Recovery of Volatile Fatty acids by Reactive Extraction using

Tri-N-Octylamine and Tri-Butyl Phosphate in Different Solvents:

Equilibrium Studies, pH and Temperature Effect, and Optimization
using Multivariate Taguchi Approach

Sumalatha Eda,1,2 Alka Kumari,1 Prathap Kumar Thella,1* Bankupalli Satyavathi1

and Parthasarathy Rajarathinam2
1. Chemical Engineering Division, CSIR-IICT, Tarnaka, Hyderabad, India
2. School of Engineering, RMIT University, Melbourne, 3000, Australia

Recovery of volatile fatty acids from fermentation broth has been investigated by adopting an intensified approach using extractants tri-n-
octylamine and tri-butyl phosphate dissolved in 1-decanol and methyl isobutyl ketone. The effects on distribution coefficient (KD) and
extraction efficiency (%E) were studied by varying the operating conditions like temperature (293.15–323.15) K, pH (2.5, 3.5, and 4.5), and
compositions of extractant (10, 20, and 30 %). Taguchi (L36) orthogonal design with five factors, namely diluents, extractant type,
composition, temperature, and pH, was employed for the multivariate optimization of reactive extraction of volatile fatty acids. In the Taguchi
approach, a “larger is better” criterion was adopted to maximize K D and %E. The statistical analysis indicated that the degree of influence on
K D by experimental variables follows the following trend: extractant type (X2 ) > pH (X5 ) > diluent type (X1 ) > temperature (X4 ) > extractant
concentration (X1 ). The trend for %E observed is as follows: extractant type (X2 ) > pH (X5 ) > temperature (X4 ) > extractant concentration
(X1 ) > diluent type (X1 ). The combination of optimum parameters were obtained as X1 ¼ 1-decanol, X2 ¼ tri-n-octylamine, X3 ¼ 20 %,
X4 ¼ 293.15 K, and X5 ¼ 3.5. A confirmation run was conducted using these parameters and K D and %E values from this run were determined
to be 8.65 and of 89.64 %, respectively, which were very close to the predicted values K D ¼ 10.36 and %E ¼ 91.26 %.

Keywords: reactive extraction, volatile fatty acids, TOA, TBP, desirability approach

INTRODUCTION adulterating it. However, the concerns can be overcome by using

effective techniques like evaporation and distillation to remove the

L
iquid-liquid reactive extraction is an economical and
feasible technique to separate the carboxylic acids and it
has been used in pharmaceutical, food, and fermentation
industries to reduce the environmental contamination. There is a
need to separate value added volatile fatty acids (VFAs) from
aqueous effluent stream or fermentation broth because the use of
VFAs is widespread in the pharmaceutical, food, and biotechno-
logical industries. In the present work, the separation of volatile
fatty acids from fermentation broth was carried out using reactive
extraction, i.e. the process intensification method, using tri-butyl
phosphate (TBP) and tri-n-octylamine (TOA) as extractants and
1-decanol and methyl isobutyl ketone as diluents to facilitate the
desired recovery. VFAs such as formic acid (FA), acetic acid (AA),
propionic acid (PA), and butyric acid (BA) are the short-chain fatty
acids produced either synthetically from fossil resources or
obtained as metabolic intermediates during fermentation. Recov-
ery of VFAs from the broth is important as they are used in the
production of important chemicals including alcohols, aldehydes,
ketones, esters, and olefins. Acetic acid and propionic acid, in
particular, have widespread applications as taste-enhancing
additives and preservatives in the food industry, as buffer
solutions in pharmaceutical industry and skin lightening, or
anti-acne agents in the cosmetic industry. Butyric acid is more
often used as an animal feed supplement, fishing bait additives and
food additives, or for flavouring.[1,2] Although the reactive
extraction process is effective, it is used cautiously in the
fermentation and food industries. It is possible that the extractants
or diluents may be carried over by the final food product, thereby
traces of extractants using specialized process equipment or by
effective process control.
The major challenge in the production of VFAs by
fermentation istheir recovery from the resultant broths.
Existing conventional processes for recovery of organic acids
from fermentation broths include stripping, adsorption,
electrodialysis, direct distillation, solvent extraction, evapo-
ration chromatographic methods, ultrafiltration, reverse
osmosis, and drying, each process having its advantages
and disadvantages. Current research is therefore aimed
towards the selection of an intensified approach, such as
reactive extraction, that would overcome the weaknesses of
the existing processes. The literature cites a few examples for
the recovery of carboxylic acids by reactive extraction from
aqueous waste streams in cellulose and vinyl acetate
manufacturing units. As the process is a combination of
both chemical and physical phenomena, a higher distribution
coefficient is achieved, making the process a promising
technique for the recovery of carboxylic acids.[3,4]
* Author to whom correspondence should be addressed.
E-mail address: prathap@iict.res.in
Can. J. Chem. Eng. 95:1373–1387, 2017
© 2017 Canadian Society for Chemical Engineering
DOI 10.1002/cjce.22803
Published online 5 April 2017 in Wiley Online Library
(wileyonlinelibrary.com).

VOLUME 95, JULY 2017 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1373
 The efficiency of liquid-liquid extraction process for recover-
ing carboxylic acids from aqueous streams is strongly influenced
by the nature of the solvent used. Classical extractants (alcohols,
ketones, aliphatic, aromatic, etc.) give low distribution coef-
ficients, resulting in the poor extraction efficiency of carboxylic
acids. On the other hand, the use of aminic or organo-phosphoric
compounds as extractants has yielded higher efficiency with
better distribution coefficients compared to classical extractants.
Preferred extractants for reactive extraction of carboxylic acids
from aqueous phase include long chain aliphatic primary,
secondary, and tertiary amines. The primary amines are
characterized by a high solubility in the aqueous phase, while
the secondary amines have the highest distribution coefficient but
tend to form amides in the downstream regeneration by
distillation. The tertiary amines offer advantages over the other
extractants on the grounds of its lower cost and generally higher
equilibrium distribution coefficient KD. Acid-amine complex
stability is affected by the basicity of the amine, which can be
altered using various diluents.
The majority of the literature is focused on the reactive
extraction of single acid in the aqueous phase. The literature is
scarce for the extraction of multi-acids from aqueous streams.
Only a very few reports are available on the extraction of multi-
acids. Among them, Juang et al.[6] studied the recovery of two
acids in the aqueous phase using TOA dissolved in xylene and
the recovery of multi-acids, i.e. VFAs (AA, BA, PA, and VA)
using tri-n-octylphosphine oxide (TOPO) dissolved in kerosene.
Alkayya et al.[7] studied the recovery of VFAs using tri-
octylphosphine oxide (TOPO) in kerosene and their results
indicate the partition coefficient of VFAs is relatively low.
In the present study, the potential of the reactive extraction
technique is explored for the effective recovery of VFAs (AA, PA,
and BA) from an aqueous stream using TOA and TBP dissolved in
1-decanol and MIBK. The aim of the extraction experiment is to
assess the applicability of reactive extraction process to influence
the distribution coefficient and extraction efficiency by varying
the pH and the composition of extractants (TOA and TBP in
1-decanol and MIBK) on the recovery of VFAs from fermentation
broth at different temperatures. Taguchi L36 mixed design was
used to determine the optimum combination of factors for
multivariate response characteristics based on the desirability
approach. In the present work, the influences of diluents type, X1
(1-decanol, MIBK), extractant type, X2 (TOA, TBP), extractant
concentration, X3 (10, 20, and 30 %), temperature, X4 (293.15,
303.15, 313.15, and 323.15 K), and pH, X5 (2.5, 3.5, and 4.5) on
the reactive extraction of VFAs were studied.
MATERIALS AND METHODS
Chemicals and Reagents
Glacial acetic acid (AA) (mass fraction purity > 0.995) was
supplied by Molychem, Mumbai, India. Propionic acid (PA) and
butyric acid (BA) of mass fraction purity > 0.99 were procured
from Sigma Aldrich. Tri-n-octylamine (TOA) (mass fraction
purity > 0.98) and tri-n-butyl phosphate (TBP) (mass fraction
purity > 0.97) supplied by Sigma Aldrich were used as extractants.
1-decanol (mass fraction purity > 0.99, density 0.819 g/mL) and
methyl iso-butyl ketone (MIBK) (mass fraction purity > 0.99,
density 0.802 g/mL) supplied by S. D. Fine-Chem Ltd., Mumbai,
India, were used as diluents. Phenolphthalein indicator (pH
8.2–9.8) was supplied by S. D. Fine-Chem. Ltd., Mumbai, India.
The sodium hydroxide used in the experiments was of reagent
1374 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING
grade. Hydrochloric acid (HCl) (mass fraction purity > 0.36
supplied by Molychem, Mumbai, India) of 0.1 N was prepared by
dissolving an appropriate amount in de-mineralized water. All the
chemicals were used without further purification. De-mineralized
(DM) water was used for all the experiments.
Preparation of VFA Solution
The synthetic mixture of acids was prepared based on the
composition of the sludge, which was collected from a facility
that uses bio production to produce volatile fatty acids. The
synthetic mixture was prepared by dissolving 8.26 g/L of AA,
3.08 g/L of PA, and 2.66 g/L of BA in de-mineralized (DM)
water to mimic the real composition of fermentation broth.
The extraction of acids from the aqueous stream into the
organic stream may not be affected by the byproducts of
fermentation and biomass because the solubility acids in the
organic phase (solvent þ extractant) and affinity of acids
towards the bases (extractants) are relatively higher than
those of byproducts.
Instrumentation
The acid composition was measured using high-performance
liquid chromatography (HPLC) (Model: Shimadzu SPD-M20A)
equipped with a UV detector. The equilibrium studies were carried
out using a shaking incubator (Model No.-LSI4018R) provided by
Daihan Labtech India Pvt. Ltd., capable of maintaining the
temperature within  0.1 K. The pH of the solution was measured
using a Sartorius PB–11 pH meter. A Radleys-RR98072 magnetic
stirrer was used for stirring the VFA mixture. A high precision
Citizen balance (accuracy of  0.0001 g) was used for weighing
the accurate amount of acids.
Reactive Extraction Experiments
All reactive extraction experiments were conducted in batch
mode. The experimental procedure used is as detailed in the
following steps: (1) Preparation of aqueous solutions by
diluting acids of total concentration 14 g/L (containing AA,
PA, and BA in the concentrations of 59, 22, and 19 %,
respectively) with DM water. (2) The total concentration of the
acids in the aqueous solution was evaluated by the APHA
method. (3) The individual acid concentration was determined
by HPLC method. (4) A fixed quantity of organic phase
(containing extractants TOA or TBP mixed with diluents
1-decanol or MIBK) was added to the aqueous solution. (5)
Aqueous and organic phases were mixed in a conical flask and
kept under shaking motion at 200 rpm for 5 h, which was
determined to be a suitable equilibrium time from preliminary
evaluations. (6) The solution was then allowed to settle for 3 h
during which samples were withdrawn periodically to analyze
and ensure the attainment of equilibrium. (7) After the
extraction of acids into the organic layer, the total concentra-
tion of acid in the aqueous layer was determined using the
APHA method. (8) Recovery of individual acids was deter-
mined by measuring their concentrations in the aqueous layer
using HPLC method.
In reactive extraction experiments, the main focus was to
determine the recovery of acids by the extractants, whereas the
main focus of the equilibrium study was to determine the
distribution of acids between the aqueous and the organic phases.
Experimental results were used to determine distribution coeffi-
cient (KD ) and extraction efficiency (%E). The distribution
coefficient KD is expressed by Equation (1):
VOLUME 95, JULY 2017

CA;org
KD ¼
CA;aq
The degree of extraction (%E) is expressed by Equation (2):
KD
%E ¼  100
ð1 þ KD Þ
Equilibrium studies
The equilibrium studies were carried out by mixing 20 mL
aqueous solution containing 14 g/L of VFAs with equal volume
organic phase (containing TOA or TBP (v/v) in 1-decanol or
MIBK). The mixture was kept at a constant temperature
(chosen in the range of 293.15–333.15 K) in a shaking
incubator for 5 h at 200 rpm, and then allowed to settle for
3 h before the two separate layers were separated. The total
acid concentration (C A,aq) in the aqueous phase was deter-
mined by titration. The corresponding total VFA concentration
in the organic phase (C A, org) was determined by mass balance.
Then, Equation (1) was used to determine the KD value.
Figures 1–6 show the experimental data. These values are also
reported in Table 3. All of these experiments were repeated
Figure 1. A plot of distribution coefficient (KD), extraction efficiency (%E) (a) for 1-decanol and MIBK, (b) 10, (c) 20, and (d) 30 % composition of
extractants in 1-decanol and MIBK diluents. Temperature ¼ 303.15 K, pH‘ ¼ 3.5.
VOLUME 95, JULY 2017
twice and the standard deviation in experimental results was
observed to be 1.
ð1Þ
Optimization of Reactive Extraction Process: Taguchi L36 Mixed
Design
A Taguchi design was employed to determine the combination of
ð2Þ optimum parameters, which influence the response character-
istics the most. In the present study, a Taguchi ðL36 Þ mixed array
design consisting of mixed levels of factors as listed in Table 1
was used for the optimization of process parameters in the
reactive extraction of volatile fatty acids (VFAs) from aqueous
solution. The process parameters that would influence the multi-
response variables were identified to be the extractant type (X1 ),
diluents type (X2 ), extractant concentration (X3 ), temperature
(X4 ), and pH of the aqueous solution (X5 ). The distribution
coefficient (KD ) and extraction efficiency (%E) were selected as
the response variables. In the Taguchi method, signal (S)
denotes the desired characteristics and noise (N) represents the
undesired value. S/N ratio is the log transformation of mean
square deviation of desired performance, which measures the
response characteristics deviating from the desired value.[8–10]
An appropriate S/N ratio (larger-the-better) was employed as the
criterion opted. Thus, a Taguchi ðL36 Þ mixed design and a larger-
the-better approach was adopted in the present study so as to
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1375
 where n is the number of replicates of runs and yi is the response
Table 1. L36 factors along with their levels by Taguchi mixed design
variable. The unit of S/N ratio is decibel (dB), often used in
Levels communication engineering. The S/N ratio values calculated
using Equation (3) for individual responses is reported in Table 4.
Factors 1 2 3

Diluent type ðX1 Þ 1-decanol MIBK –

Desirability approach
Extractant type ðX2 Þ TOA TBP –
Extractant 0.229 kmol m-3 0.458 kmol m-3 0.678 kmol m-3 Desirability approach is an established and useful technique for
concentration ðX3 Þ the simultaneous optimization of factors, which verifies the
Temperature (K) (X4 ) 293.15 303.15 323.15 optimum performance of multiple responses. In this approach,
pH (X5 ) 2.5 3.5 4.5
each response is converted into a desirability function that varies
from 0 to 1 (lowest to highest). The numerical multivariate
maximize KD and %E. The selected experimental factors with optimization has been carried out to identify the combination of
their respective levels suggested that 36 runs in total are to be factors that maximizes the overall desirability of the process. In the
carried out to study the influence of these factors on response present study, the desirability function for two response character-
variables. The details of the design matrix along with their istics (KD and %E) was calculated for satisfying the larger-
experimental results are presented in Table 2. For the criteria the-better (LTB) criterion according to the following equation:
opted, the equation representing the response variable (yi ) for n
observations is shown below: If b
y  ymin di ¼ 0
" #  r
S 1X n
1 ^y  ymin
¼ 10 log10 ð3Þ If b
y  ymax di ð4Þ
N n i¼1 y2i ymax  ymin

Table 2. L36 mixed array (levels of five different factors and obtained results)

Exp. No. X1 X2 X3 X4 X5 KD S=N ratio (db) %E S=N ratio (db) D

1 1 1 1 1 1 4.28 12.63 81.06 38.18 0.62

2 1 1 2 2 2 7.33 17.78 87.99 38.95 1.00
3 1 1 3 3 3 2.35 7.42 70.14 36.92 0.36
4 1 1 1 1 1 4.28 12.63 81.06 38.18 0.62
5 1 1 2 2 2 7.33 17.78 87.99 38.95 1.00
6 1 1 3 3 3 2.35 7.42 70.14 36.92 0.36
7 1 1 1 1 2 6.06 15.65 85.84 38.67 0.81
8 1 1 2 2 3 2.44 7.75 70.90 37.01 0.37
9 1 1 3 3 1 5.06 14.08 83.50 38.43 0.71
10 1 2 1 1 3 1.00 -0.03 49.90 33.96 0.07
11 1 2 2 2 1 1.21 1.66 54.75 34.77 0.13
12 1 2 3 3 2 1.17 1.36 53.92 34.63 0.12
13 1 2 1 2 3 0.90 -0.92 47.37 33.51 0.05
14 1 2 2 3 1 1.04 0.34 50.98 34.15 0.09
15 1 2 3 1 2 1.82 5.20 64.54 36.20 0.26
16 1 2 1 2 3 0.90 -0.92 47.37 33.51 0.05
17 1 2 2 3 1 1.04 0.34 51.05 34.15 0.09
18 1 2 3 1 2 1.82 5.20 64.54 36.20 0.26
19 2 1 1 2 1 2.14 6.61 68.15 36.67 0.32
20 2 1 2 3 2 1.92 5.67 65.75 36.36 0.28
21 2 1 3 1 3 1.85 5.34 64.95 36.25 0.27
22 2 1 1 2 2 2.16 6.69 68.35 36.70 0.32
23 2 1 2 3 3 1.25 1.94 55.56 34.89 0.14
24 2 1 3 1 1 2.61 8.33 72.30 37.18 0.40
25 2 1 1 3 2 1.86 5.39 65.03 36.26 0.27
26 2 1 2 1 3 1.76 4.91 63.77 36.09 0.25
27 2 1 3 2 1 2.42 7.68 70.76 37.00 0.37
28 2 2 1 3 2 1.34 2.54 57.26 35.16 0.16
29 2 2 2 1 3 1.11 0.91 52.61 34.42 0.10
30 2 2 3 2 1 1.80 5.11 64.32 36.17 0.26
31 2 2 1 3 3 0.85 -1.41 45.95 33.24 0.00
32 2 2 2 1 1 1.57 3.92 61.12 35.72 0.21
33 2 2 3 2 2 1.82 5.20 64.60 36.20 0.26
34 2 2 1 3 1 1.32 2.41 56.93 35.11 0.15
35 2 2 2 1 2 1.63 4.24 62.00 35.85 0.22
36 2 2 3 2 3 1.14 1.14 53.35 34.54 0.11
X1 : Diluent type, X2 : Extractant type, X2 : Extractant concentration, X3 : Temperature, X3 : pH, D: Overall desirability.

1376 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING VOLUME 95, JULY 2017
Table 3. Individual distribution coefficients of AA, PA, and BA, overall distribution coefficients and extraction efficiency %E at different pH and
temperatures

pH ¼ 2.5

293.15 K 303.15 K 313.15 K 323.15 K

10 20 30 10 20 30 10 20 30 10 20 30

AA-2.99 AA-5.55 AA-6.16 AA-2.33 AA-4.02 AA-4.53 AA-1.94 AA-3.12 AA-3.52 AA-1.74 AA-2.92 AA-3.31
TOA in 1-decanol Individual PA-6.68 PA-11.96 PA-13.10 PA-4.05 PA-9.48 PA-11.29 PA-4.62 PA-6.95 PA-7.81 PA-4.82 PA-6.12 PA-8.49
KD BA-24.34 BA-46.25 BA-53.33 BA-24.18 BA-35.06 BA-45.46 BA-18.36 BA-31.04 BA-37.93 BA-18.09 BA-33.85 BA-37.72
Overall KD 4.27 8.01 8.86 3.38 5.96 6.77 3.01 4.67 5.24 2.80 4.32 5.06
Efficiency 81.06 88.91 89.85 77.19 85.63 87.13 75.07 82.37 83.98 73.70 81.34 83.50

AA-0.73 AA-0.79 AA-0.98 AA-0.73 AA-0.78 AA-0.915 AA-0.47 AA-0.56 AA-0.61 AA-0.45 AA-0.50 AA-0.57
TBP in 1-decanol Individual PA-2.74 PA-2.75 PA-3.07 PA-0.92 PA-1.69 PA-2.65 PA-1.52 PA-1.72 PA-2.03 PA-1.66 PA-1.86 PA-2.16
KD BA-10.38 BA-11.11 BA-17.83 BA-3.41 BA-3.84 BA-5.71 BA-7.10 BA-8.26 BA-9.5 BA-6.85 BA-7.65 BA-9.06
Overall KD 1.40 1.53 1.76 1.00 1.21 1.52 0.96 1.08 1.18 0.94 1.04 1.16
Efficiency 58.36 60.61 63.80 50.17 54.90 60.34 49.01 52.12 54.32 48.66 51.04 53.72

AA-1.28 AA-1.47 AA-1.51 AA-1.24 AA-1.34 AA-1.38 AA-1.05 AA-1.12 AA-1.15 AA-1.03 AA-1.09 AA-1.12
TOA in MIBK Individual PA-3.68 PA-4.20 PA-4.62 PA-3.44 PA-3.78 PA-4.21 PA-3.07 PA-3.11 PA-3.12 PA-3.34 PA-3.45 PA-3.37
KD BA-50.62 BA-54.20 BA-57.08 BA-28.04 BA-31.11 BA-57.08 BA-14.75 BA-13.52 BA-13.79 BA-14.31 BA-13.01 BA-12.8-
Overall KD 2.23 2.51 2.60 2.13 2.29 2.41 1.83 1.90 1.94 1.82 1.89 1.92
Efficiency 69.07 71.58 72.29 68.07 69.60 70.70 64.66 65.59 66.0 64.64 65.48 65.82

AA-0.76 AA-0.86 AA-1.08 AA-0.76 AA-0.858 AA-1.01 AA-0.73 AA-0.85 AA-1.01 AA-0.69 AA-0.81 AA-0.96
TBP in MIBK Individual PA-2.70 PA-2.81 PA-3.47 PA-2.49 PA-2.82 PA-3.33 PA-2.20 PA-2.54 PA-2.81 PA-2.37 PA-2.71 PA-3.23
KD BA-10.17 BA-11.23 BA-14.06 BA-10.35 BA-11.28 BA-12.90 BA-10.38 BA-11.28 BA-13.08 BA-9.8 BA-10.06 BA-12.6
Overall KD 1.44 1.57 1.90 1.41 1.56 1.80 1.34 1.52 1.74 1.32 1.48 1.73
Efficiency 59.02 61.11 65.56 58.54 60.96 64.32 57.35 60.44 63.59 56.93 59.77 63.45

pH ¼ 3.5

293.15 K 303.15 K 313.15 K 323.15 K

10 20 30 10 20 30 10 20 30 10 20 30

AA-4.38 AA-6.01 AA-7.19 AA-3.28 AA-5.01 AA-5.60 AA-2.43 AA-3.71 AA-4.15 AA-1.838 AA-2.83 AA-3.14
TOA in 1-decanol Individual PA-7.57 PA-13.12 PA-16.01 PA-7.72 PA-11.35 PA-13.19 PA-5.68 PA-8.77 PA-10.44 PA-4.58 PA-6.49 PA-8.47
KD BA-28.55 BA-47.36 BA-60.86 BA-33.10 BA-43.3 BA-55.59 BA-2.29 BA-38.11 BA-45.66 BA-21.73 BA-30.66 BA-37.55
Overall KD 6.06 8.65 10.36 4.93 7.32 8.24 3.72 5.55 6.26 2.91 4.28 4.862
Efficiency 85.83 89.64 91.2 83.16 87.99 89.17 78.84 84.74 86.2 74.44 81.08 82.95

AA-0.82 AA-0.88 AA-0.91 AA-0.80 AA-0.92 AA-0.94 AA-0.506 AA-0.53 AA-0.58 AA-0.49 AA-0.53 AA-0.585
TBP in 1-decanol Individual PA-2.93 PA-3.26 PA-5.28 PA-3.18 PA-3.19 PA-3.52 PA-1.98 PA-2.04 PA-2.40 PA-1.78 PA-1.86 PA-2.083
KD BA-5.63 BA-8.81 BA-15.62 BA-4.28 BA-5.22 BA-12.03 BA-7.61 BA-8.01 BA-9 BA-7.44 BA-8.81 BA-9.35
Overall KD 1.45 1.60 1.82 1.41 1.56 1.72 1.05 1.08 1.19 1.02 1.07 1.16
Efficiency 59.25 61.6 64.57 58.50 60.96 63.33 51.2 52.1 54.35 50.50 51.87 53.81

AA-1.54 AA-1.60 AA-1.61 AA-1.22 AA-1.30 AA-1.40 AA-1.06 AA-1.12 AA-1.15 AA-1.03 AA-1.18 AA-1.22
TOA in MIBK Individual PA-4.31 PA-4.61 PA-4.76 PA-4.62 PA-5.474 PA-.23 PA-3.24 PA-3.51 PA-3.59 PA-3.93 PA-2.81 PA-2.87
KD BA-16.97 BA-18.13 BA-20.11 BA-15.73 BA-17.22 BA-18.41 BA-13.37 BA-13.86 BA-14.2 BA-11.66 BA-11.78 BA-1.91
Overall KD 2.52 2.62 2.66 2.16 2.32 2.50 1.84 1.95 1.99 1.86 1.92 1.96
Efficiency 71.6 72.42 72.73 68.36 69.92 71.44 64.87 66.10 66.62 65.03 65.79 66.32

AA-0.79 AA-0.89 AA-1.09 AA-0.79 AA-0.83 AA-1.01 AA-0.71 AA-0.83 AA-0.95 AA-0.70 AA-0.81 AA-0.95
TBP in MIBK Individual PA-2.68 PA-3.13 PA-3.66 PA-2.54 PA-3.26 PA-3.55 PA-2.38 PA-2.80 PA-3.49 PA-2.38 PA-2.80 PA-3.49
KD BA-10.71 BA-11.97 BA-14.11 BA-8.5 BA-11.03 BA-14.11 BA-10.56 BA-11.66 BA-3.1 BA-10.56 BA-11.66 BA-13.14
Overall KD 1.46 1.63 1.93 1.43 1.57 1.82 1.34 1.53 1.75 1.34 1.51 1.74
Efficiency 59.50 62 65.91 58.86 61.1 64.60 57.37 60.46 63.63 57.28 60.14 63.57

pH ¼ 4.5

293.15 K 303.15 K 313.15 K 323.15 K

10 20 30 10 20 30 10 20 30 10 20 30

Individual AA-1.55 AA-2.18 AA-2.28 AA-1.32 AA-1.52 AA-1.67 AA-1.32 AA-1.52 AA-1.58 AA-1.03 AA-1.40 AA-1.43
TOA in 1-decanol KD PA-4.01 PA-4.43 PA-4.61 PA-3.49 PA-3.99 PA-4.11 PA-2.95 PA-3.57 PA-4.03 PA-2.69 PA-3.76 PA-3.91

(Continued.)

VOLUME 95, JULY 2017 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1377
 pH ¼ 4.5

293.15 K 303.15 K 313.15 K 323.15 K

10 20 30 10 20 30 10 20 30 10 20 30

BA-16 BA-20.53 BA-23.8 BA-14.10 BA-15.15 BA-16.80 BA-14.92 BA-17.38 BA-20.53 BA-11.35 BA-14.38 BA-17.08
Overall KD 2.48 3.24 3.39 2.19 2.43 2.62 2.10 2.41 2.54 1.72 2.28 2.34
Efficiency 71.3 76.46 77.26 68.70 70.89 72.41 67.75 70.68 71.78 63.34 69.51 70.13

AA-0.575 AA-0.59 AA-0.71 AA-0.36 AA-0.37 AA-0.40 AA-0.414 AA-0.43 AA-0.45 AA-0.32 AA-0.35 AA-0.359
TBP in 1-decanol Individual PA-1.43 PA-1.45 PA-1.54 PA-2.56 PA-2.81 PA-2.90 PA-1.05 PA-1.25 PA-1.30 PA-1.20 PA-1.21 PA-1.41
KD BA-4.50 BA-5.09 BA-6.27 BA-5.33 BA-6.27 BA-7.53 BA-5.85 BA-6.56 BA-7.04 BA-5.91 BA-6.06 BA-6.69
Overall KD 0.99 1.03 1.18 0.90 0.94 0.99 0.808 0.86 0.89 0.739 0.77 0.81
Efficiency 49.9 50.79 54.16 47.37 48.46 49.79 44.69 46.34 47.32 42.51 43.5 44.81

AA-1.02 AA-1.06 AA-1.16 AA-0.703 AA-0.74 AA-0.76 AA-0.68 AA-0.70 AA-0.72 AA-0.68 AA-0.70 AA-0.72
TOA in MIBK Individual PA-2.50 PA-2.57 PA-2.37 PA-2.37 PA-2.43 PA-2.65 PA-1.99 PA-2.01 PA-2.04 PA-1.79 PA-1.88 PA-1.92
KD BA-10.36 BA-12.2 BA-14.59 BA-8.0 BA-9.37 BA-9.87 BA-8.42 BA-8.96 BA-9.18 BA-8.42 BA-8.70 BA-8.83
Overall KD 1.68 1.75 1.85 1.30 1.37 1.41 1.24 1.27 1.29 1.22 1.255 1.27
Efficiency 62.7 63.74 64.95 56.53 57.80 58.6 55.35 56.11 56.51 54.99 55.67 56.08

AA-0.52 AA-0.59 AA-0.62 AA-0.50 AA-0.55 AA-0.59 AA-0.46 AA-0.49 AA-0.54 AA-0.47 AA-0.52 AA-0.55
TBP in MIBK individual PA-1.49 PA-1.70 PA-1.87 PA-1.37 PA-1.73 PA-1.90 PA-1.22 PA-1.86 PA-1.90 PA-0.99 PA-1.09 PA-1.57
KD BA-6.98 BA-7.83 BA-11.42 BA-6.22 BA-7.79 BA-8.70 BA-5.28 BA-8.19 BA-8.62 BA-5.2 BA-5.63 BA-8.70
Overall KD 1.0 1.11 1.18 0.95 1.06 1.14 0.88 1.02 1.08 0.85 0.91 1.45
Efficiency 50.02 52.64 54.28 48.76 51.62 53.35 46.77 50.60 52.03 45.98 47.90 51.11

systematically the relative importance of factors, which indicate

If b
y  ymax di ¼ 1
where ^y indicates the response characteristic, ymin and ymax are the
lower and upper tolerance limits of response characteristics,
respectively, di is the desirability of ith response, and r is the
desirability index. In a multi-response condition, the ideal case is
achieved if the desirability of all responses equals one, and the case
is unacceptable if the desirability of response reaches 0. Therefore,
the overall desirability is computed as the geometrical mean of all
the individual desirability (di ) according to Equation (5):
1
D ¼ ðd1 d2 . . . . . . :dn Þn
where D is the overall desirability and n is the total number of
responses. MINITAB 17 was used for the design of experiments,
analysis of variance analysis (ANOVA), and optimization of the
extraction process. The ANOVA was performed to study
the contribution of each factor on responses relative to the total
contribution.
Analytical Method
Individual carboxylic acid (AA, PA, and BA) concentration was
analyzed using a HPLC (Model: Shimadzu SPD-M20A) method
equipped with C18 reverse phase column with dimensions of
250  4.6 mm and specifications including particle size of
5 mm, flow rate of 0.8 mL/min, and wavelength of 210 nm. The
mobile phase consisted of 75/25 acetonitrile/water and
0.1 N H2SO4 at a pH of 2.5. A sample injection volume of
ð5Þ 20 mL was used. The overall acid concentration in the aqueous
phase was analyzed using the APHA method.[11] Aqueous
sample (5 mL) was diluted to 100 mL by adding DM water and
the mixture was titrated with 0.1 N HCl up to the end point
with pH 3.0 and then titrated with 0.1 N NaOH up to the end
point with pH 6.5.

Table 4. Response table for signal-to-noise ratios (S/N; larger is better)

Factors

X1 X2 X3 X4 X5

Levels KD %E KD %E KD %E KD %E KD %E

1 6.83 36.28 8.52 37.04 4.86 35.70 6.11 36.25 6.27 36.34
2 4.25 35.77 2.04 34.88 4.91 35.82 5.77 36.10 6.97 36.50
3 6.09 36.35 3.97 35.52 2.61 35.04
Dmaxmin 2.57 0.47 6.47 2.16 1.22 0.66 2.13 0.74 4.36 1.46
Rank (K D ) 3 1 5 4 2
Rank (%E) 5 1 4 3 2
Dmaxmin : Range is the difference between maximum and minimum levels of factors.

1378 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING VOLUME 95, JULY 2017
Figure 2. Distribution coefficient (KD) and extraction efficiency (%E) for different organic phases. (a) TOA þ 1-decanol, (b) TBP þ 1-decanol,
(c) TOA þ MIBK, and (d) TBP þ MIBK.
RESULTS AND DISCUSSION
The reactive extraction data for a system involving aqueous
solution of VFAs (14 g/L) and organic phase involving extractants
(TOA or TBP) and diluents (1-decanol or MIBK) are presented in
Table 3 for pH ranging from 2.5 to 4.5 and temperature ranging
from 293.15 to 323.15 K. The results are presented in terms of
distribution coefficient (KD ) and extraction efficiency (%E). In the
following sections, the influences of different parameters are
discussed and the optimization of the reactive extraction process is
explained.
Effects of Diluent and Extractant
The present study was carried out with two types of diluents and
extractants in the extraction of VFAs (AA, PA, and BA). The
diluents, 1-decanol and methyl isobutyl ketone (MIBK), belong to
alcohol and ketone groups, respectively. The extractants investi-
gated in the present study are TOA and TBP. Distribution
coefficient (KD ) and extraction efficiency (%E) for both the
diluents without extractant and with different concentrations of
extractants are shown in Figure 1.
VOLUME 95, JULY 2017
Tertiary amines (tri-n-octylamine) and organo-phosphorous
(tri-butyl phosphate) compounds are effective extractants and
provide a higher distribution coefficient and extraction
efficiency than carbon-bonded oxygen-bearing extractants.
The chemical stability of the chosen extractants plays an
important role in determining the efficiency of the extracting
solution, enabling good separation with solutions containing
chemically similar elements. TOA and TBP compounds contain
amine and phosphoryl groups respectively, which are stronger
Lewis bases than other extractants like the carbonyl group,
which leads to a higher extraction efficiency. TOA has a higher
viscosity (8.39 mPa  s) than TBP (3.39 mPa  s), but when it is
used with diluents, it leads to good separation. TOA in 1-
decanol shows the highest value of distribution coefficient
(KD ¼ 8.24) with an extraction efficiency (%E) of 89.17 %. The
KD and %E values for TBP in 1-decanol are found to be 1.72 and
63.25 %, respectively. From Figures 1b, 1c, and 1d, it can be
observed that the capability of TBP in extracting volatile fatty
acids is lower than that of TOA. It is also clear that 1-decanol
has a greater influence on the extraction efficiency than MIBK.
Among the four organic systems used in this work to extract
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1379
Figure 3. Distribution coefficient ðK D Þ and extraction efficiency and (%E) for individual acids (AA, PA, and BA) at different temperatures. (a) TOA þ 1-decanol,
(b) TBP þ 1-decanol, (c) TOA þ MIBK, and (d) TBP þ MIBK.
VFAs, the combination of TOA in 1-decanol has the highest
extraction efficiency followed by TOA, MIBK, TBP in MIBK,
and then TBP in 1-decanol (Figures 1b, 1c, and 1d). This trend
can be attributed to the higher solubilities of VFAs in 1-decanol
compared to those in MIBK. Although the 1-decanol-TOA
mixture has a higher viscosity compared to other systems, the
effect of viscosity on mass transfer would have been overcome
by the turbulent flow conditions produced during the operation
of shaking incubator at 200 rpm which could lead to an
increase in the convective mass transfer coefficient and the
generation of fine droplets which increase the interfacial area
for mass transfer.
Effect of Extractant Concentration
The reactive extraction experiments were carried out with 10,
20, and 30 % extractants (TOA and TBP) dissolved in
1-decanol and MIBK. The concentrations of extractants in
the extractant-diluent mixture were restricted to 30 % due to
high viscosities of TOA and TBP. Higher concentration of
extractant in the organic phase was found to result in
the formation of a three phase mixture during the
extraction process. The KD and %E values obtained for
the extraction of total acids and individual acid using the four
1380 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING
extractant-diluent systems are shown in Figure 2. For all
four systems, KD and %E values are higher for butyric acid
followed by propionic acid and then acetic acid. Also, the KD
and %E values increase with increasing TOA and TBP
concentrations in the organic phase regardless of the diluent
used. Among the four systems studied, TOA þ 1-decanol have
the highest KD and %E values followed by TOA þ MIBK,
TBP þ MIBK, and TBP þ 1-decanol.
The extraction efficiency usually depends on the following
factors:
1. The association ability between the acid and extractant
(pKb – pKa.)
2. The hydrophobicity of the solute (logP)
3. The steric effect between the solute and the extractant[12]
The association ability and hydrophobicity values for the three
acids are as follows: pKa ¼ 4.75 and log P ¼ 0.33 for acetic acid,
pKa ¼ 4.67 and log P ¼ 0.33 for propionic acid, and pKa ¼ 4.82
and log P ¼ 0.79 for butyric acid. On analyzing these values along
with the data shown in Table 3, it can be stated that the higher KD
and %E observed for butyric acid are due to its hydrophobic
nature.
VOLUME 95, JULY 2017
Figure 4. Overall extraction efficiency (%E) and overall distribution coefficient (K D ) at different temperatures. (a) TOA þ 1-decanol, (b) TBP þ1-decanol, (c)
TOA þ MIBK, and (d) TBP þ MIBK.
Effect of Temperature
In the large-scale production of carboxylic acids, fermenters
usually operate in the temperature range of 305.15 to 313.15 K.
Therefore, an extractant will be suitable only if it is effective in this
temperature range. Reactive extraction is an exothermic process
and therefore a decrease in the extractability is expected with
increasing temperature because the extractability is a function of
the properties of extractants and diluents, which are expected to
change with changing temperature. Therefore, it is important to
study the effect of temperature on the reactive extraction. Many
researchers have studied the influence of temperature on %E and
reported that the extractability decreases with increasing
temperature.[13–15]
In the present study, the effect of temperature on the extraction
of VFAs was studied by varying the temperature from 293.15 to
323.15 K. Figure 3 shows the experimental results of KD and %E
for individual acids (AA, PA, and BA) as a function of temperature
for the extractant concentration of 30 %. Figure 4 presents KD and
%E values for the mixture of VFAs as a function of temperature for
the extractant concentration of 30 %. It can be observed that the
VOLUME 95, JULY 2017
distribution of both total and individual acids between the organic
and aqueous phases decreases with increasing temperature. The
KD values for all three acids exhibit nearly a flat trend with
increasing temperature for TBP in MIBK system and non-linear
trends for other systems. Effect of temperature on the extraction of
VFAs was studied also for 10 and 20 % of extractants (TOA and
TBP) in the solvents (1-decanol and MIBK). The trends in the
results of these experiments are similar to those observed for 30 %
extractant concentration. The changes in KD and %E with
changing temperature can be attributed to the effect of temper-
atures on different parameters such as pKa, the amine-acid
interaction, the solubility of the acid in the extractant, the
extractant basicity, and water co-extraction. Especially, the
solubility of acids in both the aqueous and extractant phases is
affected by temperature. It is well known that the solubility of
acids in water increases as the temperature increases. Similarly,
the solubility and the distribution coefficient (KD ) of acetic and
propionic acids is significantly affected by the temperature during
the reactive extraction. On the other hand, only slight variation in
solubility (or KD ) was observed for butyric acid as temperature
THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1381
Figure 5. Extraction efficiency (%E) and distribution coefficient (KD) for individual acids (AA, PA, and BA) at different pH. (a) TOA þ 1-decanol,
(b) TBP þ 1-decanol, (c) TOA þ MIBK, and (d) TBP þ MIBK.
changes from 303.15 to 323.15 K for all the systems investigated.
Therefore, it can be concluded that the extraction of VFAs can be
carried out over a range of temperatures. The results of the overall
extraction of acid mixtures at different temperatures shown in
Figures 4a to 4d indicate that both KD and %E decrease with
increasing temperature.
Effect of pH
In the acid-producing sludge fermentation process, the accumula-
tion of free volatile fatty acids inhibits the action of acid-producing
bacteria. The pH adjustment is an important step to control the
equilibrium concentration between the aqueous and organic
phases. The extraction of carboxylic acids depends on the pH value
in the aqueous phase. Carboxylic acids are extracted more
effectively at low pH values, i.e. lower than the pKa value of the
respective acid where the acids are available in their un-
dissociated forms. In the mixtures used in the present study,
amines and phosphates extract carboxylic acids from an aqueous
solution forming an acid-extractant complex with un-dissociated
acid. The effect of pH on the extraction of organic acids was
demonstrated by Chen et al.,[16] who studied the production of
organic acid during sludge digestion under different pH conditions
through the addition of methanogen inhibitor. They found that
there was a good ecological niche for acid-producing bacteria
1382 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING
when the pH is 6 and the accumulation of fermentation products
improves the inhibition of acidogenic bacteria.
The effect of pH variation on the extractability of acetic,
propionic, and butyric acids in mixtures containing 30 %
extractant is shown in Figure 5. For all four systems used in
this work, the KD and %E values decrease as pH increases. Also,
the variation in KD and %E for acetic and propionic acids is
slight as pH increases from 2.5 to 4.5. On the other hand, KD
values of butyric acid decrease dramatically especially when pH
varies from 3.5 to 4.5. Figure 6 illustrates the effect of pH on the
overall extractability of VFAs. The maximum extraction is
achieved at a pH of 3.5 for all four systems studied. Furthermore,
it is clear that KD and %E values increase with decreasing pH
except at extremely high or low pHs, where they do not change
significantly. A decrease in pH means an increase in the
concentration of un-dissociated acids. Therefore, it can be
concluded that the extractability is influenced by the concentra-
tion of un-dissociated acids.
Optimization of Reactive Extraction Process: Taguchi L36 Mixed
Design
In the Taguchi multivariate optimization method, KD and %E
were chosen as the two response variables. Figures 7a and 7b
present the effects of experimental variable parameters on
VOLUME 95, JULY 2017
Figure 6. Overall extraction efficiency and overall distribution coefficient at different pH. (a) TOA þ 1-decanol, (b) TBP þ 1-decanol, (c) TOA þ MIBK, and
(d) TBP þ MIBK.

Figure 7. Main effect plots (S/N ratios) for (a) distribution coefficient, K D , (b) % extraction efficiency, %E.

VOLUME 95, JULY 2017 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1383
response variables. Experimental data were transformed into
S/N ratio (using larger is better criterion) and analyzed. It can be
inferred that 1-decanol and TOA had the largest influence on both
KD and %E. Also, there is negligible effect on the response
characteristics as the extractant concentration increases from 10 to
20 %. However, both KD and %E increase when the extractant
concentration increases from 20 to 30 %. Both KD and %E
decrease as the temperature increases from 293.15 to 323.15 K.
The largest changes in response are achieved for changes in the pH
value especially at pH ¼ 3.5. From the ranks obtained for each
factor, it is evident that the extractant type is the factor which has
the most influence on both KD and %E whereas extractant
composition and diluents type have the least influence (Table 4).
For each response, optimum conditions were obtained as 1st factor
at level 1, 2nd factor at level 1, 3rd factor at level 3, 4th factor at level
1, and 5th factor at level 2.
Analysis of variance test was performed to detect the relative
significance of each factor to response. Fischer ratio (F-test)
values were calculated from the experimental data and
compared with the critical values shown in Table 5. On the
basis of the calculated F-values, it can be concluded that the
extractant type and pH have statistical significance (i.e. a
significant effect) on both KD and %E at 95 % level of
confidence. It can be observed from Table 5 that extractant
type (X2 ) has the largest sum of squares indicating that it is the
most effective factor for KD followed by pH (X5 ), diluent type
(X1 ), temperature (X4 ), and extractant concentration (X3 ).
However, in the case of %E, the most effective factor is the
extractant type (X2 ) followed by pH (X5 ), temperature (X4 ),
extractant concentration (X3 ), and diluent type (X1 ). The p-
values in the ANOVA table were used to check the significance
of each coefficient. They indicate that all coefficients
(X1 ; X2 ; X4 ; and X5 ), except the extractant concentration (X3 ),
were highly significant for KD and %E with very low p-values
(p < 0.05). It can also be observed from the ANOVA table that
there are possible interactions between the factors. The
interaction of diluents type on extractant type is highly
significant statistically with very low p-values. However, no
interaction was observed between the extraction concentration
and temperature, and between the temperature and pH. For
distribution coefficients, all interactions are statistically signif-
icant except for X2 X5 and X3 X5 . In the case of %E, the
interaction between X1 X2 and X2 X3 is statistically significant
with a value of p < 0.05. F-critical values were also calculated
and included in Table 5, for factors of statistical significance
such that F > Fcr . In the case of distribution coefficient, all the

Table 5. ANOVA for distribution coefficient (KD) and extraction efficiency (%E) in L36 array design

Factors SSa DOFb MSc Fd Fcre p-valuef PCg

For distribution coefficient ðK D Þ

X1 49.207 1 36.451 207.64 161.4 0.000 7.65

X2 305.414 1 138.115 786.78 18.51 0.000 47.49
X3 9.284 2 1.536 8.75 5.14 0.035 1.44
X4 26.173 2 6.651 37.89 4.46 0.003 4.07
X5 110.058 2 37.309 212.54 4.10 0.000 17.12
X1 X2 116.265 1 80.58 459.03 18.51 0.000 18.08
X1 X3 0.341 2 1.585 9.03 5.14 0.033 0.05
X1 X4 0.877 2 3.053 17.39 4.46 0.011 0.14
X1 X5 8.119 2 6.984 39.78 4.10 0.002 1.26
X2 X3 6.588 2 4.797 27.33 3.89 0.005 1.02
X2 X4 10.424 2 3.684 20.99 3.63 0.008 1.62
X2 X5 0.081 2 0.055 0.31 3.49 0.747 0.01
X3 X5 0.217 4 0.054 0.31 2.53 0.859 0.03
Residual 0.702 4 0.176
Total 643.751 29

For extraction efficiency (%E)

X1 1.870 1 1.603 53.35 161.4 0.002 2.93

X2 33.849 1 13.965 464.87 18.51 0.000 53.09
X3 2.205 2 0.022 0.72 5.14 0.539 3.46
X4 2.963 2 1.092 36.34 4.46 0.003 4.65
X5 12.680 2 3.020 100.52 4.10 0.000 19.89
X1 X2 9.187 1 5.962 198.47 18.51 0.000 14.41
X1 X3 0.058 2 0.058 1.92 5.14 0.261 0.09
X1 X4 0.110 2 0.101 3.37 4.46 0.139 0.17
X1 X5 0.024 2 0.074 2.47 4.10 0.200 0.04
X2 X3 0.222 2 0.245 8.16 3.89 0.039 0.35
X2 X4 0.372 2 0.173 5.75 3.63 0.067 0.58
X2 X5 0.133 2 0.037 1.22 3.49 0.387 0.21
X3 X5 0.083 4 0.021 0.69 2.53 0.634 0.13
Residual Error 0.120 4 0.030
Total 63.878 29
a
sum of squares; bdegrees of freedom; cmean square; dFischer’s ratio; eF0.05(n1, n2) where n1 is the degree of freedom and n2 ¼ (a n –a), a is factor number and n
is number of levels; fprobability of significance; gpercentage contribution.

1384 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING VOLUME 95, JULY 2017
Figure 8. Percentage contribution of factors for (a) distribution coefficient ðK D Þ, (b) % extraction efficiency (%E).

Figure 9. 3-D surfaces showing effect of combination of factors {(X3, X4), (X4, X5), and (X3, X5)} on responses. (a–c) distribution coefficient (KD), (d–f)
extraction efficiency (%E).

VOLUME 95, JULY 2017 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1385
 extraction of VFAs was investigated to determine a suitable pH for
Table 6. Multivariate optimization of KD and %E and confirmation of
the optimum condition
the recovery of VFAs from aqueous streams or fermentation
broths. The highest extraction efficiency and distribution coeffi-
Confirmation run cient were achieved at a pH of 3.5. The effect of temperature is an
(Experimental Predicted important variable in the process design; therefore the effect of
Multiple optimization values) values temperature was studied on the separation of VFAs. The
K D and %E simultaneously optimized
maximum extraction efficiency and distribution coefficient values
Diluent type (X1) 1-decanol 1-decanol were obtained at 303.15 K, which is the operating temperature of
Extractant type (X2) TOA TOA fermentation broths. The extraction efficiency of organic phases
Extractant concentration (X3) 0.458 kmol  m-3 0.458 kmol  m-3 TBP þ 1-decanol, TOA þ 1-decanol, and TBP þ MIBK was tested
Temperature (X4) 293.15 K 293.15 K in batch reactive extraction process using process parameters
pH (X5) 3.5 3.5 that were optimized using multivariate response Taguchi (L36)
Distribution coefficient (K D ) 8.65 10.36 mixed design adapting desirability approach. The optimum
Extraction efficiency (%E) 89.64 91.26 condition obtained were X1 ¼ 1-decanol, X2 ¼ TOA, X3 ¼ 20 %,
X4 ¼ 293.15 K, and X5 ¼ 3.5 with a maximum distribution
coefficient of 8.65 and an extraction efficiency of 89.64 %. The
factors are found to be significant. Figures 8a and 8b show the reactive extraction process was found to be a promising technique
percentage contribution of each factor to the responses. The for the effective removal of the VFAs (AA, PA, and BA) from
percentage contribution on both KD and %E is maximum fermentation broth.
from the extractant type (X2) (47.49 and 53.09 %) followed
by pH (17.12 and 19.89 %) and the interaction of X1 X2 (18.08
NOMENCLATURE
and 14.41 %), respectively. 3-D surface plots shown in
Figure 9 depict the effects of combination of factors {(X3 , X4 ), CA concentration of acid (kmol  m3)
(X5 , X6 ), and (X3 , X5 )} on the responses KD and %E. KD distribution coefficient
Individual desirability (di ) of responses KD and %E were %E degree of extraction
calculated and combined to evaluate the overall desirability ^y response characteristics
(Table 2). The ymax and ymin are the maximum and minimum d desirability of response
values of %E and KD obtained in L36 experiments; L31 D overall desirability
combination has the ymin values (0.85 for KD and 45.95 % S
signal to noise ratio
N
for %E), and L1 and L5 have the ymax values (7.33 for KD and
87.99 % for %E). The desirability index, r is taken as 1 as it
approaches the target value. From Table 2, it can be seen that Subscripts
the combination 2, 2, 1, 3, and 3 (Experiment 31) shows the
i ith desirability
overall desirability value as zero, which means that the S/N
aq aqueous phase
ratio cannot be calculated for this case. Therefore, a mean of
org organic phase
overall desirability has been determined to find out the
min lower tolerance
optimum combination.[17,18] It can also be noticed that the
max upper tolerance
combination 1, 1, 2, 2, and 2 (Experiments 2 and 5) show an
overall desirability value of 1, which signifies that the
maximum target responses have been achieved. From Table 4 Superscripts
and Figure 6, the optimum combination of parameters is 
equilibrium
predicted as: X1 ¼ 1-decanol, X2 ¼ TOA, X3 ¼ 20 %,
r desirability index
X4 ¼ 293.15 K, and X5 ¼ 3.5 and the responses for KD and
%E at the optimum combination are 10.36 and 91.26 %,
respectively, which is different from the overall desirability
value one combination. A confirmation run was carried out to REFERENCES
verify the validity of the optimum condition and the values for
KD and %E in the confirmation run were found to be 8.65 and [1] M. P. Zacharof, R. W. Lovitt, Waste and Biomass Valorization
89.64 %, respectively. These values are close to the predicted 2013, 4, 557.
one reported in Table 6 (i.e. 10.36 and 91.26 %, respectively), [2] M. S. Solichien, D. O’Brien, E. G. Hammond, C. E. Glatz,
which suggests that the model adequately represents the Enzyme Microb. Tech. 1995, 17, 23.
investigated system. [3] J. M. Wardell, C. J. King, J. Chem. Eng. Data 1978, 23, 144.
Therefore, from the statistical analysis of experimental data, it is [4] R. Wennersten, J. Chem. Technol. Biot. 1983, 33, 85.
concluded that the maximum extraction efficiency of VFAs can be
[5] N. L. Ricker, E. F. Pittman, C. J. King, Journal of Separation
achieved using 20 % of TOA as an extractant diluted in 1-decanol
Process Technology 1980, 1, 23.
at a temperature of 293.15 K and a pH of 3.5.
[6] R. S. Juang, R. H. Huang, R. T. Wu, J. Membrane Sci. 1997,
136, 89.
CONCLUSIONS [7] E. Alkaya, S. Kaptan, L. Ozkan, S. Uludag-Demirer, G. N.
The present study revealed that a TOA þ 1-decanol system was Demirer, Chemosphere 2009, 77, 1137.
effective in the separation of volatile fatty acids (AA, PA, and BA) [8] S. Varala, A. Kumari, B. Dharanija, S. K. Bhargava, R.
compared to TBP þ 1-decanol, TOA þ MIBK, and TBP þ MIBK in Parthasarathy, B. Satyavathi, Journal of Environmental
the reactive extraction process. The effect of pH on the reactive Chemical Engineering 2016, 4, 405.

1386 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING VOLUME 95, JULY 2017
 [9] M. Radhakumari, A. Ball, S. K. Bhargava, B. Satyavathi,
Bioresource Technol. 2014, 166, 534.
[10] N. Daneshvar, A. R. Khataee, M. H. Rasoulifard, M.
Pourhassan, J. Hazard. Mater. 2007, 143, 214.
[11] APHA, Standard methods for examination of water and waste
water, 20th edition, American Public Health Association,
Washington 1998.
[12] Z. Li, Q. Wei, Y. Dai, J. Chem. Eng. Data 2002, 47, 843.
[13] A. Keshav, K. L. Wasewar, S. Chand, Desalin. Water Treat.
2009, 3, 91.
[14] A. Keshav, K. L. Wasewar, S. Chand, Ind. Eng. Chem. Res.
2009, 48, 888.
[15] K. L. Wasewar, A. Keshav, International Journal of Research
and Reviews in Applied Sciences 2010, 3, 290.
[16] Y. Y. Chen, The study on the acetate production by
fermentation from municipal sewage sludge, Master’s thesis,
Jiangnan University, Wuxi 2007.
[17] G. Nandi, S. Datta, A. Bandyopadhyay, P. K. Pal, “Analyses of
hybrid Taguchi methods for optimization of submerged arc
weld,” National Seminar on Joining Processes: Challenges for
Quality, Design and Development, Mechanical and Production
Engineering Department, National Institute of Technology,
Agartala, 5–6 March 2010.
[18] K. Akhtar, M. K. Akhtar, A. M. Khalid, J. Hazard. Mater.
2008, 156, 108.

Manuscript received August 7, 2016; revised manuscript received

November 7, 2016; accepted for publication December 2, 2016.

VOLUME 95, JULY 2017 THE CANADIAN JOURNAL OF CHEMICAL ENGINEERING 1387