You are on page 1of 8

Science China Press and Springer-Verlag Berlin Heidelberg 2010 csb.scichina.com www.springerlink.

com

*Corresponding author (email: zhaoliangwu@163.com)
Articles
SPECIAL TOPICS:
Bioengineering April 2010 Vol.55 No.12: 12131220
doi: 10.1007/s11434-010-0110-x
A drainage-enhancing device for foam fractionation of proteins
WU ZhaoLiang
*
, QIAN ShaoYu, ZHENG HuiJie & ZHAO YanLi


Department of Bioengineering, Hebei University of Technology, Tianjin 300130, China
Received September 11, 2009; accepted January 9, 2010

The foam fractionation of nisin from its fermentation broth was studied. Two types of devices consisting of a rubber piston and a
foam riser were developed to enhance foam drainage. The separation performance of these two devices was investigated. Experi-
mental results indicated that the second device could significantly reduce the liquid fraction of the foam leaving the column, c
out
,
leading to a higher enrichment of the out-flow stream. As its mounting height increased from 0 to 15 cm, c
out
declined from
7.07 to 6.13 and the maximum nisin activity in the foamate could reach 39.6 IU/L. The slight increase in nisin inactivation
rate indicated the applicability of this method in the recovery and concentration of proteins. Finally, the mechanism of the process
was primarily explained by invoking recent work on pneumatic foams. This research provides a basis for the design of multistage
draining foam fractionator which could potentially be an effective separation equipment.
bioseparation, foam fractionation, drainage, liquid holdup, bubble column

Citation: Wu Z L, Qian S Y, Zheng H J, et al. A drainage-enhancing device for foam fractionation of proteins. Chinese Sci Bull, 2010, 55: 12131220, doi:
10.1007/ s11434-010-0110-x



Foam fractionation is a member of a group of processes
known as adsorptive bubble separation technique [1]. It is
based on the differences in surface activity of molecules in a
mixture coupled with the very high surface-to-volume ratio
of a foam. The foam separation technique has been widely
used in ore flotation and waste water treatment, etc. [2,3].
Recently, it appears that the separation of proteins by this
method has been studied on the basis of its possible ap-
plicability to food and biochemical industries [47]. A re-
view of the purification of proteins using the foam fraction-
ation technique can be found in [8]. The enrichment in pro-
tein concentration in foam is known to be a combined effect
of interfacial adsorption and foam drainage. Both of them
would be a function of liquid physicochemical properties
and operating conditions [9].
There are different approaches proposed to improve the
performance of the foam separation technique. Briefly, they
can be divided into three categories: (1) the methods used to
improve adsorption performance in the liquid layer includ-
ing changes of the format of gas distributor [10,11] and
adding components to the liquid pool [12,13]; (2) changes
of the operation mode, such as using counter-current [13,14]
or multistage operation [1517] and (3) enhancement of
foam drainage [1820].
Due to the short resident time of the bubbles in the liquid
pool, the effect of the first improved method was not obvi-
ous. And yet, the high cost involving in the preparation of
various facilities and the limit of adsorption saturation [21]
are disadvantageous for the application of multistage process.
However, unlike these two methods, the improvement of
foam drainage, which has greatly enhanced the quality of
separation performance according to the report of Tsubomizu
et al. [19], is more economical and promising. So far, the
theory of foam drainage is already quite mature, at least for
free liquid foam drainage [9,2225], while only a few stud-
ies have been made on the methods of accelerating the
drainage process, which can give rise to lower liquid frac-
tion in the foam.
Furthermore, even though some procedures may provide
excellent results in terms of enrichment and recovery for
certain surfactant solutions, the performance may be limited
in the case of recovering biologically active compounds
1214 WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12
such as bacteriocins and enzymes from fermentation broths.
Since the fermentation broths usually contain very high lev-
els of peptides and proteins (10~30 g L
1
compared to a
bacteriocin concentration of 0.01~0.1 g L
1
[26]), the in-
crease in surface concentration of the target protein by in-
creasing its concentration in the aqueous phase may be lim-
ited due to adsorption saturation. Once the surface is satu-
rated, operation in multistage will not significantly enhance
the enrichment of the target component [21]. And due to the
enhanced surface dilatational and bulk viscosities, the rate
of drainage of the foam will be slowed [27]. What is more,
there are also problems of denaturation for biologically ac-
tive molecules like bacteriocins and enzymes [2830], thus
some improving methods may lead to a serious loss of bio-
logical activity.
To solve these problems, it is necessary to develop a
foam fractionation equipment especially for separating bio-
active substances, which can enhance the foam drainage and
meanwhile has moderate impact on the bioactivity of the
target objects. Here we designed a rubber piston with a
foam riser mounted at the center of the column to achieve
this objective. The foam fractionation of nisin from its fer-
mentation broth was carried out. The mechanism of this
device was simply explained by invoking recent work on
pneumatic foams of Stevenson [31].
1 Materials and methods
1.1 Materials
A schematic diagram of the foam fractionator used in this
study is illustrated in Figure 1.
The column (B+C) was constructed of a transparent
acrylic resin tube of 5.0 cm inside diameter and 120 cm in
height. A plastic elbow (E) was connected to the top of the
column to deliver the foam to a collector cup (F), which was
placed on a digital balance (G) to weigh the foamate online.
A rubber piston (D) with a foam riser mounted at the center
was inserted into the column. The diameter of the piston
was 5.0 cm, corresponding to the inside diameter of the
column. Two types of foam risers were used. The type I
foam riser was a cylindrical plastic tube of 2.6 cm inside
diameter and 3.0 cm in height, with both ends open. The
type II foam riser was the same cylindrical plastic tube, with
the upper end sealed and horizontal pores drilled through
the top 1 cm of the pipe wall. The pore diameter and pore
number were 3 mm and 12 respectively, giving an opening
ratio 4.32% (with respect to the cross-sectional area of the
column). Four outlet ports (Vs, only two were shown for
clarity) were equipped at 100, 105, 110 and 115 cm from
the bottom of the column, respectively. They could be con-
nected to another port at the lower part of the column via a
constant flow pump (H
1
). This pump was used to drain the
liquid accumulated on the rubber piston back to the bulk
liquid solution in continuous operation experiments. And
another constant flow pump (H
2
) was used to feed addition-
al solution into the liquid layer in some experiments. A
stainless steel plate (A) with laser-burned pores (150 m
10 m), arranged in equilateral triangle patterned with a
center to center distance of 4 mm, was installed at the bot-
tom of the column and served as the gas sparger.
Nisin-containing fermentation broth (a gift from Kangyi
Bioengineering Co., Ltd., Tianjin, China) was used as the
model surfactant solution. It was chosen since nisin fer-
mentation solution proved to be of fine foamability and
nisin was traditionally recovered from its fermentation broth
by the foam fractionation method [32]. Fresh broth was
clarified by microfiltration and stored at 4C for no more
than one week before use. The labeled nisin titer and total
protein content were 100 g mL
1
and 0.0024 g mL
1
,
respectively. Fine bubbles with an arithmetical mean diam-
eter of 3.35 mm were generated within a wide range of gas
flow rates (200400 cm
3
min
1
). Nisin activity in all solu-
tions, including initial solution, residual solution and col-
lapsed foam solution, was determined by a modified du-
al-dosage agar diffusion assay [33]. All experiments were
performed at room temperature (20C2C).
1.2 Methods
(1) Measurement of c
out.
The mass accumulation of the
foamate m
F
in the collector was tracked and plotted against
bubbling time and the rate of mass change can be expressed
as
F
L F
d
,
d
m
Q
t
= (1)
where Q
F
is the volumetric liquid flow rate of the foamate
and
L
=1.037 g cm
3
was the density of the bulk liquid and
here it was approximately taken as the density of the
foamate. And a mass balance for the total foam volume is
G F
out out
.
1
Q Q
c c
=

(2)
Therefore,
out F
L G
out
d
.
d 1
m
Q
t
c

c
=

(3)
Since c
out
is much smaller than 1, eq. (3) can be simpli-
fied as
F
out
L G
d 1
.
d
m
Q t
c

= (4)
This method was validated by tracking m
F
during the
bubbling process. Experiments were carried out in batch
mode and m
F
was plotted against bubbling time. The rubber
piston and foam riser were removed in those experiments.
WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12 1215

Figure 1 Experimental apparatus. Components: A, Gas sparger; B, Liquid pool; C, Foam Layer; D, Rubber piston & foam riser; E, Elbow; F, Foam col-
lector; G, Digital balance; H1,2. Constant flow pump. Ports: I, Feed port; II, Liquid outlet in continuous operation; III, Air inlet; IV, Discharge port in batch
operation; V, Downcomer ports; VI, Sample ports. Dashed line in the foam column indicated the liquid level.
(2) Fed-batch foam fractionation. Foam fractionation
with the rubber piston & foam riser was firstly performed in
fed-batch mode to find proper operating conditions for the
final continuous operation. The fed-batch mode was a
continuous mode without discharging the bubbly liquid at
port II (Figure 1).
The fermentation broth was warmed to 20C in a tank
(50 L) and then pumped into the column to obtain a liquid
pool depth of 100 cm. Compressed air (0.18 MPa) was in-
troduced at the bottom through the sparger at a constant
volumetric flow rate 240 cm
3
min
1
, corresponding to a su-
perficial gas velocity of 12 cm min
1
. A liquid pool started
to accumulate on the upper surface of the rubber piston as
the foam passed through the foam riser (both type I and type
II). The liquid accumulation rate, Q
D
,was calculated by
( )
D
D C R
,
H
Q A A
t
A
=
A
(5)
where A
C
and A
R
were the inner cross sectional area of the
column and the outer cross sectional area of the foam riser,
respectively. H
D
was the rise of the level of the liquid pool
on the rubber piston in a certain time interval At; it was read
directly on the wall of the foam column. Samples of this
liquid were taken and analyzed. c
out
was also measured.
Generally, a steady state of foam generation (i.e. a
foam with a constant c
out
) could last for 3 hours in a
fed-batch operation under the following initial conditions:
liquid pool depth=100 cm, foam height=20 cm and volu-
metric gas flow rate, Q
G
=240 cm
3
min
1
. However, since the
distance between the liquid/foam interface and the rubber
piston, H
R
, had a profound influence on Q
D
, additional feed
solution was pumped into the column via the constant flow
pump (H
2
) during the foaming process to roughly keep the
liquid level in the liquid pool constant. The feed rate was
approximately 40 cm
3
min
1
. That was the so-called
fed-batch operation.
Here it is necessary to reinterpret the accurate definition
of Q
D
. In essence, Q
D
means the volumetric drainage rate of
interstitial liquid from the foam layer above the rubber pis-
ton, whose value is equal to the liquid accumulation rate. So
it should be noted that our following discussion in Section 2
is all about its physical meaning.
(3) Continuous foam fractionation. Continuous foam
fractionation was conducted at the proper operating condi-
tions which were pre-determined by the fed-batch experi-
mental results. Those continuous operating conditions in-
cluded: volumetric gas flow rate, Q
G
, volumetric feed rate,
Q
IN
, volumetric discharge rate, Q
OUT
, and the flow rate at
which the constant flow pump, Q
P
. Q
P
was firstly set at the
value of Q
D
calculated from eq. (5) using the fed-batch ex-
perimental data. Necessary adjustments were made to keep
the liquid level on the rubber piston constant, until the sys-
tem had attained a steady state. In a steady state operation a
mass balance gives
IN F OUT
= + , Q Q Q (6)
and with eq. (2), eq. (6) becomes
out
IN G OUT
out
.
1
Q Q Q
c
c
= +

(7)
The volumetric feed rate Q
IN
was set at 40 cm
3
min
1
,
1216 WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12
giving a HRT=49 min. Since the constant foam generation
in a fed-batch operation can last for about 3 hours, no de-
crease in c
out
should be observed in a continuous operation
at this feed rate. The downward flow rate of the bubbly liq-
uid in the liquid pool caused by the feed (2.04 cm min
1
)
was much smaller than the rising rate of bubbles, and thus
its effect on the pool residence time of the bubbles was ig-
nored.
2 Results and discussion
2.1 Validation of the method of calculating c
out

Solid squares in Figure 2 represented a typical foamate
mass vs. bubbling time profile. The straighter portion of the
plot (t>240 s) could be fit into a linear regression with a
high correlation coefficient (y=7.189+0.04449x, R
2
=
0.9997), indicating the stability of eq. (4). This could be
attributed to a quasi-steady state of foam generation in a
batch operation which was also observed by Neely et al.
[34]. However, the most interesting part of the plot lain in
the portion before the quasi-steady state, which may be
missed in a continuous operation, or if the time scales for
data collecting were too large. An example could be found
in Webb et al. [35]. NMRI studies of the free drainage of
egg white and meringue mixture froths performed by Ste-
venson et al. [36] also missed this since the froths were
generated by shaking rather than bubbling. Such an ap-
proach generally results in an initially spatially uniform
foam [25].
Open squares in Figure 2 were the derivative of mF with
respect to t. It was even clearer that there was a growing
phase of c
out
before the steady state. Therefore, all data in the
following experiments were collected 5 min after the intro-
duction of the air to bypass the unsteady state of foam gen-
eration.

Figure 2 Mass of foamate collected over bubbling time. Operating con-
ditions were: QG=200 cm
3
min
1
, liquid pool depth=100 cm, foam height
=20 cm, pressure of gas supply=0.18 MPa. The time required for the foam
to pass through the elbow (16 s) was subtracted from the time axis.
2.2 Proper operating conditions
The operating conditions used in the continuous operation
were shown in Table 1. H
R
and Q
G
were predetermined op-
erating conditions in the fed-batch operation. c
out
and Q
D

were calculated from eqs. (4) and (5), respectively. Q
IN
was
the volumetric feed rate in the continuous operation as de-
scribed above. Q
OUT
was calculated from eq. (7). It must be
pointed out that the values of Q
OUT
and Q
P
shown in Table 1
were the starting points where the continuous operation was
carried out. Adjustments of those two parameters were
made until the system had attained a steady state, which
usually took no more than one hour due to the high dilution
rate.
2.3 Effect of H
R
on Q
D
and c
out

Figure 3 showed the variation of Q
D
and c
out
as a function of
H
R
. In the case of type I foam riser. Both Q
D
and c
out
had
their maximum values when the rubber piston & foam riser
were placed just above the bubbly liquid/foam interface.
However, they displayed different trends after this point as
H
R
increased from 0 cm to 15 cm: Q
D
continued to decrease
near exponentially while only slight decreases of c
out
were
observed. A variable, F
n
, which was similar to Q
D
was pro-
posed in Neely et al. [37] for modeling a batch foam frac-
tionation process. It was defined as the flow rate of the in-
terstitial liquid draining from a postulated equilibrium stage.
Although an exponential regression of their data gave a
higher correlation coefficient (R
2
>0.99), they developed an
empirical linear drainage equation with a relatively lower
correlation coefficient (R
2
=0.97). However, Q
D
here should
not be taken as equivalent to F
n
since the rubber piston
stopped the interstitial liquid from draining down and thus
might have an effect on the liquid holdup profile of the
foam below it. The fact that Q
D
decreased much faster than
F
n
might partially be due to this effect.
In the case of type II foam riser, Q
D
exhibited a similar
exponential-decay trend with the type I foam riser while c
out

behaved very differently compared with the experimental
results of type I foam riser. It decreased near linearly as the
piston & foam riser moved from the liquid/foam interface to
15 cm above. It is easy to understand that since Type I foam
riser could not change the structure and geometry of the
foam no matter where it had been placed, the foam leaving
the column might possess a nearly constant liquid fraction.
However, the sudden change in the flow type at the small
holes of Type II foam riser might induce a complex change
in the foam structure, especially the bubble size distribution.
Tsubomizu et al. proposed that the foam lamella could be
partially wiped off as the foam squeezed through the small
pores on a horizontal plate inserted into the column [19]. If
that was the fact, the foam under the perforated plate should
have a greater liquid fraction. However, simple visual ob-
servation showed that the foam under our rubber piston was
WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12 1217
Table 1 Operating conditions for the continuous foam fractionation
Initial conditions Fed-batch experimental results Derived conditions for continuous operation
Foam riser HR(cm)
QG
cout ()
QD QIN QOUT QP
(cm
3
min
1
) (cm
3
min
1
) (cm
3
min
1
) (cm
3
min
1
) (cm
3
min
1
)
I
0
240
6.87 6.71
40
38.34 6.71
5 6.61 0.914 38.4 0.914
10 6.5 0.321 38.43 0.321
15 6.43 0.125 38.45 0.125
II
0 7.07 6.872 38.29 6.872
5 6.63 1.33 38.4 1.33
10 6.32 0.273 38.47 0.273
15 6.13 0.135 38.52 0.135


Figure 3 Effect of the position of foam riser on QD and cout
much dryer than if there was not such a piston.
The decrease in liquid fraction of the foam after it pass-
ing through the foam riser was considered to be a conse-
quence of enlarged bubble size in our study. It must be
pointed out that bubbles within the foam were always
growing in size even if there were not any add-ons in the
column. As the small holes in foam riser II might prompt
bubble coalescence, a lower liquid fraction of the foam
leaving the column could be expected when such a device
was inserted into the foam column. And since dry foam was
less stable than wet foam, more intensive bubble coales-
cence might occur if the foam riser II was placed higher
above the foam/liquid interface, where the foam was al-
ready dried due to gravity drainage. However, since the
liquid fraction of the foam was too low (Figure 3), the en-
hanced drainage had no obvious effect on Q
D
.
2.4 Predicting the effect of foam riser I on Q
D

A model for the effect of surface and internal bubble coa-
lescence on the hydrodynamic conditions of a foam was
proposed by Stevenson [31]. Although bubble coalescence
occurred at the foam front as well as in the bulk of the foam
phase, it seemed that the net liquid flux up the column was
determined by the radius of the bubbles at the top of the
foam layer.
Now consider a pneumatic foam rising in a column
which is separated into two segments by the rubber piston
with Type I foam riser. The foam riser is placed at the
height where the bubble radius is 0.6 mm. Since the foam
riser virtually serves as the exit of the foam beneath it, the
volumetric liquid flow through the foam riser, Q
f
can still be
calculated by simply invoking Stevensons method. As-
suming that there is no bubble coalescence at the foam riser,
Q
f
can be read from the peak of the corresponding Q
f
vs.
curve in Figure 4.
After the foam has passed the foam riser, bubbles within
the foam continue to burst and finally give a mean radius of
0.8 mm at the top of the column. Similarly the volumetric
liquid flow rate leaving the column can be obtained by in-
voking Stevensons method. It was pointed out by Steven-
son [31] that Q
f
was a function of bubble radius; it de-
creased as bubble size increases. Thus the incoming side
and outgoing side liquid flows of the segment above the
foam riser will not balance each other, i.e. the incoming
(corresponding to r
b
= 0.6 mm bubbles) is larger than the
outgoing (corresponding to r
b
= 0.8 mm bubbles). The ex-
cess will accumulate on the rubber piston and the volumet-
ric accumulation rate can be calculated graphically as Q
D
in
Figure 4.
Consider the case of a pneumatic foam with initial r
b
=
0.5 mm, but in which bubble coalescence takes place
throughout the foam so that at the top of the column r
b
=0.8

Figure 4 Equilibrium volumetric liquid flow rate relative to bubble radi-
us. Data were taken from Stevensons work [31,37,38].
1218 WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12

Figure 5 Relationships between bubble size, liquid fraction, liquid flux
and foam height. Data were derived from Figure 4.
mm. The liquid fraction at any height rather than the top
will be enhanced due to drained interstitial liquid as a con-
sequence of bubble coalescence. This was clearly showed in
Stevensons work [31] and here in Figure 4. Given the axial
bubble size distribution profile, c
0
, c
1
, c
2
and c
out
in Figure 4
can be used to plot the axial liquid holdup profile of the
foam. To predict Q
D
as a function of H
R
, the axial bubble
size distribution profile is necessary. We note that it is not
the intention to present a model for predicting axial bubble
size distribution herein, but to predict Q
D
given a certain
axial bubble size distribution. For simplicitys sake here we
assume that the bubble size grows linearly with the foam
height and the axial liquid holdup profile and Q
D
profile are
plotted in Figure 5. The near exponential decay trend of Q
D

were consistent with the experimental data in Figure 3. The
horizontal dashed line indicated a constant Q
f
and c
out
.
Also plotted in Figure 5 is the equilibrium liquid holdup
with respect to different bubble sizes. It was clearly showed
that the liquid fraction of the lower part of the foam could
be enhanced as the foam height grows. Therefore any ap-
proach trying to predict c
out
by assuming an initial liquid
holdup which is independent on the foam height, like
Maruyama et al.s work [38], will bring unacceptable errors.
2.5 Predicting the effect of foam riser II on c
out

It could be concluded from the above experimental results
and theoretical analysis that the bubble size, especially the
bubble size at the top of the foam layer, was the major de-
terminant of Q
f
and c
out
. The enrichment of the target pro-
tein could be enhanced by reducing c
out
. By applying foam
riser II, c
out
was effectively reduced as shown in Figure 4.
The effect of foam riser II can also be explained by Steven-
sons theory.
Reconsider the example used above. Due to unprompted
bubble coalescence, r
b
grows from 0.5 mm at the bottom to
0.6 mm, 5 cm above the foam/liquid interface. However, the
liquid holdup of the foam at this height is greater than the
equilibrium value of a foam constituted of 0.6 mm bubbles.
Simply inserting foam riser I at this height can change the
liquid holdup profile of the foam beneath it but cannot af-
fect the foam above it, therefore it has no significant effects
on c
out
. However, due to the small openings on foam riser II,
bubble coalescence can be prompted. Assuming that 5% of
bubbles burst after passing through the foam riser, the av-
erage bubble radius will be enlarged to 0.7 mm just above
rubber piston. Consequently, the bubble size at the top of
the foam layer will be greater than if there is no such a foam
riser, leading to a reduced Q
f
and c
out
. If the linear bubble
size distribution is still adopted, the bubble radius at H
R
=15
cm will be 0.9 mm. And since the dryer the foam is the
more unstable the bubbles will be, inserting of foam riser II
at higher locations of the foam may lead to more serious
bubble coalescence and an even smaller c
out
, which was
verified by the experimental results in Figure 4.
It must be pointed out that burst of bubbles may have a
negative effect on mass recovery of the target substance.
Therefore for the best performance of the foam riser, the
structure of it and the position where it is placed need to be
optimized. The linear axial bubble size distribution profile
was arbitrarily adopted in the example and had no physical
basis.
2.6 Effect of foam riser II on nisin activity
In order to examine the effects of our newly developed de-
vice on the denaturation of protein activity, a comparison
between bare column and drainage-enhanced column under
the same operating conditions (as in Table 1, H
R
=15 cm)
was implemented. The experimental results were shown in
Table 2. It was found that the recovery rate of the foam frac-
tionator with drainage-enhancing device II was slightly de-
creased to less than 95% of that of the bare column; however,
the enrichment ratio was significantly increased from 4.25 to
10.3, up to 2.42 times as long as the bare column. As our
previous analysis, this was due to the changes of the foam
structure, which reduced the entrainment capacity of the
foam when passing through the device and thus the liquid
holdup reduced as well as the enrichment ratio rose. At the
same time, we can see only a small increase, from 9.02 to
10.6, in nisin inactivation rate, which indicated that this
drainage-enhancing device indeed had mild effect on the
nisin antibacterial activity. For our column, the maximum
nisin activity can reach 39.6 IU/L. Currently, there have
been no reports related to nisin denaturation in foam frac-
tionation. Hence our data provide initial evidence that the
foam column with this drainage-enhancing device is appli-
cable for recovery and concentration of biological products.
WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12 1219
Table 2 Comparison of experimental results between bare column and drainage-enhanced column

Initial solution Foamate Residual solution
Enrichment
ratio
b)
(-)
Recovery
rate
c)
(%)
Residual
rate
d)
(%)
Inactivation
rate
e)
(%) Volume
(L)
Nisin
activity
a)
(IU/L)
Volume
(L)
Nisin
activity
a)
(IU/L)
Volume
(L)
Nisin
activity
a)

(IU/L)
Bare column 1.96 3.85 0.375 16.4 1.59 0.450 4.25 81.5 9.48 9.02
Column with
device II
1.96 3.85 0.147 39.6 1.82 0.507 10.3 77.2 12.2 10.6
Comparison 0.395 2.42 1.14 1.12 2.42 0.947 1.29 1.18
a) The errors in measuring nisin activity of the initial solution and foamate were 5% and 10%, respectively; b) Enrichment ratio=Nisin activity of
foamate/Nisin activity of the initial solution; c) Recovery rate=Nisin quantity of foamate/Nisin quantity of the initial solution100%; d) Residual rate=Nisin
quantity of the residual solution/Nisin quantity of the initial solution100%; e) Inactivation rate=1recovery rateresidual rate.

3 Conclusions
A device constituted with a rubber piston and a foam riser
was designed to enhance the drainage of the foam in a foam
fractionation process. Preliminary results on continuous
foam fractionation of nisin-containing fermentation broth
suggested that this device was capable of enhancing foam
drainage. It was evident that the device changed the overall
liquid holdup properties of the foam phase and only a small
rise in the overall loss of nisin activity indicated the ap-
plicability of this method in the recovery and the concentra-
tion of biological products. Furthermore, theoretical analy-
sis by invoking recent work on pneumatic foam demon-
strated that the main mechanism of this device was its en-
larging effect on the bubble. Therefore in future studies
multiple of this device can be inserted into the foam column
to form a multistage draining foam fractionator and we ex-
pect that this novel equipment will greatly enhance foam
drainage and enrich the target molecule though more de-
tailed design is necessary, e.g. the length of foam riser, the
diameter of the pore.
This work was supported by Natural Science Foundation of Tianjin (Grant
No. 08JCZDJC25200) and Natural Science Research Program of Hebei
Province (Grant No. Z2008310). The authors express their great thanks to
Li Xueliang for proposing the concept of the foam riser and his contribu-
tion to the experiments as a former postgraduate student in Hebei Univer-
sity of Technology.
1 Lemlich R. Adsorptive bubble separation methods. Ind Eng Chem,
1968, 60: 1629
2 Wungrattanasopon P, Scamehorn J F, Chavedej S, et al. Use of foam
flotation to remove tert-Butylphenol from water. Sep Sci Technol,
1996, 31: 15231540
3 Mathews A, Bishnoi P R, Svrcek W Y. Treatment of oil contaminat-
ed waste waters by foam fractionation. Water Res, 1979, 13: 385391
4 Maruyama H, Seki H, Suzuki A, et al. Batch foam separation of a
soluble protein. Water Res, 2007, 41: 710718
5 Aksay S, Mazza G. Optimization of protein recovery by foam separa-
tion using response surface methodology. J Food Eng, 2007, 79:
598606
6 Linke D, Zorn H, Gerken B, et al. Laccase isolation by foam frac-
tionationNew prospects of an old process. Enzyme Microb Technol,
2007, 40: 273277
7 Gerken B M, Nicolai A, Linke D, et al. Effective enrichment and re-
covery of laccase C using continuous foam fractionation. Sep Purif
Technol, 2006, 49: 291294
8 Lockwood C E, Bummer P M, Jay M. Purification of proteins using
foam fractionation. Pharm Res, 1997, 14: 15111515
9 Kruglyakov P M, Karakashev S I, Nguyen A V, et al. Foam drainage.
Curr Opin Colloid Interface Sci, 2008, 13: 163170
10 Prakash A, Briens C L. Porous gas distributors in bubble columns.
Effect of liquid presence on distributor pressure drop. Effect of
start-up procedure on distributor performance. Can J Chem Eng,
1990, 68: 204210
11 Andou S, Yamagiwa K, Ohkawa A. Effect of gas sparger type on op-
erational characteristics of a bubble column under mechanical foam
control. J Chem Technol Biotechnol, 1999, 66: 6571
12 Boonyasuwat S, Chavadej S, Malakul P, et al. Anionic and cationic
surfactant recovery from water using a multistage foam fractionator.
Chem Eng J, 2003, 93: 241252
13 Bando Y, Kuze T, Sugimoto T, et al. Development of bubble column
for foam separation. Korean J Chem Eng, 2000, 17: 597599
14 Ito Y. Method for continuous countercurrent foam separation. 1986
15 Darton R C, Supino S, Sweeting K J. Development of a multistaged
foam fractionation column. Chem Eng Process, 2003, 43: 477482
16 Leonard R A, Blacykl J D. Multistage bubble fractionator. Ind Eng
Chem, 1978, 17: 358361
17 Morgan G. Single and multistage foam fractionation of rinse water
with alkyl ethoxylate surfactants. Sep Sci Technol, 2001, 36:
22472263
18 Kruglyakov P M, Khaskova T N. Adsorption accumulation of pro-
teins and dyes in foams of solutions and waste water. Colloids Surf
Physicochem Eng Aspects, 2005, 263: 400404
19 Hidenori T, Rie H, Kazuaki Y, et al. Effect of perforated plate on
concentration of poly(vinyl alcohol) by foam fractionation with ex-
ternal reflux. J Chem Eng Jpn, 2003, 36: 11071110
20 Linkea D, Zorna H, Gerkenb B, et al. Foam fractionation of
exo-lipases from a growing fungus (pleurotus sapidus). Lipids, 2005,
40: 323327
21 Maruyama H, Seki H, Suzuki A, et al. Variation of saturated surface
density of ovalbumin on bubble surface in continuous foam separa-
tion. J Colloid Interface Sci, 2006, 299: 416420
22 Koehler S A, Hilgenfeldt S, Stone H A. A generalized view of foam
drainage: Experiment and theory. Langmuir, 2000, 16: 63276341
23 Sun Q C, Huang J, Wang G. Foam drainage wave coalescing and its
energy evolution. Chinese Sci Bull, 2008, 53: 31383141
24 Sun Q C, Ge W, Huang J. Influence of gravity on narrow input forced
drainage in 2D liquid foams. Chinese Sci Bull, 2007, 52: 423427
25 Bhakta A, Ruckenstein E. Drainage of a standing foam. Langmuir,
1995, 11: 14861492
26 Carolissen-Mackay V, Arendse G, Hastings J W. Purification of bac-
teriocins of lactic acid bacteria: Problems and pointers. Int J Food
Microbiol, 1997, 34: 116
27 Arnaud S J. Physical chemistry in foam drainage and coarsening. J
Mater Chem, 2006, 2: 836849
28 Clarkson J R, Cui Z F, Darton R C. Protein denaturation in foam I.
1220 WU ZhaoLiang, et al. Chinese Sci Bull April (2010) Vol.55 No.12
Mechanism study. J Colloid Interface Sci, 1999, 215: 323332
29 Clarkson J R, Cui Z F, Darton R C. Protein denaturation in foam II.
Surface activity and conformational change. J Colloid Interface Sci,
1999, 215: 333338
30 Liu Z, Liu Z, Wang D, et al. On the denaturation of enzymes in the
process of foam fractionation. Bioseparation, 1998, 7: 167174
31 Stevenson P. Hydrodynamic theory of rising foam. Miner Eng, 2007,
20: 282289
32 Wu Z, Hu G, Yu G. Isolation of Nisin preparation method. PRC, CN
1743339A, 200638
33 Wu Z, Wang L, Jing Y, et al. Variable volume fed-batch fermenta
tion for nisin production by Lactococcus lactis subsp. lactis W28.
Appl Biochem Biotechnol, 2009, 152: 372382
34 Neely B C, Eiamwat J, Du L, et al. Modeling a batch foam fractiona-
tion process. Bio Bratislava, 2001, 56: 583589
35 Webb S D, Page R C, Jay M, et al. Characterization and validation of
the gamma-scintigrqaphic method for determining liquid holdup in
foam. Appl Radiat Isot, 2002, 57: 243255
36 Stevenson P, Mantle M D, Hicks J M. NMRI studies of the free
drainage of egg while and meringue mixture froths. Food Hydrocoll,
2007, 21: 221229
37 Stevenson P, Stevanov C, Jameson G J. Liquid overflow from a
column of rising aqueous froth. Miner Eng, 2003, 16: 10451053
38 Maruyama H, Suzuki A, Seki H, et al. Enrichment in axial direction
of aqueous foam in continuous foam separation. Biochem Eng J,
2006, 30: 253259

You might also like