Appl Biochem Biotechnol (2012) 167:197–213

DOI 10.1007/s12010-012-9682-z

Reactive Extraction of Citric Acid Using Tri-n-octylamine

in Nontoxic Natural Diluents: Part 1—Equilibrium
Studies from Aqueous Solutions

Amit Keshav & Prakriti Norge & Kailas L. Wasewar

Received: 18 June 2011 / Accepted: 10 April 2012 /

Published online: 27 April 2012
# Springer Science+Business Media, LLC 2012

Abstract Use of cheap, nontoxic, and selective solvents could economically provide a
solution to the recovery of carboxylic acids produced by the bioroute. In this regard in the
present paper, reactive extraction of citric acid was studied. Problems encompassing the
recovery of the acid ([H3A]aqo 00.1–0.8) was solved by using tertiary amine (tri-n-octylamine,
TOA) in natural diluents (rice bran oil, sunflower oil, soybean oil, and sesame oil). TOA was
very effective in removal of acid providing distribution coefficient (D) as high as 18.51
(E%095 %), 12.82 (E%093 %), 15.09 (E%094 %), and 16.28 (E%094 %) when used with
rice bran oil, sunflower oil, soybean oil, and sesame oil, respectively. Overall extraction
constants and association numbers for TOA + rice bran oil, TOA + sunflower oil, TOA +
soybean oil, and TOA + sesame oil were evaluated to be 35.48 (mol/l)−1.46, 29.79 (mol/l)−1.30,
33.79 (mol/l)−1.51, and 37.64 (mol/l)−1.65 and 1.46, 1.30, 1.51, and 1.65, respectively. Specific
equilibrium complexation constants (KE(n/m)) were also predicted using mathematical
modeling.

Keywords Reactive extraction . Citric acid . Tri-n-octylamine . Natural diluents . Toxicity

Nomenclature
TOA Tri-n-octylamine
[B] Concentration of extractant (amine), moles per liter
[H3A] Concentration of undissociated citric acid, moles per liter
D Distribution coefficient
E% Degree of extraction

A. Keshav (*) : P. Norge

Department of Chemical Engineering, National Institute of Technology (NIT) Raipur, Raipur,
Chhattisgarh, India
e-mail: dr.amitkeshav@gmail.com

K. L. Wasewar
Department of Chemical Engineering, Visvesvaraya National Institute of Technology (VNIT) Nagpur,
Nagpur, Maharashtra, India
e-mail: k_wasewar@rediffmail.com
198 Appl Biochem Biotechnol (2012) 167:197–213

[A3−] Concentration of citric acid anion, moles per liter

H2A− Concentration of dissociated citric acid obtained after first dissociation,
moles per liter
HA−2 Concentration of dissociated citric acid obtained after second dissociation,
moles per liter
[H+] Concentration of hydroxyl ion, kilomoles per cubic meter
Kai Dissociation constant (i01 to 3 depending upon the stage of dissociation)
n Number of acid molecules involved in the association reaction
m Number of extractant molecules involved in the association reaction
α and β Limits of number of acid and extractant molecules considered for mass
action law modeling
z Overall loading ratio

Subscripts
aq Aqueous phase
org Organic phase

Superscript
o Initial

Introduction

Reactive extraction as a process could be an advantage for simultaneous removal of acidic

mass from the fermentation broth since it could increase the reactor productivity, as the
product is removed continuously, and also the process feasibility, as the product inhibition is
eliminated. Usually, it is desirable to have high distribution coefficients when any solvent
medium is employed for reactive extraction. Many a time, a compromise has to be made
amount the various desirable parameters of solvents required, listed in literature such as high
selectivity, high distribution, low toxicity, low viscosity, inflammability low surface tension,
etc. The search is though always to acquire as close to the desire. Toxicity of the employed
solvents is the major factor concerning liquid–liquid extraction process to be applied to a
bioreactor. There are two modes by which any reactive extraction process could be applied
to a fermentation broth: in situ and ex situ. In recovery from fermentation broths, it is
desirable the solvent possesses minimum toxicity, high selectivity, and high distribution
ratios. The problem of toxicity is more of a concern in in situ recovery of acid from
fermentation broth. Honda et al. [1] defined toxicity as a ratio of growth or production
velocity at a given concentration of organic solvent and in control cultivation. Toxicity is
caused on two different levels: molecular level, which is due to dissolved organic solvent
which can inhibit enzymes or modify cell membrane permeability, and phase level, which is
due to direct contact of solvent phase with the cells which may block nutrients diffusion
from medium to cells due to solvent coating and may disrupt the cell wall due to increased
surface tension.

Literature Data on Toxicity of Solvents

Laane et al. [2] express toxicity in terms of function log P which is logarithm of the KD of the
solvent in a standard octanol–water system. If a log P value is found less than 4 for any
solvent, the solvent is considered toxic whereas if it is greater than 6 it is nontoxic. The range
between 4 and 6 represents onset of toxicity. A more sophisticated (albeit tedious) method
Appl Biochem Biotechnol (2012) 167:197–213 199

based on principal component analysis using three descriptors (log P, ε, and melting point)
was developed by Andersson [3].
However, in a particular operating system of the fermentation broth, toxicity depends on the
microorganism used, and widely varying results could be found depending on whether yeast or
bacteria were used and on the particular strain of bacteria. Roffler et al. [4] studied the effect of
toxicity of organic solvents for in situ solvent extraction of carboxylic acids using various
species of bacteria (Bacillus subtilis and Acetobacter spp.), yeasts (Kluyveromyces marxianus
and Pichia fermentans), and fungi (Aspergillus terreus and Rhizopus arrhizus). They found that
trialkylamine and tributyl phosphate are much more toxic than the other carriers such as
trialkylphosphine oxide, triisobutylphosphine sulfide, and trihexyl phosphate. Pogorevc et al.
[5] studied the biocompatibility of organic solvents on four enzymes from three different
classes. They encompass completely water-miscible solvents, such as dimethyl sulfoxide, N,
N dimethyl formamide, methanol, acetone, and tetrahydrofuran, and little miscible (or virtually
immiscible) analogs, e.g., ethyl acetate, di-isopropyl ether, toluene, and hydrocarbons
(n-hexane, n-dodecane). The relative degree of deactivation was then plotted against the
respective log P value of the organic solvent. Remarkably little or no loss of activity was
observed for all enzymes for nonpolar solvents. A strong and irreversible enzyme deactivation
was predominant in more polar water-immiscible solvents. The reason of this may be the strong
hydrogen bonds, which contribute to enhanced deactivation.
Lewis and Yang [6] used Alamine 336/2–octanol mixture as the extractant in an extractive
fermentation process for propionate production from whey lactose. The extractant was found to
inhibit the cell growth, especially at concentrations higher than 60 % saturation. More substrate
was left and less products produced when more solvent was present, indicating that fermenta-
tion was inhibited by the solvent. However, even at 100 % saturation, the solvent toxicity on the
fermentation was not severe. The solvent toxicity was also tested with immobilized cells in the
continuous process.
There are different techniques by which the problem of toxicity could be resolved.
Cell immobilization [7], employing membrane that can prevent cells to come in direct
contact with the solvent, using microorganism strains which are unaffected by the
solvent used, and employing nontoxic solvents are some of the methods that could
reduce problem of toxicity. Removal of solvent toxicity by the above methods though
seems to be good, yet they suffer from few problems. Use of membrane leads to problems of
membrane fouling, high cost of membrane, membrane ruptures during operation, etc., and
practically insoluble solvents provide very low distribution coefficients. Thus, the best option is
to use nontoxic solvents or blend a toxic solvent with a nontoxic solvent to yield a biocompat-
ible mixture.
Solichien et al. [8] presented the solvent toxicity of the cultures of Propionibacterium
acidipropionici using pure diluents and extractant with diluents. Using pure diluents, diluent
was used directly as well as swollen in rubber. The immobilization was found to reduce the
toxicity. Corn oil and decanol were tolerable even with direct exposure. When extractant–
diluent combination system was employed, all solvent systems were found to be toxic except
for tri-n-octylphosphine/kerosene system. Secondary and tertiary amines were tested to be
toxic to propionibacteria. The diluents which were found to be nontoxic in the earlier studies
were found to be toxic when used with these extractants.
Prasad and Sirkar [9] discussed the various configurations of hollow fiber membrane for
extraction. Supported liquid membrane (SLM) and membrane extraction are also gaining
importance for recovery of solute using reactive extraction. Tri-n-octylamine (TOA) dissolved
in kerosene has been used for removal of acetic acid and butyric acid using SLM [10, 11].
Stripping was done by n-sodium hydroxide solution.
200 Appl Biochem Biotechnol (2012) 167:197–213

Martak et al. [12] studied solvent toxicity on different strains of bacteria (B. subtilis,
Acetobacter spp.), yeast (K. marxianus and P. fermentans), and fungi (A. terreus and R.
arrhizus) for the production of various carboxylic acids. Alkanes, alfols, Cyanex 471X (8,
12 %), and Cyanex 923 (100 and 50 %) do not affect the biomass growth of the yeasts
employed whereas Alamine in different percentages do. In the production of itaconic acid by
A. terreus, similar trend was observed and Alamine was found to be more toxic. For the
production of lactic acid by R. arrhizus, toxicity is more dominant for tri-n-butylphosphate
and Alamine, whereas THP and Cyanex 923 were found to be relatively less toxic. Playne
and Smith [13] found that TOA is nontoxic to anaerobic bacteria.

Citric Acid and Need for Its Recovery

Citric acid is a widely used carboxylic acid. As food additives, citric acid is used as a
flavoring and preservative in food and beverages, especially soft drinks. Citrate salts of
various metals are used to deliver minerals in a biologically available form in many dietary
supplements. The buffering properties of citrates are used to control pH in household
cleaners and pharmaceuticals. Citric acid’s ability to chelate metals makes it useful in soaps
and laundry detergents. By chelating the metals in hard water, it lets these cleaners produce
foam and work better without need for water softening. In a similar manner, citric acid is
used to regenerate the ion exchange materials used in water softeners by stripping off the
accumulated metal ions as citrate complexes. The saturation point for citric acid and water is
59 %. Citric acid is used in biotechnology and the pharmaceutical industry to passivate high-
purity process piping (in lieu of using nitric acid). Nitric acid is considered hazardous to
dispose once used for this purpose, while citric acid is not. Citric acid is the active ingredient
in some bathroom and kitchen cleaning solutions. A solution with a 6 % concentration of
citric acid will remove hard water stains from glass without scrubbing. In industry, it is used
to dissolve rust from steel. Citric acid is one of the chemicals required for the synthesis of
hexamethylene triperoxide diamine, a highly heat, friction, and shock-sensitive explosive
similar to acetone peroxide. For this reason, purchases of large quantities of citric acid rouse
suspicion of potential terrorist activity. Citric acid can be added to ice cream to keep fat
globules separate and can be added to recipes in place of fresh lemon juice as well. Citric
acid is used along with sodium bicarbonate in a wide range of effervescent formulate, both
for ingestion (e.g., powders and tablets) and for personal care.
The most economical method for producing citric acid since the 1930s has been fermen-
tation. The substrate used for the fermentation has been molasses, pumpkin, lower alcohols,
sucrose, glucose, fructose, galactose, etc. Recovery of acid is important in view of produc-
tion of acid. The conventional method of acid removal is calcium hydroxide precipitation
which is an unclean process since it employs chemical which causes disposal and handling
problems. Solvent extraction has created lot of scope for the removal of this acid and in the
last two decades, and a lot of work has been done in this direction. Wenneresten [14, 15], Yi
et al. [16], Sergieuskii [17], Sirman et al. [18], Bauer et al. [19, 20], Baniel et al. [21], Juang
and Huang [22], and Bizek et al. [23, 24] are some of the authors who had done some great
contribution toward the removal of citric acid from aqueous mediums using various extractant
and diluents.

Present Work

Very little work on the usage of nontoxic diluent for reactive extraction has been found in
literature [25–27]. None of any author before it employed a nontoxic/natural solvent like
Appl Biochem Biotechnol (2012) 167:197–213 201

vegetables oil, etc. for extraction of citric acid. Employing membrane for fighting
against the toxicity of solvent though had been worked out in some literatures, yet the
problem of increased price of unit along with problems associated with membrane like
stability and rupture, etc. has always surround these works. Toxicity can be combated
if no use of toxic solvent is made. Rice bran oil, the oil extracted from inner husk of
rice; sunflower oil, the non-volatile oil expressed from sunflower (Helianthus annuus)
seeds; soybean oil, extracted from the seeds of the soybean (Glycine max); and
sesame oil, the oil derived from sesame seeds, have been chosen in the present work
for reactive extraction of citric acid using TOA. The low cost, easy availability, high smoke
point, and low viscosity of these oils are the attractive features to be used in
extraction purposes. Distribution coefficient (D), degree of extraction (E%), and equilib-
rium complexation constant (KE) have been evaluated. Mathematical modeling was performed
to calculate the various types of complexation constant, and the KE values were assessed
through the R2 values of the fit.

Materials and Methods

Citric acid (C6H8O7; molecular wt 210.14; density 1.665 gm/cm3) was obtained from Merck
Co Ltd., Germany. TOA, a commercially available tertiary amine supplied by Aldrich Co.,
India with molecular weight of 353.66 and density of 0.809 g/cm3, was used as extractants.
Rice bran oil, sunflower oil, soybean oil, and sesame oil were used as diluent and were
obtained from local market. Sodium hydroxide was used for analysis and was purchase from
Fisher Scientific. Phenolphthalein was used as indicator (pH 8–10). Distilled water was used
in the experiment.
Equal amount of organic (extractant in diluent) and aqueous (citric acid) phase were taken
and mixed for 6 h in hot water bath shaker (Remi Equipment Pvt. Ltd., India). Different
concentrations of citric acid (0.1–0.8 mol/l) and different percent of TOA (10–30 %) in
diluents were taken for the equilibrium studies. Six hours was found to be sufficient time for
equilibrium. The samples after stirring were kept for phase separation for 4 h. Settling time
for complete phase separation was also standardized by keeping the same type of samples
for 2, 4, and 6 h. There has been negligible change in the values of acid concentration
obtained in all the cases, and hence, 4 h has been selected as the settling time. Citric acid
concentration in the aqueous phase was determined by titration using standard NaOH
solutions (0.05–0.4 N) using phenolphthalein as an indicator. To remove problems of alkali
reaction with environmental carbon dioxide, it was always prepared fresh and a soda lime
absorbent guard was used at the top of the burette. A few experiments were carried out in
duplicate, and consistency was within 5 %.

Theory

Physical Extraction

Carboxylic acids (HkA) mainly exist as dimmers in the organic phase owing to strong
intermolecular hydrogen bonding, especially in nonpolar or slightly polar diluents. On the
contrary, in the aqueous phase, they existed as monomers because the intermolecular
hydrogen bonding between the acids is destroyed owing to their preferential hydrogen
bonding with the water molecules. However, at pH > pKa, various forms of acid ionization
202 Appl Biochem Biotechnol (2012) 167:197–213

are possible. For citric acid (k03), dissociations reactions and dissociation constants can be
represented as
H2 A þ H þ
H3 A ¼ H 2 A  þ H þ Ka1 ¼ ð1Þ
H3 A

HA2 þ Hþ
H2 A ¼ HA2 þ Hþ Ka2 ¼ ð2Þ
H2 A

A3 þ Hþ
HA2 ¼ A3 þ Hþ Ka3 ¼ ð3Þ
H2 A
The overall dissociation constant (Kai) can be represented as:


½Hþ  H3i Ai
Kai ¼
4i i1  i¼13 ð4Þ
H A

In physical extraction, the acid is extracted into the organic phase by partition. In addition
to this, the acid partitioned into the organic phase may form dimmer or higher association.
Dimerization usually happens in solvents which possess poor solvation characteristic such as
inert diluents like hexane, etc. in which only dispersion forces are present. Extractability of
acid depends on extent of hydration of the acid and energy of the bond to water molecules.
Aliphatic hydrocarbons and inert diluents such as natural diluents have very low solubility in
water. To obtain complete miscibility in the phases, very high concentration of citric acid is
required. At high acid content, i.e., in water-deficient situation, the solvation sheath around
acid is composed of both water and solvent molecules, thus making the solute species more
like organic solvent. These diluents are unable to act as a hydrogen bond donor, so these
solvents interact only slightly with the acid since only the nonspecific directional, induction,
and dispersion forces are operating, giving very low distribution coefficients. Protic solvents
such as alcohols contain hydrogen atoms bound to electronegative element O (−O–H) and
are good anion solvators due to their hydrogen bonding ability. The tendency becomes more
pronounced as the charge density (ratio of charge to volume) of the anion to be solvated
increases. Further, alcohols are capable of being both acceptors and donors, as are the acids
and hence have the ability of high distribution, though most of these alcohols are toxic and
hence a compromise has to be made in their selection for recovery of acid from bioreactor.
Overall distribution coefficient is defined as the ratio of total (analytical) concentration of
acid in all its forms (by partition, dimmers and as complexes) in organic phase and total
(analytical) concentration of all its existing forms (dissociated and undissociated) in aqueous
raffinate. It includes the effects like ionic strength, nature of ion concentration of H+ etc. of
solution constituents. For physical extraction, distribution coefficient and degree of extrac-
tion can be defined as:
½H3 Aorg þ ½H3 A2org
D¼

 ð5Þ
½H2 A aq þ HA2 aq þ A3 aq þ ½H3 Aaq


D  100
E% ¼ ð6Þ
ð1 þ DÞ
Appl Biochem Biotechnol (2012) 167:197–213 203

Owing to the low dissociation constants of the acids in the study (Ka1 01.742×10−3, Ka2 0
8.892×10−5, Ka3 07.551×10−6) and pH (2–2.5) less than pKa of the acid (pKa1 02.759±
0.009; pKa2 04.051±0.005; pKa3 05.122±0.007), the solute is mainly in molecular form in
the aqueous medium. Thus, the ratios [H2A−]aq +[HA−2]aq +[A−3]aq have negligible value
and hence can be neglected. Further, the overall distribution coefficient is evaluated instead
of distribution ratio, and since titration is employed in the determination of concen-
trations in aqueous and organic phase, the overall distribution coefficient (Eq. 5) may be
reduced to
½H3 Aorg þ ½H3 A2org
D¼ ð7Þ
½H3 Aaq

Chemical Extraction

Chemical extraction systems make use of liquid ion exchangers, more commonly called as
extractants or reactants, to perform selective separations. There are many different extrac-
tants commercially available. The four main types of liquid ion exchangers may be classified
as anion exchange, cation exchange, solvatizing, and chelate forming. For ease of handling
and other practical purposes, the extractant is diluted preferably in a nonaromatic, high-
boiling, water-immiscible diluent. Since most extractants are highly viscous, the physical
properties of the organic phase can be tailored by diluting the extractant to give desired
properties like high interfacial tension, low viscosity, and low density. The reactive sub-
stance forms a reversible complex with the target substance and promotes its transfer from
the aqueous phase to the organic phase, remarkably enhancing the selectivity of the
separation compared to the physical liquid–liquid extraction. The distribution data can be
interpreted by a set of equilibria involving the formation of complexes with “n” citric acid
molecules and “m” amine molecules.

100

80

60
E%

40

20

0
0.23 0.46 0.69 0.91 1.14 1.37
o
[B] (mol/l)
Rice bran oil sunflower oil Soybean oil Sesame oil
Fig. 1 Effect of TOA concentration of extraction of citric acid (0.1 mol/l) in different natural diluents: finding
the optimum TOA concentration
204 Appl Biochem Biotechnol (2012) 167:197–213

Fig. 2 Chemical equilibria for 0.6

the extraction of citric acid
10% TOA
(0.1–0.8 mol/l) with TOA 0.5
20% TOA
(0.23–0.69 mol/l) in rice

[H3A]org (mol/l)
30% TOA
bran oil 0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8
[H3A]aq (mol/l)

Distribution of non-dissociated acid

 
nH3 A þ mðBÞorg ¼ ðBÞm H3m A org m ¼ 1; 2; 3 ð8Þ

Distribution of dissociated acid

 
nHþ þ H3n An þ mB ¼ ðBÞm H3m A org m ¼ 1; 2; 3 ð9Þ

Both sets of equations are characterized by the thermodynamic extraction constant


ðBÞm ðH3 AÞn org
KE ¼ ð10Þ
½H3 Anaq ½Bm
org

If assuming n01, the distribution coefficient (D) for the undissociated fraction extracted can
be defined in terms of equilibrium complexation constant as


ðBÞm H3 A org
D¼ ¼ KE ½Bm ð11Þ
½H3 Aaq org

Taking log
logD ¼ log KE þ m½Borg ð12Þ

Fig. 3 Chemical equilibria for 0.6

the extraction of citric acid (0.1–
0.8 mol/l) with TOA (0.23– 0.5
10% TOA
0.69 mol/l) in sunflower oil
[H3A]org (mol/l)

0.4 20% TOA

30% TOA
0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8
[H3A]aq (mol/l)
Appl Biochem Biotechnol (2012) 167:197–213 205

Fig. 4 Chemical equilibria for 0.6

the extraction of citric acid (0.1–
0.8 mol/l) with TOA (0.23– 0.5 10% TOA
0.69 mol/l) in soybean oil
20% TOA

[H3A]org (mol/l)
0.4
30% TOA

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8
[H3A]aq (mol/l)

If n is not 1, a general model could predict the various types of KE values as:

a X
X b
½H3 Aorg ¼ mKEðm:nÞ ½Bm n
org ½H3 Aaq 1 < m < a; 1 < n < b ð13Þ
m¼1 n¼1

The equilibrium concentration of “free” amine R3N can be calculated from the balance
equation of amine:

a X
X b
½Bo org ¼ ½Borg þ mKEðm:nÞ ½Bm
org ½H3 A
n
ð14Þ
m¼1 n¼1

The extent to which the organic phase can be loaded with citric acid is expressed as the
loading ratio. The loading of the extractant z is defined as the total concentration of acid in
the organic phase divided by the total concentration of amine in the organic phases.
½H3 Aorg
z¼ ð15Þ
½Borg

The value of z depends on the extractability of the acid, i.e., strength of the acid–base
interaction, and its aqueous concentration. The stoichiometry of the overall extraction
reaction depends on the loading ratio in the organic phases, z. If the organic phase is not
highly concentrated, i.e., at very low loading, the (1:1) complex is formed and this happens

Fig. 5 Chemical equilibria for 0.6

the extraction of citric acid (0.1–
0.8 mol/l) with TOA (0.23– 0.5
0.69 mol/l) in sesame oil 10% TOA
[H3A]org (mol/l)

0.4 20% TOA

0.3 30% TOA

0.2

0.1

0
0 0.2 0.4 0.6 0.8
[H3A]aq (mol/l)
206 Appl Biochem Biotechnol (2012) 167:197–213

20
18
16
0.23 mol/l TOA
14
0.46 mol/l TOA
12 0.69 mol/l TOA
KD

10
8
6
4
2
0
0.1 0.2 0.4 0.6 0.8
[H3A]oaq (mol/l)
Fig. 6 Effect of TOA concentration on distribution of citric acid using rice bran oil as diluent

when z<0.5. If z values are higher than 0.5, other types of complexations are considered
such as (1:2), (1:3), etc. Usually in citric acid extraction, other types of aggregates such as
2:3, 3:2, etc. also need to be considered because such types of complexations are also
contributing to the overall loading ratio.

Results and Discussion

Reactive liquid–liquid extraction has been energetically favorable for separating out acid in
less than 5 mass %. A lot of literature has cited that extraction using traditional carbon-
bonded oxygen donor solvents such as inert diluents, dipolar aprotic solvents, apolar aprotic
solvent, and polar diluents results in not very high acid recovery. In the present case, even

14

12
0.23 mol/l TOA
10
0.46 mol/l TOA
KD

8
0.69 mol/l TOA
6

0
0.1 0.2 0.4 0.6 0.8
[H3A]oaq (mol/l)
Fig. 7 Effect of TOA concentration on distribution of citric acid using sunflower oil as diluent
Appl Biochem Biotechnol (2012) 167:197–213 207

16

14
0.23 (mol/l) TOA
12
0.46 (mol/l) TOA
10
0.69 (mol/l) TOA
KD

0
0.1 0.2 0.4 0.8
o
[H3A] aq (mol/l)

Fig. 8 Effect of TOA concentration on distribution of citric acid using soybean oil as diluent

the physical extraction in the natural diluent—rice bran oil, sunflower oil, soybean oil, and
sesame oil—results in very poor distribution coefficient. Degree of extraction in no case, i.e.,
any concentration of acid (0.1–0.8 mol/l), could result in values greater than 2 %. Hence,
these natural diluents are poor extracting medium for citric acid. Thus, for complete recovery
of acids, high flow rates of these diluents are required that further increase the cost of back
recovery of acid from the loaded organic phase. In addition, the solubility and tendency to
form azeotropes can be the additional problems associated with these traditional solvents in
that case.
For the reactive extraction of acid, hydrophobic tertiary amines as extractants have been
successfully used in literature. These extractants have high selectivity with acid over water
and other non-acidic species. Since acid has strong attraction to aqueous phase, it is desirable
that the interaction of these extractants with acid should be in the range of hydrogen bonding

18

16
0.23 (mol/l) TOA
14
0.46 (mol/l) TOA
12
0.69 (mol/l) TOA
10
KD

0
0.1 0.2 0.4 0.6 0.8
[H3A]oaq (mol/l)

Fig. 9 Effect of TOA concentration on distribution of citric acid using sesame oil as diluent
208 Appl Biochem Biotechnol (2012) 167:197–213

Fig. 10 Plot of loading ratio (z) 0.9

vs log[H3A]aq for the different 0.8 0.23 mol/l TOA
concentrations of TOA in rice 0.46 mol/l TOA
0.7
bran oil 0.69 mol/l TOA
0.6
0.5

z
0.4
0.3
0.2
0.1
0
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
log (H3A)oaq

or ion–ion interactions and could result in high distribution coefficients. Amines have the
tendency to provide high distribution coefficient due to the tendency to form ion pair
association of the alkyl ammonium cation with the acid anion. The choice of diluent is,
however, important in this regard since it is desirable that diluent should form hydrogen
bond with the oxygen accessible on the complexed acid. The diluent also affects the physical
properties, such as viscosity, surface tension, and specific gravity, of the extractants. Many
of the diluents are toxic and for the extraction of acids which are specifically produced by
bio route, it become very essential to opt for a nontoxic diluent. The major problem
associated with using inert nontoxic diluents is the poor extraction obtained. In some studies,
modifiers along with these nontoxic diluents were employed for the recovery of acids, yet
the increase in complexity is another problem that needs to be dealt with in those situations.
In the present text, TOA has been used with rice bran oil, sunflower oil, soybean oil, and
sesame oil, respectively. Extractant was used in the volume percentage of 10 to 30 % in
respective diluents. Higher concentrations were also checked with these natural diluents
(Fig. 1), but employing TOA greater than 30 % could not result in significant increase in
distribution of acid. Further, this also would result in increase of the cost of the operating
unit. Higher concentration of extractant is also not advisable owing to two reasons. First, the
higher the concentration of these extractants, the higher is its effect in terms of toxicity to the
microorganisms available in the fermentation broth. Secondly, most of the extractants are
active only in acidic conditions, and with the raise of pH, the recovery falls very largely [28]
and the presence of large excess of extractant would not participate in extraction process.
As seen in the batch experiments, the presence of extractant results in significant
improvement of the extraction of acid as compared to physical extraction (Figs. 2, 3, 4,

Fig. 11 Plot of loading ratio (z) 1

vs log[H3A]aq for the different 0.9
concentrations of TOA in 0.8 0.23 mol/l TOA
sunflower oil
0.7 0.46 mol/l TOA
0.6
0.69 mol/l TOA
z

0.5
0.4
0.3
0.2
0.1
0
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
log [H3A]oaq
Appl Biochem Biotechnol (2012) 167:197–213 209

Fig. 12 Plot of loading ratio (z) 0.8

vs log[H3A]aq for the different
0.7
concentrations of TOA in soybean 0.23 mol/l TOA
oil 0.6
0.46 mol/l TOA
0.5 0.69 mol/l TOA
0.4

z
0.3
0.2
0.1
0
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
log [H3A]oaq

and 5). TOA results in the high distribution coefficient for the extraction of citric acid using
natural diluents. In all the cases, the higher is the percentage of TOA, the higher is the
distribution coefficient obtained (Figs. 6, 7, 8, and 9). E% increases with raising TOA
concentration for all acid concentration and in all the diluents. Average increases in E%
values as TOA varied from 10 to 30 % in rice bran oil, sunflower oil, soybean oil, and
sesame oil were found to be 51, 47, 51, and 53 %, respectively. With 30 % TOA in rice bran
oil, sunflower oil, soybean oil, and sesame oil, the highest D values were found to be 18.51,
12.82, 15.09, and 16.28, respectively. The D values were found to decrease with increase in
acid concentration and hence suggesting that reactive extractions using TOA are very
successful for removal of acid from dilute solutions. For the reactive extraction of citric
acid using TOA, D values decreases by 82, 86, 81; 82, 89, 78; 87, 88, 82; and 82, 87, 82 %,
as acid concentration was raised from 0.1 to 0.8 for 0.23, 0.465 and 0.69 mol/l TOA,
respectively, in rice bran oil sunflower oil, soybean oil, and sesame oil. All the diluents
chosen in the study were equally good in providing the solvation of acid–TOA complexes.
The best diluent among all was sunflower oil, which provides the highest E% values on most
of the studied data points.
Experiments involving 30 % TOA concentration and lower acid concentrations provided
the highest degree of extractions. This extraction could have raised the pH of the solution. So
even if excess amount of TOA is present, it would have gone without recovering anything,
and hence, additional cost would have to be faced for recycling this TOA back to the broth.
So it is optimum to use 30 % TOA. This is also justified in view of the problem of toxicity it
could raise. Further, it has been mentioned that distribution coefficient of acid using various

Fig. 13 Plot of loading ratio (z) 0.9

vs log[H3A]aq for the different 0.8
concentrations of TOA in sesame 0.7
0.23 mol/l TOA
oil 0.46 mol/l TOA
0.6
0.5 0.69 mol/l TOA
z

0.4
0.3
0.2
0.1
0
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
log [H3A]oaq)
210 Appl Biochem Biotechnol (2012) 167:197–213

Fig. 14 Fit of Eq. 12 for predict- 2

ing overall KE values and associ-
1.5 2
ation number for reactive y = 1.46x + 1.55; R = 0.92
extraction of citric acid using 1
TOA in rice bran oil

log D
0.5

-0.5

-1

-1.5
-1.5 -1 -0.5 0
log [B]org

concentrations of amine goes through a maximum [29, 30]. The reason for this being that
TOA alone is a relatively poor solvation medium for the complexes.
Loading ratios (Eq. 15) were also obtained for the reactive extraction of citric acid using
TOA in rice bran oil, sunflower oil, soybean oil, and sesame oil, respectively. In the
extraction using TOA in these natural diluents, loading values were found to lie between
0.27:0.81, 0.28:0.86, 0.27:0.74, and 0.26:0.77 for TOA/rice bran oil, TOA/sunflower oil,
TOA/soybean oil, and TOA/sesame oil systems, respectively, for acid concentrations of 0.1–
0.8 mol/l and extractant concentrations of 0.23–0.69 mol/l. Figures 10, 11, 12, and 13 show
the loading curves (plot of z vs log initial acid concentrations) for extraction involving TOA
in natural diluents (rice bran oil, sunflower oil, soybean oil, and sesame oil). In extraction
involving TOA, there is a marked effect of amine concentration on the loading of acid.
Kertes and King [31] stated that if z values are less than 0.5, (1:1) acid amine complexes are
formed. Thus, higher complexation is definitely possible in extractions involving TOA. This is
more pronounced at the higher concentration of acid as can be seen in Figs. 10, 11, 12, and 13.
For (1:1) complexations, Kertes and King [31] suggested the plot of z/1−z vs aqueous
phase acid concentration, for the computation of equilibrium complexation constant (KE).
Because of wide variation of loading ratios values for extraction using TOA, this fit is not
justified. Plot of Eq. 12 is shown in Figs. 14, 15, 16, and 17, and it suggests that 1.46, 1.30,
1.51, and 1.65 number of TOA molecules, respectively, react with one molecule of particular
acid when rice bran oil, sunflower oil, soybean oil, and sesame oil, respectively, were used as
diluents. The overall equilibrium complexation constant was computed to be 35.48 (mol/l)−1.46,
29.79 (mol/l)−1.30, 33.79 (mol/l)−1.51, and 37.64 (mol/l)−1.65 for reactive extraction of citric acid
in rice bran oil, sunflower oil, soybean oil, and sesame oil, respectively. Thus, in extraction
involving TOA, other types of associations are also involved. To predict the KE values for the

Fig. 15 Fit of Eq. 12 for predict- 1.4

ing overall KE values and associ- 1.2 2
ation number for reactive y = 1.3019x + 1.474; R = 0.95
1
extraction of citric acid using 0.8
TOA in sunflower oil 0.6
log D

0.4
0.2
0
-0.2
-0.4
-0.6
-2 -1.5 -1 -0.5 0
log [B]org
Appl Biochem Biotechnol (2012) 167:197–213 211

Fig. 16 Fit of Eq. 12 for predict- 1.4

ing overall KE values and associ- 1.2
ation number for reactive 1 y = 1.5073x + 1.5307; R2 = 0.87
extraction of citric acid using
0.8

log D
TOA in soybean oil
0.6
0.4
0.2
0
-0.2
-0.4
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
log B org

extraction of citric acid using TOA (0.23–0.69 mol/l), Eqs. 13 and 14 were applied. Various
different types of complexations combinations were considered. For reactive extraction of citric
acid using TOA in rice bran oil, when (1:1) and (1:2) acid/amine complexations were consid-
ered, KE(1:1) and KE(1:2) values of 4.68 l/mol and 32.04 (l/mol)2 were obtained with R2 value of
0.93. Assume that any higher complexations such as (1:3), (1:4), etc. could improve R2 value
but result in no positive values of respective complexation constants. This suggests that
aggregate formation could have occurred. Various aggregates were considered in the process,
and it was found that considering (1:1), (1:2), and (3:2) type of complexations, the data predict
R2 value of 0.98. KE(1:1), KE(1:2), and KE(3:2) values were found to be 4.55 l/mol, 30.44 (l/mol)2,
and 73.46 (l/mol)4, respectively.
For TOA in sunflower oil, considering only the (1:1) and (1:2) complexation had given
the good fit of the experimental data, and KE(1:1) and KE(1:2) were computed to be 8.85 l/mol
and 22.37 (l/mol)2 and R2 of the fit was 0.97. For reactive extraction of citric acid using TOA
in soybean oil, it was not possible to get a good fit of overall data points taking all points
together. However, when the data were broken and regression was done for the upper two
points where the loading have been low, two types of complexations (1:1) and (1:2) were
found to be prevalent as seen from the fit of data. KE(1:1) and KE(1:2) were computed to be
9.62 l/mol and 13.51 (l/mol)2, and R2 of the fit was 0.95. The points having loading higher
than 0.5, (1:1), (1:3), and (1:5) acid/amine complexations were found with the coefficient
values to be 0.35 l/mol, 83.90 (l/mol)3, and 251.27 (l/mol)5 and R2 of the fit was 0.96.
Similar was the results with the extraction involving TOA in sesame oil, where it was found
that at lower acid concentration in the entire studied TOA concentration, (1:1) and (1:2)
complexations, and at higher acid concentration in the entire studied TOA concentration,
(1:2) and (1:3) complexations were prevalent. The corresponding equilibrium complexation

Fig. 17 Fit of Eq. 12 for predict- 1.4

ing overall KE values and associ- 1.2
ation number for reactive 1 y = 1.6479x + 1.5756; R2 = 0.94
extraction of citric acid using 0.8
TOA in sesame oil 0.6
log D

0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1.5 -1 -0.5 0
log Borg
212 Appl Biochem Biotechnol (2012) 167:197–213

constants were found to be: for [H3A]o 00.1–0.3, TOA 10–30 % in sesame oil; KE(1:1) 0
6.01 l/mol and KE(1:2) 023.69 (l/mol)2 (R2 00.95) and for [H3A]o 00.3–0.8, TOA 10–30 % in
sesame oil; KE(1:2) 021.50 (l/mol)2 and KE(1:3) 045.06 (l/mol)3 (R2 00.96).
TOA has been found to be the good extractant for the reactive extraction of citric acid
from aqueous solutions when rice bran oil, sunflower oil, soybean oil, and sesame oil,
respectively, have been used as diluent. Further, TOA is not very toxic to bacteria respon-
sible for growth in microbial fermentation for citric acid production and hence could be
easily used for in situ or ex situ recovery units. Fermentation is usually carried at pH of 7.
Production of acid results in fall of pH. Even the slight presence of acid, i.e., the presence of
acid in very little concentration, decreases the pH significantly. TOA has been suggested to
be effective extractant only at low pH conditions. Thus, in order to employ TOA for
recovery of acid, a control mechanism is proposed which measures the pH of systems,
and as soon as pH falls below 4, the TOA is be inserted. The presence of TOA removes the
acid and hence again raises the pH to 7. Since the microorganisms responsible for acid
production are effective at pH of 6 to 7, productivity of acid by microorganisms will not fall
due to the presence of reactive extraction step. Effect of temperature is also needed to be
studied since extraction decreases with temperature. However, in the other work of the
author [32] and other researchers [33, 34], it has been found that in the operating range of
fermentor (30–40 °C), the decrease in extraction due to temperature increase has not been
excessive and hence reactive extraction could be successfully employed for recovery of acid
under these conditions. The study on use of natural diluents for acid recovery is a long way
to go.

Conclusion

Natural diluents such as rice bran oil, sunflower oil, soybean oil, and sesame oil could be a
good nontoxic system for removal of citric acid from fermentation broth particularly when
used with a tertiary amine (TOA). High degree of extractions (>95 %) could be possible
when 30 % TOA is employed with these diluents for removal of citric acid. Citric acid
concentrations chosen for the study were as obtained in fermentation broths (0.1–0.8 mol/l).
High removal of acid was obtained at lower acid concentration signifying the success of the
extraction system for removal from dilute solutions. Mass action modeling for predicting KE
values results in good fit of experimental data. The data provided could be efficiently used
for the removal of citric acid using nontoxic natural diluents.

References

1. Honda, H., Taya, M., & Kobayashi, T. (1986). Journal of Chemical Engineering Japan, 19, 268–273.
2. Laane, C., Boeren, S., & Vos, K. (1985). Trends in Biotechnology, 3, 251–252.
3. Andersson, M. (1998). Biocatalysis and Biotransfusion, 16, 259–273.
4. Roffler, S. R., Randolph, T. W., Miller, D. A., Blanch, H. W., & Prausnitz, J. M. (1991). Extractive
bioconversions with non-aqueous solvents. In B. Mattiasson & O. Holst (Eds.), Extractive bioconver-
sions. New York: Marcel Dekker.
5. Pogorevc, M., Stecher, H., & Faber, K. (2002). Biotechnology Letters, 24, 857–860.
6. Lewis, V. P., & Yang, S. T. (1992). Biotechnology, 8, 104–110.
7. Yabannavar, V. M., & Wang, D. I. C. (1991). Biotechnology and Bioengineering, 37, 716–722.
8. Solichien, M. S., OBrien, D., Hammond, E. G., & Glaz, C. E. (1995). Enzyme and Microbial Technology,
17, 23–31.
9. Prasad, R., & Sirkar, K. K. (1988). AICHE Journal, 34, 177–188.
Appl Biochem Biotechnol (2012) 167:197–213 213

10. Nuchnio, P., Izawa, I., Nishio, N., & Nagai, S. (1987). Journal of Fermentation Technology, 65, 2821–
2828.
11. Christen, P., Minier, M., & Renon, H. (1990). Biotechnology and Bioengineering, 36, 116–123.
12. Martak, J., Rosenbergl, M., Schlosser, S., & Kristofikovi, C. (1995). Biotechnology Techniques, 9, 247–
252.
13. Playne, M. J., & Smith, B. R. (1983). Biotechnology and Bioengineering, 25, 1251–1265.
14. Wennersten, R. (1980). Proceedings of the International Solar Extractive Conference, 2, 80–87.
15. Wennersten, R. (1983). Journal of Chemical Technology and Biotechnology, 33(B), 85–94.
16. Yi, M., Pen, Q., Chen, D., Pen, L., Zhang, M., Wen, R., et al. (1987). Beijing Daxue Xuebao Ziran
Kexuebn, 4, 30–37.
17. Sergieuskii, V. V. (1989). Izvestiia Vysshei Vchebn. Zaved. Khimii. Tekhnol., 32, 79–81.
18. Sirman, T., Pyle, D. L., & Grandison, A. S. (1990). In D. L. Pyle (Ed.), Separation for biotechnology 2
[Pap. Int. Symp.] (pp. 245–296). Amsterdam: Elsevier.
19. Bauer, U., Marr, R., Rueckl, W., & Siebenhofeer, M. (1988). Chemical and Biochemical Engineering, 2,
230–232.
20. Bauer, U., Marr, R., Rueckl, W., & Siebenhofeer, M. (1989). Berichte der Bunsengesellschaft für
Physikalische Chemie, 93, 980–984.
21. Baniel, A. M., Blumberg, R., Hajdu, K. (1982) U.S. patent 4,275,234.
22. Juang, R. S., & Huang, T. H. (1995). Journal of Chemical Engineering Japan, 28, 274–281.
23. Bizek, V., Horacek, J., Kousova, M., Heyberger, A., & Prochazka, J. (1992). Chemical Engineering
Science, 47, 1433–1440.
24. Bizek, V., Horacek, J., & Kousova, M. (1993). Chemical Engineering Science, 48, 1447–1457.
25. Harington, T., & Hossain, M. M. (2008). Desalination, 218, 287–296.
26. Keshav, A., Wasewar, K. L., & Chand, S. (2009). Journal of Chemical Technology and Biotechnology, 84
(4), 484–489.
27. Wasewar, K. L., Shende, D., & Keshav, A. (2011). Journal of Chemical Technology and Biotechnology,
86(2), 319–323.
28. Keshav, A., Wasewar, K. L., & Chand, S. (2009). AICHE Journal, 55(7), 1705–1711.
29. Keshav, A., Wasewar, K. L., & Chand, S. (2010). Chemical Engineering Communications, 197, 606–626.
30. Malmary, G., Albet, J., Putranto, A., Hanine, H., & Molinier, J. (1998). Journal of Chemical and
Engineering Data, 43(5), 849–851.
31. Kertes, A. S., & King, C. J. (1986). Biotechnology and Bioengineering, 28, 269–293.
32. Keshav, A., Wasewar, K. L., Shri Chand, S., Uslu, H., & Inci, I. (2009). i-manager’s Journal of Future
Engineering and Technology, 4(2), 41–49.
33. Tamada, J. A., & King, C. J. (1990). Industrial and Engineering Chemistry Research, 29, 1333–1338.
34. Canari, R., & Eyal, A. M. (2004). Industrial and Engineering Chemistry Research, 43, 7608–7617.