of solvent.
The water was analyzed to be pH 1.3. The column oven temperature was maintained at 40 °C.
Beginning with a mobile phase of 100% acidified water, the acetonitrile concentration was
ramped linearly to 60 vol.% from zero to 20 min and then ramped linearly up to 90% from 20 min
to 25 min. The mobile phase composition was maintained constant at 90% to 28 min and then
returned to 100% water.

For analysis of solution concentrations above 85 superficial wt.% lactic acid, the total flow rate
and column temperature were maintained as above, but the gradient was modified. The mobile
phase was ramped linearly from 10% to 100% acetonitrile from 0 to 25 min. Acetonitrile
concentration of mobile phase was brought back to 10% at 35 min.

2.3.3.1. Response factor for LA 1. Dilute solutions of lactic acid (<20 superficial wt.%) contain >
98% EMLA LA]; their concentrations can be accurately determined by titration as described in
Section 2.3.1. To prepare a standard containing only LA], a dilute solution containing 7—8
superficial wt.% total lactic acid in water was prepared and heated for 6 h in presence of
Amberlyst-IS cation exchange resin to facilitate hydrolysis of any LAz or higher oligomers
present. Titration of this solution with 0.1N NaOH showed a value of 7.3 true wt.% LA]. This
solution was used to create HPLC calibration standards for LA] that spanned the range of LA,

68

concentrations (0.1 — 1 true wt. %) used in HPLC analysis. A linear UV response was observed
from the calibration curve obtained by sample dilution. The response factor for LA. obtained
from this calibration was used for quantitative determination of LA. in concentrated lactic acid
solutions.

2.3.3.2. Response factor for LA ;. A 50 superficial wt.% lactic acid solution, containing LA. and
LA;, was titrated/hydrolyzed/back-titrated with standardized 0.1N NaOH solution as described in
Section 2.3.1. By this method the composition of LA. and LA; were quantified as 46 and 3 true
wt.%, respectively. HPLC analysis was performed on the sample and LA. was quantified using
the response factor from calibration described in Section 2.3.3.1. GC analysis of the sample
showed the presence of 51 true wt.% water, and closed the material balance. This standardized
solution was diluted in water to provide a series of calibration standards that spanned the
pertinent range of true wt.% of LA. (0.1 to 1 wt.% by appropriate dilution with water) and LA;.
A linear UV response with concentration was observed for LA; following prompt analysis. The
response factor from this calibration curve for LA; was used for quantitative determination of the
superficial LA; concentration in lactic acid solutions. The ratio of response factors for superficial
wt.% was found to be LA;/ LA. = 1.43 in all HPLC analyses.

2.3.3.3. Response factors for LA 3 and LA... In a solution with approximately 93 superficial wt.%
aqueous lactic acid solution, the linear oligomers LA3 and LA.. are observed in significant
quantities in addition to LA;. HPLC analyses of the solution showed compositions of 58 and 22
true wt.% for LA. and LA;, respectively, with the remaining lactic acid in the form of higher
oligomers. GC analysis of the solution showed the presence of 12 true wt.% water. The presence
of lactic acid oligomers up to LA. was also verified by GC—MS analysis. The assignment of
response factors for higher oligomers was based on the following premises: (1) the difference in
successively higher oligomers of lactic acid is the presence of an additional ester group; (2) the
UV detector response is related to the presence of carbonyl groups in the ester functionality; and
(3) the ratio of LA;/LA. response factors was 1.43. Therefore, the same ratio of response factors
was assigned to each of the successively higher oligomers of lactic acid for superficial wt.%
(LA/LA. = 1.43). Using these response factor ratios for LA; and LA.., the concentrations of LA;
and LA. were determined from HPLC to be 6 and 2 true wt.% respectively. Using these values,
the material balance closed (58 + 22 + 6 + 2 + 12 = 100).

To further test the calibration, a series of dilutions where prepared from a solution that was
determined by titration to be 73.8 superficial wt.% lactic acid. The dilutions spanned the range of
various wt.% of LA., LA;, LA;, and LA.. acids (0.1—1 wt.% by appropriate dilution with water),
and the HPLC analysis showed a linear concentration response. Using the response factors
determined above, the total superficial concentration was determined to be 74%, in excellent
agreement with titration and thus verifying the reliability of the oligomer HPLC response factors.

2.3.3.4. Analysis of higher (>LA4) lactic acid oligomers. High oligomers of lactic acid are
insoluble in water, but they are miscible in acetonitrile. Mixtures of acetonitrile + water have
intermediate solvent strength. To dilute a sample of 115 superficial wt.% lactic acid to an overall
concentration of 2 wt.% in a homogeneous phase, a solution of at least 50 wt.% acetonitrile was
needed. However, this composition was not suitable for injection because HPLC could not
provide reliable resolution between LA. and LA; if more than 20 wt.% acetonitrile was present in
an injected sample containing large quantities of LA. and LA;. The difficulties did not arise
when the quantities of LA. and LA; were small. To provide reliable results, lactic acid solutions
greater than 105 superficial wt.% were analyzed in two fractions. Approximately 0.1 g lactic acid
solution was transferred to a microcentrifuge tube and weighed. Approximately 1mL of water
was added; the solution was shaken, and then centrifuged at 4000 rpm in a desktop

69

microcentrifuge for 4 min. The water phase was carefully removed using a pipette. The water
extraction was repeated four to five times. This water-soluble fraction was weighed and held for
analysis. Next, the water-insoluble high oligomers were recovered in 100% acetonitrile and this
acetonitrile phase was weighed. All steps were done at room temperature. The oligomer contents
in both water and acetonitrile were combined in calculation of superficial wt.% oligomer
distribution in the two fractions, and then combined to calculate the superficial wt.% of the
original sample and % EMLAj. The response factors for the higher oligomers where assumed to
be the same as the values for LA3 and LA... The HPLC results for total lactic acid content
determined by adding the superficial wt.% of the individual oligomers is in good agreement with
the results from titration as shown in Table 1.

3. Mathematical model

We present here a model of infinite oligomer formation using chemical theory. There are a few
examples in the literature of compounds whose phase equilibria properties have been described
with the help of chemical theory or chemical theory along with physical intermolecular forces.
The most strikingly related example is that of formaldehyde in aqueous and/or methanolic
solutions, which reveals extreme deviations from ideality caused mainly by chemical reactions.
Formaldehyde in the presence of water gives methylene glycol and polyoxomethylenes; in the
presence of methanol it gives hemiforrnal and higher hemifonnals [13].

VLE for formaldehyde-containing systems has been described using chemical theory by Kogan
[14], Kogan and Ogorodnikov [15,16], Brandani et al. [17] and Masamoto and Matsuzaki [l8].
Maurer [13] presented for the first time a model in which chemical reactions together with
physical intermolecular forces were used successfully to describe the VLE and enthalpy for
formaldehyde-containing systems containing both reactive and inert components such as trioxane.
Maurer’s model was subsequently extended and tested using new data; for an update on the
model up to 1992 the reader is referred to Hahnenstein et a1. [19]. This approach has also been
used by Brandani et a1. [20—22].

For the system formaldehyde-water, the mole fraction of compounds in the liquid phase is
calculated by modeling the oligomerization as two equilibrium constants—one for methylene
glycol formation from formaldehyde and water and the second for subsequent higher methylene
glycol oligomer formation.

 

 

l xMG _ r VMG ]
K = ‘— —-—— (3)
1 llxw xFA)_ L(7w 7E4)
'l
Kn=[_wl_ Land 25,, (4.
(xn-l xMG)_ Kin-1 mall

These assumptions are reasonable since methylene glycol is a chemically different structure than
formaldehyde, while the higher oligomers of methylene glycol are chemically similar to each
other. The formaldehyde formaldehyde—methanol system is treated in a similar way.

70

6&3 <4 3688280 A.
.a\oom <4 EBBEEOU _.

.v 23m... :0 3.8:. 8... u... use :0303055 05 E 352qu 8 3.23.8 o8 £9398 0.3: 3 moms—.38;

 

 

 

.2... 42.. m2. .3. 8.: 2:.. mm... m3 2% 2.... EN 2:.2 3:...

2:: 32 NS. 2.5 3:: 2.... com. 2.2 3... . :2... N3 8...: 2%..

$3 2:. E... 3.. Es mm... «3 E... 3.: on: 3.8 8.2.. .22.

.....N a... 2:. 2:. .2. a; one 3.... 8.: 2.2 2.2 8.2.. .22..

e2. 2.... 3... mm... ..:. t... 2.... ”2o 8.... :2 3.2 $2.. 8.2..

.. . ... 2:. 2:. 2:. 2... 2.. 3m 2... a2... 8.8 2.2. 3.2.. 22..

wow 2:. 2:. 2:. 2:. mm... 2.... a...“ 2.... .2: ~32 8.3 33

Cum 2:. 2:. 2:. 2:. 2:. 2.... NS 2... 2.3 3.8 $2. 2......

N3 2:. 2.... 2:. 2:. 2:. .22. .8. .23 .33 .33 .82. .22.

a... 2:. 2:. 2:. 2:. 2:. mm... 2:. 23 2%. 8.? 8.... 3....

2... 2:. 2:. 2.... 2:. 2:. 2:. 2... 8.. an: .23. 8.: 8.2

2:. 2:. 2.... 2:. 2:. 2:. 2:. 2:. Q... A? 32. SS 3%

2:. 2:. 2:. 2:. 2:. 2.... 2:. 2:. 3:. a3 .38 .3; .2»?

2:. 2:. 2:. 2:. 2:. 2:. 2:. 2:. 2:. 2.6 3.3 $5. 2......

2:. 2:. 2:. 2:. 2:. 2:. 2:. 2:. 2... as .30 3.3. :3

2:. 2:. 2:. 2:. 2:. 2:. 2:. 2:. 2:. in... 8.3 .2... «mm.

+23 2.3 :3 3.. b3 :3 :3 J... na... :3 .3 one: 8.3:;
.3222... were... one: 2.5 2...; 3.6.2.3 :85

 

count: 3 Eon .floceomzm .89 FEB eaten—Eco can $.32 0.5: .«o @9556

_ 033.

71

3.]. Literature models for lactic acid based on chemical theory

Prior modeling work to determine the distribution of lactic acid oligomers in solutions above 20
wt.% concentration has been performed by Bezzi et al. [23] and reported by Holten [1]. In the
first modeling approach, only the dimers of lactic acid (LAz) were considered. This approach,
however, becomes inaccurate at higher concentrations of lactic acid (>50 wt.%), where
significant oligomerization occurs. In a second modeling approach, polylactic acids were taken
into account, giving a more realistic representation at high concentrations. However, this model
was limited in that solutions were characterized only by concentration of free lactic acid (LA) and
total oligomer species; no distributions of oligomers was generated. This polylactic model works
poorly at low concentrations, and is interpretative rather than predictive in its application

We are unaware of published mathematical models, apart from the ones described above, that
attempt to represent the liquid phase distribution of lactic acid and its oligomers in solution.
Therefore, we propose here a model that is based on chemical theory and incorporates an infinite
series of oligomer components. The model accurately predicts liquid phase compositions of
lactic acid in a method similar to Maurer’s for formaldehyde systems, and represents a clear
advancement of the characterization of concentrated lactic acid solutions. In order to compare the
present model to those in the literature, this work utilizes the terminology used by Montgomery
[7] and Ueda and Terashima [8] as clarified in Section 1.1.

3. 2. Infinite series polymer model

From a thermodynamic standpoint, the formation of oligomeric intermolecular esters of lactic
acid can be described as the set of successive reactions shown below, where W denotes water

 

2LA, :3— LA,+ W (5)
LA2 + LA, :3 LA3 + W (6)
LA3+ LA, \——-‘ LA4 + W (7)
Generally, oligomer formation can be written as

LAM, + LA, -‘—-—‘ LA4 + W (8)
The chemical reaction equilibrium constants for the above reactions in the generalized form is
given by

"LA ."W
Kj = j j > 2 (9)

("MU-1) "“1 )

Note that since the number of moles of products and reactants is equivalent regardless of the
degree of oligomerization, the equilibrium constant written in Eq. (9) is equivalent to an
equilibrium constant written in mole fractions.

Since lactic acid oligomers (LA;, LA;, etc.) are all formed via identical reaction pathways and are

themselves chemically similar, it is reasonable to assume that the esterification reactions (Eqs. (5)
- (8) above) have the same value of equilibrium constant.

72

 

K=K1=K2=K3=K4= ..... =Kj (10)

This reasoning is analogous to the treatment of the formaldehyde model, where all
polyoxomethylenes have the same equilibrium constant since they are chemically very similar but
the formaldehyde to methylene glycol reaction involves different chemical structures and
therefore has a different equilibrium constant [13].

Eq. (9) can be rearranged to the following form

"LAJ- = "1.414)’ , (11)
where
n K
r = LA: (12)
"W

and it is recognized that n LA] and n W are properties of the solution, identical for all oligomers
at a specific superficial concentration. Because of the recursion, it is possible to write

nu, = nw‘j‘“ (13)

A total lactic acid superficial mole balance is given by
ngA = z jnLAj = nu] (1+ 2r +3r2 +4r3 +...)

"LA (14)

 

where the left hand side is the superficial number of moles of lactate in solution, the second and
third expressions represent the infinite converging series obtained by inserting Eq. (13), and the
final term represents the closed form solution. The water superficial mole balance is given by
taking the difference between the true moles present, and those consumed by hydrolysis of
oligomers

"li/ = "W _z(j—1)nLAj

= nW -nLAlr(l+2r+3r2 +4r3 +...)

 

(15)

where Eq. (13) is substituted into the summation between the second expression and the third,
and the right hand side is the closed form solution. The left-most variable in Eq. (15) is the
superficial number of moles of water. Eq. (14) can be inserted into (15) to give

nW=n;V+n2A-r (l6)

Inserting Eqs. (14) and (16) into Eq. (12) provides a relation between K and r in terms of the
superficial concentrations of lactic acid and water

73

r ni +ni r
K: (1W “2) (17)
nLA(1—-r)

Free acid and all oligomers contribute to titratable acidity that can be calculated by

 

"LA
2711A] = nLAl(l+r+r2+r3+...)=(—1—:—;5

3. 3 .A pplication

To apply the model, an overall superficial number of moles nly , n2 A and K are specified. Eq.

(17) is rearranged as a quadratic in r and solved explicitly for the value of r. The value of r is
then used to calculate "LAI from Eq. (14), and subsequently the distribution of oligomers from

Eq. (13) as well as the remaining balances.

The equations can be manipulated to express the various oligomer concentrations in terms of the
overall superficial wt.% lactic acid.

The %EMLA for LA is

%EMLA]- : 1W") (1 — r2) (19)

The superficial wt% of LA,- is
(Superficial wt% of LA,) = (%EMLA,)(overall superficial wt.% LA) (20)

The true wt.% of water is
(True wt.% water) = 100 + (overall superficial wt.% LA)(0.2r — 1) (21)

The true wt.% of a LA, is
(True wt% LA!) = (0.8j + 0.2)(overall superficial wt% LA) r0 ‘ 1’ (1 — r)2 (22)

4. Results and discussion
4.1. Analytical results and modeling

Aqueous solutions of lactic acid were prepared and analyzed for oligomer concentrations up to
120 superficial wt.% lactic acid. Table 1 gives a summary of the HPLC results and a comparison
with total acidity of the solution determined by titration. The HPLC results for overall superficial
wt.% were calculated by summing the peak areas for the individual oligomers. As a check of the
HPLC method, the total acid content by the HPLC and titration agreed within i3 wt.% for
solutions up to 105 wt.% lactic acid.

The value of the equilibrium constant K = 0.2023 was obtained by least squares regression of

%EMLA for species LAl through LA4 simultaneously. Using this value, the distribution is
modeled with an average deviation of :1: 0.12% of the reported %EMLA. For each composition

74

from Table 1, calculated %EMLA of the oligomers is presented in Table 2. From the HPLC
results, the material balance provided the superficial number of moles of lactic acid and water.
Using the value of K and the superficial moles, the value of r was determined for each overall
composition, and then Eq. (19) was applied.

Fig. 1 shows a GC/MS result for an 85 superficial wt.% lactic acid solution, demonstrating by
molecular weights that only linear oligomers of lactic acid are present. All four components,
namely LA], LAz, LA; and LA.,, were identified and verified by their respective mass

fragmentation data obtained from GC/MS.
772

42000000 471 LA2
35000000- LA 1
28000000-
21 000000-
1 4000000-

LA
7001000- 1183 4
,L limo

TIC I | I I I I I I ' I i I I I I I‘-
Scan 800 1000 1200 1400 1500
l ' ' ' ' ' ' ' ' ' l ' ' ' ' ' ’ ' ' l ' ' ' ' ' ' ' '

. """l""" 'l""""' '
Min. 5 10 15 20 25

995

LA;,

 

 

 

 

 

 

 

r

.4
d

§-.
:§
§

Fig. 1. GC/MS of 85 wt.% LA. The mass fragments (not shown) were used to verify that linear
oligomers of LA are present. No lactide was observed.

Fig. 2 shows an example HPLC chromatograph of a 115 superficial wt.% solution of lactic acid.
Fig. 3 shows total titratable acidity as a function of lactic acid concentration as summarized by
Holten [1] from various sources and from this work. The titratable acidity reflects a balance
between increasing total acid content and increasing degree of oligomerization that eliminates
free acid groups. The titratable acidity goes through a maximum at about 90 wt.% lactic acid.
The model represents the experimental data with an average deviation of at 2% of titratable

acidity.

Fig. 4 shows the experimental distribution of LA., LA;, LA3 and higher oligomers collected in
this work and compared to data from Ueda and Terashima [8] and Montgomery [7]. Higher
oligomers are denoted by LA4+, i.e. sum of tetramcrs and higher oligomers. The abscissa of Fig. 4
denotes the superficial lactic acid concentration; note that it runs through 125% as explained in
the introduction. The ordinate of Fig. 4 denotes the %EMLA distribution of lactic acid between
monomer and its oligomers on a water-free basis. The percentages are calculated as described in
the introduction. The lines shown in Fig. 4 are the calculated values of LA., LA;, LA3, LA., and
LA4+ from the model. Excellent agreement is seen between the experimental values of this work
and the values calculated from the model.

It can be seen from the experimental data of this work and also from Montgomery [7], that there
is a maximum value of approximately 15% EMLA LA3 occurring at 1 14 superficial wt.% and a
maximum value of 29% EMLA LA; occurring at 105 superficial wt.%. Experimental data from
Ueda and Terashima [8] are also presented; this set of experimental data runs up to 87% total
acidity. Watson’s [9] experimental data are not plotted because he reports the presence of lactide,
which is known to be unstable in aqueous solutions.

75

Fig. 5 compares the experimental analysis and model concentrations of LA5 through LAID for
solutions with superficial lactic acid content of 80 to 125 wt.%. The agreement is excellent for
analyzed solutions up to 108 superficial wt.% of acid. The agreement is not as good for the
solutions with superficial concentrations of 116 wt.% and 120 wt.%. These samples were
analyzed in two fractions

as discussed above. Since the total acid content is in good agreement by HPLC and titration
(Table l), we believe that the disagreement between the model and HPLC results is due to the
incomplete separation of oligomers in the HPLC, even though distinct peaks appear on the HPLC
chromatogram. Attempts to refine the HPLC method further for these very high molecular
weight solutions have not been successful.

Concentrated solutions of lactic acid (>105 superficial wt.%) are fluid at 120 °C, but are very
viscous at room temperature. The solutions had a very slight amber tint, but none of the dark
coloration indicated by Montgomery [7]. Our results are in good agreement with those of
Montgomery [7] except at the highest concentration. Montgomery reported incomplete
separation of LA; and higher oligomers—a problem that we experienced only for higher
oligomers (>LA5). To test for hydrolysis under analysis conditions in this work, ethyl lactate was
analyzed using the same HPLC method as for the lactic acid oligomers and was found to be
stable. Also, our results are also consistent with those of Montgomery, who tested extensively for
hydrolysis.

In discussion of the distribution of weight percentages in lactic acid solutions, it is appropriate to
express the concentrations in terms of superficial wt.%. The superficial wt.% for oligomers can
be quickly calculated from the values in Table 1 by multiplying the total acid superficial wt.% by
the % EMLA. A summary of true weight percentages calculated by the oligomer model is shown
in Table 3.

76

LA;

 

 

 

 

 

 

 

 

156‘
LA,
‘4 LA] LA4
:2 100
c
>
E LAs
50“

 

 

1 1 1 1 1
2.5 5.0 7.5 10.0

Minutes

Fig. 2. HPLC chromatograph of the water soluble fraction from 115 superficial wt.% lactic acid
demonstrating the separation of oligomers.

10

 

Titratable acid
(mol/100g-solution)

 

 

 

o l l l l
0 25 50 75 100 125

Superficial wt % lactic acid

Figure 3. Total titratable acidity tabulated from various workers by Holten [1] and measured in
this work compared with the model proposed in this work. 1:1 data compiled by Holten, I this

work.

77

 

100 — —~
so - 0
so ~ LA, LA,+
7o .
so -
50 -
4o —
30 -

%EMLA,

20 " LAz LA3
10 ~ LA A
D I 4
O - A ‘
0 25 50 75 100 125
Superficial wt % lactic acid

Figure 4. Experimental oligomer distribution compared with the model expressed as %EMLA.
Solid lines represent the model, solid symbols are measured in this work and open symbols are
from literature as reported by [7] and [16]. The curve labeled LA4+ indicates the sum of all
oligomers LAj where j 2 4.

D

 

 

 

 

 

 

 

o
l 2
12.0 - 1:1
0
B
9.0 1 .
j- .
E . .
o\° 6 0 - A M9
M10
3.0 - 5’ A
’/
. _ I .
LA;, ‘ - 0 °
0.0 4 . — . 7————--‘f#--"'"‘. {1‘ , ,
80 90 100 110 120

Superficial wt % lactic acid

Figure 5. Experimental oligomer distribution compared with the model expressed as %EMLA.
Experimental difficulties in analyzing the two highest concentrations are discussed in the text.

78

:MNONO H V— wcmmfl _DfiOE or: :0 “Emma town—nao—ao 0.8 want—:00 wcmcmmEOH 0:» fig a$_=mOu _8=08_.~0me 80¢ 0.5 WEED—00 hack wmhc OFF

 

 

 

79

E” 83 33 88.8 83 8% 8% 88.8 8} NE. SN 838 «3 83- 8.8. 83
”.8 8.4 :8 8.8 8.5 8.8 2: 5: mi N.: 8.5 a; 3.. .8..- S: :8
8.8 82 8:. ME 88 84. N3 2. Z: ”.8 8.8 33.8 988 83- :2: 38.8
8.8 83 23 83. 82 m; 88 2: 8.: SN .8 was :88 82- 32 88.8
8.: 88.8 88 83. 8._ Ba 8+ :8 a: 8.8 m8 32 2: 83- 32 M88
v.2 ~88 ~88 82 82 82 :8 a: 8.2 8.8 we. 82 88.8 88.8- 8.2: :88
88 88.8 88.8 88 8; 82 :._ 88 3: m8 m? 88 88.8 :3 4.8 88.8
8.~ 88.8 8.8 88.8 28.8 was 82 82 Ge 2% m8 8; :88 82 8.8 88.8
a; 88.8 88.8 88.8 88.8 :88 88.8 88.8 8..“ we v.8 83 :88 84 8.; 83
82 88.8 88.8 88.8 88.8 888 8.8.8 83 o: 8.: 8.8 38.8 .83 88 8.: 23
ES 88.8 88.8 88.8 88.8 88.8 88.8 88.8 38.8 2: 8.8 88.8 $2 8.2 8.8 :3
88.8 88.8 88.8 88.8 88.8 88.8 8.8 8.8 82. 8.5 28 88.8 :88 88 58 23
28.8 88.8 88.8 88.8 88.8 88.8 :88 28.8 82 8.8 08 38.8 m: :8 8.: 23
8.8 88.8 88.8 88.8 88.8 88.8 88.8 :88 .88 88 :8 28.8 828 8.: 8.8 88
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 :88 8.8 8.8 88.8 8.8 88 M2: :88
+13 25 .5 .3 :3 :3 :3 :3 £4 :3 .3 85.5 8.55 3.23 E

«323; 83.3.5 L 5.: A: 33:83 22 “V35%

 

33m... 80¢ mcoEmanoo 13:255qu 2: (8 some 8 mSEow=o 88 <Ei~m$ toga—=28 .3 $5.55
N 03m...

 

 

_.ww omm om.m o_.m no; mm: mo; _w._ m _ ._ mmwd oomd o_mod mm- mm_

m.ww mm.m wod odd nmd wwd mwd mmd cod $.w odd ow_d om- om—

mmd mod mow mmd and and ed. m.m_ _.w_ m.w_ 9: oood m T m:
oood womd 34 do; mm.m wm.m _m.w w.m_ 5.: omm 5mm E..— o_. o:
omod wood Emd oowd S; m_.m wm.w omd o.o_ omm odm $.m m- mo_
mood o .od omod god owmd wood 9: mww o.: odm omw omd o ooo
oood _ood wood m .od omod womd mmnd wwm wod 9mm w.mm owd m mo
oood oood Sod mood Edd .ood mwmd mm; mm.m wd— foo mg.“— o_ oo
oood oood oood _ood mood o _od 2 _.o omod mmd md— m.mo WC 2 mo
oood oood oood oood .ood mood mwod ommd w _ .m ox: mmo o. .m om ow
oood oood oood oood oood mood omod mm _ .o mw._ w.o_ m. G wdm m m mm
oood oood oood oood oood .ood oood wood Sod Id 9% :m om on
oood oood oood oood oood oood wood Sod _ Ed wmd odm w.mm mm 3
oood oood oood oood oood oood mood wmod oowd w _ .m mam odw ow do
oood oood oood oood oood oood Sod m .od oomd mow mdm m.mw mw mm
oood oood oood oood oood oood oood wood 52 .o _ _.m mdw wdm om om
oood oood oood oood oood oood oood wood mo_ .o mmm mmw m.mm mm mw
oood oood oood oood oood oood oood mood wood m5. odm mdo oo ow
oood oood oood oood oood oood oood Sod wmod om; 9mm fimd 3 mm
oood oood oood oood oood oood oood oood _mod wmwd odm mom on om
oood oood oood oood oood oood oood oood _ .od mmmd m.wm mm mm mm
oood oood oood oood oood oood oood oood mood ommd o.o_ odw ow om
oood oood oood oood oood oood oood oood mood \iw _ .o w.w_ odd mm 2
oood oood oood oood oood oood oood oood Sod omod 5d odo do do
oood oood oood oood oood oood oood oood oood o_od wow odd mo m
+:<A 851— 35 8’3 £3 :5 “<1— ..ti— «<4— ~<Q J:— 3225 .533 3&3 <4 $43

Saigon—8 2.09.2. 539$ 9...; 3653.5 .aBEon—sm

 

.mcoEmooEoU E65853 mzocg 5o mLoEow:o Box 28m. 98 583 mo cows 25 do macaw—:23 .232 .m 05m...

80

4. 2. Implementation of lactic acid model into ASPEN plus

Implementation of the model is extended to ASPEN Plus, which is the most widely used
simulation software in the chemical process industry. Use of this model will be shown in future
publications for the esterification of lactic acid with ethanol from the authors’ laboratories [24].

The proposed model could be incorporated into the process simulator via a user-written
subroutine. As an alternative, we assume that oligomerization is adequately approximated by a
truncated series. Fig. 4 implies that solutions up to 90 wt.% can be represented by monomer
lactic acid and the first four oligomers (LA2—LA5). We have used this assumption to simulate a
distillation column for the purpose of evaluating its suitability for process simulation.

Stream 1 - 22 wt% (superficial) Lactic Acid

 

 

 

 

 

 

 

Flow 100 kg/s Stream 2
Wt % Flow 46.27 kg/s
Wt %
LA 22.0
Water 78.0 LA1 0.00
LA2 0.00
LA3 0.00
‘ . LA4 0.00
A LA5 0.00
Water 100.00

 

 

 

Reflux ratio : 0.5
Boilup ratio : 19.2

 

 

Stream 3 - Concentrated Lactic Acid

0 23.73 kgls
Full model

 

 

LA 58.15 57.90
LA2 21.86 21 .85
LA3 6.59 6.62
LA4 1.80 1.82
LA5 0.46

Water 11.13 11.17

 

 

 

Fig. 6. Process flow diagram and results for the truncated ASPEN simulation compared to the
complete oligomer model. The comparisons of composition are for a superficial composition of
92.72 wt.% lactic acid.

Fig. 6 shows the ASPEN Plus simulation to remove water from a 22 superficial wt.% lactic acid
solution (non-equilibrated) and form an equilibrated 92.72 superficial wt.% solution. The
reactive distillation column is assumed to operate with equilibrium stages, so the bottoms product
contains an equilibrium mixture of lactic acid oligomers at an overall concentration of 92.72
superficial wt.%. The oligomer concentrations obtained from the ASPEN Plus simulation with
the truncated model compare well with those from the non-truncated oligomer model as
summarized in the inset table within Fig. 6. The simulation verifies that the model can be used to
model a distillation column where a dilute solution of lactic acid is converted to concentrated
solution consistent with oligomer distribution represented by the full model. Other options for
comparison of the truncated and full model could have been used, such as an equilibrium reactor
with a non-equilibrium feed; the selection of a distillation column was arbitrary.

81

4. 3. Eflect of temperature and its effect on equilibrium constant (K)

There are no experimental reports available on heats of formation of oligomers of lactic acid.
Other esterification reactions involving carboxylic acids and alcohols are either therrnoneutral or
have very low heats of formation in the range of 2—6 kJ/mol [25—27], resulting in negligible to
modest changes (IO—15%) in equilibrium constants with temperature changes of 80 K. In this
work, the series of esterification reactions leading to formation of oligomers are assumed to be
thermoneutral, resulting in a temperature-independent K= 0.2023. Also, the oligomerization
reactions are extremely slow at room temperature, which makes it very difficult to assess the
reaction kinetics and time required to reach any redistribution at room temperature [1].
Experiments over a period of eight weeks showed no measurable redistribution of oligomers from
the solutions that were prepared at the elevated temperatures reported above.

5. Conclusions

In this work, we provide new data to complement and extend literature data for oligomerization
of lactic acid in aqueous solutions. We present a model based on chemical theory that consists of
an infinite sequence of equilibriumhomo—esterification reactions between successive oligomers of
lactic acid. We show that a single value of the equilibrium constant (K= 0.2023) applied to all
oligomerization reactions accurately predicts titratable acidity and oligomer concentrations for
solution concentrations ranging from very dilute to greater than 100 superficial wt.% lactic acid.
We demonstrate that inclusion of oligomers only up to LA5 is suitable for process modeling of
lactic acid solutions up to 90 wt.%.

List of Symbols
K,- chemical reaction equilibrium constant for j order oligomer
LA monomeric lactic acid

LA2 dimer lactic acid, lactoyllactic acid
LA3 trimer lactic acid, lactoyl-lactoyllactic acid
LAj polymeric lactic acid consisting on j units of lactic acid

n, molar concentration of component j

r defined by Eq. (12)

x, Mole fraction of componentj

)j activity coefficient of component j
Superscripts

i initial (used for superficial number of moles)
Subscripts

FA formaldehyde

j component

LAj polymeric lactic acid consisting of j units of lactic acid
MG methylene glycol
MG" higher polyoxomethylene glycols

to order of oligomer
W water
Acknowledgement

The authors extend appreciation to the National Corn Growers Association and the Department of
Energy for financial support.

82

References

[1]
[21
[3]

[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]

[l8]
[19]
[20]
[21]
[22]
[23]
[24]

[25]
[26]

[27]

CH. Holten, Lactic acid: Properties and Chemistry of Lactic Acid and Derivatives, Verlag
Chemie, 1971.

CH. Scheele, Om Mjolk: Kgl. Vetenskaps-Academiensnya Handlingar, 1 Stockholm (1780) 1 I6—
124.

M.H. Hartmann, in: D.L. Kaplan (Ed), Biopolymers from Renewal Resources, Springer-Verlag,
Berlin, 1998, pp. 367-41 1.

SK. Ritter, Chem. Eng. News 82 (22) (2004) 31-34.

W. Robert, Chem. Week. April 10 (2002) 31.

D. Garlotta, J. Polym. Environ. 9 (2002) 63—84.

R. Montgomery, J. Am. Chem. Soc. 74 (1952) 1466—1468.

R. Ueda, T. Terashima, Hakko Kogaku Zaashi. 36 (1958) 371—374.

P.D. Watson, Ind. Eng. Chem. 32 (1940) 399—401.

K. Tanaka, R. Yoshikawa, C. Ying, I-I. Kita, K. Okamoto, Chem. Eng. Sci. 57 (2002) 1577—1584.
Holten, ibid, pp. 200.

A. Engin, H. Haluk, K. Gurkan, Green Chem. 5 (2003) 460—466.

G. Maurer, AlChE J. 32 (1986) 932—948.

L.V. Kogan, Zhur. Prikl. Khim. 52 (1979) 2149.

L.V. Kogan, S.K. Ogorodnikov, J. Appl. Chem. USSR 53 (1980) 98—101.

L.V. Kogan, S.K. Ogorodnikov, J. Appl. Chem. USSR 53 (1980) 102.

V. Brandani, G.D. Giacomo, P.U. Foscolo, Ind. Eng. Chem. Process. Res. Dev. 19 (1980) 179—
185.

J. Masamto, K. Matsuzaki, Chem. Eng. Jpn. 27 (1994) 6—1 1.

I. Hahnenstein, M. Hasse, Y.-Q. Liu, G. Maurer, AlChE Symp. Ser. 298 (1994) 141—157.

S. Brandani, V. Brandani, G.D. Giacomo, Ind. Eng. Chem. Res. 30 (1991) 414—420.

S. Brandani, V. Brandani, G.D. Giacomo, Fluid Phase Equil. 63 (1991) 27—41.

S. Brandani, V. Brandani, G.D. Giacomo, Ind. Eng. Chem. Res. 31 (1992) 1792—1798.

S. Bezzi, L. Riccoboni, C. Sullam, Mem. cl. sci. fis. mat. nat. 8 (1937) 181—200.

N.S. Asthana, A.K. Kolah, C.T. Lira, D]. Miller, US. Patent Appl. 2004. Improved Process for
Production of Organic Acid Esters.

J. Gangadwala, S. Mankar, S. Mahajani, Ind. Eng. Chem. Res. 42 (2003) 2146—2155.

W. Song, G. Venimadhavan, J .M. Manning, M.F. Malone, M.F. Doherty, Ind. Eng. Chem. Res. 37
(1998) 1917-1928.

W. Jiu, C. Tan, Ind. Eng. Chem. Res. 40 (2001) 3281—3286.

83

Chapter 6 - Vapor-Liquid Equilibrium

6.1 - Overview

This chapter summarizes the works in vapor-liquid equilibrium measurements,
and application of the lactic acid oligomer distribution model to the systems where
concentrated lactic acid is used.

A custom-made P-x-y apparatus was built to reliably perform isothermal VLE and
VLLE measurements down to about 0.7 kPa and temperature up to 80 0C. The use of this
apparatus can easily be extended to higher temperatures with a minor modification.
Isothermal VLE data of the systems of ethyl lactate + ethanol, ethyl lactate + water,
triethyl citrate + water, triethyl citrate + ethanol, and diethyl succinate + ethanol, which
did not exist in the open literature, were measured and become readily available using
this apparatus.

As reviewed in the previous chapter, isothermal VLE measurements are preferred
over the isobaric measurements, and are more suitable for thermally labile and reactive
substances. But, for a system with slow kinetics, isobaric VLE measurements using T-x-y
can be successfully obtained. Therefore, a commercial Fischer still was used to
isobarically measure the VLE of the system lactic acid + water and lactic acid + ethanol +
ethyl lactate + water. Details of experiments, results and limitations of the T—x-y Fischer

are presented in the next sections.

6.2 - P-x-y Apparatus and VLE Measurements
Overview of the P-x-y apparatus is shown in Figure 6.1. Details of the operating

procedure and schematic for this apparatus are described in the paper (Vu et al., 2006),

84

included in section 6.2.1. The additional schematics not included in the published paper

are shown in Figures 6.2 and 6.3.

 

Figure 6.1 Set up of the P-x-y apparatus

85

Vs

Vs
Erlemeyer U

 
   
 

 

 

 

flas

Magnetic bar

Figure 6.2 Configuration of the Agitator and Liquid Sampling Valves

He He

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—1><1—> Vacuum WVacuum
v chw .,_ a _ ____ W 1Y1“?
Sample loop , 1 Sample loop
60 : ’ 4M 1 GO 81L
5 i __ '
_ 6 ~ : '

   

Vapor line Vapor line

Figure 6.3 Vapor Sampling Valve at the Inject (left) and Load (right) Position

86

6.2.1 - P-x-y Data of Ethyl Lactate Systems and (Ethanol + Water at 40 °C)
— A Reprint of the Paper “Vapor-Liquid Equilibria in the Systems Ethyl

Lactate + Ethanol and Ethyl Lactate + Water”

Included in this section is a copy of the paper Vapor-Liquid Equilibria in the
Systems Ethyl Lactate + Ethanol and Ethyl Lactate + Water, by D.T. Vu, C T. Lira, N. S.
Asthana, A. K. Kolah, and D. J. Miller. The paper is reformatted from Journal of
Chemical Engineering Data, 2006, 51, 1220-1225, for the same purpose as was explained

in section 5.1 of chapter 5.

87

J. Chem. Eng. Data 2006, 51, 1220-1225

Vapor-Liquid Equilibria in the Systems Ethyl Lactate + Ethanol and
Ethyl Lactate + Water

Dung T. Vu, Carl T. Lira,* Navinchandra S. Asthana, Aspi K. Kolah, and Dennis J. Miller

Department of Chemical Engineering and Materials Science, 2527 Engineering Building, Michigan State
University, East Lansing, Michigan 48824

Abstract

A simple vapor-liquid equilibrium (VLE) apparatus has been constructed to successfully measure
the VLE of binary ethyl lactate systems that have relatively high differences in volatility
(P2'°““/Plsat ~ 7.0). Degassing is done in situ, reducing the experimental time considerably.
Isothermal VLE of the ethyl lactate + ethanol system was measured at (40.0, 60.1, and 80.2) °C,
and the isothermal VLE of the ethyl lactate + water system was measured at (40.0 and 60.0) °C.
The ethyl lactate + ethanol system is slightly nonideal, and the ethyl lactate + water system forms
a minimum boiling azeotrope. Isothermal data for ethanol + water were measured at 40.0 °C to
demonstrate reliability of the apparatus.

 

* Corresponding author. Phone: (517)355-9731. E-mail: lira@egr.msu.edu.
© 2006 American Chemical Society
Published on Web 06/28/2006

88

Introduction

Interest in lactate esters is increasing due to emphasis on environmentally friendly solvents from
bio-derived sources. Lactate esters (primarily ethyl lactate) have excellent solvent properties and
low toxicity and are candidates to replace many halogenated solvents including ozone-depleting
CFCs, carcinogenic methylene chloride, toxic ethylene glycol ethers, and chloroform.l Lactate
esters such as ethyl lactate have the ability to dissolve a wide range of chemicals. They can be
used to remove greases, silicone oils, and adhesives in cleaning a variety of metal surfaces for
fabrication and coating applications. Because ethyl lactate exists in beer, wine, and soy products,
it has been approved by the FDA for use in food industries for many years.

Despite their numerous attractive advantages, the production volume of lactate esters used has
been small in industry. Traditional batch processing is expensive compared to the potential for
continuous processing. New technologies have been developed to yield lactate esters from
carbohydrate feedstocks via esterification using reactive distillation or pervaporation
membranes.2'3

Esterification usually requires distillation to purify the esters. For column designs and process
simulation, thermodynamic properties such as reliable vapor-liquid equilibrium (VLE) data of the
related components are valuable. Recently, phase equilibrium of the methyl lactate system has
been studied, and VLE of some lactate esters with their associated alcohols at 101.33 kPa were
made available."’5 However, no information for the ethyl lactate + water system has been found
in the existing literature. This work presents the equilibrium P-x-y data of the ethyl lactate +
ethanol and ethyl lactate + water systems. We have chosen to collect P-x-y data isothermally
because the temperature can be kept low where the reactive system ethyl lactate + water is
kinetically more stable.

Experimental Details

Chemicals. Ethyl (S)-(-)-lactate 98 % and ethyl alcohol (200 proof) were purchased from Sigma
Aldrich. Water (HPLC grade) was obtained from J. T. Baker, Inc. Water and ethyl alcohol were
used as received. Ethyl lactate was further purified by vacuum distillation. Only 85-90 % of the
pre-distilled volume was collected for the VLE experiments. Both the first overhead fraction (5-
10 %) and the reboiler residue (5 %) were discarded. No detectable water or ethanol remained in
the ethyl lactate after distillation as determined using gas chromatography (GC). The GC
procedure will be described in the analytical method section.

Apparatus. A P-x-y apparatus was constructed for VLE measurements of binary systems from
ambient temperature to 353K (Figure 1). The apparatus is based on the design of similar
equipment described in the literature.6 The apparatus has three main sections: an equilibration
section, a feed section, and a sampling section.

(a) Equilibrium Chamber and Isothermal Bath. A modified 125 mL Erlenmeyer flask was used
as an equilibrium cell. The cell was placed on a submersible stir plate immersed in the isothermal
water bath. Temperature was maintained by a PolyScience series 730 circulator. To minimize
water bath evaporation, approximately 1 in. of mineral oil was added to the bath to cover the
water’s surface when conducting experiments at 80 °C. The bath had fluctuations less than i
0.0] °C at 40 °C below, but the variation was i 0.05 °C at (60 and 80) °C. Temperature was
measured using a thermometer calibrated against a NIST traceable thermometer; the accuracy
was better than i 0.001 °C. Pressure inside the cell was measured using a MKS Baratron model
PDR 2000 dual capacitance diaphragm absolute pressure gauge. The pressure gauge provides
reliable values between 0.13 and 133 kPa with the resolution of 0.013 kPa and an accuracy of
0.25 % of the reported reading.

89

The cell was connected to the feed and gas sampling systems using 1/16 in. o.d. 316 stainless
steel tubing sealed to the chamber using ACE glass Teflon adapters (Catalog No. 5801-07) and
connectors (Catalog Nos. 5854-07 and 5824-24). The Baratron gauge was attached to the top of
the cell using a length of glass tubing with a tapered ground glass joint to provide a vacuum tight
connection. The Baratron and glass were joined using a Cajon union (SS-4-UT-6).

The liquid and vapor phases were both stirred. Two different vapor-phase stirrer configurations
were used in the course of this work. In the first configuration, a vertical length of 1/8 in.
stainless steel rod was used to support the vapor-phase agitator. The rod was placed vertically in
the center of the equilibrium cell; the bottom end was soldered to a small clip mounted onto a
magnetic stir bar. At the middle of the vertical rod, two small arms were created by soldering a
wire to the rod. Teflon plumbing tape (1/2 in. X l in. x 0.04 in.) was wrapped around the arms to
create the agitator. The bar and Teflon tape provided the means of mixing for the liquid and
vapor phases simultaneously. However, when the apparatus was modified by adding a liquid-
phase sampling section, the equilibrium chamber had to be placed 3/4 in. above the submersible
stir plate. Consequently, the magnetic field was considerably reduced, the bottom of the flask was
no longer flat, and the vapor stirrer did not work reliably. Thin polypropylene strips (0.06 in. X 3
in. x 0.04 in.) were wrapped around the center of the magnetic stir bar, and small supports were
fabricated from Teflon sheet.

Feed He

     
    
    
  
 

 

 

 

 

 

 

 

 

 

 

 

chamber Liquid l Pressure
. Sampling
GC ‘ valve
.Pressure
V1A .- ge Vapor
line

 

 

 

 

 

19W! pgnbn

 

rain! 91an

 

 

 

Constant Temperature Bath

Figure 1. Schematic of the apparatus

(b) Feed Section. Two 125 mL flasks and two liquid injectors were connected using 1/4 in. o.d.
polypropylene and 316 stainless steel tubing and Swagelok adapters. Polypropylene tubing
provided flexibility for the connection between glass (feed chambers) and stainless steel valves
(VIA, VIB) and permitted observation of the liquid level in the feed section. The length of
polypropylene tubing was minimized to limit permeability of air from the environment. The
flasks were mounted 3 ft above the injectors, providing a hydrostatic head to load the injectors
with liquids from the flasks when valves VIA and V”; were opened (Figure 1). The liquid

90

injectors were 30 mL calibrated pumps (High-Pressure Equipment Company 62-6-10) used to
meter liquids to the equilibrium cell with the accuracy of :1: 0.003 mL of the injected volume.
Pressure of the liquids inside the injectors was monitored using inexpensive pressure gauges.

(c) Liquid-Phase Sampling. Degassing of the liquids in the feed section (flasks and injectors)
was tedious. However, we found that the liquids could be degassed reliably within the
equilibration chamber. Complete degassing was easy to identify by a reliable stable pressure in
the chamber after repeatedly pulling the pressure down about 1 kPa. Because of the expected
minor shift in composition during degassing after liquids were charged to the equilibrium
chamber, a liquid sampling section was added to the apparatus. This modification was done for
the ethyl lactate + water system, reducing considerably the experimental time. High vacuum
needle valves, purchased from Chemglass (CG-553-02, CG-534-02) were connected by a 4 in.
length of 1/4 in. o.d glass tubing. To take a liquid sample, valve V6 was first opened to permit
evacuation of the sample region. Then valve V6 was closed before valve V5 was cracked opened
for 10 s to collect approximately 0.2 mL of liquid from the equilibrium cell. No fluctuation in
pressure of the equilibration cell was noted when valve V5 was opened. After sample collection,
valve V5 was closed entirely and valve V6 was opened fully to permit a narrow Teflon tube
connected to a syringe to be inserted for withdrawal of most of the liquid sample. To remove all
residual traces of liquid, acetone was added through V6 and then removed via the syringe
apparatus. Any remaining acetone was evaporated under vacuum while the cell was undergoing
the next equilibration.

(d) Vapor Phase Sampling. The vapor sample system was based on a Valco six-port switching
valve (00V-l375V) positioned immediately above the water bath, approximately 8 in. from the
equilibrium cell. A high-temperature rotor (SSAC6WE, 225 °C) and preload nut (PLAW30)
were chosen as part of the valve assembly. The vapor line was 1/16 in. stainless steel with a 1/16
in. stainless steel valve. The vacuum line was a 6 in. length of 1/16 in. stainless steel connected
to a 1/16 in. valve and adapted to vacuum tubing. The He carrier gas entered through 1/16 in.
stainless tubing connected to the outlet of the gas chromatography (GC) injector, and 1/ 16 in.
stainless tubing was used to return the sample to the GC oven where it was fed onto the column.
The GC was placed as close as practical to the apparatus, using about 24 in. of tubing between the
GC and the sample valve. A 1.8 mL sample loop was created by adapting a coiled length of 1/4
in. tubing to the Valco ports. Each vapor sample was equivalent to about 0.3 ,uL of the related
liquid mixture directly injected into the GC. To avoid condensation of the high boiling
components, the vapor line was heat-traced and maintained 15-20 °C above the temperature of
the equilibrium cell. To collect a vapor-phase sample, the sample loop was evacuated by placing
the valve in the “load” position with the vapor line valve V3 closed and the vacuum valve We
opened; then the valve W, was closed, and the vapor line valve was opened. The loading was
done within 1 min, and then the valve V3 was closed and the sampling valve was switched
quickly to the “inject” position. No pressure drop in the equilibrium cell was observed during
the course of vapor sampling, since the volume of vapor sample was small as compared to the
volume of the chamber. Additional details on the vapor and liquid sampling configurations are
available from the corresponding author.

Experimental Procedure. A Sargent-Welch two-stage vacuum pump (model 1400) was used to
evacuate the apparatus and sample sections and to provide degassing of liquids. Prior to the
experiment, the entire system was evacuated and checked for the leaks. A stable base pressure of
0.07-0.09 kPa for 3-4 h indicated that the chamber was leak tight. Liquids were degassed before
they were loaded into the injectors. During the degassing process, fluids in the flasks were shaken
and tested using the click test for degassing as described by Van Ness and Abbott7 and Campbell
and Bhethanabotla.8

91

When performing experiments where the liquid composition was determined from the quantities
of liquids injected, the following tests supplemented the click test to verify complete degassing in
the feed lines and injectors and to verify a leak-tight feed section: (1) Pressure of fully loaded
injectors with degassed liquids observed from gauges PA and PB had to be steady and equal to the
vapor pressure of liquids. If the pump A (or B) was operated while VIA (or V13) was opened and
VM (or V23) was closed, the displacement of liquid level in the polypropylene feed line had to be
proportional to the displacement inside the injector. (2) If the V1 A (or V13) and VM (or V23) were
closed, the pressure of the injector A (or B) had to increase instantaneously when the pump
started to compress the liquid inside that injector.

To inject liquid A (or B) to the equilibrium cell, pressure PA (or P3) was raised to approximately
0.3 MPa before valve V2,, (or V23) was opened. After the pressure of the injector dropped, the
valve was closed, the injector pressure was restored, and the injected volume was recorded.

To carry out the experiment, 10-20 mL of component 1 of the studied binary system was charged
to the equilibrium cell. After the vapor pressure of this pure liquid was measured, a
predetermined quantity of the component 2 was added to the cell. After equilibration, vapor and
liquid samples were collected. These steps were continued until the liquid mole fraction of
component 1 approached 0.1. Afterward, the equilibrium chamber was emptied; the entire system
was cleaned and degassed thoroughly. Then, the process was reversed, charging the equilibrium
cell first with component 2 and then adding component 1.

The volume of the initial charge in the experiments with the ethyl lactate + ethanol system was
selected to ensure that error in calculation of liquid compositions from the injected volume would
be negligible. For the ethyl lactate + water system, 5mL of liquid inside the equilibrium chamber
was found to be sufficiently large to ensure accurate composition measurements, because the
volumes of liquid injections were not critical with the liquid sampling section in place. Both
liquid and vapor of the studied binary mixture were well-mixed and were allowed to reach
equilibrium before any measurement was performed. Equilibration was identified by the
consistency of the equilibrium pressure reading from the Baratron following vapor withdrawals
using vacuum and by the reproducibility of the equilibrium vapor-phase composition.

Analytical Methods. Liquid compositions in the ethanol + water and ethyl lactate + ethanol
mixtures were calculated from the known volume of each component charged to the cell. For
ethyl lactate + water, samples of the liquid phase were taken via the liquid sampling section, and
the compositions were determined from GC analysis. Vapor samples of the studied binary
mixtures were injected to the gas chromatograph using the vapor sample valve.

The GOW-MAC 350 gas chromatograph was operated under isothermal conditions using a
carrier stream of helium at 35mL/min. The column temperature was 220 °C in experiments
involving ethyl lactate, but it was reduced to 150 °C for the ethanol + water system. A
thermoconductivity detector was set at 290 °C and 110 mA filament current. The column packing
used was Poropak Q 50/80, packed in 6 it long x 1/8 in. o.d. X 0.085 in. wall stainless steel
tubing. To ensure that all vapor samples were analyzed in the column without loss via
condensation,l ft of 1/16 in. o.d. 316 stainless steel tubing was added to the column and used as a
precolumn heater within the GC oven.

Calibrations of known compositions of mixtures were done for each binary system to obtain the
correlation between the ratio of GC peak areas and the mixture compositions. From the
calibration, the unknown compositions of the injected samples were determined. The amounts of
each component in the calibrated mixtures were weighed using an electronic balance with its
readability of 0.1 mg. The standard mixtures were prepared gravimetrically in an approximate
size of 1.0 :i: 0.3mg; therefore, the deviation in calculation of molar compositions was negligible.
To reduce the error due to the possible evaporation of the more volatile component, two duplicate

92

mixtures were prepared for each calibration point. Three GC injections were done for every data
point, in both calibration and sample analyses. The difference in the ratio of peak areas of the
triplicate GC injections was less than :1: 0.05 % of the calculated value.

Results and Discussion

Ethanol + Water System. Isothermal VLE data for the ethanol + water system at 40.0 °C were
collected and compared to literature data for validation of reliability of the constructed VLE
apparatus (Table 1). The ethanol + water system was chosen to study because its components are
in the system of interest, and 40.0 °C isothermal literature data are available from two
independent sources. Both literature and experimental data were regressed using the Britt-Luecke
algorithm, maximum-likelihood principle, provided by ASPEN PLUS 12.1. The area test of
Redlich-Kister and point-to-point test of Van Ness and Fredenslund were used to check for data
reliability?“ The data are considered to pass the area test if the difference between the positive
and negative areas is less than 10 %. However, to pass the point-to-point test, the absolute mean
deviation between the calculated and experimental vapor compositions should be S 0.01.

Table l. VLE data for ethanol (1) + water (2) at 40.0 °C

P40.O/kPa x140.0 y140.0

40.0 40.0 40.0
P /kPa X] y]

 

 

) 7.41 0 0 14.25 0.158 0.541
7.83 0.005 0.036 14.93 0.201 0.573
8.08 0.007 0.069 15.51 0.256 0.598
8.27 0.010 0.096 15.79 0.319 0.612
8.55 0.014 0.133 16.37 0.418 0.655
8.97 0.020 0.181 16.57 0.448 0.660
9.32 0.026 0.221 16.96 0.518 0.697
9.85 0.035 0.269 17.21 0.583 0.730
10.64 0.050 0.332 17.51 0.682 0.767
11.72 0.075 0.407 17.71 0.748 0.805
12.17 0.085 0.421 17.81 0.828 0.841
13.01 0.108 0.478 17.95 0.892 0.893
13.12 0.111 0.474 18.00 0.943 0.960
13.77 0.136 0.519 18.00 1.000 1.000

UNIQUAC with the Hayden and O’Connell (HOC) virial coefficient correlation were used to
evaluate thermodynamic consistency. The point-to-point test value was 0.011, significantly
smaller than that of 0.063 from Udovenko and Fatkulina'2 and 0.248 from Mertl.l3 In the
available literature, these are the only isothermal VLE data that can be found for the ethanol +
water system at 40.0 °C. Neither data from Udovenko and Fatkulina nor this work passed the
area test, but the value of 10.40 %, which is obtained from this work, is smaller than Udovenko
and Fatkulina’s value and close to the accepted value. The smoothness of the P-x-y curve in
Figure 2 and results from the thermodynamic consistency tests show that the VLE data of ethanol
+ water from this work are very reliable and more consistent than existing literature data at 40 °C.

Ethyl Lactate + Ethanol System. VLE were measured at (40.0, 60.1, and 80.2) °C for this
system (Table 2). To minimize the effects of any systematic errors in particular run, the VLE
experiments were performed at least five times using different increments and decrements of each

93

component molar fraction at the reported temperature. All the activity coefficient models listed
in Table 3 provide similar correlation of experimental data. The value of R used in the NRTL—
HOC equation is 0.3. Figure 3 shows the representation of the UNIQUAC with the HOC
correlation. The same nonlinear regression method and consistency tests were used as described.
For the HOC method, the n values were assumed to be 1.3 for ethyl lactate + ethanol and 0.53 for
ethyl lactate with itself. These values were based on the assumption that solvation of ethyl lactate
would be similar to that of ethyl acetate in ethyl acetate + ethanol mixture and that ethyl lactate
pure self-interactions would be similar to ethyl acetate pure self-interactions. It should be noted
that the calculated vapor fugacity coefficient of ethyl lactate is in the range of 0.990 to 0.998 and
that for ethanol is from 0.993 to 0.999 at the system pressure.

 

 

 

 

 

 

 

 

20 ‘ I I Y 7 I T 1 v 120 r 1 v r i I r 1* w
¢ .
8‘. a” i 9‘ 100 ‘6 l
.0' o .
'88 A % 1
15 — o o : 3°
. 0° .3
.. . l r .
3‘: ° <> 8 so
it ~ 5’ g‘ r ‘
i. a 5 0
10 ”.0 g 0 1 40 0
8. o 1
a' . .
20 ' a
1
.>
5 . i , w * ‘ d ‘ ‘ o L , *
0.0 0.2 0.4 0.6 0-8 1-0 0.0 0.2 0.4 0.6 0.3 1.0
X 1- Y1 x1- y 1
Figure 2. P—x-y of ethanol ( l) + water (2) at Figure 3. P-x-y of ethyl lactate (l) + ethanol
400°C: o-this work, A- Udovenko and (2) system. A 400°C; 060.1°C; 0 802°C;
Fatkulina”, and <>-1Vl€fl'tl-l3 solid lines are the representation of

UNIQUAC with HOC correlation.

Data are combined from at least five different runs for each reported temperature as described.
All P-x-y diagrams are smooth and do not exhibit any trends of systematic error within specific
runs. All experimental data satisfied the point-to-point test, but only data at 40.0 °C passed the
area test. The area test results were 31 % and 19 % for data at (60.1 and 80.2) °C, respectively.
The inconsistency could be due to the error in measuring the vapor phase at low concentration of
ethyl lactate where the GC detection was limited. Another potential source of error could be
minor decomposition of the ethyl lactate in the GC detector during vapor-phase analysis. It was
noted during runs that the outlet lines of the thermal conductivity detector gradually became
restricted due to deposits over a period of several hours. The lines were kept clear using a syringe
cleaning wire, but this method did not allow determination of the extent of decomposition.
Plugging of lines was not noted on the GC used to analyze the liquid samples. Additional
experimental runs were consistent with each other, as compiled in the tables and figures, and did
not improve the results of the consistency tests.

94

Table 2. VLE data for ethyl lactate (l) + ethanol (2) systems at (40.0, 60.1, and 80.2) °C

 

P40.0/kPa

40.0

40.0

P60.l/kpa

60.1

60.1

P80.2/kPa

80.2

80.2

 

 

xI Y1 X1 5’1 X1 Y1

1.12 1.000 1.000

2.57 0.951 0.433 3 .03 1.000 1.000

3.59 0.893 0.271 5.97 0.946 0.482

4.28 0.862 0.219 8.55 0.897 0.306

5.45 0.814 0.16 11.41 0.836 0.205

6.55 0.754 0.125 14.51 0.774 0.148 7.63 1.000 1.000
7.91 0.689 0.093 17.64 0.722 0.101 14.08 0.935 0.488
9.21 0.608 0.074 19.24 0.675 0.095 22.08 0.863 0.283
9.92 0.554 0.061 20.86 0.641 0.073 30.82 0.775 0.184
11.76 0.43 0.042 25.05 0.559 0.06 38.54 0.705 0.133
13.31 0.329 0.029 25.52 0.532 0.052 47.94 0.62 0.101
14.23 0.283 0.015 29.38 0.448 0.039 57.66 0.534 0.075
14.76 0.239 0.024 32.1 0.386 0.034 67.94 0.443 0.059
15.81 0.172 0.008 33.26 0.354 0.027 81.3 0.316 0.032
16.81 0.102 0.012 36.97 0.266 0.022 81.37 0.316 0.036
16.99 0.097 0.004 37.17 0.259 0.019 92.05 0.203 0.02
17.31 0.073 0.003 39.81 0.195 0.011 100.31 0.121 0.013
16.37 0.120 0.000 42.46 0.128 0.012 101.42 0.106 0.007
18.01 0.000 0.000 47.21 0.000 0.000 109.12 0.000 0.000

Table 3. Binary Parameters of Ethyl Lactate (1) + Ethanol (2) System and Average Absolute
Percent Deviation 4%) for Equilibrium Pressure (P) and Vapor-Phase Mole Fractions (y,, y;)’1

 

 

 

Binary Average Absolute
Equation Parameters Percent Deviation
blz/K bZI/K P/% y1/% y2/%

UNIQUAC — 1G 11.]. = eXP(bij /T) -43.00 -23.10 3.3 23.2 1.5
UNIQUAC-HOC Tij = exP(bij / T) -40.03 -29.40 3.1 24.7 1.4
NRTL-HOC Gij = eXP(-0-3bij /T) -298.69 585.62 3.8 24.8 1.5
Van-Laar-HOC Aij = by. /T 169.19 65.21 3.3 24.7 1.5
WILSON-HOC -198.48 71.55 3.7 24.8 1.5

 

° The vapor-phase Hayden-O’Connell parameters are given in the text.

The prediction of isobaric VLE data of ethyl lactate + ethanol at 101.33 kPa using the binary
parameters obtained from the reported data are in good agreement with Pefia-Tejedor et a1.” For
the ethyl lactate + water system at 40.0 °C, with Pefia-Tejedor’s binary parameters, the activity
coefficients at infinite dilution of ethanol and ethyl lactate are predicted to be 1.38 and 1.35,
respectively, using the UNIQUAC-HOC model. From this work, these values are 1.25 and 1.67,

95

respectively. Similar results were obtained for the data at (60.1 and 80.2) °C.

The P-x bubble line is nearly linear, and the infinite dilution activity coefficients are not large.
The ethyl lactate + ethanol system thus can be considered slightly nonideal. This is due to the
presence of the hydroxyl group in ethyl lactate, such that the interaction between ethyl lactate
molecules is similar to their interaction with the ethanol molecule.

Ethyl Lactate + Water System. VLE at (40.0 and 60.0) °C were measured for the ethyl lactate +
water binary system (Table 4). Ethyl lactate was hydrolyzed significantly at 80°C, as verified by
the presence of ethanol in CC analyses. Hydrolysis was not detected in the experiments
performed at (40.0 and 60.0) °C. The VLE experiments at each listed temperature were
performed five times; the same methods as described for the ethyl lactate + ethanol system were
used. Figure 4 shows that the system has a minimum boiling azeotrope, occurring at 5-7 mol %
ethyl lactate. Due to the narrow phase envelope at high water concentrations, it was not possible
to determine the exact azeotrope composition using gas chromatography, even though the
analysis was very reproducible.

 

25
20

15»

P/ kPa

10-

 

 

 

 

o 1 . 1 l
0.0 0.2 0.4 0.6 0.8 1.0
X1, Y1

Figure 4. P-x-y of ethyl lactate (1) + water (2) system. 0 400°C; 0 600°C;
solid lines are the representation of UNIQUAC with HOC correlation.

96

Table 4. VLE data for ethyl lactate (1) + water (2) system at (40.0 and 60.0) °C

 

 

P400 /kP a xl40.0 y140.0
1.12 1.000 1.000
1.23 0.994 0.941
1.44 0.985 0.811
1.63 0.975 0.722
1.76 0.970 0.661
1.87 0.964 0.626
2.00 0.958 0.584
2.12 0.952 0.560
2.28 0.945 0.500
2.44 0.935 0.474
3.05 0.903 0.388
3.25 0.874 0.361
3.93 0.834 0.272
4.56 0.770 0.240
5.20 0.699 0.197
6.07 0.620 0.153
7.01 0.502 0.111
7.27 0.433 0.103
7.47 0.374 0.073
7.48 0.367 0.094
7.56 0.300 0.068
7.48 0.252 0.087
7.64 0.225 0.061
7.67 0.171 0.050
7.61 0.137 0.085
7.65 0.124 0.046
7.61 0.073 0.039
7.49 0.025 0.015
7.47 0.000 0.000

 

P600 /kP a x1600 2[160.0
3.03 1.000 1.000
4.91 0.973 0.594
6.01 0.949 0.457
7.04 0.938 0.405
7.83 0.912 0.351
9.04 0.892 0.319
8.53 0.891 0.315
10.44 0.856 0.280
11.56 0.808 0.222
13.21 0.763 0.198
14.72 0.694 0.152
16.03 0.638 0.135
16.97 0.568 0.115
18.01 0.518 0.089
18.40 0.488 0.094
19.04 0.446 0.092
19.55 0.399 0.078
20.01 0.328 0.078
20.40 0.248 0.071
20.57 0.248 0.066
20.61 0.197 0.059
20.70 0.187 0.059
20.70 0.146 0.055
20.72 0.106 0.052
20.69 0.070 0.049
20.68 0.042 0.044
20.41 0.027 0.033
20.48 0.023 0.032
20.56 0.022 0.027
20.33 0.012 0.012
20.15 0.005 0.005
20.01 0.000 0.000
97

 

Table 5. Binary Parameters of Ethyl Lactate (1) + Water (2) System and Average Absolute
Percent Deviation (%) for Equilibrium Pressure (P) and Vapor-Phase Mole Fractions (y,, yz)a.

 

Binary Parameters Average Absolute

 

 

Equation Percent Deviation
b12/K b21/K P/% yl/o/o Xz/O/o
UNIQUAC — IG Tij = eXP(bij /T) 250.51 -l33.02 2.4 22 4.1
UNIQUAC-HOC Ti] = CXP(bij / 7‘) 248.19 -131.44 2.4 22.2 4.1
NRTL-HOC Gij = exp(-0.3b,~j / T) -87.07 967.2 3.4 21.6 3.8
Van-Laar-HOC Aij = bij /T 895.05 307.06 3.4 21.4 4.2

WILSON-HOC Aij = exP(bij /T), Vi /Vj :1 -978.35 -51.56 2.1 22.9 5.0

 

" The vapor-phase Hayden-O’Connell parameters are given in the text.

The data are fitted with several thermodynamic models, and the binary parameters determined are
listed in Table 5. All of the selected activity models fit the data equally well; the deviations are
given in Table 5. The HOC 1) value of 1.3 was used for ethyl lactate with water (based on the
literature value for ethyl acetate + water), and the same method as described above was applied

for data regression. The azeotrope composition is predicted to be at 6.5-6.7 mol% ethyl lactate,
based on the UNIQUAC-HOC fit.

The data satisfy the area test but are less satisfactory when analyzed via the point-to-point test.
The values of 8.6 % and 0.04 for area and point-to-point tests, respectively, were obtained for the
VLE data at 40.0 °C. Likewise, the values for data at 60.0 °C were 4.6% and 0.037. Because the
point-to-point test is more significant for isothermal VLE than the area test, the data were
carefully reevaluated, including the regression used to generate the GC calibration curve. It was
found that the difference in calculation of phase compositions using different representations of
the GC calibration curve is negligible. However, the consistency tests are very sensitive to a
small change in vapor phase composition. For example, if data point at P = 1.2 kPa in Table 4 is
omitted, the value of the point to-point test changes from 0.04 to 0.026. We have also evaluated
point-to-point consistency using Legendre polynomials11 and the Modified Margules15 method to
represent the excess Gibbs energy, but the differences between the calculated and measured
values in vapor composition are also larger than the target of 0.01. Consistency failure due to
inadequacy of the HOC method is unlikely because the vapor fugacity coefficients are near 0.989
and 0.993 across the composition range for ethyl lactate and water, respectively. Additional
experimental runs were consistent with each other as shown in the tables and figures and did not
improve the consistency test results.

Fitting of the ethyl lactate + water system is challenging because the infinite dilution activity
coefficients are large. These coefficients are 17.7 for ethyl lactate and 2.8 for water from
UNIQUAC-HOC in ASPEN 12.1. The UNIQUAC-HOC fails to represent the vapor phase
accurately at 40.0 °C and fails to represent the pressure maximum accurately at 60.0 °C, as shown
in Figure 4.

The vapor-phase analysis in this system may be subject to the same potential decomposition of
ethyl lactate as mentioned earlier. Degradation was more noticeable in this system than in the
ethyl lactate + ethanol system.

98

Summary and Conclusions

This work presents a simple design of an isothermal VLE apparatus that is capable of measuring
the vapor pressure of single components down to about 0.7 kPa and the VLE of nonideal binary
systems. The P-x-y apparatus is valuable for collecting data at low temperature, where reactive
chemicals are kinetically more stable. With the liquid sampling section and the ability to perform
the degassing in situ, the apparatus can be extended to multicomponent systems. Data have been
evaluated with standard consistency tests, and all data sets passed or nearly passed at least one of
the standard tests.

Acknowledgment

The authors are grateful to Professor Scott Campbell (University of South Florida)for
suggestions. Appreciation is also extended to Elisabeth Newton for her work in the assembly of
an earlier version of the apparatus.

Literature Cited

(1) US. Environmental Protection Agency, 1998 Alternative solvents/reaction conditions award,
citation to Argonne National Laboratory.
http://www.epa.govfi;reenchemistry/ascra98.htm1 (Accessed December 2005).

(2) Asthana, N.; Kolah, A. K.; Vu, D. T.; Lira, C. T.; Miller, D. Org. Process Res. Dev. 2005, 9, 599-
607.

(3) Benedict, D. J .; Parulekar, S. J .; Tsai, S.-P. Ind. Eng. Chem. Res. 2003, 42, 2282-2291.

(4) Sanz, M. T.; Calvo, B.; Beltran, S.; Cabezas, J. L. J. Chem. Eng. Data. 2002, 47, 1003-1006.

(5) Sanz, M. T.; Beltran, S.; Calvo, B.; Cabezas, J. L. J. Chem. Eng. Data. 2003, 48, 1446-1452.

(6) Pathare, S.; Bhethanabotla, V.R.; Campbell, S. W. Ind. Eng. Chem. Res. 2004, 49, 510-513.

(7) Van Ness, H. C.; Abbott, M. M. Ind. Eng. Chem. Fundam. 1978,17, 66-67.

(8) Campbell, S. W.; Bhethanabotla, V. R. Chem. Eng. Educ. 1997,34-39.

(9) Redlich, 0.; Kister, A. T. Ind. Eng. Chem. 1948, 40, 345.

(10) Van Ness, H. C.; Byer, S.M.; Gibbs, RE. AlChE J. 1973, 19, 238-244.

(11) Fredenslund, A.; Gmehling, J.; Rasmussen, P. Vapor-Liquid Equilibria using UNIFAC; Elsevier:
Amsterdam, 1971.

(12) Udovenko,V.V.; Fatkulina ,L.G. Zh. F 12. Khim. 1952, 26,1438.

(13) Mertl, 1. Collect. Czech. Chem. Commun. 1972, 37, 366.

(14) Pefia-Tejedor, S.; Murga, R.; Sanz, M.T. ; Beltran,S. Fluid Phase Equilib. 2005, 230, 197-203.

(15) Prausnitz, J.M.; Lichtenthaler, R.N.; Azevedo,E.G.d. Molecular Thermodynamics of Fluid-Phase
Equilibria, 3rd ed.; Prentice-Hall: New York, 1999.

 

Received for review December 23, 2005. Accepted May 19, 2006. The authors thank the National Corn
Growers Association and the US. Department of Energy for financial support.

JE050537Y

99

6.2.2 — P-x data of Citrate Systems

Triethyl citrate (99% purity, [CAS 77-93-0]) was purchased from Sigma-Aldrich,
and water HPLC grade [CAS 7732-18-5] was from J.T. Baker. All chemicals were used
without further purification, and the same experimental procedure [98] described for

ethyl lactate systems was applied for citrate systems.

O—CH
HC 2 5
5 2\

00
C32*"5

OH 0

Figure 6.4 Structure of Triethyl Citrate

The structure of triethyl citrate is shown in Figure 6.4. It is the only ethyl ester of
citric acid in this study, because no source of pure mono- or diethyl citrate was available
for the VLE experiments. Similar to ethyl lactate, triethyl citrate is also an alpha-hydroxy
ester. However, citric acid is a white crystalline powder at room temperature, and does
not undergo self-esterification like lactic acid; therefore there was no concern of citric

acid oligomer involved in the esterification producing citrate esters.

6.2.2.1 - Triethyl Citrate + Water
Phase behavior of triethyl citrate + water systems were studied at 25.0 °C and
60.0 °C, shown in Figure 6.5. Due to the low vapor pressure of triethyl citrate the gas
chromatography could not get significant vapor readings, only P-x data were obtained.
Results of the VLE measurements are summarized in Table 6-1. It shows that

triethyl citrate + water systems are partially miscible, represented by the horizontal

100

portion of the P-x curve. The miscibility slightly changes with temperature, and the
mixtures exhibit a positive deviation from Raoult’s law in the two-phase, vapor-liquid
region.

The system probably does not have an minimum boiling heteroazeotrope because
vapor pressure of ethyl citrate and water are very different from each other [99]. To
determine whether the triethyl citrate + water system has a minimum or maximum
hetero-azeotrope behavior, the normal boiling temperatures of triethyl citrate + water
mixtures were carefully measured as molar composition of triethyl citrate was gradually
varied between 0 and 0.5. Results showed that the change in temperature was negligible;
it was not possible to determine whether the heteroazeotrope boiling point was above or

below the boiling point of pure water within the error of uncertainty.

Table 6-1 P-x Data of Triethyl Citrate (1) + Water (2) at 25.0 °C and 60.0 °C

 

P25'olkPa x1250 st'olkPa x1251) P60.0’kpa x1601) #OJIkPa x1601)

 

 

3.19 0.07 20.11 0.00

3.19 0.13 3.19 0.64 20.13 0.07 19.84 0.54
3.19 0.17 3.04 0.66 20.14 0.13 18.98 0.57
3.19 0.31 2.81 0.69 20.15 0.18 17.91 0.60
3.19 0.37 2.91 0.69 20.18 0.23 16.50 0.64
3.17 0.43 2.73 0.72 20.19 0.27 15.54 0.69
3.20 0.43 2.65 0.74 20.21 0.31 13.11 0.75
3.21 0.54 2.58 0.75 20.21 0.34 9.74 0.81
3.21 0.60 2.15 0.82 20.46 0.39 5.26 0.89
3.19 0.62 1.33 0.90 20.38 0.48 0.63 1.00

 

For analyses of the P-x-y data, the vapor was assumed to be pure water. ASPEN

UNIQUAC model was used to fit data. The binary parameters are biz = -537.32 K, b2] =

94.18 K, andt'y. = exp(b,.j /T).

101

 

 

 

 

 

 

 

 

kuA..A.-A-A.A.A.A..‘..-A..‘ .
19.0 - u. -
.- ‘~ -
F 1‘.
l A‘
14.0 - -
, 60.0 ‘ -
E . (L+L+V) 1». ,
i"- i ‘ ‘
D. 9.0 L
i
4'0 L—o—H 4
E (L+L+v)25.0
-1.0 ‘ . ‘ '
00 02 0.4 06 08 10

Figure 6.5 P-x of Triethyl Citrate (1) + Water (2): A, at 60 °C; 0, at 25 °C
Dashed line and solid line are P-x-y representations of ASPEN UNIQUAC. (L+V) and
(L+L+V) denote for two-phase and three-phase regions.

6.2.2.2 - Triethyl Citrate + Ethanol

The triethyl citrate + ethanol system was miscible. The P-x data at 40 0C are
listed in Table 6-2, and Figure 6.6 shows the experimental measurements compared to the
UNIQUAC prediction. In fitting data, vapor was assumed to be at least 99 mole% of
ethanol, and the binary interaction parameters (1,] = exp(bu / T)) are blz = -294.68 K, and
b21 = 65.37 K for triethyl citrate (1) + ethanol (2). The UNIQUAC P-x-y generated from
ASPEN was consistent with measurements and assumption of vapor molar compositions,
but bubble pressure was poorly predicted when molar composition of ethanol is more

than 0.5 (Figure 6.6). It should be noticed that a poor prediction for a complex

102

compound such as triethyl citrate is not atypical, and triethyl citrate was not included in

the ASPEN data bank. This compound was defined using molecule connectivity and the

UNIFAC functional groups in order to be used with ASPEN.

Table 6-2 P-x Data of Triethyl Citrate (1) + Ethanol (2) at 40.0 °C

 

 

 

 

 

 

 

 

 

P‘°'°IkPa x1‘°'° P‘°'°IkPa x1‘°'° P‘°'°IkPa x1‘°'°
18.00 0.00 14.96 0.29 7.67 0.72
17.15 0.11 13.67 0.36 5.00 0.84
16.08 0.24 11.55 0.48 2.96 0.92

19.0 i

t . .A .

.. ‘ ..... A 4

.. ‘~‘.. 4

14.0 C In“ ‘

8 1 A‘ ~. 1

£- 1 ‘ J

O. 9.0 L . 1

; x. 1

_ a, 1

4.0 - ‘ , .

C ‘~ ‘

'1-0 " L 1 4 l 1 I L 41 .d
0.0 0.2 0.4 0.6 0.3 1.0

x1,1l1

Figure 6.6 P-x of Triethyl citrate (1) + Ethanol (2) at 40 °C
A , measured. Dashed line is P-x-y representation of ASPEN UNIQUAC

103

6.2.3 - P-x Data of Diethyl Succinate System

Diethyl succinate [CAS 123-25-1] was purchased from Sigma-Aldrich and was
distilled prior to use. This compound has limited solubility in water [100]. Boiling
temperatures of diethyl succinate (1) + water (2) systems were measured using a simple
recirculating still. At 98.5 kPa, pure water boiled at 99.2 °C, and both systems of (XI =
0.043, xz = 0.957) and (x1 = 0.061, X2 = 0.939) boiled at 98.0 °C. This result indicated

that diethyl succinate + water system has a minimum boiling azeotrope.
0

H502 o
\o \C2H5

0

Figure 6.7 Structure of Diethyl Succinate

For the diethyl succinate (1) +ethanol (2) system, due to a low vapor pressure of
diethyl succinate only P-x measurements could be taken. Table 6-3 lists data measured at
50.0 °C and the ASPEN UNIQUAC prediction is shown in Figure 6.8. In fitting data,

vapor was assumed to be at 99.99 % of ethanol. The binary interaction parameters used

in the UNIQUAC prediction (ti!- = exp(b,-j /T)) are bl; = -149.17 K, and b21= 17.88 K.

Table 6-3 Diethyl Succinate (1) + Ethanol (2) at 50.0 °C

 

 

 

 

P50.o,kpa x150.0 Pso'olkPa x150.0 Peon/kPa x1500
29.53 0.00

28.20 0.09 20.40 0.48 8.19 0.85
28.06 0.10 17.83 0.58 6.84 0.88
27.00 0.15 15.03 0.67 5.11 0.92
25.46 0.23 12.97 0.72 4.11 0.94
24.33 0.30 12.88 0.72 3.04 0.96
22.93 0.38 9.68 0.81 0.07 1.00

 

104

 

 

 

~ .‘~ ‘A . . 4
25.0 ‘ ' .A ~ ~ ‘
m . 1
20.0 ‘L . ~ —
a ~A. ‘
Q 15 0 “A‘ ~
0. 1. _
10.0 ‘ 1 5
Q ‘ _,
. ‘ A
5.0 ', ‘ ,‘ _
. ‘4‘
00 C'1"'L"1'-1'-r-~--1-~r-g--¢-J--¢-.1. nh.L—A----L_‘i
0.0 0.2 0.4 0.6 0.8 1.0
x1,Y1

Figure 6.8 P-x of Diethyl Succinate (1) + Ethanol (2) at 50 °C
A: measured, dashed line is P—x-y representation of UNIQUAC.

6.3 - T-x-y apparatus and measurements

Isobaric VLE data of lactic acid + water, and lactic acid + ethanol + ethyl lactate
+ water were obtained using a Fischer still (model VLE 100D) with recirculation of vapor
phase, shown in Figures 6.10 and 6.11. The still operated with absolute pressure ranging
from 0.25 kPa to 0.4 MPa, and temperature up to 250 °C. The equilibrium temperature
was measured with resolution of 0.1 °K, with the temperature sensor positioned above the
Cottrell [101] pump. Pressure resolution was 0.01 kPa. Pure water, acetone and ethanol
were used to calibrate the pressure and temperature sensors.

Heating was regulated to maintain a mean recirculation speed of 30 drops per
minute. Mixtures were equilibrated for at least 12 hours to ensure the equilibrium was

reached, before each 0.5 ml sample was taken from condensed vapor and liquid for GC

105

 

and/or HPLC analyses. The equilibrium state was indicated by a constant pressure and

temperature of the system.

 

Figure 6.9 Overview of Fischer Recirculating Still

106

 

     
 
  
  

equilibrium temperature probe

 

 

 

 

 

\\\\\\\,‘
\\\\\\

 

 

 

 

reboiler temperature probe

 

 

 

vapor-phase sampling port

 

liquid-phase sampling port heating bulb

stirrer plate discard valve

Figure 6.10 Schematic of Fischer Recirculating Still

107

 

6.3.1 — T-x-y Data of Lactic Acid + Water System

For the system lactic acid + water, data were collected at 103.33 kPa (Table 6-4).
Lactic acid oligomers were quantified using a Hewlett-Packard 1090 Liquid
Chromatograph, equipped with UV detection (Hitachi L400H) at a wavelength of 210
nm. The mobile phase was acetonitrile (ACN) + water in a gradient mode (0% ACN
(t=0) to 60% ACN (t=20 min) to 90% ACN (t=25 min) to 0% ACN (t=28 min) at 1.0
ml/min. The Novapak C18 column (3.9 mm x 150 mm) was used and both ACN and
water are acidified by 2 ml of 85% (w/v) phosphoric acid in one 1 L of solvent, equivalent

to a pH=1.3. Complete details of the HPLC analysis are referred to section 5 . 1.

Table 6-4 T-x-y Data of Lactic acid (1) + Water (2) at 103.33 kPa

Liquid molar composition

 

TlK water LA1 LA2 LA3 LA4

378.25 7.9E-01 1.9E-01 1.5E-02 1.2E-03 1.0E-04
379.25 8.0E-01 1.8E-01 1.4E-02 LIE-03 8713-05
380.25 7.3E-01 2.4E-01 2.4E-02 2.3E-03 2.3E-04
380.75 7.2E-01 2.5E-01 2.5E-02 2.5E-03 2.5E-04
381.75 6.9E-01 2.8E-01 3.0E-02 3.3E-03 3.7E-04
381.85 7.1E-01 2.6E-01 2.7E-02 2.7E-03 3.0E-04
383.35 6813-01 2813-01 3.2E-02 3.5E-03 4.0E-04
3 87.3 5 5 .8E-01 3 .6E-01 5 .4E-02 8.05-03 1.2E-03
391.65 5.3E-01 3.9E-01 6.6E-02 LIE-02 1.9E-03
399.85 4.2E-01 4.5E-01 1.0E-01 2.3E-02 5.1E-03
402.25 4.6E-01 4.3E-01 8.7E-02 1.8E-02 3.6E-03
404.05 4.6E-01 4.3E-01 8.8E-02 1.8E-02 3.6E-03
409.15 4213-01 4513-01 1013-01 2.3E-02 5.3E-03

 

108

Table 6-4

Vapor molar composition

T-x-y Data of Lactic acid "H Water (2) at 103.33 kPa (continued)

 

TIK water LA1 LA2 LA3 LA.,
378.25 1.0E+OO 5.1E-04 2.9E-07 1.7E-10 9.6E-14
379.25 1.0E+OO 7.2E-04 5.8E-07 4.7E-10 3.8E-13
380.25 1.0E+OO 1.1E-03 1.5E-06 1.9E-O9 2.4E-12
380.75 1.0E+00 8.3E-04 7.8E-07 7.4E-10 6.9E-13
381.75 1.0E+OO 9.7E-04 1.0E-06 1.1E-O9 1.2E-12
381 .85 1 .OE+OO 1 .9E-O3 1.0E-O4 0.0E+00 0.0E+00
383.35 1 .0E+00 2.4E-03 1 .0E-O4 0.0E+OO 0.0E+OO
387.35 1 .0E+OO 3 .OE-O3 1 .0E-04 0.0E+OO 0.0E+00
391.65 9.9E-01 1.2E-02 5.0E-04 0.0E+00 0.0E+OO
399.85 9.8E-Ol 2. 1 E02 9.0E-04 0.0E+00 0.0E+00
402.25 9.8E-01 2.2E-02 9.0E-O4 0.0E+00 0.0E+00
404.05 9.8E-01 1.5E-02 6.0E-04 0.0E+OO 0.0E+OO
409.1 5 9.7E-Ol 3 .OE-02 1 .3E-03 1.0E-O4 0.0E+00

 

Figure 6.11 shows the measured data in this work and literature values, reported
by Sans et al. [102]. As discussed in chapter 5, the amounts of oligomers are significant
in concentrated lactic acid solutions. Sans reports lactic acid compositions in terms of
only monomers and dimers so the data do not agree with this work at high concentration
of lactic acid when plotted vs. mole fraction of lactic acid monomer. However, when the
liquid and vapor molar compositions from Sans data are recalculated using the oligomer
distribution model discussed earlier, the new values are in excellent agreement with the
measured values listed in table 6-4. The following is an example of conversion:

From the reported data [102], at T = 438.13 °K, water (1): x1 = 0.1940, monomer

(2): x2 = 0.6202, dimer (3): x3 = 0.1858. Using molecular weights of the species:

Superficial wt. % LA = (90x2 +162x3) / (18x1 + 90x; +162x3) = 96.09

109

122,, = (90x2 +162x3) / 90 = 0.9546
and r1"... = 0.1940

Using Eqn. (17) [103] from the oligomer distribution model:

12' +n’ r
K=r(—l—’A—2)=0.2023
"121 (1_r

Solving results in r = 0.2505. Applying Eqn. (22) [103] to calculate for the true

wt. % LA j, then
wt.% LA, = 53.98, wt.% LA2 = 24.34, wt.% LA3 = 8.81, wt.% LA4 =2.88,
wt.% LA 5 =0.89, and the true wt. % of water = 8.72.
MwLA, = 90, MW“; = 162, MwLA3 = 234, MwLA4 = 306
Because LA5 content is small, assuming the amount of all oligomers higher than

LA4 is negligible. The true liquid molar compositions are:

true wt.% of water

 

 

 

 

 

water (x1) = 18 .=4 0 = 0.38
true wt.% of water 12" wt. ALA}
1 8 j=l MWLAJ-
LA ( ) MWLAJ-
. x . :
J 1+1 '=4
true wt.% of water 1:: Wt'%LAJ'
18 Il/IwL A

LA, (x2) = 0.47, LA2 (x3) = 0.12, LA3 (x4) = 0.029, LA, (x5) = 0.007

The adjusted monomer concentrations of Sanz et al. are shown in Figure 6.11, which are

in good agreement with the measurements presented here, and are plotted along with the

published monomer concentrations and T-x-y data.

110

450

 

 

 

 

 

430 -- -
65>
9% 410 -- Q o 0 ~
p. 0
684°
390 -
O
370 . ‘ t 1 i ' 1 ‘ 1 ‘ a
0.0 0.1 0.2 0.3 0.4 0.5 0.6
X LA1, .V LA1

Figure 6.11 T-x-y of Lactic acid (1) + Water (2) at 103.33 kPa as a Function of
Monomer Concentration
Lines: the true LA1 compositions from Sanz after correction as described above. Solid
symbols: measured in this work; open symbols: original data reported by Sanz [102].

6.3.2 — T-x-y Data of Lactic acid + Ethanol + Ethyl lactate + Water System

The Fischer apparatus was also used to obtain isobaric VLE data of lactic acid
oligomers (LA1, LA2, LA3, LA4) + ethanol + ethyl lactate oligomers (ElLA, EzLA,
E3LA) + water systems. Ethanol and water contents were determined from Gas
Chromatograph (GC Varian 3400, Poropak Q 50/80, 6 ft long x 1/8 in. OD column).
Acetonitrile was used as an internal standard. Oligomers of lactic acid and esters were
quantified by HPLC.

ASPEN was used to fit the experimental data. Lactic acid tetramer (LA4), ethyl
lactate dimers (EzLA), and ethyl lactate trimers (E3LA) were manually entered into the

ASPEN data bank, using connectivity of atoms and UNIFAC functional groups. For

111

example, EzLA is registered as C3H1405, having three-group 1015 (CH3—), one-group
1010 (>CH2), one-group 1005 (>CH-), one-group 1200 (-OH), and two-group 3300 (-
COO-). It was noticed that ASPEN assigns the same UNIFAC group for primary,
secondary and tertiary hydroxyl groups. There was also a concern that large errors could
be involved in estimating vapor pressures of lactic acid dimer (LA2) and trimer (LA3),
using ASPEN parameters. As shown in Table 6-5 and Figure 6, ASPEN expected vapor
pressure of lactic acid trimers about 13 times higher than that of dimer at high

temperature, but there was no significant difference between dimer and monomer (LA1).

Table 6-5 ASPEN Parameters of the Extended Antoine Equation.

 

 

aLA1 ”LA1 aLAz bbLA2 31.113 bbLA3
c1 218.2822 214.9990 94.2322 216.0467 336.2222 213.4113
C2 -18757 -17489 -11976 -17489 -30565 -17489
C3 0 0 0 0 0 0
C4 0 0 0 0 0 0
C5 -28.816 -28.816 -10.528 -28.816 -44.817 -28.816
C6 1.3OE-05 13015-05 5.07E-18 1301-3-05 1531-3-05 13013-05
C7 2 2 6 2 2 2
C8 289.9 373.15 385.7 385.65 312.2 312.2
C9 675.0 438.13 660.0 660 777.0 777.0

 

a

: Retrieved from ASPEN ver. 12.1, b: from fitting the predicted Antoine coefficient, provided
by Sans, b”: C2-7 are held the same as parameters in bLA1 (the second column), Cl is adjusted to

decrease the vapor pressure of each oligomer by a factor of 10.

 

lnPZCI'l-TSZC, +C4*T+C5*ln(T)+C6*TC7 for C8<C9 (6.1)
3

where P is calculated in kPa and T is in Kelvin.

112

 

20.0

10.0 7

0.0 7

In (PIKPa)
E

-10.0 7 LA

-20.0 7
LA:

 

 

 

.30.0 1 r T 1
0.0002 0.0003 0.0004 0.0005 0.0006 0.0007

(TIK)‘1'3
Figure 6.12 ASPEN Predicted P’“of Lactic, di-Lactic, and tri-Lactic Acids

In this work, the Antoine coefficients from the PRO II process simulation
database, provided by Sanz et al. [102], were used to refit the extended Antoine vapor
pressure equation coefficients for the monomer in ASPEN. The predicted vapor
pressures using the fitted ASPEN equation were about 16 % lower than the predicted
values using the PRO II coefficients. A fit of the Aspen vapor pressure equation to a
curve generated from the PRO II Antoine coefficients provided VLE calculations more
consistent with measurements reported by Sans et al. for lactic acid + water systems than
use of the original ASPEN vapor pressure coefficients. From the extended Antoine
coefficients of the monomer, the C1 values were adjusted for LA2, LA3, and LA4
assuming vapor pressure of each oligomer decreased by a factor of 10, shown in table 6-
5. This ratio is similar to the ASPEN ratio for lactic acid trimer and dimer.

For the ethyl lactate oligomers, parameters were arbitrarily selected to provide

low vapor pressures. The molecular area (q) and volume (r) used in the UNIQUAC

113

activity model were calculated using Bondi’s method. The binary interaction parameters,

by- and bji, of the corresponding compound (i) with water (1') in Eqn: Tij- = exp(b,-j / T) were

obtained by equating the experimental and calculated bubble temperatures. In each case,

the liquid composition was specified and the bubble T and vapor composition were

predicted.

Table 6-6 Parameters used in ASPEN Prediction

 

 

 

LA1 LA2 LA3 LA4 E1LA EzLA E3LA
C1 214.999 212.696 210.394 208.091 71.866 10.472 10.334
C2 - l 7489 -17489 -17489 -17489 -671 5 -43 82.7 -4852.5
C3 0 0 0 0 0 -58.88 -138.1
C4 0 0 0 0 0
C5 -28.816 -28.816 -28.816 -28.816 -9.567
C6 1.3E-05 1.3E-05 1.3E-05 1.3E-05 0.0145
C7 2 2 2 2 1
C8 373.15 373.15 373.15 373.15 247
C9 438.13 438.13 438.13 438.13 588
q 2.884 4.84 6.796 8.752 10.708 3.928 5.884
I' 3.179 5.529 7.879 10.229 12.579 4.456 6.806
bl] -100 -200 -300 -400 279.92 70.81 22.43
1311 135 270 405 540 55.588 -202.36 -l48.76

 

C1-C7: parameters of the extended Antoine equation (Eqn. 6.1) where T is in

Kelvin, P is calculated in kPa for LA], LA2, LA3, LA., and ElLA. For E2LA and E3LA,

P is calculated in bar. q: molecular area, r: molecular volume. bij and bji: interaction

parameters, i: the corresponding compound, j: water.

114

Table 6-7 Measurement and Prediction of LAs + Ethanol + ELAs + Water Systems

 

 

 

 

 

 

 

 

 

 

Experiment 1 Experiment 2
Measured Prediction Measured Prediction

Liquid Vapor Liquid Vapor Liquid Vapor Liquid Vapor
LA1 0.0092 0.0010 0.0093 0.0001 0.0042 0.0023 0.0042 0.0001
LA2 0.0080 0.0000 0.0080 0.0000 0.0082 0.0064 0.0082 0.0000
LA3 0.0053 0.0000 0.0054 0.0000 0.0009 0.0000 0.0009 0.0000
LA4 0.0018 0.0000 0.0018 0.0000 0.0000 0.0000 0.0000 0.0000
E1LA 0.6575 0.1592 0.6599 0.2139 0.8677 0.4956 0.8677 0.4695
EzLA 0.1104 0.0000 0.1108 0.0041 0.0415 0.0000 0.0415 0.0035
E3LA 0.0221 0.0000 0.0222 0.0000 0.0028 0.0000 0.0028 0.0000
Water 0.0088 0.0198 0.0088 0.0303 0.0048 0.0162 0.0048 0.0312
Ethanol 0.1731 0.8199 0.1738 0.7516 0.0698 0.4795 0.0698 0.4957
T(K) 374.3 374.3 373.2 373.2 395.2 395.2 389.9 389.9
P(KPaL 98.7 98.7 98.7 98.7 98.7 98.7 98.7 98.7

Experiment 3 Experiment 4
Measured Prediction Measured Prediction

Liquid Vapor Liquid Vapor Liquid Vapor Liquid Vapor
LA1 0.0605 0.0002 0.0605 0.0002 0.0488 0.0003 0.0488 0.0002
LA2 0.0180 0.0011 0.0180 0.0000 0.0145 0.0000 0.0145 0.0000
LA3 0.0045 0.0001 0.0046 0.0000 0.0039 0.0000 0.0039 0.0000
LA., 0.0018 0.0007 0.0018 0.0000 0.0005 0.0000 0.0005 0.0000
E1LA 0.4784 0.1045 0.4784 0.0850 0.6683 0.1873 0.6683 0.1796
E2LA 0.0421 0.0001 0.0421 0.0005 0.0128 0.0000 0.0128 0.0002
E3LA 0.0071 0.0004 0.0071 0.0000 0.0012 0.0000 0.0012 0.0000
Water 0.0501 0.0923 0.0501 0.0876 0.0662 0.1944 0.0662 0.1730
Ethanol 0.3375 0.8006 0.3375 0.8267 0.1838 0.6179 0.1838 0.6470
T(K) 358.9 358.9 361.2 361.2 368.1 368.1 367.8 367.8
P(KPa) 99.0 99.0 99.0 99.0 99.2 99.2 99.2 99.2

 

115

 

Table 6-7 (continued)

 

 

 

 

 

 

 

 

 

 

Experiment 5 Experiment 6
Measured Prediction Measured Prediction

Liquid Vapor Liquid Vapor Liquid Vapor Liquid Vapor
LA1 0.0095 0.0004 0.0095 0.0001 0.0396 0.0021 0.0396 0.0007
LA2 0.0006 0.0000 0.0006 0.0000 0.0009 0.0000 0.0009 0.0000
LA3 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
LA4 0.0000 0.0000 0.0000 0.0000 0.0023 0.0000 0.0023 0.0000
E1LA 0.8426 0.2711 0.8426 0.3243 0.8221 0.4338 0.8221 0.4939
EzLA 0.0078 0.0000 0.0078 0.0003 0.0555 0.0000 0.0555 0.0038
E3LA 0.0000 0.0000 0.0000 0.0000 0.0048 0.0000 0.0048 0.0000
Water 0.0417 0.1974 0.0417 0.1820 0.0396 0.2659 0.0396 0.2495
Ethanol 0.0978 0.5311 0.0978 0.4933 0.0350 0.2982 0.0350 0.2521
T(K) 374.6 374.6 375.6 375.6 384.7 384.7 384.9 384.9

P(KPa) 99.0 99.0 99.0 99.0 98.4 98.4 98.4 98.4

Experiment 7 Experiment 8
Measured Prediction Measured Prediction

Liquid Vapor Liquid Vapor Liquid Vapor Liquid quor
LA1 0.0406 0.0032 0.0406 0.0009 0.0042 0.0121 0.0042 0.0001
LA2 0.0087 0.0074 0.0087 0.0000 0.0234 0.0058 0.0234 0.0000
LA3 0.0000 0.0000 0.0000 0.0000 0.0000 0.0004 0.0000 0.0000
LA4 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
E1LA 0.8470 0.4990 0.8470 0.5682 0.9157 0.4894 0.9157 0.5820
E2LA 0.0481 0.0000 0.0481 0.0041 0.0021 0.0104 0.0021 0.0002
E3LA 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Water 0.0298 0.2333 0.0298 0.2182 0.0288 0.2235 0.0288 0.2149
Ethanol 0.0258 0.2571 0.0258 0.2085 0.0258 0.2583 0.0258 0.2029
T(K) 388.2 388.2 361.4 361.4 387.2 387.2 388.4 388.4

FjKPg) 97.3 97.3 97.3 97.3 97.6 97.6 97.6 97.6

 

116

 

6.3.3 - Limitation of T-x-y Fischer Still

The design of the Cottrell coil is excellent in eliminating liquid entrained to
equilibrium vapor in measuring saturated vapor pressure of pure components. However,
the Cottrell coil is limited to viscous mixtures, in which the heavy components fall back
to the reboiler due to gravity. The T-x-y Fisher apparatus does not provide an adequate
mixing and superheats the large portion mixture between the tip of the heating bulb and
the Cottrell pump. Before the isobaric measurements of the systems involved with lactic
acid were taken, vapor pressures of HPLC water, pure acetone, re-distilled absolute
ethanol, and re-distilled ethyl lactate were measured to verify the reliability of the T-x-y
Fischer still. All the measured Psalt were in excellent agreement with the reported values
in literature. However, it was noticed that the reboiler temperature (Teolumn) and
equilibrium temperature (Thead) were the same in experiments with acetone and ethanol,
and Thead was about 1 °C below Tcolumn in experiments with water and ethyl lactate. These
temperatures were read from the probes at locations shown in Figure 6.10. The
difference between Thead and Tcolumn became significantly large in experiments with lactic
acid + water, especially when the mixtures were highly viscous, for example, Thcad = 128
0C, Tcolumn = 162 oC, and Thead = 135 oC, Tcolumn = 166 °C. The following calculations
were used to determine whether the lactic acid + water mixtures were overheated, causing

the differences between Thead and Tcolumn,

Energy balance for adiabatic flashes:
nVHV+nLHL =ninHi" (6.2)

If the mixture is overheated, then Eqn 6.1 becomes,

117

 

"V Hsat,V + ”L HSCIIJ. = nL,in(Hsat,L + CPAT) + "VJ" (Hsaty + CPAT) (63)

Re-arranging, Eqn. 6.2 becomes,

fliHsatJ’ +fiHsaI.L :nLJn Hsat,L +nl’,in Hsat,V +CpAT (6.4)
n7 nT nT nT
where nT = nV + nL = nV’i” + nL’i" . Let reference state = saturated liquid, then
V ”V ___nV,in
Hm” T = CpAT (6.5)
Cp = a+bT+cT2 -l-dT3 +eT‘1 J/Kmol (6.6)
. 2
AH"“P = A(1-T,) B+CTr+DTr J/Kmol (6.7)

For water: a = 276370, b = 2091.1, c = 8.125, d = 0.01412, e = 9.37 x 10*, and
A = 52.053 x 103,13 = 0.3199, C = -0212 , D = 0.25795, and T, = 647.096 °K.

At 409. 15 °K (the last data point in Table 6-4):

Cp(water) = 7.7 x 1051061161, AH”“P(water) = 3.9 x 107 J/Kmol

If all water in the feed was overheated: AT = 51 0K

118

Chapter 7 — Vapor Pressure Prediction using SPEAD

7.1 — Overview

Predicting vapor pressures of the acetals, 5-hydroxy-2-methyl-l,3-dioxane and 4-
hydroxymethyl-2-methyl-l ,3-dioxolane, and ethyl lactate oligomers for process design of
reactive distillations, using SPEAD is an objective in this study. These compounds are
involved in the current research at MSU, but they are not available in pure state for direct
VLE measurements. Like ethyl lactate, the above acetals have low toxicity and low odor
and can be excellent solvents for a wide variety of applications in the pharmaceutical and
cosmetic industries [104]. They can also be used as additives in diesel fuel [105].

Vapor pressures of an ethyl lactate oligomer mixture were measured to compare
with the SPEAD prediction. This prediction was accomplished using a FORTRAN code,
written in this study. The predicted vapor pressures of ethyl lactate were in the same
order with the measurements. The normal boiling points for the acetals were predicted

within 2 % of the reported values, which are the only information available in literature.

7.2 — SPEAD Simulation Procedure

Prior to simulations, molecules were sketched using ChemSketch software and
relaxed to their lowest internal energy [106]. This generated a three dimensional
molecular file (.mol) for each studied molecule, in which the coordination and
connectivity of all atoms were specified. The molecular information in .mol files were
then carefully transferred into molecular dynamic (.3md) files in SPEAD. As described
in chapter 4, molecules are broken down into a number of interaction sites in SPEAD

simulation. Each interaction site of the simulated molecule is assigned by the repulsive

119

diameter (bond diameter) of the site, followed by an initial set of spatial coordinates
(group—coordinates), then followed by a three or four digit index (group ID) specifying
the type of site. Among the described properties of interaction sites, the group ID
identifies parameters used to correlate the potential well depth of the interactions in
computing the Helmholtz energy after simulation. Therefore, it does not affect the count
of number of interactions in wells during simulation. The .3md file also contains the
bond array showing which potential wells these bonded sites occupy. Sites directly
bonded to each other have a value of 2, sites two bonds apart have a value of 4, and sites
that are three or more bonds apart have a value of 6 [107]. Layout of the .3md files and
complete molecular structures of ethyl lactate oligomers are provided in Appendix C.

To this point, a simulation could be started by choosing Simulate Molecular
Properties from the Commands panel of the SPEAD interface. From the selected .3md
file, the necessary information for a simulation such as the size of the simulated box
corresponding to number of molecules, the effective volume of molecule, time step, etc.,
was created. In this dissertation, all simulations were performed at 5 us for 100
molecules, if not otherwise specified.

After simulation, equation of state was generated and vapor pressure of the
interested molecule was calculated. Procedures for these calculations are referred to the

SPEAD Introduction and Help [78].

7.2.1 - Pair Interaction Sites of the Interested Compounds
The SPEAD interaction sites of ethyl lactate oligomers and acetals were
designated as described in Figure 7.1. Each interaction site was specified by a three or

four digit index, identifying the main and sub groups. For example, the site 1602 was

120

made of the main group 16 and the sub group 2.

 

 

 

 

 

0 CH3
1602 102
H30 L104 11101 0 CH3
102\/ \ O/E \/ 15$/ 101
303 904 n 201
1502 .I
OH
1404 1602
Ethyl Lactate Oligomers
O o 209
1504 209 1504
H3C 301 H3C 301 301——oH
102 \ 102 1404
/ 301 CH
O 1402 O 209
1504 201 1504
4 -hydroxymethyl -2 -methyl -1 , 3 -di0xolane 5 -hydroxy-2-methyl- I , 3 -di0xane

Figure 7.1 The Interaction Sites of Ethyl Lactate Oligomers and Acetals

At present, SPEAD has not been fully developed for the multi-oxygen-
containing molecules such as ethyl lactate or acetals. Therefore, the site 1404 for
secondary —OH group and sites 904, 1502, 1602 for ester groups in Figure 7.1 are not yet
parametized. Finding the optimal parameters for these sites was crucial for reliably

predicting vapor pressures of acetals and oligomers.

7.2.2 - Approach of Optimizing Secondary -OH and -COO- groups
SPEAD was developed with the premise that parameters are transferable within

the homologous compounds. Therefore, the best parameters for secondary —OH and —

121

COO— groups (shown in Figure 7.2) could be obtained from fitting the good experimental
Psat data available in the DIPPR database for 2-a1kanols and esters. A datum was
considered good if it had the DIPPR notation “acceptance” and the DIPPR “deviation” of

less than 5 %.

H C R O
3 \ /R 1\ /1502\
C R2
“04
OH O
1404 1602

Figure 7.2 Group Indices in 2-Alkanols and Base Esters used in Optimization

7.2.3 - Mathematical Methodology

The wells for each group 904, 1404, 1502, and 1602 were Characterized by two
parameters which were the inner (8,) and outer (8,) interaction well depth. In addition,

group 1404 (-OH) and 1602 (>CO=) formed hydrogen bonds, which were described by
the three parameters: the energy (eHb), the volume (BondVoI), and the rate (BondRate) of
the bonds. As a result, the optimization of secondary -OH and ester groups involved
either five or nine parameters, respectively.

The following Eqns (7.1 — 7.4) describe the correlation of hydrogen bonding

parameters used in calculation of total Helmholtz energy:

 

A ’ Al's A0 Al Al
———=—+—+—+A“”0" (7.1)
RT RT T T2
AUSSOC A A
=2lnX +1-X (7.2)
1 11 1
XA =—1+\/2;+ 4a (7.3)

122

 

a = p————KAD [exp(B£AD)— I] (7.4)

where p is the molar density, n the packing fraction, KAD the molar hydrogen bonding
volume, CAD the bonding energy, k3 is the Boltzmann’s constant and ,6 =1/kBT.
SPEAD parameter file ParmsHb3.txt denotes SAD as eHb, and BondVol as BondV

and BondRate as Bond VolSlo [108]. The correlation of Bond Vol and BondRate is defined

in Eqn 7.5 below:

 

"sites total

2
n — . .
K A D = BondVol[1+BondRate( h b°"d’"g 3”“) J (7.5)

As stated by Korsten [64] and also observed, the logarithm of vapor pressure of
any compound is linear to T”. Therefore, a good prediction of PM for a series of
homologous compounds must have a minimum error in both PM" and slope of the ln(P“")

with respect to T13. Figure 7 .3 below illustrates the possible errors in prediction.

123

 

 

 

 

 

 

 

 

 

/\ /\
(a) (b)
-. \\\ \
"'... \
InPSalJ \\\\\PredlCted Inpsafl \\ M d
\\ \ easure
. \ \
Measured“. \\ \\
\\ Predicted \
\\
\
\
T.1'3 > T-1.3 >
/N /\
(C) (d)
\ a. “
\ Predicted \" Measured
33"“\\ P d-ct d\\
\ re 1 e \
Measured ‘\ \\
\\ \ "a
\\ \\
\
\
T-1.3 I T'1-3 I
Figure 7.3 Illustration of Error in Prediction of P“t

SPEAD developers used grid search, simplex and recursive random search [109]
algorithms for parameterization of hydrocarbons and series of simple homologous
compounds. But, these methods were not successful in finding a global optimum for a
system with hydrogen bondings.

This dissertation provided a FORTRAN compiler using the routine DBCONF
from the International Mathematical and Statistical Library (IMSL) to optimize the five

and nine parameters of secondary -OH and ester groups.

124

To minimize the errors described in Figure 7.3, the objective function (f -—> min)

was defined as follows:

 

 

f = fl * f2 (7-6)
i=n
= 2 01980" sz’re- In 853.1,. ) (7.7)
i=1
_ 1 1:" abs (In 131qu",— —ln gfgte) _ (1n Pfiflfexr -ln Hféflp) (7.8)
_ n-171-3 -1 3 -1.3
'=1 (T1413 - Ti ) (71H - Ti )

where n was the number of data points, Pisg’re and 13%,,

are the respective predicted and
experimental vapor pressures for a datum point i. The function fl is created to measure

the absolute error in Psat, while f2 measures the error in the slope.

The DBCONF routine algorithm - DBCONF uses a popular variant of the Quasi-
Newton method, which is called the BFGS (Broyden-Fletcher-Goldfarb—Shanno) method
and an active set strategy to solve a nonlinear optimization problem subject to simple
bounds on the variables [110-113]. The algorithm can be summarized as follows:

An active set A containing the indices of the variables at their bounds is built
from a given starting point 36(0) and an estimate of Hessian matrix H0 =V2 f (xw) )

The routine then computes the search direction for the “free variables”, which is not in

the active set according to the formula:

x<k+1> = x”) — Hk‘lVf(x(k)) (7.9)
50‘) = x(k+l) —x(k) (7.10)
y(k) = Vf(x(k:l))—Vf(x(k)) (7.11)

125

T T
Hksu) (300) Hz. + y(k) (Jr/0)
31k) . HkSUc) y(10.501)

 

Hk+l =Hk — (7-12)

The active set is changed only when a free variable hits its bounds during iteration
or the optimality condition is met for the free variables but not for all variables in A, the
active set. In the latter case, a variable that violates the optimality condition will be
dropped out of A.

More details on the DBCONF algorithm can be found in the IMSL
documentation. The quasi-Newton method and line search are explained by Dennis and
Schnabel [114], and the active set strategy is explained by Gill and Murray [115]. A
c0py of FORTRAN code to call DBCONF and sample of input and output data files are

included in Appendix B.

7.3 - Results of Optimization of the 2nd -OH and -COO- Groups

Existing data were divided into two sets. Some were used for parameter fitting
and made up the training set. The other data were used for evaluation of predictive
capability and made up the validation set. The training and testing sets, and results of
optimization to obtain parameters for the secondary -OH and —COO- interaction sites are
summarized in Table 7-1. More details of the output files containing experimental and

predicted vapor pressures, generated by the FORTRAN program are in Appendix C.

126

Table 7-1 Optimization and Validation of -OH and -COO— Sites

 

 

 

 

 

 

 

 

 

 

Compound Name Notation # data Deviation in Prediction References
pomts o Bias Max
Training -OH site
2-propanol 20IC3 33 6 4.1 15.7 [116-118]
2-butanol 20IC4 32 11.3 -11.3 -14.4 [119, 120]
2-pentanol 20105 33 3.3 -2 -6.7 [121, 122]
2-hexanol 20lC6 27 5.5 4.6 15.6 [122, 123]
2-octanol 20108 33 5.3 3.9 16.6 [122, 124]
2-nonanol 20IC9 2 3 2.7 5.7 [125, 126]
Testing-OH site
2-heptanol 20|C7 9 4.9 -4.9 -7 [127]
3—pentanol 30105 24 24.1 -24.1 -29.8 [122, 128]
3-hexanol 30106 22 12 -11.7 -32.2 [122, 127, 128]
3-heptanol 3oIC7 6 8 -8 -9.9 [129, 130]
cyclohexanol CZOICS 33 29.4 29.4 49.9 [117, 131]
cis 2-methylcyclohexanol cZol_2_C1C6 3 21.9 21.9 30.9 [132, 133]
cis 4-methylcyclohexanol c20|_4_C106 2 28.4 28.4 41.7 [134, 135]
2,3—butanediol diolC4 22 79.4 -79.4 -90.9 [136]
Training -COO- site
ethyl propionate CBateCZ 28 5.2 3.3 9 [137]
n-butyl propionate CBateC4 32 2.1 0.7 -8.2 [121]
methyl n-butyrate C4ateC1 30 15.3 -14.9 -38.4 [121, 138]
ethyl n-butyrate C4ateCZ 9 6 -4.7 -24.1 [132]
n-propyl n-butyrate C4ateC3 28 1.5 -1.4 -3.5 [139, 140]
isobutyl isobutyrate iC4atelC4 17 16.7 16.7 22.2 [117, 141]
methyl decanoate C10ateC1 18 6.3 -6 -13.1 [142, 143]
Testinq-COO- site
n-propyl propionate C3ate03 3 5.2 -0.3 -8.2 [137]
n-butyl n-butyrate C4ateC4 2 13 —4.6 -17.6 [132]
n-propyl isobutyrate iC4ateC3 1 16 16 16 [144]
n-butyl valerate CSateC4 2 5.2 2.5 7.7 [144, 145]
ethyl isovalerate iCSateCZ 1 18.3 -18.3 -18.3 [132]
methyl Iaurate C12ateC1 14 8.7 -1.9 -16.7 [68]
isopropyl Iaurate C12ate|03 7 4.9 3.7 1 1 [68]
isobutyl Iaurate C12atelC4 11 7 -3.5 -10.9 [68]
2-ethyl hexyl Iaurate ClZateZCZCS 9 26.1 -26.1 -36.4 [68]
methyl tetracosanoate C24ateC1 6 26.7 -26.7 -41.8 [68]

 

127

The secondary -OH group — All secondary alcohol data from DIPPR were used
in optimization (2-alkanols (CZ-C9». The 2-heptanol was not included in the training
set, because its vapor pressures in DIPPR database are not experimental but smoothed

data. The average error (0') in fitting 160 data points of the training set is 6 %. Psat were

from 0.01 kPa to 1 MPa.

Parameters obtained from optimization of 2-alkanols were used for prediction of
vapor pressure for 3-alkanols and 2-heptanol. As shown in Table 7-1 predictions are in
good agreement with the reported values in literature. The errors are large for the 3-
pentanol and 3-hexanol, but Psat data of these compounds were measured at low
temperature (Psalt < 0.01 kPa), and they were not in the same range with data used in the
training set.

The parameters of 1404 group from 2-alkanols were also tested with cyclohexanol
and cyclomethylhexanol to verify if they could be transferable to the secondary OH
group, which bonded to a non-aromatic ring. The vapor pressures were overestimated;
cyclic alcohols have higher boiling points than the straight chain alcohols, affected by
their stronger intermolecular hydrogen bonds and the current version of SPEAD did not
represent the effects precisely.

The ester -C00- group - Optimization of the ester groups used 162 data points
as summarized in Table 7-1. Deviation of the fitting data is ~ 8 % of the measured
values. Experimental Psat were limited, therefore the validation to check for
transferability of the obtainable parameters only included 56 data points. Results showed
that vapor pressure of esters containing up to 30 carbons could be predicted within 27 %

of the measured values, using parameters listed in Table 7-2.

128

Table 7-2 Parameters used in SPEAD Calculated P“It

 

Potential Well

 

 

Site Description Hydrogen Bonding
Depth
Bond Bond
8, a, Bond Vol Rate Energy
101 —CH3a 91.871 16.445
102 -CH3b 55.100 32.400
106 —CH3f 108.000 11.000
201 —CH2— 26.558 21.827
209 —CH2— in a ring 30.000 21.000

301 >CH— to a Carbon 7.100 6.946
303 >CH— to the 2nd —OH 31.500 4.400

* 1504 Cyclic ether —0— 140.25 23.65
*904 =C— 10.209 1.698
*1404 2"d —OH 142.743 41.760 0.00003587 140.00 4.247
* 1502 Ester —0— 100.198 4.087
*1602 =0 152.632 44.705 0.002 104.65 0.682

 

Sites with * are optimized in this study.

7.4 - Prediction of Psat for Ethyl Lactate and Methyl Lactate
First, vapor pressure of methyl lactate and ethyl lactate were predicted to compare
with experimental values. Results (Table 7-3) showed that SPEAD could not provide an

adequate prediction for ethyl and methyl lactates using the above-optimized parameters.

7.4.1 - Effect of Intramolecular H-bonds in Lactates

As shown in Table 7-3, all methods in DIPPR except for Othmer-Yu also
underestimated the lactates. These compounds containing both a secondary hydroxyl and
an ester group in their molecules, can form intramolecular hydrogen bonds (—
OH....O=C<). To verify whether the intramolecular hydrogen bonding could be the

cause of underestimation in SPEAD, full atom liquid simulations of 50 ns were

129

conducted using COMPASS potentials in the NVT ensemble with the Anderson
thermostat, provided by Accelrys MS Modeling 4.0. Vibrational and torsional energies
were included. Results indicated that liquid phase hydrogen bonds were both intra— and

intermolecular.

Table 7-3 Predicted P"it of Methyl Lactate and Ethyl Lactate

 

Methyl lactate

 

 

 

 

 

 

 

 

 

 

 

 

 

T = 313.15 K T = 333.25 K T = 353.35 K

Method P = 0.0012 MPa P = 0.0036 MPa P = 0.0094 MPa

Value % Dev Value % Dev Value % Dev
Riedel 0.00057 -52% 0.00213 -41% 0.00662 -41%
Othmer-Yu 0.00945 689% 0.02998 731% 0.08191 731%
Gomez-Thodos 0.00027 -78% 0.00129 -64% 0.00483 -64%
Lee-Kesler 0.00054 -55% 0.00204 -43% 0.00644 -43%
Maxwell-Bonnell 0.00182 52% 0.00501 39% 0.01 197 39%
aSPEAD 0.00003 -97% 0.00015 -96% 0.00058 -94%
bSPEAD 0.001 10 -8% 0.00340 -6% 0.00900 —6%

Ethyl lactate
T =313.15 K T = 333.25 K T = 353.35 K

Method P = 0.0012 MPa P = 0.0031 MPa P = 0.0076 MPa

Value % Dev Value % Dev Value % Dev
Riedel 0.00034 -71% 0.00133 -57% 0.00431 43%
Othmer-Yu 0.00610 430% 0.01997 538% 0.05607 638%
Gomez-Thodos 0.00012 -89% 0.00067 ~79% 0.00277 -64%
Lee-Kesler 0.00032 -73% 0.00127 -60% 0.00417 -45%
Maxwell-Bonnell 0.00098 -15% 0.00295 -6% 0.00761 0.2%
aSPEAD 0.00003 -97% 0.00015 -95% 0.00060 -92%
bSPEAD 0.00080 -33% 0.00240 -23% 0.00650 -14%

 

“SPEAD: using parameters listed in Table 7-2. bSPEAD: same as ‘SPEAD , but the BondVol = 0,
BondRate = 0, and BondEnergy = 0 for site 1404. The experimental Psat of methyl lactate and ethyl
lactate were taken from the DIPPR database.

130

Adjusting for the effect of intramolecular H-bonds in calculating the Helmholtz
energy in SPEAD was beyond the scope of this study. However, it was found that
SPEAD gives a good prediction for the lactates (still a small underestimation) if all
parameters of H-bond in 1404 group were set to zero. Table 7-4 provides an example of
Psat predictions for ethyl lactate using the described settings.

The normal boiling point for ethyl lactate is 427.15K, and the predicted value was
428.02 K. Similar results are obtained for methyl lactate (Tb = 417.95 K, predicted value
= 418.45 Tb). For the methyl 3-hydroxy butanoate (at P = 0.00132 MPa, T = 336.2 K,

and predicted T= 330.2 K).

Table 7-4 Measured T-P and SPEAD Prediction for Ethyl Lactate
Using parameters listed in Table 7-2 (except for H-bonding)

 

 

 

 

 

P (MPa) P (MPa)
1 (K) T (K)

Measured SPEAD Measured SPEAD
353.15 0.0077 0.0065 317.45 0.0013 0.0010
351.15 0.0072 0.0059 314.65 0.0010 0.0008
349.55 0.0067 0.0055 313.25 0.0011 0.0008
347.15 0.0062 0.0049 312.85 0.0010 0.0007
343.55 0.0053 0.0041 310.45 0.0008 0.0006
341.65 0.0047 0.0037 309.15 0.0007 0.0006
337.45 0.0040 0.0030 308.15 0.0007 0.0006
333.15 0.0031 0.0024 305.65 0.0005 0.0005
330.65 0.0028 0.0021 304.05 0.0005 0.0004
327.35 0.0023 0.0017 303.15 0.0005 0.0004
325.05 0.0021 0.0015 300.05 0.0004 0.0003

 

7.4.2 - A Common Point for Ethyl Lactate Oligomers

Another evaluation of SPEAD predictions for oligomers was the use of Eqn 4.1,

131

which was described by Korsten in chapter 4. Vapor pressures were generated for each
lactate ester (ElLA, EzLA, E3LA, E4LA, E5LA) at temperatures between 300—700 °K
(increment of 20 0K), using the parameters listed in Table 7-2 (all parameters for H-bonds
of site 1404 were zero, but parameters of site 1602 were not adjusted, because H-bond
energy of this site is much lower than the bond energy of site 1404 and has a very minor
effect).

Results showed that the SPEAD-predicted vapor pressure curves of lactate
oligomers were not completely linear when plotted with Korsten’s temperature
dependence. However, fitting the predictions with linear equations, the extrapolated
vapor pressures for EzLA, E3LA, E4LA, and ESLA merged at the common point a (To =
4947K, Pa = 2643.3 MPa) as illustrated in Figure 7.4. Therefore, the predictions were
generally consistent with the empirical common point analysis of Korsten. The
coefficients shown in Eqns 7.1 and 7.2 for slopes of these vapor pressure curves were

obtained from regression using the least square method.

 

__ 1 1
lnP—lnPa+B W- 1.30 (Eqn 410)
7' Ta
B = 30(93)+ 31140-65 = 4669.09- 321.4203 —1162.8M0-65 (7.13)

(Ta = 4947.5 1K, Pa = 2643.3 MPa) , and 93 = 22.596

The above equations can be combined as:

 

_. _ - 0.65 1 _ 1
lnP—7.88 (2593.72+1162.8M )[TIBO 494751.30) (7.14)

where M is molecular weight of the corresponding ethyl lactate oligomer, T is in K, and P

is in MPa.

132

 

6.1

In (PIMPa)

-9-

 

 

 

 

0

0.0001

0.0002 0.0003 0.0004 0.0005 0.0006 0.0007
mm"-3
Figure 7.4 Trend of Predicted Vapor Pressure of Ethyl Lactate Oligomers
*: methyl lactate, O: methyl 3-hydroxy-butyrate, OI LlE, A: L2E, A: L3E, I: LE, 0: L5E

7.4.3 - A Validation for P“at Prediction of Ethyl Lactate Oligomers

Psat of the ethyl lactate oligomer mixture were measured using P-x-y apparatus,

which was described in chapter 6. The molar compositions were determined using

HPLC. Table 7-5 shows the predicted values using SPEAD regressed with the Korsten

correlation as explained above; the measured pressures are the same order of magnitude.

Table 7-5 Vapor Pressure of Ethyl lactate Oligomers Mixtures

Molar compositions: ElLA = 0.656, EzLA =0.282, E3LA = 0.043, E4LA = 0.019

 

 

 

P (kPa)
T(K)
measured PSLPEAD-Kw
298.0 0.40 0.08
301.2 0.47 0.11
303.7 0.51 0.13
306.2 0.56 0.15
310.9 0.68 0.21
314.7 0.81 0.28
318.5 0.51 0.35

 

 

 

T (K) P (kPa)
P measured PSPEAD-Korste_n_

320.7 1.06 0.41
326.6 1.33 0.59
330.7 1.66 0.76
335.1 1.93 0.98
338.2 2.18 1.17
342.1 2.54 1.45
345.8 2.94 1.77

 

133

Using Eqn 7.9 to estimate total vapor pressure of mixture at 345.8 K:

 

1 1
Pmonomer = exp|:7.88 — (2593.72 + 1162.8 * 1 1813065 )(3453130 — 4947.51.30 ]]
= 2.69 x 10'3MPa
P2,... =2.85x 10'5 MPa, P3,... = 5.48 x 10'7 MPa, P4,...r = 0.15 x 10'7 MPa
If mixture is an ideal solution, then
Pmixture = 0.656*(2.69) + 0.282*(0.028) + 0.043*(5.48 x 10“) + 0019*(015 x 10“)
= 1.77 kPa

Result is in the same order with the value from measurement.

7.5 — Prediction of Psat for Acetals

Acetals of interest were the 4-hydroxymethyl-2-methyl-l,3-dioxolane (4HMD)
and 5-hydroxy-2-methyl-1,3-dioxane (5HMD). These compounds contain two cyclic -0-
groups in a molecule. Different from alcohols and esters, the ether oxygen atom does not
form intramolecular H-bonds in ethers; therefore vapor pressures of mono-ethers (each
molecular containing a single -0- group), such as tetrahydrofuran, are much higher than
vapor pressures of alcohols, esters, di-ethers, and the above acetals 4HMD and 5HMD.

The existing SPEAD parameters (.8I = 287.4, a, = 26.7) for the cyclic-ether
oxygen (group 1504) provided an excellent Psat prediction for tetrahydrofuran. But,
using these existing parameters for 1,3-dioxane and 1,4-dioxane, SPEAD underestimated
vapor pressures by at least 85 %. In addition to ether oxygens, the 4HMD and 5HMD
also contain a hydroxyl group in their structures; therefore if the existing SPEAD

parameters were not sufficient for use in 1,3-dioxane and 1,4-dioxane, they were

134

obviously not suitable for the 4HMD and 5HMD. Thus, optimization was needed for
group 1504 assuming the methylene site in a ring (group 209) was already parametized.
Experimental Psat data are very limited for acetals. Table 7-6 lists the compounds
found in the DIPPR data bank that have the most similar structure to the 4HMD and
5HMD. The 1,3-dioxane and 1,-4 dioxane were used in optimization; the tn'oxane and

tetrafurfural alcohol were used in validation.

Table 7-6 Optimization and Validation of the Cyclic -0- Site

 

# data Deviation in Prediction References

 

Compound Name
Structure points 0 Bias Max

 

 

 

 

Training-O—site
0A0
1,3-dioxane V 15 4.7 4.7 11.7 [146-148]
/—\
1,4—dioxane 0. ,0 33 2 0 4.1 [141,149,150]
Testing -0- site
Trioxane f0"
. 0 o 11 3.7 3 6.8 [121,151]
( or tnoxymethylene) v
HO.\

Tetrahydrofurfuryl alcohol 0 J\ 20 15.4 -15.4 46.1 [152-154]

\_/

Since the ether group does not associate with H-bond, optimization of the 1504

 

group only involved two variables, the inner and outer well depths of the site. A minor
modification was made in the FORTRAN program for fitting. As discussed in the

previous sections, this program was written to optimize either five or nine parameters in

alcohol and ester groups. The best parameters for group 1504 were found to be 8, =

140.25, and €,= 23.65. Using these parameters, vapor pressures of trioxane and

135

tetrahydrofirrfuryl (testing compounds) were respectively predicted within 4 and 16 % of
the reported values in literature.

Table 7-7 summarizes the prediction of Psat for acetals 4HMD and 5HMD using
the new well depths for group 1504. There is currently no convergence in the smoothed
SPEAD calculation of compressibility factor Z at below 300 °K and above 500 °K for
both 4HMD and 5HMD, so the vapor pressures were evaluated only between these

temperatures.

Table 7-7 Prediction of Psat for Acetals using SPEAD. Tbmioxam) = 449.15 °K,
SPEAD value = 453.15 °K. Tudioxom) = 460.15 °K, SPEAD value = 467.96 °K [155].

 

 

 

 

 

P (kPa) P (kPa)
T (°K) T (°K)

5HMD 4HMD 5HMD 4HMD
300 0.038 0.017
310 0.085 0.040 410 22.9 13.7
320 0.181 0.088 420 33.5 20.3
330 0.365 0.183 430 47.9 29.6
340 0.699 0.360 440 67.0 42.0
350 1.30 0.68 450 92.0 58.5
360 2.30 1.20 460 124.1 80.0
370 3.80 2.10 470 164.6 107.4
380 6.20 3.50 480 214.9 142.0
390 9.90 5.70 490 276.7 185.0
400 5.30 9.00 500 351.6 237.6

 

*All parameters of H-bond in 1404 group were used as listed in Table 7-2.

As shown in Table 7-7, SPEAD predicted values are very close to the reported
boiling points, which are the only VLE data available in literature for 4HMD and 5HMD
[155]. The linear trend of predicted vapor pressure curves follows the Korsten
correlation. In addition, regression using the least square method shows vapor pressure

curves of these homologous isomers 4HMD and 5HMD (same molecular weight and

136

same functionality) merge at a common point 01 (Ta = 1024 °K, Pa = 126.7 MPa) and the

value of 03 is 33.49 in Eqn 7.13.

ln(PIMPa)

 

 

 
 
 
  
 

5HMD: y = ~31274x + 8.7291
R2 = 0.9998

 

 

'8 4H MD: y = -32630x + 8.7567
_1 0 R2 = 0.9998
-12 1 1 1
0.0003 0.0004 0.0005 0.0006
(TIK)'1'3

Figure 7.5 SPEAD Predicted Vapor Pressure of Acetals

O: 5HMD, I: 4HMD. Lines: linear regression

0.0007

Table 7-8 and Figure 7.6 are predicted VLE of 4HMD and 5HMD mixtures at

373.15 °K. As expected, SPEAD predicts 4HMD and 5HMD form ideal solutions. It

will be difficult to separate these acetals using distillation due to their small relative

volatility.

Table 7-8 SPEAD Predicted T-P-x-y of 4HMD (1) + 5HMD (2) at 373.15 °K

 

 

X1 Y1 P (MP3) X1 Y1 P (MP3)
0.0 0.000 0.0025

0.1 0.167 0.0027 0.6 0.729 0.0037

0.2 0.310 0.0029 0.7 0.807 0.0039

0.3 0.435 0.0031 0.8 0.878 0.0041

0.4 0.545 0.0033 0.9 0.942 0.0043

0.5 0.642 0.0035 1.0 1.000 0.0045

 

137

...'——_n___._ —— —_ -fl

1.0

 

 

 

 

 

0.8
0.6
S
0.4
0.2
0.0 1 i .‘
0.0 0.2 0.4 0.6 0.8 1.0

X1

Figure 7.6 Predicted VLE of 5HMD (1) + 4HMD (2) Mixtures at 373 °K.
x, y: liquid and molar compositions.

7.6 — Bias in SPEAD Simulation and Regression of A2

SPEAD commonly underestimates vapor pressures of the high molecular weight
compounds in homologous series. This bias has been seen in Psat predictions of amines,
amides, acetates, ketones, and hydrocarbons [67, 108], also shown in the fifth column of
Table 7-1.

Error was observed in regression of A2 for high molecular molecules of
homologous series. A maxima of A2 at high density was apparent in E3LA, and it
increased in size for E4LA and E5LA (Figures 7.7). The same trend was found with 2-

alkanols and training set esters.

138

 

 

 

 

 

 

430000 —
N-160000 “ XQK . _ 94“! ________ E4LA
< ‘~ '

-24oooo — __________

-' f ’ E5LA
-320000 1
-400000 r 1 1 1 1 1 1
O 0.1 0.2 0.3 0.4 0.5 0.6 0.7

11

Figure 7.7 Trend of A2 in Simulation of Ethyl Lactate Oligomers
Notations used in Figures are referred to chapter 5. Solid lines: connecting the values of
A2 in simulation. Dot lines are the regression of A2.

As it can be seen in Figure 7.7, the values of A2 from regression at high density

are higher than the values originally obtained from simulation (the regression curve is

above the actual A2 curve), resulting in ~25-30 % positive deviation in the values of

2‘42

6171 from regression compared to the original values for n > 0.52.

139

 

Chapter 8 - Conclusion and Recommendations

8.1 - Separation of Ricinolein

This work demonstrates the potential of adsorption can be an alternative
separation process to produce ricinolein from castor oil. Recently, a patent has been
granted to researchers at Dow Chemical, Inc. for their adsorptive separation method to
separate plant oil triglycerides from mixtures [3]. The methodology and result in that
patent are similar to the work presented in this dissertation.

Future work on modeling adsorption behavior will be useful for scaling up and
processing designs. Solvent mixtures can be used to enhance the solubility of castor oil
and/or the selectivity of adsorbents to glycerides. Studies to find the optimal solvent
strength for the selected triglycerides is also important. Isohexane which is identified as
not a hazardous air pollutant [156] in the United States should be replaced for hexane if it
is an option of choice. HPLC should be used to analyze the effluents, since the
derivatization of glycerides into FAME for GC analyses limits the detection of di- and
mono-ricinolein with non-functional fatty acids in samples.

Separation of ricinolein should be improved using simulated moving beds (SMB).
This separation technology has been successfu1 in separating p—xylene from xylene
mixtures, used in the petrochemical industry for almost 30 years. The SMB is also used
in fi'uctose/glucose separation and olefins/parafins separation [157]. The use of SMB can
avoid problems related to pressure drop and solid motion in fixed beds. It is more
efficient than the other types of adsorption, since it uses less adsorbent for the same

throughput.

140

 

8.2 — Extraction and Purification of SQDG

Results from this study show that non-chlorinated solvents can replace chloroform
in extraction and recovery of SQDG from alfalfa. Their applications can be extended to
any plant extractions. The proposed method in this dissertation uses significantly less
chemicals and solvents than the reported amount in literature.

Following this work, isohexane could replace hexane because this solvent is not
an environmental concern, and-ethanol can be substituted for methanol. Blended solvent
mixtures of more than two components, for example a mixture of acetone-hexane-
methanol should be tried. The Hansen solubility parameters can be used as the basis in
designing an effective solvent system.

Activated carbon or non-polar adsorbents could be used as an alternative
technique to remove chlorophyll and pigments from the extracts, depending on the costs
of solvent recovery compared to the cost of adsorbent regeneration. Design of solvent
recovery and economic analyses should be done to determine the optimal process.
Further studies to completely remove proteins prior to the use of adsorption and ion-
exchanger in purification of SQDG should be considered. The presence of proteins
blocks the adsorption sites, and builds up the pressure of columns.

Recovery of DMDG and DGDG along with SQDG should be evaluated. Alfalfa
contains a significant amount of these lipids compared to the SQDG content. Evaluation
of ion-exchange DEAE capacity and development of a method for detection of SQDG in
the effluent from ion-exchange columns will be useful. Other exchange materials should
be investigated since DEAE is relatively expensive. An inert material may also be used

to disperse DEAE powder to avoid channeling and high back pressure in the column.

141

 

Different sources of SQDG should be evaluated for application of the
extraction/purification methods. For example, algae have been reported as a promising

source of SQDG with a high content.

8.3 — Measured and Predicted VLE and Vapor Pressure

This work provided VLE and VLLE data and predicted vapor pressure of systems
involved in the reactive distillation of ethyl lactate, diethyl succinate, triethyl succinate,
and acetals of glycerol for research and industrial process designs. Isothermal
measurements were successfully performed using a custom made P-x-y apparatus. On
the other hand, isobaric measurements were taken from a commercial Fischer T—x-y
recirculating still. The P-x-y apparatus, developed in this study is valuable for reactive
chemical systems which cannot be measured at high temperature, and the Fischer T-x-y
still is effective for systems with slow kinetic, such as lactic acid and water mixtures.

For a successful operation of the P—x-y apparatus, the entire system must
completely be degassed prior to the VLE measurements. In addition to the procedure,
which is described in the journal paper [96] (also included in chapter 5), the following
actions are recommended: 1) Ensure that there is no plugging in the sample loop, and the
vapor line, the valve V3 and the sampling valve are free from impurity of gases and
liquids: open V3 and place the sampling valve at load position then turn on the vacuum
pump. If the system is leak-tight and completely degassed, the same reading from
Baratron is obtained each time when the sampling valve is switched from injet to load
position. If there is no plugging in the sample loop, pressure reading from the Baratron is
quickly reduced as the vacuum pump starts, for example t= 0 min, P= 730 torr, t= 5 sec,

P= 450 torr, t=105ec, P=300 torr. 2) Ensure that there is no air in the liquid lines and

142

liquids can be instantly ejected from valves V221, and V23 (repeat the test for each valve):
turn the liquid injector A (or B) to ~30-40 kPa (reading from PA or PB), then slightly open
valve (V2A or V23) until the first drop of liquid enters the equilibrium chamber then close
valve V21, (or V23). Record the reading from Baratron and repeat the test. If no liquid
was ejected from the valve V2A (or V213) or readings from the Baratron changed, re-
degassing of the feed chamber and liquid line should be considered. The reading from
the Baratron should be the pressure of pure liquid A (or B) at the experimental
temperature. 3) Turn the liquid injector in counterclockwise direction (about 1 turn for
each 5 turns in clockwise direction) when loading liquid from the feed chamber to an
empty injector. This activity provides a quick equilibration inside the injector and allows
liquid to be fully loaded. 4) Interrnittently open valve V4 and turn off the vacuum pump
during the evacuation of the equilibration chamber. This activity helps to dilute the gas
inside the chamber with fresh air, and reduces the time for a complete evacuation of the
chamber.

As discussed in chapter 6, the Fischer recirculating still does not provide adequate
mixing, and overheats the large portion mixture between the tip of the heating bulb and
the Cottrell pump. These limitations can be reduced by adding a section between the feed
chamber and the returning liquid line from the Cottrell pump to recirculate liquid, and
provide a good mixing before liquid flows back to the reboiler.

The rate of recirculating vapor is regulated by the preset in operating the Fischer
apparatus. This parameter should be assigned carefully to reduce the effect of
overheating liquid. For a low boiling point liquid ~ 80-100 °C, the preset should be

~ 10 %. For an intermediate boiling point liquid such as ethyl lactate, the preset should

143

 

be ~ 15-18 %, and for high boiling point or highly viscosity liquid such as concentrated
lactic solution, the preset should be ~25-28 %.

SPEAD molecular simulation demonstrates an excellent capability to predict
vapor pressure of complex molecules. A FORTRAN program is provided from this work
for use with SPEAD. It can be modified for optimization of a larger number of SPEAD
parameters. To obtain the global optimum, the scaling factor which relates to the Hessian
matrix used in the DBCONF routine of IMSL must be set in the right order of magnitude
to avoid stepping outside the interval. In addition, the order of the initial parameters
guesses in the input file should be used carefully, because the routine DBCONF always
adjusts parameters in the order in which they are entered before performing simultaneous
searches. Different orders were used but no clear trends were observed and no definite
recommendations can be made.

Difficulties were encountered in converging SPEAD vapor pressure calculations
at high reduced temperatures, which may be due to the fitted A1 and A2 terms, or may be
due to the iteration method. SPEAD should improve the fitting of A2 in addition to any

weaknesses in the overall simulation or perturbation method. The current expression of

2 3 4
A2 = cln+c2n +C3n4 +64" indicates that coefficient C4 is the dominating factor at high
1+ 50017

 

density. A different polynomial to obtain a better regression of A2 at high densities,
and/or a correlation to constrain its coefficients should be considered. Fitting of A2 for
the intermediate densities may also need to be improved because the calculated Z shows
some deviation from the simulation data in this region. The limitations of the
perturbation convergence may be important and should be evaluated. Perhaps third or

higher order terms are needed for certain conditions, and a longer simulation should be

144

 

performed for large molecules to reduce the apparently random scatter at high densities.

To reduce the underestimation of vapor pressures for high molecular compounds,
more investigation should be undertaken to determine the cause of the curvature in the
predicted curvature of ln(Psat) versus T“, The ratio of A2/A1 is known to be a function
of molecular weight [84], but the dependence has not been quantified. Such a correlation
may be helpful in extending predictive capabilities.

Further studies of both intra and intermolecular hydrogen bonding systems are
desirable to extend the reliability of SPEAD for highly oxygenated compounds, which
usually have multiple hydrogen bonding sites. These compounds are expected to be
deve10ped from natural feedstocks. SPEAD currently does not have a method to
differentiate between hydrogen bond donors and hydrogen bond acceptors, and greatly
underestimates the vapor pressure of diols, for example the 2,3-butanediol was
underestimated by 79 %.

Another interesting modification would involve including the hydrogen bonding
in the reference simulation. This modification requires careful thought because the
simulations may need to be performed at multiple temperatures. The non-specific
attractive potentials could be added as the perturbation, therefore the DMD simulation
would still be faster than a simulation with full potentials.

The use of wells at positions 1.2, 1.5, 1.8 and 2.0 0‘ was dictated by the previously

developed SPEAD programs. The position of the wells should be adjusted to more
accurately approximate the Lennard-Jones potential. The current SiteParms. 2580. txt
(SPEAD well table) has a number of wells that do not represent the expected Lennard-

Jones shape. The well depths could be coupled to reduce the number of adjustable

145

 

parameters. The two intermediate wells are currently interpolated from the inner and
outer wells, but a large number of variables required adjustments in searching for the
optimal parameters of one functional group in some cases.

This work was performed using a compiled version of SPEAD. Therefore,
flexibility was limited for investigating the simulation results. For example, it was not
possible to explore the intra- and intermolecular events using the current SPEAD version,
and SPEAD could not be run on the MSU supercomputers. Also, it was discovered that
the code uses a fixed seed for pseudorandom number generations, which exactly provides
the same value of Z when the simulation is repeated. Therefore, the scatter in the
calculation of A2 was not eliminated by restarting the simulation. MSU should develop

an independent simulation code and smoothing programs.

146

l.

APPENDIX A

EXPERIMENTAL ABSORPTION DATA
(Reference in Chapter 2)

147

Run # 6: In ethanol, XAD-2 = 1.9296 g, flowrate = 0.16 ml/min, C = 1.0 %, S = 0%

 

‘Fraction ”Time “Bed VOI d611% ecrc0 'c.Ic.,., °C;/02,0

 

l 20 0.84 0.10% 0.100 - -
2 40 1.68 0.53% 0.530 - -
3 59 2.48 0.77% 0.770 - -
4 79 3.33 0.84% 0.840 - -
5 99 4.17 0.82% 0.820 - -
6 119 5.01 0.85% 0.850 - -
7 149 6.27 0.89% 0.890

8 189 8.00 0.98% 0.980 - -

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

e' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , Cm, C20.

148

Run # 7: In ethanol, XAD-2 = 2.802 g, flowrate = 0.16 ml/min, C = 2.46 %, S = 0%

 

 

‘Fraction ”Time °Bed VOI “oil % eolco 'c.Ic.,0 °C,/cm

1 15 0.545 0.07% 0.027 - -

2 25 0.909 0.06% 0.024 0.026 0.005
3 40 1.455 0.71% 0.290 0.317 0.090
4 56 2.036 1.80% 0.732 0.785 0.341
5 71 2.582 2.09% 0.851 0.866 0.745
7 101 3.673 2.30% 0.934 0.932 0.950
8 116 4.218 2.34% 0.953 0.957 0.925
9 131 4.764 2.36% 0.960 0.907 1.353
10 146 5.309 2.43% 0.989 0.991 0.975
1 1 162 5.891 2.43% 0.991 0.993 0.972
12 176 6.400 2.45% 0.996 1.002 0.951
13 191 6.945 2.45% 0.997 1.000 0.977
14 206 7.491 2.49% 1.013 1.016 0.994
15 221 8.031 2.48% 1.010 1.017 0.962

 

 

C and S denote for castor and soybean oil. a : sample collected in the experiment, b :

time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,
e, f, g

: concentrations of total oi1 (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , Cm, C20

149

 

1.4
1.2 r

1

CIC 0, C1/C1,o, C 2/C2,o

 

 

 

 

0 3 6 9 12

Bed volume

Figure A.1 Fixed-bed Adsorption Using Ethanol and XAD-2 (Run # 7)
0: total oil, 0: hydroxyl, A: non-hydroxy.

150

Run # 8: In ethanol, SD-2 = 2.833 g, flowrate = 0.18 ml/min, C = 0 %, S = 3.69 %

 

 

'Fraction I’Time °Bed Vol doil % °CIC0 'Cde °C,/Cm
1 12 0.44 0.07% 0.020 0.000 0.020
2 24 0.88 0.00% 0.000 0.000 0.000
3 36 1.32 0.02% 0.004 0.000 0.004
4 51 1.87 0.31% 0.087 0.000 0.087
5 68 2.49 1.28% 0.357 0.000 0.357
6 83 3.04 2.39% 0.663 0.000 0.663
7 97 3.56 2.66% 0.739 0.000 0.739
8 114 4.18 2.92% 0.812 0.000 0.812
9 130 4.77 3.22% 0.896 0.000 0.896
10 148 5.43 3.42% 0.951 0.000 0.951
11 163 5.98 3.49% 0.971 0.000 0.971
12 178 6.53 3.56% 0.988 0.000 0.988

 

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

6' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , C10, C20,

151

Run # 9: In ethanol, SD-2 = 2.904 g, flowrate = 0.17 ml/min, C = 0.80 %, S = 1.60 %

 

 

‘Fraction t’Time °Bed Vol "oil % eCICo 'C1IC1,0 gczlcm
1 16 0.55 -0.01% -0.006 - -
2 32 1.11 0.03% 0.012 0.019 0.009
3 51 1.77 0.26% 0.109 - -
4 67 2.32 0.53% 0.220 - -
5 82 2.84 0.93% 0.386 0.670 0.267
6 97 3.36 1.27% 0.527 - -
7 114 3.95 1.58% 0.660 1.008 0.512
8 133 4.61 1.81% 0.754 1.042 0.631
9 150 5.20 1.98% 0.824 - -
10 166 5.75 2.09% 0.870 - -
11 181 6.27 2.20% 0.915 - -
12 196 6.79 2.24% 0.933 - -
13 214 7.42 2.24% 0.931 0.857 0.960
14 232 8.04 2.15% 0.895 0.938 0.872
16 271 9.39 2.32% 0.968 0.908 0.988
17 293 10.16 2.38% 0.991 - -
18 319 11.06 2.40% 0.999 - -

 

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

8' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , (31,0, C20,

152

Run # 10: In ethanol, SD-2 = 2.836 g, flowrate = 0.19 mI/min, C = 1.99 %, S = 0.41 %

 

 

'Fraction ”Time °Bed V01 “oil % °CIC° 'c.7c.,., gczlcz,o
1 16 0.65 0.00% -0.002 0.000 0.000
2 31 1.26 0.19% 0.078 0.104 0.014
3 46 1.86 0.80% 0.336 0.431 0.098
4 62 2.51 1.32% 0.554 0.692 0.204
5 77 3.12 1.65% 0.690 0.838 0.317
6 92 3.73 1.91% 0.798 0.942 0.435
7 109 4.41 2.05% 0.858 0.935 0.663
8 124 5.02 2.13% 0.891 0.984 0.657
9 139 5.63 2.19% 0.915 0.000 0.000
10 154 6.24 2.19% 0.917 1.019 0.659
11 169 6.84 1.87% 0.781 0.820 0.681
12 186 7.53 2.29% 0.957 0.987 0.880
13 202 8.18 2.32% 0.970 0.000 0.000
14 219 8.87 2.33% 0.973 0.000 0.000
15 235 9.52 2.31% 0.967 0.000 0.000

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, °: number of bed volumes, d : total oil concentration,

e' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , C1,o, C2,o_

153

Run # 11: In ethanol, SD-2 = 2.837 g, flowrate = 0.20 ml/min, C = 0.38 %, S = 1.99 %

 

 

'Fraction ”Time °Bed v61 dOII % °c1co "c.Ic.,o gc,7c,,..
1 15 0.62 0.00% -0.002 0.000 0.000
2 30 1.24 0.01% 0.003 0.020 0.001
3 45 1.86 0.24% 0.103 0.403 0.050
4 61 2.52 0.66% 0.276 0.873 0.170
5 77 3.18 1.06% 0.443 0.958 0.353
6 94 3.88 1.43% 0.600 0.000 0.000
7 109 4.50 1.75% 0.734 0.944 0.697
8 126 5.20 1.93% 0.810 1.073 0.763
9 142 5.86 2.09% 0.880 1.103 0.840
10 157 6.48 2.19% 0.921 0.000 0.000
11 174 7.19 2.30% 0.965 1.187 0.925
12 191 7.89 2.26% 0.950 0.000 0.000
13 206 8.51 2.30% 0.965 1.146 0.933
14 221 9.13 2.31% 0.971 0.000 0.000

 

C and S denote for castor and soybean oil.

time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

e,f,g

(C2) in the effluent compared to their initial C0 , Cm, C20

154

: sample collected in the experiment, b :

: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

 

Run # 12: In ethanol, SD-2 = 2.826 g, flowrate = 0.20 ml/min, C = 0 %, S = 2.44 %

 

 

‘Fraction “Time °Bed v61 "oil °/. °c7co 'c.1c.,0 s'czlcz,o
1 15 0.66 0.00% -0.002 0.000 0.000
2 30 1.31 0.02% 0.007 0.000 0.007
3 45 1.97 0.23% 0.094 0.000 0.094
4 60 2.62 0.38% 0.156 0.000 0.156
5 75 3.28 0.85% 0.351 0.000 0.351
6 90 3.93 1.47% 0.606 0.000 0.606
7 105 4.59 1.95% 0.807 0.000 0.820
8 120 5.25 1.98% 0.820 0.000 0.807
9 135 5.90 1.72% 0.807 0.000 0.710
10 154 6.73 2.21% 0.914 0.000 0.914
11 169 7.39 2.36% 0.974 0.000 0.000
12 184 8.04 2.52% 1.042 0.000 0.000
14 214 9.35 2.44% 1.010 0.000 0.000

 

C and S denote for castor and soybean oil.

time (minutes) after fixed-bed starts, ° : number of bed volumes, d : total oil concentration,

e, f,g

(C2) in the effluent compared to their initial C0 , C10, C20,

155

: sample collected in the experiment, b :

: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

Run # 13: In ethanol, SD-2 = 2.835 g, flowrate = 0.19 ml/min, C = 0 %, S = 2.42 %

 

 

'Fraction I’Time cBed Vol c'oil % °CIC° 'Cde gczlcm
1 15 0.63 0.44% 0.168 0.000 0.168
2 30 1.26 0.05% 0.020 0.000 0.020
3 45 1.90 0.46% 0.174 0.000 0.174
4 60 2.53 0.68% 0.257 0.000 0.257
5 75 3.16 1.56% 0.592 0.000 0.592
6 92 3.88 1.47% 0.555 0.000 0.555
7 112 4.72 1.81% 0.687 0.000 0.687
8 127 5.35 2.06% 0.780 0.000 0.000
9 142 5.98 2.22% 0.840 0.000 0.000
10 157 6.61 1.78% 0.674 0.000 0.000
11 172 7.25 2.34% 0.887 0.000 0.000
12 187 7.88 2.47% 0.937 0.000 0.000
13 202 8.51 2.41% 0.915 0.000 0.000
14 217 9.14 2.54% 0.962 0.000 0.000
15 229 9.65 2.63% 0.996 0.000 0.000

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :

time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,
e, f, g

: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , Cm, C20

156

Run # 14: In ethanol, SD-2 = 2.832 g, flowrate = 0.19 ml/min, C = 0 %, S = 2.42 %

 

 

‘Fraction ”Time °Bed v61 “oil % °c1co 'c./c.,., °C,/cm
1 15 0.63 0.26% 0.109 0.000 0.109
2 30 1.27 0.08% 0.035 0.000 0.035
3 45 1.90 0.16% 0.067 0.000 0.067
4 60 2.54 0.58% 0.241 0.000 0.241
5 75 3.17 1.03% 0.429 0.000 0.000
6 90 3.80 1.42% 0.590 0.000 0.000
7 105 4.44 1.37% 0.571 0.000 0.000
8 120 5.07 1.92% 0.802 0.000 0.000
9 135 5.71 2.11% 0.880 0.000 0.000
10 150 6.34 1.70% 0.708 0.000 0.000
1 1 165 6.98 1.23% 0.514 0.000 0.000
12 180 7.61 2.33% 0.970 0.000 0.000
13 195 8.24 2.03% 0.847 0.000 0.000
14 210 8.88 1.80% 0.752 0.000 0.000
15 225 9.51 1.80% 0.749 0.000 0.000

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :

time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,
6' f’ 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , Cm, C20

157

Run # 15: In ethanol, SD-2 = 2.832 g, flowrate = 0.18 mI/min, C = 3.76 %, S = 0 %

 

 

‘Fraction I’Time °Bed Vol doil % °CICo 'Cde 902mm

1 15 0.59 0.07% 0.018 - -

2 30 1.17 0.31% 0.079 0.085 0.025
3 45 1.76 1.59% 0.409 0.425 0.260
4 60 2.34 2.57% 0.660 0.672 0.546
5 75 2.93 3.07% 0.789 0.819 0.620
6 90 3.51 3.42% 0.877 0.889 0.771
7 105 4.10 3.58% 0.918 0.924 0.863
8 120 4.68 3.67% 0.943 0.970 0.798
9 135 5.27 3.19% 0.819 0.799 0.997
10 150 5.85 3.72% 0.954 0.968 0.826
11 165 6.44 3.72% 0.956 - -

12 180 7.02 3.71% 0.952 - -

13 198 7.72 3.78% 0.971 0.992 0.775

 

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

6' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C2,0_

158

a : sample collected in the experiment, b t

Run # 16: In ethanol, SD-2 = 2.834 g, flowrate = 0.19 ml/min, C = 1.20 %, S = 1.21 %

 

 

‘Fraction ”Time °Bed VOI “oil °/. ecrco 'c.7c.,., we...
1 15 0.62 0.00% 0.002 0.004 0.000
2 30 1.24 0.06% 0.024 0.049 0.004
3 45 1.86 0.34% 0.144 0.293 0.027
4 60 2.48 0.78% 0.327 0.581 0.126
5 75 3.10 1.11% 0.468 0.748 0.246
6 90 3.72 1.49% 0.629 0.928 0.391
7 105 4.34 1.67% 0.704 0.970 0.493
8 120 4.96 1.88% 0.793 1.021 0.613
9 135 5.57 1.99% 0.837 1.003 0.705
10 150 6.19 2.09% 0.879 0.974 0.804
11 165 6.81 1.90% 0.800 0.958 0.674
12 185 7.64 2.28% 0.962 - -
13 202 8.34 2.24% 0.944 0.989 0.909
14 219 9.04 2.26% 0.951 0.908 0.986
15 239 9.87 2.36% 0.995 1.036 0.962

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

8' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , Cm, C2,o_

159

Run # 17: In ethanol, SD-2 = 2.835 g, flowrate = 0.17 ml/min, C = 2.40 %, S = 0 %

 

 

'Fraction ”Time °Bed VOI “oil % °CICo 'c./c.,o [Jog/c2,0

1 10 0.37 0.05% 0.019 - -

2 25 0.94 0.12% 0.050 0.055 0.017
3 42 1.57 0.34% 0.142 0.153 0.062
4 60 2.25 0.99% 0.410 0.433 0.238
5 75 2.81 1.50% 0.623 0.653 0.402
6 91 3.41 1.82% 0.756 0.784 0.552
7 108 4.04 2.36% 0.978 1.006 0.774
8 125 4.68 2.23% 0.924 0.944 0.776
9 140 5.24 2.33% 0.965 0.982 0.841
10 157 5.88 2.37% 0.983 0.997 0.885
11 174 6.51 2.37% 0.981 0.988 0.931
12 194 7.26 2.39% 0.993 1.000 0.939
13 211 7.90 2.41% 1.000 0.995 1.039

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

9' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C20

160

 

1.4 r

l

1.0
0.8
0.6

1

CIC 6, C1IC 1,6, C 2IC 2,0

 

 

 

 

0.0 ‘=’ I T I
0 3 6 9 12

Bed volume

Figure A.2 Fixed-bed Adsorption Using Ethanol and SD-2 (Run # 9)
0: total oil, O: hydroxyl, A: non-hydroxy.

 

1.4 ~
1.2 r
1.0 7

l

0.6
0.4 —
0.2
0.0

CIC 0, C 1IC 1,0, C 2IC 2,0

l

 

 

 

 

l I

0 3 6 9 12

-—1

Bed volume

Figure A.3 Fixed-bed Adsorption Using Ethanol and SD-2 (Run #10)
9: total oil, O: hydroxyl, A: non-hydroxy.

161

CIC 0, C 1IC 1,0, C ZIC 2.0

 

l

1.4
1.2 7
1.0 ~
0.8 ~
0.6
0.4
0.2
0.0

l

l

J

 

 

 

 

r I

0 3 6 9 12

Bed volume

Figure A.4 Fixed-bed Adsorption Using Ethanol and SD-2 (Run #11)

CIC 6, C1IC 1,6, C zIC 2,0

0: total oil, O: hydroxyl, A: non-hydroxy.

 

1.4 -
1.2 1
1.0 1 ,. , ._ --------

0.6 _
0.4 1
0.2 7

0.0 7 1 1
0 3 6 9 12

 

 

 

 

 

Bed volume

Figure A.5 Fixed-bed Adsorption Using Ethanol and SD-2 (Run # 15)
0: total oil, O: hydroxyl, A: non-hydroxy.

162

 

1.4 1

CIC 0, C1/C 1,0, C 2/C 2,0

 

 

 

 

l

0 3 6 9 12

Bed volume

Figure A.6 Fixed-bed Adsorption Using Ethanol and SD-2 (Run # 16)
0: total oil, O: hydroxyl, A: non-hydroxy.

 

 

 

 

 

1.4 —
,3, 1.2 —
9
g 1.0 —
2- 0.8 —
Q
[E 0.6 —
13 0.4 ~
0

0.2 —

0.0 _ T 1

o 3 6 9 12

Bed volume

Figure A.7 Fixed-bed Adsorption Using Ethanol and SD-2 (Run #17)
0: total oil, O: hydroxyl, A: non-hydroxy.

163

Run # 18: In ethanol, L-493 = 2.865 g, flowrate = 0.19 mein, C = 1.64 %, S = 0.84 %

 

 

“Fraction ”Time °Bed VOI “oil % ecrco 'c.Ic.,,, °C,/cm
1 15 0.572 001% -0003 - -
2 34 1.296 0.07% 0.029 0.048 0.006
3 54 2.059 0.60% 0.243 0.350 0.114
4 70 2.669 1.23% 0.501 0.682 0.283
5 87 3.317 1.66% 0.676 0.988 0.300
6 102 3.888 1.89% 0.770 - -
7 121 4.613 2.04% 0.831 1.141 0.459
8 141 5.375 2.14% 0.871 - -
9 162 6.176 2.22% 0.900 1.054 0.716
10 183 6.976 2.31% 0.939 - -
11 204 7.777 2.36% 0.960 1.128 0.759
12 228 8.692 2.39% 0.971 - -
13 251 9.569 2.41% 0.981 - -
14 274 10.445 2.41% 0.978 1.108 0.822

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

e' f’ 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , (31,0, C20

164

Run # 19: In ethanol, L-493 = 2.827 g, flowrate = 0.19 m1/min, C = 0.82 %, S = 1.63 %

 

 

‘Fraction I’Time cBed Vol doil % °CIC° 'Cde °C;/02,0
1 18 0.661 0.03% 0.008 - -
2 38 1.396 0.03% 0.012 - -
3 53 1.947 0.24% 0.098 0.264 0.018
4 74 2.718 0.75% 0.305 0.716 0.106
5 91 3.343 1.28% 0.520 1.018 0.281
6 106 3.894 1.62% 0.663 0.899 0.549
7 120 4.408 1.89% 0.771 0.982 0.670
8 139 5.106 2.09% 0.853 - -
9 157 5.767 2.19% 0.896 1.012 0.840
10 181 6.649 1.89% 0.772 - -
11 201 7.384 2.35% 0.960 0.982 0.950
12 223 8.192 2.40% 0.980 - -
13 243 8.927 2.43% 0.991 1.056 0.960
14 262 9.625 2.36% 0.962 - -

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, ° : number of bed volumes, d : total oil concentration,

e' f' 9: concentrations of total 0il (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C20

165

Run # 20: In ethanol, L-493 = 2.875 g, flowrate = 0.19 ml/min, C = 2.48 %, S = 0 %

 

 

'Fraction ”Time “Bed VOI ‘oil % °CICo 'c./c.,o gczlcm
1 20 0.756 0.12% 0.047 0.049 0.033
2 46 1.740 1.11% 0.447 0.440 0.486
3 62 2.345 1.81% 0.729 - -

4 79 2.987 2.11% 0.852 0.894 0.601
5 96 3.630 2.29% 0.924 0.936 0.853
6 112 4.235 2.41% 0.971 - -
7 128 4.840 2.42% 0.975 0.985 0.915
8 148 5.597 2.43% 0.982 - -
9 163 6.164 2.49% 1.003 1.018 0.910
10 180 6.807 2.47% 0.996 - -
11 199 7.525 2.46% 0.992 - -
12 214 8.092 2.48% 1.001 1.007 0.968
13 228 8.622 2.48% 0.998 - -

 

‘3 : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, °: number of bed volumes, d : total 011 concentration,

e' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , C10. C20.

166

 

l

1.4
1.2
1.0
0.8
0.6
0.4 _
0.2
0.0

l

1

CIC 0, C1/C1,o, C 2IC2,o

l

 

 

 

 

l P T

0 3 6 9 12

Bed volume

Figure A.8 F ixed-bed Adsorption Using Ethanol and L-493 (Run #18)
0: total oil, O: hydroxyl, A: non-hydroxy.

 

CIC 0, C1IC1,o, C 2IC2,o

 

 

 

 

0 3 6 9 12

Bed volume

Figure A.9 Fixed-bed Adsorption Using Ethanol and L493 (Run # 19)
6: total oil, O: hydroxyl, A: non-hydroxy.

167

 

1.4 a
1.2 -
1.0 1
0.8 1
0.6 ~
0.4 -
0.2 ~

0.0 1 1 T
0 3 6 9 12

CIC 0, C1/C1,0, C 2102.0

 

 

 

 

Bed volume

Figure A.10 Fixed-bed Adsorption Using Ethanol and L-493 (Run # 20)
0: total oil, O: hydroxyl, A: non-hydroxy.

168

Run # 21: In Ethanol, M-43 = 2.837 g, flowrate = 0.14 ml/min, C = 2.51 %, S = 0%

 

 

'Fraction ”Time °Bed Vol ‘1011 % °c1co b.7012. 902/sz
1 16 0.679 0.02% 0.010 0.000 0.000
2 31 1.316 0.97% 0.385 0.387 0.366
3 46 1.953 1.92% 0.765 0.769 0.735
4 62 2.633 2.25% 0.895 0.902 0.838
5 76 3.227 2.27% 0.903 0.910 0.852
6 91 3.864 2.44% 0.973 0.000 0.000
7 106 4.501 2.42% 0.966 0.973 0.912
8 121 5.138 2.42% 0.963 0.958 1.000
9 139 5.902 2.43% 0.967 0.000 0.000
10 156 6.624 2.46% 0.979 0.000 0.000
11 177 7.516 2.45% 0.976 0.000 0.000
12 192 8.152 2.44% 0.973 0.000 0.000
13 209 8.874 2.42% 0.963 0.000 0.000
14 222 9.426 2.41% 0.960 0.000 0.000
15 238 10.106 2.47% 0.983 0.986 0.965

C and S denote for castor and soybean oil. : sample collected in the experiment, b :
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

8' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C1,o, C20.

169

 

1.4 ~
1.2 -

CIC 0, C1/C1,o, C ZIC 2,0

 

0.0 -

Figure A.11

 

 

 

I I I

3 6 9 12

Bed volume

Fixed-bed Adsorption Using Ethanol and M43 (Run # 21)
0: total oil, 0: hydroxyl, A: non-hydroxy.

170

Run # 22: In Ethanol, Florisil = 2.369 g, flowrate = 0.19 ml/min, C = 0.81 %, S=1.61 %

* Same as Table 2-4 in page 16

 

 

‘Fraction ”Time °Bed Vol “011 % eCIc0 'c,/c.,o 902162111
1 15 0.620 0.05% 0.021 - -
2 32 1.323 1.08% 0.448 0.465 0.432
3 48 1.984 1.99% 0.824 0.756 0.840
4 65 2.686 2.21% 0.915 0.937 0.889
5 85 3.513 2.26% 0.933 0.882 0.939
6 115 4.753 2.32% 0.960 - -
7 133 5.497 2.34% 0.969 1.129 0.880
8 151 6.241 2.35% 0.971 1.291 0.810
9 168 6.943 2.38% 0.983 - -
11 183 7.563 2.37% 0.978 - -
12 222 9.175 2.41% 0.996 1.242 0.868

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :

time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,
e, f, g

: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , Cm, C20,

171

 

1.4 ..

0.6
0.4 ~
0.2 _

0.0 1 1 1 1
O 3 6 9 12

1

CIC 0, C1IC1,0, C 2IC2,o

 

 

 

 

Bed volume

Figure A.12 Fixed-bed Adsorption Using Ethanol and Florisil (Run # 22)
0: total oil, O: hydroxyl, A: non-hydroxy. * Same as Figure 2.3 in page 16

172

Run # 23: In Hexane, F lorisil = 2.408 g, flowrate = 0.14 ml/min, C = 0 %, S = 3.77 %

 

 

‘Fraction ”Time °Bed Vol d011 % °CICo 'c,1c.,o 90270,,o
1 17 0.50 0.02% 0.005 0.000 0.005
2 37 1.08 0.01% 0.003 0.000 0.003
3 59 1.72 0.01% 0.001 0.000 0.001
4 79 2.31 0.01% 0.003 0.000 0.003
5 95 2.77 0.09% 0.024 0.000 0.024
6 115 3.36 0.82% 0.219 0.000 0.219
7 139 4.06 2.08% 0.553 0.000 0.553
8 161 4.70 2.97% 0.789 0.000 0.789
9 179 5.23 3.00% 0.797 0.000 0.797
10 199 5.81 3.88% 1.033 0.000 1.033
11 219 6.39 3.93% 1.046 0.000 1.046

 

C and S denote for castor and soybean oil.

time (minutes) after fixed—bed starts, c: number of bed volumes, d : total oil concentration,

e, f,g

a

(C2) in the effluent compared to their initial Co , C10, C20,

173

: sample collected in the experiment, b :

: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

Run # 24: In Hexane, Florisil = 2.407 g, flowrate = 0.14 ml/min, C = 0 %, S = 2.46 %

 

 

“Fraction ”Time °Bed Vol (1011 % °CICo 'c.Ic.,0 g0,70,,o
1 15 0.46 -0.05% -0019 0.000 -0019
2 38 1.17 -0.02% -0.007 0.000 -0.007
3 55 1.69 0.04% 0.014 0.000 0.014
4 70 2.16 0.01% 0.003 0.000 0.003
5 87 2.68 0.10% 0.039 0.000 0.039
6 107 3.30 0.60% 0.244 0.000 0.244
7 123 3.79 1.06% 0.429 0.000 0.429
8 138 4.25 1.33% 0.537 0.000 0.537
9 157 4.84 1.30% 0.527 0.000 0.527
10 173 5.33 2.21% 0.892 0.000 0.892
11 192 5.92 2.32% 0.936 0.000 0.936
12 214 6.59 2.43% 0.981 0.000 0.981
13 234 7.21 2.46% 0.994 0.000 0.994
14 254 7.83 2.46% 0.993 0.000 0.993
15 274 8.44 2.49% 1.008 0.000 1.008
16 296 9.12 2.49% 1.005 0.000 1.005
17 318 9.80 2.51% 1.014 0.000 1.014

 

C and S denote for castor and soybean oil. a : sample collected in the experiment, b :
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

e' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C20

174

Run # 25: In Hexane, Florisil = 2.427 g, flowrate = 0.21 ml/min, C = 1.43 %, S = 0 %

 

‘Fraction hTime °Bed Vol d011% °CIC° 'Cde 902/02,.

 

1 17 0.75 0.05% 0.033 - -

2 34 1.51 0.09% 0.063 - -

3 49 2.17 0.03% 0.024 - -

4 68 3.01 0.01% 0.008 - -

5 89 3.94 0.03% 0.020 - -

6 107 4.74 0.02% 0.017 - -

7 123 5.45 0.00% 0.003 - -

8 144 6.38 0.00% 0.000 - -

9 164 7.26 0.06% 0.045 - -
10 181 8.02 0.19% 0.131 0.085 0.442
1 1 200 8.86 0.39% 0.272 0.193 0.803
12 221 9.79 0.61% 0.422 0.295 1.266
13 246 10.90 0.85% 0.591 0.465 1.440
14 266 11.78 0.84% 0.583 - -
15 283 12.53 0.96% 0.667 - -
16 298 13.20 1.01% 0.705 0.589 1.477
17 315 13.95 1.11% 0.770 - -
18 339 15.02 1.12% 0.782 - -
19 357 15.81 1.18% 0.822 0.597 2.332
20 375 16.61 1.22% 0.852 - -
21 394 17.45 1.15% 0.798 - -
22 413 18.29 1.28% 0.893 0.710 2.122

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, °: number of bed volumes, d : total oil concentration,

9" f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C20,

175

Run # 26: In Hexane, Florisil = 2.413 g, flowrate = 0.18 ml/min, C = 1.29 %, S = 1.3 %

 

 

‘Fraction ”Time c3001 Vol “011 °/. 30700 'c./c.,. 90270,,o
1 20 0.77 0.02% 0.006 - -
2 40 1.55 0.03% 0.010 - -
3 64 2.47 0.08% 0.031 - -
4 87 3.36 0.14% 0.051 - .-
5 109 4.21 0.75% 0.278 - -
6 132 5.10 1.14% 0.420 - -
7 154 5.95 1.38% 0.509 - -
8 174 6.72 2.02% 0.747 - -
9 197 7.61 2.08% 0.767 - -
10 220 8.50 2.33% 0.861 - -
11 238 9.20 2.35% 0.868 - -
12 253 9.78 2.40% 0.886 - -
14 294 11.36 1.77% 0.654 - -

 

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :
time (minutes) after fixed—bed starts, c: number of bed volumes, d : total oil concentration,

9' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C20,

176

Run # 27: In Hexane, Florisil = 2.413 g, flowrate = 0.18 ml/min, C = 1.66 %, S =0.85 %

 

 

‘Fraction ”Time °Bed Vol “011 % °crco 'c.Ic.,o °C;/02.1
1 18 0.66 -0.02% -0.006 - -
2 35 1.28 001% -0003 - -
3 55 2.01 0.06% 0.024 - -
4 75 2.75 0.16% 0.061 - -
5 96 3.52 0.48% 0.183 0.008 0.419
6 119 4.36 1.06% 0.407 0.030 0.913
7 142 5.20 1.49% 0.569 0.048 1.269
8 166 6.08 1.76% 0.673 0.188 1.325
9 191 6.99 1.80% 0.687 - -
10 217 7.95 1.78% 0.682 - -
11 243 8.90 1.80% 0.687 0.438 1.023
12 270 9.89 1.89% 0.723 - -
13 295 10.80 1.93% 0.739 0.633 0.880
14 313 11.46 2.03% 0.775 - -
15 330 12.08 2.05% 0.785 0.618 1.011
16 358 13.11 2.02% 0.771 - -
17 375 13.73 2.09% 0.799 - -
18 400 14.65 2.17% 0.829 0.664 1.050
19 420 15.38 2.20% 0.841 - -

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, °: number of bed volumes, d : total oil concentration,

3' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , Cm, C20,

177

Run # 28: In Hexane, Florisil = 2.462 g, flowrate = 0.14 ml/min, C = 0.80 %, S =1.68 %

 

 

‘Fraction ”Time °Bed Vol cl011 % °CIC0 'c./c.,0 90,702,,
1 17 0.60 0.04% 0.017 - -
2 34 1.20 0.04% 0.015 - -
3 51 1.79 0.02% 0.008 - -
4 70 2.46 0.05% 0.018 - -
5 92 3.24 0.17% 0.068 - -
6 115 4.05 0.77% 0.306 0.010 0.431
7 138 4.86 1.15% 0.459 - -
8 161 5.66 1.98% 0.790 0.032 1.111
9 185 6.51 2.11% 0.841 0.062 1.171
10 208 7.32 2.30% 0.916 0.066 1.277
11 232 8.16 2.31% 0.921 0.054 1.288
12 253 8.90 2.29% 0.913 - -
13 279 9.82 2.27% 0.903 0.125 1.232
14 306 10.77 2.23% 0.890 - -
15 335 11.79 2.20% 0.877 0.353 1.099
17 378 13.30 2.34% 0.932 0.646 1.053

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

6' f' 9: concentrations of total oil (C), ricinolein (C 1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C10, C20

178

 

1.4 7
1.2 7
1.0 7

0.6 7
0.4 7

0.2 7 /

0.0 7 ' ~~ 1 1 1

CIC 0
o
00

 

 

 

Bed volume

Figure A.13 Fixed-bed Adsorption Using Hexane and Florisil (Runs # 23, 24, 26)
9: total oil (Run # 23), A: total oil (Run # 24), 0: total oil (Run # 26)

 

2.5
2.0 —
1.5 ~ ,2. ...... A":
1.0

0.5

CIC 0, C1IC1,0, C2IC2,0

 

 

 

0.0

 

Bed volume

Figure A.14 Fixed-bed Adsorption Using Hexane and Florisil (Run # 25)
0: total oil, O: hydroxyl, A: non-hydroxy.

179

 

2.5

2.0 n

 

CIC 0, C1IC1,o, C 2102.0

 

 

 

 

Bed volume

 

Figure A.15 Fixed-bed Adsorption Using Hexane and Florisil (Run # 27)
6: total oil, 0: hydroxyl, A: non-hydroxy.

 

2.5

2.0 ~*

CIC 0, C1IC1,o, C 2102.0

 

 

 

 

 

Bed volume

Figure A.16 Fixed-bed Adsorption Using Hexane and Florisil (Run # 28)
0: total oil, 0: hydroxyl, A: non-hydroxy.

180

Run # 29: In Methanol, SD—2 = 2.835 g, flowrate = 0.20 ml/min, C = 2.44 %, S = O %

 

 

'Fraction ”Time °Bed Vol "oil °/. °c1co 'c.Ic.,o gczlcz,0
1 15 0.60 0.01% 0.003 0.000 0.000
2 30 1.19 0.00% -0.002 -0.002 -0.002
3 45 1.79 0.05% 0.019 0.019 0.022
4 60 2.38 0.23% 0.093 0.101 0.039
5 75 2.98 0.76% 0.308 0.327 0.169
6 90 3.57 1.33% 0.538 0.000 0.000
7 106 4.21 1.82% 0.737 0.775 0.454
8 118 4.68 2.09% 0.846 0.000 0.000
9 133 5.28 2.25% 0.911 0.954 0.590
10 145 5.76 2.33% 0.940 0.000 0.000
1 1 157 6.23 2.37% 0.960 0.998 0.675
12 177 7.03 2.38% 0.964 0.000 0.000
13 197 7.82 2.39% 0.964 0.000 0.000
14 217 8.61 2.46% 0.992 0.985 1.051
15 237 9.41 2.40% 0.971 0.000 0.000

 

C and S denote for castor and soybean oil. : sample collected in the experiment, b :

time (minutes) after fixed-bed starts, °: number of bed volumes, d : total oil concentration,
e' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , (31,0, Cm

181

 

CIC 0, C1/C1,o, C 2/C 2,0

 

 

 

 

0 3 6 9 12

Bed volume

Figure A.17 Fixed-bed Adsorption Using Methanol and SD-2 (Run # 29)
6: total oil, 0: hydroxyl, A: non-hydroxy.

182

 

 

l‘l'JL.

Run # 30: In Propanol, SD-2 = 2.838 g, flowrate = 0.20 ml/min, C = 2.38 %, S = O %

 

 

“Fraction ”Time °Bed Vol “oil % °C./Co "c.1c1,o 902/02,.I
1 18 0.72 0.04% 0.015 - -
2 33 1.31 0.76% 0.314 0.292 0.469
3 48 1.91 1.58% 0.648 0.654 0.612
4 64 2.55 2.00% 0.823 0.827 0.792
5 80 3.18 2.09% 0.860 - -
6 98 3.90 2.17% 0.894 0.907 0.806
7 114 4.54 2.22% 0.914 - -
8 129 5.13 2.28% 0.939 - -
9 143 5.69 2.29% 0.941 - -
10 161 6.41 2.32% 0.955 - -
11 185 7.36 2.31% 0.951 - -
12 207 8.24 2.32% 0.955 0.964 0.892
13 229 9.12 2.33% 0.957 - -
14 253 10.07 2.36% 0.971 0.975 0.944

 

7

 

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, °: number of bed volumes, d : total oil concentration,

9' f‘ 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial Co , C1,0, C2,o.

183

 

Run # 31: In Propanol, SD-2 = 2.856 g, flowrate = 0.16 ml/min, C = 1.63 %, S =0.84 %

 

 

‘Fraction bTime “Bed Vol “oil % °c1co 'c,1c,,., fi'czlcz,0
1 17 0.54 0.00% 0.000 - -
2 36 1.15 0.51% 0.207 0.297 0.075
3 54 1.73 1.34% 0.540 0.735 0.254
4 75 2.40 1.78% 0.718 0.897 0.456
5 95 3.04 2.04% 0.821 - -
6 121 3.87 2.18% 0.880 1.027 0.664
7 144 4.60 2.29% 0.923 - -
8 168 5.37 2.33% 0.941 0.944 0.936
9 193 6.17 2.40% 0.968 - -
10 213 6.81 2.40% 0.967 0.988 0.937
11 238 7.61 2.40% 0.969 - -
12 263 8.41 2.43% 0.979 1.038 0.894
13 287 9.18 2.49% 1.003 - -
14 317 10.14 2.47% 0.995 - -

 

 

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, c: number of bed volumes, d : total oil concentration,

°' f' 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , C10, Cm

 

184

Run # 32: In Propanol, SD-2 = 2.835 g, flowrate = 0.19 ml/min, C = 0.81 %, S =1.63 %

 

 

‘Fraction “Time °Bed VOI “oil % °c1co 'c.Ic.,0 gczlcz,0
1 17 0.65 0.00% 0.000 - -
2 34 1.30 0.41% 0.172 0.316 0.113
3 52 1.99 1.20% 0.498 0.657 0.433
4 70 2.67 1.65% 0.686 0.777 0.649
5 88 3.36 1.92% 0.800 - -
6 110 4.20 2.11% 0.878 0.941 0.851
7 131 5.00 2.21% 0.922 - -
8 149 5.69 2.28% 0.951 0.978 0.939
9 176 6.72 2.33% 0.972 0.995 0.963
10 198 7.56 2.36% 0.984 - -
11 218 8.32 2.37% 0.989 0.975 0.994
12 242 9.24 2.39% 0.994 - -
13 265 10.12 2.38% 0.993 0.952 1.009
15 291 11.11 2.41% 1.003 1.037 0.989

 

a : sample collected in the experiment, b :

C and S denote for castor and soybean oil.
time (minutes) after fixed-bed starts, C: number of bed volumes, d : total oil concentration,

6' f’ 9: concentrations of total oil (C), ricinolein (C1), and non-hydroxylated component

(C2) in the effluent compared to their initial C0 , C10, C10.

185

 

l

1.4
1.2

l

0.8 7
0.6 -
0.4 -
0.2 7

0.0 " I 1 1
0 3 6 9 12

CIC 0, 01/6 1.0, C ZIC 2,0

 

 

 

 

Bed volume

Figure A.18 Fixed-bed Adsorption Using Propanol and SD-2 (Run # 30)
9: total oil, O: hydroxyl, A: non-hydroxy.

186

 

 

 

CIC 0, C1/C1,o, C 2/C 2,0

 

 

 

 

0.0 " 1 r r
0 3 6 9 12

Bed volume

 

Figure A.19 Fixed-bed Adsorption Using Propanol and SD-2 (Run # 31)
0: total oil, O: hydroxyl, A: non-hydroxy.

 

 

 

 

 

 

1.4 —
3- 1.2-
R _
0_ 1.0
3- 0.8 —
Q
5. 0.6—
;; 0.49
0 0.2—

0.0 — 1 1 T

0 3 6 9 12

Bed volume

Figure A.20 Fixed-bed Adsorption Using Propanol and SD-2 (Run # 32)
0: total oil, O: hydroxyl, A: non-hydroxy.

187

 

 

 

 

 

1.4 —
5- 1.2 ~
n
o 1.0 _
3- 0.8 —
Q
5 0.6 -
2° 0.4 ~
° 0.2 —
0.0 ‘ 1 1 7
O 3 6 9 12
Bed volume

Figure A.19 Fixed-bed Adsorption Using Propanol and SD-2 (Run # 31)
6: total oil, O: hydroxyl, A: non-hydroxy.

 

 

 

 

 

1.4 -
.3-
£3
0
o:
2
5
a
Q
o
0.0 1 7 1
0 3 6 9 12
Bed volume

Figure A.20 Fixed-bed Adsorption Using Propanol and SD-2 (Run # 32)
6: total oil, O: hydroxyl, A: non-hydroxy.

187

 

 

 

APPENDIX B

FORTRAN PROGRAM
(Reference in Chapter 7)

188

Program Opt2olEster

l**********************************************************************

1 There are 5 subroutines, and two functions named and used in the following order:
! 1. ConvertData.f90

! 2. CalcSlope.f90

! 3. DungErrCalcf90

! 4. ErrFunction.f90: Function error.f90

! 5. CachnPsat.f90: Function lnPErr ()

! 6. WriteParmsHb3.f90

! 7. WriteRegEoK.f90

l***********************************************************************

USE MSFLIB !To use anQ and changeDIRQQ
USE MSIMSL !To use IMSL functions

implicit none

character compName* 16(200)

integer k(20),L,n,nComponent,nPoint,NumAdj_20l,NumAdj_Ester,Nout,option

double precision T(200),ln_expP(200), 1n_preP(200), AveExpSlope(200),
AveSpeadSlope(200)

logical status, result

common/C/n,k,CompName,nComponent,nPoint
common/WT,ln_expP,ln_preP,AveExpS10pe,AveSpeadSlope
common/P/Paraml ,Param2

common/O/option

parameter (N umAdj_201=5, NumAdj_Ester=9)

integer IPARAM(7), IBTYPE

double precision PsatErr, RPARAM(7), XLB1(NumAdj_201), XUBl(NumAdj_201),
ParamGuessl (N umAdj_201), Param1(NumAdj_201), ParamScalel (N umAdj_201),
ErrScalel , XLB2(NumAdj_Ester), XUBZ(NumAdj_Ester), ErrScaleZ,
ParamGuess2(NumAdj_Ester), Pararn2(NumAdj_Ester), ParamScale2(NumAdj_Ester)

data ParamGuessl/0.0000368dO,130.d0,4.3d0,152.d0,30.d0/, ErrSca1e1/0.000001 dO/,
ParamScalel/ 1.DO, 1.D-8, 1.D-7, 5.D-9, 1.D-8/

data XLB1/.OOOOODO, 0.D0, O.DO, O.D0,0.DO/

data XUBl/.OOOO6D0, 200.D0, 10.DO, 200.D0,60.D0/

data ParamGuessZ/0.0000]573d0,104.65d0,.8d0,10.8D0, 0.6D0, 100.2d0,5.d0,
152.6d0,44.7d0/, ErrScaleZ/0.000000IdO/, ParamScale2/ 1.D-3, 1.D-8, 1.D-7, 5.D-9, 1.D-
7,5.D-9, 1.D-8, 5.D-9, 1.D-8/

data XLBZ/.ODO, O.DO, .ODO, 0.D0, .ODO, 0.DO, 0.D0, O.D0,0.DO/

data XUB2/.002DO, 140.D0, 3.DO, 20.DO, 5.D0, 200.DO,40.D0, 200.D0,100.DO/

external DungErrCalc
IBTYPE=O

189

 

call DU4INF(IPARAM, RPARAM)

open (7,f11e='C:\spead\Calchs\Input\Parmsrecord.txt')

write(7,*) (IPARAM(L),L=1,7)

write(7,*) (RPARAM(L),L=1,7)

close(7)

1*“ Set non default values for desired IPARAM and RPARARM elements. Using the
same set up as in the example UMINF/DUMINF

write(*,*) 'Enter (1): optimization of 2-018‘

write(*,*) 'Enter (2): optimization of esters'

read (*,*) option

if (option .EQ.1) then

call DBCONF(DungErrCalc,NumAdj_2ol,ParamGuess1 ,IBTYPE,XLB 1 ,
XUB 1 ,ParamScalel ,ErrScalel ,IPARAM,RPARAM,Param1,PsatErr)

call WriteParmsHb3 ( 1 4,4,param 1)

call WriteRegEok(] 4,4, paraml (4),paraml (5))

status = CHANGEDIRQQ('C:\Ddept\BaseAlcohol\2013')

result=RUNQQ('C:\Ddept\BaseAlcohol\201s\RegStep53v.Exe','-1 -r')

status = CHANGEDIRQQCC:\Temp\dung\try')

else if (option .EQ.2) then
call DBCONF(DungErrCalc,NumAdj_Ester,ParamGuess2,IBTYPE,XLBZ,
XUBZ,ParamScale2,ErrScale2,IPARAM,RPARAM,Param2,PsatErr)
call WriteParmsHb3( 1 6,2,param2)
ca11 WriteRegEok(9,4, param2(4),param2(5))
call WriteRegEok( 15,2, param2(6),param2(7))
call WriteRegEok(16,2, param2(8),param2(9))
status = CHANGEDIRQQCC:\Ddept\BaseEsters')
result=RUNQQ('C:\Ddept\BaseEsters\RegSteps3v.Exe','-l -r')
status = CHANGEDIRQQCC:\Temp\dung\try')
endif

call UMACH (2, NOUT)

write (N OUT,100) PsatErr, (IPARAM(L),L=3,5)

100 format (' The Psat Error','value is', F 15.3, H, ' The number of iterations is’,I3, //,' The
number of function evaluations is ', I3, //, ' The number of gradient evaluations is', 13)

End

l*****IIHIUIHI‘**************************************************************

Subroutine ConvertData

! Reading data from TPRhoResults.txt

! Converting text type to numeric, calculating and saving T,P,Rho, T"(-1.3), 1nP(exp),
ln(Pspead) in the Convert.txt

190

implicit none

character colHead* 16(200,6),compName*16(200),Name* 16(200)

integer i,j,k(20),n,nComponent,nPoint,option

double precision T(200),ln_expP(200), 1n_preP(200), AveExpSlope(200),
AveSpeadSlope(200), col(200,5)

common/C/n,k,CompName,nComponent,nPoint
common/WT,ln_expP,ln__preP,AveExpSlope,AveSpeadSlope

common/O/option

i=0
n=0
k(1)=1
nPoint=0
if (option == 1) then
open( 1 ,file='C :\Ddept\BaseAlcohol\201s\TPRhoResu1ts .txt')
else if (option == 2) then
open(] ,flle=’C :\Ddept\BaseEsters\TPRhoResults.txt')
endif
open(2,file='c:\temp\dung\try\Convert.txt')

do while (.not.EOF(l))
i=i+1
read(1,*) (colHead(i,j),j=1 ,6)
compName(i)= colHead(i, 1)
if (i .EQ.1) then
write (2,201) (colHead(i,j),j=l,6),' (1/Texp)"(-1.3), ln(exp_P), ln(pre_P)'
else
do j=2,6
read(colHead(i,j),*) col (i,j)
enddo
T(i)= (col (i,2))**(-1.3)
ln_expP(i)= log(col (i,3))
ln_preP(i)=log(col(i,5))
write (2,202) colHead(i,1),(col(i,j),j=2,6), T(i),ln_expP(i),ln_preP(i)
nPoint=nPoint+l
endif
lcounting number of data point for each component. This info is written at the end of the
file Covert.txt.
if ((i .GT. 1) .and. (compName(i)/=compName(i-l))) then

n=n+1
k(n)=1
else
k(n)=k(n)+ 1
endif

Name(n)=CompName(i)

191

enddo
write (2,204)'* * "‘ * * * * * * *','There are ',nPoint,' data points and ',n- 1 , ' components.’
write(2,*)'Below is number of data points for each component:'

do i=2,n
write(2,203)i-1,Name(i),k(i)
enddo
nComponent=n
close(l)
!close(2) will be closed in ErrFunction
201 format (A12,A8,A10,A15,A10,A10,A40)
202 format(Al6,F5.1,5X,4F8.5,5X,E13.6,5X,2F13.6)
203 format (I3,5x,A16,I3)
204 format (//A10//Al 1,13,A17,12,A12)

End subroutine ConvertData
l*********************************IIUIIIIUIUI‘*Ilnli********#*********************

Subroutine CalcSlope
!Calculating the average experimental and predicted slopes. Writing info onto inter1.txt

implicit none

character CompName* 16(200)

integer i,j,k(20),m,n,nComponent,nPoint

double precision SumExpSlope(200),SumSpeadSlope(200),ln_expP(200), 1n_preP(200),
AveExpSlope(200),AveSpeadSlope(200),ExpSlope(200),SpeadSlope(200),T(200)
common/C/n,k,CompName,nComponent,nPoint
common/V/T,ln_expP,ln_preP,AveExpSlope,AveSpeadSlope

m=1

do i=1,nComponent
SumSpeadSlope(i)=O.dO
SumExpSlope(i)=0.d0

if (k(i)>1) then
do j=2,l<(i)
!calculate and compare the average experimental slope with predicted
m=m+1
if (T(m+1)-T(m) ==O.d0) then
k(i)=k(i)-1
ExpSlope(m)=(ln_expP(m+2)-ln_expP(m+ 1 ))/(T(m+2)-
T(m+1)) ~
SpeadSlope(m)=(ln_preP(m+2)-ln_preP(m+1))/(T(m+2)-
T(m+1))
else
ExpSlope(m)=(ln_expP(m+ 1 )—ln_expP(m))/(T(m+ 1 )-T(m))
SpeadSlope(m)=(ln_preP(m+1 )-ln_preP(m))/(T(m+1)-
T(m))

192

endif
SumSpeadSlope(i)=SumSpeadSlope(i)+ SpeadSlope(m)
SumExpSlope(i)=SumExpSlope(i)+ ExpSlope(m)

enddo

AveExpSlope(i)=SumExpSlope(i)/(k(i)-1)

AveSpeadSlope(i)=SumSpeadSlope(i)/(k(i)-1 )

endif
enddo

open(3,file=’c:\temp\dung\try\inter1 .txt')

!unit 3 will be kept open until error is evaluated in ErrFunction

do i=1,nComponent
write(3,301)AveExpSlope(i),AveSpeadSlope(i)

enddo

301 format (2F25.8)

End subroutine CalcSlope
1**********************************************************************

Subroutine DungErrCalcCNumAdj,Parms,PsatErr)
! Reading data from TPRhoResults.txt. Converting text type data to numeric, calculating
and saving T,P,Rho, T"(-1.3), 1nP(exp), 1n(Pspead), and calculating slope

USE MSF LIB !To use anQ and changeDIRQQ

USE MSIMSL !To use IMSL functions

implicit none

character compName* 16(200)

integer k(20),n,nComponent,nPoint,NumAdj,option

double precision T(200), ln_expP(200), 1n_preP(200), AveExpSlope(200), PsatErr,
PsatErrl, PsatErr2, lnPErr,AveSpeadSlope(200), error, Param1(5), Param2(9), Parms(9)
common/C/n,k,CompName,nComponent,nPoint
common/V/T,ln_expP,ln_preP,AveExpSlope,AveSpeadSlope

common/P/Param 1 ,Param2

common/O/option

external error, 1nPErr

logical result, status

if (option .EQ.1) then
NumAdj =5
call WriteParmsHb3(14,4,Parms)
call WriteRegEok( 14,4, param1(4),param1(5))
status = CHANGEDIRQQCC:\Ddept\BaseAlcohol\2018')
result=RUNQQ('C:\Ddept\BaseAlcohol\2013\RegSteps3v.Exe','-1 -r')
status = CHANGEDIRQQCC:\Temp\dung\try')
else if (option .EQ.2) then
NumAdj =9

193

call WriteParmsHb3(16,2,Parms)
call WriteRegEok(9,4, param2(4),param2(5))
call WriteRegEok(15,2, param2(6),param2(7))
call WriteRegEok(16,2, param2(8),param2(9))
status = CHANGEDIRQQCC:\Ddept\BaseEsters')
result=RUNQQ('C :\Ddept\BaseEsters\RegSteps3v.Exe','- 1 -r')
status = CHANGEDIRQQCC:\Temp\dung\try')
endif
call ConvertData
call CalcSlope
PsatErr1= error(3 ,AveExpSlOpe,AveSpeadSlope,nComponent)
PsatErr2= lnPErr()
PsatErr=PsatErr1 *PsatErr2
open (7,file='C:\spead\Calchs\Input\Parmsrecord.txt',access='append’)
write(7,*)PsatErr1, PsatErr2, PsatErr
write(7,*)
close(7)

End subroutine DungErrCalc
l***********************************************************************

Function error(location,parml ,parm2,times)

integer location, times
double precision parm1(200), parm2(200)

rewind (location)
error=0.d0
do i=1 ,times
read(location,*)parm1 (i),parm2(i)

error=erroi+abs(parm1 (i)-parm2(i))
enddo

close(3)
return

End function error
l***********************************************************************

Function lnPErr” ()
implicit none
character CompName* 16(200)

integer i,k(20),n, nComponent, nPoint

double precision lnPErr, ln_expP(200), 1n _preP(200), AveExpSlope(200),

194

AveSpeadSlope(200), T(200)

common/C/n,k,CompName,nComponent,nPoint
common/V/T,ln_expP,ln_preP,AveExpSlope,AveSpeadSlope

lnPErr=0

do i=1 ,nPoint
lnPErr=lnPErr+abs(ln_expP(i)-ln_preP(i))
enddo

End function lnPErr

l****#******************************************************************

Subroutine WriteParmsHb3(main,sub,parms)
Use MSFLIB

character note*10(200)

integer mainType(200),subType(200), nDs(200),nAs(200),sub,main
integer countline

double precision bondVol(200),bondSlope(200),bondEnergy(200),parms(9)

open (6,file='C:\spead\Calchs\Input\ParmsHb3.txt')
open (7,file='C:\spead\Calchs\Input\Parmsrecord.txt',access='append')

rewind(6)

i=0
countline=0
do while (.not.EOF(6))
i=i+1 '
countline=countline+ 1
if (i .EQ.1) then
read (6,*)mainType(i)
else
read(6,*) mainType(i),subType(i), nDs(i),nAs(i),bondVol(i),
bondSlope(i), bondEnergy(i),note(i)
if ((mainType(i)==main) .and.(subType(i)==sub)) then
if ((parms(1)*parms(2)*parms(3)) .GT. 0.D0) then
bondVol(i)= parms(1)
bondSlope(i) = panns(2)
bondEnergy(i)=parms(3)
endif

write(7,602)mainType(i),subType(i), nDs(i),nAs(i),bondVol(i),

bondSlope(i), bondEnergy(i),note(i)
endif

195

 

 

 

endif
enddo

rewind(6)
write (6,601)mainType(1),'nDs nAs bondVNm3 bVolSlo eHchal_mol'
do i=2,countline

write(6,602)mainType(i),subType(i), nDs(i),nAs(i),bondVol(i), bondSlope(i),
bondEnergy(i),note(i)
enddo

close(6)
close(7)

601 format (I3,A40)
602 format(4l3,3x,F15.8,3x, 2F15.3,3x,3A10)

End subroutine WriteParmsHb3
l******************************Il‘****IIUIIill***IlHlIIII***************************

Subroutine WriteRegEoK(ID1 ,ID2,parm1,parm2)

implicit none

character note* 10(200)

integer IDl ,ID2,option,count,i,mainType(200),subType(200),step1(200),step4(200)
double precision parml, parm2, eoleigh(200), eoleow(200), eok4High(200),
eok4Low(200)

common/O/option

if (option ==1) then
open (4,file='C:\Ddept\BaseAlcohol\2018\RegEoK.txt')
open (5,file='C:\Ddept\BaseAlcohol\Zols\RegEoKrecord.txt',access='append')
else if (option ==2) then
open (4,file=’C:\Ddept\BaseEsters\RegEoK.txt')
open (5,file='C:\Ddept\BaseEsters\RegEoKrecord.txt',access='append')
endif
open (7,file='C:\spead\Calchs\Input\Parmsrecord.txt',access='append')

i=0
count=0
do while (.not.EOF(4))
i=i+1
count=count+ l
read(4,*)note(i)
if (index(note(i),'#') .EQ. 0)then
backspace (4)
read

(4,*)mainType(i),subType(i),eok1High(i),eok1Low(i),step1(i),eok4High(i),eok4Low(i),st

196

 

 

ep4(i)
if ((mainType(i)==ID1) .and.(subType(i)==ID2))then
if (parml .GT. parm2 .and. (parml*parm2 .GT. 0.D0)) then
eoleigh(i)=parm1
eok4High(i)=parm2
eoleow(i)= eoleigh(i)
eok4Low(i)= eok4High(i)
write (5,201) mainType(i),subType(i),
eokl High(i),eok1 Low(i),step] (i),eok4High(i),eok4Low(i),step4(i)
wn'te (7,201) mainType(i),subType(i),
eok 1 High(i),eok1 Low(i),step 1 (i),eok4High(i),eok4Low(i),step4(i)
endif
endif
endif
enddo

rewind(4)
write(4,101)'#The epsilons to regress. The type description is in the SiteParms.tXt file.’
write(4,102)"#Please put these in ascending order. I'm not going to bother writing code"
write(4,103)'#to sort these things into the proper order.‘
write(4,104)'#Main Type Sub Type eokl Low eoleigh eokl Step
eok4Low eok4High eok4Step'
i=5
do i=5,count-1
write (4,201) mainType(i),subType(i),
eoklHigh(i),eok1Low(i),step1(i),eok4High(i),eok4Low(i),step4(i)
enddo
write (4,105)'#END'

101 format(a77)

102 format(a75)

103 format(a44)

104 format(a71)

105 format(a4)

201 format (214,2F 1 5.3,14,3x,2F1 5.3,14)
close(4)

close(S)

close(7)

End subroutine WriteRegEoK

I***********************************************************************

197

 

APPENDIX C

MISCELLANEOUS DOCUMENTATION
(Reference in Chapter 7)

198

C.1 - Vapor Pressure Data (predicted and experimental)

The tables in this section are ouput files (TptCoeff.txt) from SPEAD and the

FORTRAN program. Notations are described bellows:

compName: Name of component (referred to Tables 7.1 and 7.4).
TextpK: experimental temperature (°K)

PexpMPa: experimental vapor pressure (MPa)

RhoLexpG__cc: experimental liquid density (g/cm3)

Ptpt: SPEAD predicted vapor pressure (MPa)

Rhotpt: SPEAD predicted liquid density (g/cm3)

C.1.1 — Training -OH Site

 

compName , Tepr , PexpMPa , RhoLexpG_cc , Ptpt , Rhotpt

201C3.dat , 415.8 , .707000 , .6461 , .675817 , .6543
201C3.dat , 407.3 , .579000 , .6583 , .543495 , .6653
201C3.dat , 403.6 , .516000 , .6635 , .491923 , .6700
201C3.dat , 400.0 , .466000 , .6684 , .445497 , .6745
201C3.dat , 395.9 , .412000 , .6739 , .396436 , .6797
201C3.dat , 392.6 , .374000 , .6782 , .361008 , .6837
201C3.dat , 386.5 , .308000 , .6861 , .300577 , .6911
201C3.dat , 382.7 , .272000 , .6908 , .267468 , .6957
201C3.dat , 375.0 , .210000 , .7003 , .208824 , .7048
201C3.dat , 374.1 , .205000 , .7014 , .202622 , .7059
201C3.dat , 364.9 , .147000 , .7122 , .147967 , .7165
201C3.dat , 363.1 , .136000 , .7143 , .138799 , .7185
201C3.dat , 362.4 , .133000 , .7151 , .135144 , .7194
201C3.dat , 359.7 , .120000 , .7182 , .122408 , .7224
201C3.dat , 355.1 , .100000 , .7235 , .102924 , .7276
201C3.dat , 354.8 , .099000 , .7238 , .101673 , .7280
201C3.dat , 353.1 , .092200 , .7256 , .095564 , .7298
201C3.dat , 350.0 , .081300 , .7291 , .084468 , .7333
201C3.dat , 349.6 , .080000 , .7296 , .083137 , .7337
201C3.dat , 345.3 , .066600 , .7343 , .069818 , .7385
201C3.dat , 343.1 , .060600 , .7366 , .063923 , .7408
201C3.dat , 340.2 , .053400 , .7397 , .056533 , .7440
201C3.dat , 333.9 , .040000 , .7464 , .042936 , .7507
201C3.dat , 333.1 , .038500 , .7472 , .041420 , .7516
201C3.dat , 325.5 , .026500 , .7552 , .028989 , .7597
201C3.dat , 325.0 , .025900 , .7557 , .028343 , .7602
201C3.dat , 323.1 , .023600 , .7576 , .025915 , .7622
201C3.dat , 313.1 , .014100 , .7676 , .015600 , .7726

199

201GB.
201GB.
201C3.
201C3.
2olC3.
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
201C4.
.dat
.dat
201C4.
.dat
201C4.
.dat
.dat
.dat
201C5.
.dat
201C5.
201C5.
.dat
.dat
201C5.
.dat
201C5.
201C5.
201C5.
201C5.
201C5.
201C5.
201C5.
201C5.
201C5.
.dat

201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4
201C4

201C4
201C4

201C4
201C4
201C5
201C5
201C5
ZolCS

201C5

201C5

201C5

dat
dat
dat
dat
dat

dat

dat

dat

dat

dat
dat

dat

dat
dat
dat
dat
dat
dat
dat
dat
dat

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ~ ‘ ‘ ‘ N \ ‘ ‘ ‘ \ \

\ ‘

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

§

‘

‘ ‘ ‘ \ ‘

303.
300.
298.
293.
283.
453.
443.
443.
433.
432.
423.
422.
387.
382.
380.
377.
375.
372.
370.
367.
364.
363.
358.
356.
353.
349.
348.
344.
340.
339.
335.
331.
327.
323.
317.
311.
301.
391.
387.
383.
383.
383.
379.
375.
374.
374.
371.
367.
364.
364.
363.
359.
355.
353.
353.
351.
346.

C1304\IWU‘luD-OOLDQLDUJGDKDWDKO\IKDKDOKOONAOONKOUJKOQCDU'IAl—‘NkobxlmLfiWWNF-JQCDNle—‘l—‘Hl—‘OH

\ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘

‘

‘ ‘ ‘ ‘ \ ‘ ‘ ‘

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘

‘ ‘ ‘ ‘ ‘ \ ‘

‘ ‘ ‘ ‘ ‘ \ ‘

.007880
.006480
.005870
.004320
.002270
.993000
.794000
.794000
.624000
.623000
.491000
.474000
.169000
.143000
.133000
.121000
.114000
.101000
.093100
.084500
.073300
.070100
.057800
.053400
.047400
.039900
.038600
.031200
.026600
.025000
.019900
.016700
.013300
.010700
.008000
.005330
.003000
.099300
.085100
.074200
.073400
.073400
.064000
.054600
.052000
.052000
.046400
.039000
.034100
.034100
.032900
.027700
.022900
.021600
.021600
.018800
.015300

200

‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ \ \ ‘ ‘ ‘ \ \ § ‘ \ § ‘

\ ‘ ~ ‘ ‘ ‘ \ ‘ ‘ ‘

‘ \ \ \ ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

\

.7774
.7805
.7822
.7870
.7963
.6148
.6300
.6305
.6456
.6462
.6595
.6616
.7073
.7133
.7158
.7192
.7212
.7249
.7276
.7305
.7348
.7361
.7415
.7436
.7469
.7512
.7521
.7573
.7610
.7624
.7674
.7712
.7757
.7801
.7855
.7926
.8018
.7160
.7204
.7243
.7249
.7250
.7285
.7327
.7341
.7342
.7368
.7410
.7444
.7444
.7450
.7488
.7529
.7543
.7544
.7569
.7609

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘ N ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘ ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ N ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ \ \ ‘ ‘ ‘ 5 ‘ ‘ ‘ ‘ ‘ ‘ \ ‘

‘

.008998
.007495
.006716
.004951
.002586
.851507
.689902
.685073
.542036
.537211
.426711
.410726
.148148
.125749
.117193
.106219
.099914
.089305
.082152
.074696
.064907
.062120
.051375
.047575
.042228
.035731
.034503
.027993
.023931
.022556
.018046
.015136
.012161
.009769
.007365
.004946
.002814
.092845
.080343
.070352
.068910
.068885
.060878
.052191
.049331
.049312
.044543
.037672
.032688
.032674
.031950
.027013
.022479
.021004
.020995
.018564
.015231

‘ ‘ \ \

‘ ‘ ‘ ‘ ‘ S ‘ ‘

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ § ‘ ‘ ‘ § ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ N ‘ ‘ ‘ ‘ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘

.7828
.7860
.7879
.7929
.8029
.6445
.6569
.6573
.6700
.6705
.6819
.6837
.7250
.7306
.7329
.7360
.7380
.7414
.7439
.7467
.7507
.7519
.7571
.7591
.7622
.7663
.7672
.7721
.7757
.7770
.7818
.7855
.7899
.7941
.7992
.8061
.8151
.7336
.7384
.7426
.7432
.7432
.7470
.7516
.7532
.7532
.7561
.7607
.7644
.7644
.7650
.7693
.7738
.7754
.7754
.7783
.7828

 

201C5
201C5
201C5

201GB
201GB

201C5
201C5
201C5
201C5
201C5
201C5

201GB
201GB
201GB
201C8

201C8
201C8
201C8
201C8
201GB
201GB
201GB

201GB
201GB
201GB

.dat
.dat
.dat
201C5.
.dat
.dat
201C5.

dat

dat

.dat
.dat
.dat
.dat
.dat
.dat
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
201C6.
.dat
.dat
.dat
.dat
201C8.
201C8.
.dat
.dat
.dat
.dat
.dat
.dat
.dat
201C8.
.dat
.dat
.dat

dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat

dat
dat

dat

\ ‘ ‘ ‘ ‘ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘ V ‘ ‘ ‘ ~ ‘ \ ‘ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘

‘

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

343.
343.
342.
338.
335.
334.
333.
333.
330.
326.
323.
323.
322.
415.
413.
413.
408.
407.
401.
398.
396.
391.
388.
385.
380.
378.
375.
370.
368.
364.
358.
354.
349.
348.
345.
341.
338.
336.
328.
318.
473.
464.
453.
453.
447.
446.
442.
440.
434.
432.
427.
423.
421.
415.
413.
410.
404.

(DOKDCDQ\JmO'XOOCDle-‘wobUJI-Jl-‘kOI—‘Ol—‘I-‘ONUOH\Jl-‘Nl—‘l—‘sb\IHF-‘ONi-‘kOstle—‘NNNKObwwml-mebb

\ ‘ ‘ ‘ ‘ \ N V ‘ ‘ \ ‘ ‘ \

\ ‘ N ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ \ §

‘ ‘ ‘ ‘ ‘

‘ ‘ ‘ ‘ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘

\

.013200
.013200
.012300
.010100
.008000
.008230
.007720
.007720
.006670
.005264
.004320
.004320
.004008
.108190
.101630
.101390
.086286
.084227
.069861
.060662
.057912
.047425
.042730
.038501
.031216
.028838
.025077
.020232
.018665
.015822
.011466
.009617
.007576
.007066
.005970
.004770
.004400
.003787
.002507
.001333
.172000
.137000
.103000
.102000
.085800
.084300
.075600
.069400
.057700
.055200
.047200
.041400
.038400
.031300
.029400
.025300
.020800

201

‘ ‘ ~ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘ \ ‘ ‘ ‘

‘ ‘ ‘ fi ‘ ‘ ‘ N \ ‘ ‘ ~ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ § ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘ ‘ ‘ \ ‘ ‘ ‘

.7641
.7641
.7650
.7686
.7718
.7721
.7735
.7735
.7761
.7792
.7826
.7826
.7835
.7013
.7033
.7034
.7086
.7094
.7151
.7189
.7205
.7260
.7289
.7314
.7365
.7387
.7416
.7464
.7483
.7515
.7577
.7612
.7655
.7668
.7696
.7733
.7759
.7770
.7847
.7934
.6619
.6713
.6825
.6829
.6888
.6894
.6932
.6959
.7017
.7031
.7077
.7115
.7135
.7189
.7206
.7243
.7290

‘ ‘ ‘ § ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘ ‘ ‘ ‘ ‘ \ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘

‘ ‘ ‘ ‘ ‘ ‘ \

.012976
.012970
.012344
.010200
.008538
.008365
.007734
.007730
.006614
.005485
.004442
.004440
.004176
.105188
.099371
.099009
.084789
.082933
.069359
.061154
.057952
.047894
.043053
.039290
.032206
.029509
.026189
.021365
.019636
.016928
.012646
.010580
.008440
.007854
.006743
.005454
.004687
.004386
.002676
.001455
.165363
.132844
.099930
.098964
.084026
.082624
.074252
.068403
.057292
.054781
.047169
.041463
.038716
.031859
.029886
.026024
.021533

‘ ‘ ~ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ V ‘ ‘ ‘ ‘ ‘ \ ‘

‘ ‘ ‘ \ ‘ ~ ‘ ‘ N ‘ ‘ ‘ ‘

\

‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘ ‘

‘ \

‘

\ ‘ ‘ ‘ \

.7863
.7863
.7874
.7914
.7950
.7954
.7969
.7969
.7999
.8034
.8072
.8072
.8083
.7172
.7192
.7193
.7247
.7254
.7314
.7354
.7371
.7429
.7460
.7487
.7542
.7566
.7598
.7650
.7671
.7707
.7775
.7814
.7863
.7878
.7909
.7951
.7980
.7992
.8081
.8181
.6820
.6906
.7011
.7014
.7072
.7077
.7113
.7141
.7198
.7212
.7258
.7296
.7316
.7372
.7390
.7428
.7478

 

201C8.dat , 403.9 , .020100 , .7297 , .020907 , .7486
201C8.dat , 400.1 , .017400 , .7332 , .018131 , .7522
201C8.dat , 398.0 , .015800 , .7351 , .016692 , .7543
201C8.dat , 394.8 , .014000 , .7379 , .014727 , .7574
201C8.dat , 392.7 , .012700 , .7398 , .013564 , .7594
201C8.dat , 391.8 , .012300 , .7406 , .013061 , .7603
201C8.dat , 388.2 , .010500 , .7437 , .011304 , .7637
201C8.dat , 386.6 , .009720 , .7451 , .010567 , .7652
201C8.dat , 384.4 , .008840 , .7470 , .009639 , .7673
201C8.dat , 380.8 , .007450 , .7502 , .008236 , .7708
201C8.dat , 379.6 , .007130 , .7512 , .007814 , .7720
201C8.dat , 376.5 , .006090 , .7538 , .006821 , .7749
201C8.dat , 373.9 , .005420 , .7561 , .006046 , .7775
201C8.dat , 371.5 , .004760 , .7581 , .005414 , .7798
201C8.dat , 367.5 , .003830 , .7615 , .004473 , .7836
201C8.dat , 366.3 , .003700 , .7625 , .004232 , .7847
201C9.dat , 364.1 , .001733 , .7664 , .001836 , .7889
201C9.dat , 358.1 , .001333 , .7714 , .001331 , .7945
(3.1.2 - Testing —OH Site

compName , Tepr , PexpMPa , RhoLexpG_cc , Ptpt , Rhotpt
201C7.dat , 424.9 , .084121 , .7008 , .078249 , .7180
201C7.dat , 410.4 , .053077 , .7153 , .049515 , .7329
201C7.dat , 397.3 , .033489 , .7281 , .031323 , .7463
201C7.dat , 385.1 , .021130 , .7396 , .019709 , .7587
201C7.dat , 374.3 , .013332 , .7496 , .012614 , .7697
201C7.dat , 364.3 , .008413 , .7586 , .008066 , .7798
201C7.dat , 354.9 , .005307 , .7669 , .005152 , .7892
201C7.dat , 346.1 , .003349 , .7746 , .003266 , .7980
201C7.dat , 337.9 , .002113 , .7817 , .002072 , .8062
3olC5.dat , 245.1 , .000007 , .8619 , .000006 , .8892
3olC5.dat , 249.8 , .000013 , .8580 , .000010 , .8848
301C5.dat , 258.1 , .000033 , .8510 , .000025 , .8771
3olC5.dat , 262.8 , .000053 , .8470 , .000040 , .8727
3olC5.dat , 267.4 , .000084 , .8431 , .000062 , .8684
3olC5.dat , 283.4 , .000353 , .8292 , .000254 , .8533
301C5.dat , 290.3 , .000627 , .8231 , .000440 , .8467
3olC5.dat , 293.8 , .000807 , .8200 , .000574 , .8433
301C5.dat , 317.7 , .003839 , .7982 , .002908 , .8200
3olC5.dat , 321.8 , .004922 , .7943 , .003729 , .8159
3olC5.dat , 325.5 , .006091 , .7908 , .004618 , .8122
3olC5.dat , 329.1 , .007452 , .7874 , .005654 , .8086
301C5.dat , 334.0 , .009707 , .7828 , .007372 , .8037
3OlC5.dat , 338.7 , .012416 , .7782 , .009440 , .7989
3olC5.dat , 343.5 , .015793 , .7735 , .012012 , .7940
3olC5.dat , 348.3 , .019930 , .7688 , .015189 , .7890
3olC5.dat , 353.0 , .024810 , .7641 , .018942 , .7841
301C5.dat , 358.1 , .031050 , .7590 , .023763 , .7788
3olC5.dat , 363.0 , .038285 , .7539 , .029390 , .7736
301C5.dat , 368.0 , .046950 , .7488 , .036148 , .7683
3olC5.dat , 373.4 , .057937 , .7432 , .044780 , .7626
3olC5.dat , 378.2 , .069435 , .7381 , .053895 , .7575
301C5.dat , 384.1 , .086016 , .7317 , .067112 , .7510
301C5.dat , 388.7 , .100990 , .7267 , .079047 , .7460
301C6.dat , 398.1 , .071700 , .7209 , .064848 , .7429

202

301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6.
301C6
301C6.
301C6.
301C6.
301C6
301C7
301C7
301C7
301C7
301C7
301C7

dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat

.dat

dat
dat
dat

.dat
.dat
.dat
.dat
.dat
.dat
.dat
c201_2_C1C6.
c201_2_C1C6.
c201_2_C1C6.
c201_4_C1C6.
c201_4_C1C6.
c201C6.
c201C6.
c201C6.
c201C6.
c201C6.
c201C6.
c201C6.
c201C6.
c201C6.
c201C6.
C201C6.
c201C6.
c201C6.
c201C6.
c201C6.

dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat
dat

dat
dat
dat
dat
dat

N N N N N N N N N N N N N N N N N N N

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

388.
378.
370.
368.
360.
358.
356.
353.
348.
341.
339.
338.
333.
330.
318.
313.
311.
308.
305.
303.
299.
296.
293.
290.
287.
284.
283.
281.
278.
273.
263.
429.
429.
429.
339.
337.
303.
440.
338.
333.
441.
345.
444.
440.
434.
434.
433.
428.
428.
424.
421.
418.
415.
415.
411.
409.
404.

KOebsbNO'N\IkDCOCDKOKONQOUTmmml-‘l—‘le—‘HH\Ol—‘HUJUJNUJ-bwwwwmsbbLfiUJl—‘Ol—‘l—‘Ol—‘l—‘Ol—‘l—‘l—‘l—‘Ol—‘l—J

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N

.050760
.035200
.032700
.023410
.014500
.015300
.016600
.013300
.009530
.008000
.007280
.005870
.004900
.003270
.001549
.001121
.001020
.000821
.000656
.000563
.000423
.000327
.000253
.000200
.000152
.000119
.000120
.000090
.000067
.000051
.000022
.100000
.098100
.098730
.002400
.002200
.000196
.100390
.002133
.001600
.099325
.002000
.137140
.120710
.103620
.101850
.099192
.086046
.086544
.076327
.069487
.062690
.056449
.055883
.049053
.045356
.038859

203

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

.7310
.7409
.7487
.7505
.7581
.7599
.7618
.7647
.7692
.7755
.7775
.7782
.7827
.7855
.7958
.8001
.8016
.8042
.8068
.8087
.8119
.8144
.8169
.8194
.8218
.8244
.8253
.8268
.8293
.8334
.8415
.6899
.6907
.6908
.7827
.7841
.8160
.7936
.8964
.9006
.7776
.8760
.8086
.8134
.8189
.8195
.8198
.8250
.8251
.8293
.8322
.8355
.8387
.8390
.8428
.8449
.8493

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N N N N N N N N N N N N

N

.045425
.030971
.022174
.020495
.014386
.013121
.011951
.010284
.008100
.005643
.005029
.004802
.003637
.003035
.001470
.001066
.000947
.000764
.000618
.000526
.000394
.000313
.000248
.000194
.000153
.000117
.000106
.000090
.000069
.000043
.000016
.091404
.089406
.089142
.002162
.001983
.000194
.116149
.002020
.001547
.114277
.002834
.159662
.141446
.122457
.120484
.119439
.103566
.103326
.091797
.084149
.076250
.069187
.068488
.060629
.056631
.048746

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N

.7536
.7642
.7728
.7748
.7831
.7852
.7872
.7905
.7955
.8026
.8048
.8056
.8107
.8139
.8257
.8305
.8322
.8352
.8382
.8403
.8441
.8470
.8498
.8527
.8555
.8584
.8595
.8613
.8641
.8689
.8783
.7211
.7219
.7220
.8127
.8142
.8478
.7937
.8978
.9024
.7875
.8848
.8122
.8171
.8228
.8234
.8237
.8291
.8292
.8335
.8366
.8400
.8433
.8437
.8477
.8499
.8546

 

 

c201C6.dat , 404.1 , .037543 , .8501 , .047432 , .8555
c201C6.dat , 398.2 , .030307 , .8559 , .038648 , .8617
c201C6.dat , 393.5 , .025221 , .8605 , .032544 , .8667
c201C6.dat , 393.5 , .025265 , .8605 , .032544 , .8667
c201C6.dat , 389.9 , .021813 , .8640 , .028470 , .8704
c201C6.dat , 385.9 , .018252 , .8679 , .024353 , .8747
c201C6.dat , 385.8 , .018418 , .8680 , .024286 , .8747
c201C6.dat , 381.1 , .015041 , .8725 , .020150 , .8797
c201C6.dat , 378.0 , .013159 , .8754 , .017764 , .8829
c201C6.dat , 375.3 , .011609 , .8780 , .015855 , .8857
c201C6.dat , 372.9 , .010426 , .8803 , .014325 , .8882
c201C6.dat , 371.7 , .009872 , .8814 , .013610 , .8894
c201C6.dat , 367.9 , .008251 , .8849 , .011543 , .8933
c201C6.dat , 366.0 , .007522 , .8867 , .010611 , .8952
c201C6.dat , 361.9 , .006125 , .8905 , .008783 , .8995
c201C6.dat , 357.1 , .004803 , .8949 , .007014 , .9043
c201C6.dat , 354.3 , .004130 , .8975 , .006101 , .9072
c201C6.dat , 350.8 , .003420 , .9007 , .005125 , .9108
diolC4.dat , 468.1 , .159990 , .8002 , .042083 , .8739
diolC4.dat , 462.1 , .133320 , .8085 , .033824 , .8794
diolC4.dat , 459.1 , .119990 , .8125 , .030234 , .8821
diolC4.dat , 455.1 , .106660 , .8179 , .025952 , .8857
diolC4.dat , 454.6 , .103990 , .8187 , .025357 , .8862
diolC4.dat , 454.1 , .102660 , .8192 , .024966 , .8865
diolC4.dat , 453.9 , .101320 , .8196 , .024676 , .8868
diolC4.dat , 453.4 , .099992 , .8201 , .024295 , .8872
diolC4.dat , 453.1 , .098658 , .8207 , .023918 , .8875
diolC4.dat , 452.1 , .095992 , .8218 , .023088 , .8883
diolC4.dat , 451.4 , .093325 , .8229 , .022371 , .8890
diolC4.dat , 447.1 , .079993 , .8284 , .018909 , .8927
diolC4.dat , 441.1 , .066661 , .8361 , .014761 , .8980
diolC4.dat , 435.1 , .053329 , .8437 , .011417 , .9032
diolC4.dat , 423.1 , .033331 , .8586 , .006634 , .9134
diolC4.dat , 417.1 , .026664 , .8659 , .004979 , .9185
diolC4.dat , 410.1 , .019998 , .8743 , .003513 , .9243
diolC4.dat , 389.1 , .007999 , .8986 , .001119 , .9416
diolC4.dat , 381.1 , .005333 , .9076 , .000693 , .9480
diolC4.dat , 375.1 , .004000 , .9143 , .000476 , .9528
diolC4.dat , 367.1 , .002666 , .9231 , .000282 , .9592
diolC4.dat , 355.1 , .001333 , .9360 , .000121 , .9686
(3.1.3 - Training -COO- Site

compName , Tepr , PexpMPa , RhoLexpG_cc , Ptpt , Rhotpt
C3ateC2.dat , 533.1 , 2.776400 , .4773 , 2.971444 , .5752
C3ateC2.dat , 523.1 , 2.395800 , .5189 , 2.577247 , .5980
C3ateC2.dat , 513.1 , 2.056500 , .5510 , 2.223511 , .6184
C3ateC2.dat , 503.1 , 1.752500 , .5780 , 1.906773 , .6371
C3ateC2.dat , 493.1 , 1.492500 , .6019 , 1.624263 , .6546
C3ateC2.dat , 483.1 , 1.260700 , .6235 , 1.373538 , .6711
C3ateC2.dat , 473.1 , 1.057400 , .6433 , 1.152323 , .6869
C3ateC2.dat , 463.1 , .882460 , .6619 , .958436 , .7021
C3ateC2.dat , 453.1 , .731540 , .6794 , .789748 , .7168
C3ateC2.dat , 443.1 , .600620 , .6960 , .644176 , .7311
C3ateC2.dat , 433.1 , .487560 , .7118 , .519671 , .7451

204

 

C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC2
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4
C3ateC4

C3ateC4
C3ateC4
C3ateC4
C3ateC4
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1

C4ateC1
C4ateC1

.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
C3ateC4.
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
C4ateC1.
.dat
.dat

dat

dat

N N N N N

N

N N N N N N N N N N N N N N N N N N N N N

N N N N N N N N N N N N N N N N N N N N N N N

N

423.
413.
403.
393.
383.
373.
363.
353.
343.
333.
323.
313.
303.
293.
283.
273.
263.
417.
416.
416.
415.
414.
414.
413.
412.
412.
411.
410.
410.
409.
408.
407.
407.
406.
405.
404.
403.
403.
366.
359.
354.
349.
343.
337.
330.
323.
317.
309.
305.
513.
503.
493.
483.
473.
463.
453.
443.

i—Ii—li—aHHi—Ii—ii—imbHambwmbmHommmi—Ji—Ibi—ai—Imqwi—Immi—I00.50415HHHHHHI—ii—ii—H—lh—ii—il—Ii—JHHI—J

N N N N N N N

N N N N N N N N N N

N

N N N N N N N N N N N N

N

N N N N N N N N N N N

N

N N N N N N N N N N N

N

1—11—41—11—4

.389300
.308770
.240110
.184780
.139720
.104660
.075927
.053809
.037317
.025065
.016399
.010386
.006366
.003700
.002073
.001107
.000540
.099490
.097520
.095690
.094120
.092620
.090860
.089160
.087430
.085730
.083920
.082370
.081220
.078650
.076530
.074990
.074200
.072120
.070310
.068880
.067300
.065820
.019732
.014932
.011732
.009333
.006933
.005200
.003733
.002533
.001733
.001133
.000867
.897200
.613900
.371200
.157800
.971520
.808330
.669280
.548090

205

N N N N N N N N N N N N N N N N N

N

N

N N N N N N N N N N N N N

N N N N N N N N

N N N N

N N N N N N N N N N N N N

.7269
.7415
.7556
.7693
.7825
.7954
.8080
.8203
.8322
.8440
.8555
.8668
.8779
.8888
.8995
.9100
.9204
.7511
.7519
.7527
.7533
.7541
.7548
.7556
.7563
.7571
.7580
.7587
.7592
.7605
.7616
.7623
.7627
.7638
.7647
.7655
.7663
.7672
.8062
.8125
.8179
.8224
.8290
.8346
.8412
.8478
.8541
.8612
.8648
.5854
.6093
.6307
.6504
.6687
.6858
.7021
.7175

N N N N N N N N N N N N

N N N N N N N N N N N N N N N N N

N N

N N N N N N N

N

N N N N N N N N N N N N N

N N N N

N

l—‘l—‘l—‘l—i

.414226
.325885
.252748
.192988
.144859
.106712
.077006
.054323
.037374
.025013
.016236
.010187
.006156
.003567
.001973
.001035
.000512
.101984
.099904
.097942
.096262
.094522
.092753
.090931
.089085
.087320
.085381
.083725
.082512
.079816
.077542
.075963
.075137
.072970
.071028
.069556
.067919
.066312
.018108
.014095
.011231
.009233
.006816
.005178
.003696
.002564
.001784
.001148
.000906
.938385
.649130
.393173
.167923
.970943
.799897
.652530
.526654

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N

.7587
.7721
.7852
.7982
.8110
.8236
.8362
.8486
.8610
.8733
.8856
.8979
.9101
.9224
.9347
.9470
.9594
.7766
.7774
.7782
.7789
.7796
.7803
.7811
.7819
.7827
.7835
.7842
.7848
.7860
.7871
.7879
.7883
.7894
.7903
.7911
.7920
.7928
.8337
.8405
.8464
.8513
.8586
.8650
.8724
.8801
.8873
.8956
.8999
.6568
.6732
.6889
.7039
.7183
.7323
.7459
.7592

 

C4ateC1
C4ateC1
C4ateC1
C4ateC1

C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC1
C4ateC2
C4ateC2
C4ateC2
C4ateC2
C4ateC2
C4ateC2
C4ateC2
C4ateC2
C4ateC2

C4ateC3

C4ateC3
C4ateC3
C4ateC3
C4ateC3
C4ateC3
C4ateC3
C4ateC3

C4ateC3

C4ateC3

C4ateC3

.dat
.dat
.dat
.dat
C4ateC1.
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
C4ateC3.
C4ateC3.
.dat
C4ateC3.
C4ateC3.
C4ateC3.
C4ateC3.
C4ateC3.
.dat
.dat
.dat
C4ateC3.
.dat
C4ateC3.
.dat
.dat
.dat
C4ateC3.
.dat
C4ateC3.
C4ateC3.
.dat
C4ateC3.
C4ateC3.
.dat
C4ateC3.

dat

dat
dat

dat
dat
dat

dat
dat

dat

dat

dat

dat
dat

dat
dat

dat

N N N N N N N N N N N N N N

N N N N N N

N

N

N N N N N N N

N N N N N N N N

N N N N N N N N N N N N

N N

N

N N N N N

433.
423.
413.
403.
393.
383.
375.
375.
375.
373.
363.
353.
343.
333.
323.
313.
303.
302.
293.
283.
273.
263.
392.
394.
392.
392.
373.
339.
321.
321.
298.
415.
412.
412.
411.
410.
409.
408.
407.
406.
405.
404.
404.
402.
400.
398.
395.
394.
392.
391.
389.
381.
373.
370.
366.
364.
358.

Akal—‘Ol—l\Il-JU'INCDI—‘bkaomem\IOONWQCDl—‘kookobbbkaI—IHi—JF-‘COl—‘i—‘l—Jl—‘i—‘l—‘l—‘l—Jibmebl—‘l-‘Hi—‘HH

N N N N N N N N N N N N N N

N N N N N N N N N N N N N N N N N N N N N N N

N

N

N N N N N N N N N N

.443700
.354240
.279980
.216910
.166390
.125460
.101720
.101590
.101500
.093419
.067594
.048183
.033370
.022331
.014619
.009226
.005593
.005333
.003273
.001840
.000973
.000473
.101240
.100920
.096285
.095445
.052782
.015945
.006693
.006666
.002267
.101450
.093430
.092480
.090680
.088160
.085730
.083260
.079940
.078860
.076830
.074900
.073290
.068970
.065500
.061320
.057170
.054400
.051310
.049200
.046990
.033690
.025330
.022740
.020160
.018430
.014520

206

N N N N N N N N N N N N N N

N N N N N N N N N N N N N N N

N N N N N N N N N N

N N N N N N N N N N

N N

N

N N N N N

.7323
.7464
.7601
.7733
.7861
.7985
.8079
.8076
.8079
.8106
.8223
.8338
.8451
.8561
.8669
.8774
.8878
.8882
.8980
.9080
.9179
.9276
.7719
.7700
.7724
.7725
.7945
.8312
.8499
.8499
.8739
.7460
.7496
.7500
.7508
.7519
.7530
.7543
.7559
.7564
.7574
.7584
.7592
.7615
.7635
.7661
.7687
.7704
.7723
.7738
.7753
.7847
.7933
.7964
.7998
.8023
.8086

N N N N N N N N N

N N N N N

N N N N N N N N N N N N N N N N N N N N N N N

N

N

N N N N N N N N

.420150
.330972
.257153
.196815
.148178
.109568
.085552
.086254
.085552
.079436
.056355
.039038
.026342
.017267
.010962
.006716
.003957
.003870
.002231
.001198
.000609
.000291
.098826
.103568
.097482
.097186
.052692
.014445
.006274
.006274
.001721
.102267
.093531
.092523
.090693
.088055
.085554
.082813
.079443
.078368
.076230
.074163
.072608
.068164
.064488
.059837
.055568
.052764
.049894
.047627
.045444
.033713
.025146
.022490
.019852
.018056
.014102

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N N N N N N N N N N N N

N N

N

N N N N N

.7722
.7849
.7975
.8098
.8221
.8341
.8434
.8431
.8434
.8461
.8579
.8697
.8815
.8932
.9048
.9165
.9281
.9286
.9398
.9514
.9632
.9750
.8016
.7997
.8021
.8022
.8244
.8633
.8841
.8841
.9116
.7776
.7810
.7815
.7822
.7833
.7844
.7856
.7871
.7876
.7886
.7896
.7903
.7925
.7945
.7970
.7996
.8013
.8031
.8046
.8062
.8155
.8241
.8273
.8307
.8333
.8398

 

 

 

C4ateC3.dat , 357.3 , .013910 , .8097 , .013484 , .8410
C4ateC3.dat , 354.8 , .012580 , .8123 , .012144 , .8436
iC4atelC4.dat , 441.3 , .174650 , .6879 , .206712 , .7120
iC4atelC4.dat , 432.9 , .142250 , .6983 , .168105 , .7215
iC4atelC4.dat , 423.9 , .112390 , .7095 , .132498 , .7316
iC4atelC4.dat , 414.8 , .088259 , .7206 , .103103 , .7417
iC4atelC4.dat , 409.6 , .076260 , .7268 , .088786 , .7474
iC4atelC4.dat , 401.1 , .058795 , .7367 , .069037 , .7566
iC4ateIC4.dat , 393.4 , .046263 , .7457 , .054182 , .7649
iC4atelC4.dat , 382.6 , .032397 , .7582 , .037661 , .7766
iC4ateIC4.dat , 378.9 , .028531 , .7623 , .033208 , .7805
iC4atelC4.dat , 373.6 , .023731 , .7684 , .027342 , .7862
iC4ateIC4.dat , 367.1 , .018665 , .7756 , .021517 , .7930
iC4atelC4.dat , 358.9 , .013332 , .7847 , .015586 , .8017
iC4atelC4.dat , 354.4 , .010932 , .7897 , .012902 , .8066
iC4atelC4.dat , 347.9 , .007999 , .7968 , .009774 , .8135
iC4atelC4.dat , 338.4 , .005733 , .8070 , .006390 , .8234
iC4atelC4.dat , 298.1 , .000640 , .8494 , .000707 , .8661
iC4atelC4.dat , 293.1 , .000427 , .8546 , .000511 , .8715
C10ateC1.dat , 433.8 , .013332 , .7523 , .013152 , .7906
C10ateC1.dat , 427.9 , .010666 , .7577 , .010606 , .7958
C10ateC1.dat , 421.8 , .008186 , .7634 , .008357 , .8013
C10ateC1.dat , 415.9 , .006666 , .7688 , .006632 , .8064
C10ateC1.dat , 410.9 , .005333 , .7733 , .005398 , .8109
C10ateC1.dat , 403.8 , .003986 , .7798 , .003967 , .8173
C10ateC1.dat , 395.9 , .002866 , .7868 , .002797 , .8242
C10ateC1.dat , 387.1 , .002000 , .7946 , .001845 , .8319
C10ateC1.dat , 380.4 , .001440 , .8004 , .001323 , .8379
C10ateC1.dat , 369.8 , .000800 , .8096 , .000753 , .8473
C10ateC1.dat , 366.6 , .000667 , .8124 , .000632 , .8502
C10ateC1.dat , 362.3 , .000533 , .8161 , .000495 , .8540
C10ateC1.dat , 350.1 , .000267 , .8263 , .000240 , .8648
C10ateC1.dat , 339.1 , .000133 , .8354 , .000118 , .8746
C10ateC1.dat , 331.6 , .000080 , .8416 , .000070 , .8813
C10ateC1.dat , 326.1 , .000053 , .8462 , .000047 , .8864
C10ateC1.dat , 316.9 , .000027 , .8536 , .000023 , .8946
C10ateC1.dat , 308.6 , .000013 , .8603 , .000012 , .9021
C.1.4 — Testing —COO— site

compName , Tepr , PexpMPa , RhoLexpG_cc , Ptpt , Rhotpt
C3ateC3.dat , 394.3 , .101720 , .7712 , .103306 , .7982
C3ateC3.dat , 395.3 , .100660 , .7699 , .106452 , .7970
C3ateC3.dat , 292.6 , .001333 , .8825 , .001224 , .9168
C4ateC4.dat , 438.6 , .098100 , .7309 , .106271 , .7555
C4ateC4.dat , 328.3 , .001730 , .8390 , .001425 , .8660
iC4ateC3.dat , 407.1 , .100260 , .7439 , .116323 , .7607
C5ateC4.dat , 457.1 , .100000 , .7118 , .107715 , .7404
C5ateC4.dat , 291.1 , .000047 , .8691 , .000045 , .9043
iC5ateC2.dat , 298.1 , .001053 , .8611 , .000861 , .9040
C12ateC1.dat , 452.0 , .007848 , .7397 , .008647 , .7791
C12ateC1.dat , 442.0 , .005357 , .7485 , .005894 , .7876
C12ateC1.dat , 432.0 , .003597 , .7573 , .003920 , .7960
C12ateC1.dat , 422.0 , .002343 , .7660 , .002541 , .8045
C12ateC1.dat , 412.0 , .001476 , .7745 , .001606 , .8129

207

C12ateC1
C12ateC1
C12ateC1
C12ateC1
C12ateC1
C12ateC1
C12ateC1
C12ateC1
C12ateC1
C12ateIC3
C12ateIC3
C12ateIC3
C12ateIC3
C12ateIC3
C12ateIC3

C12ateIC3.

C12atelC4
C12atelC4
C12atelC4
C12atelC4
C12atelC4
C12atelC4

C12atelC4.

C12atelC4
C12atelC4
C12atelC4
C12atelC4
C12ate2C2
C12ateZC2
C12ate2C2
C12ate2C2
C12ate2C2
C12ate2C2
C12ate2C2
C12ate2C2
C12ate2C2
C24ateC1
C24ateC1
C24ateC1
C24ateC1
C24ateC1
C24ateC1

.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat
.dat

.dat
.dat
.dat
.dat
.dat
.dat
dat
.dat
.dat
.dat
.dat
.dat
.dat
dat
.dat
.dat
.dat
.dat
C6.dat
C6.dat
C6.dat
C6.dat
C6.dat
C6.dat
C6.dat
C6.dat
C6.dat

.dat
.dat
.dat
.dat
.dat
.dat

N N N N N N N N N N N N N N N N N N N N N N

N

N N N N

N N N N N N N N

411.
401.
391.
381.
371.
362.
355.
345.
335.
452.
442.
432.
422.
412.
401.
391.
452.
442.
432.
412.
402.
391.
381.
371.
361.
355.
345.
452.
442.
432.
422.
412.
402.
392.
381.
371.
452.
442.
442.
432.
432.
422.

i-‘NNNLONONCDOl—‘OOOOOLOACD\JflmOHNNNKOKOOUJl—‘l—‘l—‘UTWUT-DLDKDCDCDKD

N N N N N N N N N N N N N N N N N N N

N N N N N

N N N N N N N N N N

N

.001600
.000969
.000611
.000357
.000203
.000111
.000074
.000037
.000018
.004698
.003219
.002104
.001405
.000885
.000535
.000313
.002591
.001726
.001137
.000433
.000255
.000142
.000078
.000041
.000020
.000013
.000006
.000472
.000290
.000173
.000099
.000055
.000030
.000015
.000007
.000003
.000017
.000008
.000008
.000003
.000003
.000001

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N N

N

N N N N N N N N N N

208

1—‘1—a1—41—11—41—J

.7745
.7831
.7914
.7995
.8081
.8154
.8210
.8289
.8368
.7735
.7806
.7877
.7945
.8015
.8084
.8151
.7457
.7582
.7623
.7684
.7756
.7847
.7897
.7968
.8070
.8494
.4543
.7675
.7748
.7821
.7894
.7966
.8037
.8109
.8181
.8252
.7881
.8220
.8220
.8555
.8555
.8883

N N N N N N N N N

N N N N N

N N N N N N N N N N N N N N N N N N N N N N N N N

N

.001604
.000978
.000581
.000336
.000181
.000102
.000064
.000032
.000015
.005215
.003481
.002266
.001450
.000885
.000526
.000304
.002835
.001845
.001168
.000427
.000245
.000135
.000072
.000037
.000018
.000011
.000005
.000409
.000241
.000138
.000076
.000040
.000021
.000010
.000005
.000002
.000010
.000005
.000005
.000002
.000002
.000001

N N N N N N N N N

N N N N N

N N N N N N N N N N N N N N N N N N N N N N N N N N N N

.8129
.8215
.8299
.8382
.8470
.8547
.8606
.8691
.8777
.7617
.7699
.7781
.7861
.7945
.8027
.8109
.7650
.7730
.7810
.7971
.8052
.8134
.8215
.8296
.8377
.8429
.8511
.7833
.7908
.7983
.8058
.8133
.8208
.8284
.8361
.8438
.8023
.8093
.8093
.8164
.8164
.8235

 

C.2 — Structure of Ethyl Lactate and its Oligomers

o 0 CH3
H3C /\ H3C 0 CH3
0 CH3 0 V
OH OH O
Ethyl lactate Ethyl lactate Dimer
ethyl 2-hydroxypropanoate 2-ethoxy- l -methyl-2-oxoethyl 2-hydroxypropanoate

0 CH3 0
H3C o
%OXW YKOACM
OH 0 CH3

Ethyl lactate T rimer
2-(2-ethoxy- 1 -methyl-2-oxoethoxy)-1-methyl-2-oxoethyl 2-hydroxypropanoate

0 CH3 0 CH3
H3C o 0 CH3
Vlko/kn/ \‘/U\O)\’( \/
OH 0 CH3 0

Ethyl lactate T etramer
2— [2-(2-ethoxy- 1 -methyl-2-oxoethoxy)-1-methyl-2-oxoethoxy]-1-methyl—2-oxoethyl 2-

hydroxypropanoate
I) CH3 0 CH3 0
I 0/l\n/ %OJY WAC/\CW
OH 0 CH3 0 CH3
Ethyl lactate Pentamer

Ethyl 14-hydroxy—2,5,8,1 l-tetramethyl-
4,7,10,1 3-tetraoxo-3 ,6,9,12-tetraoxapentadecan-1 —oate

209

C.3 — The .m3d Files used in Psat Predictions

631 - Ethyl Lactate Dimer. 3md

1. Interaction sites, n= 13

 

 

 

 

Bond radius Group-coordinates Group ID
x Y Z Maiquroup Sub group

0235 0.65588 0.00000 0.00000 14 4
0.390 0.52621 0.05991 -0.02242 3 3
0. 363 0.40938 0.11646 0.04292 1 2
0_ 345 0.47677 -0.02100 -0.13928 9 4
0270 0.49160 -0.05429 -0.25944 16 2
0270 0.46124 -0.14741 -0.08840 15 2
0.390 0.33557 -0.17536 -0.03036 3 20
0. 345 0.25117 -0.20569 -0.14276 9 4
0.363 0.32966 -0.30924 0.04143 1 2
0270 0.15709 -0.10846 -0.15196 15 2
0.270 0.21697 -0.26136 -0.24874 16 2
0.357 0.03174 -0.11630 -0.10087 2 1
0.363 0.00000 0.00000 0.00000 1 1

2. Bond Matrix

0 1 2 3 4 5 6 7 8 9 10 11

1 2

2 4 2

3 4 2 4

4 6 4 6 2

5 6 4 6 2 4

6 6 6 6 4 6 2

7 6 6 6 6 6 4 2

8 6 6 6 6 6 4 2 4

9 6 6 6 6 6 6 4 2 6

10 6 6 6 6 6 6 4 2 6 4

1166666664626

12666666666462

210

C.3.2 — Ethyl Lactate Trimer. 3md

1. interaction sites, n= 18

 

 

 

 

Bond radius Group-coordinates Group ID
x y 2 Main group Suquroup

0.285 0.82500 -0.10234 0.22100 14 4
0.390 0.81527 -0.04031 0.09983 3 3
0.363 0.91205 0.00000 0.00000 1 2
0.345 0.73146 0.07980 0.13413 9 4
0.270 0.73418 0.17905 0.21498 16 2
0.270 0.63387 0.10412 0.04838 15 2
0.390 0.53727 0.00527 0.02053 3 20
0. 345 0.42286 0.08670 -0.03130 9 4
0.363 0.54661 -0.11142 -0.08145 1 2
0.270 0.31415 0.04083 0.03105 15 2
0.270 0.38275 0.18021 -0.10763 16 2
0.390 0.28141 0.03173 0.16284 3 20
0.345 0.15079 -0.03662 0.16725 9 4
0.363 0.34997 -0.01765 0.28444 1 2
0.270 0.09165 -0.13605 0.22339 16 2
0.270 0.05192 0.04560 0.12031 15 2
0.357 0.00000 0.00000 0.00000 2 1
0.363 0.05158 —0.12339 -0.07990 1 1

2. Bond Matrix
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 2

2 4 2

3 4 2 4

4 6 4 6 2

5 6 4 6 2 4

6 6 6 6 4 6 2

7 6 6 6 6 6 4 2

8 6 6 6 6 6 4 2 4

9 6 6 6 6 6 6 4 2 6

10 6 6 6 6 6 6 4 2 6 4

11 6 6 6 6 6 6 6 4 6 2 6

12 6 6 6 6 6 6 6 6 6 4 6 2

13 6 6 6 6 6 6 6 6 6 4 6 2 4

14 6 6 6 6 6 6 6 6 6 6 6 4 2 6

15 6 6 6 6 6 6 6 6 6 6 6 4 2 6 4

16 6 6 6 6 6 6 6 6 6 6 6 6 4 6 6 2

17 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 4 2

211

 

C.3.3 - Ethyl Lactate Tetramer. 3md

1. Interaction sites, 11 = 23

 

 

 

 

Bond radius Group-coordinates Group ID
x y 2 Main group Sub floup
0.363 1 .18538 0.00000 0.00000 1 1
0.285 0.21606 -0.00092 0.13672 14 4
0.390 0.15150 -0.01614 0.01336 3 3
0.363 0.00000 0.00000 0.00000 1 2
0.345 0.16659 -0.16314 -0.00495 9 4
0.270 0.13132 -0.27999 0.03315 16 2
0.270 0.26707 -0.19608 -0.08527 15 2
0.390 0.40473 -0.16235 -0.04931 3 20
0.345 0.46757 -0.23992 -0.15746 9 4
0.363 0.45018 -0.02212 -0.08116 1 2
0.270 0.59382 -0.26893 -0.11509 15 2
0.270 0.46812 -0.29037 -0.27660 16 2
0.390 0.62454 -0.34400 -0.00107 3 20
0.345 0.71075 -0.24401 0.06991 9 4
0.363 0.52397 -0.39057 0.09579 1 2
0.270 0.70730 -0.13888 0.14405 16 2
0.270 0.82769 -0.30396 0.08193 15 2
0.390 0.91459 -0.29869 -0.03879 3 20
0.363 0.88646 -0.42092 -0.13489 1 2
0.345 1.02727 -0.35253 0.04474 9 4
0.270 1.08707 -0.43113 0.12028 16 2
0.270 1.09711 -0.24145 0.07588 15 2
0.357 1.15519 -0.15309 -0.01526 2 1

 

212

2. Bond Matrix

.21

066

12

3642

46424

566462

6664624

76666462

 

866666642

9666666424
106666666426

66666664264
12666666664626

11

136666666666462

1466666666664624

15666666666666426

166666666666664264

1766666666666664662

18666666666666666642

196666666666666666424

2066666666666666666462

466666666666666664624
222666666666666666666462

21

213

0.3.4 - Ethyl Lactate Pentamer. 3md

1. Interaction sites, n = 28

 

 

 

 

Bond radius GrouL-coordinates GrouplD
x y 2 Main group Sub group
0.345 0.08076 -0.09219 0.03199 9 4
0.285 0.99207 0.25405 -0.00236 14 4
0.390 0.98110 0.16311 -0.10459 3 3
0.363 1.00510 0.13702 -0.25105 1 2
0.345 1.05847 0.05614 -0.03250 9 4
0.270 1.17048 0.00000 0.00000 16 2
0.270 0.98713 -0.00689 0.05169 15 2
0.390 0.87015 -0.07672 -0.00130 3 20
0.345 0.80529 -0.08187 0.13130 9 4
0.363 0.90450 -0.22855 -0.03154 1 2
0.270 0.67346 -0.09869 0.11164 15 2
0.270 0.81811 -0.04284 0.25128 16 2
0.390 0.60069 -0.21425 0.10818 3 20
0.345 0.56016 -0.22269 0.25166 9 4
0.363 0.62052 -0.35008 0.05876 1 2
0.270 0.60637 0.26066 0.36467 16 2
0.270 0.44013 —0.17823 0.28202 15 2
0.390 0.41802 -0.04203 0.25083 3 20
0.363 0.50561 0.07024 0.31366 1 2
0.345 0.28714 -0.03116 0.30773 9 4
0.270 0.20969 0.01436 0.39889 16 2
0.270 0.19267 -0.04209 0.20844 15 2
0.390 0.15397 -0.15525 0.14194 3 20
0.363 0.05175 -0.25290 0.21374 1 2
0.270 0.14117 -0.11408 -0.08815 15 2
0.270 0.00000 0.00000 0.00000 16 2
0.357 0.09034 0.21832 -0.16249 2 1
0.363 0.15233 -0.31923 -0.24503 1 1

 

214

2. Bond Matrix

(OCDNCDU'I-wa-I
.—L
'N
O)

4.34.;
(JON—KO

MAJ—244.114
AOCOCDNCDU'I

62

642
6646
66464
666626
6666462

NNNN
01-wa

N _l
O) -h
GANNJANAmmmmmmmmmmmmmmmmmmmmo

cacao:mmmmmmmmmmmmmmmmmmmmc-c-N
mmmmmmmmmmmmmmmmmmmmmhhNM
mammmmmmmmmmmmmmmmmmmmma
0500503030300503030303020503030505CDCDANN
OQQGOOQQOCDCDCDGOJODGGOOCDCDA
030003000503050030303030505050348-510
mmmmmmmmmmmmmmmmbbmm
GODOOOGODODQOOCDCDOJCDNNNN
GODGOQGOQQOJOJOOJQCDCDQO)
CDCDOCDOCDCDODCDGODCDOJANNN
mmmmmmmmmmmmmmmm
mmmmmmmmmmmanmm
mammmmmmmmnmmn
OQGQOOQQOQQOQ
OGODCDOJOJOJCDQODCD-h

02020300303030)th

QOQGOOANNN

ODOQOJCDODODCD-h

ODOODODOJ-bNN

N
N]

215

 

0.3.5 - Dioxane. 3md (5-hydroxy-2-methyl-1,3-dioxane or 5HMD)

Interaction sites, n= 8

 

 

 

 

Bond radius Group-coordinates Group ID
x y 2 Main group Sub group
0.39 0.38061 0.01813 0.06592 3 1
0.27 0.30861 -0.11056 0.03568 15 1
0.357 0.21754 -0.10558 -0.07610 2 1
0.39 0.12777 0.02395 -0.06059 3 1
0.357 0.22643 0.14493 -0.06764 2 1
0.27 0.32506 0.13956 0.02941 15 1
0.363 0.51032 0.00000 0.00000 1 2
0.285 0.00000 0.00000 0.00000 14 4

0.3.6 - Dioxolane. 3md (4-hydroxymethyI-Z-methyl-1,3-dioxolane or 4HMD)

Interaction sites, n = 8

 

 

 

Bond radius Group-coordinates Group ID
x y 2 Main group Sub ggup

0.39 0.38970 0.02521 0.03615 3 4
0.27 0.35938 0.12730 -0.07242 15 1
0.357 0.25062 0.08419 -0.16080 2 1
0.39 0.09236 -0.08703 -0.05174 1 6
0.357 0.22984 -0.07035 -0.12540 3 4
0.27 0.31654 -0.09858 -0.00529 15 1
0.363 0.54307 0.00000 0.00000 1 2
0.285 0.00000 0.00000 0.00000 14 2

Bond Matrix (for 5HMD) Bond Matrix (for 4HMD)
0 1 2 3 4 5 6 0 1 2 3 4 5 6

1 2 1 2

2 4 2 2 4 2

3 6 4 2 3 6 6 4

4 4 6 4 2 4 4 4 2 2

5 2 4 6 4 2 5 2 4 4 4 2

6 2 4 6 6 6 4 6 2 4 6 6 6 4

7 6 6 4 2 4 6 6 7 6 6 4 2 4 6 6

216

10.

11.

12.

13.

LIST OF REFERENCES

Campell & Co., Castor oil & Castor Oil Derivatives, http://www.campbell-
uv.com/castor.html (Accessed December, 2005).

 

Baker, Urethanes - Castor oil - Chemical Derivatives, in CasChem Technical
Bulletin 100. 1982. p. 1-3.

Lysenko, 2., et al., Separation of Plant Oil Triglyceride Mixtures by Solid Bed
Adsorption. US. Patent 7,097,770 82, 2006.

Sonntag, N.O.V., Structure and Composition of Fats and Oils, in Bailey's
Industrial Oil and Fat Products, D. Swem, Editor. 1979, John Wiley & Sons:
New York. p. 1-98.

Smith, M., M. H., and L. Kunst, Production of Hydroxy Fatty Acids in The Seeds
of Arabidopsis Thaliana. Biochem. Soc. Trans, 2000. 28: p. 947-950.

Offeman, R.D., Stephenson, Serena K. , Robertson, George H., Orts, William J .,
Solvent Extraction of Ethanol from Aqueous Solutions Using Biobased Oils,

Alcohols, and Esters. Journal of the American Oil Chemists' Society, 2006. 83(2):
p.153-157.

Browse, J. and C. Somerville, Glycerolipid Synthesis: Biochemistry and
Regulation. Annual Review of Plant Physiology and Plant Molecular Biology,
1991. 42: p. 467-506.

Kates, M., Glycolipids, Phosphoglycolipids, and Sulfolipids. Handbook of Lipid
Research, ed. DJ. Hanahan. Vol. 6. 1990: Plenum Press. 520.

Haines, T.H., The Chemistry of The Sulfolipids. Prog. Chem. Fats Other Lipids,
1971.ll:p.297.

Benson, AA, The Plant Sulfolipid. Advances in Lipid Research, 1963: p. 387-
394.

Benson, J .S., Isolation and Fatty Acid Composition of the Plant Sulfolipid and
Galactolipids. Journal of Lipid Research, 1964. 5: p. 432-436.

Norman, H.A., Mischke, Charles F, Allen, Bertha, Vincent, James S., Semi-
Preparative Isolation of Plant Sulfoquinovosyldiacylglycerols by Solid Phase
Extraction and HPLC Procedure. Journal of Lipid Research, 1996. 37: p. 1372-
1375.

Mudd, B.J. and K.F. Kleppinger-Sparace, The Biochemistry of Plants, P.K.

217

14.

15.

16.

17.

18.

I9.

20.

21.

22.

23.

24.

25.

26.

Stumpf and BE. Conn, Editors. 1987, Academic Press: New York. p. 159-173.

Allen, C.F., Good, R, Davis, H.F., Chisum, P., Methodology for the Separation of
Plant Lipids and Application to Spinach Leaf and Chloroplast Lamellae. The
Journal of The American Oil Chemists' Society, 1966. 43: p. 223-231.

Benning, C., Biosysthesis and Function of The Sulfolipid Sulfoquinovosyl
Diacylglycerol. Annu. Rev. Plant Physiol. Plant Mol. Biol., 1998. 49: p. 53-75.

Johnson, AR. and J .B. Davenport, Biochemistry and Methodology of Lipids.
1971: Willey-Interscience. 578.

Kim, Y.H., J .S. Y00, and MS. Kim, Structural Characterization of
Sulfoquinovosyl, Monogalactosyl and Digalactosyl Diacylglycerols by FAB-CID-
MS/MS. Journal of Mass Spectrometry, 1997. 32: p. 968-977.

Carter, H.E., R.A. Hendry, and NZ. Stanacev, Wheat Flour Lipids. III. Structure
Determination of Mono- and Digalactosyl Diglycerides. J. Lipid Res., 1961. 2: p.
223.

Murray, R.K., M. Levine, and M.J. Komblatt, Sulfatides: Principle Glycolipids of
the Testis, Spermatozoa of Chordates, in Glycolipid Methodology, L.A. Witting,
Editor. 1976, American Oil Chemists' Society: Champaign, Illinois. p. 305-327.

Gustafson, K.R., et al., Aids-Antiviral Sulfolipidsfrom Cyanobacteria (Blue-
Green-Algae). Journal of the National Cancer Institute, 1989. 81(16): p. 1254-
1258.

Fishwick, M.J. and A.J. Wright, Comparison of Methods for the Extraction of
Plant Lipids. Biochemistry, 1977: p. 1507-1510.

Harwood, J .L., Synthesis of Sulfoquinovosyl Diacylglycerol by Higher Plants.
Biochimica et Biophysica Acta, 1975. 398: p. 224-230.

Barber J. and Goumaris K., What Role Does Sulphollipid Play within the
T hylakoid Membrane? Photosynth. Res., 1986. 9: p. 239-249.

Harwood, J .L., Jones A. L., Lipid Metabolism in Algae. Adv. Bot. Res., 1989. 16:
p. 1-53.

Andrews, J. and J. Ohlrogge, in Plant Physiology, Biochemistry and Molecular
Biology, D.T. Dennis and DH. Turpin, Editors. 1990, Longman Scientific &
Technical, Singapore. p. 339.

Kates, M., Plant Phospholipids and Glycolipids. Adv. Lipid Res, 1970. 8: p. 225.

218

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

Sigma-Aldrich, Technical Information Bulletin - Ion-exchange Resins and

Related polymeric Adsorbents,

http://www.sigmaaldrich.com/Brands/Aldrich/Tech Bulletins/AL 142/Amberlite
Amberlysthtml (Accessed September, 2006).

Krishnankutty, N. and MA. Vannice, Effect of Pretreatment on Surface Area,
Porosity, and Adsorption Properties of a Carbon Black. Chem. Matter, 1995. 7: p.
754-763.

Zweig, G. and J. Sherma, Lipids. CRC Handbook of Chromatography, ed. H.K.
Mangold. Vol. 1. 2000, Boca Raton, Florida: CRC Press, Inc. 601.

Douglas M. Ruthven, Principles of Adsorption and Adsorption Process. 1984,
New York: John Wiley & Sons, Inc. 433.

Yao, J .K. and GM. Rastetter, Microanalysis of Complex Tissue Lipids by High-
Performance Thin-layer Chromatography. Analytical Biochemistry, 1985.
150(1): p. 111-16.

The Dow Chemical Company, F ILMT EC TM Membranes, DOWEXTM Ion
Exchange Resins, http://www.dow.com/1iguidseps/grod/prd_dowx.htm (Accessed
September, 2006).

 

AOCS official method Ce 2-66: Preparation of Methyl Esters of Long-chain Fatty
Acids, in Official Methods and Recommended Practices of the American Oil
Chemists' Society, D. Firestone, Editor. 1994, AOCS: Champaign, IL.

Baber, T.M., D.T. Vu, and C.T. Lira, Liquid-Liquid Equilibrium of the Castor Oil
+ Soybean oil + Hexane Ternary System. Journal of Chemical and Engineering
Data, 2002. 47(6): p. 1502-1505.

Levesque, B.A., Pierre; Tardif, Robert; F erron, Liliane; Gingras, Suzanne;
Schlouch, Emmanuelle; Gingras, Guy; Levallois, Patrick; Dewailly, Eric., Cancer
Risk Associated with Household Exposure to Chloroform. Journal of Toxicology
and Environmental Health, Part A, 2002. 65(7): p. 489-502.

Hildebrand, J.H., Solubility. XII. Regular Solutions. J. Am. Chem. Soc, 1929.
51(1): p. 66-80.

Hildebrand, J .H., A History of Solution Theory. Annual Review of Physical
Chemistry, 1981. 32: p. 1-23.

Hansen, C.M., Three-Dimensional Solubility Parameter-Key to Paint-Component
Aflinities: I. Solvents, Plasticizers, Polymers, and Resins. Journal of Paint
Technology, 1967. 39: p. 104-17.

219

 

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

Burke, J ., Solubility Parameters: Theory and Application. The Book & Paper
Group Annual, 1984.3: p. 13-58.

Barton, A.F.M., CRC Handbook of Solubility Parameters and other Cohesion
Parameters. 1983: CRC Press. 594.

P. Dormann, S.H.-B., I. Balbo, and C. Benning, Isolation and Characterization of
an Arabidopsis Mutant Deficient in the Thylakoid Lipid Digalactosyl
Diacylglycerol. Plant Cell, 1995. 7: p. 1801-1810.

Siakotos, A.N., Analytical Separation of Nonlipid Water Soluble Substances and
Gangliosides from Other Lipids by Dextran Gel Column Chromatography.
journal of American Oil Chemists' Society, 1965. 42: p. 913-919.

Hartel, K., P. Dbrmann, and C. Benning, DGDI-Independent Biosynthesis of

Extraplastidic Galatolipids after Phosphate Deprivation in Arabidopsis. PNAS,
2000. 97(19): p. 10649-54.

Gage, D.A., Z.-H. Huang, and C. Benning, Comparison of Sulfoquinovosyl
Diacylglycerol from Spinach and the Purple Bacterium Rhodobacter sphaeroides
by Fast Atom Bombardment Tandem Mass Spectrometry. Lipids, 1992. 27(8): p.
632-636.

Ishizuka, 1., Suzuki, M.,Yamakawa, T, Isolation and Characterization of a Novel
Sulfoglycolipid, 'Seminolipid, ' from Boar Testis and Spermatozoa. J. Biochem.,
1973. 73(1): p. 77-87.

E. Choe, J .L., K. Park, S. Lee, Effects of Heat Treatment on Lipid and Pigments of
Freeze-Dried Spinach. Food Chemistry and Toxicology, 2001. 66(8): p. 1074-
1079.

Benson, A.A., Plant sulfolipid. V. Lysolipid Formation. Biochimica et Biophysica
Acta, 1962. 57: p. 601-603.

Benning, C., F ormic Acid in Extractions of Plant Lipids. A Personal
Communication, 2003.

O'Brien, J.S., F illerup, Dorothy L., Mead, James F ., Brain Lipids: 1.

Quantification and Fatty Acid Composition of Cerebroside Sulfate in Human
Cerebral Gray and White Matter. Journal of Lipid Research, 1964. 5: p. 109-116.

Thomson, G.H. and AH. Larsen, DIPPR: Satisfying Data Needs. J. Chem. Eng.
Data, 1996. 41: p. 930-934.

Dvoskin, N., Thermodynamic Equilibrium of New Organic Mixtures for
Absorption Heat Pumps, in Mechanical Engineering. 2004, Ben-Gurion

220

 

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

University. p. 121.
Hala, E., et al., Vapor Liquid Equilibrium. 2nd ed. 1967: Pergamon Press, Oxford.

Abbott, M.M., Low-Pressure Phase Equilibria: Measurement of VLE. Fluid
Phase Equilibria, 1986.29: p. 193-207.

Lyman et al., Vapor Pressure, in Handbook of Chemical Property Estimation
Methods. 1982. p. 14-1, 14-20.

Chiou, CT. and V.J. Freed, Chemical Studies on Bench Mark Industrial
Chemicals. Annaual Report to National Science Foundation on Contract
AEN7617700, Report No. NSF/RA-770286, National Science Foundation,
Washington DC, 1977.

Reid, R.C., J .M. Prausnitz, and T.K. Sherwood, The Properties of Gases and
Liquids, ed. rd. 1977, New York: McGraw-Hill Book Co.

Pandharipande, S.L., Laxminarayan Institute of Technology, and N. Nagpur
University, India,, Artificial neural network: Prediction of vapor pressure of
phenol series. Chemical Engineering World, 2004. 39(7): p. 70-72.

Miller, D.G., Estimating Vapor Pressure - a Comparison of Equations. Ind. Eng.
Chem, 1964. 56: p. 46.

Thomson, G.W., Techniques of Organic Chemistry, A. Weissberger, Editor. 1959,
Interscience: New York. p. 473.

F ishtine, S.H., Reliable Latent Heats of Vaporization. Ind. Eng. Chem, 1963.
55(47).

Kistiakovskii, V.A., Latent Heats of Vaporization. J. Russ. Phys. Chem. Soc.,
1921. 53(1): p. 256-64.

Nelken, L.H. and D. Birkett, Dipole Moment, in Handbook of Chemical Property
Estimation Methods. p. 25-1 - 25-22.

Watson, K.M., Thermodynamics of the Liquid State - Generalized Prediction of
Properties. Ind. Eng. Chem, 1943. 35: p. 398.

Korsten, H., Internally Consistent Prediction of Vapor Pressure and Related
Properties. Ind. Eng. Chem, 2000. 39: p. 813-820.

Yair, O. B. and A. Fredenslund, Extension of the UNIFAC Group-Contribution

Method for the Prediction of Pure Component Vapor Pressure. Ind. Eng. Chem.
Process Des. Dev., 1983. 22: p. 433-436.

221

 

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

Jensen, T., A. F redenslund, and P. Rasmussen, Pure-Component Vapor-Pressures
Using UNIFAC Group Contribution. Industrial & Engineering Chemistry
Fundamentals, 1981. 20(3): p. 239-246.

Gray, N.H., Z.N. Gerek, and J .R. Elliott, Molecular Modeling of Isomer Effects in
Naphthenic and Aromatic Hydrocarbons. Fluid Phase Equilibria, 2005. 228: p.
147-153.

N. Bureau, J .J ., I. Mokbel, and J-C. de Hemptinne, Vapour Pressure
Measurements and Prediction for Heavy Esters. J. Chem. Thermodynamics,
2001. 33(11): p. 1485-1498.

William. E. Asher, J .F .P., Garnet. B. Erdakos, and John. H. Seinfeld, Estimating
the vapor pressures of multi-functional oxygen-containing organic compounds
using group contribution methods. Atmospheric Environment, 2002. 36(9): p.
1483-1498.

Constantinou, L. and R. Gani, New Group Contribution Method for estimatiing
Properties of Pure Compounds. AIChE J ., 1994. 40(10): p. 1697-1709.

Joback, KC. and RC. Reid, Chem. Eng. Commun., 1987. 57: p. 233-243.

Aspen Technologies, Physical Property Methods and Models, in Aspen Physical
Property System 2004.1 User Manual. p. 8-4 - 8-5.

Riedel, LA., A New Universal Vapor Pressure Equation. Chem. Ing. tech, 1954.
26: p. 83.

D. Ambrose and J. Walton, Vapor-Pressure up to Their Critical Temperatures of
Normal Alkanes and Alkanols. Pure & Appl. Chem. 61: p. 1395-1403.

Aspen Technology, 1., Physical Property Reference Manuals. Aspen Physical
Property System 2004.1, April 2005.

T. B. Thompson (Ed.) at tbthompson@dow.com, Chemical Industry of the
Future, Technology Roadmap for Computational Chemistry. 1999.

Fiona Case, et al., The second industrial fluid properties simulation challenge.
Fluid Phase Equilibria, 2005. 236: p. 1-14.

Elliott, J .R., SPEAD 1.03. 1.04, Intro and Help.
J. R. Elliott, SPEAD Progress Report, June 2003 . 2003.

Prausnitz, J .M., R.N. Lichtenthaler, and E.G.d. Azevedo, Molecular

222

 

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

Thermodynamics of F laid-Phase Equilibria. 3rd ed. In The Physical and
Chemical Engineering Sciences, ed. P.H. International. 1999: Prentice-Hall, Inc.
860.

Chapela, G.A., L.E. Scriven, and HT. Davis, Molecular Dynamics for
Discontinuous Potential. IV. Lennard Jonesium. J. Chem. Phys., 1989. 91: p.
4307.

Mackie, A.D., et al., Vapour-liquid phase equilibria predictions of methane-
alkane mixtures by Monte Carlo simulation. Molecular Simulation, 1997. 19(1):
p. 1-15.

Goujon, F., Malfreyt, P., Boutin, A. and Fuchs, A. H., Vapour-liquid phase
equilibria of n-alkanes by direct Monte Carlo simulations. Molecular Simulation,
2001. 27(2): p. 99-114.

Cui, J .Y. and J .R. Elliott, Phase Diagrams for a Multistep Potential Model of n-
Alkanes by Discontinuous Molecular Dynamics and Thermodynamic Perturbation
Theory. Journal of Chemical Physics, 2002. 116(19): p. 8625-8631.

Rice, RC. and DD. Do, Approximate Solution Methods for ODE: Perturbation
Methods, in Applied Mathematics and Modeling for Chemical Engineers. 1995,
John Willey & Sons, Inc. p. 184-189.

Barker, J .A. and D. Henderson, Perturbation Theory and Equation of State for
Fluids: The Square-Well Potential. J. Chem. Phys., 1967. 47(8): p. 2856-2861.

Cui, J .Y. and J .R. Elliott, Phase envelopes for variable width square well chain
fluids. Journal of Chemical Physics, 2001. 114(16): p. 7283-7290.

George B. Thomas, J. and R.L. Finney, Calculus and Analytic Geometry. 2003,
Library of Congress Cataloging-in-Publication Data. p. 333.

Hu, L.G., Rangwalla, H., Cui, J. Y., and Elliott, J. R., Vapor-liquid equilibria of
vibrating square well chains. Jom'nal of Chemical Physics, 1999. 111(3): p. 1293-
1301.

Carnahan, NR and K.E. Starling, Equation of State for Nonactracting Rigid
Spheres. J. Chem. Phys., 1969. 51: p. 635-636.

Miller, D.J.A., Navinchandra S. ; Kolah, Aspi; Lira, Carl T. (Board of Trustees of
Michigan State University, USA), Process for Production of Organic Acid Esters
by Reactive Distillation, in US. Pat. Appl. Publ. 2006. 20 pp.

Holten, C.H., Lactic acid: Properties and Chemistry of Lactic Acid and
Derivatives. Verlag Chemie. 1971.

223

 

93.

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

106.

Vidler, W., The Kinetics of Polymer Formation and Hydrolysis in Lactic
Solutions. Report to Stichting Ilra., 1959. Personal Communication.

Van Ness, J .H., Hydroxy Carboxylic Acids. Encyclopedia of Chemical
Technology. Kirk Othmer. Vol. 13. 1981, New York: Willey.

Montgomery, R., Acidic Constituents of Lactic Acid- Water Systems. J. Am.
Chem. Soc., 1952. 74(6): p. 1466 -1468.

LockWood, L.B., D.E. Yoder, and M. Zienty, Lactic Acid Ann. NY. Acad. Sci.,
1965. 119: p. 854-867.

Eder, R. and F. Kutter, Gleichgewicht von Milchsa'ure und Milchsa'ure-anhydrid
in Handelsmilchsa'ure. Helv. chim. Acta, 1926. 9: p. 355-364.

Vu, D.T., et al., Vapor-liquid Equilibria in the Systems Ethyl Lactate + Ethanol

and Ethyl Lactate + Water. Journal of Chemical and Engineering Data, 2006.
51(4): p. 1220-1225.

Elliott, J .R. and CT. Lira, Introductory Chemical Engineering Thermodynamics.
1999: Prentice Hall PTR. 660.

Hefter, G.T., Solubility data. Diethyl Succinate - Water System. Solubility Data
Series, 1992. 49: p. 91-4.

Cottrell, F ., On the Determination of Boiling Points of Solutions. J. Am. Chem.
Soc., 1919. 41: p. 721-729.

Sanz, M.T., et al., Vapor Liquid Equilibria of the Mixtures Involved in the
Esterification of Lactic Acid with Methanol. Journal of Chemical and Engineering
Data, 2003. 48(6): p. 1446-1452.

Vu, D.T., et al., Oligomer Distribution in Concentrated Lactic Acid Solutions.
Fluid Phase Equilibria, 2005. 236: p. 125-135.

Ferreira, A.C.D., J .C. Barbe, and A. Bertrand, Heterocyclic Acetals from Glycerol
and Acetaldehyde in Port Wines: Evolution with Aging. Journal of Agricultural
and Food Chemistry, 2002. 50(9): p. 2560-2564.

Delfort, B., et al., Diesel Fuel Compounds Containing Glycerol Acetals, Pub. No:
US 2003/01 63 949 A1, in United States Patent Application Publication. Sep. 4,
2003: France.

Advanced Chemistry Development, ChemSketch 8.17 Freeware.

224

 

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

Gray, N.H., Layout of M3D File. Report to J. Richard Elliott, 2006. Personal
Communication.

Baskaya, F .S., et al., Transferable step potentials for amines, amides, acetates,
and ketones. Fluid Phase Equilibria, 2005. 236: p. 42-52.

Ye, T. and S. Kalyanaraman. A Recursive Random Search Algorithm for Large-
Scale Network Parameter Configuration. in ACM SIGMETRICS International
Conference on Measurement and Modeling of Computer Systems. 2003. San
Diego, CA, USA.

Broyden, G., the Convergence of a Class of Double-Rank Minimization
Algorithms, Parts I and II. J. Inst. Math. Appl., 1970. 6: p. 76-90 and 222-231.

Shanno, D.F., Conditioning of Quasi-Newton Methods for Function Minimization.
Mathematics of Computation, 1970. 24: p. 647-656.

Goldfrab, D., A Family of Variable Metric Methods Derived by Variational
Means. Mathematics of Computation, 1970. 24: p. 23-26.

Fletcher, R., A New Approach to Variable Metric Algorithms. Computer Journal,
1970. 13: p. 317-322.

Dennis, J .E. and RB. Schnabel, Numerical Methods for Unconstrained
Optimization and Nonlinear Equations. Computational Mathematics. 1983:
Prentice-Hall. 378.

Gill, P.E., W. Murray, and M.H. Wright, Practical Optimization. 1981: Academic
Press. 401.

Ambrose, D., Sprake, C.H.S., Vapor Pressure of Alcohols. J. Chem. Thermo.,
1970.2:p.63l.

Timmermans, J ., Physico-Chemical Constants of Pure Organic Substances. 2nd
ed. 1965: Elsevier, New York.

Ambrose, D., Counsell, J .F., Laurenson, M., Lewis, G.B.,, Thermodynamic
Properties of Organic Oxygen Compounds. XL VII. Pressure, Volume,
Temperature Relations and Thermoaynamic Properties of Propan-Z-ol. J. Chem.
Thermo., 1978. 10: p. 1033.

Lyons, R.L., The Determination of Critical Properties and Vapor Pressure of
Thermally Stable and Unstable Compounds. MS. Thesis, Dept. of Chem. Eng.
The Pennsylvania State University, May 1985.

Ambrose, D. and R. Townsend, Thermodynamic Properties of Organic Oxygen

225

 

 

121.

122.

123.

124.

125.

126.

127.

128.

129.

130.

131.

132.

Compounds. Part [X The Critical Properties and Vapour Pressures above Five
Atmospheres of Six Aliphatic Alcohols. J. Chem. Soc., 1963: p. 3614.

Boublik, T., Fried, V., Hala, E., The Vapour Pressures of Pure Substances. 1973:
Elsevier, New York.

Sachek, A.I., Markovnik, V.S., Peshchenko, A.D., Shvaro, O.V., Andreevskii,
D.N., Vapor Pressure of Secondary C5 - C8 Alcohols of Normal Structure. Sov.
Chem. Ind., 1984. 16: p. 705.

Hovorka, F., Lankelma, H.P., Stanford, S.C., Thermodynamic Properties of the
Hexyl Alcohols. II. Hexanols-I, -2, -3 and 2-Methylpentanol-I and -4. J. Amer.
Chem. Soc., 1938. 60: p. 820.

Ambrose, D., Ghiassee, N.B., Vapor Pressures, Critical Temperatures, and
Critical Pressures of Benzyl Alcohol, Octan-2-ol, and 2-Ethylhexan-I-ol. J.
Chem. Thermo., 1990. 22: p. 307.

McKillop, A., Ford, M.E., Mercury-Assisted Solvolyses of Alkyl Halides. Simple
Procedure for the Preparation of Nitrate Esters, Acetate Esters, Alcohols and
Ethers. Tetrahedron, 1974. 30: p. 2467-2475.

Polgar, N., Robinson, R., Synthetical Experiments Bearing on the Constitution of
Phthioic Acid. J. Chem. Soc., 1945: p. 393.

Thomas, L.H., Meatyard, R., Viscosity and Molecular Association. Part IV.
Association of Monohydric Alcohols and Some Hindered Phenols. J. Chem. Soc.,
1986:p.1963.

Thomas, L.H., Meatyard, R., Smith, H., Davies, G.H., Vapor Pressures and
Molar Entropies of Vaporization of Monohydric Alcohols. J. Chem. Eng. Data,
1979. 24(3): p. 159.

N'Guimbi, J ., Berro, C., Mokbel, I., Rauzy, E., Jose, J ., Experimental vapour
pressures of I 3 secondary and tertiary alcohols--correlation and prediction by a
group contribution method. Fluid Phase Equilibria, 1999. 162: p. 143-158.

Sherrill, M.L., An Investigation of a Series of Derivatives of Normal Heptane. 1.
Preparation, Identification and Physical Constants. J. Amer. Chem. Soc., 1982.
52:p.1930.

Ambrose, D., Ghiassee, N.B., Vapor Pressures and Critical Temperatures and
Critical Pressures of C5 and C6 Cyclic Alcohols and Ketones. J. Chem. Thermo.,
l987.19:p.903.

Riddick, J .A., Bunger, W.B., Sakano, T.K., Organic Solvents: Physical

226

 

I33.

134.

135.

136.

137.

138.

139.

140.

141.

142.

143.

144.

Properties and Methods of Purification. 4th ed. 1986: Wiley Interscience, New
York.

Huckel, W., Hubele, A., Cis- and trans-1-Methyl-2-Cyclohexanol. Liebigs Ann.
Chem, 1958. 27: p. 613.

Eliel, E.L., Haber, R.G., the Boiling Points of the Methylcyclohexanols - an
Exception to the Conformational Rule. J. Org. Chem, 1958. 23: p. 2041.

Matyschok, H., Pritzkow, W., Radeglia, R., Untersuchungen zur Sulfochlorierung
von Paraflinen. II. Untersuchungen uber die Reaktion von Aliphatischen
Sulfochloriden mit Nucleophilen. J. Prakt. Chem, 1977. 319: p. 46.

Selected Values of Properties of Chemical Compounds, Data Project.
Thermodynamics Research Center, Texas A&M University, College Station,
Texas, loose-leaf data sheets, l980-extant.

Young, S., the Vapour-Pressure, Specific Volumes, Heats of Vaporization, and
Critical Constants of Thirty Pure Substances. Royal Dublin Society, Scientific
Proceedings, Ser. A. Vol., 1910. 12: p. 374.

Gonzalez, E., Ortega, J., Densities and Isobaric Vapor-Liquid Equilibria of Butyl
Esters (Methanoate to Butanoate) with Ethanol at 101.32 KPa. J. Chem. Eng.
Data, 1995. 40: p. 1178.

Ortega, J., Galvan, S., Vapor-Liquid Equilibria and Densities for Propyl
Butanoate + Normal Alcohols at 101.32 kPa. J. Chem. Eng. Data, 1995. 40: p.
699.

F arkova, J ., Wichterlc, 1., Vapour Pressures of Some Ethyl and Propyl Esters of
Fatty Acids. Fluid Phase Equil., 1993. 90: p. 143.

Riddick, J .A., Bunger, W.B., Sakano, T.K., Organic Solvents: Physical
Properties and Methods of Purification. 3rd ed. ed. 1970: Wiley Interscience,
New York.

Bonhorst, C.W., Althouse, P.M., Triebold, H.O., Esters of Naturally Occurring
Fatty Acids. Physical Properties of Methyl, Propyl and Isopropyl Esters of C 6 to
C10 Saturated Fatty Acids. Ind. Eng. Chem, 1948. 40: p. 2379.

Gros, A.T., Feuge, R. 0., Surface and Interfacial Tensions, Viscosities, and other
Physical Properties of Some n-A liphatic Acids and their Methyl and Ethyl Esters.
J. Amer. Oil Chem. Soc, 1952. 29: p. 313-17.

Vogel, A.I., Physical Properties and Chemical Constitution. Part XIII. Aliphatic
Carboxylic Esters. J. Chem. Soc., 1948: p. 624.

227

145.

146.

147.

148.

149.

150.

151.

152.

153.

154.

155.

156.

157.

Bronsted, J .N., Colmant, P., Molekulgrosse und Phasenverteilung. 111. Z. Phys.
Chem. (A), 1934. 168: p. 381.

Walker, R., Davidson, D.W., The Dielectric Constant and Dipole Moment of 1,3-
Dioxane. Can. J. Chem, 1959. 37: p. 492.

Fletcher, S.E., Mortimer, C.T., Springall, H.D., Heats of Combustion and
Molecular Structure. Part VII. 1:3 Dioxa- and 1 :3 :5-Trioxa-cycloalkanes. J.
Chem. Soc. London, 1959: p. 580.

Arbusov, B.A., Vinogradova, V.S., Parachors of the Cyclic Esters of Carbonic
Sulfurous and Phosphorous Acids. Izv. Akad. SSSR, 1950. 297.

Kobe, A.K., Ravicz, A.E., Vohra, S.P., Critical Properties and Vapor Pressures
of Some Ethers and Heterocyclic Compounds. J. Chem. Eng. Data, 1956. 1(1): p.
50.

Crenshaw, J .L., Cope, A.C., Finklestein, Rogan, R., The Dioxanates of the
Mercuric Halides. J. Amer. Chem. Soc., 1938. 60: p. 2310.

Melia, T.P., Bailey, D., Tyson, A., Thermodynamics of Addition Polymerisation.
I. The System Trioxane-Polyoxymethylene. J. Appl. Chem, 1967. 17: p. 15.

Komatsu, S., Masumoto, M., Studies on Catalytic Action at High Pressure and
Temperature. Catalytic Hydrogenation of Ethers. Bull. Chem. Soc. Jap., 1930. 5:
p. 241.

Quaker Oats Chemicals Inc., Tetrahydrofurfuryl Alcohol. Bulletin 206-A,
Chicago, IL, 1960.

Kaufmann, W.B., Adams, R., the Use of Platinum Oxide as a Catalyst in the
Reduction of Organic Compounds. IV. Reduction of F urfural and its Derivatives.
J. Amer. Chem. Soc., 1923. 45: p. 3029.

CRC Handbook of Chemistry and Physics. 63 ed, ed. M.J.A. Robert C. Weast.
1982-1983: CRC Press.

Wakelyn, P.J., Wan, Peter J ., National Cotton Council, Washington, DC, USA.,
Edible Oil extraction Solvents: FDA Regulatory Considerations. Inform, 2004.
15(1): p. 22-23.

Cavalcante, C.L., Jr., Industrial Adsorption Separation Processes: Fundamentals,

Modeling and Applications. Latin American Applied Research, 2000. 30(4): p.
357-364.

228

I1111111111111"11111