(381g) Combinatorial Design of Selective Gel Materials for Extractive Fermentation

Due
to limited global supplies of fossil fuel resources, rising prices, and
environmental problems, future energy needs must be met in part or entirely by
fuels derived from renewable resources.  
Production of ethanol (EtOH) and n-butanol (BuOH) from non-food
feedstocks, namely cellulosic biomass, has received much attention.  If cellulosic bio-alcohols are to become an economically
sustainable source of fuel in the future, drastic improvements to process
economics must be realized, however.  One
of the main problems facing the biofuels industry is the fact that
fermentations of cellulose-derived sugar mixtures tend to produce very low
concentrations of the alcohol, which in turn leads to a high energy cost for biofuel
recovery by traditional multi-stage distillation.  For example, typical concentrations of BuOH
obtained are less than 20 g/L due to the toxicity of the BuOH, while typical
concentrations of EtOH are much lower than the values found in pure sugar
fermentations, due to the presence of toxic inhibitor compounds in
cellulose-derived sugars. Thus, design of alternative separation strategies
such as pervaporation and extractive fermentation, which are amenable to low
alcohol concentrations, remains a challenge.

Because
liquid-liquid extraction suffers from fouling and emulsion problems, approaches
based on polymeric membranes or absorbent solids are favored for alcohol
recovery.  The usual problem in designing
materials for selective transport applications is a trade-off between
selectivity for the desired product and solubility of the product, however.  In other words, materials that absorb
significant quantities of the alcohol usually have poor selectivity, absorbing
significant amounts of water.  As a
result, the definition of an optimum material composition for pervaporation requires
consideration of both transport rates and product purity.  Ultimately, the material must transport alcohol
at a high enough rate to maintain reasonable productivity, while maintaining
low enough water content to reduce the cost of subsequent dehydration processes. 

This
work addresses a novel high-throughput screening approach for designing selectively
swelling copolymer gel materials for use in extractive fermentation processes.   The
overall approach is based upon polyacrylate random copolymers in which one
monomer (H) is more hydrophilic than the alcohol, and the other (B) is more
hydrophobic, so that the cohesive energy density of the copolymers covers a
wide range.  The gels that absorb the
greatest amount of alcohol are expected to be those having cohesive energy
density similar to that of the targeted alcohol molecule.  However, the gel that is most selective for
alcohol in a mixed solvent situation will likely have a different composition,
as more hydrophobic materials are less likely to absorb water.  Thus, selection of an optimal material for
extractive fermentation requires a more thorough analysis of the volumetric
productivity of the reactor.  To support
such analysis, a high throughput screening approach has been developed to
rapidly generate information concerning swelling, selectivity, and distribution
coefficient of copolymer gel materials, information that is invaluable in
process modeling.  

A
combinatorial matrix of gels is prepared, having orthogonal gradients in two
key parameters, such as the mole fraction of the hydrophilic monomer (x_H),
and the mole fraction of acrylates belonging to the crosslinker molecule (x_C).  Testing of equilibrium swelling
characteristics in pure and mixed solvents provides a library of information
regarding the parameters influencing performance in separations applications.  Typical results of the high-throughput
screening methodology are illustrated in Fig. 1.  A series of gels was prepared by
co-polymerizing hydrophobic (B = butyl acrylate) and moderately hydrophilic (H
= hydroxyethyl acrylate) chemical units. 
 All gels were thereafter swollen
in pure water and pure alcohols (n-butanol, ethanol) to obtain information
about equilibrium swelling properties.  Equilibrium
swelling data reveal that some of the gels have very high affinity for alcohol
and low affinity for water (data shown are for EtOH).  For example, gels having x_B = 0.7
to 0.9 and x_C  = 0.008 gained
more than 1300% of their initial (dry) mass in EtOH, while gaining less than 5%
their mass when swollen in pure water.  However,
examination of equilibrium swelling properties in pure components is not
sufficient to determine an "optimal" material for a separation process.
 The performance of the material in a mixed
solvent system (water + alcohol) may differ greatly, due to the greatly lowered
chemical potential of the alcohol molecules in the external liquid mixture.

To
determine values of the alcohol distribution coefficient and the selectivity
for alcohol, additional combinatorial swelling experiments are therefore conducted
in solvent mixtures of interest (e.g. 20 g/L BuOH or 100 g/L EtOH).  The composition of the external liquid at
equilibrium is measured by high-throughput HPLC.  The equilibrium gel masses are also recorded, and
a mass balance on the system permits rapid computation of selectivity and
distribution coefficient of each gel.  Fig.
2 shows calculated selectivities for the test system.  The usual trade-off between selectivity and absorption
capacity is apparent by comparing Figs. 2 and 3.   Gels that absorb significant amounts of liquid,
which are rich in the hydrophilic monomer, exhibit poor selectivity for the
alcohol.  Highly selective gels consisting
of mostly the hydrophobic (B) component absorb too little EtOH to be useful in
a practical sense.  In addition, it is
apparent that the gels that appeared optimal based upon pure solvent swelling (Fig.
1) actually have no special characteristics in mixed-solvent selectivity tests
(Fig. 2).  While the combinatorial approach
produces a rich library of information concerning competitive swelling characteristics
in the water-alcohol mixtures, no "optimal" material is identified a
priori. Thus, it is apparent that selection of an "optimal" material
requires consideration of the fermentation process characteristics. 

A
material optimization protocol is demonstrated for extractive fermentation by
considering the volumetric productivity of a batch fermentor containing an
absorbent gel, which depends both on the volume fraction of the absorbent
material and the distribution coefficient for alcohol.  While the absorbent material continuously
removes alcohol from the system during fermentation, increasing the volumetric
productivity of the liquid phase, the absorbent material also displaces some portion
of the liquid phase, decreasing the effective fermentation volume.  The optimum material maximizes the overall volumetric
productivity of the system in units of (kg alcohol) / [(L reactor volume)*h].  It can be shown that an effective absorbent
material must have a distribution coefficient for the alcohol, D_a, significantly
greater than 1 to achieve a significant improvement in volumetric productivity.  Previous studies of extractive fermentation have
often neglected this criterion, focusing only on the total amount of alcohol
produced from a fixed liquid volume and ignoring the displacement of liquid by
the absorbent material in a vessel of fixed size. 

In
practice, obtaining an absorbent material with D_a > 1 is
difficult and in fact, none of the materials in the test matrix shown in Figs. 1-3
meets this criterion, despite the attractive discrepancy in pure solvent swelling
observed.  Two strategies have recently been
pursued for obtaining polymer gel materials having D_a > 1.  First, crosslinking a highly selective
material in a solvent was explored to obtain gels having fewer trapped
entanglements, increasing the equilibrium swelling ratio in the mixed solvent
systems.  Further combinatorial studies
of the interdependent effects of solvent concentration and crosslinker
concentration have been conducted to maximize swelling of a gel material with a
constant ratio of H:B monomers.  Second,
the combinatorial methods were extended to create gels from acrylate-functional
derivatives of Castor oil, a biomass-derived macromonomer.  Based on previous studies of liquid-liquid
extraction, Castor oil has an unusually high distribution coefficient for
alcohol, motivating this portion of the work. 
The polyacrylate gels prepared from hydrophobic and hydrophilic ester modifications
of Castor oil exhibit superior performance, with D_a > 1.  The remainder of the talk will address the
performance of a model absorbent material in a batch extractive fermentation
process.