Table of Contents
I. Front Material
a. List of Figures……………………………………………………...…………….ii
b. List of Tables………………………………………………………...………….iv
c. Symbols……………………………………………………………...…………vii
d. Subscripts………………………………………………………………………viii
1. Introduction
1.1: Applications………………………………………………………...…………….1
1.2: Demand…………………………...………………………………………………2
1.3: Current Methods of Synthesis…………………………...………………………..2
1.4: Biosynthetic Alternative………………………………………………...………..3
2. Theory
2.1: Pervaporation………………………………………………………………...…...5
2.2: Extractive Fermentation…………………………………………………………10
2.3: Adsorption………………………………………………………………………11
3. Pervaporation
3.1: Hollow Fiber Membranes: Materials and Methods…………………………….14
3.2: Hollow Fiber Membranes: Results……………………………………………..16
3.3: Flat PDMS Membranes: Materials and Methods………………………………17
3.4: Flat PDMS Membranes: Results………………………………………………..19
4. Extraction
4.1: Materials and Methods…………………………………………………………..21
4.2: Results…………………………………………………………………………..22
5. Adsorption
5.1: Styrene Oxide Recovery as Proxy for Styrene Recovery……………………….28
5.2: Materials and Methods…………………………………………………………..28
5.3: Results…………………………………………………………………………...30
6. Conclusions and Recommendations…………………………………………………..36
7. Works Cited…………………………………………………………………………...39
8. Appendices
A. Raw Data for Pervaporation Studies……………………………………………...42
B. Raw Data for Determination of Partition Coefficients……………………………43
C. Raw Data for Extractive Fermentation Studies…………………………………...45
D. Raw Data for Establishment of Styrene Oxide Isotherm on L-493 Resin………..48
E. Raw Data for Adsorption Studies with Styrene Oxide Producing NST74 E. coli..50
F. Calibration Curves………………………………………………………………...53

i

List of Figures
Figure 1: Polymerization of Styrene to Polystyrene by Addition of a Free Radical…………..1
Figure 2: Synthesis of Styrene via Alkylation of Benzene and Ethylene and Subsequent
Dehydrogenation of Resulting Ethylbenzene………………………………..3
Figure 3: Biosynthesis of Styrene in E. coli Through Glucose and L-phenylalanine………...4
Figure 4: Pervaporation Process Operated in Continuous Mode…………………………...…5
Figure 5: Three Steps of Mass Transfer in General Membrane Processes…………………....8
Figure 6: Apparatus for Pervaporation Experiments Using Hollow Fibers………………….15
Figure 7: Styrene Concentration over a Period of Two Hours in Pervaporation Control
Experiment (No Membrane) with Starting Concentration of 200 mg/L and
Flow Rate of 10 L/h…………………………………………………………..16
Figure 8: Aqueous Styrene Concentration in Pervaporation Experiment with Hollow Fiber
Polyethylene/Polyurethane Membrane, Starting Concentration of 200 mg/L,
and Flow Rate of 10 L/h……………………………………………………..17
Figure 9: SEM Image of PDMS Membrane after 72 Hours in Presence of E. coli NST74
(2500X)…………………………………………………………………...…..19
Figure 10: SEM Image of PDMS Membrane after 72 Hours in Presence of E. coli NST74 and
50 mg/L Styrene (2500X)…………………………………………………….20
Figure 11: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over
72 Hours in Broth A1…………………………………………………………23
Figure 12: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over
72 Hours in Broth A2…………………………………………………………24
Figure 13: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over
72 Hours with 20 g/L Phenylalanine Supplemented Every 24 Hours in Broth
C2…………………………………………………………………………...26
Figure 14: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over
72 Hours with 20 g/L Phenylalanine Supplemented Every 24 Hours in Broth
C3…………………………………………………………………………...26

ii

Figure 15: Isotherm at 32 °C for Styrene Oxide Adsorption on L-493 (Error bars reported to
1 standard deviation)………………………………………………………….31
Figure 16: Optical Density at 600 nm of Styrene Oxide Producing Cultures. Blue = Control;
Red = Resin added at 36 hours to Flask 1; Green = Resin added at 48 hours to
Flask 2; Purple = Resin added at 60 hours to Flask 3………………………..33
Figure 17: Aqueous Styrene Concentration in Flasks Over 72 Hours. Blue = Control, Red =
Resin Added at 36 Hours to Flask 1, Green = Resin Added at 48 Hours to flask
2, Purple = Resin Added at 60 Hours to Flask 3……………………………..34
Figure 18: Mass of Styrene Oxide Desorbed from L-493 Resin over Five 10 mL Ethanol
Washes. Blue = Resin added at 36 hours to Flask 1, Red = Resin added at 48
hours to Flask 2, Green = Resin added at 60 hours to Flask 3…………..........35
Figure 19: Isotherm data plotted according to linearized Freundlich Equation……………...48
Figure 20: Isotherm data plotted according to linearized Langmuir Equation………………49
Figure 21: Calibration Curve for Aqueous Styrene on HPLC……………………………….53
Figure 22: Calibration Curve for Aqueous Phenylalanine on HPLC………………………..53
Figure 23: Calibration Curve for Aqueous Cinnamic Acid on HPLC……………………….54
Figure 24: Calibration Curve for Aqueous Styrene Oxide on HPLC………………………..54
Figure 25: Calibration Curve for Styrene in Hexadecane on GC……………………………55
Figure 26: Calibration Curve for Styrene in BEHP on GC. Samples were diluted with hexane
to final composition of 50% BEHP solution and 50% hexane……………….55
Figure 27: Calibration Curve for Styrene in Hexane on GC…………………………………57
Figure 28: Calibration Curve for Styrene in Dodecane on GC…………………………...….57
Figure 29: Calibration Curve for Styrene Oxide in Hexane on GC………………………….58

iii

List of Tables
Table 1:

Specs for Hollow Fiber Membranes Used in Hollow Fiber Pervaporation
Experiments………...………………………………………………………...14

Table 2: Partition Coefficients for Styrene between Water and Selected Solvents (% Error in
Parentheses)……...…………………………………………………………...23
Table 3: Styrene Concentration in Experiments with Hollow Fiber Membranes (1 hour)…..42
Table 4:

Styrene Concentration in Aqueous Phase of Controlled Experiments with No
Membrane (2 hours)…………………………………………………………..42

Table 5: Partition Coefficients for Styrene between Hexadecane and Water When 5 mL of a
50 mL Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying
Volumes of Double Deionized Water..………………………………………43
Table 6: Partition Coefficients for Styrene between Hexane and Water When 5 mL of a 50
mL Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying
Volumes of Double Deionized Water………………..………………………43
Table 7: Partition Coefficients for Styrene between BEHP and Water When 5 mL of a 50 mL
Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying Volumes
of Double Deionized Water…………………………………………..………44
Table 8: Partition Coefficients for Styrene between Dodecane and Water When 5 mL of a 50
mL Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying
Volumes of Double Deionized Water………………………………………..44
Table 9: Styrene Concentration (mg/L) in Organic Phase in First Preliminary Extraction
Experiment (10 mL solvent: 50 mL minimal media)………………………..45
Table 10: Styrene Concentration (mg/L) in Organic Phase in Second Preliminary Extraction
Experiment (10 mL solvent: 50 mL minimal media)………………………..45
Table 11: Styrene Concentration (mg/L) in Organic Phase in Extraction Experiments with
Cultures A1, A2, and A3 (5 mL solvent: 50 mL minimal media). Note
decreased productivity of A3 compared to A1 and A2……………………….45
Table 12: Styrene Concentration (mg/L) in Organic Phase in Extraction Experiments with
Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with 0.25 g
phenylalanine supplemented every 24 hours. B2 produced at significantly
lower rates than B1 and B3…………………………………………………..46

iv

Table 13: Styrene Concentration (mg/L) in Aqueous Phase in Extraction Experiments with
Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with 0.25 g
phenylalanine supplemented daily. Data for B3 not included because it did not
produce at same level as other two cultures…………………………………..46
Table 14: Phenylalanine Concentration (g/L) in Aqueous Phase in Extraction Experiments
with Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with
0.25 g phenylalanine supplemented daily. Data for B3 not included because it
did not produce at same level as other two cultures………...………………..46
Table 15: Cinnamic Acid Concentration (mg/L) in Aqueous Phase in Extraction Experiments
with Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with
0.25 g phenylalanine supplemented daily. Data for B3 not included because it
did not produce at same level as other two cultures…………………………..47
Table 16: Styrene Concentration (mg/L) in Organic Phase in Extraction Experiments with
Cultures C1, C2, and C3 (5 mL solvent: 50 mL minimal media) with 1.0 g
phenylalanine supplemented every 24 hours…………………………………47
Table 17: Isotherm Data for Styrene Oxide on L-493 Resin (All samples in 20 mL water) at
32 °C………………………………………..………………………………...48
Table 18: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Five Methanol
Washes (20 mL each). 2.5 g L-493 were added to each culture upon transfer to
minimal media……………………………………………………………….50
Table 19: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Five Methanol
Washes (10 mL each). 0.75 g L-493 were added to each culture upon transfer
to minimal media………………………………..…………………………...50
Table 20: Optical Densities at 600 nm of Cultures with 0.75 g L-493 Resin added upon
transfer to minimal media compared to controls……………………………...50
Table 21: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Four Methanol
Washes (10 mL each). 1.0 g L-493 were added at three different times (28, 32,
and 36 hours) after transfer to minimal media environment. All flasks were
derived from same culture…………………………………………………….51
Table 22: Aqueous phase concentration of styrene oxide (g/L) over a period of 72 hours in
flasks to which 1.0 g resin were added at three different times (28, 32, and 36
hours) after transfer to minimal media. Bad sample likely at 72 hours for
control………………………………………………………………………51
v
v

Table 23: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Five Ethanol
Washes (10 mL each). 1.0 g L-493 were added at three different times (36, 48,
and 60 hours) after transfer to minimal media………………………...……...51
Table 24: Aqueous phase concentration of styrene oxide (g/L) over a period of 72 hours in
flasks to which 1.0 g resin were added at three different times (36, 48, and 60
hours) after transfer to minimal media………………………………………..52
Table 25: Optical Density at 600 nm over a period of 72 hours of cultures to which 1.0 g
resin were added at three different times (28, 32, and 36 hours) after transfer to
minimal media…………………………………………………………….…..52
Table 26: Mass of Styrene Oxide Removed (g) from L-493 in Two Washes with Selected
Solvents (Theoretical Mass Adsorbed = 0.006 g)…………………………….52

vi

Symbols
Activation Energy (kJ/mol)…………………………………………………………………..EA
Adsorption Ratio (g/kg)……………………………………………………………………….q
Aqueous Phase Concentration (g/L)……………………………………………………….....C
Chemical Potential (J/mol)……………………………………………………………………µ
Enrichment Factor (unitless)…………………………………………………………………..β
Flux (mol/m2 h)………………………………………………………………………………..J
Freundlich Adsorption Capacity (unitless)…………………………………………………...K’
Freundlich Adsorption Favorability Parameter (unitless)…………………………………….n
Gaseous Phase Mole Fraction (mol/mol)………………………………………………………y
Langmuir Adsorbent Affinity (unitless)………………………………………………………b
Langmuir Maximum Solute Adsorbed (g)……………………………………………………Q
Liquid Phase Mole Fraction (mol/mol)………………………………………………………..x
Mass Transfer Coefficient (mol2/h m2 J)……………………………………………………….k
Membrane Transfer Coefficient (mol2/ h m J)…………………………………………………P
Partition Coefficient (unitless)………………………………………………………………...K
Resin Mass (g)…………………………………………………………………………...……M
Selectivity (unitless)……………………………………………………………………………α
Temperature (K)……………………………………………………………………………….T
Thickness (m)…………………………………………………………………………………..t
Volume (L)…………………………………………………….………………………………V

vii

Subscripts
Equilibrium……………………..………………………………………………………………e
Extract…………………………..……………………………………………………………...E
Initial Value…………...………………………………………………………………………..0
Liquid Phase…...………………………………………………………………………………..l
Membrane……………………………………………………………………………………..M
Overall………………………………………………………………………………………...ov
Raffinate.………………………………………………………………………………………R
Solute.…………………………………………………………………………………………..i
Solvent….………………………………………………………………………………………j
Vapor Phase……………………………………………………………………………………v

viii

1. Introduction
1.1. Applications
Styrene, a derivative of benzene, is one of the most widely-utilized commodity chemicals in
the world.

Though a volatile liquid in its standard molecular form, styrene can be

polymerized and combined with other chemicals to create products with applications ranging
from plastics to synthetic rubbers1. Polystyrene is the simplest and most widely demanded
styrene derivative, accounting for 59% of styrene consumption in 20082. The polymer is
generally synthesized by introducing a radicalizing agent (often hydrogen peroxide) to
styrene, setting off a chain reaction illustrated in Figure 13. In the initiation step, the radical
binds to a styrene molecule, severing one of the two bonds in the alkene and producing a lone
electron on the carbon alpha to the benzene ring. The radicalized styrene can bind to another
styrene unit, affecting the same alteration in the structure of the normal molecule. The
repetitive process results in long-chained molecules with excellent thermal and electrical
insulation capabilities, lending to the wide-spread use of polystyrene in the electronics
industry4.

Figure 1: Polymerization of Styrene to Polystyrene by Addition of a Free Radical

3

Polystyrene can also be combined with other chemicals to produce products with greater
resistance to both heat and stress4. The two most common examples are styrene-butadiene
rubber (SBR) and acrylonitrile-butadiene-styrene (ABS). SBR consists of a three to one ratio
of butadiene to polystyrene and is the most commonly used synthetic rubber in the world4.
Likewise, ABS sees extensive application in various plastics industries4.
1

1.2. Demand
Given the widespread use of styrene, it should come as no surprise that the styrene demand
has been steadily increasing over the past decade. Between 2000 and 2010, global demand
increased from 17.3 million tons to 21.7 million tons5. The three polymers described above
(polystyrene, SBR, and ABS) accounted for over 75% of the demand5. Projected demand
over the next decade is anticipated to dwarf that of the previous decade (up to 41.1 million
tons) due to expansion of the styrene industry in Asia, particularly in China, South Korea, and
India5.

1.3. Current Methods of Synthesis
On an industrial scale, styrene is most commonly synthesized by dehydrogenation of ethyl
benzene via an oxide catalyst6. The requisite ethyl benzene is obtained by alkylation of
benzene with ethylene, catalyzed by aluminum chloride6. Figure 2 (page 3) is a sample
process flow diagram for the alkylation and dehydrogenation steps6. Though the schematic
dates back to the 1950s, the general procedure has not changed significantly in the last 60
years.
From the perspective of energy efficiency and independence, several problems exist with the
process depicted in Figure 2. As of 2002, industrial styrene production in the United States
consumed 200 trillion BTUs of steam annually, making it the most energy-intensive of
commodity chemical processes7. Given the current trend in styrene demand, this number
figures to rise exorbitantly. Furthermore, the starting components benzene and ethylene are
both currently derived from pyrolysis of petroleum6, which puts styrene production at odds
with the current emphasis upon limiting petroleum dependency and carbon emissions. An
alternative method, therefore, of producing styrene which promised to consume less energy
and petroleum would be an incredible development. Such a method, however, would have to
be capable of assuming the burden of the burgeoning global demand for styrene.

2

Figure 2: Synthesis of Styrene via Alklyation of Benzene and Ethylene and Subsequent
6
Dehydrogenation of Resulting Ethylbenzene

1.4. Biosynthetic Alternative
Recently, Rebekah McKenna of Arizona State University demonstrated that styrene can be
produced by genetically engineered bacteria. Figure 3 (page 4) outlines the pathway, which
utilizes Escherichia coli (E. coli) NST74 with L-phenylalanine as a starting compound7. The
organism generates the requisite L-phenylalanine from glucose, a renewable resource
which—on an industrial level—might theoretically be supplied to the process in the form of
biomass feedstock. The first step (A) is catalyzed by the PAL2 enzyme from Arabidopsis
thaliana, while the second step (B) proceeds by the activity of FDC1 from Saccharomyces
cervisiae. The genes of these enzymes are expressed in E. coli NST74 by means of the
plasmid pACYCDuet-1.

3

Figure 3: Biosynthesis of Styrene in E. coli Through Glucose and L-phenylalanine7

It was observed that styrene and trans-cinnamate became toxic to E. coli at respective
concentrations of 300 and 800 mg/L7. To counteract this undesirable effect and make this
process economically viable on a larger scale, styrene must somehow be continuously
removed from the E. coli media. Several options for accomplishing this feat are possible, but
the three that will be explored in the remainder of this paper are pervaporation, solvent
extraction, and adsorption.

4

2. Theory
2.1. Pervaporation
Pervaporation is a relatively new technique that shows promise in achieving separations of
dilute solutions. The most common solvent currently being studied is water because of the
potential applicability of pervaporation in wastewater treatment. Generally, a pervaporation
unit consists of a membrane that is more selectively permeable to the solute than the solvent
(though in some cases membranes are designed for selectivity towards the solvent).

A

vacuum is applied on one side of the membrane to maintain a favorable gradient for diffusion
of the solute through the membrane as a gas8. The separated stream, called the permeate, is
then condensed, resulting in a liquid that is significantly richer in the solute than is the feed
stream8.

The remaining feed can be recycled to allow pervaporation to be performed

continuously. Figure 4 shows one possible set-up for a continuous pervaporation process9.

Figure 4: Pervaporation Process Operated in Continuous Mode9

The effectiveness of a pervaporation process is generally described by either a separation
factor or an enrichment factor. The separation factor between the permeate (species i) and the
solvent (species j) is defined as9:

5

 /


 = /




(1)

Likewise, the enrichment factor can be expressed as9:


 = 

(2)



For dilute solutions, Equations 1 and 2 will become approximately equal,9 an approximation
that will be valid in the current study since the desired styrene concentration is below 300
mg/L.
The degree of separation attainable by pervaporation is a function of the operating conditions
(temperature, pressure, feed concentration, and flow rate) and the membrane material and
geometry.

As mentioned previously, most investigations of these variables have been

performed with water as the solvent. Thus, unless otherwise noted, water will be assumed to
be the solvent in the following discussion.
Recent studies on pervaporation have generally operated between temperatures of 20 °C and
70 °C8. In most combinations of solute and membrane material studied, both the solute and
water fluxes were found to increase with increasing temperature8. Thus, depending on the
relative increases in the fluxes, the separation factor might either be positively or negatively
affected by a temperature increase. Separations with very hydrophobic solutes have proven
an exception to this rule, demonstrating decreased water fluxes at high temperatures10.
Further studies have shown that Arrhenius-type temperature relationships can often reliably
be applied to membrane fluxes in the form11:
=



exp( )

(3)

However, considering that the introduction of live bacteria will place severe restrictions on
operating temperature, the temperature will not be a process variable that can be significantly
adjusted for the purposes of flux optimization in this study.
Unlike temperature, pressure on the permeate side has not been shown to have a significant
impact on solute flux8. Operating too near the vapor pressure of water, though, can lead to

6

decreased water fluxes12.

This phenomenon might appear desirable at first glance, but

lowering the water flux too drastically can adversely affect convective flow across the
membrane8.

Regardless, operation at near vacuum conditions is desirable from the

perspective of maintaining a favorable concentration gradient of the solute across the
membrane.
Maintaining vacuum conditions also affects the manner in which flux varies with feed
concentrations. In the presence of a vacuum, solute flux has been shown to increase linearly
with increasing feed concentration under dilute conditions8. This observation fits logically: a
higher feed concentration results in a larger gradient and hence a larger driving force for
transfer. Interestingly, increasing solute concentration has also led to a higher water flux in
some instances. This event was shown to be correlated to the presence of membrane swelling.
In studies where steps were taken to prevent swelling, water flux was shown to be
independent of solute concentration under dilute conditions13.
The dependence of flux on feed flow rate is more complicated than those relationships
described above because of mass transfer considerations. Mass transfer across a membrane is
generally theorized as a three step process. The chemical in question must first adsorb to the
surface of the membrane. In order for a driving force for this process to exist, the solute
concentration on the membrane surface must be less than the bulk solute concentration. Thus,
the presence of a liquid boundary layer (Region 2 on Figure 512 on page 8) between the bulk
fluid (Region 1) and the membrane (Region 3) is postulated, across which the solute
concentration decreases.

Likewise, the solute concentration must decrease across the

membrane in the direction of transfer and across the space (Region 4) between the opposite
side of the membrane and the bulk vapor phase (Region 5). Each region of transfer has a
characteristic mass transfer coefficient. The inverse of the coefficient is referred to as the
resistance of the region. That is, a region with a low mass transfer coefficient will have a
large resistance to transfer. Just as the overall rate of a chemical reaction is limited by the
slowest reaction step, so the overall rate of mass transfer is limited by the step with the
greatest resistance. Due to the presence of a vacuum on the vapor-side of the membrane, the
resistance of the vapor boundary layer (Region 4) is usually considered negligible in
pervaporation processes12. In studies with hydrophobic membranes, feed flow rate was found
7

to have a large impact on solute flux when the resistance of the liquid boundary layer (Region
2) dominated that of the membrane (Region 3)14, 15.

Figure 5: Three Steps of Mass Transfer in General Membrane Processes

12

Additional experiments demonstrated that water flux is not similarly affected; that is,
increasing the feed velocity does not increase water flux when the resistance of the liquid
boundary layer is limiting16, 17. Unlike temperature and pressure adjustments—which might
increase solute flux but decrease selectivity due to simultaneous water flux increase—changes
in the feed flow rate can increase both solute flux and selectivity. Thus, feed flow rate
appears to be the optimal process variable to adjust to optimize solute separation and
recovery.
Turning attention to the membrane, the two most critical factors affecting pervaporation
performance are permeability and thickness. Together, these two properties determine the
membrane resistance, which is inversely proportional to the permeability and directly
proportional to thickness8. Permeability is determined by the membrane material. The most
commonly utilized hydrophobic material is polydimethylsiloxane (PDMS), which has
demonstrated high selectivity for many organic solutes8. A noteworthy result with respect to
membrane thickness is that water flux tends to decrease at a faster rate than solute flux as
thickness is increased. For instance, one study noted that an eight-fold increase in thickness
8

resulted in almost exactly an eight-fold decrease in water flux compared to less than a twofold decrease in toluene flux18. A final parameter that has been explored in many experiments
is the geometric orientation of the membrane. Studies have been performed with flat sheets
with cross-flow, hollow fibers (either shell-side or tube-side flow), and with coiled tubes. The
effect of these geometries on selectivity depends heavily on the membrane material and the
solute in question.
Taking all of the above factors into consideration, several mathematical models have been
developed to describe solute transfer across a pervaporation membrane. These models do not
differ in their starting points, only in the assumptions and simplifications made in evaluating
the relative activities of the solute and water. The common starting point is a general overall
mass transfer equation19:


=   ( , −  , )

(4)

Referring again to Figure 5, the bulk liquid chemical potential is characteristic of Region 1
while the bulk vapor chemical potential corresponds to Region 5. The Resistance in Series
Model presented by Ji simplifies Eq. 4 by assuming that the liquid chemical potential is equal
to the bulk solute concentration (Cib) and that the vapor solute potential is negligible under
vacuum conditions12.

Meanwhile, Ghoreyshi’s Solute-Diffusion Model, which is more

recent, presents a rigorous approach involving activity coefficients and the mole fractions of
the solute in the liquid and vapor phases19.
The inverse of the overall mass transfer coefficient is the overall resistance. The overall
resistance is the sum of the individual resistances of each region of transport (Regions 2, 3,
and 4), given by8:






"



= +# +
$



(5)

However, the final term in Eq. 5 tends to zero as complete vacuum on the permeate side is
approached. The usefulness of this simplification manifests itself in that it produces a linear
equation of kov-1 against Pm-1 with a slope of t and an intercept of kL-1. Thus, by performing
experiments in which the membrane thickness is varied (and all other parameters are held
9

constant), one can estimate the liquid-phase transfer coefficient and the membrane
permeability from the response of the overall transfer coefficient8.

2.2. Extractive Fermentation
Extractive fermentation is the application of liquid-liquid extraction (or solvent extraction)
specifically to separation of a solute from fermentation media. The method makes use of the
differing solubility of a solute in two immiscible solvents. The medium in which the solute
initially resides—the feed—is contacted with another solvent which possesses a higher
affinity for the solute. After mixing and allowing the system to come to equilibrium, the
solvent preferential for the solute is removed. This layer is called the extract, while the
remaining feed (which still contains some solute) is termed the raffinate. For a given solute
and pair of solvents, a partition coefficient describing the relative amount of solute (i) present
in the extract (E) and raffinate (R) is defined as20:


% = ,&

,'

(6)

Generally the most important factors to consider with respect to solvent extraction are the
relative polarities of the solute and solvents. Styrene, for instance, is a non-polar molecule
and therefore does not possess a high affinity for water (which constitutes the bulk of the
bacterial media in fermentation processes). If a nonpolar solvent were to be contacted with
this media, one would expect most of the styrene to migrate from the media to the solvent. Of
course, the issue becomes more complex with live bacteria because one must be careful not to
introduce a solvent that will inhibit the growth and productivity of the E. coli. Additionally,
potential solvents must possess a low viscosity, exhibit a large density difference with respect
to water, be sterilizable, and separate easily from water; all at a moderate cost21. Operational
considerations that must be taken into account include a decision between batch and
continuous processes and, in the event of continuous operation, the number of equilibrium
stages and the type of flow (countercurrent or concurrent).

10

Probably the most extensively studied biochemical application of solvent extraction has been
in A-B-E fermentations. This process utilizes the organism Clostridium spp. to produce
butanol, acetone, and ethanol in a mass ratio generally reported as 60:30:1022. Of these three
compounds, butanol and acetone are produced in sufficiently large quantities to make their
recovery desirable.

All three products inhibit microbial growth at sufficiently high

concentrations, and so must be removed from the process if large yields of butanol and
acetone are to be realized. Ishii et al. demonstrated oleyl alcohol to be well-suited for the
task, as it exhibited a large partition coefficient for butanol and did not adversely affect the
bacterial cultures23. Other studies showed that benzyl benzoate could be used in conjunction
with oleyl alcohol to extract acetone from the fermentation broth24. These findings provide an
encouraging precedent.
Screening of potential solvents for a given extractive process can itself become an exhausting
task.

Fortunately, though, some research has already been performed with respect to

extractive recovery of styrene. Panke et al. of the Swiss Federal Institute of Technology used
the compounds hexadecane and bis-(2-ethylhexyl)-phthalate (BEHP) to recover both styrene
and styrene oxide from an E. coli fermentation broth25. They determined that the partition
coefficients for styrene in each solvent are on the order of 1000 and that neither solvent
significantly inhibits microbial growth25. Since styrene oxide was the product of interest in
this experiment, the group did not attempt to optimize styrene recovery with these solvents.
However, this information provides an excellent jumping off point.

2.3. Adsorption
Adsorption is another separation technique generally reserved for removal of dilute solutes
from liquids. The adsorbent, a granular solid, is selected so that the interactions between the
solute and the adsorbent surface are preferential to those between the solute and solvent. In
order to increase the surface area across which these interactions occur, many adsorbents
contain pores of varying diameters. For each grouping of solvent, solute, and adsorbent, a
unique equilibrium is established in which the ratio of adsorbed to unadsorbed solute remains

11

constant. A typical example of an adsorption process is a water filter that uses activated
carbon to remove certain ionic species.
At a given moment during the adsorption process, the ratio of solute adsorbed to the total
mass of adsorbent is expressed as q (usually in g solute/kg adsorbent). At equilibrium, the
maximum capacity of the adsorbent is reached and q becomes constant. A mass balance on
the system at equilibrium yields the equation26:
(()* − )+ ) = ,(-+ − -* )

(7)

Usually, though, dry resin is added to a system, meaning that q0 = 0. By making this
simplification and rearranging Eq. 6, an expression for the equilibrium q can be easily
found26:
.(/ 0/1 )

-+ =

(8)

2

Many studies have been performed to relate the equilibrium q to the equilibrium concentration
of the solute in the liquid phase. Two of the most widely used models that have been
developed are the Langmuir and Freundlich isotherms. The Langmuir isotherm takes the
form27:
3 5/

4 1
-+ = 65/

(9)

1

Q0 is the maximum amount of solute that can be adsorbed and b is a constant describing the
affinity of the solute for the adsorbent. For the purpose of experimental analysis, Eq. 9 can be
linearized27:
/1

71



=3

45

/

+ 31

(10)

4

Thus, Q0 can be found from the reciprocal of the slope of a plot of

/1

71

against Ce. Once this

value is known, the intercept of the same plot can be used to find b.
The other common isotherm, the Freundlich model, is an exponential equation27:

12

9

-+ = %′)+:

(11)

K’ is the adsorption capacity and n denotes the favorability of the adsorption. Eq. 11 also can
be linearized to allow determination of its associated parameters as follows27:


;<-+ = ;<%= + > @ ;<)+
?

(12)

The Freundlich model has generally proven to be more appropriate for heterogeneous
systems27, thus it is anticipated that it will demonstrate a better fit than will the Langmuir
model to the data in this study.

13

3. Pervaporation
3.1. Hollow Fiber Membranes: Materials and Methods
Pervaporation was first investigated using hollow fiber membranes.

Two membrane

compositions were considered; specifications are presented in Table 1.

Polyurethane

membranes have been applied to pervaporation of trichloroethane (TCA), toluene, and
dichloromethane9, while composite membranes consisting of ethylene and propylene have
been explored in removing trichloroethylene (TCE) and toluene8. None of the materials have
been applied to pervaporation of styrene. A primary consideration in the utilization of these
membranes was the fact that a group headed by Dr. Hwun-Woo Kim has been employing
them in its research at the Arizona State Biodesign Institute. Dr. Kim was gracious enough to
provide the materials and guidance in creating the modules.
Table 1: Specs for Hollow Fiber Membranes Used in Hollow Fiber Pervaporation Experiments
Fiber A

Fiber B

Supplier

Mitsubish-Rayon28

Teijin Fibers29

Material

Inner Wall: Polyethylene

Polypropylene

Middle Layer: Polyurethane
Outer Wall: Polyethylene
Outer Diameter (microns)

200

280

Thickness (microns)

55

50

Cross Sectional Area (square microns)

31,400

61,544

The fibers were anchored into plastic supports at each end using Urethane adhesive from
Scotch-Weld. To characterize the impact of total surface area on membrane flux, modules
consisting of 5, 10, and 29 fibers were utilized.

Figure 6 (page 15) illustrates the

experimental apparatus. During each run, one membrane module was placed in a tubular
piece of glassware (Unit C). A vacuum pump (Unit D) was used to create a vacuum on the
14

tube side of the hollow fiber. The permeate was drawn off from Unit C via Path 2. An
aqueous solution of styrene—originally stored in Unit A—was circulated continuously
through Unit C on the shell side of the hollow fibers via Path 1. Path 1 was constructed
almost entirely of 1/4” stainless steel tubing in an effort to prevent styrene from vaporizing or
interacting with any synthetic rubber material. The only exception was a short segment, about
a half-inch long, of natural rubber tubing situated shortly upstream of the point where the feed
entered Unit C. This small segment was added for the purpose of taking samples. At this
stage of design, a method of recovering styrene present in the permeate was not included in
the apparatus. It was first necessary to demonstrate a decrease in the concentration of styrene
in the aqueous phase

Figure 6: Apparatus for Pervaporation Experiments Using Hollow Fibers

15

During each experiment, a feed with a styrene concentration of 200 mg/L was introduced to
Unit A and stirred continuously to ensure proper mixing. The feed was circulated through the
apparatus at a rate of 10 L/h. A control experiment with no membrane was performed for two
hours to determine if any styrene was lost to the system. Experiments with membranes were
performed for an hour and samples of the aqueous feed were taken at 0, 3, 6, 10, 15, 30, 45,
and 60 minutes using a syringe at the location described above and analyzed by High Pressure
Liquid Chromatography (HPLC).

3.2. Hollow Fiber Membranes: Results
The polypropylene and polyethylene-polyurethane composite tubular membranes did not
demonstrate a robust ability to separate dilute concentrations of styrene from water.
Controlled experiments with no membrane demonstrated a styrene loss on the same order as
experiments with either type of membrane, regardless of the number of membrane fibers.
Figure 7 displays the decrease in styrene concentration that occurred during the controlled
run, while Figure 8 (page 17) that when a 5-fiber composite membrane was utilized.
0.25

Concentration (g/L)

0.2
0.15
0.1
0.05
0
0

30

60

90

120

Time (min)

Figure 7: Styrene Concentration over a Period of Two Hours in Pervaporation Control Experiment (No
Membrane) with Starting Concentration of 200 mg/L and Flow Rate of 10 L/h

16

The initial concentration in Figure 8 is higher than that in Figure 7, but it is readily observed
that the change in concentration over the period of an hour is almost identical in each case
(approximately 0.06 g/L).

A similar trend was observed in experiments with a larger total

membrane surface area and with polypropylene membranes of varying areas. In spite of
efforts to ensure the contrary, styrene appeared to quit the system through volatilization faster
than through pervaporation. Experiments with this configuration were therefore discontinued.

0.25

Concentration (g/L)

0.2
0.15
0.1
0.05
0
0

10

20

30

40

50

60

Time (min)

Figure 8: Aqueous Styrene Concentration in Pervaporation Experiment with Hollow Fiber Polyethylene/Polyurethane
Membrane, Starting Concentration of 200 mg/L, and Flow Rate of 10 L/h

3.3. Flat PDMS Membranes: Materials and Methods
To enhance recovery by pervaporation, an alternative configuration and membrane material
were explored. The hollow fiber design was discarded in favor of a flat polydimethylsiloxane
(PDMS) sheet with cross-flow. PDMS was selected because it has been demonstrated to
show high selectivity towards volatile organic compounds in a number of other studies8. The
PDMS sheets, which varied in thickness from 20-200 µm, were provided by the Lind Lab at
Arizona State University.

17

Experiments with PDMS membranes operated in pervaporation mode were not performed
because the required equipment was unavailable at the time of this study. In order to provide
a starting point for future work on this subject, the degree of fouling experienced by E. coli
NST74 on PDMS was characterized by scanning electron microscopy (SEM). An NST74
inoculum was prepared by combining a seed culture, 5 mL of Luria bertani (LB) broth, 5 µL
of ampicillin, and 5 µL of kanamycin in a culture tube. The mixture was incubated at 37 °C
for 24 hours. Two 2.5 mL samples of the inoculum were spun down and added to separate
well-plates, each of which contained a 20 micron thick sample of PDMS membrane. The
total volume in each well-plate was increased to 5 mL by the addition of rich media. A 0.27
µL volume of styrene was added to one of the two well-plates to simulate a 50 mg/L styrene
solution that might be encountered by the fermentation broth during pervaporation
experiments. The styrene well-plate and control well-plate were placed in an incubator at 37
°C and allowed to sit for a period of 72 hours. The same volume of styrene (0.27 µL) was
added to the styrene well-plate every 24 hours to counteract the amount lost by evaporation.
To prepare the membrane samples for analysis, the media in the well plates was decanted,
combined with 2 mL of 2.5% gluteraldehyde, and allowed to sit for 1.5 hours. The samples
were then washed five times with 2 mL of double deionized water. The membranes were next
fixed in 2 mL of 1% osmium tetroxide for 1.5 hours. Another five washes of 2 mL double
deionized water followed. The water in the membranes was then exchanged with ethanol by
soaking the membranes in solutions of increasing ethanol concentration. The first exchange
step was performed with 2 mL of a 10% ethanol solution. Subsequent exchanges were
accomplished with 2 mL of 20%, 40%, 60%, and 80% ethanol, as well as three washes with 2
mL of 100% ethanol. The samples were then critical-point dried with 9 CO2 exchanges.
Finally, the samples were mounted on carbon tape on pin stubs and sputter coated with gold
on a Technics Sputter Coater. Images were obtained using a JEOL JSM6300 Scanning
Electron Microscope.

18

3.4. Flat PDMS Membranes: Results
SEM analysis demonstrated that the degree of membrane fouling due to the formation of E
coli NST74 biofilm is less severe in the presence of styrene. Figure 9 shows the membrane
from the control well-plate after 72 hours, while Figure 10 (page 20) displays the membrane
from the styrene well-plate after the same amount of time. Each image is magnified 2500
times.

Figure 9: SEM Image of PDMS Membrane After 72 Hours in Presence of E. Coli NST74 (2500X)

Figures 9 and 10 demonstrate that styrene somehow interferes with the hydrophobic
interactions between PDMS and the cellular membranes of the E coli. Styrene may compete
directly with PDMS for hydrophobic sites on the E. coli membrane or styrene may effect a
change in the E. coli membrane that decreases the number of hydrophobic regions.
Regardless of the mechanism, the outcome observed is highly desirable. Fouling has a
19

negative impact on both membrane flux and selectivity, which decreases achievable permeate
purity and increases the required frequency of membrane replacement or cleaning. On an
industrial design scale, such deficiencies translate into accrual of further operating costs. The
fact that styrene-producing E. coli will cause a relatively lower degree of PDMS fouling is
thus a promising result for future work on this subject.

Figure 10: PDMS Membrane After 72 Hours in Presence of E. coli NST74 and 50 mg/L Styrene Solution

20

4. Extraction
4.1. Materials and Methods
The partition coefficients of styrene in the water/hexadecane and water/BEHP systems were
determined to characterize the separation efficiency of each solvent. Solvent screening was
also performed with hexane and dodecane to determine whether hydrocarbon chain length had
a significant effect on solute partitioning. Several stock solutions were created by combining
1.7 mL of styrene with 50 mL of each solvent. Next, 5 mL of each solution were contacted
with different five volumes of deionized water (5, 10, 25, 50, and 100 mL) which, with
replication, resulted in ten samples for each solvent. When obtaining the partition coefficient
for hexadecane and BEHP, additional samples were created using 75, 125, and 500 mL of
water to mitigate the high variability encountered when analyzing the first ten samples. Each
two-phase media was shaken for one minute and then allowed to equilibrate. Once the
organic and aqueous layers settled out, two 2 mL samples were taken from each phase. The
organic samples were analyzed by Gas Chromatography (GC) and the liquid samples by High
Pressure Liquid Chromatography (HPLC). Partition coefficients at each starting volume of
water were calculated and used to determine an average partition coefficient for each solvent.
The compatibility of hexadecane and BEHP with live, styrene-producing E. coli cultures was
next investigated. Three bacterial broths were created by adding NST74 seed cultures to 5
mL of LB broth and supplementing them with 5 uL each of ampicillin and kanamycin. These
broths were placed in an Innova 44 Incubator Shaker from New Brunswick Scientific at 37 °C
for 24 hours. Bacterial samples were then transferred to a minimal media environment
consisting of 25 mL of MM1, 21.25 mL of sterile water, 3.75 mL of a 20% glucose solution,
1 mL of the inoculum and 50 uL each of ampicillin and kanamycin. MM1 is a phosphatelimited environment adapted from Qi et al30 by McKenna et al7 composed of: glucose (20
g/L), MgSO4•7H2O (0.5 g/L), (NH4)2SO4 (4.0 g/L), MOPS (24.7 g/L), KH2PO4 (0.3 g/L),
K2HPO4 (1.0 g/L) and 5 ml/L ATCC Trace Mineral Supplement (Catalog No. MD-TMS):
EDTA (0.5 g/L), MgSO4•7H2O (3.0 g/L), MnSO4•7H2O (0.5 g/L), NaCl (1.0 g/L),
FeSO4•7H2O (0.1 g/L), Co(NO3)2•6H2O (0.1g/L), CaCl (0.1 g/L), ZnSO4•7H2O (0.1 g/L),

21

CuSO4•5H2O (0.01 g/L), AlK(SO4)2 (0.01 g/L), H3BO3 (0.01 g/L), Na2MoO4•2H2O (0.01
g/L), NaSeO3 (0.001 g/L), NaWo4•2H2O (0.1 g/L), and NiCl2•6H2O (0.02 g/L).
Duplicate flasks based on each of the three seed cultures were created for a total of six flasks.
The flasks were placed in the incubator shaker at 32°C for eight hours. After this time, they
were induced with 10 uL of IPTG. Five mL of hexadecane were added to half the flasks and
5 mL of BEHP to the other half. Foam caps were initially used until it was realized that these
caps allowed styrene to evaporate from the system. Subsequently, glass caps were used. Pure
oxygen was supplemented to each flask initially and every 24 hours thereafter to ensure that
the cultures had sufficient oxygen to grow. To buffer the solution pH, 1 mL of K2HPO4 was
added to each flask 24 hours after induction. Samples were taken from the organic phase
every 24 hours, centrifuged, and analyzed by gas chromatography. This experiment was
carried out for 72 hours.
Further experiments were performed to determine if the styrene yield could be increased by
supplementing the cultures with additional phenylalanine. A positive outcome would indicate
that a ceiling on the capacity of the styrene pathway had not yet been reached. Fermentation
broths were created by the same process described above, but 0.25 g of phenylalanine was
added to each flask at 0, 24, and 48 hours after induction (for a concentration of 5 g/L). The
aqueous phase was sampled and the concentrations of styrene, phenylalanine, and cinnamic
acid were determined by HPLC. The consumption of phenylalanine by the E coli and the
efficiency of the conversion of cinnamic acid to styrene could therefore be monitored. In a
subsequent experiment, the amount of daily supplemented phenylalanine was increased to 1.0
g for a concentration of 20 g/L (the maximum solubility of phenylalanine in water is 50 mg/L
at 25 °C31).

4.2. Results
The partition coefficients for styrene that were established between water and the four
solvents described previously are presented in Table 2. Hydrocarbon chain length was shown
to have only a slight impact on partition coefficient within the error boundaries observed. The
22

values obtained for hexadecane and BEHP were lower than those reported in a study
performed by Panke et al25, though the values were on the same order. BEHP demonstrated
the highest partition coefficient, meaning that BEHP should most efficiently remove styrene
from a fermentation broth.
Table 2: Partition Coefficients for Styrene between Water and Selected Solvents (% Error in
Parentheses)
Solvent

Hexadecane

BEHP

Hexane

Dodecane

K

1220 (+/- 5%)

1580 (+/- 9%)

1180 (+/- 5%)

1350 (+/- 4%)

In in the initial set of glass-capped shake flask experiments, the broths derived from one of the
three seed cultures exhibited very poor growth—about half the activity of the other cultures.
The two more productive cultures yielded remarkably similar results. Figure 11 tracks the
concentration of styrene in the organic extractors over the 72 hour lifetime of the first broth

Styrene Concentration in Solvent (mg/L)

(culture A1), while Figure 12 (page 24) does the same for the second broth (A2).

1.60E+03

1.20E+03

8.00E+02

4.00E+02

0.00E+00
0

24

48

72

Time (hrs)

Figure 11: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over 72 Hours for Broth A1

23

Styrene Concentration in Solvent (mg/L)

1.60E+03

1.20E+03

8.00E+02

4.00E+02

0.00E+00
0

24

48

72

Time (hrs)

Figure 12: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over 72 Hours for Broth A2

Cultures A1 and A2 both approach a maximum recovery of 1600 mg/L styrene when
hexadecane is the solvent.

Given the 10:1 media volume to solvent volume ratio, the

production rate of the cultures was thus only 160 mg/L. An even lower production rate of
between 120 and 130 mg/L was observed when BEHP was the solvent. The markedly
different trends in organic phase concentration with time are worth noting. The shape of the
hexadecane curves generally follows the bacterial growth cycle; namely, an exponential phase
proceeded by a lag phase and followed by a stationary phase. Meanwhile, the rate of styrene
accrual in BEHP is strongly linear with time. Perhaps the rate of styrene transfer to BEHP is
slower than to hexadecane and, given more time for transfer to occur, the accumulation in the
BEHP phase would plateau and more closely resemble the hexadecane curve.
It cannot be determined from Figures 11 and 12 if the maximum BEHP capacity for styrene is
greater than the styrene capacity of hexadecane. But it cannot be denied that over 72 hours,
contrary to what was expected based on the measured partition coefficients, hexadecane
proved to more effectively remove styrene from the fermentation broth. It was observed
during sampling that—when the flasks were stationary—the hexadecane layers would spread
evenly across the entire aqueous surface. Meanwhile, the BEHP would coalesce towards the
24

center of the flask, leaving a margin of uncovered broth around the flask circumference. The
BEHP may be so averse to interacting with the aqueous phase that sufficient mixing to fully
transfer styrene to the organic phase does not occur.
In the next set of experiments when phenylalanine was supplemented daily to the cultures at a
concentration of 5 g/L, the styrene yield increased marginally. Three broths were again
created. However, one culture again did not grow as well as the other two and so was not
factored in to the results of the experiment. The average styrene production from the two
producing cultures with hexadecane as a solvent was 230 mg/L, while that from the same
cultures in the presence of BEHP was 400 mg/L.

The calculated averages represent a

departure from the trend noted above in which hexadecane demonstrated superior recovery
capability. However, the variability of the averages was quite large—54% error with respect
to hexadecane and 63% error with respect to BEHP. When the variability was considered in
conjunction with the results of the experiment that followed (see below), the change in
behavior was not assumed to be significant.
The aqueous phase was also sampled for styrene, phenylalanine, and cinnamic acid when 5
g/L of phenylalanine were provided for the cultures daily. No trends were readily apparent
with respect to the choice of solvent and, unfortunately, consistency between cultures was not
observed. For instance, the respective concentrations of styrene and phenylalanine in the
aqueous media of Culture 1 remained fairly constant at 4 mg/L and 3 g/L with BEHP as a
solvent. But when Culture 2 was in contact with BEHP, styrene accumulated daily to a final
aqueous concentration of 26 mg/L in spite of phenylalanine concentrations remaining near 3
g/L. Likewise, the styrene concentration decreased every 24 hours in Culture 1 but increased
every 24 hours in Culture 2 when hexadecane was the solvent of choice. Culture to culture
variability thus made it difficult to draw any conclusions about the relationship between
extraction solvent and the aqueous concentrations of the compounds in the styrene pathway.
Additional trials should be performed to clarify the trends, or lack thereof, noted.
Further increasing the concentration of supplemented phenylalanine to 20 g/L resulted in even
greater styrene yields. As Figures 13 and 14 show, the styrene concentration in the organic

25

phase was between 10,000 and 12,000 mg/L when hexadecane was the solvent, translating to

Styrene Concentration in Sovlent (mg/L)

media production rates between 1 and 1.2 g/L.

1.40E+04
1.20E+04
1.00E+04
8.00E+03
6.00E+03
4.00E+03
2.00E+03
0.00E+00
0

24

48

72

Time (hrs)

Styrene Concentration in Sovlent (mg/L)

Figure 13: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over 72 Hours with 20 g/L Phenylalanine
Supplemented Every 24 Hours in Broth C2

1.40E+04
1.20E+04
1.00E+04
8.00E+03
6.00E+03
4.00E+03
2.00E+03
0.00E+00
0

24

48

72

Time (hrs)

Figure 14: Concentration of Styrene Extracted in Hexadecane (Blue) and BEHP (Red) Over 72 Hours with 20 g/L
Phenylalanine Supplemented Every 24 Hours in Broth C3

26

The remaining culture, C1, showed styrene concentrations in BEHP on the same order as
cultures C2 and C3, but a maximum concentration in hexadecane of 6000 mg/L. Thus, in two
of the three broths supplemented with 20 g/L of phenylalanine daily, hexadecane proved to be
the superior solvent. Unlike during the first set of extraction experiments, the concentration
of styrene in BEHP began to level off after a period of 72 hours. The increased rate of
production may have facilitated faster transfer of styrene to BEHP.
Most importantly, the maximum styrene titer observed was four times the toxicity limit of 300
mg/L. Increasing the concentration of supplemented phenylalanine up to the solubility limit
of phenylalanine in water may further increase styrene yield. From an industrial scale-up
standpoint, continuous addition of another chemical agent in such large quantities is not
desirable. But such an obstacle might be overcome in the future by working to engineer
NST74 to produce phenylalanine in even greater quantities, assuming such a task is possible.
At the very least, the results obtained in this study demonstrate that the flux through the
styrene pathway can be increased up to 1.2 g/L.

27

5. Adsorption
5.1. Styrene Oxide Recovery as Proxy for Styrene Recovery
To this point, the two main difficulties continually encountered with increasing styrene yields
from E. coli NST74 have been the high volatility of styrene and the low toxicity threshold of
NST74 for styrene. Initial pervaporation experiments met limited success because the system
could not be kept sufficiently closed to stop styrene from evaporating. Extraction had to be
performed with glass caps to avoid similar difficulties, which made supplementing oxygen to
the E. coli necessary to prevent premature cell death.
To circumvent the volatility and toxicity obstacles, attention was turned to the recovery of
styrene oxide from fermentation broth. Styrene oxide can be synthesized by E. coli NST74
through the addition of a single gene—styAB from Pseudomonas putida32—to the
pACYCDuet-1 plasmid. Compared to styrene, styrene oxide has a much lower volatility and
is not as toxic towards E coli (toxicity limit of 1.6 g/L32). Styrene oxide should thus be less
difficult to recover than styrene. Since the biochemical pathway for styrene oxide is identical
to that for styrene, merely lengthened by a step, the styrene oxide yield also represents a
theoretical styrene yield. Though the issues with styrene recovery would remain, a high flux
through the styrene/styrene oxide pathway would at least indicate that the recovery issues are
worth addressing. Adsorption was proposed as a method by which styrene oxide might be
most efficiently removed from fermentation media. The adsorbed styrene oxide could then be
recovered by washing the resin with a hydrophobic solvent.
5.2. Materials and Methods
Isotherms for styrene oxide and, for the sake of comparison, styrene in the presence of the
hydrophobic resin L-493 were established. The initial concentration of each compound was
varied from 100 to 500 mg/L at intervals of 100 mg/L. Two different masses of resin were
investigated at each starting concentration, generating ten total samples. In the first trial, the
resin masses were 1.0 g and 2.0 g. The masses were decreased to 0.1 g and 0.25 g in the
second trial. Two additional samples were created, each with a starting concentration of 300
mg/L and a resin mass of 1.0 g (Trial 1) or 0.1 g (Trial 2). These replicates were created for
28

the purpose of error analysis. A control vial with no resin was made for each compound to
determine if any solute was lost to the system through evaporation. The vials were oriented
horizontally and shaken for 24 hours to ensure establishment of equilibrium.
The choice of solvent for solute recovery from the L-493 adsorbent was dictated by two
factors. First, the solvent had to display greater affinity for styrene oxide than did the resin.
However, the solvent could not be so hydrophobic as to cause sequestering and clumping of
the resin due to the presence of water on some of the adsorption sites. Methanol was
therefore selected as the washing solvent. The suitability of methanol as a recovery solvent
was established by contacting 20 mL of a 300 mg/L solution of styrene oxide with 0.25 g of
resin for 24 hours. The final concentration in the aqueous phase was determined by HPLC
and used with Eq. 13 (page 31) to determine the theoretical amount of adsorbed styrene oxide.
The resin was subjected to two washes with methanol, each wash with a volume of 5 mL.
During each wash, the vial containing methanol and the resin was shaken vigorously for a
minute. The methanol was then analyzed by GC to determine the concentration of styrene
oxide in the methanol and the thus the amount of adsorbed styrene oxide recovered. An
identical procedure was performed with hexadecane—which is more hydrophobic than
methanol—as a solvent for the purposes of comparison.
Four sets of experiments were performed with L-493 resin in contact with active, styreneoxide producing cultures. All fermentation broths were created in the same manner as during
the extraction experiments. The pH of the cultures was buffered by the addition of 1 mL of
potassium phosphate 24 hours after transfer to minimal media environment. In the first and
second experiments, three seed cultures were used to create six flasks (two flasks per culture).
Resin was added to one of the two cultures in each set, while the other acted as a control. The
mass of resin added in the first experiment was 2.0 g, while the mass of resin added in the
second experiment was 0.75 g. In both cases, the resin was added at the same time that the
cultures were induced (8 hours after transfer to minimal media). The aqueous phase was
sampled periodically over the course of 72 hours by HPLC. In the first experiment, each of
the five methanol washes was 20 mL, while in the second experiment each wash was 10 mL.
The methanol washes were analyzed by GC. The optical density of the cultures at 600 nm
was recorded at 24 and 72 hours after transfer to minimal media in the second experiment. .
29

The purpose of the third and fourth experiments was to investigate the effect of resin addition
time on the styrene oxide recovery. Four flasks were created from one seed culture during
each experiment. One of the four flasks acted as a control. In the third experiment 1.0 g of
resin were added selectively to the other three flasks at 28, 32, and 36 hours after transfer to
minimal media (that is, 4, 8, and 12 hours after pH buffering). The resin was not added until
this time because the cultures do not begin producing styrene oxide at high rates until after
buffering has been performed. After 72 hours, the resin was subject to four methanol washes,
each with a volume of 10 mL. An identical mass of resin was added to the three flasks in the
fourth experiment, but the resin addition times were 36, 48, and 60 hours after transfer to
minimal media. The procedure for the fourth trial differed from the previous three in two
other respects.

First, ethanol was used as the washing solvent instead of methanol to

determine if the percent recovery of the adsorbed styrene oxide could be increased. Second,
the five wash samples (10 mL each) were analyzed by HPLC with the aqueous samples so as
to remove any platform-to-platform variability in measured styrene oxide concentration that
might have been encountered using the GC.

5.3. Results
An isotherm for styrene could not be constructed because of the high volatility of styrene.
The control vial possessed a final styrene concentration on the order of the vials which
contained resin. Adsorption, therefore, should not be considered a viable option for styrene
recovery unless a method can be found that will keep styrene in solution. More success,
though, was encountered with respect to styrene oxide.

The controls for styrene oxide

demonstrated negligible loss in comparison to the amount adsorbed. Figure 15 (page 31)
shows the resulting isotherm.
The corresponding data set was plotted according to Eq. 10 and Eq. 12, the linearized forms
of the Langmuir and Freundlich models, respectively. A line of best fit was determined in
each instance, and the Freundlich model demonstrated a stronger correlation. Based on the
slope and intercept of the linearized Freundlich equation, the model for styrene oxide was
constructed and is given by Eq. 13 (page 31).
30

9

-+ = 147.5) 9.FG

(13)

60
50

q (g/kg)

40
30
20
10
0
0

0.05

0.1

0.15

0.2

C, Equilibrium Concentration (g/L)

Figure 15: Isotherm at 32°C for Styrene Oxide Adsorption on L-493 (Error bars reported
to 1 standard deviation)

The equilibrium concentration of styrene oxide in the aqueous phase during the solvent test
was 0.027 g/L. The theoretical q for this final concentration according to Eq. 13 was 23.8
g/kg. The amount of styrene oxide adsorbed by 0.25 g of resin was therefore estimated to be
5.95 mg. According to GC analysis, two methanol washes removed 4.7 mg of styrene oxide
and two hexane washes removed 5.3 mg of styrene oxide. Methanol thus recovered 79% of
the adsorbed styrene oxide, while hexane recovered 89%.

However, during the hexane

washes it was observed that the resin clumped together and stuck to the sides of the vial,
which made mixing more difficult. Methanol was therefore selected as the wash solvent
under the assumption that performing additional washes would increase the percent recovery.
The first and second experiments performed with resin in contact with live cultures did not
yield satisfactory results. In both experiments the amount of styrene oxide remaining in the
aqueous phase after 72 hours was nearly undetectable. Based on the mass of styrene oxide
recovered via methanol, only 0.15 g/L (+/- 24%) of styrene oxide were produced with 2.5 g of
31

resin and only 0.015 g/L (+/- 18%) on average with 0.75 g of resin. Meanwhile, the controls
in each experiment produced 1.50 g/L and 1.35 g/L, respectively. After 24 hours, the average
optical density of the resin-containing flasks was 0.62 (+/- 13%) while that of the controls was
1.64 (+/-11%). Clearly, the resin somehow inhibited the growth and productivity of the
cultures. It was proposed that the rapid swirling of the resin subjected the cultures to a high
degree of shear stress, stifling bacterial growth in the lag phase. Subsequent experiments
therefore focused on adding resin to growing cultures several hours after induction to allow
the cells to reach the exponential phase without interference.
The third experiment offered an improvement over the previous two. The concentration of
styrene oxide in the aqueous phase after 72 hours was found to be independent of the resin
addition time when such time was between 4 and 12 hours after pH buffering. On average, the
final styrene oxide concentration was 0.085 g/L (+/- 4%). According the Eq. 13, the resin
theoretically adsorbed 0.043 g of styrene oxide. The maximum amount of styrene oxide
recovered after four methanol washes (each 10 mL) was 0.035 g, representing an 82%
recovery. However, even if 100% of the styrene oxide had been recovered, the total amount
of styrene oxide produced by the cultures would have been 0.84 g/L, less than the 1.03 g/L
generated by the control culture. Adsorption, therefore, still did not improve styrene flux
under the conditions investigated in the third experiment.
Further increasing the resin addition time to 48 and 60 hours after transfer to minimal media
resulted in styrene oxide titers nearly identical to those obtained in the previous trial. Unlike
the first two trials in which culture growth was clearly inhibited by the presence of resin, the
cultures in the final group of flasks did not show markedly different growth from the control
broth. Figure 16 (page 32) shows the optical density (OD) at 600 nm of the control and the
three resin containing broths. For each flask to which resin was added, the final OD at 72
hours was greater than the OD at 36 hours. Furthermore, Flasks 2 (48 hour addition time) and
3 (60 hour addition time) showed a final OD greater than that of the control flask. The OD of
Flask 1 (36 hour addition time) lagged behind the others by the time 72 hours has passed, at
which point the media in Flask 1 had been in contact with the resin for 36 hours. At this
point, however, the E. coli cells had begun to enter the death phase of their growth cycle, and
so the small downturn in OD at this point should not be a significant concern.
32

4

Optical Density at 600 nm

3.5
3
2.5
2
1.5
1
0.5
0

36

48

60

72

Time After Transfer to Minimal Media (Hours)

Figure 16: Optical Density at 600 nm of Styrene Oxide Producing Cultures. Blue = Control; Red = Resin added at
36 hours to Flask 1; Green = Resin added at 48 hours to Flask 2; Purple = Resin added at 60 hours to Flask 3

The average final concentration of styrene oxide in the aqueous phase was approximately 0.1
g/L (+/- 17%) and appeared to be nearly independent of resin addition time, as Figure 17
(page 34) shows.

Figure 17 also demonstrates that the styrene oxide production rate was

uniform up to the time that resin was added to each of the three experimental broths.

The

sharp decrease in styrene oxide concentration in Flask 1 between 36 and 48 hours shows that
equilibrium was established between the resin and the media in at most 12 hours. Similar
equilibrium settling times were observed for Flask 2 (between 48 and 60 hours) and for Flask
3 (between 60 and 72 hours). A potentially promising trend that was unfortunately not
observed was a rebound in aqueous styrene oxide concentration following the minimum
achieved after resin addition. Such a concentration increase would have indicated that the
adsorption capability of the resin had been exhausted and that the NST74 cultures were still
producing styrene oxide. However, the results in Figure 17 suggest instead that removal of
styrene oxide from the fermentation broth does not spur the cultures on to further
productivity.

33

Styrene Oxide Concentration (g/L)

1.4
1.2
1
0.8
0.6
0.4
0.2
0
30

36

48

60

72
Time After Transfer to Minimal Media (Hours)
Figure 17: Aqueous Styrene Concentration in Flasks Over 72 Hours. Blue = Control, Red = Resin Added at 36 Hours
to Flask 1, Green = Resin Added at 48 Hours to flask 2, Purple = Resin Added at 60 Hours to Flask 3

Finally, ethanol proved to be a more efficient solvent than methanol for recovery of adsorbed
styrene oxide. Figure 18 (page 35) shows that three washes of 10 mL were sufficient to
recover the bulk of adsorbed solute. The results also reveal that the resin addition time did
not have an impact on the amount of styrene oxide that was adsorbed by the L-493. All told,
an average of 0.044 g (+/- 4%) of styrene oxide was recovered. Based on the average final
aqueous concentration, Eq. 13 predicts a maximum theoretical recovery of 0.046 g styrene
oxide. Thus, 95% theoretical recovery was achieved by using ethanol as the wash solvent,
corresponding to an average styrene production of 0.88 g/L on the part of the media. But the
control exhibited a total production of 1.3 g/L, as Figure 17 indicates. Adsorption and
recovery of styrene oxide from fermentation media using L-493 resin and ethanol was
therefore demonstrated to be a feasible separation technique, but achievable titers in the
presence of L-493 were not greater than titers without resin.

34

0.03

Mass Styrene Oxide (g)

0.025
0.02
0.015
0.01
0.005
0
1

2

3

4

5

Wash

Figure 18: Mass of Styrene Oxide Desorbed from L-493 Resin over Five 10 mL Ethanol Washes. Blue = Resin added
at 36 hours to Flask 1, Red = Resin added at 48 hours to Flask 2, Green = Resin added at 60 hours to Flask 3

35

6. Conclusions and Recommendations
It has been previously demonstrated that the commodity chemical styrene can be synthesized
by genetic engineering of the phenylalanine overproducer E. coli NST74. Achievable styrene
titers, however, are limited to 300 mg/L due to the low toxicity threshold of E. coli NST74 for
styrene. In this study, the separation techniques of pervaporation and extractive fermentation
were investigated as potential methods of improving styrene yields from active E. coli
cultures. Additionally, adsorption using the hydrophobic resin L-493 was explored as a
method of recovering styrene oxide—synthesized by the addition of one step to the styrene
pathway—in an effort to demonstrate that enhanced flux through the styrene pathway was
possible.
Pervaporation experiments focused on two configurations. A hollow-fiber configuration with
two different membrane materials (polypropylene and polyethylene/polyurethane composite)
did not demonstrate an ability to separate styrene from a binary styrene-water solution. The
system could not be kept sufficiently closed to prevent styrene from leaving through
evaporation.

The second configuration and material combination analyzed was a flat

polydimethylsiloxane (PDMS) sheet with cross-flow.

Experiments with PDMS in full

pervaporation mode were not performed due to unavailability of the required apparatus.
However, fouling of PDMS due to the formation of E. coli NST74 biofilm, both in the
presence and absence of styrene, was characterized. Scanning electron microscopy showed
that fouling is less severe when styrene is present, which bodes well for future pervaporation
experiments with PDMS.

Once all equipment is available, future work with PDMS

membranes should investigate the ability of PDMS to separate dilute binary solutions of
styrene and water.

If the results are favorable, research can progress to characterizing

separations of active, styrene-producing fermentation broths, and the achievable titers can be
compared to those obtained using extraction and adsorption.
Bis-(2-ethylhexyl)-phthalate and hexadecane were then evaluated as potential solvents for
styrene recovery from fermentation broths of E. coli NST74. The goal was to increase
achievable styrene titers by continuously removing styrene from the fermentation media to
avoid surpassing the styrene toxicity threshold of the E. coli. Though a lower partition
36

coefficient was calculated for hexadecane than for BEHP, hexadecane was shown to extract
more styrene than BEHP when contacted with bacterial media, in most trials.

Initial

experiments in minimal media exhibited a maximum production of 160 mg/L of styrene,
obtained when hexadecane was the solvent. Supplementing the fermentation broth with 20
g/L of phenylalanine every 24 hours was shown to increase styrene yields to 1.2 g/L (again
with hexadecane as a solvent). The concentration of phenylalanine could theoretically be
increased up to the solubility limit of phenylalanine in water, though such a method would not
be sound from a scale-up perspective. But the increased styrene titer demonstrated the
potential for further improvement in the process. One possible step that might be taken in the
future is attempting to engineer NST74 to produce greater amounts of phenylalanine, thereby
removing the need to externally supplement the amino acid.
Adsorption of styrene oxide using L-493 resin did not lead to a greater styrene flux than was
obtained using extractive fermentation with supplemented phenylalanine. Adding the resin
upon induction of NST74 cultures proved to be detrimental to the growth of the cultures.
Higher titers were obtained when the resin was added after the cultures had been allowed to
grow uninhibited for several hours. One gram of resin was added at five different times after
transfer to minimal media: 28, 32, 36, 48, and 60 hours. Ethanol was shown to be an
effective solvent for recovery of adsorbed styrene oxide, demonstrating 95% recovery when
five washes were performed. The resin addition time was not shown to have a significant
impact on the final styrene oxide concentration in the aqueous phase or on the mass of styrene
oxide adsorbed. Equilibrium styrene oxide concentrations in the aqueous phase consistently
fell between 0.08 and 0.11 g/L, meaning that the theoretical mass adsorbed was
approximately 0.44 g. Between the adsorbed styrene oxide and the styrene oxide remaining
in the aqueous phase, titers never exceed 0.88 g/L. The titer in a control flask without resin
was 1.3 g/L.
Of the three separation techniques studied, pervaporation deserves the most future
consideration. Comparatively little pervaporation data was collected in this study, but other
studies have documented the potential benefits of the method. Furthermore, a ceiling on
achievable styrene flux was encountered using extraction and adsorption, and that ceiling is
not likely to be significantly raised unless NST74 can be engineered to further overproduce
37

phenylalanine. Demonstrating that titers above the toxicity threshold of E. coli for styrene are
possible was a positive result, but further work will have to be performed before biosynthetic
styrene becomes feasible on a grander scale.

38

Works Cited
[1] HIS (2011) Polystyrene, http://www.ihs.com/products/chemical/planning/ceh/
polystyrene.aspx (IHS Chemical, Englewood, CO)
[2] HIS (2009) Styrene, http://www.ihs.com/products/chemical/planning/ceh/styrene.aspx
(IHS Chemical: Englewood, CO)
[3] McMichael, Kirk, Organic chemistry pages from Washington State University hosted by
California State University, Dominguez Hills,
http://chemistry2.csudh.edu/rpendarvis/Polymer.html (California State University:
Dominguez Hills, CA).
[4] Lewis J.R. (2007) Hawley’s Condensed Chemical Dictionary (15 ed.), Wiley and Sons:
Hoboken, NJ.
[5] Global Business Intelligence (2011) Styrene global markets to 2020—substitution of
polystyrene by polypropylene and polyethylene terephthalate to impact growth in
developed regions,
[6] Ward A.L., Roberts J.W. (1951) Styrene, Interscience Publishers: New York, NY
[7] McKenna R and Nielsen D.R. (2011) Styrene biosynthesis from glucose by engineered E.
coli. Metabolic Engineering 13: 544-554.
[8] Peng M, Vane L, Liu S (2003) Recent advances in VOCs removal from water by
pervaporation. Journal of Hazardous Materials B98: 69-90
[9] Vane L and Alvarez F (2002) Full-scale Vibrating Pervaporation Membrane Unit: VOC
Removal From Water and Surfactant Solutions. Journal of Membrane Science 202:
177-193
[10] Yeom C.K., Kim H.K., and Rhim J.W. (1999) Removal of trace VOCs from water
through PDMS membranes and analysis of their permeation behaviors. Journal of
Applied Polymer Science 73: 601-611.
[11] Neel J, in: Huang R.Y.M. (1991), Pervaporation Membrane Separation Processes,
Elsevier, Amsterdam, The Netherlands, p. 1.
[12] Ji W, Sikdar S, and Hwang S (1994) Modeling of multicomponent pervaporation for
removal of volatile organic compounds from water. Journal of Membrane Science 93:
1-19.

39

[13] Mishima S. and Nakagawa T (2000) Sorption and diffusion of volatile organic
compounds in fluoroalkyl methacrylate-grafted PDMS membrane. Journal of Applied
Polymer Science 75: 773-783.
[14] Dutta K. and Sikdar S (1999) Separation of Volatile Organic Compounds from Aqueous
Solutions by Pervaporation Using S-B-S Copolymer Membranes. Environmental
Science and Technology 33: 1709-1716.
[15] Jou J, Yoshida W, and Cohen Y (1999) A novel ceramic-supported polymer membrane
for pervaporation of dilute organic compounds. Journal of Membrane Science 162:
269-284.
[16] Psaume, R., Aptel P, Aurelle Y, Mora J.C., and Bersillon J.L. (1988) Pervaporation:
importance of concentration polarization in the extraction of trace organics from
water. Journal of Membrane Science 36: 373-384.
[17] Urtiaga A.M., Gorri E.D., Beasley J.K., and Ortiz I (1999) Mass transfer analysis of the
pervaporative separation of chloroform from aqueous solutions in hollow fiber
devices. Journal of Membrane Science 156: 275-291.
[18] Nijhuis H.H., Mulder M.H.V., and Smolders C.A. (1991) Removal of trace organics
from aqueous solutions and effect of membrane thickness. Journal of Membrane
Science 61: 99-111.
[19] Ghoreyshi A.A., Jahanshahi M, and Peyvandi K (2008) Modeling of volatile organic
compounds removal from water by pervaporation process. Desalination 222: 410-418.
[20] Seader J.D. and Henley E.J. (2006) Separation Process Principles (2nd Ed.), Wiley &
Sons: Hoboken, NJ
[21] Diaz, Mario (1988) Three-phase extractive fermentation. Trends in Biotechnology 6:
126-130.
[22] Shi Z, Zhang C, Chen J, and Mao Z (2005) Performance evaluation of acetone-butanol
continuous flash extractive fermentation process. Bioprocess Bisyst Eng 27: 175-183.
[23] Ishii S, Taya M, Kobayashi T (1985) Monitoring and control for extractive fermentation
of Chlostridium acetobutylicum. Journal of Fermentation Technology 63(2): 181-187
[24] Shi Z, Shimizu K, Iijima S et al (1990) Theoretical development and performance
evaluation for extractive fermentation using multiple extractants. Biotechnology
Bioengineering 36: 520-529
40

[25] Panke S , Wubbolts M, Schmid A, and Witholt B (2000) Production of enantiopure
styrene oxide by recombinant Escherichia coli synthesizing a two-component styrene
monooxygenase. Biotechnology and Bioengineering 69: 91-100.
[26] Armenante P. Adsorption, http://cpe.njit.edu/dlnotes/CHE685/Cls11-1.pdf (New Jersey
Institute of Technology: Newark, NJ).
[27] Hameed B.H., Mahmoud D.K, and Ahmad A.L. (2008) Equilibrium modeling and
kinetic studies on the adsorption of basic dye by a low-cost adsorbent: Coconut (Cocos
nucifera) bunch waste. Journal of Hazardous Materials 158: 65-72.
[28] Chung J, Nerenberg R, and Rittmann B (2006) Bioreduction of selenite using a
hydrogen-based membrane biofilm reactor. Environmental Science and Technology
40: 1664-1671
[29] Tang Y, Zhou C, Van Ginkel S. W., Ontiveros-Valencia A, Shin J, and Rittman B (2012)
Hydrogen permeability of the hollow fibers used in the H2-based membrane biofilm
reactors. Journal of Membrane Science 407-408: 176-183.
[30] Qi W.W., Vannelli T, Breinig S, Ben-Bassat A, Gatenby A.A., Haynie S.L., and
Sariaslani F.S. (2007) Functional expression of prokaryotic and eukaryotic genes in
Escherichia coli for conversion of glucose to p-hydroxystyrene. Metabolic
Engineering 9: 268-276.
[31] Chemical Book (2010) Phenylalanine. http://www.chemicalbook.com/
ChemicalProductProperty_EN_CB2488691.htm
[32] McKenna R (2013) Presentation to Nielsen Research Group at Arizona State University.

41

Appendix A: Raw Data for Pervaporation Studies
Table 3: Styrene Concentration in Experiments with Hollow Fiber Membranes (1
hour)
Concentration (g/L) by Membrane (See below for key)
Time
A
B
C
D
(min)
0
0.21
0.09
0.22
0.17
3
0.21
0.15
0.22
0.18
6
0.21
0.18
0.22
0.18
10
0.21
0.18
0.21
0.19
20
0.21
0.19
0.21
0.18
30
0.19
0.18
0.19
0.17
45
0.18
0.17
0.18
0.16
60
0.18
0.15
0.17
0.14

Table 4: Styrene Concentration in Aqueous Phase of Controlled Experiments with No
Membrane (2 hours)
Time (min)
0
10
40
70
90
120
Concentration (g/L)
0.19
0.19
0.19
0.19
0.19
0.19

Key
•

A = Polypropylene, 29 fibers, 18 cm long

•

B = Polypropylene, 5 fibers, 18.5 cm long

•

C = Composite Polyethylene/Polyurethane, 5 fibers, 18.3 cm long

•

D = Composite Polyethylene/Polyurethane, 10 fibers, 13.2 cm long

42

Appendix B: Raw Data for Determination of Partition Coefficients
Table 5: Partition Coefficients for Styrene between Hexadecane and Water When 5 mL of a
50 mL Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying Volumes of
Double Deionized Water
Vol. Water (mL)
Organic C (mg/L)
Aqueous C (mg/L)
K
5
33900
32.6
1040
5
33700
33.7
1000
10
38900
32.8
1180
10
39000
32.0
1220
25
27800
24.1
1150
25
27800
24.8
1170
50
33300
44.6
746
50
33200
29.1
1150
75
28500
16.3
1740
75
28400
16.2
1750
100
38400
47.9
800
100
38300
32.7
1170
200
28400
13.7
2080
200
28500
16.7
1700
500
37500
93.7
400
500
37300
29.2
1250

Table 6: Partition Coefficients for Styrene between Hexane and Water When 5 mL of a 50
mL Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying Volumes of Double
Deionized Water
Vol. Water (mL)
Organic C (mg/L)
Aqueous C (mg/L)
K
5
43340
44.7
969
5
43225
33.0
1309
10
43275
32.1
1348
10
43240
33.7
1283
25
44215
35.5
1245
25
44525
35.6
1250
50
44665
42.0
1063
50
44770
43.6
1027
100
43650
34.2
1277
100
43550
38.7
1126

43

Table 7: Partition Coefficients for Styrene between BEHP and Water When 5 mL of a 50 mL
Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying Volumes of Double
Deionized Water
Vol. Water (mL)
Organic C (mg/L)
Aqueous C (mg/L)
K
5
47200
29.5
1600
5
43500
24.5
1770
10
45100
21.0
2150
10
42800
17.8
2410
25
37500
22.2
1690
25
34800
24.4
1430
50
43900
225
195**
50
42000
52.0
808
75
43800
29.5
1480
75
41400
26.5
1560
100
47500
26.2
1820
100
41900
613
68.3**
125
29200
11.9
2440
125
36100
47.3
762
500
39200
9.4
4620**
500
20.6
1900

Table 8: Partition Coefficients for Styrene between Dodecane and Water When 5 mL of a 50
mL Hexadecane Solution with 1.7 mL Styrene is Contacted with Varying Volumes of Double
Deionized Water
Vol. Water (mL)
Organic C (mg/L)
Aqueous C (mg/L)
K
5
40150
32.6
1231
5
40125
32.4
1238
10
40000
33.6
1189
10
39850
90.9
438**
25
39750
28.7
1384
25
39520
56.1
704**
50
39400
29.2
1348
50
39900
25.5
1567
100
40525
27.6
1467
100
39620
26.2
1514

**Partition Coefficients were not included in final analysis because they were greater than 1.5
times the inter-quartile range from the median and were thus considered outliers.

44

Appendix C: Raw Data for Extractive Fermentation Studies
Experiment 11
Table 9: Styrene Concentration (mg/L) in Organic Phase in First Preliminary Extraction
Experiment (10 mL solvent: 50 mL minimal media)
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
1
2324
0
1810
2
2181
1000
1750
3
0
76
648
BEHP
1
2915
0
2926
2
3025
2915
2431
3
0
0
525

Experiment 21
Table 10: Styrene Concentration (mg/L) in Organic Phase in Second Preliminary Extraction
Experiment (10 mL solvent: 50 mL minimal media)
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
1
0
56
720
2
66
955
925
3
25
29
30
BEHP
1
15
54
2090
2
873
1309
1397
3
20
0
646

Experiment 3
Table 11: Styrene Concentration (mg/L) in Organic Phase in Extraction Experiments with
Cultures A1, A2, and A3 (5 mL solvent: 50 mL minimal media). Note decreased productivity
of A3 compared to A1 and A2
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
A1
995
1535
1565
A2
845
1445
1540
A3
478
640
720
BEHP
A1
475
889
1293
A2
521
857
1188
A3
420
597
851

1

Preliminary results not used in body of paper due to problems with sampling errors. Errors were corrected by
Experiment 3

45

Experiment 4
Table 12: Styrene Concentration (mg/L) in Organic Phase in Extraction Experiments with
Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with 0.25 g phenylalanine
supplemented every 24 hours. B2 produced at significantly lower rates than B1 and B3
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
B1
820
1120
1205
B2
496
501
510
B3
978
1891
1776
BEHP
B1
550
2440
3333
B2
474
830
996
B3
815
1357
1869

Table 13: Styrene Concentration (mg/L) in Aqueous Phase in Extraction Experiments with
Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with 0.25 g phenylalanine
supplemented daily. Data for B3 not included because it did not produce at same level as
other two cultures
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
B1
5.3
0.4
0.0
B3
4.4
14.8
21.1
BEHP
B1
4.1
4.1
4.2
B3
10.2
24.4
26.5

Table 14: Phenylalanine Concentration (g/L) in Aqueous Phase in Extraction Experiments
with Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with 0.25 g
phenylalanine supplemented daily. Data for B3 not included because it did not produce at
same level as other two cultures
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
B1
2.6
0.2
0.1
B3
0.0
3.0
3.4
BEHP
B1
2.6
3.4
2.9
B3
2.4
0.1
3.3

46

Table 15: Cinnamic Acid Concentration (mg/L) in Aqueous Phase in Extraction Experiments
with Cultures B1, B2, and B3 (10 mL solvent: 50 mL minimal media) with 0.25 g
phenylalanine supplemented daily. Data for B3 not included because it did not produce at
same level as other two cultures
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
B1
4.8
0.9
0.1
B3
0.1
66.9
162.9
BEHP
B1
3.9
58.3
28.8
B3
4.6
0.1
122.7

Experiment 5
Table 16: Styrene Concentration (mg/L) in Organic Phase in Extraction Experiments with
Cultures C1, C2, and C3 (5 mL solvent: 50 mL minimal media) with 1.0 g phenylalanine
supplemented every 24 hours
Culture
24 Hours
48 Hours
72 Hours
Hexadecane
B1
533
4405
4965
B2
605
9350
10250
B3
785
11150
12050
BEHP
B1
397
7656
7931
B2
692
6380
7183
B3
640
5841
6985

47

Appendix D: Raw Data for Establishment of Styrene Oxide Isotherm on L493 Resin

Table 17: Isotherm Data for Styrene Oxide on L-493 Resin (All samples in 20 mL water) at
32 °C
Resin Mass (g)
0.103
0.108
0.098
0.104
0.116
0.247
0.25
0.25
0.253
0.253
0.105
0.101

C0 (g/L)

Cf (g/L)
0.1
0.2
0.3
0.4
0.5
0.1
0.2
0.3
0.4
0.5
0.3
0.3

q (g/kg)
0.010
0.041
0.089
0.134
0.174
0.005
0.014
0.027
0.040
0.059
0.086
0.088

17.5
29.4
43.1
51.1
56.2
7.7
14.9
21.8
28.5
34.8
40.8
42.0

5

4

ln (q)

3

2

y = 0.5184x + 4.9939
R² = 0.9681

1

0
-6

-5

-4

-3

-2

-1

ln (C)

Figure 19: Isotherm data plotted according to linearized Freundlich Equation
48

0

0.0035
0.003

C/q

0.0025
0.002
y = 0.0146x + 0.0007
R² = 0.9652

0.0015
0.001
0.0005
0
0

0.05

0.1

0.15

Concentration (g/L)

Figure 20: Isotherm data plotted according to linearized Langmuir Equation

49

0.2

Appendix E: Raw Data for Adsorption Studies with Styrene Oxide
Producing NST74 E. coli

Experiment 1
Table 18: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Five Methanol
Washes (20 mL each). 2.5 g L-493 were added to each culture upon transfer to minimal
media.
Wash
Culture 1
Culture 2
Culture 3
1
0.0021
0.0030
0.0023
2
0.0024
0.0038
0.0027
3
0.0011
0.0018
0.0013
4
0.0004
0.0007
0.0005
5
0.0002
0.0003
0.0002

Experiment 2
Table 19: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Five Methanol
Washes (10 mL each). 0.75 g L-493 were added to each culture upon transfer to minimal
media.
Wash
Culture 1
Culture 2
Culture 3
1
0.0002
0.0001
0.0001
2
0.0004
0.0003
0.0003
3
0.0002
0.0002
0.0002
4
0.0001
0.0001
0.0001
5
0.0001
0.0000
0.0000

Table 20: Optical Densities at 600 nm of Cultures with 0.75 g L-493 Resin added upon
transfer to minimal media compared to controls
Culture
OD at 24 Hours
OD at 72 Hours
Resin Flasks
1
0.6656
0.663
2
0.6744
0.6428
3
0.5316
0.4638
Controls
1
1.834
2.238
2
1.566
1.736
3
1.496
1.75

50

Experiment 3
Table 21: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Four Methanol
Washes (10 mL each). 1.0 g L-493 were added at three different times (28, 32, and 36 hours)
after transfer to minimal media environment. All flasks were derived from same culture.
Wash
28 Hour Flask
32 Hour Flask
36 Hour Flask
1
0.008
0.010
0.003
2
0.015
0.013
0.003
3
0.009
0.008
0.001
4
0.004
0.004
0.001

Table 22: Aqueous phase concentration of styrene oxide (g/L) over a period of 72 hours in
flasks to which 1.0 g resin were added at three different times (28, 32, and 36 hours) after
transfer to minimal media. Bad sample likely at 72 hours for control
Flask
24 hours
48 hours
72 hours
Control
0.007
1.028
0.010
28 Hour Addition
0.023
0.003
0.087
32 Hour Addition
0.010
0.074
0.082
36 Hour Addition
0.217
0.022
0.086

Experiment 4
Table 23: Mass of Adsorbed Styrene Oxide Removed (g) from L-493 Using Five Ethanol
Washes (10 mL each). 1.0 g L-493 were added at three different times (36, 48, and 60 hours)
after transfer to minimal media.
Wash
Culture 1
Culture 2
Culture 3
1
0.0247
0.0257
0.0257
2
0.0125
0.0137
0.0131
3
0.0033
0.0040
0.0040
4
0.0011
0.0011
0.0011
5
0.0003
0.0003
0.0003

51

Table 24: Aqueous phase concentration of styrene oxide (g/L) over a period of 72 hours in
flasks to which 1.0 g resin were added at three different times (36, 48, and 60 hours) after
transfer to minimal media.
Flask
30 hours
36 hours
48 hours
60 hours
72 hours
Control
0.80316
1.0053
1.2699
1.30392
1.30644
36 Hour
Addition
0.80298
1.03176
0.07524
0.07866
0.08298
48 Hour
Addition
0.80946
1.01214
1.27512
0.10188
0.10692
60 Hour
Addition
0.86706
1.07784
1.3068
1.3176
0.1161

Table 25: Optical Density at 600 nm over a period of 72 hours of cultures to which 1.0 g
resin were added at three different times (28, 32, and 36 hours) after transfer to minimal
media.
Time
Control
36 Hour
48 Hour
60 Hour
Addition
Addition
Addition
36 Hours
2.812
2.758
2.779
2.976
48 Hours
3.142
3.161
3.246
2.914
60 Hours
3.102
3.027
3.316
2.959
72 Hours
3.237
2.943
3.331
3.41

Table 26: Mass of Styrene Oxide Removed (g) from L-493 in Two Washes with Selected
Solvents (Theoretical Mass Adsorbed = 0.006 g)
Wash 1

Wash 2

% Recovery

Hexane

0.0014

0.0039

88%

Methanol

0.0036

0.0011

79%

52

Appendix G: Calibration Curves

0.6
y = 0.00003833x
R² = 0.96909779

Concentration (g/L)

0.5
0.4
0.3
0.2
0.1
0
0

2000

4000

6000
8000
10000
Peak Area at 258 nm

12000

14000

Figure 21: Calibration Curve for Aqueous Styrene on HPLC

2.5
y = 0.000096431x
R² = 0.967392562

Concentration (g/L)

2

1.5

1

0.5

0
0

5000

10000
Peak Area at 215 nm

15000

Figure 22: Calibration Curve for Aqueous Phenylalanine on HPLC
53

20000

0.9
0.8

y = 0.000052x
R² = 0.972488

Concentration (mg/L)

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

5000

10000
Peak Area at 258 nm

15000

20000

Figure 23: Calibration Curve for Aqueous Cinnamic Acid on HPLC

Styrene Oxide Concentration (g/L)

0.7
0.6

y = 0.000115x
R² = 0.996934

0.5
0.4
0.3
0.2
0.1
0
0

1000

2000
3000
4000
Peak Area (6.1 min)

5000

Figure 24: Calibration Curve for Aqueous Styrene Oxide on HPLC

54

6000

14000

Styrene Concentration (mg/L)

12000

y = 0.0005x
R² = 0.9828

10000
8000
6000
4000
2000
0
0.00E+00

5.00E+06

1.00E+07

1.50E+07

2.00E+07

2.50E+07

3.00E+07

Peak Area (1.37 min)

Figure 25: Calibration Curve for Styrene in Hexadecane on GC

Styrene Concentration (mg/L)

14000
12000

y = 0.0011x
R² = 0.9808

10000
8000
6000
4000
2000
0
0.00E+00

2.00E+06

4.00E+06

6.00E+06

8.00E+06

1.00E+07

1.20E+07

Peak Area (1.31 min)

Figure 26: Calibration Curve for Styrene in BEHP on GC. Samples were diluted with
hexane to final composition of 50% BEHP solution and 50% hexane

55

10000

Styrene Concentration (mg/L)

y = 0.0005x
R² = 0.9881
8000

6000

4000

2000

0
0.00E+00

5.00E+06

1.00E+07

1.50E+07

2.00E+07

2.50E+07

Peak Area

Figure 27: Calibration Curve for Styrene in Hexane on GC

Styrene Concentration (mg/L)

10000
y = 0.0005x
R² = 0.9903

8000

6000

4000

2000

0
0.00E+00

5.00E+06

1.00E+07
Peak Area

1.50E+07

Figure 28: Calibration Curve for Styrene in Dodecane on GC

56

2.00E+07

Styrene Oxide Concentration (g/L)

60
y = 9E-06x
R² = 0.954
50
40
30
20
10
0
0

1000000

2000000

3000000

4000000

5000000

Peak Area (2.25 min)

Figure 29: Calibration Curve for Styrene Oxide in Hexane on GC

57

6000000