Research Article

Received: 4 January 2012 Revised: 14 March 2012 Accepted: 16 April 2012 Published online in Wiley Online Library:

(wileyonlinelibrary.com) DOI 10.1002/jctb.3847

Investigation of liquid–liquid phase equilibria

for reactive extraction of lactic acid with
organophosphorus solvents
Susanne Lux∗ and Matthäus Siebenhofer

Abstract
BACKGROUND: Lactic acid is a versatile commodity chemical with a variety of applications. Synthesis of lactic acid either through
fermentation of carbohydrates or through chemical synthesis is state of the art. Separation from dilute aqueous solution is
complex and accounts for the major part of production costs. Reactive extraction based on reversible adduct formation is a
promising alternative for the separation of lactic acid.

RESULTS: Extraction was carried out with the organophosphorus solvents tri-n-butyl phosphate, tri-n-octylphosphine oxide
and Cyanex 923. Shellsol T was used as diluent. Partition coefficients increase with increasing extractant content and decreasing
temperature. Cyanex 923 has several advantages compared with tri-n-butyl phosphate and tri-n-octylphosphine oxide with
respect to lactic acid distribution and hydrodynamic properties. Liquid-liquid phase equilibria for lactic acid extraction with
Cyanex 923 were modeled. Selectivity of lactic acid extraction with respect to glycolic acid and formic acid was discussed.

CONCLUSION: The organophosphorus extractant Cyanex 923 was found to be an appropriate solvent for lactic acid extraction
from aqueous solutions. Experimental data and model data based on the Law of Mass Action showed good agreement. Lactic
acid extraction from multi-acid solution showed good selectivity compared with glycolic acid. Lactic acid selectivity is low with
respect to formic acid.

c 2012 Society of Chemical Industry

Keywords: reactive extraction; lactic acid; Cyanex 923; liquid–liquid equilibrium

SYMBOLS AND NOTATION Synthesis of lactic acid either through fermentation

a acitivity of carbohydrates or through chemical synthesis from 2-
aqu aqueous phase hydroxypropanenitrile (lactonitrile) are state-of-the-art routes.2
α degree of dissociation Glycerol based synthesis of lactic acid is a new approach in pro-
c concentration cessing fats and oils.3 Enantiopure lactic acid can be produced
γ activity coefficient via fermentation. Consequently, 90% of lactic acid is produced
HA carboxylic acid through this route.2
KD partition coefficient The quality grade of lactic acid depends on the usage.
Kdiss dissociation constant Purification requirements for polylactic acid production are
Kst stoichiometric equilibrium constant considerable. The conventional route for lactic acid recovery is
Kth thermodynamic equilibrium constant based on precipitation of poorly soluble calcium lactate with
n association coefficient Ca(OH)2 or CaCO3 . The precipitated calcium lactate cake is filtered
org organic phase off, washed and reacidified with strong mineral acids, leading to
RPO phosphorus-bonded oxygen donor extractant free lactic acid and CaSO4 .4 This conventional downstream process
nondiss non-dissociated needs huge amounts of chemicals and generates large quantities
V volume of gypsum sludge. The sludge is difficult to dispose of due to
retained organic impurities. There is loss of calcium lactate along
INTRODUCTION with the washing (5% at 20 ◦ C in the mother liquor) and additional
2-Hydroxypropionic acid (lactic acid, LA) is a versatile commodity
chemical with a variety of applications in food, beverages,
chemical, pharmaceutical, and cosmetics industries. The food ∗
Correspondence to: Susanne Lux, Institute of Chemical Engineering and
industry used to be the main consumer of lactic acid and lactates.
Environmental Technology, Graz University of Technology, Inffeldgasse 25C,
In the last ten years consumption of lactic acid for industrial use has 8010 Graz, Austria. E-mail: [email protected]
become the leading market. Future growth in demand is expected
to be driven mainly by lactic acid-based polymers and, to a lesser Institute of Chemical Engineering and Environmental Technology, Graz
degree, lactate solvents.1 University of Technology, Inffeldgasse 25C, 8010 Graz, Austria

J Chem Technol Biotechnol (2012) www.soci.org

c 2012 Society of Chemical Industry
 www.soci.org S Lux, M Siebenhofer

purification steps are required to produce commercial grade lactic Morales et al.17 investigated extraction equilibria of different
acid.5 monocarboxylic acids, including lactic acid, with tri-n-butyl
Formation of calcium sulfate waste can be avoided by extractive phosphate in dodecane. Extractant concentration was varied
separation of lactic acid. Due to high polarity and the hydrophilic between 10 and 80 vol%. Matsumoto et al.18 investigated the
character of lactic acid, it is poorly extracted with conventional synergistic effect of alkylamine and tri-n-butyl phosphate on lactic
organic extractants. High consumption of solvent is inevitable. acid extraction.
Jenemann6 used iso-propyl ether as solvent. For one volume of Yabannavar and Wang13 used tri-n-octylphosphine oxide
aqueous lactic acid solution 50 volumes of iso-propyl ether were dissolved in dodecane and Isopar-M for lactic acid extraction
applied due to unfavorable distribution. from model solutions and fermentation broths. Feed lactic acid
Addition of reactive components to the organic solvent may concentration was varied between 1 and 6 g L−1 and TOPO content
increase lactic acid distribution. The organic extractant reacts between 5 and 15%. Diluents dodecane and Isopar-M showed a
with the solute at the water/solvent interface. The hydrophobic similar behavior.
reaction adduct formed is then solubilized into the organic phase. Extraction equilibria of lactic acid and other monocarboxylic
Reactive extraction of carboxylic acids may be based on solvation acids with trialkylphosphine oxide CH3 (CH2 )7 – 9 )3 =O in kerosene
with carbon-bonded or phosphorus-bonded oxygen-bearing were studied by Wang et al.11 Extraction equilibrium constant was
extractants and on ion-pairing formation with high-molecular found to be 1.0 mol L−1 .
weight aliphatic amines.7 High partition coefficients are achieved Frieling and Schügerl5 investigated lactic extraction with Cyanex
with secondary and tertiary amines. But re-extraction is often 923 in kerosene. Extraction equilibrium studies were carried out
carried out with bases because of high thermal stability due to the for Cyanex 923 concentration ranging from 0.11 to 0.85 mol L−1 .
ionic character of amine salts.8 With increasing concentration of Lactic acid production through fermentation could benefit
bases the tendency for emulsification increases. from integrated recovery by extraction.19 Biocompatibility of
Extraction with phosphorus-bonded oxygen-bearing extrac- extractants is often unsatisfactory though. Alex et al.20 suggested
tants is promising. Selectivity is affected by strength of the use of hollow fibre liquid–liquid contactors. This membrane
donor–acceptor interaction and solvation. Due to their higher process technology could broaden the use of liquid–liquid
Lewis basicity phosphorus-bonded oxygen donor extractants pro- extraction.
vide better distribution of carboxylic acids than carbon-bonded Main objective of this work was an extensive study on lactic
oxygen donor extractants. Basicity of the phosphoryl group is acid extraction with different organophosphorus extractants. Tri-
determined by its substituents. Alkoxy groups suppress basicity n-butyl phosphate, tri-n-octylphosphine oxide and Cyanex 923
whereas alkyl substituents increase it. Generally, extraction ca- diluted with Shellsol T were used as extactants. Equilibrium
pability of carboxylic acid uptake increases in the order: trialkyl data were given for highly concentrated and also undiluted
phosphate ((RO)3 P=O) < alkyl dialkylphosphinate ((RO)2 P=O) < extractant, which had been scarce in the literature so far. The
trialkylphosphine oxide (R3 P=O). With tributyl- and trioctylphos- effect of extractant content, temperature, and initial lactic acid
phine oxides, higher partition coefficients are achieved than with concentration on lactic acid distribution and water co-extraction
triphenylphosphine oxides.9,10 was investigated. Phase equilibrium data were compared with a
According to Wang et al.11 the degree of extraction is controlled model based on the Law of Mass Action.
by two effects. One is the strength of the acid. The stronger the
acid, the stronger the interaction with the organophosphorus
extractant will be. Second, hydrophobicity of the acid will affect
extraction. Hano et al.12 concluded that the extraction equilibrium MODELING LACTIC ACID EXTRACTION WITH
is only controlled by the hydrophobicity of the acid, not by the ORGANOPHOSPHORUS SOLVENTS
pKa value. As the extractant forms adducts with the carboxylic The extraction equilibrium of lactic acid with organophosphorus
acid, the pH-value of the aqueous solution should be below the compounds can be represented by the general reaction equation
pKa value of the acid.13 Partition coefficients of reactive extraction
with organophosphorus extractants increase with decreasing acid nHAaqu + RPOorg ↔ (RPO(HA)n )org (1)
concentration and decreasing temperature.5,11
Typically, reactive extractants are highly viscous. Tri-n-butyl
phosphate (TBP) is highly viscous at room temperature (3.8 mPas with n representing the number of reacting molecules of lactic
at 20 ◦ C) and density is close to one. Tri-n-octylphosphine oxide acid (HA) per molecule of organophosphorus extractant (RPO).
(TOPO) is in solid state at ambient temperature. Improvement The overall partition coefficient KD of the extracted acid represents
of hydrodynamics is obtained with diluents.14 According to the ratio of the total acid concentration in the solvent phase to the
Yabannavar and Wang13 tri-n-octylphosphine oxide can only total acid concentration in the aqueous phase.
be applied after dissolving in nonpolar diluents. Polar diluents
interfere with the extraction step by disintegrating the laden cHA,org
KD = (2)
solvent phase. A liquid phosphine oxide (Cyanex 923) was cHA,aqu
developed by Watson et al.15 It consists of a mixture of four
trialkylphoshine oxides with n-hexyl-and n-octyl substituents. Deviation from ideal (Nernst) partition has to be assumed in acid
Use of tri-n-butyl phosphate for the extraction of carboxylic acids extraction with organophosphorus based solvents. This includes
was already reported by Pagel and McLafferty16 in 1948. Decrease dissociation of the acid in the aqueous phase and association
of extraction behavior with increasing acid concentration was in the solvent phase. Distribution equilibria of carboxylic acids
observed. But due to minor change of partition coefficients it was with organophosphorus extractants are therefore analytically best
concluded that TBP does not associate with the acid. described by an algorithm according to the Law of Mass Action

wileyonlinelibrary.com/jctb

c 2012 Society of Chemical Industry J Chem Technol Biotechnol (2012)
Reactive extraction of lactic acid with organophosphorus solvents www.soci.org

with the thermodynamic equilibrium constant Kth : 0.14

aRPO(HA)n ,org 0.12

Kth =
anHA,aqu · aRPO,org

Degree of dissociation a
cRPO(HA)n ,org · γRPO(HA)n ,org 0.1
= n (3)
cHA,aqu · γHA,aqu
n
· cRPO,org · γRPO,org
0.08
Whereas the thermodynamic equilibrium constant Kth is inde-
pendent of acid concentration, the stoichiometric equilibrium 0.06
constant Kst may change with increasing acid concentration, since
the activity coefficients are likely to change as well. 0.04

cRPO(HA)n ,org 0.02

Kst = (4)
n
cHA,aqu · cRPO,org
0
0 20 40 60 80
The thermodynamic equilibrium constant Kth and the stoichio-
clactic acid (g/L)
metric equilibrium constant Kst are correlated by the product of
activity coefficients. With the assumption of γ = 1 under relatively Figure 1. Degree of lactic acid dissociation for different lactic acid
high dilution conditions for the substances involved, Kth equals concentrations (T = 20 ◦ C).
Kst .
It has to be taken into account that non-dissociated carboxylic
acid is extracted into the organic phase due to adduct formation plotting the inverse equilibrium concentration c 1 against
HA,org
c 1
with the extractant.21 Substitution of cRPO(HA)n ,org through HA,org
n n . This results in the linear equation shown in
c cHA,aqu,nondiss
and cRPO,org = cRPO,0,org − HA,org
n then results in a simplified version Equation (10).
of Equation (3)
1 1 1
cHA,org = + (10)
Kst = (5) cHA,org n · Kst · cHA,aqu,nondiss
n
· cRPO,0,org cRPO,0,org · n
n · cHA,aqu,nondiss
n
· (cRPO,0,org − cHA,org /n)
With specified initial concentration of the extractant cRPO,0,org and
Dissociation of monobasic acids is expressed by the dissociation by altering the degree of association the stoichiometric equilibrium
constant KDiss constant Kst is calculated from the slope.
cH+ · cA− cHA,0 2 · α 2
KDiss = = (6)
cHA cHA,0 · (1 − α) MATERIALS AND METHODS
Chemicals
The non-dissociated fraction of the carboxylic acid is expressed Lactic acid (pur., Riedel-de Haën) was provided in aqueous
through solution with a concentration of 1 mol L−1 . Glycolic acid (2-
cHA,aqu,nondiss = cHA,aqu · (1 − α) (7) hydroxyethanoic acid, 98%, p.a.) and formic acid (methanoic acid,
with the dissociation degree α 99%, p.a.) were purchased from Merck.
The extractants tri-n-octylphosphine oxide (TOPO) and Cyanex

923, a mixture of tertiary n-octyl- and n-hexylphosphine oxides,
−KDiss ± KDiss 2 + 4 · cHA,0 · KDiss were supplied by Cytec Industries Inc. The content of trialkylphos-
α= (8)
2 · cHA,0 phine oxides was 93% in both extractants. Tri-n-butylphosphate
(TBP, 99%, pur.) was purchased from Merck. All extractants were
Degree of dissociation is low at high acid concentration but, as used without further purification. The diluent Shellsol T (SST), an
shown in Fig. 1, it cannot be neglected at low acid concentration. isoparaffinic hydrocarbon solvent with a hydrocarbon range of
Consequently, Equation (9) represents the analytical expression C11 to C13 and a distillation range of 187–213 ◦ C, was supplied by
of liquid–liquid phase equilibria for extraction with chemical re- Shell Chemicals.
action. The equilibrium concentration of the species extracted Tri-n-butyl phosphate, tri-n-octylphosphine oxide, and in some
into the solvent phase cHA,org is given as a function of feed experiments Cyanex 923 were diluted with Shellsol T. Cyanex
concentration of reactive extractant cRPO,0,org and the correspond- 923 and tri-n-butyl phosphate were also applied without dilution.
ing concentration of non-dissociated acid in the aqueous phase Aqueous solutions of lactic acid were prepared by dissolving
(cHA,aqu,nondiss ). different quantities of the acid in deionized water without pH
adjustment.
n · Kst · cHA,aqu,nondiss
n
· cRPO,0,org
cHA,org = (9)
1 + Kst · cHA,aqu,nondiss
n
Extraction experiments
Extraction experiments were carried out in temperature controlled
Determination of equilibrium acid concentration in the organic mixing funnels. A solvent to water ratio of 1 : 2 was applied,
phase, degree of association n, stoichiometric equilibrium constant yielding in 25 mL of organic and 50 mL of aqueous phase. The
Kst and maximum concentration n · cRPO,0,org is achieved by solvent phase and the aqueous phase were equilibrated for

J Chem Technol Biotechnol (2012)

c 2012 Society of Chemical Industry wileyonlinelibrary.com/jctb
 www.soci.org S Lux, M Siebenhofer

30 min on a mechanical shaker (Janke&Kunkel HS-500) with a 5

frequency of 150–160 rpm. In previous experiments it was proven

Partition coefficient KD,lactic acid

that equilibrium is achieved within 30 min. After equilibration the Cyanex 923
mixture was allowed to settle in the temperature controlled flasks 4
TBP
for 20 h. In most experiments time was sufficient for complete TOPO
phase separation. In some cases, when TBP was used as extractant 3
at 25 ◦ C, further centrifugation was required for separation of the
two emulsified phases. Emulsification tendency of Cyanex 923 is
low, even when it is applied without dilution. 2
After phase separation the concentration of lactic acid in the
aqueous phase and the water content in the solvent phase were
1
determined. Equilibrium concentration of lactic acid in the solvent
phase was calculated via mass balance. Mass and density of the
aqueous and the solvent phase were measured before and after 0
extraction. The pH of the aqueous feed lactic acid solution was 0 20 40 60 80 100
determined. Concentration of trialkylphosphine oxides cRPO,0 in Extractant content (%)
TOPO and Cyanex 923 were used as specified by the supplier.
Figure 2. Partition coefficients of lactic acid extraction with Cyanex 923,
Extractant solubility in water was specified by the supplier too. As
TBP and TOPO for different extractant contents at 25 ◦ C (black), 40 ◦ C (grey)
Cyanex 923 solubility in water is 0.001 wt%, losses of extractant and 60 ◦ C (white) (clactic acid,0 = 2 g L−1 ; Vorg : Vaqu = 1 : 2).
with the aqueous phase are of minor relevance.

Sample analysis Partition coefficients of lactic acid extraction with TBP are low
Lactic acid was analyzed by high-performance liquid chromatogra- with KD = 0.30 for 50 wt% TBP and KD = 0.85 for undiluted TBP
phy (HPLC). The HPLC system Ultimate 3000 (Dionex) was equipped at 25 ◦ C. Further decrease of partition coefficients to 0.16 and 0.63
with the column Acclaim OA (Dionex). Detection was carried out is observed with increasing temperature to 60 ◦ C. Similar results
with a UV/VIS detector (Dionex) at 210 nm. Elution of substances were obtained by Matsumoto et al.18 for lactic acid extraction
from the column was carried out with 0.1 mol L−1 Na2 SO4 in ul- (0.1 mol L−1 ) with TBP in hexane. Partition coefficients were 0.4
trapure water at 30 ◦ C. The pH-value of 2.65 was adjusted with with 50 wt% TBP (2 mol L−1 ) and 0.7 when using 80 wt% TBP
methanesulfonic acid. (3 mol L−1 ) at 25 ◦ C.
For Karl Fischer titration the titration manager Radiometer Highest partition coefficients are achieved with Cyanex 923. At
Copenhagen equipped with VIT 90 video titrator, ABU 93 triburette 25 ◦ C partition coefficients increase from 2.14 (50 wt% Cyanex 923)
and SAM 90 sample station was used. Mean deviation of lactic acid to 4.46 (undiluted Cyanex 923). Partition coefficients decrease to
concentration and water content was determined to be 3%. Mean 1.69 (50 wt% Cyanex 923) and 3.27 (undiluted Cyanex 923) with
error for KD values was 0.05. increasing temperature to 60 ◦ C.
Partition coefficients as a function of extractant content are
comparable for TOPO and Cyanex 923, expected due to the similar
RESULTS AND DISCUSSION hydrocarbon substituents of both extractants. All extractants show
Screening of extractants TBP, TOPO and Cyanex 923 linear correlation of partition coefficients and extractant content,
Extractability of the extractants TBP, TOPO and Cyanex 923 for confirming that the diluent Shellsol T has no impact on lactic acid
lactic acid extraction was compared at different temperatures distribution.
(25 ◦ C, 40 ◦ C and 60 ◦ C) and feed extractant contents (Fig. 2). TBP The amount of water dissolved and co-extracted into the organic
and Cyanex 923 were used in the concentration range 30 to 70 phase is shown in Fig. 3 for various extractant contents at 25 ◦ C,
wt% in Shellsol T, and without dilution. Mass fraction of TOPO in 40 ◦ C and 60 ◦ C. Maximum water content is 8.7 wt% when using
Shellsol T was varied between 5 and 20% due to limited solubility undiluted Cyanex 923 at 25 ◦ C.
of TOPO at ambient temperature. Feed lactic acid concentration
was 2 g L−1 . Extraction with Cyanex 923
Partition coefficients for reactive extraction of carboxylic acids For lactic acid extraction from dilute aqueous solution Cyanex
with organophosphorus compounds increase from phosphate 923 shows several advantages over TBP and TOPO. Apart from
via phosphonate to phosphine oxide.9,10 Increasing temperature achieving higher partition coefficients compared with TBP, Cyanex
results in a decrease of partition coefficients.5,11 Both trends 923 does not tend to form emulsions. In contrast to TOPO
are confirmed for lactic acid extraction with TBP, TOPO and the liquid phosphine oxide Cyanex 923 can be used at high
Cyanex 923. concentration and even without dilution. Cyanex 923 can be used
Extraction capacity of TOPO based solvents is low due to its at elevated temperature as well as ambient temperature. All further
poor solubility in Shellsol T resulting in low TOPO concentration investigation of extraction equilibria was carried out with Cyanex
in the solvent phase. At 25 ◦ C partition coefficients increase from 923 at 20 ◦ C.
0.20 to 0.86 with increasing TOPO content from 5 wt% to 20 wt%. As may be concluded from Equation (9) partition coefficients
Increasing temperature results in decrease of partition coefficients for acid extraction with organophosphorus compounds are
to 0.10 and 0.64, respectively. Partition coefficients derived in this influenced by the initial acid concentration. The higher the
work are comparable to those given by Yabannavar and Wang13 initial acid concentration in the aqueous phase, the smaller is
for lactic acid extraction with TOPO in dodecane. For a feed lactic the partition coefficient.
acid concentration of 1 g L−1 partition coefficients were reported Liquid–liquid phase equilibria for different concentrations of
to be 0.35 and 0.95 with 5 wt% and 15 wt% TOPO extractant. Cyanex 923 in the solvent phase are given in Table 1. As may be

wileyonlinelibrary.com/jctb

c 2012 Society of Chemical Industry J Chem Technol Biotechnol (2012)
Reactive extraction of lactic acid with organophosphorus solvents www.soci.org

9 120
Water content of the organic phase (%)

8
Cyanex 923 100
7 TBP
TOPO
6 80

corg-1 (kg/mol)
5
60
4

3 40
30%
2 50%
70%
20
1 100%

0
0 20 40 60 80 100 0
0 100 200 300 400
Extractant content (%) (caqun·(1-α)n)-1 (kg/mol)
Figure 3. Equilibrium water contents of solvent phases for lactic acid
extraction with Cyanex 923, TBP and TOPO for different extractant contents Figure 4. Determination of the association factor n, representing the
at 25 ◦ C (black), 40 ◦ C (grey) and 60 ◦ C (white) (clactic acid,0 = 2 g L−1 , number of molecules of lactic acid reacting per molecule of phosphine
Vorg : Vaqu = 1 : 2). oxide, and the stoichiometric equilibrium constant Kst according to
Equation (10).

deduced from Table 1, partition of lactic acid depends both on

the initial acid concentration and the Cyanex 923 content in the Table 2. Association factor n and stoichiometric equilibrium constant
solvent phase. Kst for different extractant contents at 20 ◦ C

Determination of the association factor n wCyanex 923 (%) n Kst (kg mol−1 )
The association factor n represents the number of reacting 30 0.93 1.51
molecules of lactic acid per molecule of organophosphorus 50 0.96 1.49
compound. For determination of the association factor n the 70 0.94 1.58
inverse concentration of lactic acid in the organic phase is plotted 100 0.95 1.51
over the inverse concentration of non-dissociated lactic acid in the
mean: 0.95 ± 0.01 1.52 ± 0.04
aqueous phase according to Equation (10) (Fig. 4). The assumption
γ = 1 for all substances involved was taken as a basis, and proven
as valid.
Association factors n for different extractant contents were The total water content of the organic phase arises both from
derived from best fit data. The association factors are given in water dissolution in the organic phase and water co-extraction
Table 2, indicating that 1 : 1 adducts of lactic acid and Cyanex with the lactic acid extraction. The amount of water co-extracted
923 are formed. The stoichiometric equilibrium constant is then into the organic phase is given by subtraction of the amount
obtained from the slope with Kst = 1.52 ± 0.04 kg mol−1 . of water dissolved in the organic phase from the total water
content. 1 mole of water is co-extracted per mole of lactic acid
Investigation of water co-extraction when using 30 wt% of Cyanex 923 in Shellsol T. With increasing
Water co-extraction into the organic phase was investigated for Cyanex 923 content the amount of lactic acid extracted into the
Cyanex 923 at 25 ◦ C. Feed lactic acid concentration was 2 g L−1 . organic phase increases. Higher concentration of lactic acid in the

Table 1. Liquid–liquid phase equilibria for lactic acid (LA) extraction with Cyanex 923 at 20 ◦ C (Vorg : Vaqu = 1 : 2; mean deviation 3%)

c (mol kg−1 ) 30% Cyanex 923

cLA,aqu 0.007 0.014 0.069 0.102 0.161 0.199 0.290 0.378 0.608
cLA,org 0.010 0.019 0.090 0.119 0.133 0.188 0.239 0.280 0.350
50% Cyanex 923
cLA,aqu 0.006 0.011 0.056 0.081 0.124 0.169 0.264 0.315 0.525
cLA,org 0.013 0.025 0.119 0.166 0.219 0.255 0.308 0.427 0.545
70% Cyanex 923
cLA,aqu 0.004 0.009 0.044 0.067 0.102 0.157 0.131 0.200 0.266 0.358
cLA,org 0.012 0.030 0.130 0.193 0.270 0.290 0.392 0.440 0.526 0.603
100% Cyanex 923
cLA,aqu 0.004 0.007 0.036 0.054 0.079 0.104 0.099 0.159 0.214 0.374
cLA,org 0.016 0.031 0.149 0.212 0.302 0.374 0.433 0.549 0.597 0.931

J Chem Technol Biotechnol (2012)

c 2012 Society of Chemical Industry wileyonlinelibrary.com/jctb
 www.soci.org S Lux, M Siebenhofer

with glycolic acid. Lactic acid selectivity is low with respect to

Table 3. Partition coefficients KD for carboxylic acid extraction with
Cyanex 923 from aqueous solution containing lactic acid (LA), glycolic formic acid.
acid (GA) and formic acid (FA) (Vorg : Vaqu = 1 : 2, cLA,0 = 0.06 mol L−1 ,
molar feed ratio (LA : GA : FA) = 1 : 1 : 2, diluent: Shellsol T, 20 ◦ C)
Cyanex Lactic Glycolic Formic REFERENCES
923 (wt%) acid acid acid 1 Malveda MP, Blagoev M and Kumamoto T, Lactic acid, its salts and
esters. CEH Marketing Research Report, SRI Consulting (2009).
100% 3.55 1.87 10.75 2 Chahal SP and Starr JN, Lactic acid, in Ullmann’ s Encyclopedia
of Industrial Chemistry. Electronic Release, Wiley-VCH, Weinheim
70% 2.53 1.08 8.50
(2006).
3 Lux S, Stehring P and Siebenhofer M, Lactic acid production as a new
approach for exploitation of glycerol. Sep Sci Technol 45:1921–1927
(2010).
organic phase provides higher hydrophilic behavior, which makes 4 Inskeep GC, Taylor GG and Breitzke WC, Lactic acid from corn sugar.
water co-extraction favorable. The amount of water co-extracted Ind Eng Chem 44:1955–1966 (1952).
increases to 6 and 8 moles of water per mole of extracted lactic acid 5 von Frieling P and Schügerl K, Recovery of lactic acid from aqueous
for 70 wt% of Cyanex 923 and undiluted Cyanex 923, respectively. model solutions and fermentation broths. Process Biochem
34:685–696 (1999).
6 Jenemann JA, Lactic acid, Canadian Patent 338853 (1934).
Lactic acid extraction from aqueous multi-acid mixtures 7 Kertes AS and King CJ, Extraction chemistry of fermentation product
carboxylic acids. Biotechnol Bioeng 28:269–282 (1986).
The performance of lactic acid extraction with Cyanex 923 8 Choudhury B and Swaminathan T, Lactic acid extraction with trioctyl
from aqueous solution containing various carboxylic acids was amine. Bioprocess Biosyst Eng 19:317–320 (1998).
investigated with respect to glycolic acid and formic acid. Feed 9 Schügerl K and Degener W, Gewinnung niedermolekularer
molar ratio of lactic acid, glycolic acid, and formic acid was organischer Verbindungen aus komplexen wässrigen Gemischen
durch Extraktion. Chem Ing Tech 61:796–804 (1989).
1 : 1 : 2. Feed lactic acid concentration was 5.3 g L−1 . Extraction
10 Wardell JM and King CJ, Solvent equilibria for extraction of carboxylic
temperature was 20 ◦ C. Overall lactic acid extractability decreases acids from water. J Chem Eng Data 23:144–148 (1978).
significantly through the presence of glycolic acid and formic 11 Wang Y, Li Y, Wang J, Li Z and Dai Y, Extraction equilibria of
acid in the feed. Lactic acid partition coefficients decrease from monocarboxylic acids with trialkylphosphine oxide. J Chem Eng
3.17 to 2.53 for 70 wt% Cyanex 923 and from 4.24 to 3.53 for Data 46:831–837 (2001).
12 Hano T, Matsumoto M, Ohtake T, Sasaki K, Hori F and Kawano Y,
undiluted Cyanex 923 with glycolic acid and formic acid in the Extraction equilibria of organic acids with tri-n-octylphosphine
feed. Selectivity for carboxylic acid extraction with Cyanex 923 oxide. J Chem Eng Jpn 23:734–738 (1990).
increases in the order glycolic acid < lactic acid < formic acid. 13 Yabannavar VM and Wang DIC, Bioreactor system with solvent
Partition coefficients for lactic acid, glycolic acid and formic acid extraction for organic acid production. Ann NY Acad Sci
506:523–535 (1987).
extraction from the multi-acid mixture are given in Table 3. Lactic
14 Achour D, Malmary GH, Putranto A, Nezzal G, Mourgues JF and
acid shows good selectivity compared with glycolic acid. Partition Molinier JR, Liquid–liquid equilibria of lactic acid between water
coefficients are of a factor of 2 higher for lactic acid extraction than and tris(6-methylheptyl)amine and tributyl phosphate in various
for glycolic acid extraction. Compared with formic acid, selectivity diluents. J Chem Eng Data 39:711–713 (1994).
for lactic acid is low. Even though feed formic acid concentration 15 Watson EK, Rickelton WA, Robertson AJ and Brown TJ, A liquid
phosphine oxide: solvent extraction of phenol, acetic acid and
is twofold feed lactic acid concentration, partition coefficients for ethanol. Solvent Extr Ion Exch 6:207–220 (1988).
formic acid are 3 times higher than lactic acid partition coefficients. 16 Pagel HA and McLafferty FW, Use of tributyl phosphate for extracting
organic acids from aqueous solution. Anal Chem 20:272–272 (1948).
17 Morales AF, Albet J, Kyuchoukov G, Malmary G and Molinier J,
Influence of extractant (TBP and TOA), diluents, and modifier on
CONCLUSIONS extraction equilibrium of monocarboxylic acids. J Chem Eng Data
The organophosphorus extractant Cyanex 923 was found to 48:874–886 (2003).
be an appropriate solvent for lactic acid extraction from dilute 18 Matsumoto M, Takahashi T and Fukushima K, Synergistic extraction
aqueous solution. Compared with TBP and TOPO, Cyanex 923 of lactic acid with alkylamine and tri-n-butylphosphate: effect of
amines, diluents and temperature. Sep Purif Technol 33:89–93
shows several advantages regarding hydrodynamic properties
(2003).
and partition coefficients. Liquid–liquid phase equilibria for lactic 19 Oudshoorn A, van den Berg C, Roelands CPM, Straathof AJJ and van
acid extraction were modeled and effects of acid concentration der Wielen LAM, Short-cut calculations for integrated product
and extractant content on the stoichiometric equilibrium constant recovery options in fermentative production of bio-bulk chemicals.
were investigated. It was shown that both initial lactic acid Process Biochem 45:1605–1615 (2010).
20 Alex S, Biasotto F and Aroca G, Extraction of organic and inorganic
concentration and extractant content have a significant effect compounds from aqueous solutions using hollow fibre liquid-liquid
on lactic acid distribution. The algorithm gives access to complete contactor. Desalination 241:337–341 (2009).
data interpretation. 21 Le TC, Baird MHI and Hanson C, Handbook of Solvent Extraction. John
Lactic acid extraction from multi-acid solution containing Wiley & Sons, New York (1983).
glycolic acid and formic acid shows good selectivity compared

wileyonlinelibrary.com/jctb

c 2012 Society of Chemical Industry J Chem Technol Biotechnol (2012)