You are on page 1of 9

Bioprocess Biosyst Eng (2011) 34:779–787

DOI 10.1007/s00449-011-0526-y

ORIGINAL PAPER

New reactive extraction systems for separation of bio-succinic acid


Tanja Kurzrock • Dirk Weuster-Botz

Received: 17 December 2010 / Accepted: 10 February 2011 / Published online: 25 February 2011
Ó Springer-Verlag 2011

Abstract Biotechnologically produced succinic acid has Introduction


the potential to displace maleic acid and its uses and to
become an important feedstock for the chemical industry. In Succinic acid is applied in the food and pharmaceutical
addition to optimized production strains and fermentation industry and other chemical industry sectors and is used as
processes, an efficient separation of succinic acid from the feedstock for the syntheses of biodegradable polyester
aqueous fermentation broth is indispensable to compete resins. Nowadays, succinic acid is mainly produced pet-
with the current petrochemical production processes. In this rochemically by hydrogenation of maleic anhydride and
context, high molecular weight amines are known to be subsequent hydration or direct hydrogenation of maleic
effective extractants for organic acids. For this reason, as a acid [1]. The current annual market size is about
first step of isolation and purification, the reactive extraction 20,000–30,000 tonnes (*400 million US $) [2].
of succinic acid was studied by mixing aqueous succinic The petrochemical production uses heavy metal cata-
acid solutions with 448 different amine–solvent mixtures as lysts in organic solvents at high temperature and pressure.
extraction agents (mixer-settler studies). The extraction Alternatively succinic acid can be produced by microor-
agents consist either of one amine and one solvent (208 ganisms from renewable resources at ambient temperature
reactive extraction systems) or two amines and two solvents and pressure. If bio-succinic acid could be produced
(240 reactive extraction systems). Maximum extraction competitively, it will have the potential to displace maleic
yields of succinic acid from an aqueous solution with acid and its uses. The estimated market potential is up to
423 mM succinic acid at pH 2.0 were obtained with more 270,000 tonnes and 15 billion US $ per year [2–4].
than 95% yield with trihexylamine solved in 1-octanol or In the recent years, bio-succinic acid yields of up to
with dihexylamine and diisooctylamine solved in 1-octanol 1.78 mol per mol glucose and final succinic acid concen-
and 1-hexanol. Applying these optimized reactive extrac- trations of up to 146 g L-1 have been reported with opti-
tion systems with Escherichia coli fermentation broth mized fermentation processes [5, 6]. The cost-effective
resulted in extraction yields of 78–85% due to the increased separation and isolation of succinic acid from the aqueous
ionic strength of the fermentation supernatant and the co- fermentation broth remains a so far not finally solved prob-
extraction of other organic acids (e.g., lactic acid and acetic lem for industrial production of bio-succinic acid. A variety
acid), which represent typical fermentation byproducts. of different downstream processing methods for bio-succinic
acid, like precipitation with calcium hydroxide or ammo-
Keywords Succinic acid  Reactive extraction  nium, electrodialysis, ion exchange and adsorption, crys-
Fermentation broth  Co-extraction  Ionic strength tallisation and liquid–liquid extraction have been reviewed
recently [7]. Most of these separation technologies showed
low selectivities and/or low yields with the exception of
T. Kurzrock  D. Weuster-Botz (&) reactive extraction with high molecular weight amines
Institute of Biochemical Engineering,
solved in non-water-miscible organic solvents [8–14].
Technische Universität München,
Boltzmannstr. 15, 85748 Garching, Germany The objective of this study is to broaden the number of
e-mail: d.weuster-botz@lrz.tum.de reactive extraction systems which may be applicable for

123
780 Bioprocess Biosyst Eng (2011) 34:779–787

the extraction of succinic acid from aqueous solutions and 0.033 and 1.0 mol kg-1 based on the solvent mixture were
to identify new improved reactive extraction systems. The applied in the second screening.
application of primary, secondary and tertiary amines as Succinic acid and lactic acid were obtained from Roth,
well as alkylamines and arylamines will be studied with whereas acetic acid was purchased from Merck. NaCl
simple mixer-settler configurations. All of the amines will (Roth) was used for adjustment of ionic strength of the acid
be applied as solutions in organic solvents due to their solutions. The inorganic salts NH4Cl (Roth), NaCl (Roth),
physical properties such as high viscosity and corrosivity. MgCl2 (Roth), MgSO4 (Merck), CaCl2 (Merck) and
It is known that the solvents will have a high impact on the KH2PO4 (Roth) were dissolved in aqueous succinic acid
stoichiometry of the formed amine–acid complexes. In this solutions in different concentrations to achieve a constant
context, solvents can be classified into active solvents ionic strength of 1.0 mol L-1 in the particular solution.
(polar, proton-donating), which can stabilize the amine– Adjustment of pH was always carried out with NaOH
acid complexes, and inert solvents, which are poorly sol- (Roth) or HCl (Roth).
vating media [15–17]. To identify the reactive extraction
systems with the highest yields and selectivities for suc- Methods
cinic acid, simple reactive extraction systems (one amine
and one solvent) and complex reactive extraction systems Succinic acid solutions
(two amines and two solvents) will be designed and stud-
ied. Reactive extraction systems already described in the For the extraction experiments from fermentation broth,
literature will be used as reference. The most efficient succinic acid was dissolved in fermentation supernatant
reactive extraction systems will be applied for succinic acid (without cells and proteins) of a cultivation of E. coli K12
extraction from fermentation broth to identify the effects of DSM 428. The cultivation medium was a defined salt
ions, ionic strength and other organic acids formed in the medium [18] with NH4Cl 3.7 mM, (NH4)2SO4 15.1 mM,
fermentation as byproducts on the succinic acid extraction KH2PO4 23.9 mM, K2HPO4 8.6 mM, NaH2PO4 12.5 mM,
yield. MgSO4 4.2 mM, CaCl22H2O 0.07 mM, ZnSO47H2O
0.002 mM, CuCl22H2O 0.0015 mM, MnSO41H2O 0.015
mM, CoCl26H2O 0.007 mM, H3BO3 0.002 mM, AlCl3
Materials and methods 6H2O 0.01 mM, Na2MoO42H2O 0.002 mM and FeSO4
0.04 mM. Glucose was applied as carbon source in dif-
Materials ferent concentrations according to the particular process
phase. After batch cultivation for 4 h with an initial glu-
Octylamine was used as primary amine extractant, dibu- cose concentration of 2 g L-1, the substrate limited expo-
tylamine, dihexylamine, dioctylamine, diisooctylamine nential fed-batch cultivation phase was started [19]. This
and diphenylamine as secondary amine extractants and phase was divided into two feeding phases: Feeding phase
tripropylamin, tributylamin, tripentylamin, trihexylamin, 1 with a glucose concentration of 100 g L-1 to control the
trioctylamin, triisooctylamin, tridodecylamin, methyldioc- growth rate to 0.5 h-1 (4 h) and feeding phase 2 with
tylamine, methyldidecylamine and methyldodecylamine as 300 g L-1 glucose to assure a growth rate of 0.1 h-1
tertiary amine extractants. All amines were obtained from (16 h). Fermentation was performed at 37 °C and pH was
Merck (Darmstadt, Germany), apart from tripentylamine controlled to 7.0 with the addition of 25% NH4OH.
and trihexylamine which were purchased from Sigma The final byproduct concentrations after the fermenta-
(Hamburg, Germany). The amines were always dissolved tion were 0.3 g L-1 pyruvate, 0.114 g L-1 acetate and
in different active and inert solvents. As polar and proton- 0.5 g L-1 fumarate. No glucose remained in the fermen-
donating solvents 1-butanol, 1-hexanol, 1-octanol, 2-octa- tation broth, because of the substrate limited conditions.
nol, 1-nonanol, 1-decanol and 1-dodecanol were used. The cells were separated by a centrifugation step (30 min,
Applied polar aprotic solvents were methylisobtylketone 4,500 rpm, 3,270g). After addition of succinic acid, pre-
(MIBK) and diisobutylketone (DIBK). As non-polar sol- cipitated proteins were removed by a second centrifugation
vents n-pentane, n-hexane, n-heptane, n-octane, n-nonane, (30 min, 4,500 rpm, 3,270g).
n-decane, n-dodecane, kerosene and toluene were applied. For extraction of succinic acid from an aqueous solu-
Most of them were purchased from Merck. 1-decanol, tion, succinic acid was dissolved in distilled water.
1-dodecanol and kerosene were obtained from Sigma and The pH of the aqueous succinic acid solution and the
MIBK and n-heptane from Roth (Karlsruhe, Germany). fermentation broth was adjusted with NaOH or HCl. A low
The amines were used without further purification with an pH of about 2–3 was preferred for succinic acid extrac-
initial amine concentration of 1.0 or 1.5 mol kg-1 based on tions, because only the undissociated acid molecules are
the solvent (first screening). Amine concentrations between extractable. Anyhow, some experiments had to be carried

123
Bioprocess Biosyst Eng (2011) 34:779–787 781

out with higher pH up to 4.5 to assure the dissolution of all reduced probability. That way, a new generation of reactive
additives such as NaCl or other salts. extraction systems is generated, which has to be evaluated
with further parallel experiments. This procedure is repe-
Extraction ated as long as optima are identified or the experimental
effort becomes too high.
For extraction studies, 500 lL of the succinic acid solution To tune into different objectives, a strict Pareto principle
(aqueous succinic acid solution or fermentation broth) were is used. The reactive extraction systems with the highest
mixed with 500 lL of the respective simple or complex yields of one generation are clustered and copied in a so-
reactive extraction system. The mixing was carried out in called external population, if they show improved results
1.5 mL glass vials with a mixer mill MM200 from Retsch with respect to the objectives compared to the already
(Haan, Germany) at 30 Hz for 10 min to assure thermo- copied reactive extraction systems from the previous
dynamic equilibrium conditions. Afterwards, the phases generations.
were separated by centrifugation at 4,500 rpm (3,270g) for The parameter settings of the GAME.opt software
5 min. applied for the second screening are listed in Table 1.

Analytics Effects of ions

The acid concentrations in the aqueous phase were mea- To study the effects of salts and ions on reactive extraction
sured by a HPLC 1100 from Agilent (Waldbronn, Ger- from fermentation broth, succinic acid was dissolved in
many) with an Aminex HPX-87H column from Biorad distilled water (0.423 mol L-1, pH 4.5). The ionic strength
(München, Germany) using 0.5 mM H2SO4 (from Merck) of the solution was varied between 0.5 and 3.0 mol L-1 by
as mobile phase and a flow rate of 0.6 mL min-1. The addition of NaCl.
concentrations of succinic acid in the organic phase were Experiments to study the effect of different salts were
calculated by applying mass balances. carried out with 0.423 M succinic acid solutions (pH of
2.5) containing NH4Cl, NaCl, MgCl2, MgSO4, CaCl2 or
Screening KH2PO4. The salt concentrations varied whilst the ionic
strength of the solutions was constant 1.0 mol L-1.
The first screening of simple reactive extraction systems
was carried out with an aqueous 0.5 M succinic acid Co-extraction experiments
solution with pH 2.3. The amines were applied with a
concentration of 1.0 mol kg-1 based on the particular Aqueous mixtures of 0.423 mol L-1 succinic acid,
solvent. Altogether, 208 simple reactive extraction systems 0.083 mol L-1 acetic acid and 0.022 mol L-1 lactic acid
were compared. were used for the co-extraction experiments. The pH was
16 amines and 16 solvents were applied in varying varied between 2.0 and 8.0 by adding NaOH or HCl.
amounts to form the complex reactive extraction systems
under study (combination of two solvents and two amines).
This would have resulted in a high number of Results and discussion
3,317,760,000 different reactive extraction systems in total.
Therefore a stochastic search strategy was chosen to reduce Screening of simple reactive extraction systems
the experimental effort: Genetic Algorithm for Multiob-
jective Experimental Design (GAME.opt) [20]. GAME.opt The nature of the amine and of the solvent has a high
generates binary bitstrings of the scalar quantities. These impact on the extraction efficiency (yields), as well as the
bitstrings encode the composition of the particular reactive stability and the stoichiometry of the generated amine–acid
extraction system. The first set of parallel experiments complexes. In Fig. 1, the final yields of all the tested
(generation) is designed randomly. All reactive extraction simple reactive extraction systems are compared. The yield
systems of one generation are experimentally validated in was not determined if aggregate formation was observed
parallel and the results are valuated on the basis of the (marked with grey), because back-extraction of succinic
objectives (maximization of extraction yield, minimization acid from aggregates will not be possible. 94 from 208
of amine concentration). The best reactive extraction sys- simple reactive extraction systems resulted in aggregate
tems are selected with highest probability; two bitstrings formation.
are randomly coupled and varied by recombinant cross- Regarding the applied solvents, two different factors are
overs. Premature convergence to local optima is prevented influencing the extraction yield of succinic acid from
by applying randomly generated point mutations with aqueous solutions: the polarity of the solvent and the

123
782 Bioprocess Biosyst Eng (2011) 34:779–787

Table 1 Parameter settings of


Seven variables Amine 1 (16 possible amines, 4 bits)
the GAME.opt software (second
screening) Concentration of amine 1 (0–500 mM, 4 bits)
Amine 2 (16 possible amines, 4 bits)
Concentration of amine 2 (0–500 mM, 4 bits)
Solvent 1 (16 possible solvents, 4 bits)
Faction of solvent 1 (0–100%, 4 bits)
Solvent 2 (16 possible solvents, 4 bits)
Fraction of solvent 2 (was calculated)
Possible 3,317,760,000
combinations
Two objectives Maximization of the extraction yield
Minimization of the amine concentration
Number of 30 individuals per generation (three parallel experiments per individual)
experiments
Crossover points 2
Mutation rate 1%
N extern 15
N cluster 15
Applied amines Octylamine, dibutylamine, dihexylamine, dioctylamine, diisooctylamine,
tripropylamine, tributylamine, tripentylamine, trihexylamine, trioctylamine,
triisooctylamine, tridodecylamine, methyldioctylamine, dimethyldodecylamine,
methyldidecylamine and diphenylamine
Applied solvents 1-Octanol, 2-octanol, 1-hexanol, 1-nonanol, 1-decanol, 1-dodecanol, kerosin, toluol,
n-pentane, n-hexane, n-heptane, n-nonane, n-decane, n-dodecane, MIBK and DIBK
Extraction 0.847 mol L-1 succinic acid in H2O; pH 4.0
conditions

presence of additional functional groups. Polar solvents


like alcohols or ketones show higher extraction yields of
succinic acid from aqueous solutions than non-polar sol-
vents like alkanes. The higher the polarity, the higher is the
ability of the solvent to solvate the generated amine–acid
complexes and the higher are the achievable yields. This
effect is also observable with respect to the tested alcohols.
The polarity decreases with an increasing chain length of
the alcohols. Therefore, higher extraction yields are pos-
sible with short-chain alcohols. Furthermore, solvents with
additional functional groups (alcohols) show much higher
yields than expected from their polarity [15, 16, 21–23].
These solvents are able to release a proton. Thereby, the
amine–acid complexes are stabilized and high yields can be
achieved.
Comparing the different groups of applied amines, two
effects were observed. A combination of amines with
alcohols shows the following order of extraction yields,
which was already described for amines in polar solvents
[24, 25]:
Fig. 1 Extraction yields and aggregate formation (grey fields) of
simple reactive extraction systems. Extraction of succinic acid from a Primary\Secondary [ Tertiary:
0.5 M aqueous solution (pH 2.3) was carried out with an amine
concentration of 1.0 mol kg-1 solvent in a mixer mill. Bold Reactive On the one hand, increasing extraction yields of succinic
extraction systems described in literature. Framed Most promising acid from aqueous solutions are observed with increasing
simple reactive extraction system (T1) in this screening alkylation and therewith increasing basicity of the amines.

123
Bioprocess Biosyst Eng (2011) 34:779–787 783

On the other hand, lower extraction yields occur with long- systems which are the most applicable for the succinic acid
chain tertiary amines because of steric hindrance due to the extraction:
three long side chains. Therefore, the highest extraction
1. diisooctylamine ? 1-octanol (yield: 92.3%)
yields of 95–96% are measured with secondary long-chain
2. trihexylamine ? 1-octanol (yield: 91.5%)
amines, such as diisooctylamine (C8), or shorter tertiary
3. dioctylamine ? 1-octanol (yield: 90.3%)
amines, such as trihexylamine (C6).
4. trihexylamine ? 2-octanol (yield: 89.9%)
The order of extraction yields of succinic acid from
5. trioctylamine ? 1-octanol (yield: 86.5%).
aqueous solutions differs in non-polar solvents. Secondary
amines are also more efficient than primary ones, but ter- The reactive extraction system with trioctylamine and
tiary amines show very low extraction yields for succinic 1-octanol is the most efficient reactive extraction system
acid. According to this, the order of extraction yields of for succinic acid described in the literature so far [10, 12].
reactive extraction systems with non-polar solvents (like According to this, four reactive extraction systems with
alkanes) follows: higher yields were identified. Altogether, the reactive
Primary  Tertiary\Secondary: extraction system with trihexylamine and 1-octanol (in the
following called ‘T1’) seems to be more adequate than the
Primary amines are less alkaline than secondary and reactive extraction system with diisooctylamine, because
tertiary amines and achieve only low extraction yields. The no aggregate formation has occurred in all tested systems,
steric hindrance of tertiary amines and the low solvating independent on the polarity of the applied solvent.
ability of alkanes lead also to strongly reduced extraction
yields. Therefore, only dibutylamine or short-chain tertiary Screening of complex reactive extraction systems
amines (BC5) are able to extract succinic acid, even though
with low yields. To minimize the aggregate formation and the water mis-
As described generally in the literature [26], arylamines cibility of some reactive extraction systems, another
achieve lower extraction yields than alkylamines, due to screening was carried out with the software GAME.opt. At
their low basicity. Comparing the extraction yields of this, a combination of two amines in two organic solvents
diphenylamine (arylamine) with dihexylamine (alkyla- should minimize aggregation by an improved solvation of
mine), diphenylamine is less adequate for succinic acid the amine–acid complexes with the application of a polar
extraction, although both amines are secondary amines solvent. At the same time, a reduction of the polar water-
with two C6 side chains or two C6 rings. miscible solvent (short-chain alcohol) is achieved by a
About 45% of the 208 reviewed simple reactive combination with a less water-miscible solvent. To assure
extraction systems showed aggregate formation. First of enough open space for the optimization, the extraction
all, combinations of non-polar alkanes with long-chain conditions were modified. The initial succinic acid con-
secondary and tertiary amines tend to aggregation. This centration was increased to 0.847 mol L-1 and the pH to
correlates to the data by Tamada et al. [16]. Reason for this 4.0. As a consequence, the maximum batch extraction
is the low solvation ability of alkanes and the low solubility yields are lower, due to the strong influence of the pH on
of amine–acid complexes containing long-chain secondary the extraction.
or tertiary amines, due to steric hindrance. Aggregation has 180 batch extractions with varying combination of
also occurred in combinations with long-chain alcohols amines and solvents were performed within six sets of
(CC8). Independent of the applied solvent, secondary and parallel experiments (generations). In Fig. 2, the Pareto-
tertiary amines with more than eight carbon atoms tend to fronts of the first and the sixth generation of reactive
aggregation. extraction systems are compared. The highest batch
Moreover, the reactive extraction systems, which show extraction yields were achieved at the highest initial amine
the highest yields, contain partially water-miscible short- concentrations. The average yields of the sixth generation
chain alcohols (1-butanol or 1-hexanol). First of all, are about 5% higher than the yields of the first generation.
1-butanol is water miscible up to a concentration of In the following, the complex reactive extraction system
1.04 mol L-1. The water miscibility of 1-hexanol is about 29 of the sixth generation (in the following called ‘29/6’) is
0.06 mol L-1. However, they are much more miscible in used for further experiments in comparison to the simple
comparison to the other applied solvents with a maximum reactive extraction system T1 of the first screening, because
miscibility in water of 8.6 mmol L-1. it has shown an effective extraction of succinic acid under
Ranking of reactive extraction systems with respect to these (and many other conditions, data not shown) without
yields, no aggregate formation and low water solubility of aggregate formation. To assure comparable conditions to
maximal 2.7–8.6 mM results in the following list of reactive system T1, the overall amine concentration of

123
784 Bioprocess Biosyst Eng (2011) 34:779–787

Fig. 3 Batch extraction yields of succinic acid from an aqueous


solution (grey bar) or an E. coli fermentation supernatant (white bar)
Fig. 2 Best results of the first (grey triangles) and the sixth (black with the reactive extraction systems T1 and 29/6 with amine
rhombus) generation of the screening of complex reactive extraction concentrations of 1.0 or 1.5 mol kg-1 solvent. Succinic acid concen-
systems with a stochastic search strategy (GAME.opt). Extraction of a tration in the aqueous model solution and the fermentation superna-
0.847 M aqueous succinic acid solution (pH 4.0) was carried out with tant was set to 0.423 mol L-1 at pH 2.0
an amine concentration of 1.0 mol kg-1 solvent in a mixer mill.
White circle is the complex reactive extraction system 29/6 (system amine concentrations of the reactive extraction systems T1
29 of the sixth generation) which is used for all following experiments and 29/6.
in comparison to reactive extraction system T1
The succinic acid extraction yields with fermentation
Table 2 Composition of reactive extraction system T1 and 29/6
supernatant are reduced to 84.4–85.1% compared to a pure
(1.5 mol amine kg-1 solvent) aqueous solution (91.4–95.5%) with reactive system 29/6.
With the simple reactive extraction system T1, yields are
Reactive extraction Components Volume Concentration
system (mL L-1) (mol L-1) solvent still lower. Instead of 89.2–94.7% (aqueous solution) only
78.3–82.9% of the succinic acid present in the fermentation
29/6 Diisooctylamine 101.66 0.34 supernatant was extracted. Reasons for the decreased yields
Dihexylamine 79.04 0.34 may be other ions and salts as well as other organic acids
1-Octanol 162.25 1.03 present in the fermentation supernatant.
1-Hexanol 657.06 5.27
T1 Trihexylamine 217.19 0.64 Effects of ionic strength
1-Octanol 783.54 4.99
It was shown that the extraction yields with reactive
extraction system T1 as well as with reactive extraction
system 29/6 was also set to 1.0 or 1.5 mol kg-1 based on system 29/6 decrease with increasing ionic strength of the
the solvent mixture. succinic acid solution (Fig. 4).
The detailed composition of the best simple reactive An ionic strength of above 2.0 mol L-1 makes the
extraction system T1 and of the chosen complex reactive extraction of succinic acid completely impossible. An ionic
extraction system 29/6 is specified in Table 2 for an overall strength of 0.6 mol L-1 represents the ionic strength of the
amine concentration of 1.5 mol kg-1 based on the solvent applied E. coli fermentation broth. The corresponding
or solvent mixture. reduction in extraction yield of about 7% is the same as
compared to the data of the extraction of succinic acid from
Reactive extraction of succinic acid from fermentation real fermentation broth, shown in Fig. 3. The maximum
broth batch extraction yields are lower than those shown in
Fig. 3, because of the higher pH of 4.5, which was nec-
Batch extraction of succinic acid from E. coli fermentation essary to assure the dissolution of the added NaCl.
supernatant was studied with the reactive extraction sys-
tems T1 and 29/6. The succinic acid concentration of the Effects of ions
fermentation supernatant was adjusted to 0.423 mol L-1
and a pH of 2.0. In Fig. 3, the extraction yields from water Batch extraction yields of succinic acid were measured in
and fermentation supernatant are shown for two different the presence of different salts like NH4Cl, NaCl, MgCl2,

123
Bioprocess Biosyst Eng (2011) 34:779–787 785

minimized concentration of these components should be


one objective of fermentation process optimization if
reactive extraction is applied for downstream processing.

Reactive extraction of other organic acids

Main byproducts of succinic acid fermentations with pro-


duction strains of Corynebacterium glutamicum, E. coli,
Anaerobiospirillum succiniciproducens, Actinobacillus
succinogenes, Mannheimia succiniciproducens or Basfia
succiniciproducens are ethanol, pyruvate, acetate, formate,
propionate, malate and lactate [5, 6, 28–31]. Alcohols, such
Fig. 4 Extraction yield as function of ionic strength (varied between as ethanol, show no influence on the reactive extraction of
0.5 and 3.0 mol L-1) of an aqueous succinic acid solution succinic acid, whilst other organic acids can be co-
(0.423 mol L-1, pH 4.5). The extractions were carried out with the
reactive extraction systems T1 (white circles) and 29/6 (black
extracted according to their pKa and the pH in the aqueous
rhombus) with amine concentrations of 1.0 mol kg-1 solvent. The phase. The dissociation constants pKa of these organic
extraction yield marked with (star) corresponds to the ionic strength acids are listed in Table 3.
of the applied E. coli medium (*0.6 mol L-1) All organic acids, with the exception of pyruvic acid,
have a pKa in the same range as succinic acid. Pyruvic acid
MgSO4, CaCl2 or KH2PO4. The ionic strength of each has a lower pKa. Therefore, pyruvic acid formed as a
solution was kept constant at 1.0 mol L-1 with exception byproduct will not be extracted in significant amounts at
of the reference conditions with \0.5 mol L-1. The pH pH [ 3, whereas the byproduct formation of all other
was set to pH 2.5 to assure the dissolution of succinic acid organic acids discussed above has to be avoided. Due to the
and the particular salt. The type of added salt changes the fact that acetic and lactic acid are the predominantly
maximum extraction yields dramatically (Fig. 5). formed byproducts, mixtures of them with succinic acid
NH4Cl, NaCl and MgCl2 show a high impact on suc- were studied in more detail making use of reactive
cinic acid extraction, whereas the other salts are less extraction systems T1 and 29/6 (Fig. 6).
inhibiting. A clear correlation between the ion concentra- As expected, the extraction yield of all organic acids
tion and the observed yields was measured. Increasing ion increases with decreasing pH because solely undissociated
concentrations result in the decrease of the extraction acids can be extracted by secondary and tertiary amines.
yields of succinic acid with both reactive extraction sys- All organic acids show dissociation constants around
tems. This can be explained by the salt anions, which pKa = 4–5. This explains why high extraction yields are
compete with the carboxylic groups of succinic acid for the observed at pH below pKa.
protons [27]. The corresponding acids (HCl, H2SO3 or At pH 5 or 6 acetic acid can be extracted with 10–15%
H3PO4) were co-extracted with succinic acid, which leads higher yields compared to the other two organic acids
to a decrease in the extraction yields. Therefore, a under study, because acetic acid has a higher pKa compared
to the other organic acids. At pH 4 the extraction yields
correlate with the pKa of the acids. Lactic acid has a pKa of
3.86 and shows the lowest yield, acetic acid has a pKa of
4.76 and shows the highest yield and the yield of succinic
acid is in between the two other organic acids (pKa1 of

Table 3 Dissociation constants pKa of some organic acid


Acid pKa1 pKa2

Succinic acid 4.17 5.64


Acetic acid 4.76 –
Lactic acid 3.86 –
Fig. 5 Effects of NH4Cl, NaCl, MgCl2, MgSO4, CaCl2 and KH2PO4
Formic acid 3.77 –
on the extraction yields of succinic acid (0.423 mol L-1, pH 2.5) with
the reactive extraction systems T1 (white bar) and 29/6 (grey bar) Malic acid 3.46 5.10
with amine concentrations of 1.0 mol kg-1 solvent. In addition, the Propionic acid 4.87 –
extraction yields under reference conditions (-) and the ion concen- Pyruvic acid 2.49 –
tration of the particular solution are indicted (black square)

123
786 Bioprocess Biosyst Eng (2011) 34:779–787

Consequently, one focus in strain and fermentation process


optimization for succinic acid production should be the
development of low pH fermentation processes if fermen-
tation process integrated reactive extraction of succinic
acid is envisioned.
Another important issue is the high impact of ions and
other organic acids on succinic acid extraction yields,
which was shown in this study. Fermentation process
development for succinic acid production should thus focus
on minimization of ionic strength (salt additions) as well as
strain optimization to minimize byproduct formation,
which is a general strategy to achieve higher fermentation
yields.
Reactive extraction processes will need efficient back-
extraction to isolate the extracted succinic acid from the
organic phase. The development of an efficient back-
extraction process in combination with crystallization of
the isolated succinic acid is presently under study in our
Fig. 6 Reactive extraction yields of succinic acid (black bar), acetic labs. One important issue of these studies will be the
acid (striped bar) and lactic acid (grey bar) from an aqueous solution achievement of total recyclability of the organic phase with
as function of pH. The applied acid concentrations were 0.423 the solved amines as often as possible to minimize the
mol L-1 succinic acid, 0.083 mol L-1 acetic acid and 0.022 mol L-1 downstream processing costs.
lactic acid. Reactive extraction systems T1 and 29/6 were applied
with an amine concentration of 1.0 mol kg-1 solvent
Acknowledgments This study was generously supported by the
Deutsche Bundesstiftung Umwelt (DBU), particularly by a DBU
succinic acid = 4.17). Low pH of 2–3 allows for a selec- Ph.D. grant for Tanja Kurzrock. The authors gratefully acknowledge
tive extraction of succinic acid due to its two carboxylic for the support of the TUM Graduate School.
groups. Nevertheless, these results show that organic acids
produced as byproducts in the fermentation will be co-
References
extracted in significant amounts. The suppression of
organic acids formation as byproducts during bio-succinic 1. Cornils B, Lappe P (1989) Aliphatic dicarboxylic acids. Ull-
acid fermentation should be another objective of strain and mann’s Encyclopaedia of Industrial Chemistry, vol A8. VCH,
fermentation process optimization if reactive extraction is Weinheim
applied for downstream processing. 2. Zeikus JG, Jain MK, Elankovan P (1999) Biotechnology of
succinic acid production and markets for derived industrial
products. Appl Environ Microbiol 51:545–552
3. Delhomme C, Weuster-Botz D, Kühn FE (2009) Succinic acid
Conclusions from renewable resources as a C4 building-block chemical—a
review of the catalytic possibilities in aqueous media. Green
The new identified reactive extraction systems with trihe- Chem 11:13–26
4. Willke T, Vorlop K-D (2004) Industrial bioconversion of
xylamine solved in 1-octanol or with dihexylamine and
renewable resources as an alternative to conventional chemistry.
diisooctylamine solved in 1-octanol and 1-hexanol result in Appl Microbiol Biotechnol 66:131–142
the best yields of succinic acid extraction from aqueous 5. Okino S, Noburyu R, Suda M, Jojima T, Inui M, Yukawa H
solutions so far reported (up to 95.5% with a simple mixer- (2008) An efficient succinic acid production process in a meta-
bolically engineered Corynebacterium glutamicum strain. Appl
settler). These high batch extraction yields can only be
Microbiol Biotechnol 81:459–464
achieved at a pH of the aqueous fermentation broth far 6. Vemuri GN, Eiteman MA, Altman E (2002) Succinate production
below the pKa1 of succinic acid (pH  4.17). The appli- in dual-phase Escherichia coli fermentations depends on the time
cation of reactive extraction in the bypass to a fermentation of transition from aerobic to anaerobic conditions. J Ind Micro-
biol Biotechnol 28:325–332
process (process integrated product separation) will thus
7. Kurzrock T, Weuster-Botz D (2010) Recovery of succinic acid
demand for a microbial succinic acid production process at from fermentation broth. Biotechnol Lett 32:332–339
very low pH or titrations of the fermentation broth. The 8. Tung LA, King CJ (1994) Sorption and extraction of lactic and
succinic acid producers presently under study (C. glutam- succinic acids at pH [ pKa1. 1. Factors governing equilibria. Ind
Eng Chem Res 33:3217–3223
icum, E. coli, Saccharomyces cerevisiae, A. succinogenes,
9. Hong YK, Hong WH (2000) Reactive extraction of succinic acid
A. succiniciproducens, M. succiniciproducens, B. succi- with tripropylamine (TPA) in various solvents. Bioproc Eng
niciproducens) operate best at 6.0 \ pH \ 7.8 [5, 6, 28–34]. 22:281–284

123
Bioprocess Biosyst Eng (2011) 34:779–787 787

10. Hong YK, Hong WH (2000) Equilibrium studies on reactive 22. Yerger EA, Barrow GM (1955) Acid-base reactions in non-dis-
extraction of succinic acid from aqueous solutions with tertiary sociating solvents. Acetic acids and diethylamine in carbon tet-
amines. Bioproc Eng 22:477–481 rachloride and chloroform. J Am Chem Soc 77:4474–4481
11. Hong YK, Hong WH (2000) Extraction of succinic acid with 23. Yerger EA, Barrow GM (1955) Acid-base reactions in non-dis-
1-octanol/n-heptane solutions of mixed tertiary amine. Bioproc sociating solvents n-butylamine and acetic in carbon tetrachloride
Eng 23:535–538 and chloroform. J Am Chem Soc 77:6206–6207
12. Hong YK, Han DH, Hong WH (2002) Water enhanced solubil- 24. Sykes P (1988) Reaktionsmechanismen der Organischen Chemie,
ities of succinic acid in reactive extraction using tertiary amines/ 9. Aufl. VCH Verlagsgesellschaft, Weinheim S.76
alcohol systems. Korean J Chem Eng 19:83–86 25. Reichardt C (1988) Solvents and solvents effects in organic
13. Hong YK, Hong WH (2004) Influence of chain length of tertiary chemistry, 2. Aufl. VCH Verlagsgesellschaft, Weinheim
amines on extractability and chemical interactions in reactive 26. Hart H (Hrsg.) (1989) Organische Chemie. VCH Verlagsge-
extraction of succinic acid. Korean J Chem Eng 21:488–493 sellschaft, Weinheim
14. Song H, Huh YS, Lee SY, Hong WH, Hong YK (2007) Recovery 27. Jun YS, Lee EZ, Huh YS, Hong YK, Hong WH, Lee SY (2007)
of succinic acid produced by fermentation of a metabolomically Kinetic study for the extraction of succinic acid with TOA in
engineered Mannheimia succiniciproducens strain. J Biotechnnol fermentation broth; effects of pH, salt and contaminated acid.
132:445–452 Biochem Eng J 36:8–13
15. Tamada JA, Kertes As, King CJ (1990) Extraction of carboxylic 28. Scholten E, Dägele D (2008) Succinic acid production by a newly
acids with amine extractants: 1. Equilibria and law of mass action isolated bacterium. Biotechnol Lett 30:2143–2146
modeling. Ind Eng Chem Res 29:1319–1326 29. Guettler MV, Jain MK, Rumler D (1996) Method for making
16. Tamada JA, King CJ (1990) Extraction of carboxylic acids with succinic acid, bacterial variants for use in the process, and
amine extractants: 2. Chemical interactions and interpretation of methods for obtaining variants. US Patent 5,573,931
data. Ind Eng Chem Res 29:1327–1333 30. Lee SJ, Song H, Lee SY (2006) Genome-based metabolic engi-
17. Tamada JA, King CJ (1990) Extraction of carboxylic acids with neering of Mannheimia succiniciproducens for succinic acid
amine extractants: 3. Effects of temperature, water coextraction production. Appl Environ Microbiol 72:1939–1948
and process considerations. Ind Eng Chem Res 29:1333–1338 31. Glassner DA, Datta R (1992) Process for the production and
18. Schaefer U, Boos W, Takors R, Weuster-Botz D (1999) Auto- purifcation of succinic acid. US Patent 5,143,834
mated sampling device for monitoring intracellular metabolite 32. McKinlay JB, Vieille C, Zeikus JG (2007) Prospects for a bio-
dynamics. Anal Biochem 270:88–96 based succinate industry. Appl Microbiol Biotechnol 76:727–740
19. Jenzsch M, Gnoth S, Beck M, Kleinschmidt M, Simutis R, 33. Raab A, Gebhardt G, Bolotina N, Weuster-Botz D, Lang C
Lübbert A (2006) Open-loop control of the biomass concentration (2010) Metabolic engineering of Saccharomyces cerevisiae for
within the growth phase of recombinant protein production pro- the biotechnological production of succinic acid. Metabol Eng
cesses. J Biotechnol 127:84–94 12:518–525
20. Link H, Weuster-Botz D (2006) Genetic algorithm for multi- 34. Arikawa Y, Kuroyanagi T, Shimosaka M, Muratsubaki H, E-
objective experimental optimization. Bioprocess Biosyst Eng nomoto K, Kodaira R et al (1999) Effect of gene disruptions of
29:385–390 the TCA cycle on production of succinic acid in Saccharomyces
21. Barrow GM, Yerger EA (1954) Acid-base reactions in non-dis- cerevisiae. J Biosci Bioeng 87:28–36
sociating solvents. Acetic acid and triethylamine in carbon tet-
rachloride and chloroform. J Am Chem Soc 76:5211–5216

123

You might also like