Journal of Membrane Science 296 (2007) 51–57

Recovery and purification of surfactin from fermentation

broth by a two-step ultrafiltration process
Mohd Hafez Mohd Isa, Diego Esteban Coraglia, Richard A. Frazier, Paula Jauregi ∗
Department of Food Biosciences, University of Reading, Whiteknights, PO Box 226, Reading RG6 6AP, UK
Received 30 October 2006; received in revised form 26 February 2007; accepted 12 March 2007
Available online 15 March 2007

Abstract
Surfactin is a bacterial lipopeptide produced by Bacillus subtilis and it is a powerful surfactant, having also antiviral, antibacterial and antitumor
properties. The recovery and purification of surfactin from complex fermentation broths is a major obstacle to its commercialization; therefore, two-
step membrane filtration processes were evaluated using centrifugal and stirred cell devices while the mechanisms of separation were investigated
by particle size and surface charge measurements. In a first step of ultrafiltration (UF-1), surfactin was retained effectively by membranes at above
its critical micelle concentration (CMC); subsequently in UF-2, the retentate micelles were disrupted by addition of 50% (v/v) methanol solution
to allow recovery of surfactin in the permeate. Main protein contaminants were effectively retained by the membrane in UF-2. Ultrafiltration was
carried out either using centrifugal devices with 30 and 10 kDa MWCO regenerated cellulose membranes, or a stirred cell device with 10 kDa
MWCO polyethersulfone (PES) and regenerated cellulose (RC) membranes. Total rejection of surfactin was consistently observed in UF-1, while
in UF-2 PES membranes had the lowest rejection coefficient of 0.08 ± 0.04. It was found that disruption of surfactin micelles, aggregation of
protein contaminants and electrostatic interactions in UF-2 can further improve the selectivity of the membrane based purification technique.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Surfactin; Ultrafiltration; Zeta potential; Micelle size; Regenerated cellulose (RC); Polyethersulfone (PES)

1. Introduction
The total quantity of chemical surfactants and biosurfactants
produced in the US and worldwide is estimated at more than 10
billion pounds sterling and 25 billion pounds sterling, respec-
tively [1]. Almost all surfactants currently in commercial use are
chemically derived from petroleum; however, interest in micro-
bial surfactants, including surfactin, has been steadily increasing
in recent years due to their diversity, environmentally friendly
nature, the possibility of their production through fermentation,
and their potential applications in the environmental protection,
crude oil recovery, health care and food-processing industries.
Surfactin is a cyclic lipopeptide biosurfactant, produced by
various strains of B. subtilis. Surfactin consists of a heptapeptide
headgroup with the sequence Glu-Leu-d-Leu-Val-Asp-d-Leu-
Leu closed to a lactone ring by a C14–15 ␤-hydroxy fatty acid.
A variety of important applications and physiological activities
∗ Corresponding author. Tel.: +44 118 3788728.
E-mail address: p.jauregi@reading.ac.uk (P. Jauregi).
have been proposed for surfactin, including reports that it has
hemolytic, antiviral, antibacterial and antitumor properties [2].
The application of surfactin as a strong, biodegradable detergent
for technical and household purposes can also be envisaged, but
would require much cheaper production methods [3].
Downstream processing in many biotechnological processes
is responsible for up to 60% of the total production cost, and the
major obstacle for the commercialization of surfactin is its recov-
ery and purification from complex fermentation broths [4]. One
of the most widely used approaches for the recovery and purifica-
tion of surfactin involves precipitation at extreme pH followed by
extraction with organic solvents, adsorption chromatography, or
thin layer chromatography [5]. However, these techniques usu-
ally lead to the generation and release or hazardous wastes and
thus, it is necessary to develop a more economic and environ-
mentally friendly approach. One of the unique characteristics of
surfactant molecules is that above the critical micelle concen-
tration (CMC) they associate to form supramolecular structures,
such as micelles or vesicles, with nominal molecular diameters
up to two to three orders of magnitude larger than the single
unassociated molecules. Surfactant molecules in this form can

0376-7388/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2007.03.023
52 M.H.M. Isa et al. / Journal of Membrane Science 296 (2007) 51–57
be retained in the retentate of ultrafiltration (UF) membranes of
certain molecular weight cut-offs (MWCOs). Mulligan et al. [6]
successfully employed this principle for the recovery and purifi-
cation of surfactin and rhamnolipids from complex fermentation
broths with one step of UF. Recovery and purification of surfactin
from broth samples with one step of UF was further studied by
Sen and Swaminathan [7] by using a stirred cell device and
evaluating some filtration characteristics and their effects on the
recovery and purity of the final products. In this latter study, they
managed to recover surfactin with a purity value based on CMC
of 70%.
Another reported method by Lin and Jiang [8] for the recovery
and purification of surfactin from fermentation broths consists
of two-step UF. In that study, they reported recovery of 95%
of surfactin from their fermentation broth, although there was
no report on the purity of the final product. For a procedure or
a method to be successful and potentially to be commercial-
ized, it must be able to achieve a very good recovery rate with
a high purity value. Although UF has been reported to be effec-
tive, detailed investigation of the mechanism and evaluation of
the separation has not been carried out. In this study character-
ization of surfactin in terms of size and surface charge has led
to an improved understanding of the mechanism of the separa-
tion based on a two-step UF process which will lead to further
optimization of the separation.
2. Materials and methods
2.1. Culture conditions, media and fermentation
The strain used in the fermentation was B. subtilis ATCC
21332, which was obtained from the American Type Culture
Collection (Rockville, MD, USA). Inocula were prepared in two
stages. Firstly the microorganism was recovered from the strain
bank and grown in the nutrient agar. Then, a loopful of cells from
this culture was added to 25 ml of nutrient broth containing 40 g/l
of glucose, which was incubated for 24 h on an orbital shaker
(300 rpm) at 30 ◦ C. The obtained culture broth was used to inoc-
ulate five conical flasks, each one containing 45 ml of Cooper’s
medium [9], using 5 ml of inoculum per flask (10%). These flasks
were incubated under the same conditions as the previous cul-
ture but, only for 16 h. The five conical flasks (total volume:
250 ml) were used to inoculate a 5-l bioreactor (BioFlo 110
New Brunswick Scientific, UK) containing Cooper’s medium
[9] (5% inoculum).
The agitation system of the bioreactor contained two six-
blade impellers on a single drive shaft connected to a motor
whose speed was controlled by a fermentation control unit. Both
pH and dissolved oxygen probes were also connected to a fer-
mentation control unit and pH was maintained constant by the
automatic addition of either 1 M NaOH or 1 M HCl, according
to the signal received from the pH electrode.
The fermentation was conducted at 30 ◦ C, pH 7, for 55 h,
under depleted oxygen conditions, which was accomplished
using a low air flow rate (1 vvm−1 ) and low stirrer speed
(100 rpm). Culture broth samples of approximately 40 ml were
taken during the course of the fermentation at regular intervals
in order to determine biomass and surfactin content. Similarly,
temperature, dissolved oxygen and pH values were recorded
prior to each sampling procedure. The final fermentation broth
obtained was harvested, clarified by centrifugation for 10 min at
8000 rpm at room temperature, divided in fractions and finally
frozen for further studies.
2.2. Recovery and purification of surfactin
2.2.1. Ultrafiltration centrifugal device
Small scale UF procedures were carried out using Amicon-
Ultra 15 centrifugal filter devices (Millipore, USA) with a
MWCO of 10 or 30 kDa, containing in both cases a regener-
ated cellulose membrane of 7.6 cm2 of active area. In general,
a feed volume of 10–15 ml was added to the filter unit and
the device assembled and centrifuged at 4000 rpm for approx-
imately 5–10 min as recommended by the supplier. After
that, the retentate in the filter unit and the permeate in the
centrifugal tube were recovered, weighed and analysed by
HPLC for surfactin concentration. The rejection of surfactin
by a membrane was defined as rejection coefficient (R) as
below:
CF − CP
R= (1)
CF
where CF and CP are the concentration of surfactin in the feed
and permeate, respectively.
Feed, as well as retentate and permeate volumes were esti-
mated considering the density of the aqueous solutions as being
equal to 1 g/ml. The density of solutions in 50% (v/v) methanol
was estimated by weighing 50 ml of this solvent system. The
recovery and purification by two-step UF which was conducted
at room temperature. After the first step (UF-1), the surfactin
micelles were concentrated in the retentate. Then the solution
conditions of the retentate were altered to disrupt surfactin
micelles, thus allowing them to permeate the membrane in
the second step (UF-2), whereas the high molecular weight
compounds were retained in the retentate. This procedure was
conducted with 50% (v/v) methanol solution as reported by Lin
and Jiang [8]. Therefore, using the 30 kDa MWCO membrane,
the retentate obtained in UF-1 was subjected to 1:10 dilution
with 50% (v/v) methanol solution to favour the disruption of
micelles and subsequent passage of surfactin to the permeate in
UF-2.
2.2.2. Stirred cell device
Lab scale UF of the fermentation broth was performed with
a magnetically stirred UF cell (Gyrosep 300, Techmate Ltd.,
UK). UF was conducted with two sets of membrane materials
with 10 kDa MWCO and an effective filtration area of 40 cm2
at a constant pressure of 2 bar, namely Biomax polyethersul-
fone (PES) and Ultracel regenerated cellulose (RC) membranes
(Millipore, USA). Throughout the UF procedures, the flow rate
across the membrane (filtration rate) were estimated by collect-
ing permeates of certain volumes during an exactly controlled
period of time. Such permeates were then weighed and trans-
lated to volume according to the density of the solution. Flux of
 M.H.M. Isa et al. / Journal of Membrane Science 296 (2007) 51–57
permeates were calculated by using the following equation:
flow rate (ml/ min)
flux (LMH or L/m2 h) = × 600
membrane area (cm2 )
UF-1 was carried out with a feed volume of 100 ml and
flow rate was monitored during the course of UF. The result-
ing retentate was diluted 1:10 in a solution of methanol 50%
(v/v) and ultrafiltered again in the UF-2 under the same condi-
tions. Both steps of UF were conducted at room temperature.
At each step of UF, the rejection of surfactin was determined
by measuring surfactin concentration in permeate by using Eq.
(1). UF-2 filtrations with both PES and RC membranes were
carried out in triplicates by using the same set of membranes.
Each run is indicated as cycle 1, cycle 2 and cycle 3. After
each step/cycle, the membranes were washed in situ with dis-
tilled water for 30–60 min, followed with 0.1 M NaOH solution
for 30–60 min and then were rinsed with distilled water for
approximately 30 min. This procedure was carried out at room
temperature, 2000 rpm stirring and at 2 bar pressure. The mem-
branes were then stored in ethanol 10% (v/v) solution at 4 ◦ C as
recommended by the manufacturer.
2.2.3. Membrane permeability measurements
The condition of membranes used in UF-2 was evaluated
by comparing the flux of distilled water of membranes before
cleaning (after passing feed for UF-2) and clean membranes,
where in both cases the membranes had been exposed to filtra-
tion of feed in 50% (v/v) methanol solution and new membranes
which had only distilled water passed through it. This experi-
ment was carried out at room temperature, 2 bar pressure and
at 2000 rpm stirring except for the membranes before cleaning.
This experiment is to monitor the effects of either concentra-
tion polarization, membrane fouling or exposure to 50% (v/v)
methanol solution used in UF-2. Flux of water was calculated
by using Eq. (2).
2.3. Analytical procedures
2.3.1. Surfactin concentration analysis
Surfactin concentration was determined by reverse phase
HPLC. The system used was a Gilson 306 (Rockford, IL, USA),
with a C18 column of dimensions 250 mm × 4.6 mm, and a
particle size of 5 ␮m. The flow rate of the mobile phase was
1.1 ml/min with initial gradient starting from 50 to 85% acetoni-
trile in 0.1% trifluoroacetic acid in the first 15 min. The gradient
then remained at 85% for 20 min before increasing to 100% for
5 min as a washing step before returning back to 50%. A 50 ␮l
sample was injected in each run which lasted for 60 min, and
eluent absorbance monitored at 214 nm. The system was cali-
brated using purified surfactin obtained from Sigma. The area of
the peaks eluting between 19 and 31 min, which were identified
as having the same retention times as those peaks eluting from
the Sigma standard, were added to give the total surfactin peak
area. This value was used to determine the surfactin concentra-
tion in fermentation broth samples, as well as samples from the
purification procedures.
53
2.3.2. Zeta potential measurements
Electrophoretic mobility of surfactin standard in 50% (v/v)
(2) methanol solutions and UF-2 feeds (UF-1 retentate diluted with
50% (v/v) methanol) were determined with a ZetaMaster anal-
yser (Malvern Instrument Ltd., Worcestershire, UK). The pH
of each set of samples was controlled by using 0.1 M HCl and
0.1 M NaOH. Measurements were performed in triplicate and
each sample was scanned three times by the instrument for each
measurement.
2.3.3. Particle size measurements
The presence and size of surfactin micelles in the fermenta-
tion broth, UF-1 retentate and its treatment with the addition of
50% (v/v) methanol were evaluated by dynamic light scattering,
using a Zetasizer Nano ZS system (Malvern, UK), courtesy of
Malvern Instruments. This instrument determines the intensity
of the light scattered at a certain angle, which is in turn, related
to the Brownian motion of the particles, their diffusion coeffi-
cient and finally their size. It is able to detect particles ranging
from 0.6 nm to 6 ␮m. Each sample was analysed five times and
information about size distribution by intensity and by volume
as well as total intensity were recorded and related to particle
size and number of particles present in solution, respectively.
2.3.4. Biomass measurements
Biomass was determined by dry cell weight, which was mea-
sured by membrane filtration, using nylon filters with a pore size
of 0.2 ␮m and diameter of 47 mm (Whatmann, Abror, USA). The
individual weight of each membrane was recorded. An aliquot
of culture broth (5 ml) was filtered through the membrane under
vacuum; the membrane was dried in a microwave for 10 min and
placed in a desiccator for approximately 24 h.
2.3.5. Total protein measurements, Pp
The total amount of protein present at each stage of the purifi-
cation procedures was determined by the bicinchoninic acid
method (BCA) [10]. A calibration curve was produced using
bovine serum albumin (BSA) as the protein standard solution.
Fermentation residual glucose, as a reducing sugar, would be
the main interfering compound of the technique [11]. For that
reason, it was decided to carry out protein precipitation with ace-
tone at −20 ◦ C for 30 min on all the samples prior to the analysis,
as suggested by Pierce Biotechnology (BCA assay supplier).
3. Results and discussion
3.1. Fermentation of surfactin
B. subtilis ATCC 21332 was cultivated under conditions in
which the levels of dissolved oxygen were allowed to be depleted
(Fig. 1). Dissolved oxygen (DO) levels became depleted dur-
ing the late exponential growth phase after 24 h of cultivation.
Despite this, the biomass concentration increased for a further
8 h after the DO concentration reached 0%. The offset between
the DO depletion and the onset of stationary phase was as a
result of anaerobic growth. This set of fermentation data showed
almost a similar trend to previous data reported by Davies et
54 M.H.M. Isa et al. / Journal of Membrane Science 296 (2007) 51–57
Fig. 1. Production of surfactin through fermentation of B. subtilis ATCC 21332:
(䊉) surfactin concentration (mg/l); () biomass (g/l).
al. [12]. The final yield of surfactin in the broth was 583 mg/l,
which is slightly higher than that obtained by Davies et al. [12]
of 439 mg/l under the same experimental conditions.
3.2. Recovery and purification of surfactin
3.2.1. First ultrafiltration step (UF-1)
As reported by Lin and Jiang [8], surfactin micelles could be
effectively retained with membranes of 30–10 kDa MWCO. It
was therefore decided to concentrate the surfactin micelles in
the retentate in UF-1, achieving also some degree of purifica-
tion, with contaminants such as glucose, salts and low molecular
weight proteins passing through the membrane.
As shown in Table 1, surfactin was totally rejected by all
membranes regardless of their configuration, membrane chem-
istry and MWCO, which confirms the macromolecular-like
behaviour of surfactin micelles. In all UF performed, at least 90%
of surfactin was recovered. The total recovery of surfactin of up
to 90% with centrifugal device indicates some surfactin might
have been lost due to membrane fouling. The results demonstrate
that surfactin micelles can be effectively recovered by using
either 30 or 10 kDa RC membranes as was found by Lin and
Jiang [8]. Also no significant differences (P > 0.05) were found
between the different UF devices and the different types of mem-
branes. Thus both RC and PES membranes (MWCO = 10 kDa)
can retain surfactin micelles effectively. The retentate contain-
ing surfactin micelles was then diluted with 1:10, 50% (v/v)
methanol prior to UF-2.
Table 1
Surfactin yield and rejection coefficient (R) in UF-1 (average of duplicate# and
triplicate measurements* )
Method and membrane type Total recovery (%)
Centrifugal device (RC 30 kDa)# 90 ± 8
Stirred cell (RC 10 kDa)* 98 ± 4
Stirred cell (PES 10 kDa)* 95 ± 1
Table 2
Surfactin yield in the final permeate and rejection coefficient (R) in UF-2 (average
of duplicate# and triplicate* measurements)
Method and membrane type Rejection Total recovery Purity
coefficient, R (%) (%)
Centrifugal device (RC 10 kDa)# 0.32 ± 0.13 91 ± 7 94 ± 1
Stirred cell (RC 10 kDa)* 0.24 ± 0.11 96 ± 5 93 ± 1
Stirred cell (PES 10 kDa)* 0.08 ± 0.04 96 ± 5 94 ± 2
3.2.2. Second ultrafiltration step (UF-2)
Based on the method suggested by Lin and Jiang [8], it was
decided to alter the retentate conditions in UF-1 to disrupt the
micelle structure and then recover surfactin in the permeate
of UF-2, while high molecular weight compounds were still
rejected by the membrane. The general approach was to achieve
a surfactin concentration lower than the CMC by means of dilut-
ing the retentate in the UF-1 and increasing the original intrinsic
surfactin CMC.
The data in Table 2 shows that UF-2 consistently led to more
than 50% of surfactin being recovered in the permeate. The total
recovery of surfactin of up to 96 ± 5% indicates that almost all
surfactin is accounted for and therefore no losses due to fouling
occur. No significant differences (P > 0.05) were found in the
rejection of surfactin with 10 kDa MWCO RC membranes for
both devices, which indicated that the membrane configuration
had no significant effect on the rejection of surfactin. However,
the membrane material had an effect on the recovery of sur-
factin as less surfactin was rejected with PES (R = 0.08) than
with RC (R = 0.24) membrane. The purity however remained
the same, thus both types of membrane rejected protein contam-
inants effectively.
Although the highest recovery of surfactin was obtained with
PES compared to RC membrane, the flux decreased quite sig-
nificantly over time with the former (Fig. 2). This shows that
PES membrane was affected more by concentration polarization
than RC, which was due to hydrophobic interactions between
Rejection
coefficient, R
1
1 Fig. 2. Flux of UF-2 permeate with PES and RC 10 kDa membranes: (䊉) PES
1 cycle 1; () PES cycle 2; () PES cycle 3; () RC cycle 1; () RC cycle 1;
() RC cycle 3.
 M.H.M. Isa et al. / Journal of Membrane Science 296 (2007) 51–57
Fig. 3. Measurements of flux of water with PES 10 kDa MWCO: (A) (䊉) clean
membrane; (B) () before cleaning (after passing feed for UF-2); (C) () new
membrane.
PES and the aggregated proteins contaminants in UF-2 (refer to
particle size measurements; Section 3.4). Further experiments
were carried out to investigate these differences in permeabil-
ity by measuring flux of water through: (A) membrane which
has been cleaned after UF-2 (B) membrane after UF-2 (C) new
membrane.
Significant differences in flux rates of A and B in Fig. 3 shows
that some degree of fouling occurred with PES membrane, while
overall there was no significant difference between A and B in
Fig. 4 with RC membrane. This shows that there is no occurrence
of fouling with RC membrane since they are hydrophilic.
Another factor to be considered in UF-2 is the effects of 50%
(v/v) methanol solution on membrane resistance. Comparison of
flux rates of A–C in Fig. 3 shows the effect of pore constriction
which led to the increase in membrane resistance for PES. While
in Fig. 4, the opposite effect was observed for RC membrane
as comparison of flux rates of A–C shows the effect of pore
dilation which lowered the membrane resistance. Both results
are in agreement with the observation by Lencki and Williams
[13] that organic solvents can cause pore constriction or dilation
of UF membranes, therefore affecting their permeability.
Therefore, higher selectivity of surfactin separation achieved
by PES membrane cannot be explained on the basis of dif-
Fig. 4. Measurements of flux of water with RC 10 kDa MWCO: (A) (䊉) clean
membrane; (B) () before cleaning (after passing feed for UF-2); (C) () new
membrane.
55
Fig. 5. Zeta potential of surfactin standard and surfactin in UF-2 feeds at different
pH in methanol 50% (v/v) solution: (䊉) surfactin standard; () UF-2 feeds.
ferences in permeability. One probable factor that makes PES
membrane the more effective UF membrane might be due to the
difference in interactions between the membrane surface and
surfactin molecules. In order to investigate this further the sur-
face charge (zeta potential) of surfactin was measured under
different experimental conditions.
3.3. Zeta potential measurements
Zeta potential measurements of surfactin standard and sur-
factin in UF-2 feeds at different pH (Fig. 5) and surfactin at
different concentration in UF-2 feeds at pH ∼ 7.7 (Fig. 6), which
was the working pH during UF-2, allowed a better understand-
ing of the effects on interactions between surfactin molecules
and UF membranes.
From both Figs. 5 and 6, it can be concluded that surfactin
molecules in methanolic solution possessed an overall negative
surface charge, which means they behaved as an anionic surfac-
tant in this type of solution. In the present study, electrostatic
interactions were considered the main interactions between
membranes and surfactin. Surfactin molecules in methanolic
solution of UF-2, with pH of ∼7.7, were negatively charged
with a zeta potential of approximately −27 mV. On the other
hand, according to Salgin et al. [14], at pH 6.8, PES membrane
had a negative charge of −48.33 and −33.96 mV at the ionic
strength of 0.01 and 0.1 M KCl, respectively. While according
to Urbanski et al. [15], at pH 7, RC membrane had a nega-
Fig. 6. Zeta potential of surfactin at different concentration in UF-2 feeds in
methanol 50% (v/v) solution (pH ∼ 7.7).
56 M.H.M. Isa et al. / Journal of Membrane Science 296 (2007) 51–57
Fig. 7. Particle analysis by light scattering of the UF-1 retentate with a membrane
of 30 kDa MWCO. Count rate: 8200 kcps. Mean diameter: 9 nm.
Fig. 8. Particle size analysis by light scattering of the final fermentation broth.
Count rate: 1500 kcps. Mean diameter: 8.2 nm.
tive charge of −5.6 mV. Therefore, it can be concluded that
under the conditions of UF-2 (pH ∼ 7.7), PES membranes were
more strongly negatively charged than RC membranes. Less sur-
factin was retained with PES than with RC membranes because
electrostatic repulsive interactions between surfactin molecules
and the more negatively charged PES membranes increased the
permeability for surfactin.
Surfactin molecule separation by UF shows similar trends as
those observed for UF of SDS (sodium dodecyl sulfate) [16],
where the best separation of SDS from aqueous model solutions
was obtained with PES and PS (polysulfone) as compared to
cellulose and RC membranes, the latter of which is characterized
by its rather low permeability. Kowalska et al. [17] also reported
that the best separation of SDS solution was achieved with PES
and PS membranes.
3.4. Particle size measurements
The retentate generated in UF-1 contains particles with a
mean diameter of approximately 9 nm (Fig. 7), being compa-
rable with those found originally in the fermentation broths of
8 nm (Fig. 8). According to micelle size measurements of a
commercial surfactin aqueous solution (Fig. 9), particles with
a mean diameter of 8 and 9 nm in fermentation broths and UF-1
Fig. 9. Particle size analysis by light scattering of commercial standard surfactin
solution. Count rate: 7700 kcps. Mean diameter: 6.3 nm.
Fig. 10. Particle analysis by light scattering of the surfactin solutions in 50%
(v/v) methanol. Count rate: 3600 kcps. Mean diameter: 100 nm.
retentate, respectively, corresponded to the diameter of surfactin
micelles. Broth conditions seem to slightly increase the micelle
size, which depends on factors such as ionic strength and pres-
ence of organic compounds that may coexist in the micelle and
thus increase its size [18]. The higher count rate of 8200 kcps
measured would be an indication of a higher number of par-
ticles in the UF-1 retentate compared to fermentation broths.
In other words, micelles in the retentate would have the same
size as in the fermentation broths, regardless of the higher sur-
factin concentration in this fraction, which would be shown by
a higher count rate. After the addition of 50% (v/v) methanol
solution to this retentate, particles with a mean diameter of 9 nm
corresponding to surfactin micelles were no longer present in
solution, indicating the rupture of such structures. Moreover,
the presence of larger particles was detected, with a mean diam-
eter of 100 nm (Fig. 10). Such large particles were generated
after the addition of 50% (v/v) methanol and could be protein
aggregates induced by methanol, in a similar mechanism already
described for protein precipitation with ethanol [11].
Therefore, it can be concluded that the effective size-based
separation of surfactin from proteins occurred due to not only
the disruption of micelles into free surfactin molecules, but also
the formation of large protein aggregates.
4. Conclusions
Selective separation of surfactin from fermentation broths
can be achieved by a two-step UF process at the laboratory scale
and shows potential to be scaled-up. Regarding the design of the
process, it could be said that in the first step of UF rejection of
micellar surfactin by a semi-porous membrane was achieved. At
this stage, surfactin was mainly purified from residual glucose
and salts, as proteins were retained together with the surfactin.
In the second step, surfactin was purified essentially from pro-
teins. It was found that PES membrane is the more suitable
UF membrane for the purification of surfactin especially in the
second step of UF. The size and surface charge measurements
carried out demonstrated that disruption of surfactin micelles,
aggregation of protein contaminants and electrostatic interac-
tions between surfactin molecules and the membrane surface
have a major influence on its selective separation, hence an
improved understanding of the key factors affecting the selec-
tivity of surfactin separation has been achieved. One concerning
factor in the second step of UF process is the effects of exposure
of UF membranes to a methanolic solution; it was found that
the permeability of PES membranes reduced after subsequent
 M.H.M. Isa et al. / Journal of Membrane Science 296 (2007) 51–57
exposure to 50% (v/v) methanol solution. The process can be
further improved by, for example, choosing a membrane that
is very resistant to organic solvents. Moreover PES membranes
are particularly affected by concentration polarization therefore
further experiments will be carried out using a cross flow fil-
tration unit which will also enable to test the scalability of this
separation process.
References
[1] S.S. Cameotra, R.S. Makkar, Recent applications of biosurfactants as bio-
logical and immunological molecules, Curr. Opin. Microbiol. 7 (2004)
262–266.
[2] H. Heerklotz, J. Seelig, Detergent-like action of the antibiotic peptide
surfactin on lipid membranes, Biophys. J. 81 (2001) 1547–1554.
[3] E. Rosenberg, Z. Ron, High and low-molecular-mass microbial surfactants,
Appl. Microbiol. Biotechnol. 52 (1999) 154–162.
[4] J.D. Desai, I. Banat, Microbial production of surfactants and their commer-
cial potential, Microbiol. Mol. Biol. Rev. 61 (1997) 47–64.
[5] K. Arima, A. Kakinuma, G. Tamura, Surfactin a crystalline peptidelipid
surfactant produced by Bacillus subtilis: isolation, characterization and its
inhibition of fibrin clot formation, Biochem. Biophys. Res. Commun. 31
(1968) 488–494.
[6] C. Mulligan, R. Yong, B. Gibbs, S. James, H.P.J. Bennett, Metal removal
from contaminated soil and sediments by the biosurfactant surfactin, Env-
iron. Sci. Technol. 33 (1999) 3812–3820.
[7] R. Sen, T. Swaminathan, Characterization of concentration and purifica-
tion parameters and operating conditions for the small-scale recovery of
surfactin, Process Biochem. 40 (2005) 2953–2958.
57
[8] S.C. Lin, H.J. Jiang, Recovery and purification of the lipopeptide biosur-
factant of Bacillus subtilis by ultrafiltration, Biotechnol. Tech. 11 (1997)
413–416.
[9] D.G. Cooper, C. MacDonald, J. Duff, N. Kosaric, Enhance production of
surfactin from Bacillus subtilis by continuous product removal and metal
cation additions, Appl. Environ. Microbiol. 42 (1981) 408–412.
[10] P.K. Smith, R. Krohn, G. Hermanston, A. Mallia, F. Gartner, M. Proven-
zano, E. Fujimoto, B. Goeke, B. Olson, D. Klenk, Measurement of protein
using bicinchoninic acid, Anal. Biochem. 150 (1985) 76–85.
[11] E.L.V. Harris, S. Angal, Protein Purification Methods. A Practical
Approach, Oxford University Press, Oxford, UK, 1989.
[12] D.A. Davies, H.C. Lynch, J. Varley, The application of foaming for the
recovery of surfactin from B. subtilis ATCC 21332 cultures, Enzyme
Microb. Technol. 28 (2001) 346–354.
[13] R.W. Lencki, S. Williams, Effects of nonaqueous solvents on the flux
behaviour of ultrafiltration membranes, J. Membr. Sci. 101 (1995) 43–
51.
[14] S. Salgin, S. Takaç, T.H. Özdamar, Adsorption of bovine serum albumin
on polyether sulfone ultrafiltration membranes: determination of interfacial
interaction energy and effective diffusion coefficient, J. Membr. Sci. 278
(2006) 251–260.
[15] R. Urbanski, E. Goralska, H. Bart, J. Szymanowski, Ultrafiltration of sur-
factant solutions, J. Colloid Interface Sci. 253 (2002) 419–426.
[16] K. Majewska-Nowak, I. Kowalska, M. Kabsch-Korbutowicz, Ultrafiltration
of SDS solutions using polymeric membranes, Desalination 184 (2005)
415–422.
[17] I. Kowalska, M. Kabsch-Korbutowicz, K. Majewska-Nowak, T. Winnicki,
Separation of anionic surfactant on ultrafiltration membranes, Desalination
162 (2003) 33–40.
[18] M.R. Porter, Handbook of Surfactants, 2nd ed., Blackie Academic and
Professional, Glasgow, UK, 1994.