Biochemical Engineering Journal 42 (2008) 2027

Contents lists available at ScienceDirect

Biochemical Engineering Journal
j our nal homepage: www. el sevi er . com/ l ocat e/ bej
Cellular structure in an N-acetyl-chitosan membrane regulate water permeability
Tomoki Takahashi
1
, Masanao Imai

, Isao Suzuki
Bioresource Utilization Sciences, Graduate School of Bioresource Sciences, Nihon University, 1866 Kameino, Fujisawa, Kanagawa Prefecture 252-8510, Japan
a r t i c l e i n f o
Article history:
Received 15 October 2007
Received in revised form 30 April 2008
Accepted 21 May 2008
Keywords:
N-Acetyl-chitosan
Membrane
Deacetylation degree
Cellular structure
Water permeability
Novel permeation mechanism
a b s t r a c t
A novel model of the water permeation mechanism in an N-acetyl-chitosan membrane with a cellular
structure is proposed. Although the entire membrane structure has a hydrophilic character, the cellular
structure incorporates junction zones that practically prevent water permeation. Chitosan membranes
withacontrolleddegreeof deacetylation(DD) werepreparedusingacastingmethod. Changes inthewater
ux and total water content of the membrane were observed with a change in DD. The membrane proper-
ties were analyzed and evaluated using water permeability measurements, scanning electron microscopy
(SEM), X-ray diffraction (XRD), and differential scanning calorimetry (DSC). SEM observations indicated
that the membrane structure was an individual cellular structure and that this cellular structure grew
with decreasing DD. XRDmeasurements indicated the crystal structure of the membrane was amorphous
regardless of the DD in the experimental range. The free water content (W
f
), the freezable bound water
(W
fb
), andthe boundwater not able tofreeze (W
b
) were evaluatedby DSC. The free water mainly contained
inside the cellular structure, and resulted in swelling the chitosan membrane. Water ux was measured
using ultraltration apparatus; it was dependent on the operational pressure, membrane thickness, and
the feed solution viscosity, and obeyed the HagenPoiseuille ow. At a higher DD, water permeation
proceeds due to degradation of the cellular structure; the amount of water in permeation channels was
greater than that for lower DDmembranes even though the total water content in the membrane was less.
The water ux of the chitosan membrane was determined by the water content constructing channels
through the membrane and not on the total water content in the membrane.
2008 Elsevier B.V. All rights reserved.
1. Introduction
Membrane fouling, especially biofouling [1], negatively inu-
ences the permeationux andthe performance of some membrane
properties (e.g., reduced salt rejection and elevated operational
pressure) [25]. Chitosan is known as a biopolymer material (Fig. 1)
that inhibits the growth of mold and bacteria [6], and is expected
to provide long-life hygienic membrane processing.
Control of water ux is very important for separation processes
using a chitosan membrane. Therefore, a detailed investigation of
the inuence of the structure and water content of the membrane
on the water ux is necessary for desirable control of water
permeation.
Abbreviations: CPS, counts per second; DD, deacetylation degree; DSC, dif-
ferential scanning calorimetry; MW, molecular weight; SEM, scanning electron
microscope; XRD, X-ray diffraction.

Corresponding author. Tel.: +81 466 84 3978; fax: +81 466 84 3978.
E-mail address: XLT05104@nifty.ne.jp (M. Imai).
1
Present address: Department of Industrial Chemistry, Faculty of Engineering
Division 1, Tokyo University of Science, 1-3 Kagurazaka, Shinjuku-ku,
Tokyo 162-8601, Japan.
The degree of deacetylation(DD) of a chitosanmembrane is stoi-
chiometrically controlled to a designed level based on the observed
linear relationship between the DDand the amount of acetic anhy-
dride additive. Although a chitosan membrane produced under
lower DD conditions has more hydrophilic character and becomes
apparently swollen with water, the permeated water ux through
such membranes is dramatically reduced [7]. This observed trend
didnot agreewiththegeneral knownbehavior of hydrophilic mem-
branes.
This paper demonstrates the preparation of a chitosan mem-
brane by a casting method in combination with N-acetylation to
control molecular properties (i.e., DD). The effects of the DD of chi-
tosan on permeability and hydrophilic character are discussed and
a novel model of the water permeation mechanism in an N-acetyl-
chitosan membrane with a cellular structure is proposed.
2. Materials and methods
2.1. Preparation of an N-acetyl-chitosan membrane
6g of chitosan (low molecular weight, SigmaAldrich) and 3g
of polyethylene glycol (MW: 7500, Wako) were dissolved in 111g
of 10 (v/v)% acetic acid. The chitosan solution was diluted to 2
1369-703X/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.bej.2008.05.013
T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027 21
Nomenclature
A additive volume of acetic anhydride (l (50g chi-
tosan solution of 2wt%)
-1
)
A
m
apparent surface area of the membrane (m
2
)
A
p
total cross-sectional area of pore of the membrane
(m
2
)
DD deacetylation degree (%) previously dened in our
paper [7]
D
p
diameter of pore in membrane (m)
J
v
volumetric water ux (m
3
m
2
s
1
)
L
m
thickness of membrane (m)
L
p
length of pore membrane (m)
n total number of pore
N
p
numerical density of pores (m
2
)
Q
endo
endothermic heat per unit mass of swollen mem-
brane (J (g of w
e
)
1
)
Q
f
heat of fusion of ice; (333.9J (g of water)
1
)
R
t
volumetric ratio of total contained water of mem-
brane
R
1
volumetric ratio of contained water in the channels
R
2
volumetric ratio of immobilized water content
inside cells
u ow rate of water in pore D
p
(ms
1
)
w
b
mass of bound water (g)
w
d
mass of dried membrane (g)
w
e
mass of swollen membrane at equilibrium (= w
t
+
w
d
) (g)
w
f
mass of free water (g)
w
fb
mass of freezable bound water (g)
w
t
total mass of contained water inmembrane (= w
f
+
w
fb
+w
b
) (g)
W
b
bound water content ratio dened as w
b
w
1
d
(gg
1
)
W
f
free water content ratio dened as w
f
w
1
d
(gg
1
)
W
fb
freezable bound water content ratio dened as
w
fb
w
1
d
(gg
1
)
W
t
total water content ratio dened as w
t
w
1
d
(= W
f
+
W
fb
+W
b
) (gg
1
)
Greek symbols
P operational pressure (Pa)
angle of X-ray diffraction (degree)

L
viscosity of feed solution (Pa s)
(w/w)% with methanol and then vacuum-ltered. Desired amount
of acetic anhydride (97.0wt%, Wako) was added to the chitosan
solution after vacuum ltration. A desired amount of cast solu-
tion (515g) was introduced into a petri dish with a diameter
of 75mm and then dried for 12h at 333K, and subsequently
gelled by immersion into a sufcient amount of 4% NaOH solu-
tion. The resultant membrane was washed with distilled water,
Fig. 1. Chemical structureof chitosanandN-acetyl-chitosan. Chitosanwithanadded
acetyl group as the acyl group becomes N-acetyl-chitosan, and exhibits properties
approximately equal to those of chitin.
Fig. 2. Control of the DD in a chitosan membrane using a casting method in com-
bination with N-acetylation (at 298K). This gure was previously reported by the
authors [7].
and a membrane was nally formed by placing it into hot water
(353K).
Colloidal titration is a method for measuring free amino groups
in a chitosan solution, and the DD indicates molar percentages of
glucosamine to the sum of glucosamine and acetylglucosamine.
The DD of the chitosan membranes produced above were analyzed
using the colloidal titration method, with potassium polyvinyl sul-
fate solution (2.510
3
N) as the titrant and toluidine blue as the
indicator. The terminal point of titration was clearly conrmed by
the change of color from blue to claret.
The relationship between the quantity of acetic anhydride (A)
usedinthe preparationandthe DDpercentage of resultant chitosan
(Fig. 2) [7] is given by
DD = 0.1045A +92.22 (1)
In the preparation of a membrane by the casting method, a DD
range of over 50% is preferable to avoid solution gelation, even if
the relationship is effective over the range of 0100%. N-Acetyl-
chitosans of DD71.3%, 76.5%, 81.8%, 87.0%, and 92.2% were prepared
for analysis.
2.2. Scanning electron microscopy (SEM)
The morphology of the chitosan membranes was observed
using a scanning electron microscope (S-3500N, Hitachi High-
Technologies Corporation, Tokyo) operated at an accelerating
voltage of 10kV. As a pretreatment for clear observationof the inner
structure, the chitosan membrane was vacuum freeze-dried. The
dried membrane samples were then coated with zinc.
2.3. X-ray diffraction
X-raydiffraction(XRD) patterns wererecordedonaRigakuRAD-
C diffractometer (monochromatic Cu K radiation) with the X-ray
cathode operated at 20mA and 30kV. The samples were supported
on a glass frame and were determined at 2 angles from10

to 50

.
2.4. Water content
The total mass of water contained in the membrane (w
t
) was
dened to be the sumof three categories of water free water (w
f
),
freezable bound water (w
fb
), and bound water not able to freeze
(w
b
) as shown in the following equation:
w
t
= w
f
+w
fb
+w
b
(2)
22 T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027
The total water content ratio (W
t
) was composed of the free water
content ratio (W
f
), the freezable bound water content ratio (W
fb
),
and the bound (but not able to freeze) water content ratio (W
b
).
Each water content ratio was dened to be the ratio of water mass
per unit mass of the dried membrane (w
d
), as presented by the
following equation:
W
t
W
f
+W
fb
+W
b
=
w
f
+w
fb
+w
b
w
d
(3)
2.4.1. Measurement of total water content
Gravimetric methods were used to determine W
t
. The prepared
membranes had already achieved the equilibrium water content.
Excess water attachedonthemembranesurfacewas removedusing
a lter paper. The mass of the swollen samples (w
e
) was then mea-
sured, before the wetted membrane was dried at 353K for 24h,
and the mass of dried membrane (w
d
) was measured. Experimen-
tal mean values of w
e
and w
d
were determined by repeating the
measurements at least thrice for satisfactory reproducibility and
reliability.
The difference between w
e
and w
d
indicates the mass of total
contained water (w
t
):
w
t
= w
e
w
d
(4)
The total water content ratio (W
t
) was then calculated using the
following equation:
W
t
=
w
e
w
d
w
d
=
w
t
w
d
(5)
2.4.2. Measurement of free water and freezing bound water
Using differential scanning calorimetry (DSC) [8], W
f
and W
fb
for the N-acetyl-chitosan membrane were determined. The DSC
measurement was conducted using a calorimeter (TAS300, Rigaku,
Tokyo) under a N
2
ow. The sealed sample pan in the DSC cham-
ber was quickly frozento 243Kusing liquidN
2
, andseveral minutes
were allowed for the systemto reach thermostatic equilibrium. The
sample holder assembly was then heated at a rate of 5Kmin
1
.
Each calorimetric scan was measured with three repetitions for
reproducibility and reliability, and the mean value was employed.
An endothermic peak at 273K is regarded as the sum of the
fusion of free water (w
f
) and the freezable bound water (w
fb
), since
phase transition phenomena for chitin/chitosan are not reported
Fig. 3. Apparatus for water permeation experiment. (a) Ultraltration apparatus
UHP-62K (Advantec), (b) feed solution, (c) stirring bar, (d) sample membrane, (e)
magnetic stirrer, (f) regulator valve, (g) N
2
gas, (h) manometer, (i) electric balance,
and (j) permeated solution.
at this temperature. Some bound water (w
b
) is not frozen at 243K,
and is considered to have no effect in this thermometry.
The DSC measurement (Q
endo
) therefore indicates the endother-
mic heat per unit mass of swollen membrane (w
e
). Consequently,
the net endothermic heat per total mass of contained water in
membrane is given by Q
endo
(w
e
/w
t
). The mass ratio of the sum of
free water and freezable bound water to the total contained water
in membrane can be obtained by dividing Q
endo
(w
e
/w
t
) by the heat
of fusion of ice, Q
f
(333.9J (g of (w
f
+w
fb
))
1
) [8,9]:
Q
endo
(w
e
/w
t
)
Q
f
=
w
f
+w
fb
w
t
= (W
f
+W
fb
)
w
d
w
t
(6)
Thus, the sum of W
f
and W
fb
can be evaluated:
Q
endo
(w
e
/w
d
)
Q
f
= W
f
+W
fb
(7)
2.5. Permeability
The water permeability of the N-acetyl-chitosan membrane was
determined from the water mass ux throughput using ultral-
tration apparatus (Fig. 3, UHP-62K, Advantec, Tokyo). The initial
Fig. 4. SEM micrographs of N-acetyl-chitosan membrane cross-sections. (a) and (b) DD 71.3%; (c) and (d) DD 81.8%; (e) and (f) DD 92.2%. The full length of reference scaling
measure in (a), (c) and (e) indicates 10m. The full length of reference scaling measure in (b), (d) and (f) indicated 5m.
T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027 23
Fig. 5. X-ray spectra of N-acetyl-chitosanmembranes withdifferent DD. (a) Original
purchased reagent of chitin (DD 1.1%), (b) original purchased reagent of chitosan
(DD 92.2%), (c) DD 71.3%, (d) DD 81.8%, (e) DD 87.0%, and (f) DD 92.2%. The spectra
are arbitrarily displaced to clarify the diffraction patterns. The intensity number
indicated on the ordinate was temporary, not corresponded to each illustrated data.
volume of water was 100ml, and the apparatus was stirred con-
stantly. The operational pressure was adapted and maintained by
introducing N
2
gas at room temperature (298K). The mass of per-
meated water throughput was accurately measured based on the
indication of an electric balance (Mettler PJ3600, Z urich), and was
converted to a volumetric water ux by recalculation using the
density of the permeated water (998kgm
3
).
3. Results and discussion
3.1. Structure of the N-acetyl-chitosan membrane
Fig. 4 presents SEM micrographs of cross-sections for three of
the N-acetyl-chitosan membranes (DD of 71.3%, 81.8% and 92.2%).
The membranes were structured with individual cells separated by
cellular walls. Assuming the cellular structure to be the key fac-
tor in determining water permeability, the mean inside diameter
of the cells was 1.7m for a DD of 92.2%, 2.6m for DD 81.8%,
and 4.7m for DD 71.3%. In other words, as the DD decreased the
size of the cells increased, and the thickness of the cellular walls
decreased.
The mechanismof the cellular structure formationwas the same
as that reported for chitin gel by Hirano et al. [1013]. Molecules of
Fig. 6. Change of water composition in an N-acetyl-chitosan membrane with DD.
() Wt, () W
f
+W
fb
, and () W
b
.
N-acetyl-chitosan formed partial junction zones or regions where
molecular strands gatheredclosely together, due to the intermolec-
ular association of acetamide groups with the N-acetylation of
chitosan, leading to the formation of cellular walls or a honeycomb.
Hirano et al. [1013] have described the structure of chitin gel (DD
0%), but not the relationship between the cellular structure and the
DD. The junction zone and honeycomb structure were formed by a
conformational conversion often observed in polysaccharides such
as carrageenan gel [14].
3.2. Crystallinity of the N-acetyl-chitosan membrane
Fig. 5 presents the XRDpatterns of four N-acetyl-chitosan mem-
branes (DD of 71.3%, 81.8%, 87.0%, and 92.2%) and the purchased
reagents (chitin (DD 1.1%), chitosan (DD 92.2%)). Specic peaks
of chitin/chitosan were observed at diffraction angles of 19

and
22

, respectively, for all samples. In addition, broad intensity in

the range from 10

to 20

indicated that the crystallinity of the

membrane was amorphous. It is generally known that the crys-
tallinity of chitin is higher than that of chitosan. However, in
these membrane samples, no dominant intensity was observed,
and the microcrystal domain did not alter with change in the
DD.
Fig. 7. Effect of operational pressure on water ux (298K, L =0.901mPa s). () DD 92.2%, () DD 87.0%, () DD 81.8%, and () DD 76.5%. The left of the gure represents a
convenient, expanded view of the lower level water ux for DD 87.0%, DD 81.8% and DD 76.5%.
24 T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027
Fig. 8. SEM micrographs of the N-acetyl-chitosan membrane (DD 76.5%). (a) Cross-section damaged due to the operational pressure during the permeation experiment. The
full length of reference scaling measure indicates 10m. (b) Detailed picture of damaged cross-section. The full length of reference scaling measure indicated 5m.
3.3. Swelling of the N-acetyl-chitosan membrane
Fig. 6 presents the water composition of the prepared N-acetyl-
chitosanmembranes. Thetotal water content ratio(W
t
) was highest
for the chitosan with lowest DD, and then dramatically decreased
with increasing DD. The sumof the free water and freezable bound
water content ratios (W
f
+W
fb
) also decreased with increasing DD.
This sumdominatedthe total water content ratio for chitosans with
lower DDlevels but this dominance also decreased with increasing
DD. Consequently, the bound water content ratio (W
b
) gradually
increased with increasing DD.
The cellular structure for chitosans of lower DD (Fig. 4(a and b))
is remarkable, particularly in regard to cellular size. Free water and
freezing bound water are thus mainly contained inside the cellular
structure, resulting in a swelling of the chitosan membrane.
3.4. Permeability of the N-acetyl-chitosan membrane
3.4.1. Effect of operational pressure
Fig. 7 illustrates the effect of operational pressure on water ux
through the membranes. The water ux increased linearly at pres-
sures less than 100kPa. In contrast, at pressure over 100kPa, the
ux was reducedfor all membranes except that of highest DD. Fig. 8
presents SEM photographs of one of the N-acetyl-chitosan mem-
branes (DD 76.5%) used in the permeation experiment at pressures
over 100kPa. As canbe clearly seen, the cellular structure was dam-
aged, and a layer-like structure was formed by compaction, thereby
reducing water permeation.
Fig. 9 depicts the effect of the DD on water ux and water con-
tent. The water ux was calculated based on 50kPa, thus avoiding
the impact of membrane compaction. The water ux increased
exponentially with the increase in the DD ranging from 80% to
Fig. 9. Change in water ux (50kPa, 298K, L =0.901mPa s) and total water content
of membrane with regulated DD. () Water ux, Jv and () total water content, Wt.
90%. In contrast, the water content decreased in the same DD
region. This result is opposite to the general trend of water perme-
ation in hydrophilic membranes. As previously seen for cellulose
membranes, a higher water ux usually results from higher water
content in the membranes.
3.4.2. Effect of membrane thickness and viscosity of the
permeating solution
Fig. 10 presents the effect of the permeating solutions viscosity
on the water ux of the N-acetyl-chitosan membrane. An aqueous
glucose solution was employed as a feed solution and its viscosity
was changed by altering the glucose concentration. Glucose is rec-
ognized as an inert solute with respect to the membrane because it
is non-electrolyte character and hydrophilic lowmolecular weight.
Shrinking and/or swelling of membrane by glucose did not appear
beforehand.
The ux of aqueous glucose solution was regarded as essen-
tially equivalent to the ux of water with an inert solute. By
incorporating the density of the respective aqueous glucose solu-
tions into the calculations, the resulting volumetric water uxes
increased linearly with increase in the reciprocal of the viscosity
(
1
L
).
Fig. 11 presents the effect of membrane thickness on the water
ux of the N-acetyl-chitosan membrane. Increasing the reciprocal
of the thickness (L
1
m
) linearly increased the water ux. The thick-
ness of the membrane was determined only as a result of altering
the volume of cast solution placed in the petri dish before drying
(described in Section 2.1); the other membrane preparing condi-
tions, the drying temperature, the drying time and the washing
conditionwere constant throughout the course of testing the mem-
branes.
Fig. 10. Effect of the feed solution viscosity on the water ux (100kPa, 298K, DD
92.2%). The ordinate Jv in this gure indicates the ux of an aqueous glucose solution
recalculated for use as the ux of water containing an inert solute.
T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027 25
Fig. 11. Effect of membrane thickness on the water ux (50kPa, 298K, L =
0.901mPa s, DD 92.2%).
3.4.3. Water permeation model and comparison with observed
behavior of water ux of chitosan membrane
Typical ow rate of water, u (ms
1
), in a single cylindrical pore
is determined using the HagenPoiseuille equation:
u =
PD
2
p
32
L
L
p
(8)
where P is the operational pressure (Pa), D
p
is the mean pore
diameter (m), L
p
is the pore length (m), and
L
is the viscosity of
the feed solution (Pa s). This can be adapted to obtain a volumetric
water ux, J
v
(m
3
m
2
s
1
):
J
V
= u
A
p
A
m
=
PD
2
p
32
L
L
p
A
p
A
m
(9)
where A
m
is the apparent surface area (m
2
) of the membrane and
A
p
is the total cross-sectional area (m
2
) of pores in the membrane.
A
p
= n
D
2
p
4
(10)
N
p
=
n
A
m
(11)
Substituting Eqs. (10) and (11) into Eq. (9), the volumetric water
ux can be expressed in terms of the number of pores per unit area
of the membrane, N
p
:
J
V
=
PD
4
p
N
p
128
L
L
p
(12)
Eq. (12) is conventionallyemployedtoexaminethewater uxbased
on the number of pores per unit area, N
p
. However, accurate, prac-
tical determination of the number of pores is very difcult. In this
study, a parameter R
t
representing the volumetric ratio of total con-
tained water in the membrane was introduced instead of the ratio
of A
p
/A
m
.
R
t
=
A
p
L
m
A
m
L
m
(13)
The numerator of R
t
corresponds to the total pore volume of the
membrane, andis equivalent tothe total volume of containedwater
in the pores at equilibrium. The denominator of R
t
is the entire
volume of wetted membrane at equilibrium. Substituting Eq. (13)
into Eq. (9) therefore gives the volumetric water ux in terms of the
volumetric ratio of total contained water in the membrane:
J
V
=
PD
2
p
R
t
32
L
L
p
(14)
R
t
W
t
(15)
Fig. 12. Schematic illustration of water classes in the membrane.
This suggests that J
V
is linearly proportional to R
t
. In a practical
sense, R
t
corresponds to W
t
as a rst approximation.
As a further complication, the pore length (L
p
) is not yet accu-
rately known and its determination, involving the tortuosity of the
pore, is very difcult. It is assumed here for simplicity that L
p
is
linearly proportional to the membrane thickness, L
m
.
L
p
L
m
(16)
Substituting these additional approximations into Eq. (14), the vol-
umetric water ux becomes
J
V

PD
2
p
W
t
32
L
L
m
(17)
This derivative equation shows the water ux to be proportional
to operational pressure P, the reciprocal of viscosity (
1
L
), and
the reciprocal of membrane thickness (L
1
m
), as a rst approxima-
tion. This agrees well with the observed data of the dependence
of pressure (Fig. 7), viscosity (Fig. 10), and membrane thickness
(Fig. 11).
However, the correlation suggested by Eq. (17) between water
ux J
v
and the water content W
t
is not consistent with the observed
data. As presented in Fig. 9, a higher water content (W
t
) in the chi-
tosan membrane resulted in a lower water ux. Therefore, in order
to understand the permeation mechanism in an N-acetyl-chitosan
membrane with a cellular structure, a further novel consideration
is necessary.
3.5. Mechanism of water permeation in an N-acetyl-chitosan
membrane with a cellular structure
To understand water permeation in a hydrophilic biopolymer
membrane, two classes of water should be considered: water in
the channels, and the immobilized water inside cells. The volume
of total contained water R
t
is thus
R
t
= R
1
+R
2
(18)
where R
1
is volume of water in the channels, and R
2
is the
volume of water immobilized inside the cells. A schematic illus-
tration of this for the N-acetyl-chitosan membranes is presented in
Fig. 12. According to the experimental results, R
t
decreased with an
increase in DD. This is because the total water content W
t
in the
membrane decreased with an increase in DD, as shown in Fig. 6. R
2
should correspond to the experimental data of (W
f
+W
fb
). R
1
is not
simply corresponded to W
b
in our data, the W
b
is minor compo-
nent. The change of W
b
could not explain the dramatical change of
water ux. The detail of R
1
should be investigated further.
For the membranes with a lower DD, the volumetric fraction of
water contained in cells (i.e., R
2
, is very large). As a result, the vol-
umetric fraction of water in the channels (R
1
), or in other words
water permeating through the membrane, is very small. This is a
26 T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027
Fig. 13. Schematic illustration of the water permeation mechanism in an N-acetyl-chitosan membrane with a cellular structure. (a) The structure of high DD chitosan
membrane, (b) the structure of low DD chitosan membrane, and (c) detailed image for low DD chitosan membrane with cellular structure and water channels. The cellular
structure illustrated in (b) is composed of immobilized water, the cellular wall and the junction zone. The structure prevents water ux.
consequence of the larger size of the cellular structure, in combina-
tion with the presence of junction zones, which limits the channels
that could form.
For the membranes with a higher DD, R
2
should be lower
because the size of the cells is smaller, the presence of junction
zones is reduced, and the resultant cellular structure permits a
greater formation of channels. Thus, R
1
should dominate R
t
even
though the total water content in the membrane is less than for
membranes of lower DD. Consequently, there is a greater perme-
ation of water through membranes with a higher DD.
This novel model of the water permeation mechanism in an
N-acetyl-chitosan membrane with a cellular structure is shown
schematically in Fig. 13. A chitosan membrane has a greater ten-
dency to have a cellular structure containing free water with a
decrease in DD. As seen from the enlarged cells in Fig. 4 (a and b),
the cells in the chitosans of lower DD are not conducive to the for-
mation of water channels necessary for permeation. The available
water channels are decreased with the enlargement of the cellu-
lar structure, reducing the possible permeation. Even if the entire
membrane structure has hydrophilic character, the cellular struc-
ture in combination with the junction zones practically prevents
water permeation.
At a higher DD, water permeation proceeds due to degradation
of the cellular structure as a result of fewer junction zones. The
amount of water forcing channels is thus larger than for chitosans
of lower DD. The water ux of the higher DD chitosan membrane
depends on the channel constructing water content (R
1
) not on
total water content (R
t
). Thereforeexperimental relationshipfor the
volumetric water ux, previously approximated in Eq. (8), becomes
J
V

PD
2
p
R
1
32
L
L
m
(19)
The role of R
1
on the water ux will be similarly considered for
water permeation in other hydrophilic biopolymer membranes.
Detailed consideration of D
p
in cellular structured membranes
should be a focus of future investigations.
4. Conclusion
The effects of the DD of a chitosan membrane on structure,
hydrophilic character, and permeability were discussed. The mem-
brane body has an individual cellular structure involving junction
zones, andthe cellular size increases witha decrease inthe DD. Free
water mainly contained inside the cellular structure of chitosans
with a lower DD, and resulted in swelling of the membrane.
Water ux was dependent on the operational pressure, the
membrane thickness, and the viscosity of feed solution, a situa-
tionthat strongly suggests the aptness of the HagenPoiseuille ow
equation. The water ux decreased exponentially with decreas-
ing DD, despite the entire membrane structure having more
hydrophilic character. The contrasting trend of water ux and total
water content of membrane with changing DD is opposite to the
generally known behavior of hydrophilic membranes, e.g., cellu-
lose membrane. It is found that the cellular structure incorporating
the presence of junction zones in N-acetyl-chitosan membranes
prevents ready permeation of water.
The special trend of water ux through the chitosan membrane
depended on the channel constructing water content (R
1
) not on
the total water content (R
t
) in the membrane. Immobilized water
(R
2
), which corresponds to the experimental data of (W
f
+W
fb
), did
not contribute to the formation of water permeation channels, and
so is unrelated to the ux. The concept of the water ux depend-
ing on the hydrophilic cellular structure is similarly applicable
to hydrophilic membranes composed of hydro-gels (i.e., polysac-
charide gels). Future quantitative evaluation of the relationship
T. Takahashi et al. / Biochemical Engineering Journal 42 (2008) 2027 27
between cellular structure and permeability in hydrophilic mem-
branes will be necessary for their application in industrial elds.
Acknowledgements
The authors sincerely thank Professor Katsuto Otake and Dr.
Atsushi Shono of the Tokyo University of Science for useful discus-
sions. Dr. Tao Kei of the Nihon University provided technical advice
for the DSC measurements and SEM operation. Dr. Tsutomu Imai
of the Ueda Lime Co., Ltd. offered references for the XRD measure-
ments.
References
[1] H.-C. Flemming, G. Schaule, Biofouling on membranesa microbiological
approach, Desalination 70 (1988) 95119.
[2] Q. Zhao, Y. Liu, C. Wang, S. Wang, H. M uller-Steinhagen, Effect of surface free
energy on the adhesion of biofouling and crystalline fouling, Chem. Eng. Sci. 60
(2005) 48584865.
[3] A.S. Mohamad, Biofouling prevention in RO polymeric membrane systems,
Desalination 88 (1992) 85105.
[4] A.S. Kim, H. Chen, R. Yuan, EPS biofouling in membrane ltration: an analytic
modeling study, J. Colloid Interface Sci. 303 (2006) 243249.
[5] R. McDonogh, G. Schaule, H.-C. Flemming, The permeability of biofouling layers
on membranes, J. Memb. Sci. 87 (1994) 199217.
[6] Y. Uchida, Antibacterial activity of Chitin and Chitosan, Gekkan Fudo Kemikaru
34 (2) (1988) 2229 (in Japanese).
[7] T. Takahashi, M. Imai, I. Suzuki, Water permeability of chitosan membrane
involved in deacetylation degree control, Biochem. Eng. J. 36 (2007) 4348.
[8] S.J. Kim, S.J. Park, S.I. Kim, Swelling behavior of interpenetrating polymer net-
work hydrogels composed of poly(vinyl alcohol) and chitosan, Reactive Func.
Polym. 55 (2003) 5359.
[9] M.B. Ahmad, M.B. Huglin, DSC studies on states of water in crosslinked
poly(methyl methacrylate-co-N-vinyl-2-pyrrolidone) hydrogels, Polym. Int. 33
(1994) 273277.
[10] S. Hirano, S. Kondo, Y. Ohe, Chitosan gel: a novel polysaccharide gel, Polymer
16 (1975) 622.
[11] S. Hirano, Y. Ohe, Chitosan gels: a novel molecular aggregation of chitosan in
acidic solutions on a facile acylation, Agric. Biol. Chem. 39 (1975) 13371338.
[12] S. Hirano, R. Yamaguchi, N-Acetylchitosan gel: a polyhydrate of chitin, Biopoly-
mers 15 (1976) 16851691.
[13] S. Hirano, R. Yamaguchi, N. Matsuda, Architecture of chitin gel as examined by
scanning electron microscopy, Biopolymers 16 (1977) 19871992.
[14] D.A. Rees, Structure, conformation, andmechanisminthe formationof polysac-
charide gels and networks, Adv. Carbohyd. Chem. Biochem. 24 (1969) 267
332.