Desalination 244 (2009) 1223

Recovery of propionic acid from an aqueous stream by reactive

extraction: effect of diluents
Amit Keshav, Kailas L. Wasewar*, Shri Chand
Department of Chemical Engineering, Indian Institute of Technology (IIT) Roorkee, Uttarakhand 247667, India
Tel. +91 (1332) 285347; Fax: +91 (1332) 276535; email: k_wasewar@rediffmail.com, klw73fch@iitr.ernet.in
Received 27 September 2007; Accepted 18 April 2008

Abstract
Reactive extraction is a promising technique for the recovery of carboxylic acid from dilute aqueous solutions.
Reactive extraction requires a proper combination of extractant and diluent which provides higher capacity and
selectivity. The recovery of propionic acid by reactive extraction (present in concentration range of 0.050.4 kmol/m3
in the aqueous phase) was studied using tri-n- butyl phosphate (TBP) in different diluents (n-heptane, petroleum ether
and toluene). Extraction using pure diluents was found to be poor. Using the mixtures of extractantdiluents the
extraction was significantly improved. Results are presented in terms of distribution coefficient, equilibrium
complexation constant, loading ratio and extraction efficiency. Propionic acidTBP was found to form a 1:1 complex
with no overloading in any case. Though the TBPtoluene system was found to provide the highest distribution
coefficient of the three diluents tested, when extraction is to be done from a fermentation broth where a toxicity
problem is the main consideration factor for selection of solvent, TBP + n-heptane is preferred.
Keywords: Propionic acid; Reactive extraction; TBP; Diluents; Equilibrium

1. Introduction
Propionic acid is an important carboxylic acid
used primarily for animal feed preservation,
including hay, silage and grains, and in human
foods, mainly in baked goods and cheese. In
addition to these major uses, propionic acid and
derivatives have been used for manufacture of
*Corresponding author.

antiarthritic drugs, perfumes and flavors, plasticizers, mould preventives in silage and hay and as
a solvents[1].
Propionic acid is industrially produced by the
petrochemical route [2]. However with the
increase of the prices of petrochemical products,
alternative means of propionic acid production
was looked for. Fermentation technology emerges
as the best alternative providing the advantage of

0011-9164/09/$ See front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.desal.2008.04.032

A. Keshav et al. / Desalination 244 (2009) 1223

being energy efficient. Extractive fermentation

technology provides the advantages of increased
reactor productivity, produce, recovery of the
fermentation product in one continuous step and
hence reduction in downstream processing load
and recovery cost [3].
Conventionally, propionic acid is extracted
from fermentation broth by calcium salt precipitation, which was found to be an expensive and
an environmentally unfriendly technique. For the
recovery of propionic acid by reactive extraction,
common organic solvents such as ketones and
alcohols show a low distribution ratio because of
the high affinity of acid for water. Thus physical
extraction with conventional solvents is not an
efficient method for recovery of propionic acid.
Reactive extraction employs a combination of
extractant and diluent to enhance the recovery by
intensifying the separation through simultaneous
reaction and extraction and provides advantages
of high selectivity and fractional recovery.
Strongly solvating extractants like organophosphorous compounds have primarily been
used in inorganic analysis for the extraction and
separation of inorganic acids or metal species [4].
Since phosphoryl can form hydrogen bonds to
proton donors, strongly solvating extractants can
also be used for the extraction of acidic organic
compounds. Weak organic acids are extracted by
organophosphorous compounds with a significantly higher distribution ratio. The high polarity
of the phosphoryl group in tributyl phosphate
(TBP) enables it to act as a strong Lewis base and
as a result it can form acid base complex when
contacted with propionic acid [5]. The extraction
reactions are ionic in nature.
Diluent is usually added along with the
extractant to enhance its physical properties by
providing general solvation and affect the extraction power of the extractant by providing specific
interaction. The diluent may consist of one or
more components, inert or active. Various active
polar and proton or electron donating diluents
(halogenated aliphatic/aromatic hydrocarbons,

13

ketones, nitrobenzenes, higher alcohols), enhance

the extraction. On the other hand, inert diluents
(long chain paraffins, benzene etc.), limit the
solvent capacity.
Zhong et al. [6] studied the reactive extraction
of propionic acid using Alamine 304-1 in 2octanol, 1-dodecanol and Withohol 85 NF as
diluents at various amine concentrations (0
100%) and found extraction to be maximum at
the amine concentration between 2040%. Diluent significantly affects the extraction power of
the extractant, with 2-octanol providing the
highest distribution coefficient of the three. The
effect of toxicity and the methods of back extraction were also presented. The 2-octanol, though
providing the highest KD, was found to fail when
the selection based on toxicity was considered
(2-octanol being toxic to the bacteria).
Extraction of different carboxylic acids using
tri-n-butyl phosophate (TBP), tri-n-octyl amine
(TOA) and their mixtures was investigated by
Matsumato et al. [7] and the synergism effect was
highlighted in improving the extraction. Studies
on the propionic acid reactive extraction using trin-octyl amine in various diluents (hexanol, butyl
acetate and petroleum ether) show the positive
effect of the different diluents on the extraction
[8]. The acid was found to form 1:1 and 2:1 acid
amine complex during the extraction.
Malmary et al. [9] performed lactic acid
extraction with TBP dissolved in dodecane in
various ratios. They concluded that TBP appears
to be an effective extractant for lactic acid
recovery and that the viscosity of the pure TBP
has to be reduced by diluting with dodecane for
easier phase settling and mass transfer.
Ingal et al. [10] studied extraction of carboxylic acids (C2C6) using TBP at pH 22.5. The pH
was taken from environment process engineering
point of view. Effect of salts was also studied.
Extraction of five carboxylic acids (acetic,
propionic, butyric, valeric, and caproic acid) was
investigated using TBP and TOA in decanol +
dodecane and extraction isotherms was obtained

14

A. Keshav et al. / Desalination 244 (2009) 1223

by Morales et al. [11]. Experimental data show

that the estimated value of equilibrium extraction
is constant and an apparent number of extracting
molecules reacting with one molecule of monoacid depend not only on the organic phase
composition but also on how they are determined.
Much work on the reactive extraction of
different carboxylic acids using tertnary and
quarternary amine has been done but very little
work on organophosphorous extractants for
propionic acid extraction can be found (as given
in the previous paragraph). In view of this,
attempts were made to study the extraction of
propionic acid using TBP in n-heptane, petroleum
ether and toluene. The data on the distribution
coefficient, equilibrium complexation constant,
loading ratio and extraction efficiency are presented. The distribution of propionic acid
between water and TBP dissolved in various
diluents and a comparison with the extraction
capacity of pure diluent alone were studied at
isothermal conditions.

2. Theory
The recovery of propionic acid using TBP is
by the formation of a complex via the interfacial
reaction [10] and can be represented as:
HAaq + pSorg : (HA.Sp)org

(1)

As the propionic acidTBP complex is formed, it

is rapidly extracted into the organic phase. The
extraction mechanism is presented in Fig. 1.
The extraction constant and the number of
reacting molecules of extractant are computed by
applying the law of mass action that is the ratio
between concentrations of product molecules and
concentration of the reactant species, according to
the general equation of interaction between the
extractant and the extracted species as:
Ks =[(HA).(S)p]org /[HA]aq [S]p org

(2)

Fig. 1. Mechanism for the reactive extraction of propionic

acid (HA) using TBP(S) (*signifies interfacial value).

Ks is expected to depend on the properties of

the acid and the solvation efficiency of the diluent
used. The dissociation of the acid in the aqueous
phase is given as:
!

HA : H + A
+

K HA

H A

HA

(3)

KD 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. KD includes the effects like ionic
strength, nature of ion concentration of H+ etc. of
solution constituents. For physical extraction, KD
can be defined as:

K Ddiluent

HAorg
HAaq

Distribution coefficient of the acid by chemical

extraction can be defined as:

chem
D

HAorg HAdiluent
org

HAaq

where is the volume fraction of diluent and

is the acid extracted into the organic
HAdiluent
org
phase by diluent alone. The overall distribution

A. Keshav et al. / Desalination 244 (2009) 1223

coefficient is evaluated as the function of extraction constant and the number of reacting species
as:

overall
D

HA.S HAdiluent
org

HA
aq

K S HAaq S org HAorg

diluent

(4)

HAaq

HAdiluent
org
K S S org
HAaq

K s S org K

diluent
D

It is well known that carboxylic acid dissociates

in aqueous solution. Under the experimental
condition that pH of the aqueous solution was
smaller than pKa of the acid (4.67), the effect of
the acid dissociation was negligibly small. Thus
Eq. (4) can be written as

log K DEoverall K Ddiluent log K S

(5)

p log S org
where [S]org can be expressed as
0
p HAorg
S org Sorg

(6)

Thus the plot of log K Doverall K Ddiluent versus

log[S]org would yield a straight line with the slope
of p and the intercept of log(KS) from where KS
can be obtained.
The equation sets (1) and (3) are a simplified,
though realistic, version of a more general and
thermodynamically strict one that includes activity coefficient and/or allows for the simultaneous formation of two, or perhaps three,
different solvates, depending on the organic
phase loading [4]. King et al. [12] investigated
extensively the evaluation of Eq. (5) for systems
with negligible K Ddiluent values and found the

15

constants of the equation. The solvation number

of the acid was found to be the same as the
number of -COOH group on the acid molecule.
Propionic acid contains one -COOH group and
thus the solvation number (p) can be assumed to
be unity.
The extent to which the extractant is loaded
with propionic acid is expressed as the overall
loading factor (Zt) as

Zt

[ HA]org [ HA]diluent
org
[ S ]0org

(7)

3. Experimental
3.1. Materials
TBP (Himedia, India), a phosphorous bonded
oxygen donor, is a light colorless liquid with the
molecular weight of 266.32 g/mol and density of
0.92 g/cm3. Propionic acid (99%) (Himedia,
India) and the diluents petroleum ether, nheptane, toluene (S.D. Fine Chemicals, Mumbai,
India) are of technical grade and were used without pretreatment. These diluents were used to see
the effect of ether (petroleum ether), alkyl aromatic (toluene) and aliphatic hydrocarbon (nheptane) on the solvation of the complex in the
reactive extraction of propionic acid. Distilled
water was used to prepare the solutions of various
concentrations of propionic acid solutions. NaOH
used for the titration is of analytical grade and
was supplied by Ranbaxy, India. For the standardization of the NaOH, oxalic acid (99.8%) was
obtained from S.D. Fine Chemicals. Phenolphthalein solution (pH range 8.210.0) was used
as an indicator for titration and was obtained from
Ranbaxy (India). The range of pH of feed solutions for the experiment was 2.653.14 at 305 K.
The initial TBP concentrations in the range 0.37
1.65 kmol/m3 (1040%) and the initial aqueous
acid concentrations range ([HA]o) of 0.05
0.4 kmol/m3 were used. Low concentration was

16

A. Keshav et al. / Desalination 244 (2009) 1223

used because propionic acid concentration in the

fermentation broth is not greater than 0.5 kmol/m3
[13].

3.2. Methods
The extraction experiments were performed
using a temperature-controlled water bath shaker
at room temperature. Equal volumes (25 cm3) of
aqueous and organic phases were shaken for 12 h
[14] and then left to settle for at least 2 h at a
fixed temperature (305 K) and atmospheric pressure (101.2 kPa). Aqueous phase pH was measured by an Orion 3 star pH benchtop (Thermo
Electro). Aqueous phase acid concentration was
determined by titration with NaOH by a UV
spectrophotometer ( = 230) and by HPLC. The
HPLC system (Waters 1523) was composed of a
binary pump, refractive index detector (Waters
2414) and a dual absorbance detector (Waters
2487). The column used was C-18.
The results of the above methods were comparable. The acid content in the organic phase
was determined with a mass balance.

4. Results and discussion

Organophosphorous compounds are effective
extractants and provide higher distribution
coefficients than carbon bonded oxygen bearing
extractants. The chemical stability of organophosphorous compounds plays an important role
in the possibility of its use as an efficient
extracting solution with good separation effect
with solutions containing chemically similar
elements. TBP, an organophosphorous compound, contains a phosphoryl group which is a
stronger Lewis base than a carbonyl one. This
leads to a high distribution coefficient. TBP was
selected because of low water coextraction
(4.67 mass %) and very low solubility in the
aqueous phase (0.039 mass %). TBP contains =
P(O) OH group, which has a marked tendency

towards an intermolecular hydrogen bonding.

Due to the presence of both electron donor and
electron acceptor groups in =P (O) OH grouping,
it undergoes specific interactions like selfassociation and molecular complex formation
with diluents or other solutes. The knowledge of
these factors is necessary for understanding the
mechanism of extraction, the effect of diluents
and role of additional reagents.
Since TBP has a relatively high viscosity
(3.5610!3 Pa s) with specific gravity close to
unity (0.98), it is used along with low viscosity
and low density diluents, which could facilitate
good phase separation in the continuous extraction process. Diluents chosen in the study were
chosen from different chemical classes to examine the effect of diluentcomplex interactions.
These interactions were found to affect the stoichimetry of reaction and magnitude of the corresponding equilibrium constants.
The extraction equilibria of propionic acid by
diluents (physical) and extractantdiluent combinations (chemical) were studied. The physical
extraction using n-heptane, petroleum ether and
toluene is shown in Fig. 2. The regression equations for the physical extraction equilibrium of
propionic acid in three diluents are represented as
2
[HA]org = 0.0061[HA]aq + 0.5012[HA]aq

(8)

for n-heptane

Fig. 2. Physical equilibria for the extraction of propionic

acid in various diluents.

A. Keshav et al. / Desalination 244 (2009) 1223

2
[HA]org = 0.0504[HA]aq + 0.7414[HA]aq

17

(9)

for petroleum ether

2
[HA]org = 0.0875[HA]aq + 1.0942[HA]aq

(10)

for toluene
2
The coefficient of [HA]aq
and [HA]aq provides
the values of dimerization and partition coefficients respectively for the respective extractions
[12]. In all the cases the extraction was low with
the K Ddiluent value lying between 0.0010.17,
0.0020.27 and 0.130.14 for n-heptane, petroleum ether and toluene respectively for the
various initial concentrations of propionic acid in
aqueous phase. Classical extractants, such as
ketones, ethers, hydrocarbons, are thus not able to
fulfill the basic requirements of high distribution
coefficient and high selectivity. The reason for
this may be the higher affinity of propionic acid
for water. Further, solvent extraction with conventional solvents requires very high solvent flow
rates and results in substantial dilution of acid
that will increase the further processing cost [15].
It can be seen that the distribution coefficient
decreases with increase in the molecular weight
of the diluents. Molecular weight has two opposite effects on distribution coefficients. For the
solvents with similar properties, distribution
coefficients generally decrease with increasing
molecular weight, making low molecular weight
solvents more preferable over high molecular
weight ones. However an opposite effect is
observed in distribution coefficients when the
miscibility of water and solvent is considered.
The miscibility of solvents and water increases
with decreasing molecular weight, so larger
amounts of low molecular weight solvents need
to be used for extraction procedures.
Table 1 demonstrates the influence of the
organic solvent and the organic phase concentration of TBP on the distribution coefficient of
propionic acid. TBP concentration was varied
from 1040% (0.371.65 kmol/m3) in various

Fig. 3. Chemical equilibria for the extraction of propionic

acid using 20% TBP in various diluents.

diluents for the acid concentrations of 0.2 and

0.4 kmol/m3 respectively. The function of diluent
is to dilute the extractant to the desired concentration and control the viscosity and density of the
solvent phase. The TBP concentration was not
used above 40% because usually it was found that
the KD increase with the increases of the TBP was
also accompanied with viscosity increase of the
system, which causes the problem of TBP to disengage. Gu et al. [6] found that maximum KD
values were found for the extractant concentration of 2040%.
The K Doverall values for TBP in n-heptane,
petroleum ether and toluene respectively lie in the
range of 0.792.54, 0.672.53 and 0.773.54
(Table 1). The equilibrium curve for chemical
extraction are plotted in Fig. 3, for 20% TBP in
different diluents, and it was found that increasing the TBP concentration in the organic phase
increases the equilibrium concentration of the
acid in the organic phase and hence the higher
distribution. The maximum removal of propionic
acid was for 0.05 kmol/m3 propionic acid solution
with 40% (1.65 kmol/m3) TBP in toluene. This is
in agreement with the theory that reactive extraction is a successful technique to recover acids
from dilute solutions. The chemical equilibrium
can be obtained by plot of organic versus aqueous
concentrations for different concentrations of
TBP in diluent.

18

A. Keshav et al. / Desalination 244 (2009) 1223

Table 1
Summary of extraction equilibrium results for TBP/diluent/propionic acid system at 305 K for the various concentrations
of TBP and acid
Solvent

[HA]0
(kmol/m3)

[HA]aq
(kmol/m3)

[HA]org
(kmol/m3)

[S]0
(kmol/m3)

pHeq

K Doverall

Zt

Petroleum ether +TBP

0.2

n-Heptane+ TBP

0.2

Toluene + TBP

0.2

Petroleum ether + TBP

0.4

n-Heptane + TBP

0.4

Toluene + TBP

0.4

0.118
0.085
0.067
0.057
0.120
0.088
0.068
0.057
0.106
0.072
0.057
0.045
0.224
0.179
0.141
0.123
0.255
0.179
0.140
0.113
0.279
0.153
0.116
0.094

0.082
0.115
0.133
0.143
0.080
0.113
0.132
0.143
0.094
0.128
0.143
0.155
0.176
0.222
0.259
0.277
0.145
0.221
0.260
0.287
0.121
0.247
0.284
0.306

0.37
0.74
1.10
1.65
0.37
0.74
1.10
1.65
0.37
0.74
1.1
1.65
0.37
0.74
1.10
1.65
0.37
0.74
1.1
1.65
0.37
0.74
1.10
1.65

2.93
3.01
3.07
3.11
2.92
3.00
3.06
3.11
2.95
3.05
3.11
3.17
2.78
2.83
2.89
2.92
2.75
2.83
2.89
2.94
2.79
2.98
2.93
2.98

0.695
1.353
1.985
2.509
0.667
1.286
1.941
2.509
0.887
1.778
2.509
3.444
0.786
1.254
1.837
2.252
0.566
1.235
1.857
2.540
0.839
3.190
2.448
3.255

0.119
0.109
0.094
0.071
0.167
0.130
0.107
0.079
0.129
0.117
0.093
0.069
0.266
0.207
0.181
0.136
0.251
0.236
0.199
0.153
0.054
0.213
0.176
0.131

The influence of the extractant concentration

on the degree of extraction [E% = K Doverall 100/
(1 + K Doverall )] was studied at room temperature.
The degree of extraction was found to increase
with increasing the concentration of TBP from
0.37 to 1.65 kmol/m3 in various diluents (Fig. 4).
Table 1 clearly signifies that it is TBP is the main
influencing factor in the extraction and value of
K Doverall increases in a diluent no matter which
diluent was used. This predicts the limit of the
use of diluents with TBP. Thus in inactive
diluents, the value of K Doverall for the extraction of
propionic acid with TBP is proportional to the
concentration of TBP.

Fig. 5 depicts the effect of TBP concentration

on acid loading in organic phase. The plot shows
that loading is independent of the TBP concentration in different diluents. For systems with
only one extractant per complex, there is no effect
of total TBP concentration on the loading and this
is apparent from the result. Thus for nonaggregating systems in inert diluents (alkanes or
aromatic diluents), there was no effect of the TBP
concentration on loading and (1:1) TBP-acid
complex can be assumed to be formed.
The plot of log( K Doverall ! K Ddiluent ) versus
log[S]org according to Eq. (5) is shown in Fig. 6,
in which the slope of unity was obtained for all

A. Keshav et al. / Desalination 244 (2009) 1223

19

Fig. 4. Degree of extraction (E%) against the initial

concentration of TBP in various diluents for the acid concentration of 0.3 kmol/m3.

Fig. 6. Distribution coefficient of propionic acid in TBP

dissolved in various diluents as the function of free TBP
concentration in the organic phase.

Fig. 5. Loading ratio against the propionic acid concentration in the aqueous phase, obtained after equilibrium with various concentrations of TBP in toluene.

Fig. 7. Zt /1!Zt vs. [HA]aq for the estimation of (1:1)

propionic acid TBP complexation constant in n-heptane.

the TBPdiluents (n-heptane, petroleum ether and

toluene) and acid combinations. This again confirms that the acidextractant (1:1) complexes are
formed. Since the loading ratio in all diluents is
less than 0.5, the results of Eq. (5) are also
verified by the plot of following equation:

Zt
K S HAaq
1 Z t

(11)

Fig. 7 represents Eq. (11) for extraction of

propionic acid using TBP + n-heptane. The figure
shows a straight line with slope of KS. This value
is nearly equal to the value obtained in Fig. 6 for

the same system. Similarly it was verified for

other systems. This ideal behavior is observed
due to the ability of diluent to solvate the complex effectively. However, if the diluent is a poor
solvating medium for the species formed, the
polar complexes tend to form clusters in the
organic phase at higher phase loadings in the inert
diluents.
The equilibrium complexation constants for
the extraction using TBP in n-heptane, petroleum
ether and toluene are listed in Table 2. The
comparison clearly shows that maximum KS
values was obtained for the case of TBPtoluene,
thus suggesting it the best system out of the three
for the recovery of propionic acid.
The difference in KS values between with and

20

A. Keshav et al. / Desalination 244 (2009) 1223

Table 2
Equilibrium complexation constants for the extraction of
propionic acid using TBP in various diluents
Extractant

Diluent

[S]o kmol/m3

KS

TBP
TBP
TBP

Petroleum ether
n-heptane
Toluene

0.371.65
0.371.65
0.371.65

1.658
1.715
2.261

without dissociation was evaluated by Morales et

al. [11]. The deviation in all cases or propionic
acid extraction was very less with (1:1) acid:TBP
complexation in all cases. For low acid concentrations, (1:1) propionic acidTBP complex is
common. The formation of (2:1) and (3:1) complexes depends on the propionic acid concentration in the aqueous phase, and the ratio of (1:1)
to (2:1) complex formation is diluent dependent.
At higher concentrations of propionic acid, the
(2:2) and (3:1) complexes can form. This overloading phenomenon results from second propionic acid molecule hydrogen bonding to the
propionic acid that is already involved in the (1:1)
complex. However, in the present study, since no
overloading was found, only the (1:1) complex
was formed and the complexing formula is shown
in Fig. 8.
Different approaches have been used to quantify the effect of diluents on the (1:1) complexation. Both partition and self-association constants
are strongly dependent on the nature of the
diluents; i.e. the thermodynamic activity of the
species taking part in the organic phase equilibria
is changed when one diluent is replaced by
another. The effect of diluent on partition and
self-association constants was explained by specific interactions only between the organophosphorous monomeric species and the diluents;
these interactions are higher and lower than the
partition coefficient and self-association constants should be, respectively. This interpretation
is acceptable for petroleum ether; however, it
cannot apply to the significant changes in the

Fig. 8. Complexing formula of (1:1) propionic acidTBP

complex.

equilibrium constant caused by replacing typically inert alkenes diluent by other inert diluents.
Neither can toluene be assumed to interact
predominantly in the grouping in the organophosphorous compound because the overall
distribution coefficient value of organophosphorous compounds [16] decreases in the order in
which the ability of aromatic diluents to act as
donors are known to increase.
King et al. [17] suggested the use of the
Hilderband solubility parameter as the measure of
solvation of the complex by the diluent. The
sequence alkane < ethers < aromatic corresponds
to increasing acid distribution and increasing
solubility parameters. The same sequence for the
distribution coefficient was found by TBP in the
corresponding diluents (n-heptane, petroleum
ether and toluene). The large difference among KS
values for the propionic acid extraction using
TBP in different diluents indicates that solvation
of the complex in different diluents is a critical
factor in acid extraction. Alkane (n-heptane)
being non-polar provides for very little solvation
of the polar complexes. Aromatic diluent provides higher KS values, which have been
rationalized as the effect of the aromatic electron on the solvation of the complex.
In some of the works, a relationship was found
between the distribution coefficient and other
physiochemical diluent parameters such as
dielectric constant, dipole moment etc. However,
more often no correlation between these parameters was found because the effect of the
diluents is determined by the ratio of the

A. Keshav et al. / Desalination 244 (2009) 1223

contributions of the solvation of the extractant

and the extracted complex to the free energy of
extraction. Solvation, in turn, depends on
different kinds of intermolecular forces which
cannot be determined merely from the abovementioned physical parameters. However in some
cases solvation energy is represented in terms of
empirical parameters of solvent polarity. The
parameters most frequently used are the Z
parameter (proposed by Kosower [18]), which
accounts for the influence of the solvent on the
position of the charge transfer band in the
spectrum of alkylpyridine iodide, and the ET
parameter [19], which is based on the absorption
spectrum of pyridinium-N-phenol-betaine. The
greater the ET and Z values, the greater the
solvating power of the solvent. The higher value
of distribution coefficient of toluene over the
other two diluents can thus also be interpreted
because of the higher ET values of toluene
(34.1 kcal/mole).
A continual reactive extraction process for
recovery of carboxylic acid requires a solvent
with minimal toxicity and maximal extraction
capacity. Toxicity is the inhibitory effect of
solvent on the catalytic activity of enzymes.
Toxicity depends both on type of solvent and
type of microorganism used. Varying results were
obtained depending on whether yeast or bacteria
were used and on the particular strain also [20].
Alkanes are practically non-toxic in comparison
to aromatic hydrocarbons and higher alcohols.
Similarly extractants are toxic to varying degrees.
Alamine 336 (20%) in isodecanol causes a
growth rate of <2 g/l, and TBP (100%) causes a
growth rate of n1 g/l in the case of lactic acid
production [21]. So this is another reason why
100% TBP was not used in present study to
recover propionic acid.
A mixture of TBP in relatively non-toxic
diluents can be an important solvent system.
Lower toxicity at higher concentration of carriers
is probably due to their association and consequently blockage of cytotoxically active groups.

21

Strategies for reducing toxicity during reactive

extraction have been previously discussed [22].
Use of soybean oil [6] or second-stage extraction
of amines with oleyl alcohol before returning the
broth to fermentation [23], and immobilization of
cells into calcium alginate gel [24] are a few
methods to remove solvent toxicity. Demirici et
al. [25] found that 5% TBP in isooctane was
acceptable in lactic acid biofilm fermentation
with Lactobacillus casei. Thorough research is
still required to provide an efficient extractant
diluent for propionic acid recovery on which
work is in progress.
Another problem is the co-extraction of water
along with acid into the organic phase. n-heptane
has only traces of solubility in water, thus the
TBP+n-heptane system behaves closely to ideality in terms of volume changes when propionic
acid at low concentration partition between. Low
mutual solubility of n-heptane + water over
toluene + water is thus another advantage that
suggests it to be a better diluent for extraction of
propionic acid using TBP.

5. Conclusions
Fermentation is one of the oldest technologies
for the production of carboxylic and hydrocarboxylic acids. Though being an energy efficient technology, it suffers from the disadvantages of acid production in low concentration and
difficult separation and purification of the acid.
Conventional recovery method being energy
inefficient and environment unfriendly leads the
researchers to look for alternative methods.
Reactive extraction came out to be one of the
most successful recovery methods with the
advantage of providing high capacity and selectivity. Reactive extraction using amines extractant
has been widely studied for the extraction of
lactic acid, citric acid, and propionic acid etc., but
very little work has been done on the phosphorous bonded oxygen donor extractants.

22

A. Keshav et al. / Desalination 244 (2009) 1223

Recovery of propionic acid using TBP in various

diluents (n-heptane, petroleum ether and toluene)
was studied. Diluents improve the solvation and
complex stability. Of the different diluents used,
toluene was found to give a higher distribution
coefficient, equilibrium complexation constant,
and loading and extraction efficiency, thus
indicating it as the best solvent for propionic acid
recovery. Though the above respective values are
higher for the TBPtoluene system, when the
question is about the extraction from fermentation
broth, it has been found that the TBPn-heptane
system was preferable as it is less toxic and has
low water co-extraction values in comparison to
other two TBPdiluent systems.

Subscripts
aq
org
p
0

Aqueous phase
Organic phase
Solvation number of TBP
Initial

Acknowledgement
The Department of Science and Technology
(DST), INDIA, Young Scientist Project, SR/FTP/
ETA-43/2005, Reactive Extraction of Propionic
acid (Dr. Kailas L. Wasewar, IIT Roorkee, India).

References
6. Symbols
!

[A ]
E%
[H+]

Concentration of anion, kmol/m

Degree of extraction
Concentration of hydroxyl ion,
kmol/m3
[HA] Concentration of acid, kmol/m3
Concentration of acid extracted into
HAdiluent
org
organic phase by diluent alone,
kmol/m3
[HA.S] Concentration of acidextractant
complex, kmol/m3
overall
K D Overall distribution coefficient of
acid
chem
KD
Distribution coefficient of acid by
chemical complexation
K Ddiluent Distribution coefficient of acid by
diluent alone
Extraction equilibrium constant,
KS
m3/kmol
Dissociation constant, m3/kmol
KHA
[S]
Concentration of extractants,
kmol/m3
Zt
Loading ratio
Greek

Volume fraction of diluent

[1] M.J. Playne and M. Moo-Young, eds., Propionic and

Butyric acids, Comprehensive Biotechnology,
Vol. 3, Pergamon, New York, 1985, pp. 731754.
[2] Ullmanns Encyclopedia of Industrial Chemistry, 6th
ed., Vol. 26, Wiley-VCH, pp. 261278.
[3] K.L. Wasewar, A.B.M. Heesink, G.F. Versteeg and
V.G. Pangarkar, Chem. Eng. Sci., 58(15) (2003)
33853394.
[4] Y. Marcus and A.S. Kertes, Ion Exchange and
Solvent Extraction of Metal Complexes, Wiley
Interscience, London, 1969.
[5] M.N. Ingale and V.V. Mahajani, Sep. Technol., 4
(1994) 123126.
[6] G. Zhong, B.A. Glatz and C.A. Glatz, Biotechnol.
Bioeng., 57 (1998) 454461.
[7] M. Matsumoto, T. Otano and K. Kondo, Sep. Purif.
Technol., 24 (2001) 337342.
[8] K.L. Wasewar and V.G. Pangarkar, Chem. Biochem.
Eng. Q., 20(3) (2006) 325331.
[9] G. Malmary, J. Albet, A. Putranto, H. Hanine and
J. Molinier, J. Chem. Technol. Biotechnol., 75
(2000) 11691173.
[10] M.N. Ingale and V.V. Mahajani, Sep. Technol., 6
(1996) 17.
[11] A.F. Morales, J. Albet, G. Kyuchoukov, G. Malmary
and J. Molinier, J. Chem. Eng. Data, 48 (2003) 874
886.
[12] A.S. Kertes and C.J. King, Biotechnol. Bioeng., 28
(1986) 269282.

A. Keshav et al. / Desalination 244 (2009) 1223

[13] Z. Jin and S.T. Yang, Biotechnol. Prog., 14 (1998)
457465.
[14] C.J. King, J.A. Tamada and A.S. Kertes, Ind. Eng.
Chem. Res., 29 (1990) 13191326.
[15] I. Inci and A. Aydin, J. Sci. Ind. Res., 62 (2003)
926930.
[16] Z. Kolarik, J. Hejna and H. Pankova, J. Inorg. Nucl.
Chem., 30 (1968) 253258.
[17] J.M. Wardell and C.J. King, J. Chem. Eng. Data,
23(2) (1979) 144148.
[18] E.M. Kosewar, J. Am. Chem. Soc., 80 (1958) 3253
3261.
[19] K. Dimroth, C. Reichardt, T. Siepmann and F.
Bohlam, Justus Liebigs Ann. Chem., (1963) 661
670.
[20] S.R. Rofller, T.W. Randolph, D.A. Miller, H.W.

[21]
[22]
[23]

[24]
[25]

23

Blanch and J.M. Prausnitz, Extractive bioconversions

with non aqueous solvents, in: B. Matson and
O. Holst, eds., Extractive Bioconversions, Marcel
Dekker, New York, 1991, pp. 135140.
J. Martak, M. Rosenberg, S. Schlosser and L.
Kristofkova, Biotechnol. Tech., 9 (1995) 247252.
V.M. Yabannavar and D.I.C. Wang, Biotechnol.
Bioeng., 37 (1991) 716723.
H. Honda, Y. Toyama, H. Takahashi, T. Nakazeko
and T. Kobayashi, J. Ferm. Bioeng., 79 (1995) 589
793.
G. Zhong, D.A. Rickert, B.A. Glatz and C.E. Glatz,
LAIT, 79 (1999) 137148.
A. Demirci, J.C. Cotton, A.L. Pometto, K.R. Harkins
and P.N. Hinz, Biotechnol. Bioeng., 83 (2003) 749
759.