Appreciation of a membrane's surface chemistry and steric exclusion character is needed to
truly understand and predict membrane performance for specific industrial separations. The
interpreter of membrane characterization data must consider both factors in the assessment of
separation potential. The test conditions employed will strongly influence the separation data
outcome, as will inherent surface force interactions between the membrane and solution
components. Sepa CF cell solute challenges and affinity chromatography methods are useful
tools to characterize the pore size and surface character of ultrafiltration and nanofiltration
The characterization data presented demonstrate the separation potential of B-type
membranes. For example, the combination of the B-type membrane surface charge and pore
size affords economical separations of salts from organics. The anionic surface charge of B-
type membranes also makes them competitive for high fouling applications. Dye concentration,
paper pulp waste treatment and similar applications appear promising for B-type membranes
where traditional membranes are not well suited.
Key words: Anionic membrane, diafiltration, molecular weight cutoff, nanofiltration, ultrafiltration
Membrane separations in the ultrafiltration (UF) and especially the nanofiltration (NF) regions of
the filtration spectrum are governed by a complex combination of both steric exclusion and
surface force interactions. UF and NF membranes bearing formal surface charges display
unusual selectivity behavior not predicted on the basis of physical pore size alone. Therefore,
practical characterization should employ several techniques to gain insight on membrane
In this work we elucidated the separation characteristics of a novel, anionically charged
membrane with UF and NF capabilities. This study included three distinct techniques to
characterize the separation potential solute challenges to demonstrate the apparent molecular
weight cutoff (MWCO) of the membrane; a special affinity chromatography technique intended
to qualitatively demonstrate surface force interactions between the solutes and the membrane
resin; and pilot scale operation of actual industrial applications.
Molecular weight cutoff
The membrane characterized in this work is a novel thin layer composite. (TLC ). The
permselective barrier layer is produced from a unique polymer material. The commercial
membrane manufactured from this proprietary resin is referred to as B-type membrane. B-type
resin features a formal negative surface charge which is responsible for unique separation
characteristics of B-type membranes.
Solute challenges were completed using a bench-top laboratory test device known as a Sepa
CF cell (Figure 1). This device is unique among test cells: it is designed to model the fluid
dynamics of commercial spiral-wound membrane elements ("sepralators"). Like the larger
sepralators, the Sepa CF cell functions in a true cross-flow or tangential flow mode. Challenge
solutions are pumped under pressure through feed channel spacers. and permeate is routed
through a woven fabric spacer called the permeate carrier. Both of these spacers are the same
materials used in a spiral-wound element (sepralator). The presence of the mesh spacer in the
feed channel in Figure 1 is designed to promote turbulence in the flow channel to reduce the
build-up of solute near the membrane surface.
Figure 1. Sepa-CF test cell and cell holder.
For the present work cell pressures of 50 and 100 psi (3.5 and 6.9 bar) were used, each in
combination with average crossflow velocities of both 0.1 and 0.2 m/s. These operating
conditions are typical of those used in actual industrial processes employing sepralators . A
schematic diagram of the test apparatus is shown in Fig. 2.
Figure 2. Schematic of Membrane Test Apparatus
A variety of organic solutes and inorganic salts were used to characterize membrane
separation. Challenge concentration ranged from 500-2000 ppm. Solute detection in the
permeate was measured by refractive index for the organic solutes and by conductivity for salts.
Process application studies
Process application tests were conducted on-site with the actual process solution using pilot-
scale equipment. Separation objectives included product concentration or reclamation,
diafiltration for desalting and waste discharge volume reduction.
The operating parameters used in these applications were as follows:
Applied pressure: 100-300 psig (6.9-20 bar)
Crossflow velocity: 0.1 and 10.0 m/s
Process temperature: 20-50 C (68-120 F)
Mesh spacer flow channel: 0.86 mm (0.034")
Membrane surface force experiments
Qualitative measures of membrane-solute surface force interactions were determined by
injecting solutes into a chromatography column filled with the B membrane polymer. A relative
scale of adsorption affinity for each of the solutes was constructed by comparing retention
To prepare a column B-type resin was pulverized into a fine powder with a ball mill, then sieved
with mechanical agitation to obtain 38-45 µ size particles. The particulate B-type resin was
packed dry into a stainless steel chromatography column (25 cm long x 4.6 mm ID). Sintered
stainless steel 2 µ frits were used on the ends of the column to contain the particular packing.
The column was run on a high performance liquid chromatography (HPLC) system using pure
water (18.2 µS-cm treated via reverse osmosis, then deionization) as the mobile phase. The
system was allowed to equilibrate for approximately 6 hours prior to solute injection to ensure
proper bed settling in the column.
Solutes analyzed included monovalent and divalent salts, polysaccharides, polyethylene glycols
(PEGs) and alcohols. All solutes were tested at 200 ppm concentration with the exception of
deuterium oxide, a reference solute, which was tested at 1%. Sample injection volumes were
100 µL, and detection was accomplished using differential refractive index measurement. The
retention volume of each solute was used to calculate the degree to which the solutes were
partitioned between the bulk fluid flowing through the particle interstices and the thin layer of
interfacial fluid surrounding the particles of packing material. The equation describing this
KA = (Vsolute - Vmin) ÷ (Vwater – Vmin)
where the Vmin is the lowest retention volume among all solutes tested and Vwater is the retention
volume of water as measured by D2O injection.
Results and Discussion
Retention characteristics and apparent MWCO
Commercial UF and NF membranes are typically rated according to a nominal solute retention
level (e.g., 90%) commonly termed MWCO. Comparison of MWCO from various manufacturers
can be difficult when different test methods are used to determine the MWCO value. This is
frequently the case in practice because a standard test method has not yet been adopted by the
The MWCO rating of a UF or NF membrane can vary widely depending upon the operating
parameters and the chemical interactions or surface forces that occur between the solvent, the
test solute and the membrane. A membrane user should consider an MWCO rating in the
context of the chemistry, the separation device configuration and the particular test conditions
used to produce the rating. These same parameters will affect the usefulness of a given
membrane for specific applications.
Solute separation can be influenced by changing cross-flow rate and operating pressure in
many instances . This is illustrated in Table I. Variation of the operating parameters resulted
in significantly different solute separation in the particular case of B-type membrane operated in
a Sepa CF cell.
The differences in solute retention shown in Table 1 can be understood using the concept of
concentration polarization, the concentration gradient of solute molecules near the membrane
surface. Increasing cross-flow velocities through the Sepa CF cell generally reduced
concentration polarization, resulting in higher separation and higher permeate flux. Lowering the
operating pressures also decreased concentration polarization, resulting in higher separation
and higher flux in proportion to the operating pressure. Thus, high cross-flow and low operating
pressure provided the most economical separations.
Table I. Effect of Sepa CF Operating Parameters on B-Type Membrane Separation
Operating Crossflow Permeate Solute
Pressure Velocity Rate Separation
(1000 ppm) a
(bar) (m/s) (LMH) (%)
Na2SO4 3.5 0.1 12.4 50
3.5 0.2 13.0 57
6.9 0.1 24.2 33
6.9 0.2 24.7 43
Raffinose 3.5 0.1 13.5 17
3.5 0.2 14.1 24
6.9 0.1 23.6 8
6.9 0.2 17.7 12
PEG 3350 3.5 0.1 6.5 30
3.5 0.2 5.9 40
6.9 0.1 11.8 22
6.9 0.2 11.2 38
These trends between fluid dynamics and membrane results are commonly observed for UF
membranes . However, the extent to which concentration polarization affects a particular
membrane separation is not readily predicted. Accurate assessment of optimal operating
conditions is usually possible only by means of empirical studies with actual process streams
The membrane user should not presume that all separations effected by an ultrafilter or
nanofilter occur by size exclusion mechanisms alone. separation data for B-type membranes
with large nonionic organics are shown in Figure 3. The PEG and polysaccharide solutes are
typical of those used to characterize the size exclusion properties of UF membranes. The data
suggest that the B-type membrane is best described as a "tight" ultrafilter with an MWCO
ranging from about 2k000 daltons for polysaccharides to around 5,000 daltons for PEG's. The
MWCO result for PEG's is higher than that for polysaccharides because PEG's are adsorbed by
the B-type membrane while polysaccharides are repelled by the membrane. The effects of
adsorption and repulsion on apparent MWCO are addressed further in the later section on
membrane surface force characteristics.
Figure 3: B-type membrane retention of nonionic organics. All solutes tested individually.
Saccharides: Sepa ST stirred cell, 100 psig (6.9 bar), 150 rpm (0.4 m/s at outer edge). PEG's:
Sepa CF radial flow cell, 50 psig (3.4 bar), 0.9 m/s crossflow.
Figure 4: B-type membrane salt retention test conditions. 0.32 m/s crossflow velocity; 250 psig
(17 bar), 2000 ppm salt concentration.
Figure 4 demonstrates that B-type membranes can function as an NF membrane in the context
of salt separations despite high MWCO ratings for uncharged organics. Charge repulsion
between salt anions and the membrane surface complement separation due to physical sieving.
Note, for example, the membrane's ability to discriminate between NaCI and Na 2SO4. Sulfate
has twice the charge of chloride and is also much larger. Charge repulsion occurs further from
the membrane surface and hence sulfate is more easily excluded from membrane pores.
Another notable feature is the effect of calcium and magnesium on salt separation. These
results suggest that polyvalent cations effectively mask the charge character of the B-type
membrane barrier layer. Thus, these salts are not rejected as effectively by means of Donnan
exclusion. Additionally, Figure 5 shows that higher salt concentrations also decrease salt
separation. This phenomenon improves the efficiency of salt removal by B-type membranes,
especially in diafiltration and concentration of very salty process streams. Mass transfer of both
salt and solvent water benefit. Increased salt passage reduces the difference in salt
concentrations across the membrane, consequently reducing osmotic pressure and increasing
the available effective pressure. The effective pressure driving permeate flux is the result of
osmotic pressure subtracted from applied pressure as measured by a gauge. Higher effective
pressure results in higher permeate water flux.
Figure 5: Effect of salt concentration on retention.
B-Type membrane. Radial flow cell test. 5.2 m/s average crossflow, 100 psig (6.9 bar)
Table II. B-Type Membrane Retention of Ionically Charge Organics
Molecular weight % Separation
FD&C red dye #40 480 98.0-99.5
FD&C yellow dye #5 534 94.4-98.4
Tannic acid 1700 99.5
192T Sepralators, 0.1 m/s crossflow, 100 psig (6.9 bar).
Table II exemplifies the NF character of B-type membrane challenged to several commercial
anionic dyes and tannic acid, a model compound for the humic acid contaminants common to
many surface water supplies. High separation of these charged organic compounds was
achieved, yet their molecular weights are well below the B-type membrane's MWCO obtained
for uncharged solutes. Separation appears analogous to that for inorganic salts due primarily to
charge repulsion at the membrane surface.
Application case studies
Two examples of successful applications illustrate the important separation characteristics of
the versatile B-type membranes. Pilot scale application tests with actual process streams are
always advised for UF and NF separations because most of these applications have complex
solutions and hence it is difficult to predict results.
The first case involved a textile dye of a nominal 7,000 molecular weight in a process solution
contaminated by high concentrations of salts. The B-type membrane sepralators in this
application passed essentially no dye and all of the salts in the feed stream. B-type membrane
effected the separation at moderate operating pressures and high product rates.
Table III shows the performance of the B-type sepralator in a 30 L (8 gal) batch pilot test
process with diafiltration followed by concentration. This application was also tried with a
polysulfone UF membrane with larger pore size, but the surface character of the B-type
membrane afforded less fouling and higher steady-state processing flux. The application is
under development with B-type membranes.
Table III. Diafiltration and Concentration of Textile Dye (MW 6000-8000)
End of End of
concentration, 8.6 8.6
99.84 99.92 99.87
2.7 0.26 0.15
Flux, LMH 12.4 28.9 14.1
192T sepralator, 290 psig (20 bar) operating pressure, 30°C (86°F).
Liters per square meter per hour.
A second application involved waste treatment of paper and pulp production effluent . A B-
type membrane was effective at removal of lignosulfates from kraft E-stage wastewater
streams. Results showed a high reduction in waste stream color and COD in the permeate
stream. As in the dye processing example, the separation is effected with high product rates at
a moderate operating pressure. The data, summarized in Table IV, show the B-type membrane
competitive advantages over typical UF membranes. The B-type membrane effects greater
removal of environmental pollutants than UF membranes tested in previous work at comparable
or higher permeate flux rates [5, 6]. Membrane processing costs for paper and pulp waste
treatment have been historically regarded as prohibitive. Recent environmental concerns
together with improved membranes may offer new opportunities for treatment of paper and pulp
wastes. Membrane improvements in the case of B-type membrane include increased fouling
resistance and greater separation capability.
Table IV. Kraft e-stage Liquor Waste Treatment
B-type TLC modified
pressure, 6.9 (100) 9.3 (135)
Flux, LMH 33.0-105 23.0-59.5
COD rejection, % 88 50.0-65.0
50°C (120°F), pH = 10, ABB CR-250 stirred cell, 10 m/s at end of blade; feed stream rate 7.5
1pm (2 gpm).
Liters per square meter per hour
Membrane surface force characteristics
Solute retention data from HPLC adsorption affinity experiments can help demonstrate the
surface force mechanisms which contribute to the separation characteristics of a membrane.
Sourirajan et al.  developed an interpretation of solute retention as a measure of the degree
to which the solutes interact with the membrane polymer. This interaction is described either as
adsorption or repulsion. The theory allows for the calculation of a partition coefficient, KA, for
each solute injected into the column. Values of KA are measured relative to that of water (D2O
injection); values less than one indicate a net repulsion, and values greater than one indicate a
net adsorption. The general trends in resin solute partition data are shown in Figure 6 for a
variety of solutes.
Figure 6. Solute partition coefficients for B-type resin.
The partition coefficients are useful for predicting separation capabilities of UF and NF
membranes with various challenge solutes. In the case of B-type membranes, the
chromatography data show that alcohols and PEG's are adsorbed by the membrane polymer
while saccharides and salts are repelled by the polymer. This suggests that separation of
saccharides and salts by a B-type membrane will be higher than separation of similarly sized
alcohols and PEG's. This hypothesis was supported experimentally by comparing the
saccharide and PEG separation of B-type membrane depicted in Figure 3. The B-type
membrane rejected more polysaccharide than PEG's at any given molecular weight. Higher
saccharide separation occurred despite the fact that the saccharides are spatially smaller than
PEG's at equivalent molecular weights.
The grouping of "higher alcohols" in Figure 6 refers to ethanol, propanol, isopropanol, n-butanol
and n-pentanol. HPLC injection data suggest that these compounds are strongly adsorbed by
the B-type polymer packing material. Peak response for these materials was poorly defined,
typically appearing as an extended drift in the baseline of the chromatogram. The adsorption
noted in this case most probably results from hydrogen bonding between the alcohol hydroxyl
groups and hydrophilic functional groups on the B-type polymer.
Adsorption of PEG solutes by a B-type resin was similar to that of alcohols. Again the hydroxyl
end groups on these molecules probably hydrogen bond with the hydrophilic moieties on the
polymer. We also conjecture that the long linear conformation of these molecules allows
hydrophobic interactions between hydrophobic regions of the B-type polymer and the PEG
Results show that saccharide retention time decreases with increasing MW. This behavior is not
completely understood at present. It is believed that glucose coordinates with metal cations
contained in the B-type polymer. Larger saccharides may be unable to coordinate as effectively
with metallic moieties in the B-type polymer.
The HPLC column was unable to discriminate between monovalent and divalent salts on the
basis of charge repulsion; both were strongly repelled. In contrast, the B-type membrane did
separate monovalent and divalent salts; separation of the divalent species was higher. This
suggests that the membrane discriminates between monovalent and divalent salts on the basis
of both ionic size and charge repulsion.
1. M. Cheryan, Ultrafiltration Handbook, Technomic Publishing, Pennsylvania, 1986.
2. B.J. Rudie, R.A. Condiff and P.W. Kariniemi, The Influence of Operating Parameters on
Ultrafiltration Membrane Dextran Rejection, International Congress on Membranes and
Membranes Processes, Chicago, 1990.
3. S. Bruin, A. Kikkert, J.A.G. Weldring and J. Hiddink, Desalination, 35 (1980) 223.
4. H. Buisson, Personal Communication, Wastewater Technology Centre, Burlington, Ontario,
5. M.K. Hill, UF of Krafi Black Liquor--Final Report, Phase I, DOE Contract No. DE-ACO2-
82CE40606, modification A003, 1985.
6. D.J. Paulson and D.D. Spatz, RO/UF Membrane Applied to the Pulp and Paper Industry,
Proceedings, Tech. Assoc. of TAPPI International Dissolving and Specialty PuIps Conference,
7. T. Matsuura, Y. Taketani and S.J. Sourirajan, Colloid Interface Sci., 98 (1983)10.