Materials Transactions, Vol. 43, No. 5 (2002) pp. 1074 to 1077
Special Issue on Smart Materials-Fundamentals and Applications
c
2002
The Japan Institute of Metals
Pervaporation Membrane System for the Removal of Ammonia from Water
Yasuhiko Hirabayashi
Forestry and Forest Products Research Institute, P. O. Box 16, Tsukuba Norin Kenkyu Danchi-nai, Ibaraki 305-8687, Japan
Regenerated cellulose and chitosan membranes were studied for the pervaporation separation of an aqueous solution of urine component
(ammonia, uric acid or creatinine). The permeation rate of water increased with increase of the temperature of feed solution induced into the
upstream side of membrane module. Uric acid, creatine and creatinine were not found in the permeate through the all membranes investigated.
Selective permeation of water and ammonia depends on membrane. The removal of ammonia through the chitosan membrane was from 57%
to 59%. Adsorption of ammonia from the downstream vapor by silica gels was carried out. And desorption of ammonia from the adsorbents
by heating under the reduced pressure to regenerate the capacity of adsorption was also confirmed. In the case of new pervaporation system,
the combination of pervaporation and adsorption/desorption process, ammonia was almost completely removed, and finally the pure condensed
water was obtained in the cold trap.
(Received December 28, 2001; Accepted February 27, 2002)
Keywords: pervaporation, ammonia, water, urine, chitosan, membrane
1. Introduction
The environmental control and life support system
(ECLSS) in the space station is composed of food, air, water,
and waste management. The water is important to support
human activities in space. The amount of water supply is entirely limited in the space station. Thus the recovery of pure
water from urine will be a required subject. Ammonia is a
main component of urine. Because of the chemical and physical similarity between water and ammonia, the separation of
them is difficult to achieve by microfiltration, ultrafiltration,
dialysis, and reverse osmosis. A deterioration of the performance due to the change of surface property of the distillation
membrane from hydrophobic to hydrophilic might cause the
trouble of direct pass of the urine components to the downstream side.
The author has investigated about the pervaporation membrane method for desalination, and demonstrated that the hydrophilic dense membranes, such as cellulose or chitosan
membranes, showed the good performance for removing salts
from water in one-step operation.1, 2) Therefore it was expected that the pervaporation membrane method would also
be good for establishing the water recycle system of ECLSS.
The author has demonstrated that the pervaporation
through cellulose membranes completely removes nonvolatile compounds from aqueous solution of urine components, such as uric acid, creatine and creatinine, but except for
ammonia.3–7) The mechanism of separation by pervaporation
is considered to be solubility-diffusion. Pervaporation process generally includes three major steps: preferential sorption of a solute or solvent into the membrane on the feed side,
their phase transition from liquid to vapor, and the diffusion
of liquid and vapor through the membrane. This means the
possibility of separation of aqueous ammonia by pervaporation based on the difference of affinity of water and ammonia
to the membrane. However, the separability of aqueous ammonia by pervaporation membrane will be limited to some
degree. Therefore, the author designed a new water purification system based on the combination of pervaporation and
adsorption/desorption process, as shown in Fig. 1. The down-
stream of the pervaporation membrane is a gas phase under
reduced pressure; the vapor mixture goes through the adsorption column before reaching to the cold trap. The adsorbent,
which selectively adsorbs ammonia from the vapor mixture
of water and ammonia, is required. Exposing the adsorbent
to the condition of the high vacuum and heat can regenerate
the performance of adsorption. Thus the adsorbent can be repeatedly used. The purpose of this work is to find a good performance membrane with better selectivity for ammonia and
water, and show the effectiveness of pervaporation followed
adsorption/desorption process for removing ammonia.
2. Experimental Procedure
2.1 Pervaporation apparatus
A schematic diagram of the pervaporation apparatus is
shown in Fig. 2. The feed was continuously circulated from
the feed tank (2) through an upstream compartment of membrane cell (5) by a tube pump (1). Feed was kept at the selected temperature by a thermostated water bath (4). The
permeation cell was insulated (7) to minimize the heat loss
to the atmosphere. In the downstream side the permeated
vapor was withdrawn by a vacuum pump (11). In the conventional pervaporation method, the permeate was directly
introduced to a cold trap (9) cooled by liquid nitrogen. In
the new pervaporation proposed in this study, single adsorption column (8) or double columns in series (or parallel) were
set-up between the permeation cell and the cold trap. Permeate was condensed in a cold trap by liquid nitrogen. After
preliminary pervaporation for 30 or 60 min to obtain a constant flux, the permeation flax was calculated from the permeation weight of the component in a defined period time
and contact area of membrane with feed. Uric acid, creatine
and creatinine in feed and permeate solution were analyzed
by HPLC (Shimadzu LC-6A) equipped with an UV-VIS detector (Shimadzu SPD-6AV) and a column (Shimpack CLCODS (6.0 mm × 15 cm) heated to 40◦ C. Elution; 20 mmol/L
phosphate buffer (pH3.0) with 1.0 mL/min flow rate. Ammonia in feed and permeate solution were determined by an ion
chromatography (TOYOSODA) equipped with a conductive
Pervaporation Membrane System for the Removal of Ammonia from Water
1075
Fig. 1 Combination of pervaporation and adsorption/desorption process in ECLSS.
Fig. 2 Scheme of pervaporation and adsorption apparatus for experiment.
detector (TOYOSODA CD-8000) and a column (Shimpack
IC-C3 (4.6 mm × 10 cm) heated to 40◦ C. Elution; 25 mmol/L
tartaric acid with 1.0 mL/min flow rate.
formed was dipped into 1-N aqueous sodium hydroxide solution, washed with water, dried in air again, peeled off from
the acrylic resin plate.
2.2 Membranes
The film-shaped membranes investigated were cellophane,
regenerated cellulose, and chitosan. The membrane structure
was examined by a scanning electron microscope (SEM). The
regenerated cellulose membrane was obtained by coagulating
the cellulose solution of N -methylmorpholine-N -oxide into
water, washing and dry. The detail method for preparation of
the regenerated cellulose is described in the previous report.4)
The chitosan membrane was formed by casting the 1% chitosan solution containing a dilute acetic acid onto an acrylic
resin plate and dried in air for 24 to 48 h. The membrane
2.3 Preparation of the silica gel column
The adsorbent used was spherical particles silica gel white
(5–10 mesh) purchased from Junsei Chemicals Co. Ltd. 10 g
of the silica gel was packed in an acrylic resin pipe of 12 mm
in inside diameter and 14 cm long. As described in 2.1, single
adsorption column (8) or double columns in series (or parallel) were set-ups between the permeation cell and the cold
trap. The silica gel adsorbed with ammonia was transferred
from the column to vacuum flask, and then the desorption of
ammonia was carried out under vacuum (below 1.33 Pa) and
heating in the range of 180 and 200◦ C until condensed ma-
1076
Y. Hirabayashi
Fig. 3 Cross-sectional SEM photograph of chitosan membrane.
Table 1 Results of pervaporation of aqueous feed solution of uric acid, creatine, creatinine, and ammonia through cellophane, regenerated cellulose
membrane, and chitosan membrane.
Concentration of
urine component in
feed
Uric acid, 0.5 g/L
Creatine, 30 mg/L
Creatinine, 1.0 g/L
Ammonia, 0.1%
Concentration of urine components in
permeate solution
Cellophane
Regenerated
cellulose
membrane
Chitosan
membrane
ND < 0.1 ppm
ND < 1 ppm
ND < 0.1 ppm
0.08–0.1%
ND < 0.1 ppm
ND < 1 ppm
ND < 0.1 ppm
0.08–0.1%
ND < 0.1 ppm
ND < 1 ppm
ND < 0.1 ppm
0.03–0.04%
ND: Non detectable
terial was no longer observed in a replaced cold trap. The
regenerated silica gel was subjected to the adsorption of ammonia to continue the pervaporation operation.
3. Results and Discussion
Figure 3 shows the cross-sectional view of chitosan membrane. Some narrow small cracks or a stripe-like structure
running parallel to the surface of membrane were observed,
but not perpendicular to the surface. Pervaporation results
of the membranes investigated are summarized in Table 1 including the previous results.1–4) Uric acid, creatine and creatinine were not found in the all permeated solution. In accordance with the results of desalination by pervaporation
process in which the dense hydrophilic membrane was used,
these non-volatile urine compounds did not permeate through
the cellulose or chitosan membranes.
Table 1 shows that pervaporation process using hydrophilic
cellulose membrane, the permeability of ammonia and water
is almost equal. This result may indicate that the partition between ammonia and water in the cellulose membrane equals
to that in the feed solution. The affinity between cellulose
membrane and ammonia is considered to be close to that between cellulose and water. The previous report showed that
ammonia preferentially permeates through the hydrophobic
membrane. The vapor pressure of ammonia is 1.166 × 106 Pa
Fig. 4 Dependence of permeation rate through chitosan membrane on the
temperature of feed.
and that of water is 4.243 × 103 Pa at +30◦ C. The partial vapor pressure of ammonia of aqueous ammonia will be higher
than that of water. In the distillation process, the lower is the
boiling point or the higher is the partial pressure, the faster
the compound distils away. The pervaporation rate depends
on the affinity to membrane material and diffusion in membrane. The affinity of ammonia to hydrophobic material will
be stronger than that of water. Taking into account the affinity
to membrane.
On the other hand, chitosan membrane retards the permeation of ammonia as shown in Table 1. More than 50% of
ammonia in feed was rejected by chitosan membrane. Structural unit of chitosan is a derivative of glucose anhydride unit
in which the hydroxyl group at 2 position is replaced with
amino group.
Because of the structural characteristics chitosan will be
positively charged in aqueous media. Ammonia is also of
course positively charged in aqueous solution. Thus ionic repulsive force, which arises between chitosan and ammonia
molecule, impedes the penetration of ammonia into chitosan
membrane.
On the basis of above discussion pervaporation process of
aqueous ammonia through chitosan membrane was investigated in detail.
Figure 4 shows the dependence of permeation rate through
chitosan membrane on the temperature of feed. The permeation rate increases linearly with increasing in temperature.
Table 2 shows the effect of column unit number of adsorbents and the way of connection of the column units on the
removal of ammonia at different temperature. Removal (%)
is defined as, ((concentration of ammonia in initial feed) −
(concentration of ammonia in permeate)) − (concentration of
ammonia in initial feed) × 100.
Removal of ammonia decrease with increasing in temperature of feed solution. The value of removal of ammonia
Pervaporation Membrane System for the Removal of Ammonia from Water
1077
Table 2 Results of pervaporation and pervaporation/adsorption of aqueous
ammonia feed solution through chitosan membrane.
Number and
connection
type of
adsorbent
column
Temperat
ure of
feed (◦ C)
Concentratio
n of ammonia
in feed (%)
0
29.7
35.0
41.4
45.9
50.9
0.989
0.1025
0.0995
0.0999
0.0965
0.0405
0.0442
0.0432
0.0423
0.0403
59.05
56.88
56.58
57.66
58.24
1
34.6
39.7
44.6
51.2
57.1
64.1
0.1310
0.1014
0.0086
0.1044
0.1306
0.1254
0.00015
0.00002
0.00007
0.00024
0.00011
0.00048
99.89
99.98
99.19
99.17
99.92
99.62
2
(Series)
43.9
50.2
55.0
60.1
63.8
68.6
0.1473
0.1489
0.1483
0.1506
0.1459
0.1479
0.000067
0.000697
0.000038
0.000036
0.000052
0.000200
99.95
99.53
99.97
99.96
99.96
99.86
43.9
49.0
57.1
60.6
65.7
71.4
0.1379
0.1403
0.13227
0.1332
0.1311
0.1279
0.000021
0.000038
0.0000097
0.000068
0.000043
0.00022
99.99
99.97
99.99
99.95
99.97
99.83
2
(Parallel)
Concentration
of ammonia in
permeate (%)
Removal of
ammonia
(%)
ranged from 78.2% at 41.2 to 39.9% at 56.5◦ C. On the viewpoint of both safety and selectivity, pervaporation operation
at mild temperature will be preferable. In any case complete
removal of ammonia is difficult by only one step operation of
pervaporation.
Therefore the author thought out the combination of pervaporation and adsorption process as indicated in Figs. 1 and 2.
Silica gel was used as an adsorbent for ammonia, because silica gel can withstand high temperature and high vacuum, and
it would release the adsorbed compounds under these conditions to recover the adsorptive activity. Figure 5 indicates that
the way of connection of column units affects the dependence
of the desorption of ammonia from the adsorbents by heating under the reduced pressure to regenerate the capacity of
adsorption was also confirmed. In the case of new pervaporation system, the combination of pervaporation and adsorption/desorption process, ammonia was almost completely removed, and finally the pure condensed water was obtained in
the cold trap. It has been cleared that non-volatile main urine
Fig. 5 Effect of the way of connection of adsorption column units on the
relationship between the permeation rate and temperature of feed.
compounds such as uric acid, creatine and creatinine can be
completely removed by the pervaporation membrane method
using the dense hydrophilic membrane. Therefore, the pervaporation membrane method is promising the very useful system to extract pure water from urine.
Acknowledgements
This study is carried out as a part of “Ground Research
Announcement for Space Utilization” promoted by Japan
Space Forum. The authors thank Miss Toshiko Kosa for her
excellent technical assistance.
REFERENCES
1) Y. Hirabayashi: Chitin and Chitosan Research 1[2] (1995) 92–93.
2) Y. Hirabayashi and G. Kamiyama: Membrane Symposium, [9], (1997)
45–48.
3) Y. Hirabayashi and N. Onishi: Membrane Symposium, [10], (1998) 41–
44.
4) Y. Hirabayashi and N. Onishi: MRS-J Symposium, [10], (1998) 56.
5) Y. Hirabayashi and N. Onishi: The 10th Tsukuba Chemical Engineering
Symposium, (1999) 11–12.
6) Y. Hirabayashi and N. Onishi: The 10th Tsukuba Chemical Engineering
Symposium, (1999) 13–14.
7) Y. Hirabayashi and N. Onishi: Membrane Symposium, [11], (1999) 89–
92.