G Malmary et al
between organic and aqueous phases has been studied
as regard to its dependence on the acid concentration
in the aqueous solution and also the tributylphosphate
content in the solvent. The in¯uence of solvent ratio
upon the ef®ciency of the extraction process was
mainly examined because this factor has economic
signi®cance in establishing the conditions of design.8
Over the past few years, only the citric acid extraction
process has been carried through to major produc-
tion.9a,9b
settling of the phases. For batch extraction experi-
ments, both aqueous and organic phases were mixed
in glass-stoppered separation funnels maintained at a
steady temperature of 25(Æ1)°C for 3h. This time
represents a suf®cient shaking period to obtain the
maximum percentage of solute extracted by the
solvent. After a settling time of 1.5h, the phases were
separated and analysed by High Performance Liquid
Chromatography (HPLC) which gives the concentra-
tion of solutes in each of them. With regard to the
organic phase, a stripping of carboxylic acids with
H2SO4 solution (0.5M) was necessary before intro-
ducing samples into the analysis apparatus. The
HPLC method consisted of a pump (Spectra-Physics
SP 8800), an integrator (Chromjet SP 4400), a UV
spectrophotometer (Spectra 100 UV-V is detector:
wavelength=210nm) and an organic acid column
(Bio-Rad Aminex ion exclusion HPX-87H) operated
at 41°C. The mobile phase was 0.005M H2SO4 and its
¯ow rate was 0.6cm3 min 1. The analyses were made
immediately after the batch extraction assays in order
to avoid possible oxidation by air, evaporation or a
bacteriological degradation of the samples. Three
replications of extraction experiments were necessary
to con®rm the validity of results obtained in this work.
The accuracy relative to the percentages of extracted
acids and the purities of aconitic acid in the extract
(Figs 1±6) was estimated to be 2% for all batch
experiments.
2
MATERIALS
Tributylphosphate, dodecane, aconitic and lactic acids
were provided by Aldrich Chemical Co and were
analysed by HPLC in order to verify their purity,
which was close to 99%. However, lactic acid, a
concentrated aqueous solution which contains 15
mass% of water, must be distilled using a total re¯ux
condenser for 12h in order to hydrolyse dimers
present in the solution.10,11 Deionised water produced
by a Millipore Mille-Q Water System was used in this
work. The pKa values of aconitic and lactic acids12,13
are listed in Table 2.
3
EXPERIMENTAL PROCEDURE
Based on the average concentration of acids in waste
water of sugar cane industry,14 two initial aqueous
solutions of various acid concentrations were treated
by solvent extraction. The ®rst solution contained 1.7
mass% of aconitic acid and 0.4 mass% of lactic acid
(pH=1.5) while the second was composed of 0.3
mass% of aconitic acid and 0.4 mass% of lactic
(pH=1.7). With regard to the solvent composition,
the tributylphosphate concentration should be as high
as possible since the diluent does not affect the
extraction capacity of the solvent. Taking the phase
separation into consideration three types of mixtures
`extractantdiluent' containing 50 mass%, 60
mass% and 70 mass% of tributylphosphate respec-
tively were studied. The solvent was saturated with
water because small amounts of water enhance
solvation for the acid solute. This phenomenon
appears to be associated with formation of complexes
in which water molecules serve as hydrogen-bonded
bridges between the carboxylic acid and the proton-
accepting solvent.7 With regard to the liquid±liquid
systems under consideration, the speci®c gravity and
the viscosity of solvent were compatible with good
4
RESULTS AND DISCUSSION
Some authors1,15,16 have shown that undissociated
molecules of carboxylic acids were only extracted by
the organophosphorus solvent. Consequently, the
distribution of organic acids between the solvent and
the aqueous phase essentially depends on the strength
of the acids.
Since the pH values of the initial aqueous solutions
are lower than the pKa of aconitic and lactic acids
(Table 2), these solutes were mainly in molecular form
in aqueous phases. In the case of aconitic acid, the ®rst
dissociation constant essentially determines the
strength of the acid, the contribution of others being
rather marginal. Thus, aconitic acid which is relatively
stronger than lactic acid will be more dissociated
whatever the pH values of both initial aqueous
solutions. On the other hand, since the hydrophilicity
of the acid radical is increased by hydroxyl functional
groups, the differences in partition coef®cient values
for various solutes are often signi®cant.1 Thus the
hydroxyl functional group present in lactic acid
strongly depresses its extractibility by the organic
solvent. Indeed, from dilute aqueous solutions con-
taining less than 2.0 mass% for an individual acid and
a solvent composed of the mixture `tributylphosphate
(70 vol%)dodecane (30 vol%)', the partition
coef®cients, determined from liquid±liquid equili-
brium, at the temperature of 25°C, were 12.7 for
aconitic acid and 2.9 for lactic acid. Furthermore, the
Table 2. pKa values of aconitic and lactic acids
Carboxylic acid pKa (in aqueous solution at 25°C)
Aconitic
Lactic
pKa =2.80
pKa =4.46
pKa =6.30
1
2
3
pKa =3.86
1170
J Chem Technol Biotechnol 75:1169±1173 (2000)