V . 8h
‘ . . rang“...
. «W . . . . . ‘ .

. . : .

“1.5 “my 5.. ‘ V ,

9.26.
.
:L

.1"

‘ T.
.mNm?
, é?¥.-.raxr._.hum
’9 V IL‘ .. “an“.
.1? .3 ‘
r“
V» i)

‘ n3 1
‘ Iva:
. f V V ‘ .
. Air... ..

. 33%.:
:-
3 .- . .

7t!
. .v
, , ‘ 52?. t
.. 4‘ V ‘ , ‘ .
.‘i... .51“? . .

1::

:r; a ‘1.
‘54

 

1.
1"
1.
54;:
1..
i

l. t
2:14;).

 

V... Di. . ¥
«3%,

i a "a bun
. LIVE: 45:.
1.35.} I51
.9.

1.! .
31:35

it
.11.. ér:
. Pffiiufimv
if k 71!.» .
.: Tina.“ ‘2.» a
trutvi..l.!
r it 3.55. :2
3.! P. \\mLi..,.tu.I.1~
.5}. 1.3.5 93.. .
I 1431.141».

 

,, , ‘ ; . L . . rmqfi... F... ,1.
xvuué , r ,, . n.931nuuhfiv

{1: .n... s
L

t~

;
2W8.
.__. 0

This is to certify that the

dissertation entitled

CHARACTERIZATION OF INTERFACIAL STRUCTURE AND
PROPERTIES OF COLLOIDAL LIQUID APHRONS
FOR BIOCATALYSIS

presented by

Prashant Srivastava

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Chemical Engineering

K. WWW

Major professor

Date /&/25/00

MS U is an Affirmative Action/Equal Opportunity Institution 0.12771

 

 

LIBRARY
Michigan State
University

 

 

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6’01 c:/ClRC/DatoDuo.p65—p.15

Characterization of interfacial structure and properties of Colloidal Liquid
Aphrons for biocatalysis

By

Prashant Srivastava

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemical Engineering
2000

Dr. R. Mark Worden

ABSTRACT
Characterization of interfacial structure and properties of Colloidal Liquid Aphrons
for biocatalysis
BY
Prashant Srivastava

Colloidal liquid aphrons (CLA) are surfactant-stabilized droplets (1 -50
microns in diameter) of a nonpolar liquid dispersed in a continuous aqueous
phase. The small dimensions of CLA provide very large interfacial areas and
short diffusion path lengths; consequently interphase solute transfer can often be
completed within seconds. The interfacial structure and mass transfer properties
of CLA emulsions were investigated in this research. Applications of CLA
emulsions in multiphase enzymatic and whole-cell biocatalysis were also
developed.

A variety of experimental tools were used to characterize the interfacial
structure of CLA emulsions, including freeze fracture transmission electron
microscopy, differential scanning calorimetry, small angle x-ray scattering, and
lsothennal titration calorimetry. Results indicate the presence of liquid crystalline
bilayers surrounding the organic phase in CLA emulsions. These layers are
formed by the interaction between the two surfactants used in CLA formulation.
The energetics and stoichiometry of the surfactant interaction were measured. A
conceptual model of the mechanism stabilizing CLA emulsions was developed.

The mass-transfer properties of CLA emulsions were measured

experimentally using two approaches. The first approach involved the direct

measurement of heptanoic acid diffusion from the CLA emulsions into the water
phase. The second approach involved indirect measurement via coupling the
mass-transfer of the solute (p-tolyl acetate) to a chemical reaction (alkaline
hydrolysis). The local mass transfer coefficient of the surfactant shell (KLshen )
was estimated to be about ~30 x 10'3 cm/s, which is comparable to KL values for
surfactant-free, liquid-liquid systems at an equivalent void fraction (¢)- The
volumetric mass transfer coefficient (KLa ) for CLA were 5 to 20 times higher than
those for conventional emulsions.

Application of CLA emulsions to enzymatic and whole cell biocatalysis
was demonstrated. The dynamic resolution of L,D phenylalanine was carried out
in a multiphase system using porcine pancreatic lipase. The CLA emulsions were
found to reduce the time required to attain equilibrium by about four-fold over
conventional contacting methods. A statistical design of experiment approach
was used to identify the major variables affecting equilibrium conversion and
enantiomeric excess. The use of CLA emulsions in the biotransformations of
cinnemaldehyde by Baker's yeast at high cell densities (2509/L) using a non-toxic
solvent (2,2,4 trimethyl pentane) was also studied. The selectivity of this
transformation was influenced by the presence of CLA emulsions.

Reversible emulsions formed using novel block-graft copolymeric
emulsifiers (Polymethacrylic acid (PMAA)-g-Ethylene Glycol (EG)) were
investigated. Such emulsions show considerable promise in multiphase
contacting, since the emulsions produced can be maintained stable as long as

desired and then coalesced on demand by a small pH change.

Dedicated to my granddad, Mr. B.N.Srivastava, and my parents who have been a

constant source of motivation and strength

ACKNOWLEDGMENTS

I would like to thank Dr. R. M Worden for his invaluable guidance and support
during this research project. I would also like to thank Dr. Alec Scranton, Dr
Charles Petty, Dr. Bruce Dale and Dr. Pat Oriel, for their service on my guidance
committee. This work would not have been possible without my principal
collaborators, John W. Heckman, at the Center for Electron Optics, and Olaf
Hahr, at the Technical University of Berlin. Invaluable intellectual discussions
with Robert C Saunders, a fellow student at MSU, have contributed immensely to
my insight into the interfacial structure. His knowledge of interfacial science and
measurement has been a constant inspiration for this work. I would like to thank
Dr. Charles Petty, for access to the particle characterization facility, and inspiring
discussions over coffee. Special thanks goes to Dr. Wilson, Professor in the
Department of Biochemistry, for letting us use the Omega titration
microcalorimeter. I would also like to thank Dr. M.M. Sharma, Hon.Professor,
Jawaharlal Nehru Center for Advanced Scientific Research (Bangalore) and Dr.
G.J. Karabatsos, Professor Emeritus, Department of Chemistry, Michigan State
University for their guidance for the p-tolyl acetate studies. Dr. Dennis Gilliland, at
the Department of Statistics, MSU, contributed immensely to the statistical
design and analysis presented in Chapter 4. I would also like to thank Frank Jere
and Mike Shaffer, fellow students in Dr Miller's group, for being such a great
listener for all my crazy ideas. It was a pleasure working with some tremendous
undergraduates; Jason Shellnut, Derek Hodges, Andrew Brinks, Anna Johnson,

and Christopher Loose, in the course of this work. Special thanks to the

Chemical Engineering office staff : Faith Peterson, Julie Caywood, Candace
Mcmaster, Nancy Albright, Mary Lehnert and Paula Holzheur for all their support

during my stint in graduate school.

vi

TABLE OF CONTENTS

 

List of Figures ................................................................................. xiv

List of Tables ................................................................................... xix
Introduction ......................................................................................... 1
1. Interracial Structure of Colloidal Liquid Aphrons ...................................... 15
1.1 Introduction -- ....................................................................... 15
1.2 Preparation of CLA emulsions ............................................................... 20
1.3 Differential Scanning Calorimetry .......................................................... 20
1.4.1 Description ........................................................................................ 20
1.4.2 Materials and Methods ..................................................................... 22
1.4.3 Results ............................................................................................... 22
1.4 Freeze-FracturelFreeze Etch Transmission Electron Microscopy ..... 24
1.4.1 Description ........................................................................................ 24
1.4.2 Materials and methods ..................................................................... 27
1.4.3 Results ............................................................................................... 28
1.5 Small Angle X ray Scattering .................................................................. 38
1.5.1 Description ........................................................................................ 38
1.5.2 Materials and Methods ..................................................................... 42
1.5.3 Results ............................................................................................... 42
1.6 Isothermal Titration Calorimetry ............................................................ 45
1.6.1 Description ........................................................................................ 45

vii

1.6.2 Materials and Methods ..................................................................... 47

1 .6.3 Results - ............ - - ............ 48

 

1.7 Development of a conceptual model for lamellar micelles stabilizing

 

CLA interfaces _ ............................................................. 51
1.8 Interpretation of CLA properties ............................................................ 54
1 .8.1 Ease of formation and small droplet sizes ..................................... 54
1.8.2 High dispersed phase volume ratios .............................................. 55

1.8.3 Influence of nature of the dispersed phase on maximum phase

 

 

 

volume ratios ........................................................... 57
1.8.4 High stability .......... -- .......................................................... 58
1.9 Conclusions ..... ................................................. 59
1 .10 References - .............................................................. 59

 

2. Enhancement of Mass Transfer Using Colloidal Liquid Aphrons :

Measurement of Mass Transfer Coefficients in Liquid-Liquid Extraction... 64

 

 

 

 

2. 1 Introduction _ ......... ........................................ 64
2.2 Materials and Methods ............................................................................... 69
2.2.1 Mass-Transfer System ......................................................................... 69
2.2.1.1 Surfactants. ..................................................................................... 69
2.2.1.2 Continuous and Dispersed Phases ........................................................................................... 69
2.2.1.3 Transferred Solute . ......................................................................................... 70
2.2.2 Measurement of Heptanoic Acid Concentration ................................ 71
2.2.3 CLA Preparation - -- ........... . .................................................. 73

viii

2.2.4 CLA Size Measurement ........................................................................ 74

2.2.5 Mass Transfer Measurement ............................................................... 74
2.2.5.1 Stirred-Cell Experiments. ........................................................................................................ 74
2.2.5.2 Stirred-tank experiments .......................................................................................................... 74

2.3 Data Analysis .............................................................................................. 77
2.4 Results and Discussion .............................................................................. 81

2.4.1 Measurements in Stirred Cell .............................................................. 81

2.4.2 .Measurements in Baffled Tank ........................................................... 81

2.4.3 Influence of Surfactant Concentration. .............................................. 88

2.5 Conclusions ................................................................................................ 92
2.6 Nomenclature .............................................................................................. 92
2. 7 References ................................................................................................... 94

3. Enhancement of Mass Transfer using Colloidal Liquid Aphrons:

Measurement of Mass-Transfer Coefficients Using Chemical Reaction ...... 99
3. 1 Introduction ................................................................................................ 99
3.2 Theory ....................................................................................................... 100
3.2.1 Mass-transfer resistances in series .................................................. 100
3.2.2 Coupling of mass transfer and reaction ........................................... 101
3.2.3 Relative rates of mass transfer and reaction ................................... 101
3.3 Materials ................................................................................................... 105

3.4 Methods .................................................................................................... 106

3.4.1 CLA preparation ................................................................................. 106
3.4.2 CLA size measurement ...................................................................... 107
3.4.3 Mass-transfer rate measurement ...................................................... 107
3.4.4 Physical-property determination ....................................................... 108
3.5 Results and Discussion .......................................................................... 109
3.5.1 Physicochemical properties .............................................................. 109
3.5.2 Effect of ionic Strength ...................................................................... 111
3.5.3 Mass-transfer coefficients ................................................................. 111
3.5.4 Effect of surfactant concentration on I and KL ............................ 114
3.6 Conclusions ............................................................................................. 1 17
3. 7 Nomenclature ........................................................................................... 118
3.8 References ............................................................................................... 119
4. LIPASE CA TAL YZED DYNAMIC RESOLUTION OF AMINO ACIDS ......... 121
4. 1 Introduction ............................................................................................... 121
4.2 Model System ............................................................................................ 124
4.3 Materials and methods ............................................................................. 126
4.3.1 Experimental Methods ....................................................................... 127
4.3.1.1. Immobilization efficiency ...................................................................................................... [27

4.3.1.2. Chiral resolution experiments ............................................................................................... [28

 

 

 

 

 

4.3.1.3. HPLC assay .......................................................................................................................... I 29
4.3.2 Conversion and enantiomeric excess measurements .................... 133
4.3.2.1. Central Composite Design .................................................................................................... I 33

4.4 Results and discussion ............................................................................ 136
4.4.1 Enzyme interactions with CLA .......................................................... 136
4.4.2 Reaction rate measurements ............................................................ 137
4.4.3 Chiral Resolution ................................................................................ 137
4.4.3.1. Conversion (Yr) ..................................................................................................................... 140
4.4.3.2. Enantiomeric excess ( Y") ..................................................................................................... I 52

4.5 Conclusion ................................................................................................ 156
4.6 References ................................................................................................. 157
5. Whole Cell Biotransformations using Colloidal Liquid Aphrons ............ 160
5.1 Introduction--- - - . __ - ........................................... 160
5.1.1 Whole Cell Biocatalysis ........................................ 161
5.1.2 Organic solvents in whole cell biocatalysis ..................................... 163
5.1.3 Colloidal Liquid Aphrons ............................................... 164
5.1.4 Model system . -- ......................................................... 164
5.2 Experimental Methods - -- .................................... 169
5.2.1 Determination of solvent blocompatibility ....................................... 169
5.2.2 Preparation of Colloidal Liquid Aphrons .......................................... 169

xi

5.2.3 Biotransformation Methods ............................................................... 170

 

5.2.4 Analytical procedures ........................................................................ 171
5.3 Results and discussion ............................................................................ 1 74
5.3.1 Biocompatibility . ............................................. 174
5.3.2 Biotransformations in presence of CLA ........................................... 176

5.3.3 Mechanisms for changing selectivity of whole cell

 

 

 

 

 

 

 

biotransformations --------- - .................................. 1 78
5.3.3.1. Eflect of concentration on product composition ................................................................... I78
5.3.3.2. Effect of time on product composition .................................................................................. I78

5.3.4 Selectivity control using hydrophobic resins .................................. 181

5.4 Conclusion -_ -- ..................................... 181
5.5 References ................................................................................................. 184
6. Reversible emulsions ................................................................................. 187
6.1 Introduction - _________ ................. 187
6.2 Materials and Methods ............................................................................. 189
6.3 Results -- - ......................................................... 191
6.4 Conclusions ------- . ....................... 193
6.5 References - - -- ...................................................... 196
7. Summary .................................................. 197
xii

List of Figures
Figure 1 Ideal substrate (Modified from Csuk et al., 1991) ............................. 2

Figure 2. Limitations of common biocatalytic substrates and strategies to

overcome those limitations (redrawn from Van Sonsbeek et al. 1993) ............. 4
Figure 3. Possible biocatalytic systems involving an organic liquid .................. 5
Figure 1.1. Model for CLA droplets (Redrawn from Sebba, 1987) ...................... 18

Figure 1.2. Schematic diagram of Differential Scanning Calorimeter (DSC) ....... 21
Figure 1.3. DSC endotherm for CLA dispersion containing 0.1% Tergitol 15-8-3
in Limonene and 0.4%Tween 80 in Water (PVR 0.25) ........................................ 23
Figure 1.4 Freeze Fracture and Freeze etching of "membraned" cells or
emulsions. ........................................................................................................... 25
Figure 1.5 Schematic diagram of sample preparation in Freeze Fracture TEM. .26
Figure 1.6. Phase contrast micrographs of A. CLA emulsion and B.
Conventional emulsions. ..................................................................................... 32

Figure 1.7 Fracture surfaces of separate phases. A. Water and Tween 80 frozen
in liquid propane at -190°C. B. Limonene and Tergitol frozen in a nitrogen

slurry at =-21o°C. ............................................................................................... 33
Figure 1.8 Replicas of conventional emulsions. A. Freeze fracture and B.
Freeze etch. Inset shows the detail of the droplet matrix interface for each at 5X

increased magnification. ..................................................................................... 34

xiii

 

Figure 1.9 A. Freeze fracture of CLA suspension showing the multilamellar
phase boundary. B. Enlargement of the boundary showing varying surface
steps approximately 20 to 40 nm high. ............................................................... 35
Figure 1.10 A. Freeze etch of a CLA emulsion showing enhanced detail of the
phase boundary. B. Enlargement showing the 20 to 40 nm steps and additional
ridges running parallel to the step faces. ............................................................ 36
Figure 1.11 A. Freeze fracture surface of a mixture of Tergitol and Tween at the
concentrations used to form the CLA. B. The same solution, etched for 8 min
prior to shadowmg ............... 37
Figure 1.12 X-ray diffraction from lattice planes (Braggs law) ............................. 41
Figure 1.13. SAXS data for 0.1% Tergitol, 0.4% Tween 80 solution in water ..... 44
Figure 1.14. Typical ITC experimental data and analysis (MicroCal Inc, MA) ..... 46
Figure 1.15 Heat flow/sec vs. time data for ITC Runs 1. Deionized water titrated
deionized water, 2. Deionized water titrated 0.1% Tergitol, 3. 2% Tween 80
titrated deionized water, 4. 2% Tween 80 titrated 0.1% Tergitol ..................... 49
Figure 1.16 Analysis of ITC data based on two site model. ................................ 50
Figure 1.17. Conceptual model for mixed lamellar micelles formed by Tergitol 15-
3-3 and Tween 80 in water. ................................................................................ 53
Figure 1.19. CLA emulsion comprising Limonene (0.1% Tergitol 15-S-3) and

water (0.4% Tween 80) at a PVR of 14 (magnification 400x) .............................. 56

Figure 2.1. Proposed structure for a single colloidal liquid aphron (CLA) when
dispersed in a continuous aqueous phase (Redrawn from Sebba (1984)). ........ 68

Figure 2.2 Calibration curve for heptanoic acid ................................................... 72

xiv

 

Figure 2.3a: Agreement of experimental data with the linear trend predicted by
Eqn. 3. (stirred cell at 60 rpm) ............................................................................. 78
Figure 2.3b: Agreement of experimental data with the linear trend predicted by
Eqn. 3. Run 10 in baffled tank at 427 rpm ........................................................... 79
Figure 2.3c. Agreement of experimental data with the linear trend predicted by
Eqn. 3. Run 8 in baffled tank at 550 rpm, 0.4% (v/v) Tween 80 in aqueous phase
and 0.1% (v/v) Tergitol 15-S-3 in CLA phase. ..................................................... 80
Figure 2.4. Variation in experimental volumetric mass transfer coefficient (KLa)
with Nag measured in a baffled tank. ................................................................... 83
Figure 2.5. Effect of Nae on experimental KL values (0) measured in a baffled,
stirred tank and predicted ammo... values (El) calculated using Eqn. 4
(Skelland et al., 1981). ........................................................................................ 86
Figure 2.6 Variation of CLA droplet diameter with Tween 80 concentration in the
aqueous phase at Tergitol 15-S-3 concentrations of 0.1% v/v (El) and 0.01% v/v
000. ..................................................................................................................... 89
Figure 3.1: Semi log plot of specific reaction rate (R) vs. ionic strength (I) at 550
(El) and 300 (0) rpm. ....................................................................................... 112
Figure 3.2. Effect of Nae on KL values measured in baffled stirred tank at initial
alkali concentrations of 0.75M (El) and 0.5M (O). The solid line represents the
prediction of Equation 9 (Skelland et al., 1981) ................................................. 113
Figure 3.3. Variation of CLA Sauter mean diameter (0) and volumetric mass
transfer coefficient (KLa, III) with Tween 80 concentration (% v/v) in the aqueous

phase. ............................................................................................................... 1 15

XV

Figure 4.1 Lipase catalysed resolution of D,L-Phenylalanine using hydrolysis of
D,L Phenylalanine methyl ester (PAM). ............................................................ 126
Figure 4.2. Reaction products of L,D PA with TAGIT ........................................ 131
Figure 4.3. Sample HPLC Chromatogram of the enantiomeric separation of L
and D Phenylalanine ......................................................................................... 132
Figure 4.4 Conversion vs. time for chiral resolution experiments with (I) and
without ( ) CLA dispersions. Reaction conditions: pH 5, Void fraction 30%, 80
mg crude enzyme, 0.2M Phe-me in decanol. .................................................... 138
Figure 4.5 Residuals vs. fitted values for full set of experimental data. (Note: The
fitted value for one of the data points (experiment 25) is negative, and the
residual is extreme) ........................................................................................... 142
Figure 4.6a. Plot of residuals vs. order of data for regression model represented
by equation 1. ................................................................................................... 144
Figure 4.6b. Normal Probability plot for residuals based on Equation 1 ............ 145
Figure 4.60. Plot of residuals vs. Fitted values for regression model given by
Equation 1. ........................................................................................................ 146
Figure 4.7. Contour plot for Yc for P and V ranging from -1 to 1, using Equation 1.
.......................................................................................................................... 150
Figure 4.8. Plot of conversion vs. void fraction at varying initial concentrations of
Phe-Me in the decanol phase ........................................................................... 151
Figure 4.9 Plot of residuals vs. order of data for Yee. ........................................ 154
Figure 4.10 Contour plots for Yee, for P and V ranging from -1 to 1, based on

equation 2. ........................................................................................................ 155

xvi

Figure 5.1. Biotransformation of Cinnemaldehyde by Baker's yeast. ................ 165

Figure 5.2. Structure of L- amiacetose .............................................................. 167
Figure 5.3. Compounds synthesised from (4) ................................................... 168
Figure 5.4. Typical NMR spectrum of biotransformation products .................... 173
Figure 5.6. Effect of CLA volume on % dial in product ...................................... 177

Figure 5.7. Change in product composition with initial concentration of
cimmemaldehyde .............................................................................................. 1 79
Figure 5.8. Effect of time on product composition. Plotted are data at different
initial substrate (1) concentrations .................................................................... 180

Figure 5.9. Effect of hydrophobic resins on product composition ...................... 183

Figure 6.1 Complex formation in MAA-MPEGMA copolymers (redrawn from

Mathur (1998)). ................................................................................................. 188
Figure 6.2. Titration curves for block copolymeric emulsifiers ........................... 194
Figure 6.3 Percentage recovery of organic phase vs. pH for reversible emulsions,

formed using different polymeric emulsifiers, and CLA emulsions. ................... 195

xvii

List of Tables

Table 1. Literature review of CLA in chronological order, upto 2000 ............... 1

Table 2.1. Summary of conditions for mass-transfer experiments. ..................... 76
Table 2.2. Effect of Nag on K, Kmmunuous, and KLsheu. ........................................ 87
Table 2.3. Effect of surfactant concentrations on KL and KLa .............................. 90
Table 3.1. Physicochemical data for p-tolyl acetate-water system. ................... 1 10

Table 3.2. Variation of interfacial area (a), overall mass-transfer coefficient (KL),
and volumetric mass-transfer coefficient (KLa) with Tween 80 concentration in
aqueous phase. ................................................................................................ 116
Table 4.1. Buffer recipes for chiral resolution experiments ............................... 126

Table 4.2. Central Composite Design (coded variables are defined as shown in

Table 4.3) .......................................................................................................... 135
Table 4.3 Coding of variables for Central Composite Design ............................ 136
Table 4.4 % Enzyme immobilized using different nonionic and ........................ 136
emulsifiers to form CLA emulsions at pH 5. ...................................................... 136

Table 4.5 Effect of pH on % enzyme immobilized on CLA emulsions formed

using Tween 80 ................................................................................................. 137
Table 4.6. Experimental data for chiral resolution experiments ......................... 139
Table 4.7 Regression analysis for data presented in Table 4.6 ........................ 141

Table 4.8. Regression analysis results for model represented by Equation 1...143

Table 4.9 Regression analysis for the model represented by Equation 2 ......... 153

xviii

Table 6.1. Size distribution and mass transfer coefficients obtained for reversible

emulsions and CLA. .......................................................................................... 192

xix

INTRODUCTION

Challenges in Biocatalysis

Biocatalysis has enormous commercial potential. Enzymes or whole cells
can convert chemical precursors into a wide variety of product classes, including
pharmaceuticals (steroids, isoflavonoids, taxol), fragrance and flavor compounds
(lactones, pyrazines, essential oils), fuels (ethanol, butanol, biodiesel), as well as
a broad range of miscellaneous oils, fatty acids, and terpenoids. Most
biocatalysts are amazingly specific, and it is this ability to carry out a reaction on
a single substrate or, conversely, to utilize multiple substrates to produce a single
product or product family, that gives them wide applicability. An important
limitation to the use of conventional biological processes arises from the
observation that many of the feedstocks and/or resulting products have only
limited or negligible water solubility. Biocatalysts, such as microorganisms (which
may be considered as complex, multifunctional, enzyme systems) or isolated
enzymes, are designed by nature to be effective in an aqueous medium, which
would suggest that they are poorly suited for use in an application that requires
them to function in a non aqueous environment. The ability to use biocatalysts in
non-natural environments, such as non-aqueous media, would facilitate the
application of bioprocesses in a wide range of operations. Development of
efficient biocatalytic processes involving nonpolar reactants or products having
low aqueous solubility is challenging (Mattiasson, et al. 1991 ). In such cases, a
multiphase approach can be effective, in which the continuous aqueous phase

contains the biocatalyst, and the dispersed phase serves as a reservoir for the

1

reactant or product. The dispersed phase may transfer a reactant to the
aqueous phase to replenish what is consumed by reaction. Alternatively, the
dispersed phase may extract an inhibitory or labile product from the aqueous
phase as it is formed (Larsson, et al. 1989). Most multiphase contacting
processes are mass transfer limited, and high surface area approaches
(emulsions) are most effective in overcoming mass transport resistances (Scott
et al., 1995).
Mass transfer limitations in biocatalytic processes

An ideal whole-cell biocatalytic process would involve a substrate that

would conform to all the conditions listed in Figure 1.

 

 

 

 

 

 

 

 

 

 

 

 

The substrate '5 The substrate is able to
soluble In
is rm entatio n pass the cell membrane
medium \
IDEAL
SUBSTRATE
The catalytic activity of the The substrate gives
microorganism is not inactivated by high regio- and
the substrate or by the product stereospecificity

 

 

 

 

 

 

Figure 1. Ideal substrate (Modified from Csuk, R et al., 1991).

However, ideal interactions between the substrate and the microorganism are
scarcely found in practice. Some of the techniques used to overcome the
limitations of real substrates are shown in Figure 2. The techniques highlighted
involve mass transfer from one phase to another. In most aqueous organic two-
phase systems, mass transfer limitations and rapid inactivation of cells are often

perceived to be the main drawbacks (Van Sonsbeek et al., 1993).

Scott et al. (1995) have identified atleast four types of reaction systems
containing a water immiscible organic liquid (Figure 3). Three of these utilize a
dispersion of either the organic or aqueous phase in an aqueous or organic
phase, respectively, with the biocatalyst being in the aqueous phase, but with
interfacial contact with the organic phase. However, there has been only limited
work on the investigation of the potential multiphase bioreactor systems and
operating parameters. A fundamental understanding of interfacial contact and
transport mechanisms, and the effects of system parameters on intrinsic
biocatalytic kinetics, is needed so that an optimum configuration for such reactor
systems can be developed (Lily et. al., 1985). This research is aimed at
understanding the fundamental aspects of interfacial stabilization to minimize
transport limitations that occur in multiphase catalytic systems and enhance the

productivity and selectivity of biocatalytic processes.

Crystalline Chemical or physical

' ' * l ' ' treatment of the cells in
Emuls’ in ’SUbSI'aie '3 ~ '
a ent lifsyagded gyground to micron order to make substrate
9 gal-sir‘z'e particles _ accessrble e.g.

/ lyophilization, grinding
ls insoluble in the I

reaction medium Cannot pass through
‘ the cell membrane

Reaction is

 

 

 

 

 

 

I carried out
._ . g - .~ _ p by a purified
:23?2:§r‘:edium ’ The substrates dealt enzyme
, - with in biocatal ic
3‘91"“ 0* SOIIFI, 5 processes yt Genetic strain T
absorbents ~ - _- “ improvement
\ / Chemical \ I I
Modification
Affects the
viability and / of substrate Shows a slow turnover

reproduction of rate and/or low regio and

cells \ / stereospecificity

Change the

- strain of
Grow the cells In Add an Change of

presence Of IOW organism inducer physiological state
substrate . _ , . . of the organism by
concentratlon. Substrate IS dispersed after change of
the completion of growth in' aeration, pH,
small portions (fed batch) or addition of salts,
continuously (at low rates) change in medium

composition

Figure 2. Limitations of common biocatalytic substrates and strategies to
overcome those limitations (Redrawn from Van Sonsbeek et al. 1993)

4

 

 

 

   

i

it
ill

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

/
Organic dispersed in aqueous Aqueous with soluble biocatalyst
with aqueous soluble Biocatalyst dispersed in organic

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

*iisizzsééééésséséii

 

 

Aqueous with insoluble biocatalyst Organic with solubilized or particulate
dispersed in organic biocatalyst

Figure 3. Possible biocatalytic systems involving an organic liquid (Scott et al.
1995)

Colloidal Liquid Aphrons

Colloidal Liquid Aphrons (CLA) are micron sized spherical droplets of
organic solvent (1-50 microns in diameter), each encapsulated in a surfactant
shell. Sebba (1972) proposed that the CLA are stabilized by a mixture of
surfactants: one in the aqueous phase and another in the dispersed organic
phase. Since Sebba’s original reports on biliquid foams (Sebba, 1972) and
subsequent description of “minute oil droplets encapsulated in a water film
(Sebba, 1979), CLA have been investigated for use in pro-dispersed extraction
processes. The small dimensions of CLA provide a very large mass transfer area
(typically 10,000-20,000 m2/m3 CLA), and allowing extraction equilibrium to be
attained within seconds. CLA have been applied successfully to the separation of
antibiotics, epoxides and ethanol (Lye and Stuckey, 1996; Rosjidi et al, 1994,
Wallis et. al 1985). Furthermore, the addition of various ion-exchange reagents to
the organic phase of the CLA has enabled polar solutes like phenylalanine
(Scarpello and Stuckey, 1996), and several metal ions to be extracted (Save et
al. 1994). Separation of CLA from dispersions and their reconcentration can be
effected by micro-filtration or centrifugation (Lye and Stuckey, 1996; Rosjidi et. al,
1994)

A brief literature review of studies previously undertaken on CLA systems
is given in Table 1. Most studies have emphasized the applications of the CLA
dispersions, and limited fundamental aspects have been studied. However,
understanding the relationship between the structure and the macroscopic

properties of CLA is essential to their optimal use.

6

 

muozooa

 

 

 

 

 

 

coEEmELE m<_2._.o
92% Co on Bear—om
85ch E33 :3 I .2: owe 82 ._c
nowaoamfiéd cozgcoocoo <40 cam i In .8860 wow Bear—om 5 >385
822on
._ocman-F
69¢ EEomtzm AEE N-m.ov ._o-—-cmxoc .968...
we: 9E3 5255: So .898; 820
E9 ooméovv cocoa mExE Beacon. we:
com. .323 2m... 2036?... western. oEoLooim
__o E? o\om-8 cozmzcoocoo cacao... _ocm=ow
cozombxm Eo>_om .cfiomtzm omega 9.300 mme m<_2._b mm? ._m as
8335.5 .__0 a Te .352 m.__._ 5.88 mom 3.25 3535.2
mtm
8398. mmoomz -9 6:93 $2 .238
is 8-8 me: 8980
23.85 .8303 35 ct: we: Esau. Wm m8.
.6 beacon. Es 2 .c .9 cs: 8:92“. .888 mmoonz -9 .293 ._n so o.___n>>
em
2 oowoaxo moi .Emuomtzm .o
cab .cozmzcoocoo Emuomtaw
.cozmbcoocoo Cosmo 9w
. A.“ as. {we 22.8 .0 8.: ocooeox mmocz -2 .255 as. .838
Emacs—Sm
ounce =0 own—E «cogent—.5
ocozoo=no< ocoE—Eoo new 2225.5“. oooLoooE .533 omega. :0 3:203:

 

 

 

 

 

 

 

Doom 2 a:

$20 _oo_mo_ocoEo 5 «1.0.0 2633... 929.23 Foam...

 

 

 

on? Em>_om
22E o. 7N9 :ozmzcoocoo
228:3 Ami 82m

 

 

 

 

 

5.2286 :2832 .32 on 6:230
:o:m~___noEE_ In 828:: waozczcoo SEBE ocmooo: omw ommw
oE>~:m m7: :ozmzcoocoo oE>~:w 6-75880 mom Econom ._m .o 23
2
82850
:8 «mm x: 886
.2: 5:26 <._ .22 26: <.. 2. own @3530 m:o_>£m
:ozogxm 29.8 :28; 8:25 .E_E E2m>m o:_Em_< E2535 _o:oo_< vamp
82635-85 5 :0 89¢ .22 26: <00 8:88me .3250 353 ._m 8 o>wm
can mcocmooo
5:856 .528 mod cc: and .59 23¢on -N :3 3m,
oomcoamfitoi 2°88: 8:2 oE:_o> oo2<t_0 6:880; mom _o:mtom ._m .0 _o__mom_
890
88; =25 88: c .888 8.2
2mm; _oooE 2:2 .338 E::n___:oo 2288 omp
E9: moEmmB 882.2 .88” 7mm” 5 .23-: 8:230 33
928 88.8: 28: cocunfio .673 I: .8900 wow 8 888m .8 8 2..
son Tom” c 89
page, 88: .22 o. 78.3
50:28 o_:o_ ommca 2526p
8:888 .828 .38 88-3 888888 our 82
82820-2: :_o>EoE:co .3 73 I: _o.. 7:860 mom _o:2_om ._m 8 23

 

 

 

 

 

 

 

 

35.2... 2.3.
8.2.588 8% 662m

 

 

 

 

 

 

 

8.695 82.8 cow-8: 8.2.588 .85.? 08 63.2.... om... cm. 82 ..m
82820.2: :2 v6-8 5926 0.5. 6:200; o.:22.:>m _o:2._om .m 0:2...QO
A:.o>E2::cm
ncm 023.: $22.35
.0 8-.. c 939882 : 79 on .888
So .0 5.85 I: .8 .22 .v. dz .3. 8... 88m: :2 82
ncm 222:5 :32. 22 To. 592% 0.5. _o- 7580 wow _o:m:om ._m .m .23
5.: o. 5.8 8.28m... <00
.0 £80 3.... o. .m .: 8...
5.62:8 E028 5:20.“. A:.E\-_E mN -\+ Wm: m-m-mp mmmp
www.mamfi-mi 5.22%.: <00 .0 22 26E 328.3. mmoomz 42.98. ._m .m m:m:N
j... w .v 8 58¢
.. .8... 8.2.888 mmom mm ...m
8.59.6 .528- .m>mu 8m .9... 8... 89% m5: 3 82
82235-2: .2c2omtam Emma-.225 8323. mmom .9 2.98. ..m .m 9ch

 

 

 

 

 

82:22:3- 2222585

 

 

 

 

 

 

 

 

5.82:8 .828 :0 2:20.28 83228:. 03 ooom
802820-21 .o 82.... .Ia :o..o2.xm_ 6:280 000 6:230 ._m .m 23

30:82.. om<

-m 2:22.50

.mcmowu .om...O<

28:32. 052.5

.0 8-0. 8.8.88. .88 3.25

.2528 2:28:28 mcmom: com F 0

E; 08 A88. 8.2.888 m 8.2 8.

882.855. 88 :25 8..-... £898 80828 2 .8 .888 88
mE>Ncw 2:2 82.: 2.0268 8-0. I: 22.28-: 000 cm 3:280 ._m .m aEm...

om<

2:2ma:>0

882820232. .mmmSEm-a com :0

.<-mmm2o::on.. .2890. .82.: 82.4.

.mE>~8>_. oE>Ncm .o 2;. 032.5
:o..m~._.noEE. SEBE om-md. :o..2.:mo:oo 93.5 mm?
mE>~cm wE>~cm .SF-v. I: 828?: 000 on 3:260 ._m .o 9E3

 

 

1O

The principal objectives of this study may be summarized as follows:

1) To elucidate the fundamental structure of CLA, including the hypothesized

“shell” structure;

2) To characterize mass transfer properties for CLA dispersions in agitated

vessels for bio-extractive and catalytic processes;

3) To demonstrate novel applications of CLA, including enhancement of yields,

regioselectivity and enantioselectivity of biocatalytic processes.

References

Csuk, R , Glanzer B. l., 1991. Bakers Yeast mediated transformations in organic
chemistry, Chem. Rev., 91, 49-47.

Lamb, SB. and Stuckey, D.C., 1999. Enzyme immobilization on colloidal liquid
aphrons (CLAs): the influence of protein properties. Enzyme Microb. Technol.
24, 541-548

Lamb, SB. and Stuckey, D.C., 2000. Enzyme immobilization on colloidal liquid
aphrons (CLAs): the influence of system parameters on activity. Enzyme
Microb. Technol. 26, 574-581

Larsson, M., Arasaratnam, V. and Mattiasson, B. 1989. Integration of
bioconversion and down-stream processing. Starch hydrolysis in an aqueous

two-phase system, Biotechnol. Bioeng. 33:758-766.

11

Lily MD, and Woodley JM, 1985, Biocatalytic reactions involving water-insoluble
organic compounds, in Biocatalysis in organic svntheses, ed. by A Tramper, HC
Van Der Pas and Linko P, Elsevier, Amsterdam. 179-192.

Lye GJ and Stuckey DC, 1996. Predispersed solvent extraction of mlcrolide
antibiotics using colloidal liquid aphrons, in Proceedings International Solvent
Extraction Conference, University of Melbourne, Australia, ed. by Shallcross DC,
Paimin R and Prvcic LM, 1399-1405

Lye, G.J. and Stuckey, D.C., 1994. Extractions of Macrolide Antibiotics Using
Colloidal Liquid Aphrons (CLAs) In: D.L. Pyle (ed.), Separations for
Biotechnology Ill, SCl Publishing, 280-286

Lye, G.J. and Stuckey, DC. 1998. Structure and stability of colloidal liquid
aphrons. Coll. Surf. A.: Physiochem. Eng. Asp. 131, 119-136

Lye, G.J., Pavlou, O.P., Rosjidi, M. and Stuckey, DC. 1996. Immobilization of
Candida cylindracea Lipase on Colloidal Liquid Aphrons (CLAs) and
Development of a Continuous CLA-Membrane Reactor. Biotechnol. Bioeng. 51,
69-78

Lye, G.J., Poutiaimem, L.V., Stuckey, DC. 1994. Removal of dilute organics from
industrial waste waters: extraction of 3,4-dichloroaniline using colloidal liquid
aphrons. In: Proceeding 2"d lntemational Symposium on Environmental

Biotechnology, lchemE Publishers, 25-27

12

Matsushita, K., Mollah, A.H., Stuckey, D.C., del Cerro, C., and Bailey, AL, 1992
Predispersed solvent extraction of dilute products using colloidal gas aphrons
and colloidal liquid aphrons: aphron preparation, stability and size. Coll. Surf.
69, 65-72

Mattiasson B. and Olle Holst 1991. Objectives for extractive Bioconversion, in
Extractive Bio Conversions, ed. Bo Mattiasson and Olle Holst, Marcel Dekker,
New York NY.

Rosjidi M, Stuckey DC and Leak DJ, 1994. Downstream Separation of chiral
epoxides using colloidal liquid aphrons (CLAs), in Separations for Biotechnology,
Ed by Pyle DL, SCI publishing. London. 280-286.

Rosjidi, M. , Stuckey, ED. and Leak, D.J., 1994.Downstream Separation of chiral
epoxides using colloidal liquid aphrons (CLAs), Separations for Biotechnology 3,
440-446

Save SV, Pangarkar VG and Kumar SV, 1994. Liquid-Liquid extraction using
aphrons, Separation Technology, 4: 104-111.

Scarpello JT and Stuckey DC 1996. Extraction of Phenylalanine using colloidal
liquid aphrons (CLAs) : stability in dispersion and rates of mass transfer, in
Proceedings of lntemational Congress on Membranes and membrane
processes, Yokohama, Japan. 988-989.

Scarpello, J.T. and Stuckey, DC. ,1999,. The influence of system parameters on

the stability of colloidal liquid aphrons. Chem. Technol. Biotechnol. 74, 409-416

13

Scott, C.D., T.C. Scott, H.W. Blanch, A.M. Klibanov, and A.J. Russell 1995.
Bioprocessing in Nonaqueous Media: Critical Needs and Opportunities.
ORNL/TM-12849, US National Technical Information Service, Springfield, VA.
Sebba F, 1972. J.Co|loid Interface Sci. 40, 468.

Sebba, F. 1987. Predispersed solvent extraction. Chem. Ind. 185-187

Sebba, F., 1984. Preparation and properties of polyaphrons (biliquid foams),
Chem. Ind., 367-372

Stuckey, D.C., Matsushita, K., Mollah, A.H., and Bailey, AL 1993. The
Extractions of Dilute Fermentation Products Using High lntemal Phase
Emulsions (Aphrons). 3rd Int. Conf. On Effective Membrane Processes: New
Perspectives, BHR Group Publications, 3-19

Van Sonsbeek, H.M., Beeftink, H.H., Tramper, J., 1993. Two phase liquid —phase
bioreactors. Enzyme Microb. Technol. 15 722-729.

Wallis DA, Michelson DL, Sebba F, Carpenter JK and Houle D, 1985.
Applications of aphron technology to biological separations, Biotech. Bioeng.
Symp 15 : 399-408.

Zhang, C., Valsaraj, K.T., Constant, W.D. and Roy, D. 1996. Studies in Solvent
Extractions Using Polyaphrons. I. Size Distribution, Stability, and Flotation of
Polyaphrons. Sep. Sci. Technol., 31, 1059-1074

Zhang, C., Valsaraj, K.T., Constant, W.D. and Roy, D. 1996. Studies in Solvent
Extractions Using Polyaphrons. ll. Semibatch and Continuous Contercurrent
Extraction/Flotation of a Hydrophobic Organic Dye from Water. Sep. Sci.

Technol. 31 , 1 463-1 482.

14

Chapter 1

1. lnterfacial Structure of Colloidal Liquid Aphrons

1.1 Introduction

Many industrially important chemical, biochemical, and environmental
processes involve a phase-contacting operation intended to transfer a solute
between two immiscible liquids. The rate of solute mass transfer is often a rate-
limiting factor that limits the overall efficiency of the process. For example, the
nitration of aromatic substances like benzene, toluene, phenol, and napthalene is
practiced on large scales for the manufacture of intermediates for dyestuffs,
polymers, pharmaceuticals and explosives. Most of these nitrations involve two
phases: an organic phase and an aqueous, mixed acid (HN03 + H2804) phase.
Albright and Hanson (1975) determined that under industrial conditions the rate
of nitration is typically controlled by diffusional factors. Similarly, the rates of
alkaline hydrolysis of several forrnate esters and the rates of reduction of
substituted nitroaromatics in mechanically agitated, liquid-liquid contactors are
dependent on interfacial area per unit volume (a), indicating that interphase mass
transfer is rate-limiting. lnterphase mass transfer is often the rate-limiting step in
multiphase biocatalysis, due to the inherent tradeoff between blocompatibility and
solute capacity. Biocompatible solvents tend to have low distribution coefficients,
and consequently, low mass-transfer driving forces (Roffler, 1986). Increasing

the interphase mass-transfer rate would allow a more rapid approach to

15

equilibrium, thereby optimizing the use of the total solute bearing capacity of the
chosen solvent for a given contacting time.

The volumetric, interphase mass-transfer rate (RA) is given by the product
of overall, interphase mass-transfer coefficient (KL), the interfacial area per unit
volume (a) and the concentration driving force (C*-C), where C' is the equilibrium
solubility of the transferring solute. Although KL varies somewhat with the
environmental conditions, its range is relatively small. The concentration driving
force is also typically small, due to the inherently low aqueous solubilities of
nonpolar solutes. Under mass-transfer-limiting conditions, the volumetric solute-
transfer rate can be increased by decreasing the size of the dispersed-phase
droplets, and thereby increasing "a” (Kafarov et al., 1974; Bhave and Shanna,
1981 ). Mechanical agitation (e.g. a mixer settler) is commonly used in mass-
transfer-limited systems for this purpose. However, this approach is energy-
intensive and thus costly, particularly in large-scale systems (Bredwell, et al.
1998)

Colloidal liquid aphrons (CLA) are surfactant-stabilized droplets (1 -50
microns in diameter) of a nonpolar liquid dispersed in a continuous aqueous
phase. The small dimensions of CLA provide very large "a” values (typically
10,000-20,000 m2/m3); consequently solute transfer can often be completed
within seconds. A surfactant film surrounding each droplet stabilizes the CLA
against coalescence. Surfactants are added to both the aqueous and
hydrophobic phases to stabilize the CLA dispersion. An advantage of CLA is that

they can be produced at very high phase volume ratios (PVR) which defined as

16

the volume ratio of dispersed phase to continuous phase; PVR values as high as
20 have been obtained with CLA without phase inversion or coalescence. The
CLA emulsions are also very stable; they have been stored in a stoppered bottle
for years without visible deterioration.

Due to the large interfacial area offered by CLA emulsions, there have
been a number of reported applications in multiphase contacting (Table 1,
Introduction). Most studies have emphasized the applied aspects of the CLA
emulsions, such as their stability and rates of extraction for various applications,
rather than exploring their fundamental properties. However, understanding the
relationship between the structure and the macroscopic properties of CLA is
essential to their optimal use.

In order to explain the small size and high stability of the CLA dispersion,
Sebba1 postulated that a multi-layered, surfactant-stabilized “shell” surrounds the
hydrophobic core. This shell was thought to include an entrapped-water layer, as
shown schematically in Figure 1.1. Sebba (1987) originally used fluorescent dyes
to probe the structure of polyaphrons and CLA and presented indirect evidence
for the existence of a shell, but the structure of the shell has not yet been
determined. Researchers investigating CGA and CLA emulsions since 1982
have generally accepted this structure. However, as yet, no studies have
provided conclusive evidence of this or indeed any other structure. Lye et al
(1998) have investigated the structure of the CLA using cryo TEM, DSC and

other light scattering techniques. However, these studies did not

17

 

 

 

Electrical 1
double
layer ‘X
(diffuse) l ¢
I"; A i“
g: Nonpolar 30";
l 't: Hqukt .$§
l (E: phase i?)
k 3 l
x 9'
Bulk water ” Tr
containing '3’
surfactant 3% A.)
a “*«fffiluI-l—l -5 *

Outer
surface
ofsheH

4

Inner
sunace
ofsheH

 

Figure 1.1. Model for CLA droplets (Redrawn from Sebba, 1987)

18

 

 

provide direct evidence for the structure of the surfactant-laden interfaces
responsible for the stabilization of CLA.

In spite of a substantial body of literature concerning gas and liquid
aphrons, the structure of these systems has often been disputed. It is suggested
that CLA and high lntemal phase emulsions (HIPE) are in fact the same, and the
term aphron be dropped to avoid confusion (Princen, 1988). This may well be the
case since the macrosc0pic properties of the two systems are very similar,
notably creaming and high stability. However, the literature suggests that there
are some significant differences between CLA and HIPE, particularly relating to
their formulation. HIPEs generally contain between 10-50% water soluble
surfactants (Lissant et al 1974, Mannheimer 1972, Princen 1980, Schwartz and
Princen, 1985), whereas CLA emulsions have been generally formulated using
0.1-2 F/o oil soluble surfactant in oil phase and 0.1-2 % surfactant in the aqueous
phase. Lye and Stuckey (1998) emphasized a need for a definitive structural
study to rationalize the understanding and nomenclature in this area and the
effects of these differences.

The objective of this study was to characterize the structure of the CLA
dispersion and identify the factors responsible for CLA formation and stability. A
variety of techniques, including DSC, FFI'EM, SAXS and ITC, were used to
characterize the interfacial properties of CLA emulsions. Surfactant theory was
applied to gain a better understanding of the fundamental aspects of CLA

formulation and develop a model for micellar stabilization of CLA emulsions.

19

1.2 Preparation of CLA em ulsions

CLA were prepared by dropping the oil phase (0.1% v/v Tergitol (15-S-3)
in limonene (Sigma Chemical 00.)) via a burette into a well stirred aqueous
solution containing 0.4% v/v Tween 80 (Sigma Chemical 00.), until the desired

phase volume ratio (PVR: Vorganjc / Vaqueous) was attained. comparative

conventional emulsions were made using identical techniques except that the oil

phase was free of the Tergitol surfactant used in the CLA.

1.3 Differential Scanning Calorimetry (DSC)

1.3. 1 Description

Differential scanning calorimetry (DSC) measures the temperatures and
heat flows associated with transitions in materials. These measurements provide
quantitative as well as qualitative information about the physical and chemical
changes that the sample undergoes. Differential scanning calorimetry (DSC) is
widely used to determine the therrnophysical and thermochemical properties of
unknown mixtures or poorly characterized phases. Examples of thermal
transitions are melting, recrystallization, decomposition, out-gassing, or a change
in heat capacity; DSC also measures the transition onset and ending
temperatures, as well as the temperature at the transition maximum. The
calorimeter measures both endothermic (heat absorbed) and exothermic (heat
released) processes, or changes in heat capacity.

The DSC uses a two-pan system, where the sample is placed in one pan

and the other pan is left empty as a reference (Figure 1.2). Sample size is usually

20

between 5-10mg of material. The two pans are exposed to the same heat
source and the difference between the heat evolution or absorption amongst the
two pans is measured. In either case, the objective is to measure the amount of
energy absorbed or released from the sample. Samples can be tested from -
150°C to 725°C. The resulting energy absorbed or released by the sample is

then plotted against the temperature ramp.

 

sample sample reference

m\ / f...
\ l /

O
[DIE—1 (J

(I m

L... /

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

computer to monitor temperature
and regulate heat flow

 

 

 

Figure 1.2. Schematic diagram of Differential Scanning Calorimeter (DSC)

21

1.3.2 Materials and Methods

DSC measurements were carried out using a heat flux Mettler TA 4000
system. A glass disk provided the primary means of transferring heat to the
sample and reference positions, while also functioning as one element of the
temperature measuring thermoelectric junctions. The CLA sample was placed in
an aluminum pan fitted with a lid which could be sealed using a Mettler press. A
sealed empty pan was used as a reference. Both the sample and reference pans
were placed on the glass disk that was gradually heated, while the temperature
was monitored. The melting of the phases in the emulsions was accompanied by
an endothennal peak which was automatically recorded. Measurements were
carried out between —50°C and 30°C.

1.3.3 Results

Melting endotherrns of CLA emulsions are shown in Figure 1.3 between -
10°C and 15°C. At least two separate phase transitions can be seen in the CLA
samples around the melting point of water, indicating the presence of two distinct
phases containing water. The colligative properties of the surfactants in water
would be expected to lower the melting point of the “shell” phase as is indicated
by the freezing point depression of the smaller of the two peaks. The sharp
transitions observed during melting indicate the crystalline nature of the phase.
Further studies were undertaken using Freeze fracture transmission electron

microscopy to visualize the lntemal structure of this “shell” phase.

22

 

 

 

 

 

 

 

DSC
1 r
-1 85 c
0 ..-._ 2 366 J/g , -0 77 C.
-1.99 C’Z ‘
-1.46 C.
3 -0.3489 w/gl
.0 -1~
Ea
(U l
0 E if
I V -2* " ,’
_3 1 I 1 ...... .3"
so -40 -20 0 20
Temperature. C

 

 

Figure 1.3. DSC endotherrn for CLA dispersion containing 0.1% Tergitol 15-S-3
in Limonene and 0.4%Tween 80 in Water (PVR 0.25).

23

1.4 Freeze-FracturelFreeze Etch Transmission Electron Microscopy

1.4. 1 Description

Detailed studies of the interfacial regions of fine aqueous suspensions
require resolution below the approximately 300 nm limit of visible-light optical
means. However, electron microscopy must be accomplished in near perfect
vacuum conditions that preclude the direct observation of such samples.
Conventional TEM sample fixation, used in the life sciences, provides no regimen
that would yield relatively artifact free preparations with sufficient electron
contrast to study aphron matrix interfaces (Dawes, 1971 ). An alternative is to
create a high resolution topographical replica of the area of interest. Freeze-
fracture and freeze-etch are TEM replica techniques where two-staged replicas
are formed on the fractured surface of frozen samples (Rash and Hudson, 1979;
Severs and Shotton (1995)). The initial fracture surface, in both cases, follows
the path where the fracture strength of the frozen sample in at a minimum. In the
case of frozen cells, for example, the fracture path always follows the middle of
the unit membranes with integral membrane proteins either pulled from one of
the leaflets or snapped by breakage of covalent bonds. In freeze-fractu re, the
contrasting metal (Pt) is evaporated directly on the freshly fractured surface of
the sample, under vacuum. Freeze-etch incorporates a brief period of low-
temperature sublimation, again under vacuum, at a temperature sufficient to
drive the sublimation but below the rapid re-crystallization temperature of the

phases present (Figure 1.4).

24

 

Water

 
   

Freeze

1 FractUre

 

 

 

 

 

 

 

 

 

Figure 1.4 Freeze Fracture and Freeze etching of “membraned“ cells or
emulsions.

25

 

 

 

 

 

 

 

 

 

-1 as c°
, I) \\
ll —
. ,_....--'
U V\c0|d Razor
I-I Blade
0 . 0 Frozen Aphron
_1 90 60"150 C Suspension or
Gold Stage
/V\/\/\ M
B

 

 

 

 

 

 

 

 

 

 

Figure 1.5 Schematic diagram of sample preparation in Freeze Fracture TEM.

26

In both cases the first coating applied is a discontinuous film of Pt metal with a
grain size of about 1 nm. This is applied at an oblique angle to the surface and
preferentially accumulates on the evaporator side of topographical features. The
contrasting film is immediately stabilized by the deposition of 10 -20 nm of
evaporated amorphous carbon.

After completion of the replica, the sample is dissolved away in suitable
solvents and the replica cleaned and mounted on an ordinary copper TEM
support grid (Figure 1.5). The resulting preparation is viewed in brightfield mode.
Generally the image is presented as a negative, giving the impression that the
surface is illuminated by the contrasting metal.

1.4.2 Materials and methods

For this investigation, CLA emulsions with a PVR of 0.25 were used.
Aliquots (~30 uL) of the emulsions were spread on 0.15 mm thick, 10 mm

diameter copper supports. These samples were ultra-rapidly frozen by plunging

the supports in liquid propane (490°C). For the pure limonene reference
sample, the freezing was accomplished by plunging the support into nitrogen
slush. I

The frozen samples were inserted onto the pre-cooled rotary stage of a
freeze-etch unit (modified Balzers BA-510) and fractured by a knife at 150°C with

a residual pressure of approx. 6.5 x 10'5

Pa. The replicas were shadowed, with
~2nm of Pt at a 30° angle to the plane of the sample followed by 10-15 nm of
carbon (at ==90° angle) by arc evaporation. Etching was accomplished by raising

the stage temperature to -100°C, positioning the knife holder (~-196°C) over the

27

sample surface and allowing it to sublime for 3 min prior to shadowing. Deep
etching of the mixed surfactant solution was allowed to progress for 8 min. The
resulting replicas were washed in a 10% graded isopropanol series, rinsed in de-

ionized H20, and then mounted on 300 mesh grids. TEM was performed on a

JEOL 1OOCX ll operated at 120keV with images recorded on Kodak SO 163 film.

The negatives were digitized with a Polaroid® SprintScan 45 scanner and

processed with Adobe® PhotoShop. Final prints were made on a dye sublimation

printer with the shadows printed dark.

1.4.3 Results

Limonene emulsions and CLAs were morphologically similar at the light
microscopy level. Figure 1.6 shows phase-contrast light micrographs of
conventional emulsion and CLA dispersions. The overall form the CLA prepared
in this study is superficially similar to a conventional emulsion. As seen in Figure
1.6, the droplet sizes averaged about 20 pm in diameter, for both systems, with
similar size distributions for each. Sebba (1984) mentions the use of differential
staining and uses light microscopy to characterize the outer shell of the CLA;
however no comparison with similarly formed conventional emulsions was made.
Indeed, the multilameller CLA interfacial region presented is below the resolution
limit of conventional phase contrast light microscopy. In addition, accurate
separation of the details of a continuous interface from diffraction artifact at the
surface of spherical objects is problematic.

TEM allows resolution of macromolecular dimension; however, the sample

must be rendered stable at very low pressure. Typically this is accomplished

28

either by keeping the specimen at a temperature below which there is no
significant vapor pressure in the microscope column, or “fixing” the specimen
chemically to allow the removal or conversion of the volatile phases. One of the
major concerns in either direct cryogenic observation or the use of frozen
samples is the possibility of freezing-rate artifacts. In general, cooling rates on
the order of 104 °C/sec must be achieved to create glass like aqueous phases.
However, the addition of solutes to the aqueous phase depresses the freezing
point and allows accurate representation of the system at significantly slower
cooling rates. The addition of cryoprotectants to freeze-fracture systems has
been extensively investigated (e.g. Rash and Hudson, 1979).

The emulsion and CLA replicas presented here are relatively free of the
typical segregation of components seen with phase separation at low cooling
rates (Chiruvolu, et al., 1994). If TEM resolvable ice crystals are formed during
the quenching process, a eutectic is formed which gives a characteristic plate-
like surface on freeze fracture (Willison and Rowe, 1985). If there are
membranes present, the growth of ice crystals beneath them causes the fracture
face to have a characteristic roughness (Gilkey and Staehelin, 1986). None of
these features were seen in the samples prepared here. Finally, the fracture
character of the aqueous phase exhibited the characteristic topography seen in
glass fracture (Doremus, 1994).

In order to determine if there were any inherent ordering in either the
aqueous (0.4% Tween 80) phase or in the limonene (0.1% Tergitol) phases, we

compared the fracture surfaces of these two components separately. In Figure

29

1.7 the conchoidal fracture of the aqueous phase (Figure 1.7A) is consistent with
the fracture of a glass-like material with no long-range order. The fracture
characteristics of the organic phase were more planar with less evidence of the
curving surface features of the aqueous phase. The limonene phase did exhibit
a coarser surface texture, however.

In Figure 1.8, the replicas of the standard emulsion droplets are shown
freeze-fractured (Figure 1.8A) and after 3 min of etching (Figure 1.88). In both
cases the interface between the organic phase droplet and its surroundings
appear as simple fracture interfaces with no fine structure.

Freeze fracture of CLA emulsion yielded organic phase droplets with a cross
fracture habit similar to the conventional emulsion. Both the freeze fractured and
freeze etch preparations of the CLA’s show a layered interface (Figure 1.9A), a
feature clearly not present in the conventional emulsions. These lamella fractured
showing varying surface steps approximately 20 to 40 nm high (Figure 1.98).
Etching the CLA fracture surface prior to replication (Figure 1.10A) further
accentuated these lamellar steps. At higher magnification (Figure 1.108), the
steps can be seen to have an additional texture approximately parallel to the step
faces. These were also spaced about 15-20 nm.

Figure 1.11A illustrates the ordering of the detergents in an aqueous
phase, without the presence of the organic phase as can be seen in the FFT
(inset) of the deeply etched combined surfactants in Figure 1.118 there is a
regular periodic spacing widthwise in this material. Since the vapor pressure of

the surfactants is much lower than water, especially at the temperatures used in

30

freeze fracture, it can be reasonably assumed that the specific dimensions
revealed are due to the preferential sublimation of water. When the surface was
deep etched (8 min), these areas were seen as elongated ridges spaced
approximately 20 nm apart (Figure 1.118). The average spacing between
interfacial layers in the CLA (Figure 1.108) is difficult to assess due to the
problem in calculating the exact angular orientation of the cleavage plane as it
varies around the spherical aphron. Planar cross fractures were not observed in
the replicas viewed in this investigation. However, based on Figures 1.7 and 1.8,
it appears that the lamellae are spaced further apart than is observed based on
the measurements of the mixed surfactants alone. Addition of the organic phase
is likely to cause a rearrangement in the packing of the surfactants, contributing
to a dilation of the interlamellar regions.

The combined results from freeze fracture/T EM of the aphron interface at
low PVR indicate the presence of stable multilamellar interfaces in the CLA
emulsion. The lamellar liquid crystalline phase, observed in this study, is one of
the most common liquid crystalline phases observed in surfactant-water systems.
The formation of lamellar liquid crystalline micelles has been observed by various
investigators (Dimitrova et al, 1995; He et al, 1993; Lang and Morgan, 1980).
Most surfactants, at high concentrations, pack in such a manner as to form
lamellar micelles. The hydrocarbon chains are in a dynamic disordered state,
which is similar to that of paraffins in liquid state, and the amphiphile bilayers are
separated by water layers. Small angle x-ray scattering was used to obtain the

periodic spacing amongst these liquid crystals.

31

 

Figure 1.6. Phase contrast micrographs of A. CLA emulsion and 8.
Conventional emulsions. The average droplet size, =20 um was the same for
both emulsions.

32

 

Figure 1.7 Fracture surfaces of separate phases. A. Water and Tween 80 frozen
in liquid propane at —190°C. 8. Limonene and Tergitol frozen in a nitrogen slurry
at =-210°C.

33

 

Figure 1.8 Replicas of conventional emulsions. A. Freeze fracture and 8.
Freeze etch. Inset shows the detail of the droplet matrix interface for each at 5X
increased magnification.

 

Figure 1.9 A. Freeze fracture of CLA suspension showing the multilamellar
phase boundary. 8. Enlargement of the boundary showing varying surface
steps approximately 20 to 40 nm high.

35

 

Figure 1.10 A. Freeze etch of a CLA emulsion showing enhanced detail of the
phase boundary. 8. Enlargement showing the 20 to 40 nm steps and additional
ridges running parallel to the step faces.

36

 

Figure 1.11 A. Freeze fracture surface of a mixture of Tergitol and Tween at the
concentrations used to form the CLA. B. The same solution, etched for 8 min
prior to shadowing. Inset shows a fourier transform of the etched sample.

37

1.5 Small Angle X ray Scattering

1.5. 1 Description

Small angle x-ray scattering is a technique for studying structural features
of colloidal size. The detailed theory of small angle x-ray scattering can be
obtained from Glatter et al. (1982). A brief discussion is presented in this section.
Any scattering process is characterized by a reciprocity law, which gives an
inverse relationship between particle size and scattering angle.

The general scattering formula is
Ilh) =FF' = jjjjjjdV. -dV. -p<r.).p(r.)-e““":"=’ (1)
Equation 1 is a Fourier integral, and can be simplified by integrating over all r =

(r1-r2) that are equal, then by integrating over all the different r.

This first step is the mathematical operation of auto-correlation, and is defined by

autocorrelation 32 (r) = jjjdv, p(rl )p(r2) (2)

with r=(r1-r2)=constant.

The resulting function, well known as the Patterson function, and widely used in
crystallography, has the following properties: every electron pair with relative
distance r can be represented by a point in fictious C-space, say. The density of
these points can be represented by p2(r). As every pair is counted twice with r
and -r, it follows that the distribution in the C-space must show a center of

symmetry.

38

The second step consists of integration over C-space

l(h) = m dV .320) - e’"" (3)
So the intensity distribution in h or reciprocal space is uniquely determined by the
structure of the object, as expressed by p2(r). Conversly, the latter can be drawn

from l(h) by the inverse fourier transform
3%) = (1/21r)3fijdhxd1ydhz -l(h) . e‘"" (4)

One quite general conclusion can be drawn from (3) and (4) : there is a
reciprocity between ordinary and reciprocal space. As they are connected by the
phase hr only, the result will be the same, when r is enlarged or h is diminished
by the same factor. So large particles will give a diffraction pattern concentrated t
small angles.

Especially with particles of colloidal dimensions, and with a usual
wavelength of about 1A, the pattern is limited to a range of one or two degrees.
This is a typical domain of small angle scattering. There are two main restrictions
that enable considerable simplification of the theory of small angle scattering
1) The system is statistically isotropic. It makes no difference here whether this is
a property of the structure itself or a consequence of some change in time
(rotation of particles or the like).

2) There exists no long range order. This means there is no correlation between
the points separated widely enough.

From restriction (1) it follows that the distribution of p2 in C-space depends only

on the magnitude r of the distance, though this will not hold true for p(r) in

39

ordinary space. Likewise the phase factor e'ihlr can be replaced by its average
taken over all directions of r. This is expressed by the fundamental formula of
Debye:

< e'"" > ___ sinhr
hr

 

by means of which (3) can be reduced to the form

l(h) = j4m2dr - 32m 5'2?

 

Diffraction is produced by the interference of waves scattered by an
object, In case of X-rays striking an object, every electron becomes a source of a
scattered wave. The scattered waves are coherent. Coherence means that all
amplitudes are added and the intensity is then given by the absolute square of
the resulting amplitude. The amplitudes are of equal magnitude and differ only by
their phase (I), which depends on the position of the electron in space. It is
convenient to represent a single secondary wave by a complex form: ei". The
phase o is 21W» times the difference between the optical path and some arbitrary

reference point.

The path length difference of two rays shown in figure 1.10, is

AB+BC = 2 d sine (10)
where 0 is referred to as the glancing angle. For many glancing angles the path
length difference is not an integral number and the waves interfere destructively.
However, when the path length difference is an integral number of wavelengths,

the scattered waves are in phase and interfere constructively. It follows that a

40

bright reflection should be observed when the glancing angle satisfies the Bragg
Law

nl=2dsin9 (11)
Reflections with n=2,3...are called second order, third order, and so on they
correspond to pathlength differences of 2.3....wavelengths. The primary use of
Bragg's law is in the determination of the spacing between layers in the lattice,
for once the angle 0 corresponding to a reflection can be determined, d may be
readily calculated.

The characterisitic pattern of X-ray diffraction lines obtained for a liquid
crystalline lamellar structure is a series of sharp lines in the relative ratio
1:1/2:1/3:1/4. The first order long spacing corresponds to the net spacing
between the two bilayers including the interlayer water phase. The characteristic

variable reported for x-ray scattering is known as q or h, where q: (Mr/7t) sine.

where 20 is the scattering

 

 

9 9

 

 

.\
ou-‘é '-'-O-.-
I

 

 

 

 

 

Figure 1.12 X-ray diffraction from lattice planes (Braggs law)

41

angle and it is the wavelength and is related to the characteristic bragg length (d-

spacing) as d=21r/q.

1.5.2 Materials and Methods

Surfactant solutions were prepared using MilliO deionized water. Small
angle X ray scattering experiments were performed using a Kratky Camera. The
X-ray source was a Rigaku Rotaflex rotating anode with a copper target
operating at 7.5-10W. The k0, wavelength of 1.54 A was selected using Nichol
filters. The energy window on the 10 cm OED linear position sensitive detector
was set to accept only the scattering signal with energy close to K0,. The input
collimators and rectangular holes produced a 2*0.13mm X-ray spot on the
sample, which was sealed in a special glass capillary having a diameter of 1mm.
The sample to detector distance was 68.5cm. The accessible range of q was
0.01-0.4A".

The scattering data accumulated over 30-240 minutes was corrected for
background scattering by subtracting the scattering intensity of water and the
empty capillary.

1.5.3 Results

Small angle X-ray scattering by liquid crystal phases generates scattering
peaks in KO), where l is the scattered intensity and 0 is the scattering angle. The
SAXS studies (Figure 1.13) unambigiously identified the lamellar phase spacing
in this system as 188 A. This is in close agreement with the spacing observed in

the FFTEM studies discussed earlier. Only a first order peak was observed,

42

which moved to a higher wavelength (corresponding to a d-spacing of 203 A)
with dilution from (1% to 0.1%v/v Tween 80 in water, keeping the molar ratio of
Tween to Tergitol at 1:1) indicating swelling of the lamellar phase. These
observations are in agreement with those reported for lamellar phases in dilute
water-C12E5 systems (Strey et al., 1990). FFTEM studies indicated that the
lamellar phase formed only in the presence of both emulsifiers. Hence, it was
necessary to find the stoichiometry and energetics of this interaction, in order to
explain the mechanism of formation of the lamellar micelles. These studies were

conducted using lsotherrnal titration calorimetry.

43

2000

O

1500 <

-1000‘

 

 

 

1.5 2 2.5

26

Figure 1.13. SAXS data for 0.1% Tergitol, 0.4% Tween 80 solution in water

1 .6 Isothermal Titration Calorimetry

1.6. 1 Description

Isothermal titration calorimetry (ITC) is a techniquethat measures the heat
of interaction between two molecules. In ITC, a solution containing a ligand is
titrated into a cell containing a solution of the macromolecule. As the two
elements interact, heat is released or absorbed in direct proportion to the rate
and energetics of binding that occurs. When the macromolecule in the cell
becomes saturated with added ligand, the heat signal diminishes until only the
background heat of dilution is observed. The area underneath each injection
peak (Figure 1.14) is equal to the total heat released for that injection. When
this area is plotted against the molar ratio of ligand added to macromolecule in
the cell, a complete binding isotherm for the interaction is obtained. ITC is a true
in-solution method. It does not require immobilization of binding components as

in surface plasmon resonance (SPR), or chemical tagging as with fluorometry.

45

Time (min) 10 ' 20 30 40

 

 

     
 

 

 

 

0 _ ..
' 8
e “ - ‘ ' '-
3' _ . ‘ . - Raw Data "
E _ ' I 1 ' 1 t I I I 1
fl .
‘6 J . '
g. — Binding lsotherrn I
.5 3 Model 1 Site 1
.5 .2 N=1.02 :1: 0.0015 ‘
5 _ K=5.57E4:1.01E3
g : H=1.359E4:29.18 I
x _ I
l l I I I 7 I - I
0 0.5 1.0 1.5 2-0
' Molar Ratio

Figure 1.14. Typical ITC experimental data and analysis (MicroCal Inc, MA)

46

ITC has been extensively used by biochemists to study interactions in biological
systems such as: protein-small molecule & enzyme-inhibitor interactions,
protein-carbohydrate interactions, protein—protein interactions, protein-lipid
interactions, lipid & lipid-small molecule interactions, nucleic acid-small molecule
interactions, protein-nucleic acid interactions, nucleic acid-nucleic acid
interactions, antibody studies, receptor interactions, and protein folding. Several
reviews on the use of ITC for interaction studies have appeared in the literature
(Baker et al., 1996; Blandamer et al., 1996, Ladbury, et al, 1996, Wadso, l, 1995,
Cooper and Johnson, 1994).

1.6.2 Materials and Methods

Microcalorimetry experiments were performed on a Microcal, Inc VP-ITC
calorimeter. The injection syringe was filled with a concentrated stock solution
containing 2% (15.27mM) Tween 80 solution. Sixty injections of 3 uL each were
made into a 1.38 mL calorimeter cell, which contained 0.1% (3.14mM) Tergitol
15-S-3 solution. The cell was stirred at 400 rpm and was thermally equilibrated at
25°C for 30 min until a smooth baseline was obtained, and injections were
initiated. The therrnogram was allowed to return to baseline between injections (6
min between injections), and each injection had a 6s duration. In order to correct
for heats of micellization/ demicellization due to dilution in the cell, three control
runs were conducted. First, deionized water was titrated with deionized water
using the procedure described above. Second, deionized water was titrated with
0.1% Tergitol 15-S-3 to correct for heats of demicellization caused by dilution of

Tergitol in the cell. Lastly, 2% Tween 80 was titrated into deionized water to

47

correct for the heats of micelllzation/demicellization of Tween 80. Before data
analysis the data for control runs was subtracted from the data points for the
Tween80-Tergitol experiment to correct for the heats of mixing. Peaks were
integrated using software from Microcal to yield enthalpy generated for each
injection.
1.6.3 Results

Figure 1.15 shows the ITC data for the sample titration along with three
control runs (i.e. water titrated with water; water titrated with 0.1% Tergitol
solution; and 2% Tween 80 solution titrated with water). Data in Figure 1.10
shows that the heat effects for the sample run are quite large compared to those
of the control runs, and can be attributed mostly to the interaction between
Tween 80 and Tergitol. The heat effects for the sample run appear to have two
kinds of interactions. For the first 15 injections, heat effects are all exothermic
and their magnitude decreased quite rapidly. For the next 20 injections, the
injection peaks show some endothermic heat besides the exothermic heat. After
40 injections the heat effects seem to level off. Some residual heat effects were
still observed and maybe attributed to the heat of mixing and a small pH
mismatch between the Tween 80 (pH 6) and Tergitol (pH 6.8) solutions. The
integrated data was fit to two sets of independent sites as shown in Figure 1.16.
The strong sites have a binding constant of ~2.3x105 M'1 and the weak sites have

a binding constant of ~6.3x104 M". The binding heat for the strong site binding is

48

 

.293 .28 8.2...
8 :85 .8 ... 22...... 88.8.8 8.2... 8 :85 .8 .m .29.: .... ..o 8.2... 8.2. 888.8
.m :22... 8055.0... 822... .22... 83.8.00 .P m::m. 0.: .o. 220 oE: .m> 00226: .82.. m. 2 2:9“.

2...... we:

owe can

it». all! i. ll.

 

 

 

 

 

 

 

 

. ...:

h.

 

 

2..” can can :3 2:. on c

.... , .. ....— , .:. ... .l. .. .

 

ill .. t .2...—
.v

L F.EE:.E..FCEL ....CLt . ... C. E . . ...:: .... CC... .. F: ........Ct.b.C.~.E:~C...
0."

L..-.h.h..r.- bb.rrbtbt.hrp.p 55.51?er births...» . PL rblp t..-» Phi-...»..PPLFPPHL Pb at»... ..-»th _. p.- rtPsFFPFPer
I .’

l It..-;»)ll.l 51.....- -.

! .lttti.‘li.t .llill-c-I Ill . (It... 3|.) 2.... :1‘!‘..§..... ..

low

.5

 

49

 

 

 

 

 

I I r I i r 3 I 5'
-0 _ 1
.1 -
1 1
‘ N1 0.0371 10.00624 ‘
-1 _ K1 2.346E5 14.19E4 4
AH1 -1 .281 E4 1 3006 1
- 1 A81 -1340
g ‘ 1 N2 0.2660 2. 0.00435 ‘
g -l K2 6.261 E4 2 7295 -
'_:‘_' . AHZ 1319 1 108.6 J
§ -2 4 A82 26.37 _
0
g 1 .l
g . l
x d
.3 .l _
. l
1 1
l i I ' l . I ' l '
0.0 0.2 0.4 0.6 0.8 1.0
Molar ratio

Figure 1.16 Analysis of ITC data based on two site model.

50

exothermic (~12.8 kcal/mol) while that of the weak site is endothermic (1.32
kcal/mole). The N values for the strong site (0.03) are much smaller than that for
the weak site (0.27).

Since Tergitol is initially in colloidal form in water (highly turbid solution)
the strong site binding may simply be the penetration of Tween 80 into the
colloidal particles and their subsequent destabilization. Later the association of
the two surfactants accounts for the weak site binding. The weak site binding is
strongly endothermic but the entropy of the system increases considerably
(26.37 kcal/ K), indicating the spontaneity and entropy driven nature of the
interaction. A high positive change in entropy is expected for this system since
the association of the surfactants will result in an increase in the the degree of
disorder of the water molecules, surrounding the E0 groups in the Tergitol
molecule. Since both surfactants have some degrees of polydispersity in
molecular weight, the results of this experiment are somewhat qualitative.
However, an N value of 0.27 for the weak site indicates the approximate
stoichiometry of interaction being approximately 4 molecules of Tergitol per

Tween 80 molecule in the system.

1.7 Development of a conceptual model for CLA stabilization by lamellar
micelles

Based on these experimental results, a conceptual model was proposed
to explain the formation of the lamellar phase observed around the CLA droplets.

Lyotropic liquid crystals consist of ordered micelles and are formed by interaction

51

between micelles (T iddy, 1980). A number of theories to describe micelle shapes
have been published (lsraelachvili et al, 1976, Tanford, 1980, Mitchell and
Ninham, 1981). These theories involve the balance of repulsive forces between
the alkyl chain and water, and those between adjacent head groups within
micelles together. They also include surface curvature and steric effects into
consideration. These geometric limitations impose restrictions on the allowed
shapes of micelles and force some of the amphiphiles to assemble into those
shapes, which appear to be thermodynamically unfavorable. The concept of
geometry of the emulsifier, developed by lsraelachvili et al (1976) was used in
this study. According to their concept, the emulsifier is characterized by its
volume v, the length of the hydrocarbon chain I, and the cross-sectional area of
the head group a. Different values of the ratio We! give different surfactant
packing possibilities. Lamellar systems are generally found if this ratio lies
between 1/2 and 1 (Mitchell et al. 1983). For lamellar liquid crystalline phases to
occur this ratio has to be ~1. Based on our experimental evidence for these
phases and the measurement of the association stoichiometry of 4 Tergitol 15-S-
3 molecules per Tween 80 molecule the mixed micellar bilayer structure shown in
Figure 1.17 was proposed. The highlighted molecular complex in the lamellar
micelle would satisfy the geometric criteria proposed by lsraelachvilli et al (1976).
Such a complex would be held at the interface, since one of its components

(Tween 80) is appreciably soluble in water (HLB 14.0) and the other Tergitol 15 -

52

 

 

 

 

 

Figure 1.17. Conceptual model for mixed lamellar micelles formed by Tergitol 15-
8-3 and Tween 80 in water.

53

8-3 is appreciably soluble in oil (HLB). Thus, the Tergitol cannot pull the Tween
80 molecules into the oil phase and the Tween 80 cannot pull the oil soluble
Tergitol into the water. Each component of the complex holds the other
component at the interface so that a very high concentration is obtained at the
interface. The positive entropy of formation indicates that the complex is stable
when formed, and hence only minute concentrations of each component are
required to cover the oil/water interface and stabilize the emulsion. Similar
molecular complexes at interfaces have been reported by other investigators, as
early as the 1940s. Schulman and Cockbain (1940) hypothesized the presence
of such complexes between cetyl sulphate and cholesterol, for stabilization of
Nujol-water emulsions.

1.8 Explanation of CLA p roperties

1.8. 1 Ease of formation and small droplet sizes

Several investigators have reported the formation of CLA emulsions at
relatively low shear rates (Sebba et al, 1984, Matsushita et al, 1992). The most
important factor that influences the ease of formation of emulsions and their
stability is the rate at which the emulsifier is adsorbed on the surface and the
interfacial tension of the system. Ford et al. (1966) have reported that the
adsorption of relatively hydrophilic emulsifiers usually occurs within a few
milliseconds but the more hydrophobic emulsifiers may migrate slowly to the
interface. Under these conditions the formation of complete interfacial film may
take several minutes. In the case of CLA systems, however, the hydrophobic

emulsifier is present in the organic phase, so the distance traveled to the

54

interface is substantially less than that for travel through the continuous aqueous
phase to the interface. The interfacial tension of this system (Limonene/ Tergitol
15-S-3/ Water/ Tween 80) is very small (~1mN/m2, measured using Wilhelmy
plate method), so the dispersion of the oil in water involves a small increase in
surface energy. Hence the system is readily dispersible, and when the small oil
drops are formed, the two emulsifiers interact at the interface forming a rigid
interfacial film that prevents coalescence.

1.8.2 High dispersed phase volume ratios

The rigid interfacial film formed by the two emulsifiers is believed to be
responsible for the high phase volume ratios possible using CLA. Sebba (1987)
reported a PVR of 20 attainable in kerosene polyaphrons. At high lntemal phase
ratios the CLA dispersion is transformed into a polyhedral foam (Sebba, 1987,
Zhang et al., 1996a), where the individual oil polyhedral cells are prevented from
coalescence by an interfacial film formed by the molecular complex (Figure 1.18).
At high phase volume ratios (PVR>3), the interfacial properties of the CLA
emulsions are similar to those of high lntemal phase emulsions (HIPE). Both
systems consist of polyhedra surrounded by lamellar films that prevent collapse
or coalescence of cells. However, since the molecular complex is stable, and
because each surfactant holds the other at the interface at high interfacial
concentration, the amount of emulsifiers required in CLA emulsions (0.1-1%) has

been lower than that traditionally observed by HIPE researchers (5-10%).

55

 

{3.

 

 

 

 

 

 

Figure 1.18. CLA emulsion comprising Limonene (0.1% Tergitol 15-S-3) and
water (0.4% Tween 80) at a PVR of 14 (photomicrograph, magnification 400x)

56

1.8.3 Influence of nature of the dispersed phase on maximum phase

volume ratios

The polarity of the lntemal dispersed liquid phase also influences the
packing of the emulsifiers at the interface as well as the partitioning of the
emulsifiers between the bulk and the interface. These factors change the
characteristics of the interfacial complex, hence changing the net surface areas
that can be stabilized by the complex. Becher (1963) found that the
polyoxyethylene head groups of nonionic surfactants interact with polar
compounds. As a result the polar head group of the aqueous phase surfactant is
no longer completely immersed in water, and a part becomes immersed in the oil
phase. Such interactions do not occur with aliphatic hydrocarbons such as
decane. The reduced stability of CLA emulsions formed using oils with high
aqueous solubility may also be explained in terms of molecular diffusion. Small
droplets formed in the emulsification process will dissolve rapidly and the larger
droplets will grow at their expense. Matsushita et al (1992) could not form CLA
emulsions using butan-1-ol, butan-2-ol or ethyl acetate, all of which have
appreciable solubilities in water. For compounds with lower (but not negligible)
water solubilities, such as butyl acetate (0.43 wt %, Dean, 1992), or amyl acetate
(0.17%, Dean, 1992), maximum PVR obtained in stable systems was limited to 5.
In this study, a PVR of 17 was obtained with limonene (13 mg/L; (Toxicology
Update, 1995)) and up to a PVR of 32 with isooctane (0.25mg/L; McAuliffe, 1965)

as the organic phase.

57

1.8.4 High stability

Several papers have reported that CLAs have remained stable for a
period of several months or even years (Sebba, 1984; Lye and Stuckey, 1998).
The high interfacial dilational viscosity of mixed emulsifier films provides viscous
resistance against coalescence from the flocculated state (Kirikou and Sherman,
1979; Adamson and Gast, 1997). The rate of drainage of liquid films between two
parallel plates is inversely proportional to the viscosity of the medium between
the plates (Reynolds, 1886). This translates into several fold larger drainage
times for CLA emulsions after creaming, thus accounting for the observed
stability periods. In addition, drainage may be retarded by the Marangoni effect
(Adamson and Gast, 1997). In this phenomenon, the flow of the liquid in the film
and the flow of the surfactant along the interface are coupled, so that film
drainage results in surfactant depletion on the draining film surface. The resulting
increase in interfacial tension in the surfactant-depleted portions of the film
creates an interfacial tension gradient that opposes the transport of surfactant.
Because of the strong coupling between the surfactant and the film liquid, the
drainage of the liquid is also retarded by this mechanism. There is evidence in
the literature that hydrophilic polyoxyethylene chains (such as those present in
Tween 80) may contribute to the stability of emulsions due to the strong
interaction with the water in the film (Boyd et al., 1972), thus retarding film

drainage.

58

1.9 Conclusions

The interfacial structure of CLA emulsions has been characterized using
DSC, FF-TEM, SAXS and ITC. Differential scanning calorimetry results indicated
the presence of a separate phase, apart from the aqueous and organic phases
present in the emulsion. Freeze Fracture TEM demonstrated the unique lamellar
interface formed in CLA emulsions. The formation of a rigid interfacial film, due to
the interaction of the two contrary emulsifiers (Tween 80 and Tergitol 15-S-3) has
been shown to be responsible for the stabilization of these emulsions. lsothennal
titration calorimetry was used to characterize the energetics and stoichiometry of
interaction between Tween 80 and Tergitol 15-S-3. Similarities and differences
between properties and formulations of HIPE and CLA emulsions were briefly
addressed. A model of lamellar micelles stabilizing the CLA interfaces was
proposed based on the geometric theory of surfactant packing. Properties of CLA

emulsions observed in the literature were explained based on this model.

1.10 References

Adamson A.W., and Gast AP, (1997). Physical Chemistry of Surfaces, Sixth ed.
John Wiley & Sons, NY

Albright, L. F., and Hanson C. (1975). Industrial and Laboratory Nitration (Am.
Chem. Soc. Symp. Ser. No. 22), American Chemical Society, Washington DC
Baker, 8. M. and Murphy, P. K., 1996. Biophysical Journal, 71, (4), 2049-2055.

Becher P. 1963. J Colloid Science, 18, 665

59

Bhave RR. and Sharma MM. 1981. Liquid-liquid reactors: operation under
refluxing conditions: measurement of effective interfacial area, Trans ICHEME,
59 161-168.

Blandamer, M. J., Cullis, P. M. and Engberts, J. B. F. N. 1996. Pure and Applied
Chemistry, 68,(8),1 577-1582.

Boyd J, Parkinson C and Sherman P, 1972. J Coll. Interface Sci., 41, 359.
Bredwell M.D., Worden RM. 1998 Mass-transfer properties of microbubbles. 1.
Experimental studies, Biotech. Prog. 14: (1) 31-38.

Cooper, A. and Johnson, C. M. 1994. In Methods in Molecular Biology, Vol. 22
(Ed's Jones, C., Mulloy, 8., Thomas, A. H.), Humana Press, 109-24.

Dawes, C.J.,1971. Biological techniques in electron microscopy. Barnes & Noble
NY. 193.

Dean JA, 1992. Lange's Handbook of chemistry, Fourteenth Edn., McGraw Hill,
Inc, NY.

Dimitrova GT, Th. F. Tadros, and Luckham PF, 1995. Investigations of the phase
changes of nonionic surfactants using microscopy, differential scanning
calorimetry, and rheology 1. Synperonic A7, a C13/C15 Alcohol with 7 mol of
ethylene oxide, Langmuir 11, 1101 -1 1 1 1.

Ford RE and Furrnidge CGL. 1966. Studies at phase interfaces: II The
stabilization of water in oil emulsions using oil soluble emulsifiers, J Coll.
Interface. Science, 22, 331-341.

Glatter O and Kratky O, 1982. Small angle X-ray scattering, Academic press, NY.

60

He M, Hill RM, Lin Z,Scriven LE, and Davis HT, 1993. Phase behaviour and
microstrucutre of polyoxyethylene trisiloxane surfactants in aqueous solution. J
Phys. Chem. 97, 8820-8834.

lsraelachvili JN, DJ Mitchell and BW Ninham. 1976. J Chem. Soc. Faraday
Transactions 2, 72, 1525

Jauregi Paula and Julie Varley, 1999. Colloidal gas aphrons: Potential
applications in biotechnology, TIBTECH, 17, 389-394.

Kafarov, V.V.; Reutsli, V. A.; Zhuravleva, T. Yu. (1974), J. Appl. Chem USSR,
49(12), 2724.

Kirikou M and Sherman P, The influence of Tween 40/Span 80 ratio on the
viscoelastic properties of concentrated oil in water emulsions, J Coll. Interface.
Science, 71 (1) 51-54.

Ladbury, J. E. and Chowdhry, B. Z. 1996. Chemistry and Biology, 3, 10, 791-801
Lang JC and Morgan RD, 1980, Nonionic surfactant mixtures 1. Phase equilibria
in C1oE4-H20 and closed loop coexistence, J. Chem. Phys. 73(11) 5849-5861.
Lissant K.J., Peace 8W, Wu SH, Mayhan KG, 1974. Structure of high lntemal
phase ratio emulsions, J. Colloid and Interface Science, 47 (2) 416-423.

Lye G.J., and Stuckey D.C., 1998. Structure and stability of colloidal liquid
aphrons, Colloids and Surfaces A-physlcochemical and engineering aspects
131:(1-3)119-136.

Lye G.J., Stuckey DC. 1994. Extraction of macrolide antibiotics using Colloidal
Liquid Aphrons (CLAs), In; D.L. Pyle (ed.), Separations in Biotechnology. 3rd

edition. Royal society of Chemistry Cambridge UK. 280-286.

61

Lye, G.J., Poutainen, L. V., Stuckey D.C., 1994. Removal of dilute organics from
industrial wastewaters: extraction of 3-4 dichloroaniline using colloidal liquid
aphrons, In: Proceeding 2“‘1 environmental symposium on Environmental
Biotechnology, lChemE Publishers, Rugby, UK. 25-27

Mannheimer RJ, 1972. Anamalous rheological characterisitics of a High lntemal
phase ratio emulsion, J. Colloid and Interface Science, 40 (3), 370-382.
Matsushita K, A. H. Mollah, D.C. Stuckey, C. Del. Cerro and A.l. Bailey, 1992.
Predispersed solvent extraction of dilute products using colloidal gas aphrons
and colloidal liquid aphrons: aphron preparation, stability and size, Colloids Surf.
, 69, 65.

McAuliffe C, 1965. J. Phys. Chem., 70 (4), 1267-1275.

Mitchell DJ and Ninham BW, J Chem. Soc. Faraday Trans. 2, 77, 601, (1981).
Mitchell, D.J.; Tiddy, G.J.T.; Waring L., 1983, J. Chem Soc. Faraday Trans. 1 ,
79, 975.

Princen HM, 1980. J. Colloid and Interface science, 75, 246

Princen HM, 1988. Langmuir, 4, 486.

Rash, J.E. and CS. Hudson (1979) Freeze fracture: methods, artifacts and
interpretations. Raven Press, NY. 204

Reynolds O, 1886. Philosophical Transactions, 177, 157

Roffler, S. 1986. Extraction ferrnentation-lactic acid and acetone/butanol

production, Ph. D. Dissertation, University of California, Berkeley.

62

Schulman JH and EG Cockbain, 1940. Molecular interactions at oil/water
interfaces part 1. Molecular complex formation and the stability of oil in water
emulsions. Trans. Faraday. Soc. 36, 651-661

Schwartz LW, and Princen HM, 1987. J. Colloid. Interface Science. 118, 201
Sebba F. 1987 Foams and biliquid Foams-aphrons. John Wiley 8. sons, Inc.,
New York. ‘

Sebba F., J. Colloid and Interface Sci., 35 (4) (1971) 643.

Severs NJ. and 0M. Shotton (1995). Rapid freezing, freeze fracture and deep
etching. Wiley-Liss, NY. 372

Strey R, Schomaker R, Roux D, Nallet F, and Olsson U, 1990. Dilute lamellar
and L3 phases in the binary water-0128:, system. J.Chem.Soc. Faraday Trans,
86(12), 2253-2261 .

Tanford, CJ, “The Hydrophobic effect: Formation of micelles and biological
membranes”, Wiley NY 1980.

Tiddy, G.J.T., Phys. Rep., 57,1, (1980)

Toxicology Update Journal of Applied Toxicology 1995.15(6) 495-499.

63

Chapter 2

2. Enhancement of Mass Transfer Using Colloidal Liquid
Aphrons : Measurement of Mass Transfer Coefficients in
Liquid-Liquid Extraction

2.1 Introduction

Biocatalysis has enormous commercial potential. Enzymes or whole cells
can convert chemical precursors into a wide variety of product classes, including
pharmaceuticals (steroids, isoflavonoids, taxol), fragrance and flavor compounds
(lactones, pyrazines, essential oils), fuels (ethanol, butanol, biodiesel), as well as
a broad range of miscellaneous oils, fatty acids, and terpenoids. Development of
efficient biocatalytic processes involving nonpolar reactants or products having
low aqueous solubility is challenging (Mattiasson, et al. 1991). In such cases, a
multiphase approach can be effective, in which the continuous aqueous phase
contains the biocatalyst, and the dispersed phase serves as a reservoir of the
reactant or product. The dispersed phase may transfer a reactant to the
aqueous phase to replenish what is consumed by reaction. Alternatively, the
dispersed phase may extract an inhibitory or labile product from the aqueous
phase as it is formed (Larsson, et al. 1989). Such extractive fennentations have
been demonstrated for a variety of fermentation products (Roffler, et al. 1991;

Puziiss, etal. 1965; Soucaille, etal. 1987).

To date, the commercial potential of multiphase bioprocesses has not
been fully realized, because the underlying biological and engineering principles
have not been sufficiently developed to allow process development, design and
scale-up (Scott et al., 1995). Selection of an appropriate fluid to serve as the
dispersed phase is an important engineering consideration. Either nonpolar
liquids or a second aqueous phase may be used when the transferred solute is
nonvolatile (Roffler, etal. 1991; Puziss, etal. 1965), and gases may be used for
volatile or gaseous solutes (Bredwell, et al. 1999). Ideally, the fluid should be
biocompatible (i.e., not inhibit the biocatalyst), and it should have a high capacity
for the solute.

The tendency of a solute to partition into the dispersed phase is given by
its distribution coefficient (Kc), which is the ratio of the equilibrium solubility of the
solute in the dispersed phase to that in the continuous phase. The larger the Kc
value, the lower the volume of dispersed phase required. Unfortunately, solvents
with the highest distribution coefficients are typically not biocompatible (e.g.,
Roffler, 1986). Biocompatible solvents, on the other hand, tend to have low
distribution coefficients, 'and consequently, low mass-transfer driving forces
(Roffler, 1986). The relative position of a solvent in the blocompatibility/solute
capacity continuum has been correlated to its distribution coefficient between
octanol and water (Kow) (Bruce et al. 1991).

The inherent tradeoff between blocompatibility and solute capacity results
in interphase mass transfer often being the rate-limiting step In multiphase

biocatalysis. Kollerup and Daugulis (1985) have developed a mathematical

65

model to estimate the effect of key process variables on the productivity of an
extractive fermentation process to produce ethanol. A key parameter in the
model is the effective distribution coefficient (Kc’) which is the product of the
distribution coefficient and the fractional degree of equilibrium mass transfer

(1]) between the two phases. The value of n was reported to be on the order of
0.2 for a typical set of operating conditions (Kollerup, et al. 1985), indicating only
a 20% approach to equilibrium. Increasing the interphase mass-transfer rate
would allow a closer approach to equilibrium. In this case a five-fold increase in
extraction efficiency is theoretically possible.

The volumetric, interphase mass-transfer rate (RA) is given by the product
of overall, interphase mass-transfer coefficient (KL), the interfacial area per unit
volume (a) and the concentration driving force (C-C“) where C* is the equilibrium
solubility of the solute being transferred. Although KL varies somewhat with the
environmental conditions, its range is relatively small. The concentration driving
force is also typically small, due to the inherently low aqueous solubilities of
nonpolar solutes. However, the value of a can be increased by orders of
magnitude by decreasing the size of the dispersed droplets or bubbles.
Mechanical agitation (e.g. a mixer settler) is commonly used in mass-transfer-
Iimited systems for this purpose. However, this approach is energy-intensive and
thus costly, particularly in large-scale systems (Bredwell, etal. 1998). Another
approach is to use surfactants to predisperse the nonpolar phase prior to its
addition to the aqueous phase. This approach can give droplet sizes on the

order of microns or smaller (Bredwell, et al. 1998; Lye and Stuckey, 1996).

66

The utility of CLA in multiphase extraction processes has been demonstrated at
the bench scale (Lye and Stuckey, 1996, Rosjidi etal. 1994, Wallis, et al. 1985,
Scarpello and Stuckey, 1996). Sebba proposed that CLA are coated with a
surfactant-stabilized “shell” that contains entrapped water, as shown
schematically in Figure 2.1. Formation of this shell has been reported to require
two contrary surfactants (i.e., one surfactant that partitions primarily into the
aqueous phase, and another that partitions primarily into the nonpolar phase).
Freeze-fracture transmission electron microscopy and other techniques to
provide evidence for a multilamellar shell structure (See Chapter 1). The results
suggested that the two contrary emulsifiers form liquid-crystal-like bilayers that
stabilize the droplets. This ordering has been shown to be a unique property of
the surfactant system used to form the aphrons.

This shell is important from an engineering standpoint, because it may
impart significant resistance to mass transfer that could significantly reduce KL.
However, no fundamental studies of CLA mass transfer have been published to
date. Such information would be needed to optimize and scale up multiphase
biocatalytic processes involving CLA (Scarpello et al., 1999). This chapter
presents KL and a data for transfer of heptanoic acid from water into limonene
CLA. These data are combined with a literature correlation (Skelland and Lee,
1981) for the mass transfer coefficient of the continuous phase (Kmominuous) to

estimate a mass transfer coefficient for the shell (KLshen). The effect of the

67

 

 

 

. 1 Outer
Electncal surface
double 3. of shell
layer .13. A‘.\‘

(diffuse) , ’3. x
X [I .l‘ j. \\‘ ‘0
4 j;
it: ,. '=.
i 1:»: 4
3 .. \l» i'
“X‘ y 3”; 4
Bulk water I
containing "Tr
surfactant 134 >3 S? :0:
- “~~~.3f_al;liul-l-l * " ° 3 e
.. _______________

 

 

Figure 2.1. Proposed structure for a single colloidal liquid aphron (CLA) when dispersed
in a continuous aqueous phase (Redrawn from Sebba ( 1984)).

68

concentration of the two surfactants on the mass-transfer properties is also

presented.

2.2 Materials and Methods

2.2. 1 Mass-Transfer System

2.2.1.1 Surfactants.

Sebba (1984) reported that two surfactants are required to form CLA--one
that partitions primarily into the aqueous phase and another that partitions
primarily into the nonpolar phase. Cationic and anionic surfactants form
relatively strong ionic bonds with proteins that can cause denaturation. Nonionic
surfactants, on the other hand, bind through much weaker hydrophobic
interactions and are thus less likely to inactivate enzymes or denature proteins
(Schwuber et al., 1980). Bredwell et al. (1997) showed that nonionic Tween
surfactants were non-inhibitory in batch ferrnentations, whereas many ionic
surfactants were toxic to the cells. Consequently, two nonionic surfactants were
chosen for the present study. The aqueous-phase surfactant used in this study
was polyoxyethylene sorbitan monooleate (Tween 80, HLB 15.0, Aldrich
Chemical Company, Milwaukee, WI ). The nonpolar-phase surfactant was a
linear alcohol ethoxylate (Tergitol 15-S-3, HLB 8.3, Sigma Chemical Company,
St. Louis, MO). The interfacial structure of the shell, which results from this
combination of surfactants, has been described in the previous chapter.
2.2.1.2 Continuous and Dispersed Phases.

Deionized water (resistivity > IBMQ-cm) drawn from a Milli-Q Plus'E water

Purification system (Millipore Corporation, Bedford, MA) was used as the

69

continuous phase. Limonene (CAS 138-86-3, Sigma Chemical Company, 97%)
was used as the dispersed phase. Limonene is a naturally occurring terpenoid
found in etherial oils derived from lemons, oranges, grapefruit, and caraway. It is
an important by-product of the citrus-processing industry and is currently used as
a solvent in waterless hand cleaners and degreasing agents. It has extremely low
solubility in water (13 mg/L at 25°C) and a log Kow of 4.232 (Toxicology Update,
1995).
2.2.1.3 Transferred Solute

To ensure that the controlling mass-transfer resistance is in the
continuous phase, it is necessary that the solute’s Kc value be large in favor of
the dispersed phase. Heptanoic acid is highly soluble in nonpolar solvents, but its
aqueous solubility is only 2.5 g/L at 25°C (Dean, 1992). As a result, it has a high
KC value for the limonene/water system. Also, because heptanoic acid ionizes in
water, its aqueous-phase concentration can be readily measured on-line using a
conductivity probe. Heptanoic acid has been used to study interphase mass-

transfer in other multiphase systems (Skelland et al. 1981).

2.2.2 Measurement of Hepta noic Acid Concentration

The concentration of heptanoic acid in the aqueous phase was measured
by electrical conductivity, as recommended by Rushton et al. (1964). A YSI 3417
(K=1.0/cm) conductivity probe (YSI Inc, Yellow Springs, OH), V2 in OD, 5% in
overall length used with a YSI Model 32 conductance meter at a range switch
setting of 200 ms“. The probe support ring was grounded to the (~) recorder jack

to minimize electrical noise in the signal. The output of the conductance meter

70

was recorded by a CAMILE 2000® data acquisition system (Dow Chemical '
Company, Midland MI) at ts intervals. The time constant of this probe (<1 msec)
was much smaller than that of the interphase mass transfer process, so the
mass-transfer dynamics could be measured accurately during the batch runs.

The apparatus was calibrated by measuring the conductivity of standard
heptanoic acid solutions in the absence of a dispersed phase. Three sets of
calibration solutions were independently prepared and measured. The resulting
calibration curve (Figure 2.2) was linear (R2: 0.9965) with respect to
concentration. The conductivity of pure water was measured both in the
presence and absence of limonene. At the low limonene volume fractions used,
limonene had no effect on the conductivity of the aqueous phase (<1% change in
conductivity).

The diffusivity value for heptanoic acid in water (6.8 x 10‘6 cm/s) was
calculated using the Wilke-Chang correlation (Wilke et al., 1955). The molal
volume of heptanoic acid at its normal boiling point was obtained using Le Bas

group contributions method (Reid et al., 1958).

2.2.3 CLA Preparation

The CLA-preparation conditions have been shown to influence the size
distribution and stability of the CLA phase (Matsushita et al., 1992).
Consequently, the CLA were prepared under consistent conditions in all runs.
The CLA emulsions were formed in a 50 mL glass beaker having an lntemal

diameter of 40 mm. Five mL of aqueous Tween 80 solution was added to the

71

 

 

0.014 /
0.013
0.012
0.011 /

0.01 /
0.009
0 008 / y = 0.0132x - 0.0064

’ R2 = 0.9965
0.007 «
0.006 , , . $ ,

0.9 1 1.1 1.2 1.3 1.4 1.5 1,5

Reading (V)

 

 

Concentration (mol/L)

 

 

 

Figure 2.2 Calibration curve for heptanoic acid

72

beaker. The concentration of the Tween 80 solution was adjusted to achieve the
desired final concentration (0.001% v/v to 0.4% v/v Tween 80). Five mL of
limonene/Tergitol 15-S-3 solution (0.01% v/v to 0.1% v/v Tergitol 15-S-3) was
then dispensed into the aqueous solution at a rate of 0.5 mlJmin while mixing at
1,600 rpm using a 25 mm magnetic stirrer bar. The resulting CLA emulsions
were creamy white, with a phase volume ratio (PVR) of 1.0. The PVR is defined
as the ratio of the volume of the nonpolar phase to that of the aqueous phase.
The CLA emulsions were very stable, with no phase separation evident over a

period of 6 months.

2.2.4 CLA Size Measurement

Particle-size distributions of CLA emulsions were measured using a
Malvem Mastersizer X particle analyzer (Malvem Instruments, Worcestershire,
UK). The instrument was fitted with a 15-mL sample chamber containing
deionized water. The contents of this cell were mixed using a 10 mm by 8 mm
Nalgene Star Head magnetic stir bar (Fisher Scientific, Pittsburgh, PA) to ensure
that the CLA remained dispersed. The size measurement was made within 10 s

of the CLA being added to the cell. The average particle size was reported in
terms of the Sauter mean diameter (ii—,2) which is defined in Eqn. 1. The

precision of the particle-size measurements is i5%.

Enid? l
Enid? ( )

 

532 =

73

2.2.5 Mass Transfer Measurement

2.2.5.1 Stirred-Cell Experiments.

In stirred-cell experiments, the two phases are not dispersed, but rather
overlaid on one another. This approach has been widely used to measure KL
values because the interfacial area is accurately known (Nanda et al., 1966;
Nanda etal., 1967, Femandes et al., 1967; Bhave etal., 1981). The stirred cells
were glass beakers having lntemal diameters of either 8.3 cm or 10.5 cm. Five
hundred mL of 0.0141 M aqueous heptanoic acid solution was overlaid with 50
mL of pure limonene. A Rushton impeller located just below the limonene-water
interface was used to gently mix the water phase without disrupting the
water/limonene interface. The stainless-steel impeller had a diameter of 5 cm, a
blade height of 1 cm, and blade width of 1.2 cm. The impeller was driven using a
Lightnin® Labmaster Il unit (Mixing Equipment Company, Rochester, NY). A
wooden shaft was used to minimize the electrical noise in conductivity
measurements. The impeller rate (60 rpm) allowed the interface to be
continuously renewed without creating a significant vortex. The conductivity of
the aqueous phase was recorded as a function of time until no further decrease
was detected. The equilibrium concentration was then measured after 24 h.
2.2.5.2 Stirred-tank experiments

The mass-transfer properties of the CLA were measured in a baffled,
stirred tank having standard dimensions (McCabe et al., 1993). The Plexiglass
vessel had an ID of 15 cm and a height of 20 cm. It was fitted with four, equally

spaced, vertical, wall baffles. The liquid height in the vessel was maintained at 15

74

cm. The Rushton impeller described above was axially centered in the tank 5 cm
from the bottom. The conductivity probe was mounted close to the impeller
blades. The continuous phase (2.8 L of 0.0141 M aqueous heptanoic acid) was
placed in the tank, and mixing was initiated at the target rpm. Then the second
phase was poured rapidly (within 1 3) near the center of the vessel to promote
rapid dispersion. For CLA experiments, 10 mL of CLA dispersion having a PVR
of 1.0 was added. In one control experiment conducted without any surfactants
(Run 10), pure limonene was used as the dispersed phase. After the two phases
were combined, the conductivity of the aqueous phase was recorded as a
function of time until no further decrease was observed. A final reading was
taken 30 min later to serve as the equilibrium value. All experiments were
conducted at 25°C.

Table 2.1 summarizes the experimental conditions used for all runs.

75

 

Table 2.1. Summary of conditions for mass-transfer experiments.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Run Vessel Stirring Aqueous-phase Nonpolar-phase

No. rate surfactant conc. (% surfactant conc. (%
(rpm) v/v) WV)

1 Stirred Cell' 60 - -

2 Stirred Cell” 60 - -

3 Baffled tank 200 0.4 0.1

4 Baffled tank 300 0.4 0.1

5 Baffled tank 375 0.4 0.1

6 Baffled tank 427 0.4 0.1

7 Baffled tank 500 0.4 0.1

8 Baffled tank 550 0.4 0.1

9 Baffled tank 600 0.4 0.1

10 Baffled tank 427 - -

11 Baffled tank 427 0.02 0.01

12 Baffled tank 427 0.04 0.01

13 Baffled tank 427 0.4 0.01

14 Baffled tank 427 0.04 0.1

 

 

 

 

 

1 Stirred cell interfacial area = 56.72 cm"2 +Stirred cell interfacial area = 83.32 cm2

76

 

 

2.3 Data Analysis

The unsteady-state mass balance on heptanoic acid in the continuous phase is

 

given below.
Kai—1?: KLaV(C-C) (2:
Equation 2 was integrated from the initial time (to) and concentration (Co) to
obtain Eqn. 3.
(C -C) v
1 =- a— (t-to) (3)
“LC mi [KL ti

The experimental data were analyzed by plotting the left-hand side of Eqn. 3 as a
function of time. The value of KLa was then obtained from the slope of the plot

using the relation

 

VC
KLa = - slope[ V ]

Example plots showing the degree to which the data followed the expected linear
trend are given in Figures 2.3 (a), (b), and (c) for each of the three types of mass-

transfer experiments performed.
The value of a was computed from the a data as shown below:

i
(132

a=6

 

where 0 is the void fraction of the dispersed phase. The void fraction 0 was

77

L0910(C*'C)/(C*'Co)

 

 

 

 

 

 

 

 

 

< u 828.. - 8.88
m. n 988

 

 

 

 

l

 

 

 

 

 

 

 

 

 

 

0 N50 000

.moo .moo mooo NAoo mmoo wmoo 0000

....Bo Am.

30:3 Nam. .630303 0.. 289.3222 :20 2.3 3m .58: .32. 88222. 3. man. w. .232. no: 2 mo 83.

78

 

 

 

 

< ... -obommx - 0.83
m” n 988

  

 

 

 

 

 

 

 

L0910(C*'C)/(C*'Co)
c'>
on

 

 

 

u.— [i ‘ \*,

 

 

 

....— .N _ d fl _ _ A fl d ~ A _ fi _
0 mm mo um 30 4mm Amo 3m moo mmm mmo mum woo wmm wmo

.235 Amy

Inca Nu? >c8m3m2 2 969.3ng 38 2:: :6 .53163 98.23 3 man. w.
3:: 8 .3 33mm Sax 2 ANN 63

79

L0910(C*'C)/(C*'co)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. < n -obwwwx - obmow
N '-
-oL -IRI » m u 0.33
-o.m ,
-o.m i
...— Illir‘ \i ‘ I
-4 .m . a d A fi fi a
o m ‘0 Am mo mm mo mm

4.3m A3

macaw moo. >c3m3maoflmxnm13m3a 88 2:: 5m :38: "Ban 93.93 3 mg? m. 3:: m 5 330a 83x
m. mmo 63. 9%». ES ...Emm: mo 5 3.505 25.8 mag 91$ 35 406.8. Am-m-m 5 Or) 35$.

80

equal to 0.0018. At such a small void fraction, collisions between droplets, and
subsequent coalescence, can be ignored during the short duration of the
experiments (Femandes et al., 1967). Hence, the interfacial area was assumed
to be constant throughout the experiments. In the stirred-cell experiments, a was
calculated as the interfacial area between the aqueous and limonene layers
divided by the volume of the aqueous phase.

2.4 Results and Discussion

2. 4. 1 Measurements In Stirred Cell

Runs 1 and 2 were conducted in stirred cells of known interfacial area
(56.7 cm2 and 83.3 cm2, respectively). The resulting KL values (0.0012 and
0.0011 cm/s, respectively) were well within the range of values reported for
similar systems (Shanna et al., 1968), indicating the suitability of the
Iimonene/water/heptanoic-acid system for measuring interphase mass transfer.
Even though the interfacial area increased by almost 50% from Run 1 to Run 2,

the KL values for the two runs were almost identical, as expected

2.4.2 Measurements in Baffled Tank

Runs 3 through 9 were conducted using CLA containing 0.1% Tergitol 15-
8-3 in the nonpolar phase and 0.4% Tween 80 in the aqueous phase. The
interfacial area was identical in all runs. As shown in Figure 2.4, KLa increased
with Nag up to about 15,000 and then became essentially independent of NR5.
The effect of Nag on KLa is through the KL term, rather than a. Since the droplets

in the CLA emulsions were much smaller than the maximum stable drop size

81

under the hydrodynamic conditions used, breakage should have been negligible
in these experiments.

The KL values obtained from experimental data are shown as a function of
Nae in Figure 2.5. These values reflect the overall resistance to mass transfer of
heptanoic acid from the dispersed phase to the bulk continuous phase. Due to
the high solubility of heptanoic acid in the nonpolar phase, the resistance to mass
transfer in the non-polar phase should be negligible, and hence, there are two
primary resistances to mass transfer: 1) the resistance of the continuous phase
and 2) the resistance of the surfactant shell surrounding the aphrons. The
resistance of the continuous phase would be expected to decrease with the
degree of mixing, while the shell resistance should be relatively unaffected by

mixing.

82

 

 

 

 

 

 

 

330 .......
o
280 /’ . '
r 230
a o
u
..l
3‘ 180
130 /
80 i . .
0 5000 10000 15000 20000 25000 30000
"as

Figure 2.4. Variation in experimental volumetric mass transfer coefficient (KLa) with NR5
measured in a baffled tank.

83

Skelland et al. (1981) developed the following correlation for KLCOmmUOUS values in

stirred tanks having standard dimensions:

K tinuous d
(231/2 =1.86410—7 ¢—0.287(_]i‘)0.548(NRE)1.37l(NWE)—0.()95 (4)

Although this correlation was developed for mass-transfer systems without
surfactants, it should reasonably predict KLcommuous values for systems containing
surfactants, provided the effect of the surfactant on the surface tension is taken
into account through the NW£3 term. Additional mass-transfer resistance due to
surfactant accumulation at the interface can be accounted for through a separate
mass-transfer coefficient for the shell (KL,she..).

Figure 2.5 shows the KLconfinuous values predicted by this correlation along
with the experimentally measured (overall) KL values. At low NR5 values, the
KLcommuous values agree well with the KL values, both in magnitude and the effect
of impeller rate. However, above a Nae of 15,000 (375 rpm), the KL values
approach an asymptote, while the KLWWGUS values continue to increase steadily
with Nag. These trends are consistent with the hypothesis that the continuous
phase resistance is rate-limiting at low Nag, but the shell resistance becomes
rate-limiting at high Nae. Based on this hypothesis, the resistances-in-series
model shown in Eqn. 5 was used to estimate KLshe" from the KL and KL, commuous

values.

1 _1 1
Klghell KL K14:

 

(5.

ontinuous

The results are summarized in Table 2.2.

84

A control experiment (Run 10) was conducted to determine how well the
Skelland et al., (1981) correlation (Eqn. 4) applied to the
limonene/water/heptanoic acid system in the absence of surfactants. This run
was conducted at a Np; of 17,800. Because the limonene was not predispersed
as CLA, the droplet size distribution represented a dynamic equilibrium between

droplet coalescence and disintegration. Under these conditions, the following

correlation for 3; published by Skelland et al. (1981) applies:

8 |w°'|

2 = 6.7 1&1 0-4¢-l.034(NRE)-0.558(N0H)-l.025 (6)

I

Equations 4 and 6 were used to calculate a predicted KLa value for Run 10 of 29

i 5 h'1 and a d_32 value of 241 i 48 um. This KLa value agreed well with the

experimentally measured value of 23 h". However, the predicted 21: value

could not be experimentally verified; in the absence of surfactants, the droplets

coalesce too fast for the Malvem Mastersizer to be used.

85

 

(0
01

[0'00
0010

 

KL (cm/s) x103
5
l

 

 

.3
O
I

01

 

 

 

0 i l l T

0 5000 10000 15000 20000 25000 30000
"RE

Figure 2.5. Effect of NR5 on experimental KL values (0) measured in a baffled, stirred
tank and predicted KLconu-nuous values (El) calculated using Eqn. 4 (Skelland er al., 1981).

 

86

 

Table 2.2. Effect of N RE on KL, KLcommuous, and KLshc“.

 

 

 

 

 

 

 

 

Run No. NpE KL (cm/s) x 103 KLcominuous (cm/s) x 103 KLShe" (cm/s) x 103
3 8333 4.5 5.1 40.3
4 12500 7.7 10.0 32.2
5 15625 10.3 14.6 35.3
6 17792 12.1 18.2 36.5
7 20833 14.0 23.7 34.1
8 22917 12.9 27.8 24.2
9 25000 13.4 32.2 23.1

 

 

 

 

 

87

 

2.4.3 Influence of Surfactant Concentration.
The influence of the concentrations of the two surfactants on the Sauter mean
diameter of CLA droplets (d32) is shown in Figure 2.6. While the 21; values

were essentially unaffected by the Tergitol concentration, they decreased as
much as 70% as the Tween 80 concentration increased from 0.02 to 0.4 g/L.
This value is slightly higher than Critical Micelle Concentration (CMC) of 0.013
g/L for Tween 80. The CMC value is defined as the concentration at which the
surfactant concentration forms a saturated monomolecular layer at the interface
(Porter, 1994). Stable CLA emulsions could not be formed below a Tween 80
concentration of 0.015 g/L.

Table 2.3 shows the effect of surfactant concentration on the

experimentally measured values of 2;, a, KL, and KLa for the limonene CLA

(Runs 6, 11, 12, 13, and 14). The d32 , a, and KL values predicted by Eqns 4 and

6 and the measured KLa value for the control run without surfactants (Run 10)
are also listed for comparison. The KLa values for CLA increased by a factor of
about three as the Tween 80 concentration was increased from 0.02 to 0.4% v/v
at a constant Tergitol-S concentration of 0.01% v/v. This effect was due almost
exclusively to an increase in a, which increased by a similar factor. The value of
KL remained essentially constant for all the CLA runs. The average KL value of

the CLA runs was 14.3 x 10’3 cm/s, and the standard deviation was 2.2 x 10'3

88

 

 

 

 

 

 

 

 

 

 

 

 

60 ‘ l
l .
50 T l l
l l
40 . i i %
s. l s
T: (I) \ I i
‘5” \
Z) i ;
N
10 '
0 i L

o 0.1 0.2 0.3 0.4 0.5

Tweenmoonc. °/e(\llv)inaq1eol.lsfl’ase

Figure 2.6 Variation of CLA droplet diameter with Tween 80 concentration in the
aqueous phase at Tergitol lS-S-3 concentrations of 0.1% v/v (El) and 0.01% WV (9 ).

89

Table 2.3. Effect of surfactant concentrations on KL and KLa

 

 

 

 

 

 

 

 

 

 

 

 

 

Run Tween 80 Tergitol (132011“) a (cmZ/cmg)+ KLoveran KLa (h‘1)
(%) (ii/0) (cm/s) x 103

1o - - 241” 0.44 14.5 23

11 0.02 0.01 49 2.19 14.9 117

12 0.04 0.01 40 2.65 17.9 170

13 0.4 0.01 14 7.07 13.1 334

14 0.04 0.1 22 4.84 13.8 240

6 0.4 0.1 17 6.41 12.1 280

 

 

“ Value calculated using Eqn. 6.

 

* Interfacial areas calculated from measurements of diameter using the equation

a=6¢ld32

90

the KL values were essentially independent of impeller rate (see Figure 2.5). In
this regime, the shell resistance is believed to be significant. Thus, the fact that
KL was largely unaffected by surfactant concentration in these experiments
suggests that the surfactant concentration does not significantly affect the
properties of the shell (e.g., shell thickness) under these conditions. Moreover,
since the measured KL values are similar in magnitude to those predicted by
Eqn. 4 for the control experiment, the mass-transfer resistance contributed by the
shell does not appear to be excessive. Consequently, at an equivalent
dispersed-phase void fraction, the 7- to 16-fold higher a values provided by the
CLA translate into 5- to 14-fold higher KLa values.

This mass-transfer advantage provided by CLA would be expected to be even

more significant at lower NR5 values. The value of a is inversely proportional to
Z- Thus, according to Eqn. 6, a should be proportional to Net!"558 for

conventional emulsions. The size distribution of the CLA (and hence a), on the
other hand, is determined exclusively by the conditions under which they are
formed. As a result, decreasing Nag would decrease a for conventional
emulsions but have little effect on a for CLA.

Because CLA emulsions can provide several-fold higher interfacial areas
and volumetric mass-transfer rates than conventional emulsions, they offer
significant advantages for mass-transfer—limited applications of multiphase
biocatalysis. However, a major impediment to the commercial application of CLA

(and other surfactant-stabilized emulsions) is the difficulty in coalescing the

91

dispersed phase and separating the two phases once the contacting step is
complete. It is possible to overcome this difficulty by stabilizing the emulsions
with novel polymeric emulsifiers whose emulsification properties are pH sensitive
(Mathur, et al., 1998). The emulsions produced with these emulsifiers can be
coalesced on demand by making a small change in the pH. These emulsions
shall be discussed further in Chapter 6.
2.5 Conclusions

The mass-transfer properties of CLA emulsions of limonene in water have
been measured experimentally. The CLA size distribution was found to be
strongly affected by the aqueous-phase surfactant (Tween 80) concentration, but

virtually unaffected by the nonpolar-phase surfactant (T ergitol-1 5-S-3)
concentration. The d; values decreased from 49 to 14 pm as the Tween 80

concentration increased from 0.02 to 0.4% v/v. The KL values measured for CLA
emulsions averaged about 15 x 10'3 cm/s and were essentially unaffected by
surfactant concentration. The KL,she.. value was estimated to be about 30 x 10’3
cm/s, which is comparable to KL values for surfactant-free, liquid-liquid systems
at an equivalent 4) value. The KLa values for CLA were 5- to 15-times higher than
those for conventional emulsions. These results suggest that CLA may offer
significant advantages over conventional emulsions for mass-transfer-Iimited,
multiphase biocatalytic processes.

2.6 Nomenclature

C = Concentration of solute (mol L")

KL = Overall mass-transfer coefficient, (cm s")

92

KLcominuous = Mass-transfer coefficient of the continuous phase (cm s")

KL,she.. = Mass-transfer coefficient of the CLA shell (cm s")

D = Diffusivity, (cm2/s)

a = interfacial area per unit volume in two phase mixture (m2/m3), calculated as

21
C132

.d?
at32 = Sauter mean diameter defined by in, ' (m)

Enid? ’

 

d. = diameter of the impeller, (m)
N = impeller speed, (rps)
K = cell constant for conductivity probe (cm")

T = tank diameter. (m)

NOH = Ohnesorge number, defined by _Ec__
1/pcd,o
2
N95 = Reynolds number, defined by fl
dei‘p.

ng = Weber Number, defined by

Vc = Volume of the continuous phase, (L)

V = Filled volume of the tank, given by (1+¢ )VC, (L)
Greek

42 = void fraction of dispersed phase, dimensionless
[J = viscosity, (Ns/mz)

p = density, (kg/m3)

93

o = interfacial tension, (N/m)

2.7 References

Bhave RR. and Shanna MM. 1981. Liquid-liquid reactors: operation under
refluxing conditions: measurement of effective interfacial area, Trans ICHEME,
59 161-168.

Bredwell M.D., Worden RM. 1998 Mass-transfer properties of microbubbles. 1.
Experimental studies, Biotech. Prog. 14: (1) 31-38.

Bredwell, M.D., Telgenhoff M.D., Barnard S., and R. M. Worden 1997. Effect of
surfactants on Carbon monoxide fennentations by Butyribacterium
methylotrophicum, Applied Biochemistry and Biotechnology, 63-5: 637-647
Bredwell, M.D., Srivastava, P., and R. Mark Worden 1999. Reactor Design
Issues in Synthesis Gas Ferrnentations, Biotechnology Prog., 15(5) 834-844.
Bruce, L.J and Daugulis, A.J. 1991. Solvent Selection Strategies for Extractive
Biocatalysis, Biotechnol. Prog. 7, 116-124.

Dean J.A. 1992. Lange’s Handbook of Chemistry, Fourteenth edition, McGraw
Hill, Inc. New York, NY.

Femandes J.B., Shanna MM. 1967. Effective interfacial area in agitated liquid
liquid contactors, Chem. Engg.Sci. 22, 1267-1282.

Kollerup, F. and Daugulis, A.J. 1985. A mathematical model for ethanol
production by extractive fermentaion in a continuous stirred tank ferrnentor,

Biotechnol. Bioeng. 27: 1335-1346.

94

Lye G.J., Pavlou O.P., Rosjidi M., Stuckey D.C. 1996. Immobilization of Candida
cylindracea lipase on colloidal liquid aphrons (CLAs) and development of a
continuous CLA-membrane reactor, Biotechnol. Bioeng. 51: (1) 69-78.

Lye G.J. and Stuckey D.C. 1996. Predispersed solvent extraction of microlide
antibiotics using colloidal liquid aphrons, in Proceedings lntemational Solvent
Extraction Conference, Melboume, Australia, ed. by Shallcross DC, Paimin R
and Prvcic LM, 1399-1405.

Mathur, A.M., B. Drescher, A.B. Scranton, and J. Klier, Polymeric Emulsifiers
Based on Reversible Formation of Hydrophobic Units. 1998. Nature, 392, 567-
370.

Mattiasson B. and Olle Holst 1991. Objectives for extractive Bioconversion, in
Extractive Bio Conversions, ed. Bo Mattiasson and Olle Holst, Marcel Dekker,
New York NY.

Larsson, M., Arasaratnam, V. and Mattiasson, B. 1989. Integration of
bioconversion and down-stream processing. Starch hydrolysis in an aqueous
two-phase system, Biotechnol. Bioeng. 33:758-766.

Matsushita K, Mollah A.H., Stuckey D.C., del Cerro C and Bailey AL 1992.
Predispersed solvent extraction of dilute products using colloidal gas aphrons
and colloidal liquid aphrons — aphron preparation, stability and size. Colloids
Surfaces, 69, 65-72.

McCabe W.L., Smith J.C. and Harriot P. 1993. Unit Operations of Chemical

Engineering, Fifth Edition McGraw Hill Inc, New York NY.

95

Nanda A.K.and Sharrna M.M.1966. Effective interfacial area in liquid liquid
extraction, Chem. Engg. Sci. 21 707-714.

Nanda A.K., Sharma MM. 1967, Kinetics of fast alkaline hydrolysis of esters,
Chem. Engg. Sci. 22, 769-775.

Porter M. R. 1994, Handbook of surfactants, 2nd ed. Chapman and Hall, London
UK.

Puziss, M. and Heden 1965. C.G., Toxin production by Clostridium tetani in
biphasic cultures, Biotechnol. Bioeng. 7: 355-366.

R Reid, and T. Sherwood 1958. Properties of Gases and Liquids: their
estimation and correlation, 2"d ed., McGraw-Hill, New York, NY

Roffler, S. 1986. Extraction fermentation-lactic acid and acetone/butanol
production, Ph. D. Dissertation, University of Califomia, Berkeley.

Roffler S.R., Theodore W Randolph, Douglas A Miller, Harvey W. Blanch and
John M Prausnitz 1991. Extractive Bio-conversions with non-aqueous solvents in
Extractive Bioconversions ed. by Bo Mattiason and Olle Holst, Marcel Dekker,
Inc. New York, NY.

Rosjidi M., Stuckey D.C. and Leak DJ. 1994. Downstream Separation of chiral
epoxides using colloidal liquid aphrons( CLA), in Separations for Biotechnology,
Ed by Pyle D.L. , SCI publishing. 280-286.

Rushton J.H., Nagata S., Rooney TR 1964. Measurements of mass transfer
coefficients in Liquid-Liquid mixing, AICHE J. 10 (3), 298-302.

Save S.V., Pangarkar V.G. and Kumar S.V. 1994. Liquid-Liquid extraction using

aphrons, Separation Technology, 4: 104-111.

96

Schwuber, M.J. and Bartnik, F. G. 1980, in Anionic surfactants: Biochemistry,
Toxicology, Dermatology, Ed. by Christian Gloxhuber, Marcel Dekker, New York.
1-49

Scott, C.D., T.C. Scott, H.W. Blanch, A.M. Klibanov, and A.J. Russell 1995.
Bioprocessing in Nonaqueous Media: Critical Needs and Opportunities.
ORNL/TM-12849, US. National Technical lnforrnation Service, Springfield, VA.
Sebba F. 1984. Preparation and properties of polyaphrons (biliquid foams),
Chemistry and Industry, 367-372.

Scarpello J.T. and Stuckey D.C. 1996. Extraction of phenylalanine using colloidal
liquid aphrons (CLA): stability in dispersion and rates of mass transfer, in
Proceedings of lntemational Congress on Membranes and Membrane
Processes, Yokohama, Japan. 988-989

Scarpello J.T. and Stuckey D.C. 1999. The influence of system parameters in the
stability of Colloidal liquid aphrons, J. Chem Technol. Biotechnol. 74: 409-416.
Sharma MM, and A. K. Nanda 1968. Extraction with second-order reaction,
Trans Inst. Chem. Engrs., 46, 44-52.

Skelland A.H.P., Jai Moon Lee 1981. Drop size and continuous phase mass
transfer in agitated vessels, AICHE Journal 27 (1) 99-111.

Soucaille, P., Joliff, G., lzard, A, and Goma G. 1987. Butanol tolerance and
autobacteriocin production by Clostridium acetobutylicum, Curr. Microbiol. 14:
295-299

Toxicology Update Journal of Applied Toxicology 1995.15(6) 495-499.

97

Wallis D.A., Michelson D.L., Sebba F., Carpenter J.K. and Houle D. 1985.

Applications of aphron technology to biological separations. Biotech. Bioeng.

Symp. 15 : 399-408
Wilke, C.R., and P. Chang 1955. Correlation of Diffusion coefficients in Dilute

solution, AICHE J., 1, 264.

98

Chapter 3

3. Enhancement of Mass Transfer using Colloidal Liquid
Aphrons: Measurement of Mass-Transfer Coefficients

Using Chemical Reaction

3.1 Introduction

Many industrially important chemical, biochemical, and environmental
processes involve a phase-contacting operation intended to transfer a solute
between two immiscible liquids. The rate of solute mass transfer is often a rate-
Iimiting factor that limits the overall efficiency of the process. For example, the
nitration of aromatic substances like benzene, toluene, phenol, and napthalene is
practiced on large scales for the manufacture of intermediates for dyestuffs,
polymers, pharmaceuticals and explosives. Most of these nitrations involve two
phases: an organic phase and an aqueous, mixed acid (HN03 + H2804) phase.
Albright and Hanson (1975) determined that under industrial conditions the rate
of nitration is typically controlled by diffusional factors. Similarly, the rates of
alkaline hydrolysis of several forrnate esters and the rates of reduction of
substituted nitroaromatics in mechanically agitated, liquid-liquid contactors are
dependent on interfacial area per unit volume (a), indicating that interphase mass
transfer is rate-limiting.

Under mass-transfer-limiting conditions, the volumetric solute-transfer rate

can be increased by decreasing the size of the dispersed-phase droplets, and

99

 

thereby increasing a (Kafarov at al., 1974; Bhave and Sharma, 1981). CLA offer
an attractive means to enhancing mass transfer in such processes.

Rational design of contacting processes involving CLA requires the overall
interphase mass-transfer coefficient (KL) and a values be predicted (Femandes
and Sharma, 1967). The value of KL will depend on the continuous phase mass-
transfer coefficient (KLmnunuous) and the shell mass-transfer coefficient (KLWH). In
this chapter, the rate of mass transfer of an ester (p-tonl acetate) from CLA into
aqueous solution was measured in a stirred tank having standard dimensions.
The data were used to determine both KL and a for a variety of impeller rates and
surfactant concentrations. These results were then used, along with a literature
correlation for KL,confinuous and a resistances-in-series model, to estimate the
Kama value.

3.2 Theory

3.2. 1 Mass-transfer resistan ces in series

The overall resistance to mass transfer between the dispersed and
continuous phases (UK) is the sum of the individual resistances. In this study,
the CLA droplets contained pure p-tolyl acetate, which is sparingly soluble in the
continuous, aqueous phase. Under these conditions, the main resistances are
(1) the resistance of the continuous phase, and (2) the resistance of the shell.
Assuming the thickness of the shell is small compared to the diameter of the
aphron, a resistances-in-series model may be written:

1 1 1
= + (Equation 1)
I(I..shell

 

L.continuous

100

3.2.2 Coupling of mass tran sfer and reaction

Acetate esters (CHacOOR) are readily hydrolyzed in aqueous solutions by

sodium hydroxide (NaOH) as shown below:

CH3COOR + NaOH —9 CH3000Na + ROH (Reaction 1)
If the R group is aromatic, the acidic phenol group (ROH) formed in Reaction 1
reacts instantaneously with a second molecule of NaOH to give the
corresponding phenoxide:

FiOH + NaOH ——> RONa + H20 (Reaction 2)
Because the solubility of the ester in the alkali solution is very low, the ester and
NaOH must exist in two different phases, and the ester must be transferred into
the aqueous phase for the reaction to proceed. Depending on the experimental
conditions, either the mass-transfer step or the reaction step may be rate-limiting.
Under conditions in which mass transfer is rate-limiting, the mass-transfer rate
may be measured indirectly by measuring the rate of NaOH consumption in the

aqueous phase.

3.2.3 Relative rates of mass transfer and reaction

Nanda and Sharma (1966) developed methods to indirectly measure mass-
transfer coefficients for an ester (referred to below as A) by monitoring the
change in concentration of NaOH (referred to below as B) based on the following
assumptions:

(1) A is soluble in the aqueous phase, but B is insoluble in the nonpolar

phase. Thus A must transfer into the aqueous phase before it can react,
and the reaction occurs only in the aqueous phase.

101

(2) The reaction is slow enough that A diffuses through the interface into the
bulk aqueous phase, and the reaction occurs uniformly throughout the
aqueous phase containing B.

(3) The reaction is irreversible and first order with respect to both A and B.

To ensure that the reaction occurs uniformly throughout the aqueous phase and
that the rate is controlled by the transfer rate of A into the aqueous phase, two
conditions must be satisfied (Doraiswamy and Shanna, 1983). First, the rate of
mass transfer must be less than the rate of reaction. This condition may be

expressed using the following inequality:
KLa << K RC,0 (Condition 1)

Second, the amount of A that reacts in the film adjacent to the phase boundary
should be negligible compared to that which reacts in the bulk B phase. This

condition is satisfied when the following inequality holds:

(DAKIECBOVI << 1 (Condition 2)

L

The reaction system used in this study was the alkaline hydrolysis of p-
tolyl acetate. Because the aqueous solubility of p-tolyl acetate is low, the driving
force for mass transfer is comparatively small. At the same time, the reaction
rate is among the highest obtained for alkaline hydrolysis for aromatic esters of
acetic acid (Tommila and Hinshelwood, 1938). Consequently this reaction
system satisfies conditions 1 and 2, and is thus well-suited for determination of

mass-transfer coefficients through the measurement of reaction rates.

102

When Conditions 1 and 2 are both satisfied, CA is approximately zero, and the

mass balance on A in the continuous phase may be written

dC,
dt

 

= —KLaCA,. (Equation 2)

where CA1 is the equilibrium solubility of A in the aqueous phase. Assuming the
formation of the phenoxide goes to completion, two moles of B are consumed per

mole of A transfened, giving

dC‘ =15 (Equation 3)
dt 2 dr

 

Combining Equations 2 and 3 gives an expression for the interphase transfer rate

of A per unit surface area (RA):

 

1 IdC .
R =-— — B =K C. E uatlon4

Equation 4 indicates that KL can be calculated from a, the rate of disappearance
of B, and CA1.

The equilibrium solubility of non-electrolytes can be strongly affected by
the presence of salts. This phenomenon, known as the “salting” effect, results
from the activity coefficient of the non-electrolyte changing with the electrolyte
concentration. Using thermodynamic considerations, Long and McDevit (1952)
proposed the following expression to describe this effect for low non-electrolyte

concentrations:

 

C
ln[ CM ]= K51 (Equation 5)

A1

103

where
l = concentration of the electrolyte in solution

Ks = salt parameter for electrolyte-nonelectrolyte interaction

The above expression, which is of the same form as the Setshenow Equation
(Setshenow, 1889), was used by Nanda and Shanna (1966) to correlate the

solubility of liquids in electrolyte solutions.

Using Equation 5, CA1 can be expressed as

CA, = C”, exp(— K51) (Equation 6)
where CA“, is the equilibrium solubility concentration of the ester in pure water.
Substituting Equation 6 into Equation 4 gives:

R, = K,C,,_ exp(- K51) (Equation 7)

Equation 7 indicates that a semi-log plot of RA vs. I should give a straight line

having a slope of -K5.

The value of a can computed from the Sauter mean diameter 21: as

shown below:

a = 6 —.¢_-— (Equation 8)
d 32

where 4), the void fraction of the dispersed phase, is defined as the ratio of the

volume of nonpolar phase to the total volume. For most of the experiments

presented in this paper, 42 was 0.0035. At such a low void fraction, the CLA

104

rarely collide, so coalescence can be ignored (Femandes and Shanna, 1967).
The following expression (Tsouris and Tavlarides, 1994) can be used to describe

the effect of it on the maximum stable drop size (dcr).

+ 0 4 / .
dc, : CMNWE—Md'[l + 25¢ Li ] (Equatlon 9)
”.1 '1' Ii.

The dcIr values calculated using Equation 9 ranged between 180 and 440 pm for
the hydrodynamic conditions used in this study. Above this size, drop breakup
due to agitation becomes significant and can no longer be neglected. However,
the mean drop size of the CLA was in the range of 20-75 urn, with 99% of all
drops measured below the dc,. Hence breakage of CLA was neglected.

The KL,confinuous values were estimated using the following correlation by Skelland

etal. (1981).

K . d 0.548
L.cont1nuous = 1.864x10-7 ¢-0.287 [TIP] (NR5 )I.3‘ll(1qWE )—().()‘)5 (Equation 10)

M

The diffusivity value for p-tolyl acetate in water was calculated from the Wilke-
Chang correlation (Wilke and Chang, 1955). The molal volume of the p-tolyl
acetate at its normal boiling point was obtained using Le Bas group contributions
method (Reid and Sherwood, 1970).
3.3 Materials

Deionized water (Resistivity > 18MQ-cm) drawn from a Milli-Q PlusE water
purification system (Millipore Corporation, Bedford, MA) served as the aqueous
phase. The dispersed, nonpolar phase was p-tolyl acetate (Aldrich Chemical

Company, Milwaukee, WI). Nonionic surfactants were used in both the aqueous

105

and nonpolar phases. The aqueous-phase surfactant was polyoxyethylene
sorbitan monooleate (Tween 80, Aldrich Chemical Company, Milwaukee, WI ),
and the nonpolar-phase surfactant was a linear alcohol ethoxylate (Tergitol 15-S-
3, Sigma Chemical Company, St. Louis, MO). Because these surfactants do not
ionize in aqueous solution (Schick et al., 1967), they do not interfere with
measurement of NaOH concentrations using conductivity. The interfacial
structure created by this combination of surfactants has been characterized in
Chapter 1.

3.4 Methods

3. 4. 1 CLA preparation

The preparation conditions were found to influence the size distribution
and stability of the CLA phase; consequently, a consistent preparation procedure
was used in all runs. The CLA were prepared in a 50 mL glass beaker with an
internal diameter of 40 mm. Five mL of p-tolyl acetate/1' ergitol (0.01% v/v to
0.1% v/v ) solution were dispensed at a rate of 0.5 mL/min into a 5-mL volume of
aqueous/T ween 80 (0.001% v/v to 0.4% v/v ) solution. A 20 mm magnetic
stirring bar spinning at 1600 rpm was used to provide agitation. The resulting
CLA emulsions were creamy white, with a phase-volume ratio (PVR) of 1. The
PVR is defined as the ratio of the dispersed-phase volume to the aqueous-phase
volume. The emulsions were very stable, with no phase separation evident over

a period of 6 months.

106

3.4.2 CLA size measurement

Palticle-size distributions of CLA emulsions were measured using a
Malvem Mastersizer X particle analyzer (Malvem Instruments, Worcestershire,
UK). The instrument was fitted with a 15-mL sample chamber that was mixed
using a small Nalgene Star Head magnetic stir bar (Fisher Scientific, Pittsburgh,
PA). A 30 uL sample of the CLA dispersion was added to 15 mL of deionized

water in the cell; the size measurement was then made within 10 s of the CLA

addition. The particle size was reported as a 21: value.

3. 4.3 Mass-transfer rate measurement

Mass-transfer properties of the CLA were measured in a baffled, stirred
tank having standard dimensions (McCabe et al., 1993). The Plexiglass tank had
an ID of 15 cm and a height of 20 cm. The liquid height in the vessel was
maintained at 15 cm. The vessel was fitted with four, equally spaced, vertical,
wall baffles with standard dimensions. The six bladed, Rushton impeller had a
diameter of 5 cm and was axially centered in the tank 5 cm from the bottom. The
impeller was driven using a Lightnin® Labmaster ll unit (Mixing Equipment
Company, Rochester, NY). A wooden shaft was used to minimize the electrical
noise in the conductivity measurements. All experiments were conducted at
25°C. To initiate an experiment, the dispersed phase was injected into the
vessel close to the rotating impeller using a pipette. The rate of change in alkali
concentration was then measured by monitoring electrical conductivity, as

recommended by Rushton et al. (1964). A YSI 3417 (K=1.0/cm) conductivity

107

probe (YSI Inc, Yellow Springs, OH), 1/2 in OD, 5% in overall length used with a
YSI Model 32 conductance meter at a range switch setting of 200 ms". The
probe support ring was grounded to the (-) recorder jack to minimize electrical
noise in the signal. The output of the conductance meter was recorded by a
CAMILE 2000® data acquisition system (Dow Chemical Company, Midland MI) at
1-s intervals. The time constant of this probe was much smaller than that of the
interphase mass-transfer process, so the mass-transfer dynamics could be
measured accurately during the batch runs. The experimental values of KL were
determined using Equation 4 along with experimental measurements of a, CA1,
and the slope of the CB vs. time plot. The slope was determined over the region

of the curve during which half of the ester was hydrolyzed.

3.4.4 Physical-property determination

The aqueous solubility of p-tolyl acetate was measured using a Perkin-
Elmer Autosystems gas chromatograph (Perkin Elmer, Nonlvalk, CT) equipped
with a flame ionization detector and a packed AT-1000 column (Alltech
Associates, Deerfield, IL) 0.085” ID x 6’ length. Saturated aqueous solutions of
p-tolyl acetate were prepared by adding 200 uL of the ester to 1.3 mL of water
and vortexing for 15 min. The mixture was then centrifuged at 14,000 rpm for 2
min to separate the phases. Eight hundred microliters of the top (aqueous)
phase was drawn, and 4 11L of HPLC grade ethyl acetate was added as an
internal standard. One microliter of the resulting sample was injected to the gas
chromatograph. The flow rate of the carrier gas was 17 mL/min. The inlet and
detector temperatures were 150°C and 250°C, respectively. The following

108

temperature program was used: 60°C for 2 min, followed by a ramp of 20°C/min
to 225°C, where the temperature was held for 3 min. The residence times for
ethyl acetate and p-tolyl acetate were 2.8 min and 10.1 min, respectively. A
response factor (Ric) of 0.48 was determined using the following set of standards:
1 uL p-tolyl acetate and 25 uL ethyl acetate in 5 mL of water; 3 uL p-tolyl acetate
and 25 uL ethyl acetate in 5 mL of water; and 5 11L p-tolyl acetate and 25 uL ethyl
acetate in 5 mL of water. The response factor was then used to calculate the
concentration of the p-tolyl acetate in the saturated aqueous solution.

Viscosity was measured using a Brookfield DV-I+ viscometer (Brookfield
Engineering Laboratories Inc., Middleboro, MA) using the small-sample adapter.
A SC4-18/13R spindle size was used for these measurements at a shear rate of
132 s". The sample volume was 8 mL. Surface tension measurements were
made with a Krilss Processor Tensiometer K 12 (Kriiss GmbH, Hamburg) using
the Wilhelmy Plate method. The sample volume was 20 mL.

3.5 Results and Discussion

3.5. 1 Physicochemical properties

The measured physicochemical properties of the fluids are listed in Table

3.1.

109

Table 3.1 Physicochemical data for p-tolyl acetate-water system.

 

 

 

 

Solubility of p-tolyl acetate in pure water 0.0135 mol/L
Viscosity of p-tonI acetate 3.75 cP
Surface tension of p-tolyl acetate 31.66 mN/m
Interfacial tension of p-tolylacetate/0.4% Tween 80 4.6 mN/m

 

Interfacial tension of p-tolylacetate,0.1% Tergitol 15-S-3/0.4% 1.75 mN/m

 

 

 

 

Tween 80

Density of p-tolyl acetate 1.048 g/L
Reaction rate constant for hydrolysis of p-tolyl acetatea 0.3191 L/(mol s)
Viscosity of water containing 0.4% Tween 80 1.44 cP

 

 

aobtained from Tommila and Hinshelwood, 1938

110

 

3.5.2 Effect of Ionic Strength

The rate of reaction was measured for initial alkali concentrations from 0.5
to 0.75 M (ionic strength of 1.0 to 1.5 M). As shown in Figure 3.1, the semi-log
plot of R vs. I is linear, confirming the validity of Equation 7. This experiment was
repeated at 300 and 550 rpm. The slopes of the lines obtained at both speeds

were almost identical and gave an average Ks value bf 0.60.

3.5.3 Mass-transfer coefficients

Experimentally measured KL values are plotted as a function of Neg in
Figure 3.2 for B values of 0.5 and 0.75 M. The KL,commuous values predicted by the
Skelland et al. (1981) correlation (Equation 10) are shown for comparison. The
KL values agree well with the KLconfinuous values up to a Net; of about 12,500.
Above this value the KL values become independent of N95, while the KLCOminuous
values continue to increase.

These trends suggest that the controlling resistance shifts from the
continuous phase to the shell as NR9 increases. The structure of the
multilamellar surfactant shell, and hence its mass-transfer resistance, should be
largely unaffected by the turbulence in the continuous phase. On the other hand,
Equation 10 shows that the resistance of the continuous phase (1/KLcommuous)
should decrease with increasing Npg. Equation 1 was used to calculate KLshe"
from the experimentally measured KL values and the KLcommuous values obtained

using Equation 1. The resulting Kushe" values ranged from 0.033 to 0.040 cm/s.

111

 

-9.2 .
-9.25 —
-9.3 1
-9.35 —
-9.4
-9.45 1
-9.5 ~
-9.55 -
-9.6 1
-9.65 -

-..J

 

ln(R)

   
 

y = -0.6067x - 8.67

y = -0.6079x - 8.78

  

 

 

-9.7

0.8

T

T l

0.9 1 1.1 1.2 1.3 1.4 1.5

Ionic Strength (mol ion/L)

Figure 3.1: Semi log plot of specific reaction rate (R) vs. ionic strength (I) at 550
(El) and 300 (9) rpm.

112

 

0.03

0.025

 

 

0.02

0.015

 

KL (6 W3)

 

0.01

 

 

 

0.005

 

 

 

 

0 l I I
2000 7000 1 2000 1 7000 22000

Figure 3.2. Effect of Npg on KL values measured in baffled stirred tank at initial
alkali concentrations of 0.75M (CI) and 0.5M (O). The solid line represents the
prediction of Equation 9 (Skelland etal., 1981).

113

The experimentally measured KLa values varied from 115 to 450 h'1 over the Net;
range of 4,000 to 25,000, respectively. At a Npg of 20,000, the KLa was 400 h".
Rushton et al. (1964) measured a volumetric mass-transfer coefficient of 300 h'1
in a batch vessel in a system with similar dispersed phase viscosity (2.5 cP) at
the same NpE but a 11> value 25 times higher. To compare the mass-transfer
efficiency of these two systems on an equivalent basis, the KLa values were
normalized by dividing by the (I) values. The resulting KLa/¢ values were 114,000
h'1 for CLA system and 5,400 for the system described by Rushton et al. This
20-fold difference illustrates the potential of CLA to enhance interphase mass
transfer in liquid-liquid contacting operations. The enhancement would be
expected to be even greater for contacting operations conducted at lower
Reynolds numbers. Thus, the use of CLA offers the simultaneous advantages of
extremely high mass-transfer rates and the potential for low agitation power

requirements.

3.5.4 Effect of surfactant concentration on 21,— and KL

The effect of Tween 80 concentration on 21—,2 and KLa is shown in Figure

3.3, and the corresponding KL values are listed in Table 3.2. The 21: values for
the CLA decreased by about 50% as the concentration of the aqueous-phase
surfactant increased from 0.02% to 0.4%. Because a varies inversely with 21:,

and the KL values were essentially independent of surfactant concentration, the
Km values increased by a comparable amount (about 50%) with increasing

surfactant concentration.

114

 

 

60 450

 

 

 

 

 

 

 

 

 

—- 400
50 /
t / 350
E 40 300
3- x
a -- 250 rm
9 30 1 A
g A ~ 200 =3
3 20 150
— 100
10
»- 50
o a t t . o
o 0.1 0.2 0.3 0.4 0.5

Tween 80 Concentration % (v/v)
Figure 3.3. Variation of CLA Sauter mean diameter (0) and volumetric mass

transfer coefficient (KLa, El) with Tween 80 concentration (% v/v) in the aqueous
phase.

115

Table 3.2. Variation of interfacial area (8), overall mass-transfer coefficient (KL),
and volumetric mass-transfer coefficient (KLa) with Tween 80 concentration in
aqueous phase.

 

 

 

 

 

 

Tween 80 conc dag (um) KLa (h") a (cmz/cma) KL (cm/s) x
(%v/v) 103
0.4 25.8 405 8.3 13.5
0.07 30.2 286 7.1 11.1
0.025 41.2 214 5.2 11.5
0.018 48.9 167 4.4 10.6

 

 

 

 

 

 

116

The result that the KL values were relatively unaffected by the surfactant
concentration suggests that the thickness of the shell is not significantly affected
by the surfactant concentration. Alternatively, any differences in shell thickness
may dissipate quickly after the CLA are transferred into the continuous aqueous
phase. This transfer step suddenly decreases in the surfactant concentration
surrounding the aphrons from greater than the critical micelle concentration
(CMC) to far below the CMC. This change creates a large driving force for
surfactant migration from the shell into the bulk solution and could lead to a rapid
thinning of the shell. This explanation was offered by Bredwell and Worden
(1998) to explain the observation that KL values measured for surfactant
stabilized microbubbles were considerably higher when measured in pure water
than when measured in an aqueous surfactant solution above the CMC.

3.6 Conclusions

The interphase mass-transfer properties of CLA emulsions in water have
been measured by coupling the mass-transfer of the solute (p-tolyl acetate) to a
chemical reaction. The overall KL values, which ranged from 0.002 to 0.015
cm/s, increased with Na; below a Npg value of about 12,500. Above this value,
KL was essentially independent of Nae. These trends suggest that as the degree
of agitation is increased, the controlling resistance shifts from the continuous
phase to the shell phase. The KL values were essentially independent of the
surfactant concentration used to prepare the CLA. A resistances-in-series model
was used, along with a literature correlation for KLcominuous, to estimate the KLShe”

value to be about 0.04 cm/s. At a nominal NpE value of 20,000, the KLa/o value

117

 

measured for CLA was about 20 times higher than that for conventional
emulsions.

3.7 Nomenclature

a = Interfacial area per unit volume in two phase mixture (mfg/"13), defined by £133—
2 (3

C = Concentration of solute (mol L")
CAw= Equilibrium solubility concentration of the ester in pure water (mol L")

o = Diffusivity, (cm?- s")
_ 2nd d°
(132 = Sauter mean diameter defined by 2119 d;,( m)

d.= Diameter of the impeller, (m)

dc, = Maximum stable drop size, (m)

I = Concentration of the electrolyte or salt in solution (mol ion L")

KL = Overall mass-transfer coefficient, (cm s")

KLcomtnuous = Mass-transfer coefficient of the continuous phase (cm s“)
KL_she.. = Mass-transfer coefficient of the CLA shell (cm s")

Kp = Reaction rate constant (L mol'1 3")

Ks = Salt parameter for ion - nonelectrolyte interaction (L mol")

N = Impeller speed, (3").

u

Jpcd 6

9.1%?

NOH- — Ohnesorge number, defined by

NnE = Reynolds number, defined by

C

118

"'51

-_
A

2 3
Nwe = Weber Number, defined by —————N :19,

RC = Response factor defined by (Canalwe/Cstd) x (Astd/Aaname)
T = Tank diameter, (m)
Vc = Volume of the continuous phase, (L)

V = Filled volume of the tank, (L)

Greek

¢ = Void fraction of dispersed phase, dimensionless

u = Viscosity, (Ns me)

p = Density, (kg ma)

0 = Interfacial tension, (N m")

Subscripts

d = dispersed phase

c = continuous phase

3.8 References

Albright, L. F., and Hanson C. (1975). Industrial and Laboratory Nitration (Am.
Chem. Soc. Symp. Ser. No. 22), American Chemical Society, Washington D.C.
Bhave, RR. and Shanna MM. (1981). Trans. Inst. Chem. Eng., 59, 161-170.
Bredwell M.D., Worden RM. (1998). Mass-transfer properties of microbubbles. 1.
experimental studies, Biotechnology Progress, 1 4: (1 ) 31 -38.

Doraiswamy L.K., Shanna MM, (1983), Heterogeneous Reactions: Analysis,
Examples, and Reactor Design, Vol 2: Fluid- Fluid-Solid reactions, John Wiley &
Sons, New York, NY p 17.

119

 

Femandes J.B., and Shanna MM. (1967) Chem. Engg. Sci., 22,1267-1282.
Kafarov, V.V.; Reutsli, V. A.; Zhuravleva, T. Yu. (1974), J. Appl. Chem USSR,
49(12), 2724.

Long FA, and McDevit W.F. (1952). Activity coefficients of nonelectrolyte
solutes in aqueous salt solutions, Chem. Rev. 51, 119-170.

McCabe W.L., Smith J.C., Harriot P (1993). Unit Operations of Chemical
Engineering, Fifth Edition, McGraw Hill Inc, New York NY.

Nanda A.K. and Shanna MM. (1966), Effective interfacial area in liquid-liquid
extraction, Chemical Engineering Sci., 21, 707-714.

Reid R.C., Shenlvood T.K. (1970). The Properties of Gases and Liquids: their
estimation and Correlation, 2"d ed., McGraw Hill book company, New York, NY.
Rushton J. H., Nagata S., Rooney TR (1964). Measurements of mass-transfer
coefficients in liquid liquid mixing. AICHE J. 10 (3), 298-302.

Schick, M.J. (1967), in Nonionic surfactants, ed. by Schick M.J., Marcel Dekker
Inc. New York. 1-5.

Setschenow,J., Z. Physik. Chem, (1889) 4, 117.

Tommila E, and Hinshelwood ON. (1938). The activation energy of organic
reaction. Part IV. Transmission of substituent influences in ester hydrolysis, J.
Chem. Soc. (London), 2, 1 801 -181 0.

Tsouris, C., and LL. Tavlarides (1994). Breakage and coalescence models for
drops in turbulent dispersions, AICHE J., 40, 395-406.

Wilke, C.R., and P. Chang (1955). Conelation of diffusion coefficients in dilute

solution, AICHE J., 1, 264.

120

Chapter 4

4. LIPASE CATALYZED DYNAMIC RESOLUTION OF
AMINO ACIDS

4.1 Introduction

Chiral drugs are enormously important to both public health and the
pharmaceutical industry. Worldwide sales of single-enantiomer drugs surged
from $36 billion in 1993 to $90 billion in 1997 (Stinson et. al, 1998). The market
of single-isomer fine chemicals (drug precursors) to pharmaceutical companies
was $25 billion in 1998 (Rogers, 1999).

Combinatorial chemistry methods are now providing an essentially
unlimited number of chemical entities for screening (Brennan, 2000). Once new
chemical entities are identified that bind to the desired biological targets and
have desirable phannacokinetics, they must be synthesized in sufficient
quantities for clinical trials and, in some cases, commercial production. The
ability to synthesize chiral compounds cost-effectively at the commercial scale is
important to keeping health-care costs manageable. Enzyme-catalyzed
biotransformations are commonly used for production of chiral drugs and drug
precursors (Stinson et. al, 1999), often in tandem with conventional chemistry
(Broxtennan et. al, 1990).

The use of enzymes for biotransformations of drug precursors presents

severe processing challenges. Ideal drug candidates would be hydrophobic to

121

allow passage through cell membranes (Brennan, 2000). Hence, many chiral
resolution opportunities involve organic chemicals that are sparingly soluble in
water. Low substrate solubility presents special problems when enzyme-based
resolution methods are considered. One challenge is finding an enzyme that has
activity and the desired stereoselectivity on what usually may be an unnatural
substrate. Enzymes have been identified that are enantioselective, yet have
broad substrate specificities to convert a non-natural substrate. These enzymes
are particularly useful in the production or resolution of carboxylic acids, esters,
and alcohols. Another challenge is to design bioreactors that make enzyme-
based schemes economically feasible on a commercial scale.

The multiple mass transfer and reaction steps involved in dispersed-phase
bioreactor systems make engineering of such systems complicated. Consider
the generalized case in which it is desired to produce a resolved carboxylic acid.
A prefened enzymatic approach is to use a hydrolytic enzyme, such as a lipase
or esterase, to stereoselectively convert an ester to its respective acid and
alcohol:

R’002 R” + H20 -) R’COzH + R”OH
In many of the chiral compounds of commercial interest, the acyl moiety R’
contains substituted benzyl, naphthyl, and/or aryl oxy functionalities, which
reduce the aqueous solubility of the ester. The resulting acids, however, exhibit
significant water solubility since the optimum pH for the reaction is usually above

their pKa values.

122

Lipase enzymes have proven particularly valuable for such
biotransformations (Patel et. al, 1996). Lipases (triaglycerol acylhydrolase, EC.
3.1.1.3) can be viewed as a special class of esterases. Esterases preferentially
catalyze the hydrolysis of soluble esters, whereas lipases are distinguished by
their ability to catalyze the hydrolysis of insoluble fatty acid esters. Interfaces
provided by either aggregated or dispersed substrate in the aqueous medium
constitute the primary sites for lipase catalysis. A stepwise increase in activity is
observed for lipases when the solubility limit of the substrate is surpassed and
substrate aggregates are formed (Entessangles and Desnuelle, 1968).
Benzonana and Desnuelle (1965) have pointed out the importance of interfacial
area rather than the bulk substrate concentration in determining the reaction rate,
and have shown that lipolysis will apparently conform to Michaelis Menten
kinetics when substrate concentration is expressed on the basis of interfacial
area rather than bulk system volume. Adsorption of the enzyme to the interface
may also change its intrinsic catalytic activity either as an immediate
consequence of adsorption, or as a slow post adsorptive conformational change
(Brockman, 1984).

The commercial potential of lipases and other multiphase enzymatic
biotransformations has not been realized, due to processing limitations
associated with conventional multiphase contacting systems. These limitations
include poor reproducibility, low volumetric productivity, high energy requirement
for dispersing the substrate during the reaction step, and difficulty in product

separation (Patel et al. 1996). Clearly, more effective methods are needed to

123

provide intimate contact between the enzyme and its nonpolar substrate (Scott
et. al, 1995).

The mass transfer properties of CLA in Chapters 2 and 3 have been
characterized. CLA emulsions provide a 5-15 fold rise in interphase volumetric
mass transfer rates over conventional emulsion systems. This result suggests
that CLA emulsions can be used to enhance the rate of reaction in multiphase
enzyme catalysed systems.

4.2 Model System

One multiphase biocatalytic system that could benefit from accelerated
mass transfer, is the dynamic resolution of amino acids to produce unnatural D-
amino acids. As components of drugs, such amino acids would retain the chiral
chemistry needed to bind with the biological target, but might be poor substrates
for enzymes that would break the drug down (Stinson, 2000). Recently there has
been enormous interest in D-phenylalanine for the development of HIV protease
inhibitors (Bosswell, 1999). The Strecker reaction (the addition of cyanide to
imines) is a direct strategy for asymmetric synthesis of alpha-amino derivatives.
Although significant progress has been made in the stereoselective version of
this reaction using imines bearing covalently attached chiral auxiliaries, gains
have been limited for an enanioselective catalytic version (Jacobsen, 1999). One
approach to purifying unnatural amino acids (D enantiomers) from racemic
mixtures is dynamic resolution. This study investigated on the dynamic resolution
of D-L Phenylalanine (D, L-PA) using lipase catalyzed hydrolysis of D-L

phenylalanine methyl ester (D, L-PAM) (Figure 4.1). The use of CLA as enzyme-

124

immobilization supports has also been investigated. The amount of lipase
immobilized on the CLA droplets formed using a variety of anionic and non-ionic
surfactants was measured. Lye et al. (1997) found that when a-chymotrypsin
was immobilized on CLA, only about 1% of the activity was retained. Lamb et al
(1999) investigated the immobilization of several other enzymes on CLA. An
(it-amylase displayed superactivity when immobilized, while lipase, trypsin and
Iysozyme were inactivated. Both the above studies utilized sodium dodecyl
sulphate, an anionic emulsifier) to form CLA emulsions. Sodium dodecyl sulphate
(SDS) is used for estimating polypeptide sizes, either by gel-electrophoresis or
by molecular-sieve chromatography, and is known to bind with water soluble
proteins causing them to form rod like shapes that are then separated by the gel
(Muga et al. 1993). Such a binding would disrupt the protein structure and
function. The objective of this study was to identify a surfactant that interacts
only weakly with the enzyme, so as to prevent protein refolding and denaturation.
CLA emulsions formed using such a surfactant would provide high rates of mass
transfer, while retaining the activity of the enzyme responsible for the resolution
reaction.

Further, the influence of various environmental parameters such as pH,
enzyme concentration and interfacial area on lipase catalyzed dynamic resolution

of D-L phenylalanine in CLA emulsions has been studied.

125

  
 

Porcine Pancreatic Lipase O

O /’
NH2 H0
D, L- Phenylalanine methyl ester' L Phen l '
. - alamn
Reactant soluble 1n Decanol Water soluble pioduct

Figure 4.1 Lipase catalysed resolution of D,L-Phenylalanine using hydrolysis of
D,L Phenylalanine methyl ester (PAM). ‘

4.3 Materials and Methods

All the solvents used were HPLC grade. Water and methanol were
obtained from JT Baker (Philipsburg, NJ, USA). Acetonitrile was obtained from
EM Science (Gibbstown NJ). Buffer solutions ranging from pH 3 to pH 7 were
prepared using the recipes described in Dean (1992). Table 1 outlines the
recipes used to make the buffer solutions. Both the sodium phosphate dibasic
and the citric acid were obtained from Sigma Chemical Company (Milwaukee,
WI) and were ACS reagent grade.

Table 4.1. Buffer recipes for chiral resolution ex eriments

 

 

 

 

 

 

pH x mL of 0.2 M N82HPO4 . 2H20 y mL of 0.1 M Citric Acid (19.213
(35.599 9 - L") g - L")
3 - 20.55 79.45
4 .' 38.55 61.45
5 51.5 48.5
6 63.15 36.85
7 82.35 17.65

 

 

 

 

 

126

Crude porcine pancreatic lipase (SIGMA 3126, Sigma Chemical Company,
Milwaukee, WI) was used as the biocatalyst in this study. This crude extract was
analyzed for protein content using a Bradford DC protein assay. Bovine Serum
Albumin was used as a standard for protein quantitation. Surfactants used in this
study include Polyoxyethylene sorbitan monooleate (Tween 80), Polyoxyethylene
sorbitan monolaurate (Tween 20), Sodium dodecyl sulphate (SDS), Hexadecyl
trimethyl ammonium bromide (HTAB), and Tetradecyl trimethyl ammonium
bromide (TT AB). Tween 80 and Tween 20 are non ionic emulsifiers, SDS is
anionic and HTAB and TTAB are cationic emulsifiers. All the above emulsifiers

were obtained from Sigma Chemical Company, Milwaukee, WI.

4. 3. 1 Experimental Methods

4.3.1.1. Immobilization efficiency

Immobilization experiments were carried out in 50ml total volume,
containing 40mL of pH5 buffer solution, 80mg of surfactant (0.2%), 80mg of
crude enzyme. A 0.5 mL sample was drawn from this solution. 10mL of decanol
was added, followed by vigorous agitation to form CLA emulsions. The emulsion
was allowed to cream andthe aqueous phase was sampled using a micropipette.
The protein content was determined using the Bradford DC protein assay.
Calibration curves were constructed for each surfactant system and used to
determine the amount of protein present in the samples. The percentage of
enzyme immobilized was calculated based on the difference between the amount
of enzyme present in the samples before the decanol was added and those

taken after the CLA dispersion was formed.

127

4.3.1.2. Chiral resolution experiments

D,L- Phenylalanine methyl ester hydrochloride (Sigma Chemical
Company) was dissolved in water in a separatory funnel and overlaid with a
layer of diethyl ether. The water phase was neutralized with 25% stoichiometric
excess of 10M NaOH solution accompanied by rigorous shaking. During this
process the phenylalanine methyl ester (Phe-Me), formed by neutralization of
the hydrochloride, was extracted into the ether phase. The phases were allowed
to separate, and than pure Phe-Me was obtained by evaporation of the ether
under vacuum. The ester was weighed, and stock solution of the ester in decanol
was prepared. Over 90% yield was obtained in this process.

The experiments were conducted in 100 mL glass bottles. The appropriate
amount of buffer was added to the bottle. Tween 80 (Aldrich Chemical Company,
Milwaukee, WI) was added to the bottle and vortexed to ensure proper mixing.
Decanol, which wasused as the dispersed organic phase, forms molecular
complexes with Tween 80 leading to the formation of lamellar micelles that
stabilize CLA emulsions, as discussed in Chapter 1. The Phe-Me stock solution
was dispersed into the aqueous phase and vigorously stirred to form CLA
emulsions. The CLA emulsions formed were creamy white and extremely stable,
with no phase separation evident over a period of several months. Appropriate
amounts of crude lipase were weighed and added to the bottle, followed by
vortexing for 30 seconds to ensure complete mixing. The bottle was then placed
in a water bath (25°C) in an INNOVA 3000 platform shaker (New Brunswick

Scientific Co. NJ) and shaken at 200 rpm. Samples were taken using 1 mL

128

syringes and placed in 10 mL vials after passing through a 25 mm Acrodisc‘g' 0.2
pm syringe filter (Pall Gelman, Ann Arbor, MI) to remove CLA droplets. This
process was also effective in removing protein, as confirmed by adsorption of the
filtered sample at 280nm.
4.3.1.3. HPLC assay

A common approach to separate enantiomers by chromatography is the
use of a homo-chiral (single-enantiomer) derivatizing agent (HDA), which reacts
with the two enantiomers to be distinguished (Gal, 1987). In this study, 2,3,4,6
tetra-O-acetyl- B-D- glucopyranosyl isothiocyanate (TAGIT) was utilized. This
reagent reacts with primary or secondary amines to yield corresponding thiourea
derivatives. The reaction products of TAGIT with L-D phenylalanine are shown in
Figure 4.2. The two products of the reaction are no longer enantiomerically
related, but are diastereomers. Unlike enantiomers, diastereomers have slightly
different physical properties and can be separated under reversed-phase
conditions using a Cm HPLC column. The derivatization procedure used is a
modification of the method of Kinoshita et al. (1981). Forty uL of sample was
added to a 20 ml glass scintillation vial, along with 1 mL of 1% (v/v) triethyl amine
(Spectrum Quality Prod., Gardena, CA) solution in water and 1 mL of TAGIT
(Fluka Chemie AG, Buchs, Switzerland) stock in acetonitrile (0.1 g/ 10 g). The
vial was vortexed for 10 seconds and placed on a shaker for 1hr at 200 rpm.
The chromatographic system consisted of a WATERS 600 solvent delivery
system, a reversed phase C18 HPLC column (Novapak® C18, 3.9x150 mm, 4 pm

particles, WAT086344, Waters Corp, Milford, MA) and a Water 490 multiple

129

wavelength UV detector. The column was eluted at room temperature and at a
flow rate of 1 mL/min with a mobile phase prepared by mixing 0.2 M pH 5 acetate
buffer and methanol in a ratio of 55:45. The wavelength used for detection of the
eluent was 250 nm. The retention times for L-phenylalanine and D-
phenylalanine were 6.26 min and 9.41 min respectively. Response factors for the
L and D diastereomers (429,726 and 474,434 respectively), were determined by
running standards (in triplicate) between 0.125 g/L and 20 g/L. The reagent
peaks were well separated from those of the amino acids and did not interfere
with the detection. A sample chromatogram of the separation obtained is shown

in Figure 4.3.

130

 

 

 

 

 

Figure 4.2. Reaction products of L,D PA with TAGIT

131

 

0.25 :1

0.2 .

 

1250

0 .1 5 - L-PA
D-PA

0.05 -

 

 

O I ‘ I l l . l

012 3 4 5 6 7 8 9101112131415
time (minutes)

 

 

 

 

Figure 4.3. Sample HPLC Chromatogram of the enantiomeric separation of L
and D Phenylalanine

132

4. 3.2 Conversion and enantiomeric excess measurements

A statistical experimental design, (central composite design) was
implemented to determine the effect of process parameters and physical
conditions on the equilibrium conversion and enantiomeric excess of the lipase
catalyzed dynamic resolution reaction. Five factors were identified for the
experiments, namely:

1) Surfactant concentration in aqueous phase

2) Void fraction

3) Amount of enzyme present

4) pH

5) Concentration of Phe-Me in the decanol phase

The experimental design involved running 32 experiments. Details on the factors

studied and the levels chosen are discussed in the next section.

4.3.2. 1. Central Composite Design

The central composite design (CCD) is an efficient design for fitting
second order models that account for curvature effects in the response variable
(Montgomery, 1997). A quadratic response to some of the variables was
expected, most notably, to pH, because most enzymes exhibit an activity
maximum at an intermediate pH. The design included a half fraction of 25
experiments with seven replicates on the center point of the design, augmented

with nine points outside the five dimensional cubic space explored by the factorial

133

experiments. Regression was used to determine the linear main effects,
curvature effects and interaction effects of the above mentioned variables.

The solubility of L- phenylalanine is 29.6 g/L at 25°C (Kroschwitz et al., 1987).
This places a constraint on the void fraction of the oil phase and concentrations
of the Phe-Me in the oil phase that can be utilized in the chiral resolution process.
The range of void fraction studied (20-40%), combined with the concentration
range (0.1 M to 0.3 M) satisfies this constraint, maintaining maximum possible
concentration of the product, L-phenylalanine, below 209/L.

Surfactant concentrations were chosen to be in the range of 0.2-0.4%. The
center point of the CCD was chosen as void fraction 30%, pH 5 buffer containing
0.3% surfactant, 80 mg crude enzyme and 0.2M Phe-Me in decanol.

The entire statistical design is shown in Table 4.2. Table 4.3 provides the coding
for each individual variable. For example, Table 4.3 indicates that scaled variable
P is given by the relation (pH-5)/1. Therefore, a value of P=1 in Table 4.2 would
correspond to a pH of 6 in the experiment. The half fraction of the 25 set and the
six center points were run first in the (random) order listed. An additional center

point experiment and the axial experiments were run next.

134

(coded variables are defined as shown in Table 4.3)

Table 4.2. Central Composite Design

 

Odd-41101101

44440140110440000002000

 

14140110110440022000000

 

 

14410140440

410000220000

 

04.410.14.01

410000000220

 

 

Run

 

123456789

 

 

135

Table 4.3 Coding of variables for Central Composite Design

 

(Surf conc (%)-0.3)/0.05

(Enzyme amount(mg)-80)/25

(Void fraction(%) -30)/10

(pH-5)/1

(Concentration in decanol (Molar)-0.2)/0.1

 

 

 

 

O‘O<ITIU)

 

 

 

 

4.4 Results and discussion

4. 4. 1 Enzyme interactions with CLA

The percentage enzyme immobilized for different surfactant systems is
shown in Table 4.4. CLA emulsions made with nonionic surfactants (Tween 80
and Tween 20) appear to be have low carrying capacity for the lipase enzyme.
However, cationic or anionic surfactants appeared to immobilize 30-50% of the
enzyme. Accounting for the purity of the crude enzyme (33%), it appears that
about 2mg of pure lipase/ mL CLA is immobilized in the ionic surfactants. This
value agrees well with the result obtained by Lye and Stuckey (1998). Lye and
Stuckey (1998) have reported that pH had a strong effect on the binding of
enzyme to CLA formed using ionic emulsifiers. However, in case of nonionic
emulsifiers pH is not expected to affect enzyme binding to CLA, since their

interactions are mainly hydrophobic. This expected trend was confirmed by the

experimental results shown in Table 4.5.

136

Table 4.4 % Enzyme immobilized using different nonionic and

emulsifiers to form CLA emulsions at pH 5.

 

 

 

 

 

 

 

 

Surfactant used % immobilized
Tween80 8
Tween20 26

HMB 47.5
'I'I'AB 45.3
SLS 31 .2

 

 

Table 4.5 Effect of pH on % enzyme immobilized on CLA emulsions formed
using Tween 80.

 

 

 

 

 

 

JH % immobilized
4 12%
5 8%
6 1 1.20%
7 7.80%

 

 

 

4. 4.2 Reaction rate measurements

Figure 4.4 shows the variation of conversion with time for experiments
with and without CLA emulsions. It is evident that the equilibrium is attained
within 1 h in the presence of CLA under these conditions, compared to over 4 h
in the absence of CLA. This indicates that the CLA enable better contacting and

faster rates of approach to equilibrium for this process.

4. 4.3 Chiral Resolution

Data for equilibrium conversion (Yc) and enantiomeric excess (Yee) are
shown in Table 4.6. The data were analyzed using a linear regression procedure
within the statistical computer package MINITAB. Polynomial models, which
relate Yc and Yea to the process variables through linear, quadratic and

interaction coefficients, were fitted.

137

 

 

Conversion (%)

88868838

 

0 L/ i 1 l l

o 0.51 1.5 2 2.5 3 3.5 4
time(h)

Figure 4.4 Conversion vs. time for chiral resolution experiments with (I) and
without ( ) CLA dispersions. Reaction conditions: pH 5, Void fraction 30°/e, 80
mg crude enzyme, 0.2M Phe-me in decanol.

138

Table 4.6. Experimental data for chiral resolution experiments.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Run# Surf Enzyme Void pH Gone. in Conversion Enantiomeric
conc. amount Fraction Decanol Yc Excess
% m % (Molar) Y...e

1 0.30 80 30 5 0.2 70.69 92.44
2 0.25 105 40 6 0.1 65.44 86.32
3 0.25 55 20 4 0.3 22.79 87.67
4 0.35 105 40 6 0.3 69.68 83.86
5 0.30 80 30 5 0.2 63.46 94.01
6 0.35 55 40 4 0.3 28.59 83.47
7 0.25 105 40 4 0.3 66.46 95.26
8 0.30 80 30 5 0.2 67.84 94.26
9 0.35 105 20 4 0.3 23.90 87.91
10 0.35, 105 20 6 0.1 96.20 90.94
11 0.35 55 20 4 0.1 7.09 21.68
12 0.25 55 20 6 0.1 93.77 91.35
13 0.25 105 20 4 0.1 7.38 31.26
14 0.30 80 30 5 0.2 67.94 95.56
15 0.25 105 20 6 0.3 62.55 86.73
16 0.35 55 40 6 0.1 51.67 85.49
17 0.30 80 30 5 0.2 60.31 84.37
18 0.25 55 40 6 0.3 53.35 85.54
19 0.35 55 20 6 0.3 53.83 80.87
20 0.30 80 30 5 0.2 61 .20 88.00
21 0.25 55 40 4 0.1 18.81 59.06
22 0.35 105 40 4 0.1 25.74 63.22
23 0.3 80 10 5 0.2 39.63 88.87
24 0.3 80 50 5 0.2 63.52 91.20
25 0.3 80 30 3 0.2 5.67 25.02
26 0.3 80 30 7 0.2 58.00 87.05
27 0.3 30 30 5 0.2 55.93 88.62
28 0.3 130 30 5 0.2 69.16 91.29
29 0.3 80 30 5 0.4 68.72 90.56
30 0.2 80 30 5 0.2 64.50 91.27
31 0.4 80 30 5 0.2 62.92 91.59
32 0.30 80 30 5 0.2 57.80 90.51

 

 

 

 

 

 

 

 

139

 

4.4.3. 1. Conversion (Yc)

The results of a statistical analysis of the data are shown in Table 4.7.
The plot of the residual vs. the fitted Yc values (Figure 4.5) revealed the Run 25
has an unusually high residual compared to the rest of the data. The fitted value
for this data point is negative, which is physically unrealizable. This experiment
contributes heavily to the large standard deviation (8.14) observed for this model
compared to the standard deviation (4.74) of the replicates for the center point of
the design. Ignoring experiment 25 as an outlier, the rest of the data were refitted
using the linear regression technique. This analysis was used to select the
statistically most significant variables (p< 0.05). The model was then refitted
using only the selected variables. The results of the statistical analysis are given
in Table 4.8. This fitted linear regression model is given as Equation 1.

c = 63.83 + 4.40 E + 2.85 v + 21.4 P - 3.34 v2 - 12.7 P2 + 3.31 SE

+ 3.38 EV - 9.56 VP 4» 6.63 VC - 9.92 PC (Equation 1)

This equation shows that the most significant variable are the main effects
E, V, P, quadratic terms for V, P, and some interaction effects namely VP, PC,
PE, PC, SE. Plotting the residuals in time order of data collection is helpful in
detecting the correlation between the residuals. A tendency to have runs of
positive and negative residuals indicates positive correlation. This would imply
that the independence assumption on the errors has been violated. A plot of
residual vs. order of data is shown in Figure 4.6a. The variance from each run

seems completely uncorrelated, so there is no reason to suspect any violation of

140

Table 4.7 Regression analysis for data presented in Table 4.6

The regression equation is
Yc = 64.9 + 4.4 E + 18.4 P - 3.97 v2- 8.90 P2 - 9.56 VP
+ 6.63 VC - 9.92 PC

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Predictor Coef Stdev T P
Constant 64.905 3.029 21 .43 0
S -1.196 1.663 -0.72 0.487
E 4.4 1.663 2.65 0.023
V 2.846 1.663 1.71 0.115
P 18.42 1.663 11.08 0
C 2.098 1.978 1.06 0.312
S" 0937 1 .509 -0.62 0.547
E" -1.227 1.509 -0.81 0.433
V" -3.968 1 .509 -2.63 0.023
P" -8.904 1.509 -5.9 0
C" -1.371 1.986 -0.69 0.505
SE 3.307 2.037 1.62 0.133
SV -0.913 2.037 -0.45 0.663
SP 1.13 2.037 0.55 0.59
SC -1.01 2.037 -0.5 0.63
EV 3.377 2.037 1.66 0.125
EP 0.21 2.037 0.1 0.92
EC 2.019 2.037 0.99 0.343
VP -9.56 2.037 -4.69 0.001
VC 6.631 2.037 3.26 0.008
PC 9919 2.037 -4.87 0
Analysis of variance
Source DF SS MS F P
Regression 20 15936.73 796.84 12.01 0
Residual Error 1 1 730.12 66.37
Total 31 1 6666.86
S = 8.147 R-Sq = 95.6% R-Sq(adj) = 87.7%

141

 

 

 

Residuals Versus the Fitted Values

 

 

 

 

O
10—
O
.0
O
O
— O
E ' ' '
“a ' l
0 0d.----------------------------------.. ........ 9 ..................
C: O . . C
O
’ O O
O . . g
0 O
O
.10—q
l l T
0 50 100
Fitted Value

 

 

 

Figure 4.5 Residuals vs. fitted values for full set of experimental data. (Note: The

fitted value for one of the data points (experiment 25) is negative, and the
residual is extreme)

142

Table 4.8. Regression analysis results for model represented by Equation 1.
\rc = 63.83 + 4.40 E + 2.85 v + 21.4 P - 3.34 v2 -12.7 P2 + 3.31 SE

+ 3.38 EV - 9.56 VP + 6.63 VC - 9.92 PC

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Predictor Coef StDev T

Constant 63.83 1 .364 46.79 0
E 4.4 1.021 4.31 0
v 2.846 1.021 2.79 0.011
P 21.369 1.199 17.82 0
v2 33433 0.9254 -3.61 0.002
P2 -12.703 1.221 -10.4 0
SE 3.307 1.251 2.64 0.016
Ev 3.377 1.251 2.7 0.014
VP -9.56 1.251 -7.64 0
vc 6.631 1.251 5.3 0
PC 9919 1.251 -7.93 0

 

s = 5.004 R-Sq = 96.5% R-Sq(adj) = 94.8%

Analysis of variance

 

 

 

 

 

 

 

 

 

Source DF SS MS

Regression 10 13863 1386.3 55.37
Residual 20 500.7 25

Error

Total 30 14363.8

 

143

 

 

 

 

 

Residual
O

 

 

 

—
_
—1
——i
—
..1

Observation Order

 

 

 

Figure 4.6a. Plot of residuals vs. order of data for regression model represented
by equation 1.

144

 

 

 

 

 

2— o
O
O
O
14 . o
r ' .
2 I
3 I
a) 0" O.
'5
g , 1
2° 3'
.1— .
O
O
O
.2— .
1 l l l l l l l l
8 6 4 2 o 2 4 6 8
Residual

 

 

 

Figure 4.6b. Normal Probability plot for residuals based on Equation 1.

145

 

 

 

 

 

 

8—
O
.0
5—
O
O
4— l
O
2% . .
2 0'— ------------------------------ .- ----------------------
0) . O .0
a) . 0
I 2.4
' 2 .
.4—
. O
.6... .
O
-8— '
l l l l l l l l l I
0 10 20 30 40 50 60 70 80 90 100
FittedValue

 

 

 

Figure 4.6c. Plot of residuals vs. Fitted values for regression model given by
Equation 1.

146

the independence or constant-variance assumptions. The normal probability plot
is shown in Figure 4.6b. The plot of the residuals vs. fitted values (Figure 4.60)
does not reveal an obvious pattern, indicating the adequacy of the model and its
assumptions. The model presented as Equation 1 is obviously empirical, but it
does highlight the effects in the response variables caused by the various
factors.

Mechanistic aspects of the significant effects of each parameter is
considered separately below:
1) Concentration of surfactant (S):
Within the range studied (0.2% -0.4%) Tween 80 concentration does not seem to
affect the equilibrium conversion or the enantiomeric excess significantly. Weak
interaction effects are present between amount of enzyme (E) and surfactant
concentration (S). Although these effects are statistically significant, their
influence on conversion is small from a process standpoint. In the presence of
CLA, the volumetric rate of reaction may be limited by enzyme concentration and
hence raising the concentration of the surfactant (i.e. reducing droplet size

(Chapter 2) does not affect the reaction.

2) Amount of Enzyme (E):

The positive coefficient in the regression model for the main effect of enzyme
concentration indicates that the higher the concentration of the enzyme the
greater the conversion. This would imply that the system is reaction rate limited,

rather than mass transfer limited. This interpretation is consistent with the

147

 

previous observations of the effect of surfactant concentration. However the
increase in conversion obtained by raising enzyme concentration is not
significant, due to the counteracting effects of the interaction terms (EV, since V
is favored at low levels as discussed later in this section).

3) pH of the aqueous phase (P)

The pH of the continuous phase buffer seems to be the single most significant
factor in the system. The presence of the quadratic term indicates curvature in
the response variable. Since most enzymes have optimal pH, this trend is as
expected. The interaction term between void fraction and pH, represented by VP,
is also significant, suggesting a change in conformation of the active site for the
enzyme. The conformation of the site would be affected by ionization (pH) as well
as presence of water affecting protein folding collectively.

4) Void fraction (V):

The void fractions of substrate and water are critically important to the course of
hydrolytic reaction. Both the main effect and the quadratic terms for void fraction
are statistically significant, in addition to interactive effects with other variables. If
a large amount of water is used, a high degree of hydrolysis may be achieved.
However, higher amounts of water imply lower volumes of the oil phase
containing substrate, and hence the total mass produced in the reactor would be
lower. On the other hand, with smaller amounts of aqueous phase, the reverse
reaction becomes significant due to increased concentration of methanol and
decreased availability of water needed for hydrolysis. There is a significant

interaction term between pH and void fraction. These terms combined, lead to

148

high equilibrium conversion at high pH and low void fractions as shown in the
contour plot (Figure 4.7).

5) Concentration of reactant in the decanol phase (C)

The signs of the various interaction terms (PC, VC) containing the reactant
concentration C as a variable, coupled with the main effect of those variables (V
and P) indicate that low reactant concentrations favor high equilibrium
conversions. However, a lower mass yield also would result due to low

concentrations in the organic phase. Figure 4.8 shows the effect of high

116* " fi’f.:l

concentrations at different void fractions, indicating that if a high mass yield is
desired, high concentration would be favored at the expense of poorer

equilibrium conversion.

149

 

v r I r

1 _- ' ' ‘ 7 ' T r ‘ ' ' ' ' ‘ ' ' ' ' .

l i

r f- .

0.5 - /// .
l 30 .

O - 40 / .

’ 50 j

60 4

» 7O 1

80 ‘

 

*7

l

j,

 

 

 

 

 

 

 

-1 -0.5 O 0.5 1

P
Figure 4.7. Contour Plot for Yc for P and V ranging from -1 to 1, using Equation 1.

150

 

 

 

 

Conversion Yc

 

 

 

 

 

20 25 30 35 40
void fraction (%)

Figure 4.8. Plot of conversion vs. void fraction at varying initial concentrations of
Phe-Me in the decanol phase

151

4.4.3.2. Enantiomeric excess (Yea)

The statistical analysis of the data showed that pH, void fraction and
concentration in the decanol phase are the most significant factors affecting
enantiomeric excess. The results of the statistical analysis are given in Table 4.9.
The enantiomeric excess showed no significant variation with surfactant
concentration and amount of enzyme. The best-fit linear regression model is
shown in Equation 2 with only the statistically significant terms
V.,., = 90.5 + 3.22 v + 9.94 P + 11.1 c - 6.59 P2 - 6.27 02 + 4.36 SE

- 5.62 VP - 2.83 VC - 12.8 PC (Equation 2)

A plot of the residuals vs. the order of experiments is presented in Figure 4.9.
Under the optimal conditions identified for conversion, notably pH value of 6, high

enatiomeric excess (90%) is predicted by equation 2 (Figure 4.10).

152

 

Table 4.9 Regression analysis for the model represented by Equation 2

V.,, = 90.5 + 3.22 V + 9.94 P +11.1 C - 6.59 P2 - 6.27 02 + 4.36 SE
- 5.62 VP - 2.83 vc - 12.8 PC

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Predictor Coef StDev T P
Constant 90.45 1 .007 89.84 0
V 3.2155 0.7729 4.16 0
P 9.9391 0.915 10.86 0
C 11.0504 0.915 12.08 0
P2 -6.5868 0.9519 -6.92 0
02 -6.2651 0.9519 -6.58 0 1
SE 4.362 0.9466 4.61 0 9
VP -5.6166 0.9466 -5.93 0 1
V0 28262 0.9466 -2.99 0.007 4
PC -12.8041 0.9466 -13.53 0
S = 3.786 R-Sq = 96.6% R-Sq(adj) =
95.1%
Analysis of Variance
Source DF SS MS F P
Regression 9 8543.99 949.33 66.22 0
Residual Error 21 301.05 14.34
Total 30 8845.04

 

 

 

 

 

 

 

153

 

 

 

 

 

 

Residuals Versus the Order of the Data
(1
5- ' 0
0 ° 0 o
O O
Q 0
_ . o ' ' .
g 04 ............................. 0-0 ............................ c
D
'g o o '
r: . o . o . o
.5.—
O O . .
I I I T I I
5 1o 15 20 25 30
Observation Order

 

 

 

Figure 4.9 Plot of residuals vs. order of data for Yee.

154

 

' 1 1 v j 1 1 v fi T

1 _. . . . . . . .2 . . .
l .
I .
0.5 . .
l- 60 <
65

0 r 70 J
i

75
i
3 .

. 55
’ O 85 1
-0.5 r 90
-1 _

l 4 J A n L l I

 

‘1

 

 

 

,-1 -0.5 0 0.5- 1

Figure 4.10 Contour plots for Yea, for P and V ranging from -1 to 1, based on
equation 2.

155

4.5 Conclusion

The dynamic resolution of L,D phenylalanine was carried out in a
multiphase system comprising CLA emulsions using crude pancreatic porcine
lipase. A non-ionic emulsifier (Tween 80) used for the formation of CLA
emulsions, exhibited minimal interactions with the biocatalyst. CLA emulsions
provided a four-fold decrease in time required to attain equilibrium over
conventional contacting methods. A statistical-design-of-experiment approach
was used to identify the major variables affecting equilibrium conversion and
enantiomeric excess in presence of CLA. The optimal conditions for high
conversion (>90%) were identified as low void fraction (10%) and high pH (6).
Under conditions that are optimal for high equilibrium conversion, the
enantiomeric excess attained was also high (>90%), thus precluding a tradeoff

between maximizing conversion and enantiomeric excess.

156

 

4.6 References

Dean JA, 1992, Lange’s Handbook of Chemistry, Fourteenth Edition, McGraw-
Hill, lnc., p 8.111

Gal, J, 1987. Stereoisomer separations via derivatization with optically active
reagents, Applications to Compounds of pharmacological Interest, LC-GC 5: 106-
126.

Kroschwitz JI and Howe Grant M, Eds., Kirk Othermer Encyclopedia of Chemical
Technology, 4th Edition, John Wiley and Sons, NY., 2, 514-515.

Kinoshita T, Kasahara Y, Nimura N, 1981. Reversed phase high performance
liquid chromatographic resolution of non esterified enantiomeric amino acids by

derivatization with 2,3,4,6 tetra-O-acetyl- B-D- glucopyranosyl isothiocyanate

and 2,3,4 tri-O-acetyl- 01-D- arabino-pyranosyl isothiocyanate. Journal of
Chromatography, 210 (1) 77-81.

Montgomery DC 1996. Design and analysis of experiments 4th edn John Wiley
and Sons, New York NY

Bonzonana G and Desnuelle P. 1965. Kinetic study of pancreatic lipase on
triglycerides in emulsion: enzyme action in heterogeneous medium. Biochim.
Biophys. Acta., 105, 121.

Bosswell C, 1999. Building fine chemicals muscle with amino acids, Chemical
Market Reporter, New York, NY June 14th.

Brennan MB, 2000. “Drug Discovery: Filtering Out Failures Early in the Game,”

C&EN, June 5, 2000, 63-73.

157

Brockman HL, 1984. General features of lipolysis: reaction scheme, interfacial
structure and experimental approaches. in “lipases,” B. Borgstrom and HL
Brockman (ed.) Elsevier, New York, Chapter.1 p3.

Broxterrnan, QB, WHJ Boesten, J Kamphuis, M Kloostennan, EM Meijer, and HE
Entessangles, B., and Desnuelle, P. 1968. Action of pancreatic lipase on
aggregated glyceride molecules in an isotropic system, Biochim. Biophys. Acta,
159, 285

Jacobsen E, 1999, Assymetric synthesis of alpha-amino acids, Chemical Market

Reporter 255 (24) p FR12-FR13.

Lamb SB, Stuckey DC. 1999.Enzyme Immobilization on colloidal liquid aphrons
(CLAs): the influence of protein properties. Enzyme and Microbial Tech. 24 (8-9)

541 -548.

Lye GJ 1997. Stereoselective hydrolysis of D-L phenylalanine using aphron
imobilized a-chymotrypsin. Biotechnology Letters, 11, 611-616.

Patel, MT, R Nagarajan, and A Kilara, 1996. Lipase Catalyzed Biochemical
Reactions in Novel Media: A Review, Chem. Eng. Comm., 152-153, 365-404.
Rogers, RS, 1999. Companies Turn to Biocatalysis: Drive to meet Customer
Demand for Single-lsomer lnterrnediates Takes on Urgency,” C&EN, July 19, 87-
91.

Schoemaker, 1990. Chemo-enzymatic Production Methods for Optically Pure
lntennediates of Potential Commercial Interest, In: Qpportunities in

Biotransformations, Elsevier Applied Science, New York, 148-169.

158

 

Scott, CD, TC Scott, HW Blanch, AM Klibanov, and AJ Russell, 1995.
“Bioprocessing in Aqueous Media: Critical Needs and Opportunities,” ORNL/TM-
12849, Department of Energy, 19-21.

Stinson, SC, 1998. Counting on Chiral Drugs, C&EN, Sept. 21, 83-104.

Stinson, SC, 1999. Prosperity for Fine Chemicals, C&EN, July 19, 1999, 65-87.

Stinson, SC, 2000. Fertile Ferment in Chiral Chemistry, C&EN, May 8, 59-70.

159

Chapter 5

5. Whole Cell Biotransformations using Colloidal Liquid
Aphrons

5.1 Introduction

The search for selective enzyme inhibitors and receptor agonists or
antagonists is one of the keys for target-oriented research in the pharmaceutical
industry. Increasing understanding of the mechanism of drug interaction on a
molecular level has shown that chirality is the key to the efficacy of many drug
products. It is now known that, in many cases, only one stereoisomer of a drug
is effective, and the other stereoisomer may be inactive, exhibit reduced activity,
or even cause detrimental side effects. Thus, pharmaceutical companies are
aware that many new drugs should be homochiral. In many cases, where the
switch from a racemic mixture to an enantiomerically pure compound is feasible,
there is the opportunity to double the yield of an industrial process (Patel, 1997).

Chiral synthons can also be prepared by asymmetric synthesis using
either conventional catalytic or biocatalytic processes. The advantages of
biocatalytic reactions are that they are stereoselective and can be carried out at
near ambient temperature and atmospheric pressure. This minimizes problems
of isomerization, racemization, epimerization, and rearrangement that generally
occur in conventional catalytic processes. Biocatalysis can be carried out using

either individual enzymes or whole cells. The application of CLA emulsions for

160

 

enzyme based biocatalysis was described in Chapter 4. In this chapter the use of

CLA emulsions for whole cell biocatalysis is demonstrated.

5. 1. 1 Whole Cell Biocatalys is

The majority of reactions of organic-chemical interest require cofactors
such as NAD(P)H or ATP. These cofactors are too expensive to be used in the
stoichiometric amounts required (Jones, 1985). Viable cells can regenerate these
cofactors efficiently in their metabolic cycles, and hence be used as efficient
stereoselective catalysts for such reactions. Organic chemists prefer bakers
yeast among microorganisms for biotransformations (Servi, 1990) because it is
inexpensive, versatile, and is easy to grow to high cell densities. In particular, the
asymmetric reduction of carbonyl compounds by Bakers yeast (Sacchromyces
cerevisiae) has been widely used for preparing optically active alcohols largely
because the reaction is easy to carry out and Baker’s yeast is commercially
available in large quantity (Brooks, et al. 1982; Deol, et al. 1976; Meyers, et al.
1980). Several review articles devoted to the use of baker’s yeast in organic
chemistry have been published (Jones, 1986; Simon et al. 1985; Van
Middlesworth and Sih, 1987), and over 300 references to scientific works in this
area have been reported (Servi 1990).

Development of yeast biotransformation processes suitable for
implementation by the pharmaceutical industry presents significant engineering
challenges. The two main issues involved with these tranforrnations are:

1) Low aqueous solubility of substrates

161

 

Ideal drug candidates would be hydrophobic to allow passage through cell
membranes (Brennan, 2000). Hence, many relevant biotransformations involve
substrates that are sparingly soluble in water. Biocatalysts, such as
microorganisms, are designed by nature to be effective in an aqueous medium.
This results in low volumetric productivities in biocatalytic processes involving
sparingly soluble substrates. In addition products streams are dilute and present
challenges in downstream processing. {1
2) Competing reactions in yeast biotransformations Ill
The presence of several intracellular enzymes that may compete for the
same substrate may result in the formation of more than one product, leading to
low selectivities. For instance, of the more than one thousand six hundred
enzymes listed in the Handbook of Enzymes (Barman TE, 1985), several
hundred have been purified from yeast and probably more are present in the
organism. Yeast can catalyze the asymmetric reductive biotransformations of a
variety of compounds containing a carbonyl group or carbon-carbon double
bond. Oxidoreductases participating in these reactions have a relatively broad
substrate specificity (Ward, et al. 1990). However, other side reactions such as
acyloin condensation and Michael type of addition of a carbon nucleophile have
also been reported (Lintner and Von Leibig, 1913, Oldrich H et al., 1986). Also,
the presence of another group of enzymes (lyases), has been reported to cause
water addition across the double bond of an 01,8 unsaturated aldehydes such as

crotonaldehyde (Fronza et al, 1988). Since most enzymes present in yeast will

162

display their properties under Similar conditions, the problem of low selectivity, in

biocatalytic processes involving yeast, is a major problem (Servi, 1990).

5. 1.2 Organic solvents in whole cell biocatalysis

There is much current interest in conducting biotransformations in the
presence of organic solvents (Brink and Tramper, 1985; Adlercreutz and
Mattiason, 1987; Laane et al., 1987a,b; Brink et al., 1988; Zaks and Klibanov,
1988). The organic solvent may serve as a reservoir for a hydrophobic substrate.
It may also shift the equilibrium conditions in order to improve selectivity.

However, the presence of a nonpolar liquid phase can create processing
difficulties. First, interphase mass transfer of sparingly soluble substrates is
typically slow. Yeast biotransformations are often carried out using high cell
densities (~250 g/L) and even cell slurries. At high cell densities the substrate
consumption capacity can far surpass mass transfer capacity. Mechanical
agitation can be used to disperse the solvent phase and thereby increase the
interfacial area and mass transfer. However, this approach is energy-intensive
and thus costly, particularly in large-scale systems (Bredwell, et al. 1998).
Second, many organic solvents are toxic to cellular biocatalysts. This toxicity can
either arise from the action of the trace amount of solvent that is dissolved in
water, or from direct contact between the cells and the organic phase. The latter

has been termed "phase" toxicity (Salter and Kell, 1995).

163

5. 1.3 Colloidal Liquid Aphrons

As shown in Chapters 2 and 3, CLA provide a 5-15 fold increase in interphase
volumetric mass transfer rates over conventional emulsion systems even without
the need for energy intensive agitation. They can be formed at low Shear rates
(Chapter 1) and pumped into the reactor as predispersed droplets of organic
solvent. Scott et al. (1995) have suggested that cells inhibited in the presence of
an organic phase may be at least partially protected by immobilization or the use
of membranes. The presence of a lipid-like liquid crystalline "shell" stabilizing the
CLA could prevent direct contact between cells and the solvent, thus reducing
phase toxicity effects. Thus, CLA emulsions have potential to enhance whole
cell multiphase bioconversions.

This research described in this chapter had two main objectives

1. To demonstrate the use of CLA produced using biocompatible organic
solvents in whole cell biotransformations involving high cell densities.

2. To determine whether selectivity of whole cell biocatalysis can be influenced
through the use of CLA emulsions and explore mechanisms for the variation

in selectivity (if any).

5. 1.4 Model system

Carbonyl reductions are probably the most thoroughly studied and
exploited biotransformations performed by Bakers yeast (Ward et al, 1990). In
this chapter, the biotransformation of an 01,8 unsaturated carbonyl compound 1
was studied (Figure 5.1). Baker’s yeast fed D-glucose converts CS 0.8
unsaturated aromatic

164

 

CHOOH
0V 2

 

 

. F'— ...-.4_.____,_

Figure 5.1. Biotransformation of cinnemaldehyde by baker's yeast.

165

aldehydes into corresponding 3-phenyl-propan-1-ols (2) and 3-phenylprop-2en-1-
ols (3) and into the 4 substituted (2S, 3R)-5-phenylpent-4-en,2,3-ols (4) (Fuganti
and Grasselli, 1985)

The formation of (2) and (3) is interesting from a mechanistic point of view
towards understanding and controlling selectivity whereas the production of (4),
which contains two more carbon atoms than the precursor aldehyde and two
chiral centers of the type R,R’CHOH, is interesting from both the mechanistic and
synthetic points of view. The (ZS, 3R)-diol possesses unique structural features
that make it useful as starting material in the synthesis of enantiomerically pure
forms of natural products (Fuganti, 1986). The (28,3R) configuration matches the
configuration at positions 5 and 4 of 6-deoxysugars of the L-series like L-
amiacetose 5 (Figure 5.2). The double bond of the diol can be stereospecifically
functionalized, the stereochemistry of the process being dictated by the
stereochemistry of the adjacent allylic center. Once the double bond has been
saturated it is possible to functionalize the derived products regioselectively at
the benzylic position. In saturated products, degradation of the benzene ring with
ozone gives 6-deoxyhexonic acids, which bear all the chiral centers present in
the side chain of the parent compound. From suitably protected forms of the diol
it is possible to extrude C4, Cs, Ce carbonyl compounds with ozone. The erythro
forms of these compounds can be converted to 01 epimers, with threo
stereochemistry, on treatment with base. This diol can be considered as
carbohydrate like chiral synthon, which can sometimes be used advantageously

as an alternative to natural carbohydrates. For example, it may be used in the

166

 

synthesis of 2,3,6 tri-deoxy-3 amino hexoses of the L- series, which are present
in the therapeutically important anthracycline glycosides adriamycin (doxorubicin)
and 4’-adriamycin. The conventional route using D hexoses requires a critical
inversion of the configuration at position 5 at some stage in the sequence which
can be avoided using the 28,3R diol as the starting compound.

Other products synthesized from the diol are shown in Figure 5.3.

" ......

OH

O
H 5
4

Figure 5.2. Structure of L- amiacetose

167

 

 

 

 

f
6 H 0“
A.JV
O ‘ """ Q 4
OH
H t

Figure 5.3. Compounds synthesised from (4)

168

5.2 Experimental Methods

5.2. 1 Determination of solve nt biocompatibility

Five different experiments conducted under identical conditions (50 mL
water, 12.5 mL yeast slurry, 28°C in 125 mL Erlenmeyer flasks, shaken in a
horizontal shaker at 220 rpm), were started simultaneously, and 10 mL of organic
solvent were added to each flask to screen for biocompatibility. The organic
solvents tested were hexane, benzene, toluene, oleyl alcohol and 2,2,4-
trimethylpentane. All solvents were HPLC grade and were obtained from Sigma
Chemical Company (Milwaukee, WI). At time intervals of 30 min, 1 mL samples
were centrifuged at 4000xg for 2 min. Ten microliters of the clear, cell free
supernatant were added to 1 mL of glucose (Trinder) reagent (Sigma Chemical
Company, Procedure 315). Each sample was incubated for exactly 18 min and
its absorbance measured at 505 nm with a PE Lambda 3A UVNIS
spectrophotometer (Perkin Elmer Corp., Norwalk, CT). Conclusions about
biocompatibility of each organic solvent were drawn by comparing the glucose
concentration profile in each flask to that of a control experiment without organic

solvent.

5.2.2 Preparation of Colloidal Liquid Aphrons

Two surfactants are typically used for the preparation of CLA emulsions.
One of the two partitions primarily in the aqueous phase, and the other partitions
primarily in the organic phase. The surfactant used in the aqueous phase was

Tween 80 (Sigma Chemical Company, Milwaukee, WI). The surfactant in the

169

-r:

organic phase was a linear alcohol ethoxylate Tergitol 15-S-3 (Union Carbide
Corp., Danbury, CT). The CLA preparation conditions influence the size
distribution and stability of the CLA phase. Consequently, a consistent
preparation procedure was used. The CLA emulsion was prepared in a 50 cm3
glass beaker with an lntemal diameter of 40 mm. The solvent (2,2,4 trimethyl
pentane)/Tergitol (0.5%) solution) was added to the aqueous phase (Tween 80
(2%)) at a rate of 0.5 mL per min, while stirring at 1600 rpm using a 25 mm
magnetic stirrer bar. The resulting CLA phases were creamy white, with a phase
volume ratio (volume of organic phase/volume of aqueous phase) of 1. They
were very stable with no phase separation evident over a period of 4 months.
One hundred milligrams of Cinnemaldehyde (1) was dissolved in the organic

solvent prior to the formation of CLA dispersions.

5.2.3 Biotransformation Methods

Trans-cinnemaldehyde (3-Phenyl-2-propenal) of the highest quality
available was obtained from Aldrich Chemical Company (Milwaukee, WI). A 250
mL Erlenmeyer flask containing 50 mL tap water and 2.5 g Glucose was
inoculated with 12.5 g of bottom fermented Saccharomyces cerevisiae (Harper’s'
Brewing Company, East Lansing, MI). The culture was grown in a controlled
environment shaker (New Brunswick Inc, NJ) at 28°C, with 220 rpm shaking for 2
h; then the biotransformation was started by adding a known amount of substrate
1 (either dissolved in the organic phase of CLA or pure). The biotransformations
were performed for 3, 8, 12 and 24 h. Whenever the glucose concentration

dropped below 10 g/L, 2.5 g of glucose were added.

170

 

In some experiments, the substrate was immobilized onto a hydrophobic resin
before being fed into the flasks. The resin, 250 mg of Amberlite XAD-4 (Supelco,
Bellefonte, PA) was washed with de-ionized water (3 mL per 1 mL of resin) and
acetone (3 mL for 1 mL of resin). The substrate was dissolved in acetone, and
the resin, once dried, was added to this solution. The mixture was shaken for 2
min, and the acetone was evaporated. The resin so obtained was poured into the
biotransformation flask.

The substrate (1) as well as all the biotransformation products, are
nonpolar organic compounds that can be separated from the biotransformation
medium by extraction with an organic solvent , e.g. chloroform (CHCI3). After
fermentation the cells were centrifuged for 5 min at 4000 xg, and the cell-free
aqueous supernatant was extracted under vigorous shaking for 2 min with 40 mL
of CHCI3'. Once again the mixture was centrifuged for 5 min at 4000 xg. The
solvent was removed by evaporating the clear organic bottom phase at 75°C
under vacuum for 10 min. The oily product mix was dissolved in 1.5 mL of

deuterated chloroform (CDCI3) for 1H-NMR analysis.

5.2.4 Analytical procedures

All spectra were recorded using 8 1H-NMR INOVA 300 (Varian Inc., Palo
Alto, CA) spectrometer operating at 300 MHz with a sample temperature of 200C;
the sample volume was 600 pl. Samples were recorded in CDCI3 solutions
containing succinic anhydride [‘H-NMR 6 (CDCI3) 2.9 (4H, CH2, d)] as an internal

standard of known concentration.

171

All product peaks in the NMR spectra were identifiable from published
data (Pouchert et al., 1993), except 5-phenyl-pent-4-ene-2, 3 diol. (4). One-
dimensional total correlation spectroscopy (TOCSY) was performed on this
compound to confirm its structure. Irradiation of the quartet of doublets
(multiplet)(J=6.5, J=3.6) peak at 3.97 ppm and observing the TOCSY spectrum
at increasing mixing times indicated that the peak at 3.97 ppm was directly
bonded to the methyl doublet (J=6.5Hz) at 1.19 ppm. No further relays were
observed in this direction. From the peak at 3.97, a relay was observed to quartet
of doublets (J=7.15, J=3.6, J=1.01) 4.25 ppm and further to the doublet of
doublets (J=16.24, J=7.0) at 6.27 ppm and further to the doublet of doublets
(J=16.0, J=1.0) at 6.67 ppm from which no other relays were observed.

A typical NMR spectmm is shown in Figure 5.4.

172

 

 

 

'Y '.1VY\ [IVY .‘Y' 1"1'rfif'

4.20 4.10 4.00

H, Li

a l J
J M OJ ‘
a. W “~—
‘IVI Ut‘Tr ‘ ‘IY' r. I~Ir 'Ifll TIT! I I7! r- I

W U " U ‘ Ii} 111' - l W 'I ri' j I i l U I '7‘ I r r 1 '

 

 

 

 

 

 

 

 

4.4 4.3 4.2 4.1 4.0 3.9 3.8

 

 

 

Figure 5.4. Typical NMR spectrum of biotransformation products

173

5.3 Results and discussion

5. 3. 1 Biocompatibility

The solvents selected for the biocompatibility experiments were oleyl
alcohol, hexane, benzene, toluene and 2,2,4-trimethylpentane. Oleyl alcohol has
been widely accepted as a biocompatible solvent for extractive ethanol
ferrnentations (Bruce and Daugulis, 1991). However, since the products of this
biotransformation were non- volatile, product recovery considerations mandate
the use of a volatile solvent. Among commonly used solvents, hexane, benzene,
toluene, and isooctane were selected due to their high volatilities and low
solubilities in water. Figure 5.5 shows the glucose utilization rate by yeast in the
presence of organic solvents (20% v/v). Among the solvents studied oleyl alcohol
and 2,2,4 trimethylpentane (also referred to as isooctane) exhibited minimal
toxicity on yeast. Because of its favorable boiling point (992°C at 105 Pa) that
allows easy solvent removal in the recovery process and its low water solubility
(0.244 mg/100 9 water), 2,2,4 trimethylpentane (isooctane) was chosen as the
organic solvent for the biotransformation experiments. Using ‘H-NMR, the
solubility of cinnemaldehyde was determined to be 30 g/L in 2,2,4 *

trimethylpentane. However keeping in view the inhibition of the cells in

174

O)
O
O
O

 

‘5”

 

 

‘5’

 

 

\\
\

 

 

Glucose concentration(mgl100ml)

 

 

 

 

3000
2000 \ \
1000 \
0 . 1
O 50 100 150 200 250 300
Time (min)
I - H
I + isooctane -l- oleyl alcohol + benzene
-I- toluene -O- hexane + control

 

Figure 5.5. Biocompatibility of various solvents

175

presence of the solvent, the ability to use a large amount of organic solvent is

limited.

5.3.2 Biotransformations in presence of CLA

The biotransformation experiments were performed with varying amounts
(0-20 mL) of CLA emulsions present in the flasks. Over 95% of the
cinnemaldehyde was converted after 3 h, in all experiments. The product
compositions resulting from varying amounts of CLA are shown in Figure 5.6.
The fraction of (2R, 3S) diol in the product increases as the volume of CLA
emulsions increases, confirming that the selectivity of the biotransformation can
be modified through the use of CLA. The yield of the diol obtained using CLA
was three-fold higher than that in the absence of CLA.
Substrate concentration is known to be a major factor affecting the selectivity of
microbial reductions (Wipf et al. 1983). The presence of the organic phase in the
form of CLA dispersions could cause partitioning of the substrate between the
phases, thus changing the apparent substrate concentration and the selectivity.
Further experiments were conducted to identify whether the changes in

selectivity could be attributed to the effect of substrate concentration.

176

25

 

01
O

 

.....
.....
.....
.....
.....
.....

 

 

E Diol

A
01

I
.....
.....
II...
ttttt
a...“
l...-

 

 

l Volume of CLA

 

 

 

 

 

oooooooo
nnnnnnnn
cccccccc
00000000
........
nnnnnnnn
oooooooo
llllllll
........
........
on .....

 

--------
0000000000000000

 

........
...-”o.
0......-
........

 

IIIIIIII
cccccccc
IIIIIIII
IIIIIIII
nnnnnnnn

 

ION

00000000
...“...
........
........
oooooooo
........
........
oooooooo
oooooooo
........
........
........
........
........
oooooooo

000000000
oooooooooooooooo
nnnnnnnnnnnnnnnn
........
oooooooo
........
oooooooo
........

--------
ooooooooo
eeeeeeee
cccccccc
........
nnnnnnnn
oooooooo
........

oooooooo
cccccccc
oooooooo
oooooooo
........
cccccccc
.........

........
00000000
I" .....
........
IIIIIIII
........
IIIIIIII
IIIIIIII
IIOIICII nun-“us
Ciuouun
llllllll

% dlol In final product
_. _. w 00 is
O 01 O 01 O 01 O 01 O
L, I .1 l

 

1 2 3 4 5 6
Experiment #

Figure 5.6. Effect of CLA volume on % diol in product

177

O”
a.
I.
I.
M
I.
O.
I.
n.
‘-
-------- on
I O
nnnnnnnn
........
oooooooo
oooooooo
I...
oooooooo
OOOOOOOO
0.0"...
not..."
D. 0....

0......-
no“...
........
........

...-0...
0.00.”.

........
IIIIIIII
.....
........
........
u mo...
........
IIIIIIII
........
........
........
........

om

...-..

::23:23:3232¥833§83: 3::::

~20

.15

~10

 

(1111) we
to alUI’IlOA

5.3.3 Mechanisms for changing selectivity of whole cell biotransformations

5.3.3.1. Effect of concentration on product composition

The initial amount of cinnemaldehyde (1) added was varied from 6.6 mg to
132 mg per 50mL water in the biotransformation flask. Using 1H-NMR
cinnamaldehyde solubility was determined to be 1.0 g/L in water. At initial
concentrations of (1) above 100 mg per 50 mL total-batch volume, product (2)
was almost 50% of total product moles. This value rose to 100% when the initial
amount of (1) was below 32 mg (Figure 5.7). This trend can be explained by
competition for the same substrate between enzymes with different Km and Vmax
values. However, in all the experiments, the yield of the diol (4) was low (<15%)

compared to the experiments conducted in the presence of the CLA emulsions.

5.3.3.2. E fiect of time on product composition

Biotransformations were performed at different initial substrate
concentrations (6.6, 33, 66, and 132 mg of (1) per 50mL water in flask) and the
percentage of 3-phenyl-propanol (2) was determined after 8, 12 and 24 h,
respectively. Results are shown in Figure 5.8. Initial cinnemaldehyde amounts
less than 150 mg were completely converted after 3 h of fermentation. The
product profile remained constant after 3 h, indicating that products 2,3, and 4

are not interconverted to a significant degree.

178

 

 

 

\/
./

0 ‘ 1 T
0 50 100 150
initial amount of cinnemaldehyde in mg

Figure 5.7. Change in product composition with initial concentration of

cimmemaldehyde. Products represented are ( I )3 phenyl 1 propanol (2), ( C) )
cinnamyl alcohol (3), and (A) 2R, 38 Diol (4).

 

/
I

 

% in the final product mix.

 

 

 

179

 

 

 

 

 

 

1004 + 4 2
m-
fl
w.’>*<:
: A
E 705
D.
‘6 6°“
8 50—
.9
5 401
e 3
3 30- '
..-/M
I .
10
0 f I I l
3 6 9 12 15 18 21 24
time(h)

Figure 5.8. Effect of time on product composition. Plotted are data at different
initial substrate (1) concentrations (9)6.6 mg, (I) 33mg, (A)66mg, (A)132 mg
per 50ml.

180

5.3.4 Selectivity control using hydrophobic resins

Adding the starting material (1) adsorbed onto a resin would be expected
to lower the concentration of the substrate in the broth (due to substrate
partioning between the resin phase and the aqueous phase) and change
selectivity based on the experiments shown in Section 5.3.3.1.
Biotransformations were done with the same amount of substrate (100 mg);
either in pure form or adsorbed onto 250 mg resin. The final product composition
after 3 h is shown in Figure 5.9. This composition corresponds to what would be
expected if the concentration in the broth was below 32 mg per 50 mL water.
Thus it may be concluded that the presence of the resins lowered the apparent
concentration in the broth, resulting in a change in selectivity towards (2), which
is the favored product at low substrate concentration.

5.4 Conclusion

This study has demonstrated the use of CLA emulsions in the
biotransformations of cinnemaldehyde by Baker's yeast at high cell densities
(2509/L). The solvent 2,2,4 trimethyl pentane was identified as a suitable
biocompatible solvent for yeast biotransformations. The presence of the organic
phase in the form of CLA emulsion increased the fraction of 4 substituted (2S,
3R)-5-phenylpent-4-en,2,3-ols (4) by almost three-fold. The selectivity of the
yeast biocatalyst appears to depend on substrate concentration. The change in
selectivity due to the presence of hydrophobic resins is consistent with this
hypothesis. However the variation in selectivity in presence of CLA, is not

attributed to this phenomenon. More studies are required to clarify the

181

mechanism by which the presence of organic solvents affects selectivity.

182

 

 

I Product 2 [3 Product 3 70 ~

:1}; PrOdUCt 4 60 L

 

 

 

 

% in final product

 

 

 

 

 

 

 

 

 

 

100 mg of (1) 100 mg (1) +250 mg resin

Figure 5.9. Effect of hydrophobic resins on product composition

183

5.5 References

Adlercreutz and Mattiason, 1987. Aspects of biocatalyst stability in organic
solvents, Biocatalysis 1, 99-108

Barman TE, 1985. Enzymes handbook, Springer Verlag, Berlin.

Brink et al., 1988. LES, Tramper J, Luyben K Ch. A. M. and vant riet K:
biocatalysis in organic media. Enzyme Microbial technology, 10, 736-743
Brink LES and Tramper J, 1985. Optimization of organic solvent in multiphase
biocatalysis. Biotechnol. Bioeng. 27, 1258-1269.

Brooks,D.W. ,Grothaus,P.G.,and Irwon,W.L.(1982).J Org. Chem. 47,2820.
Deol, B. S., Ridley, DD, and Simpson, G. W. (1976) Aust. J. Chem. 29, 2459
Fronza G, Fuganti C, Grasselli P, Poli G, Servi S, Zorzella, A. (1988). J. Org.
Chem. 53, 6153

Fuganti C, 1985, Bakers Yeast mediated preparation of carbohydrate-like
synthons, in Enzymes as catalysts in organic synthesis, ed. M.P. Schneider,

Kluwer Academic Publishers, NOI‘WS", MA, USA. 3-17.

Fuganti, C.; and Grasselli, P.; in Enzymes in Organic Synthesis, Pitman London
UK, 1985, pp. 112-127.

Jones JB, 1986, Tetrahedron, 42,3351

Laane C, Boeren S, Vos K., and Veeger C. ,1987b . Optimization of biocatalysis
in organic media, In Biocatalysis in Organic media Eds Laane C, Tramper J and

Lilly MD., Elsevier, Amsterdam, 65-84

184

Laane C, Tramper J and Lilly MD, 1987a. Eds. Biocatalysis in organic media,
Elsevier, Amsterdam,

Lintner CJ and Von Leibig HJ, 1913, Z. Physiol. Chem. 88,109

Meyers, A. |., and Amos, R. A. (1980) J. Amer. Chem. Soc. 102, 870

Oldrich H, Kakac B, Zvacek J, Tuma J, 1986. Czech Patent 231,401
(C1.C12Pz/24); C.A. 1987,106,3871.

PateI,R.N., Stereoselective Biotransformations in Synthesis of some
pharmaceutical intermediates, Advances in Applied Microbiology,(1997),43,91-
140

Pouchert, c.; Behnke, J.; The Aldrich library of ’3Cand 1H FT-NMFI spectra, 1993,
Aldrich Chemical Company,Milwaukee,Wl.

Scott CD, Scott TC, Blanch HW, Klibanov AM, and Russel AJ 1995.
Bioprocessing in nonaqueous media: Critical needs and opportunities.
ORNL/TM-12849, US. National Technical Information Service, Springfield, VA.
Simon H, Bader J, Gunther H, Neumann S, Thanos J 1985, Angew. chem. int.
Edn. Engl., 24, 539

Van Middlesworth F, Sih CJ 1987. Biocatalysis 1, 117

Wang, KC, and Sih, C.J. (1963) Biochemistry 2, 1238.

Ward, O.P.; Young, C.S.; Enzyme Microbial Technol. 1990,12, 482-493.
Ward,O.P., and Young C.S., Reductive biotransformations of organic compounds
by cells or enzymes of yeast,Enzyme Microb. Technol., 1990, 12,482-493.

Wipf, B.; Kupfer, E.; Bertazzi, R.; Leuenberger, H.G. W. Helv. Chim. Acta.
(1983). 66,485.

185

Zaks A and Klibanov AM, 1988. J. Biol .Chem. 263, 3194-3201.
Zhou, B, Gopalan, A.S.; VanMiddlesworth, F.; Shieh, W.R,; Sih, C.J., J. Am.

Chem. Soc. (1983), 105, 5925.

186

Chapter 6

6. Reversible emulsions

6.1 Introduction

Experimental evidence for the structure of CLA, and measurements of
volumetric mass transfer coefficients for CLA in agitated vessels, has been
presented in Chapters 1-3. Results presented in Chapters 2 and 3 showed a 5
to1510ld increase in mass transfer rate using pre-dispersed CLA compared to
conventional liquid dispersions. However, the high stability of the CLA emulsions
makes phase separation, upon completion of the mass transfer, difficult.
Separation of CLA from dispersion and their reconcentration can be effected by
micro-filtration or centrifugation (Rosjidi et al., 1994). However, these processes
can be energy intensive and substantially impact the economics of the extraction
process.

Recently Mathur et al. (1998) have developed a novel class of co-
polymeric block/graft emulsifiers whose emulsification properties are lost in
response to small, predetermined changes in pH. The reversible stabilizer is a
random-substitution copolymer of methacrylic acid (MAA) and methoxy-
poly(ethylene glycol) methacrylate (MPEGMA). When mixed in the appropriate
ratio the MAA forms a continuous linear polymer “backbone" into which the
MPEGMA methacrylate groups are inserted randomly. The resulting comb-type
co-polymer has a MAA backbone with poly(ethylene glycol) (PEG) side chains. In
common with other polyacids such as poly(acrylic acid), the MAA backbone is

187

highly soluble in water at high pH, since its acid group are dissociated, and is
less soluble at low pH (undissociated state). The PEG side chains are only
slightly less soluble at low pH. However, when in close proximity, the hydroxyl
groups of the PEG side chain form a hydrogen-bonded complex with the
carboxylic acid groups of the MMA backbone, causing the side chains to fold

against the backbone as illustrated in Figure 6.1.

H OH OH OH OH 0
=0 C=O C=0 C=O C=O C=0

‘1’
c

 

Figure 6.1 Complex formation in MAA-MPEGMA copolymers (redrawn from
Mathur (1998)).

188

 

6.2 Materials and Methods

The block/graft copolymers are synthesized via a free-radical solution
polymerization of methacrylic acid (MAA) with methoxy poly(ethylene glycol)
methacrylate (MPEGMA) using hydrogen peroxide (30% aqueous solution) as
initiator and a 50/50 w/w mixtUre of ethanol/water as the solvent in a Single step
process. By choosing the ratio of methacrylic acid to ethylene glycol repeat units
it is possible to form pseudo-multi-block copolymers consisting of a hydrophilic
acid backbone interspersed with hydrophobic segments due to complexation.
Such a polymer behaves like a multi-block polymeric emulsifier under acidic
conditions and as a hydrophilic graft copolymer under basic conditions. Thus this
reversible emulsification behavior can be repeatedly tumed "on" or "off" by
controlling the pH or other variables that affect complexation. As a result, the
emulsions produced by using these novel polymeric emulsifiers can, in principle,
be maintained stable as long as desired and then coalesced on demand by a
small pH change. This study investigated the reversibility and mass transfer

properties of emulsions formed using these novel emulsifiers.

The reversible emulsifier was prepared using the method of Mathur et al
(1998). To produce an emulsifier with a 10:1 repeat ratio of MAA to EG: 0.55 g of
MPEGMA 1000 was combined with 9.45 g of MAA, containing hydroquinone
inhibitor, in 509 of a 50/50 mixture (v/v) of ethanol and water in a clear glass
vessel. The mixture was briefly agitated by hand and then purged with a slow

nitrogen feed for 5 min. The vessel was then capped with a silicone septum and

189

crimp cap under nitrogen gas. Fifty pL of 30% hydrogen peroxide was then
added by injection through the septum, and the needle was left in place in the
septum to act as a vent. The vessel was then placed in a temperature-controlled
oven at 70°C for 3 days, with occasional agitation. The polymer was poured from
the vessel and the solvent left to evaporate. For a repeat ratio of 20:1, the same
procedure was followed with 0.275 g of MPEGMA and 9.725 g of MAA. This
procedure resulted in a very viscous solution of polymer at the end of the three-
day period (regardless of the ratio used). The emulsifiers synthesized contained

20:1, 15:1 and 10:1 ratio of the MA to EG repeat units.

The reversible emulsifiers were used as the stabilizer in the aqueous
phase, and Tergitol 15-S-3 was added as the surfactant in the organic phase to
lower the interfacial tension. Emulsions were prepared in a 50 mL glass beaker
with an internal diameter of 40 mm. Five mL of Limonene/Tergitol (0.1% v/v)
solution were dispensed at a rate of 0.5 mL/min into a 5-mL volume of
aqueous/PMAA-PEG block graft emulsifier (0.4% v/v) solution. A 20 mm
magnetic stirring bar spinning at 1600 rpm was used to provide agitation. The
size distribution of the CLA dispersion was measured using a Malvem
Mastersizer X. Mass transfer coefficient measurements were performed as

previously reported using a conductivity probe (Chapter 2).

Coalescence properties of these emulsifiers were characterized by the
percentage of organic phase recovered vs. pH. Emulsifier solutions (0.4% w/v)

were titrated with 0.1 N NaOH solution to find the amount of NaOH required to

190

 

change the pH of the emulsion (PVR 1). Twenty mL of emulsion (PVR 1) was
stirred in a beaker for 30 min after addition of the appropriate amount of NaOH to
change the pH to a desired value. The volume of organic phase recovered was
recorded after 30 min. The experiment was repeated for emulsions stabilized by
both 20:1 and 15:1 emulsifiers.
6.3 Results

The 10:1 emulsifiers could not form stable emulsions with Tergitol 15-S-3
perhaps because of low hydrophilicity of these emulsifiers. However the 20:1 and
15:1 emulsifiers, being more hydrophilic, formed emulsions spontaneously. The
Size distribution measured for emulsions formed using the new emulsifiers and
their comparison with sizes reported previously (Chapter 2) is shown in Table
6.1. The size data obtained indicate that the average size is about 2 fold larger
than previously reported for CLA formed using Tween 80. The higher the
MAAzEG repeat ratio, the more hydrophilic the emulsifier. Hydrophilicity aids in
partitioning of the stabilizer in the aqueous phase, thus permitting better steric
stabilization. Volumetric mass transfer coefficients measured for emulsions
formed using reversible emulsifiers are similar to those measured in conventional
emulsifier systems (Chapter 2). The comparison of measured KL values to those
measured for CLAS made using Tween 80 as the aqueous surfactant is shown in
Table 6.2. The volumetric mass transfer coefficient KLa is about half that for CLA
emulsions. However this still represents a 4-fold increase over conventional liquid

liquid systems (29 h") at the same Reynolds number (N95).

191

Table 6.1. Size distribution and mass transfer coefficients obtained for reversible
emulsions and CLA.

 

 

 

 

 

Aqueous phase surf. Tergitol Diameter KLa (h'1) KL (cm/s)x103
concentration conc. (um)
0.4% (20:1) 0.1 43.9 126 14.4
0.4% (15:1) 0.1 81.7 53 11.2
0.4 % Tween 80 0.1 16.8 280 12.1

 

 

 

 

 

 

The titration curves for the two emulsifiers are presented in Figure 6.2.
Figure 6.3 shows that after 30 minutes 95% of the oil phase was recovered from
reversible emulsions formed using 20:1 emulsifiers at a pH of 8. At a pH of 9
about 80% of the oil phase was recovered from emulsions formed using the 15:1
emulsifier. Comparatively, less than 5% of the oil phase could be recovered from
CLA formed using Tween 80 and Tergitol 15-S-3 at a pH of 10. Thus, the
reversible emulsifiers hold considerable promise for applications where rapid
dispersed phase recovery is desired. In order to undertake further studies on the
reversible emulsions systems, the interfacial characterization of these reversible
emulsions is essential. Phase behavior of several other polymeric emulsifiers is
available in the literature. Hill et al. (1994) have characterized the L2, L3 and La
phases exhibited by four trisiloxane surfactants using small angle x-ray scattering
(SAXS), small angle neutron scattering (SANS) and differential interference
contrast microscopy. He et al. (1993) have studied the microstructure of

polyoxyethylene trisiloxane surfactants in water and Showed the presence of

192

hexagonally packed cylindrical micelles, using cryo-TEM, SAXS and SANS.
However, before such studies can be undertaken current methods of
polymer synthesis have to be improved. The polymers obtained by the current
methods are highly polydisperse (polydispersities ranging from 2-5)(Drescher et
al., in press). High polydispersities would significantly alter the aggregation
behavior of the polymers making the interpretation of such results difficult. It is
suggested that further work on the reversible stabilizer aggregation behavior be
undertaken after further work on polymer synthesis and processing for this
polymer class, so that a more practical and well-behaved system can be
developed.
6.4 Conclusions

Preliminary studies indicate a high reversibility of emulsions formed using
the 15:1 and 20:1 emulsifiers. Mass transfer studies indicate fourfold higher
volumetric mass transfer coefficients for such emulsions over surfactant free
systems. However, these rates are three- fold lower than those reported for CLA
(Chapter 2). The ease of coalescence due to the unique molecular architecture
involves a tradeoff in mass transfer enhancement compared to CLA. The
reversible emulsifiers hold considerable promise for applications where rapid

dispersed phase recovery is desired.

193

 

 

 

1 0
o 20:01 P(MAA-g-EG) copolymer

 

 

9‘ .15:01 P(MAA-g-EG)copolymer i 1* u

 

 

 

 

 

 

 

 

0 500 1 000 1 500 2000 2500 3000
volume of 0.1M NaOH added full

 

 

 

 

 

 

Figure 6.2. Titration curves for block copolymeric emulsifiers

194

 

I 15:01
E] 20:01
Tween 80 (CLA)

 

 

1 00%
90%
80%
70%
60%
50%
40%
30%
20%
1 0%

0%

 

Percent Recovery of Organic
phase

 

9 10

7 8

pH

Figure 6.3 Percentage recovery of organic phase vs. pH for reversible emulsions,
formed using different polymeric emulsifiers, and CLA emulsions.

195

6.5 References

Rosjidi M., Stuckey D.C. and Leak DJ 1994. Downstream Separation of chiral
epoxides using colloidal liquid aphrons( CLA), in Separations for Biotechnology,
Ed by Pyle D.L. , SCI publishing. 280-286.

Drescher, Bernhard, A.B. Scranton, and J. Klier, 2000. Synthesis and
Characterization of Polymeric Emulsifiers Containing Reversible Hydrophobes:
Poly(methacrylic acid-g-ethylene glycol), Submitted to Polymer (in press).

Arvind M. Mathur, Bernhard Drescher, Alec. B. Scranton and John Klier, Nature,
392(1998) 367. I

Hill RM, Mengtao He, Davis HT, and Scriven LE. 1994. Comparison of the liquid
phase behavior of four trisiloxane supervvetter surfactants. Langmuir, 10, 1724-
1734.

He M, Hill RM, Lin Z, Scriven LE, and Davis HT, 1993. Phase behavior and
microstructure of polyoxyethylene trisiloxane surfactants in aqueous solution.

J Phys.Chem. 1993, 97, 8820-8834.

196

Chapter 7

7. Summary

The interfacial structure and mass transfer properties of CLA emulsions
were investigated in this research. Applications of these emulsions in multiphase
biocatalysis were also developed.

Chapter 1 dealt with the characterization of the interfacial structure of CLA
emulsions using FF-TEM, DSC, SAXS and ITC. Freeze fracture TEM
demonstrated the lamellar interface formed in CLA emulsions. The formation of a
rigid interfacial film, due to the interaction of the two unlike emulsifiers (Tween 80
and Tergitol 15-S-3) was found to be responsible for the stabilization of these
emulsions. Differential scanning calorimetry results showed that this interfacial
film forms a separate phase, apart from the aqueous and organic phases present
in the emulsion. lsotherrnal titration calorimetry was used to characterize the
energetics and stoichiometry of interaction between Tween 80 and Tergitol 15-S-
3. Similarities and differences between properties and formulations of HIPE and
CLA emulsions were elaborated. A model of lamellar micelles stabilizing the CLA
interfaces was proposed based on the geometric theory of surfactant packing.
Properties observed in the CLA literature were explained based on this model.

The mass-transfer properties of CLA emulsions of limonene in water were
measured experimentally and presented in Chapter 2. The CLA size distribution
was found to be strongly affected by the aqueous-phase surfactant (Tween 80)

concentration, but virtually unaffected by the nonpolar-phase surfactant (Tergitol-

197

 

15-S-3) concentration. The d32 values decreased from 49 to 14 pm as the
Tween 80 concentration increased from 0.02 to 0.4% v/v. The KL values
measured for CLA emulsions averaged about 15 x 10'3 cm/s and were
essentially unaffected by surfactant concentration. The KLshen value was
estimated to be about 30 x 10'3 cm/s, which is comparable to KL values for
surfactant-free, liquid-liquid systems at an equivalent 0 value. The KLa values for
CLA were 5- to 15-times higher than those for conventional emulsions. These
results suggest that CLA may offer significant advantages over conventional
emulsions for mass-transfer-Iimited, multiphase biocatalytic processes.

Chapter 3 presented measurements of the interphase mass-transfer
properties of CLA emulsions in water by coupling the mass-transfer of the solute
(p-tolyl acetate) to a chemical reaction. The overall KL values, which ranged from
0.002 to 0.015 cm/s, increased with Na; below a Nag value of about 12,500.
Above this value, KL was essentially independent of N95. These trends suggest
that as the degree of agitation is increased, the controlling resistance shifts from
the continuous phase to the shell phase. The KL values were essentially
independent of the surfactant concentration used to prepare the CLA. A
resistances-in-series model was used, along with a literature correlation for
KLconfinuous, to estimate the KLshe" value to be about 0.04 cm/S. At a nominal NRE
value of 20,000, the KLa/o values measured for CLA were about 20 times higher
than that for conventional emulsions. These values, measured using chemical

reaction, compare well to those obtained in Chapter 2 that were measured under

extractive conditions.

198

 

The application of CLA emulsions in multiphase enzymatic biocatalysis
was presented in Chapter 4. Dynamic resolution of L,D phenylalanine was
carried out in a multiphase system using porcine pancreatic lipase. A non-ionic
emulsfier (Tween 80), used for the formation of CLA emulsions, exhibited
minimal interactions with the biocatalyst. The CLA emulsions provided a four-fold
increase in reaction rates over conventional contacting methods. A statistical
design of experiment approach was used to identify the major variables affecting
equilibrium conversion and enantiomeric excess in presence of CLA. The optimal
conditions for high conversion (>90%) were identified as low void fraction (10%)
and high pH (6). Under conditions that are optimal for high equilibrium
conversion, the enantiomeric excess attained was also high (>90%), thus
precluding a tradeoff between maximizing conversion and enantiomeric excess.

Chapter 5 demonstrated the use of CLA emulsions in the
biotransformations of cinnemaldehyde by Baker's yeast at high cell densities
(2509/L). The solvent, 2,2,4 trimethyl pentane, was identified as a suitable
biocompatible solvent for yeast biotransformations. The presence of the organic
phase in the form of CLA emulsion increased the fraction of 4 substituted (2S,
3R)-5-phenylpent-4-en,2,3-ols (4) by three-fold upto 45%. The selectivity of the
yeast biocatalyst appears to depend on substrate concentration. The observed
change in selectivity due to the presence of hydrophobic resins was consistent
with this hypothesis. However the variation in selectivity in the presence of CLA

could not be attributed to this phenomenon. More studies are required to clarify

199

 

the mechanism by which the presence of dispersed organic solvents affects
selectivity. f

Chapter 6 presented preliminary studies on reversible emulsions formed
using the 15:1 and 20:1 pH reversible emulsifiers. Results indicated a high
degree of reversibility for emulsions formed using 15:1 and 20:1 emulsifiers.
Mass transfer studies indicated a four-fold increase in volumetric rates of mass
transfer for such emulsions compared to surfactant-free systems. The volumetric
mass transfer rate was three-fold less than that measured for CLA systems.
Thus, the ease of coalescence due to the unique surfactant architecture involves
a trade off in mass transfer enhancement compared to CLA. In order to
undertake further studies on the reversible emulsions systems, the interfacial

characterization of these reversible emulsions is essential.

200