C.

Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

GRAHAMTEK PROJECT
FINAL REPORT
PART C

Effect of EMF and Bubbles on Critical Flux of Particles

Using DOTM

Team members:
Professor Tony Fane
Associate Professor Adrian Law
Zhang Yanpeng
School of Civil and Environmental Engineering
Nanyang Technological University, Singapore

Page 1
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Team Members: Prof. Tony Fane, A/Prof Adrian Law and Zhang Yanpeng

C. 1. Introduction and Objectives

Grahamtek has developed novel technologies, including an EMF device and a flow
distributor (that may produce micro-bubbles) that appear to show significant effects in
reducing fouling in spiral-wound RO modules. The reasons for the beneficial effects are
not yet well established. Therefore this project is focused on investigating the effects of
an EMF device and micro-bubbles on fouling reduction in a model system.

An electromagnetic field (EMF) is present when an electrical current flows through a

wire coil. The EMF device of GrahamTek uses a three-phase AC current. It can be
applied either upstream of the membrane module or around the spiral-wound RO module
(this is the usual arrangement). It has been reported that membrane fouling can be
significantly reduced with EMF applied compared to the situation without EMF. It has
been postulated that particle aggregation could have occurred when the particles in the
feed solution experienced EMF. A detailed analysis of possible AC field effects has been
presented in section A of this report (‘Effect of AC fields on bacteria and particles:
implications for RO).

When fluid flows under high pressure and passes through a well-designed restriction it is
possible that small bubbles are produced by cavitation effects. Within a spiral wound
module such small bubbles may be able to enhance polarization control and remove or
prevent particle deposition. In the GrahamTek system it is believed that small bubbles are
controlled and produced by the novel Integrated Flow Distributor (IFD). The IFD is a
perforated plate located in the feed side. The holes in the IFD are positioned with a
certain angle to the distributor surface. It is postulated that the production of small
bubbles can be enhanced by the IFD and this is being investigated in a separate subtask
(see section E of this report, ‘Assessment of the formation of micro-bubbles by Integrated
Flow Distributor).

This section (C) reports the findings of Task 3 which is to investigate the effects of EMF
and small bubbles on membrane filtration in a ‘model’ system with well-controlled
hydrodynamics, flux and feed properties. Measurements have been made of the critical
flux of particle suspensions in crossflow microfiltration using a non-invasive, in-situ
direct observation technique, known as Direct Observation Through the Membrane
(DOTM). The objectives of Task 3 include:
(i) To measure the critical fluxes for organic and inorganic particles with and
without EMF;
(ii) To study the effects of placing the EMF device upstream or winding it around
the membrane cell;
(iii) To monitor the changes in particle size distribution after the EMF exposure;
(iv) To optimize the operating conditions with EMF to enhance critical flux;
(v) To observe the effect of small bubbles on fouling removal and prevention.

Page 2
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

C. 2. Theoretical basis
This task studies the effect of EMF and small bubbles on particle deposition during
crossflow microfiltration in terms of the concept of critical flux. The concept of critical
flux has been often used to describe the conditions where incipient fouling occurs. In
1995, Field et al. first proposed the concept of critical flux from their experiments in
filtration of yeast suspension in the constant flux mode [1]. Field et al. stated that:

“There exists a flux below which a decline of flux with time does not occur; above his
flux, fouling is observed. This flux is termed as the critical flux and its value depends
on the hydrodynamics and probably also on the other variables.’’ [1]

Many experiments and theories show that the value of critical flux is a function of
numerous parameters such as suspension properties (particle size and concentration),
surface interactions (pH and zeta potential) and hydrodynamic operating conditions
(tangential velocity) [2]. In microfiltration where the sieving mechanism applies, the
overall filtration rate (known as permeate flux) can be expressed as a ratio of the pressure
driving force to the filtration resistance. The driving force is the transmembrane pressure
and the filtration resistance includes the membrane hydraulic resistance due to the
membrane microporous structure as well as the cake layer resistance due to the particle
deposition in or on the membrane. The general relationship between the permeate flux
and the filtration driving force described by Darcy’s law:

1 dV ∆P
J≡ = (C-1)
Amem dt µ ( Rm + Rs )

where J is the permeate flux, V is the total volume of permeate, Amem is the membrane
area, t is the filtration time, ΔP is the pressure drop imposed across the cake layer and
the membrane, µ is the viscosity of the permeate, Rm is the membrane hydraulic
resistance which can increase with time due to membrane fouling and compaction and Rs
is the cake resistance which can increase with time due to the cake build-up and
compression [3].

There are several mathematical models to predict permeate flux in the literature. Among
them, the mostly used models are Brownian diffusion model, shear-induced diffusion
model and inertial lift model. The application of these models depends on the particle
size in the suspensions. In microfiltration, micron particles are to be separated. When it
comes to micron particulates (0.2~30 µm in diameter), the conventional concentration
polarization model cannot predict the flux properly since Brownian diffusion is no longer
significant for such large particles. Zydney and Colton replaced the Brownian diffusivity
by the shear-induced diffusivity first measured by Eckstein et al. in 1977 [4]. They
observed that the shear-induced diffusivity increases with increasing shear rate and
concentration. Depending on the concentration, the shear-induced diffusivity (SID) has
the following forms:
D s ≈ 0.1φ r 2γ For 0< φ <0.2 (C-2)
Ds ≈ 0.025φ r 2γ For 0.2< φ <0.5 (C-3)

Page 3
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Zydney and Colton (1986) applied D = 0.03r 2γ o to the Leveque equation in the
concentration polarization Film Model giving [4]:
r 4 1/ 3 φ
J = 0.078( ) γ o ln( w ) (C-4)
L φb

From Equation (C-4), the permeate flux increases linearly with the shear rate and
proportionally to four-third power of the particle radius. Both the concentration
polarization model and the SID model predict that the flux decreases when the bulk
concentration increases.

Usually it is difficult to determine the wall concentration ( φw ). To determine its value,

many assumptions have to be made. In Li’s work [5], they used the membrane area
coverage by the deposited particles to calculate φw . They proposed for spherical
particles, φw = 2 / 3ξ , where ξ is the fraction of membrane area covered by the deposited
particles . They found that the shear-induced diffusion model predicted reasonable fluxes
for 6.4 and 11.9 µm latex particles and under-predicted the critical flux for 3 µm latex.
They also provided a modified shear-induced diffusion model which is applicable to 5-12
µm model particles.

C.2.1 Implications for RO

Although most of the critical flux studies have been applied to microfiltration there is no
reason why the phenomena will not be observed in RO. Critical flux is based on the
competition between convection (flux) and crossflow-induced backtransport. For
example, we have recently shown, using the techniques described in section D, that
critical flux can be measured in model RO systems with saline feed and colloidal silica
foulant [6]. Critical flux behaviour has also been observed on large scale RO plant [7].

C. 3. Experimental Methods
C. 3.1 Equipment
The tests on the effect of EMF were conducted with two EMF devices, the EMF ‘cell
winding’ and the EMF ‘upstream’. The EMF ‘cell winding’ is a copper coil around the
membrane cell and produces the electromagnetic field right above the membrane surface.
The EMF ‘upstream’ is a PVC tube allowing fluid flow through with a copper coil
around it. Due to the different configurations, they are described separately in the
following sections. Most of the tests were with the EMF cell winding. The results ‘with
EMF’ denote the experiments using the cell winding if not otherwise specified.

C.3.1.1 EMF cell winding

A typical crossflow microfiltration system was set up in the laboratory. The direct
observation through the membrane (DOTM) technique was integrated into the filtration
line. Figure C.1 shows the crossflow microfiltration rig used in this project including the
DOTM facilities.

The feed solution was stored in a reservoir and pumped by a gear pump (Cole-Parmer
Instrument) into the filtration line and through the membrane module which was

Page 4
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

positioned on the working stage of a microscope. The feed solution was circulated in the
pipeline and recycled back to the feed tank. Constant flux stepping was achieved by a
peristaltic pump (MasterFlex, Cole-Parmer Instrument) on the permeate line. The
permeate liquid was collected in a container on an electronic balance. The mass weight
was recorded by the lab PC with an attached timer and transformed into permeate flux
according to the time interval and membrane area. The operation pressure was controlled
by a pressure gauge fixed in the pipeline after the feed side outlet of the module. Three
pressure transducers in the up- and downstream of the membrane module and permeate
side were used to collect pressure data. The pressure data of P1, P2 and P3 were recorded
separately in the PC and used to calculate the average transmembrane pressure (TMP)
from the equation, TMP = 0.5x(P1+P2)-P3. An electronic flowmeter was located in the
outlet of the pipeline to record the flowrate. All the electrical signals were fed to an A/D
data acquisition board connected to the PC for recording and the parameters were
displayed in separate graphs automatically using the LabView software.

The key to the DOTM technique is using the Anopore inorganic membrane (Anodisc,
Whatman, UK) which is transparent when wet. The observation was viewed via a digital
CCD camera which was connected directly to a PC. The images could be snapped with
pre-determined time intervals and stored in the PC while the preview could be made at
the same time.

Figure C.1 Filtration set-up including DOTM facilities.

The membrane module used in DOTM is a crucial component as it has to be transparent

like the membrane. A membrane module was specially designed and fabricated from
perspex to allow the transmission of light (Figure. C.2). The membrane module consisted
of the two parts, permeate channel and feed channel, with the transparent membrane in
between. An observation window was cut in the permeate channel half allowing the
objective to move close to the membrane in the module. A piece of thin optical glass was

Page 5
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

used to seal the window. The O-ring in the feed channel half prevented any leakage and
held the pressure in the flow channel. The feed flow channel was 110 mm in length, 40
mm in width and 1 mm in height. A perforated plate with 3 mm holes made of plastic
(instead of metal) was placed in the permeate channel to support the membrane and
prevent membrane breakage. An objective lens with 32x magnification (Carl Zeiss,
Germany) was mounted on the permeate side of the module. Another two objectives with
10x and 20x magnifications were also available for observing larger particles. The image
observed by the microscope (Axiolab Zeiss, Carl Zeiss, Germany) was viewed via a
digital camera (MicroPublisher 5.0 RTV, QImaging, Canada) and previewed on a
computer monitor.

Figure C.2 Schematic of the membrane module.

Figure C.3 shows a picture of the EMF device provided by GrahamTek, including the
power box and the cell winding. The winding was a specially designed and fabricated
coil using copper wires. The winding was fitted into the microscope used in DOTM and
the membrane cell was inserted into the winding. The hole in the middle of the winding
provided a viewing window allowing observation of the membrane surface. The feed
solution experienced the electromagnetic field when passing through the membrane cell.
Pressurize air was passed over the cell in order to remove the heat produced by the EMF
device.

Page 6
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Figure C.3 Picture of EMF cell winding working with DOTM.

C.3.1.2 EMF upstream

The EMF upstream device was used to find evidence of particle aggregation during
membrane filtration. Due to the power input to the EMF upstream tube the feed tended to
heat up. To assess this the feed solution was circulated through the upstream tube and the
temperature changes with time were monitored. It was found that the temperature of the
feed solution increased from 25 to 40 ºC within 3 hours. Yeast cells used in this task
would tend to ferment at such a high temperatures. It was necessary to keep the
temperature constant in order to eliminate this effect and to control the filtration
temperature. An independent circulating line was added to the original filtration system
to control filtration temperature, as shown in Figure C.4. The circulating line on the left
in Figure C.4 was used to circulate the feed solution through the EMF upstream tube at a
higher flowrate (3 l/min). A water bath with a cooling coil was used to keep the feed
temperature constant at 25 ºC. The original crossflow filtration loop was run at the usual
flowrate (0.85 l/min).

Page 7
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Figure C.4 Integrated system with a circulating line.

C.3.1.3 Bubbling setup

Several difficulties must be solved in order to use DOTM to observe the effect of small
bubbles. These problems includes,
(1) How to introduce bubbles into the feed channel;
(2) How to control the air pressure and flowrate;
(3) How to control bubble size.

The bubbles tended to grow or connect together along the system tubing. The bubble size
was difficult to control in the millimeter range. The air was introduced by a syringe
needle inserted as near as possible from the inlet of the feed channel. The air was
provided by an air pump since the membrane cell could not resist a very high pressure. It
was observed that the bubbles tended to accumulate together and sometimes stayed inside
the feed channel when the membrane cell was positioned horizontally. Therefore, the
microscope was rotated by 90 degree and mounted on a customer-made stand to make the
cell vertical (Figure C.5). Figure C.6 shows the vertical DOTM configuration used for the
bubble study. A mass flow controller (5058E, Brooks Instrument) was used to control the
air flowrate.

Page 8
C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Figure C.5 Picture of vertical DOTM.

Figure C.6 Vertical DOTM configuration for bubble study.

Page 9
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

C. 3.2 Methods and Protocols

C.3.2.1 Materials
Polystyrene latex particles were used as the model inorganic particles in this project.
They have several advantages which are perfectly suitable for the fouling study, such as
stability in solution, buoyancy in water, spherical shape and narrow particle size
distribution. Micro particles based on polystyrene (Fluka, Sigma-Aldrich, U.S.A.) are
negative charge-stabilized colloidal particles. Latex beads are produced by the
polymerization of styrene under conditions that induce spontaneous coalescence bead
formation. Normally latex beads are supplied as aqueous suspensions with a solid content
of 10% and a similar density (1.05 x 103 kg/m3) to pure water. They consist mainly of the
polymer particles and water, with small amounts of surfactant, sodium bicarbonate and
potassium sulfate. The 3.0 µm latex particles used in this study had a standard deviation <
0.1 µm. The latex suspension with a specified concentration was diluted and re-dispersed
in MilliQ○R water to make a feed solution for filtration. The solution pH was adjusted
using 0.1 molar acid or alkaline solutions. The salinity was controlled by introducing
sodium chloride into the solutions. Latex particles larger than 0.6 µm tend to settle down
slowly in suspensions. The prepared feed suspensions had to be stirred to maintain a
uniform suspension using a magnetic stirring bar in this study. The latex supplies and
reused suspensions were stored in a refrigerator at 4 ºC for sterilization.

Yeast cells were selected as the model organic particles in this project. The proteins
extracted from yeast are often used to study protein fouling and adsorption [8]. Unlike
latex suspensions, yeast suspensions are usually selected as a substitute for bio-
suspensions to study microbial fouling in microfiltration. Yeast cells are not spherical as
the latex particles. However its elliptic shape is often regarded as approximately spherical.

Baker’s yeast (Saccharomics cerevisiae) was selected in this study obtained from the
NTUC Fair Price supermarket in Singapore. In order to keep the yeast properties constant,
only one product of yeast (Saf-instant, France) was used to prepare the yeast suspensions.
The yeast powder was washed before used to remove cell debris and extra-cellular
polymeric substances (EPS) attached to the yeast cells. The washing process was similar
to that used by Mores [9]. 0.25 gram of powder with 50 ml MilliQ○R water was mixed in a
50 ml centrifuge tube and then centrifuged at 2500 rpm for 10 minutes. The supernatant
liquor was discarded. The yeast was re-dispersed in 50 ml fresh MilliQ ○R water and
centrifuged again. This washing procedure was repeated three times before making a
yeast suspension. Some of the mass would be lost during washing. The washed yeast was
dried at a temperature of 105 ºC for 24 hours. The weight loss was about 30 % after
washing for three times. Only 70 % of the final product was re-dispersed to make a feed
solution. The particle size distribution was measured by the Mastersizer (Malvern, UK).
The measurement of particle size distribution shows a mean size (d50) around 5 µm with a
narrow distribution. Fresh yeast suspensions were prepared every time just before the
filtration and discarded within 4 hours to prevent fermentation.

Page 10
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

C.3.2.2 Experimental protocols

The critical flux for filtration of particles is defined according to the critical flux concept
by the flux stepping filtration test. All the tests were conducted in constant flux mode.
The pre-prepared membrane was inserted into the membrane module and the two halves
of the module were screwed together. The module was then connected into the filtration
pipeline and mounted on the working stage under the objective. Clean water from the
feed tank was pumped slowly into the system. After all the air and bubbles were
displaced from the lines and module the flowrate was set to the desired value. The
crossflow was allowed to stabilize for 10 minutes before the concentrated particles were
spiked into the tank. The feed solution was stirred by a magnetic bar to make the
suspension uniform and prevent particle aggregation. The data logger was started to
record all the parameters and the observed images were captured simultaneously. The
objective lens focus had to be frequently adjusted to get clear images. The average
crossflow velocity (computed by dividing the crossflow discharge by the flow area) was
fixed at 0.2 m/s. The corresponding Reynolds number was 600 and hence the flow regime
was laminar.

The protocol for critical flux determination involved operations with a series of flux steps.
The tests started from a low imposed constant flux. If there was no particle deposition or
fouling layer on the membrane surface during 15 minutes, the operating flux was
regarded as being below the critical flux for this situation. The flux was then increased
step by step with an increment of 15 l/m2h until about 2/3 of the observing membrane
area was covered by the particles during the 15 minutes step. This operating flux was
regarded as above the critical flux. A reported critical flux is the average value of the
neighboring fluxes where the fouling transient happens. Due to the non-uniformity of the
deposition on the membrane, the experiments were repeated at least three times with an
even smaller flux increment to verify the critical flux value. The smallest increment in
flux which can be achieved by the peristaltic pump on the permeate side is equivalent to
about 1.5 l/m2h. The error bar shows the standard deviation of three experiments.

After use the filtration system was flushed by a 4 g/l detergent solution (Tergazyme,
Alconox, U.S.A.) at 40 ºC for at least 1 hour. Then the detergent solution was drained out
and MilliQ○R water was used to wash the system again several times. Sodium azide (NaN3)
solution at a concentration 2 g/l was used to remove microorganisms possibly stained on
the wall of the system once in a while. After that, the system was washed with MilliQ○R
water.

C. 4. Results

C.4.1 EMF winding and hematite particles

Preliminary tests with the EMF winding were conducted using iron hematite particles.
For a feed containing hematite flocs (prepared in our lab), the alignment of the flocs was
changed by 90 degree compared to the non-EMF conditions. As a result, the nature of
particle deposition was changed. However, the fouling was not alleviated when EMF was
applied (Figure C.7). This established that the EMF had an effect on conductive materials,
but under these conditions it did not alter critical flux.

Page 11
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Without EMF (flow direction: right to left)

5 min 10 min 20 min

With EMF (flow direction: right to left)

Figure C.7 Hematite particle deposition pattern with and w/o EMF winding.

20th min 30th min

35th min 44th min

(a) Images of baker’s yeast (with EMF before 30 min, without EMF after 30 min).

Page 12
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

0.8 50
40
0.6

Flux (l/m2h)
TMP (kPa)
30
0.4 without EMF
with EMF 20
0.2 TMP 10
Flux
0 0
0 15 30 45
Time (min)
(b) TMP and Flux of the measurement of baker’s yeast.
Figure C.8 Critical flux of baker’s yeast with EMF at CFV 0.1 m/sec, Con. 0.55g/l.

C.4.2 EMF winding and yeast cells

Figure C.8 shows the effect of the applied EMF (from 0-30 minutes), followed by a
period (30-44 minutes) without EMF. Figure C.8 (b) shows that at the 30th minute the
imposed flux was 37 l/m2h which was higher than the critical flux of about 25 l/m2h
without EMF [note that the TMP values were very low and cannot be taken as a true
measurement of TMP changes]. The coverage of the membrane at 30 minutes (with EMF)
is shown in Figure C.8 (a) and was very low. When the EMF was switched off there was
no significant additional deposition. This may mean that the yeast inventory had been
“pre-conditioned” by the EMF as it would have cycled an average 6 times through the
cell.

5th min 15th min

Page 13
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

20th min 30th min

(a) Images of baker’s yeast (without EMF before 15 min, with EMF after 15 min).
1.2 50

Flux (l/m2h)
TMP (kPa)

40
0.8
without EMF with EMF 30
0.6 TMP
Flux
0.4 20
0 15 30
Time (min)
(b) TMP and Flux of the measurement of baker’s yeast.
Figure C.9 Critical flux of baker’s yeast with EMF at CFV 0.1 m/sec, Con. 0.55g/l.

Figure C.9 shows what happened when the experiment was repeated with fresh feed at a
flux of 37 l/m2h, but now starting without the EMF. Figure C.9 (a) shows that at the 15th
minute there was a more evident deposition of yeast, as expected for operating above
critical flux. Between the 15th to 30th minute, the EMF was applied, and close inspection
of the Figure C.9 (a) image at the 30th minute suggests some removal may have occurred.

Figure C.10 (a) shows images at the 15th, 30th and 45th minute with the EMF at fluxes of
37, 52 and 65 l/m2h respectively. At all these supra-critical fluxes (measured as 25 l/m2h
without EMF) the yeast deposition was sparse (compare with Figure C.9 (a) at 15th
minute). Figure C.10 (a) also shows that when the EMF was switched off (images at 50th
and 60th minute) the deposition did not increase. However, the yeast feed had already
circulated an average 10 times through the EMF cell before the power was switched off.
Again, this could imply some “pre-conditioning” has occurred.

Page 14
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

5th min 15th min

20th min 30th min

35th min 45th min

50th min 60th min

(a) Images of baker’s yeast (with EMF before 45 min, without EMF after 45th min).
th

Page 15
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

1.5 90
80

Flux (l/m2h)
70

TMP (kPa)
1
60
50
0.5 with EMF without EMF 40
TMP
30
Flux
0 20
0 15 30 45 60 75
Time (min)
(b) TMP and Flux of the measurement of baker’s yeast.

Figure C.10 Critical flux of baker’s yeast with EMF at CFV 0.1 m/sec, Con. 0.55g/l.

(a) 15th min in Figure C.9 (b) 45th min in Figure C.10

Figure C. 11 Comparison of images on gray scale from Figures C.9 and C.10.
(a) flux = 37 l/m2hr,no EMF
(b) 65 l/m2hr,with EMF

Figures C.9 and C.10 provide visual comparisons of deposition without and with the
EMF. To highlight the differences selected images have been processed in ‘gray scale’
using image analysis software. Figure C.11 (b) shows much less deposition with EMF,
even though the flux was significantly higher (65 vs 37 l/m2hr) and the filtration period
was longer.

Page 16
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

120

100 no salt

80

Jcrit (LMH)
60

40
EMF
20
no EMF
0
0 2 4 6 8 10 12
pH

Figure C.12 Critical flux of 0.525 g/l yeast with EMF at various pH, CFV=0.2 m/sec.

Figure C.12 shows the critical flux data for 0.525 g/l yeast suspensions under various pH
values with and without the EMF. Both lines show that the critical flux for the yeast
suspensions increased with the pH values in a large range. The gap between the two lines
is about 10 l/m2h of flux which means greater increase in permeate flux was achieved
with the EMF. The error bars shown are the result of three repeats of each experiment.

Figure C.13 shows the effect of EMF on the critical flux for 0.525 g/l yeast suspensions
in the presence of salt in the system at pH 7.0. The addition of salt caused more fouling
compared with the same conditions without any salt whose critical flux was 60 l/m2h
when the EMF was applied. Significant relative increases in the critical fluxes were found
when the EMF was applied which can be seen in Table C.1.

70
60 EMF
50 no EMF
Jcrit (LMH)

40
30
20
10
0
0 1 2 3 4 5 6 7
Salt (g/l)
Figure C.13 Critical flux of 0.525 g/l yeast at pH=7 at various salt concentrations.

Page 17
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

Table C.1 Increase in critical flux with EMF

Salt (g/l) 0 2 4 6
no EMF (LMH) 48 40 20 15
EMF (LMH) 60 50 30 20
Increase % 25 25 50 33.3

Figure C.14 presents the critical flux of yeast at various concentrations in the presence of
2 g/l salt with and without EMF at a CFV=0.2 m/sec. It shows the anticipated trend that
the critical flux decreases with the feed concentration. This occurred with and without
EMF, but the critical flux was clearly enhanced by the EMF at all conditions.
70

60 salt 2 g/l
50
Jcrit (LMH)

40

30

20 EMF
10 no EMF

0
0 0.2 0.4 0.6 0.8 1 1.2

Yeast concentration (g/l)

Figure C.14 Critical flux of yeast at different concentrations in presence of 2 g/l

salt; with and without EMF.

C.4.3 EMF winding and latex particles

Figure C.15 shows the effect of the EMF cell device on the polystyrene latex particles
with a concentration of 0.05% vol. in brackish water. All the tests were conducted at the
isoelectric point of the membrane (pH 8) with a crossflow velocity 0.2 m/sec. The results
show that the critical flux decreased with the salt concentration in the solution. This is
probably due to a decrease in the electric double layer interaction between the particles
caused by the presence of salt. It also can be seen that the critical fluxes resulting from
the two situations, with and without EMF, were comparable. The differences were
within experimental error. Although this is a null result it does show the reproducibility
of the measurement technique.

Page 18
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

80
70
60

Jcrit (LMH)
50
40
30
20 EMF
10 no EMF
0
0 1 2 3 4 5 6 7
Salt concentration (g/l)
Figure C.15 Critical flux of 3.0 µm latex in brackish water (0.05% vol., pH 8, CVF
0.2 m/sec).

Shown in Figure C.16 are the results on the effect of the EMF cell device on the critical
flux of polystyrene particles at various pH values in the absence of salt in the solution.
The critical fluxes were all around 60 l/m2h for all pH values whether there was EMF or
not. At the isoelectric point of the membrane (pH 8), the critical flux may show a small
peak. This implies the effect of surface charge has no significant impact on the critical
flux up to 3.0 µm. Importantly for this type of feed the EMF had no detectable effect on
critical flux. It is plausible that this material would respond to EMF at different
frequencies (see section A), but our system did not allow frequency variation.

80
70
60
Jcrit (LMH)

50
40
30
20 EMF
10 no EMF
0
4 5 6 7 8 9 10 11
pH
Figure C.16 Critical flux of 3.0 µm latex at various pHs (0.05% vol., CVF 0.2 m/sec).

Page 19
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

C.4.4 EMF upstream on yeast cells

Figure C.17 shows the critical fluxes for 0.525 g/l washed bakers’ yeast suspensions at
various pH values when there was EMF, EMF ‘upstream’ and no EMF in the absence of
salt. The critical flux increased when the EMF winding device was applied around the
membrane cell. However, there was no significant improve in the critical flux when the
EMF upstream device was used. Shown in Figure C.18 are critical fluxes for 0.525 g/l
washed bakers’ yeast suspensions at pH 7 in the presence of salt. Again the critical flux
increased in the case of using the EMF winding but there was no significant change in the
case of using the EMF upstream device.
120
100 without salt, yeast 0.525 g/l

80
Jcrit (LMH)

60
40 EMF upstream
20 EMF
no EMF
0
3 4 5 6 7 8 9 10 11 12
solution pH
Figure C.17 Critical fluxes for 0.525 g/l yeast at different pH values.
70
60 yeast 0.525 g/l, pH 7.0 EMF upstream
50 EMF
Jcrit (LMH)

no EMF
40
30
20
10
0
1 2 3 4 5 6 7
Salt concentration (g/l)
Figure C.18 Critical fluxes for 0.525 g/l yeast at different salt concentrations.

C.4.5 Effect of bubbling

Tests on small bubbles were conducted using 10 µm latex particles at a concentration of
0.05 % vol. and CFV 0.2 m/s. The suspension, mixed with small bubbles (1-2 mm), was
passed through the feed channel vertically. Preliminary tests showed that the bubble size
was a function of fluid flowrate in the vertical position. The bubble size decreased with

Page 20
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

increasing crossflow rate. Conditions were found to produce small bubbles around 1-2
mm.

There are two purposes of introducing bubbles to enhance filtration, to remove deposited
particles on the filter media and to prevent particles from landing on the filter surface. In
this study, cake layers were purposely formed on the membrane surface first above the
critical flux (Jcrit = 100 l/m2h). A mono cake layer was formed within 3 minutes (Figure
C.19 (a)) and an estimated four layers were form in the following 3 minutes (Figure C.19
(b)) at a flux 140 l/m2h. After the 6th minute, bubbles were introduced into the feed
channel. The air flowrate was set at 25 ml/min which was about 3 % of the fluid flowrate.
The interval capturing image was adjusted to 10 seconds to observe closely the particle
removal. The movement of the bubbles was unable to be observed due to the large bubble
size and high flowrate. The layer on the top of the cake was removed almost as soon as
the bubbling started (Figure C.19 (c)). Although the images do not show a clear view of
each layer, the changes in image darkness indicate that the small bubbles peeled off the
cake layer by layer instead of removing the cake all together (Figure C.19 (d)). 110
seconds after bubbling started, only a mono layer of particles was left which could be
clearly seen from the image (Figure C.19 (e)). The layer was loosened (Figure C.19 (f))
and then started flowing on the membrane surface (Figure C.19 (g)). After 230 seconds,
all the fouling was removed and the membrane was almost clean again (Figure C.19 (h)).

(a) 3rd min (b) 6th min

(c) 20 sec after (b) (d) 70 sec after (b)

Page 21
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

(e) 110 sec after (b) (f) 130 sec after (b)

(g) 200 sec after (b) (h) 230 sec after (b)
Figure C.19 Fouling removal process by bubbling.

After the fouling removal, the operating flux was increased further with bubbling to
observe how the bubbles prevented particle deposition. At a flux 160 l/m2h, the particles
were barely able to touch the membrane surface. At a flux 210 l/m2h, an equilibrium was
reached as some of the particles landed on the surface for a while and were then removed
by the continuous bubbles. The critical flux for these bubbling conditions was not found
as the operating flux had to be very high and was beyond the pump working range.
Although the results with bubbling were qualitative, due to difficulties in producing well
characterized bubble sizes, the trends are clear. The critical flux with bubbling was
significantly higher than with the non-bubbling operation.

C.5. Discussion
The results with the hematite particles and yeast cells show that the EMF winding had an
effect on particle deposition during the crossflow microfiltration. The critical fluxes for
the yeast cell suspensions have been increased by around 30 % with EMF winding.
However, particle aggregation may not be the reason for the increase in critical flux since
the particle size distribution did not change noticeably after circulation through the EMF
winding. This points to a mechanism of dielectrophoretic mobility as an enhancement
velocity for augmenting critical flux. Thus the Film Model becomes,

J CRIT = k S ln (CW C B ) + V E (C-5)

where VE is the enhancement velocity.

The fact that there was no significant change in the critical fluxes for the latex particles
with the EMF does not necessarily invalidate this conclusion. The difference is probably
due to the low dielectric constant of the latex particles (2.3-2.7). As explained in section
A the effect of the EMF depends on the material of the particles. Zeta potential also
influences the effect of the EMF. This can be seen from the experiments at various pH
values and with the presence of salt in the feed solutions.

The EMF upstream was shown to be non-effective in the yeast experiments. One
postulated mechanism of EMF enhancement was particle aggregation, and this was not
observed. If it had been an enhancement mechanism the EMF upstream would have been
expected to help. However the EMF upstream arrangement did show an effect in the RO
tester with colloidal silica (section D), so the lack of an effect with yeast is not readily

Page 22
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

explained. These results underline the fact that we have not confirmed the mechanisms at
work with the EMF devices. Further work with the ability to vary frequency and power
may reveal the mechanism(s).

Another assumed aspect of the GrahamTek technology that could enhance filtration
performance is the presence of micro bubbles. Bubbles smaller than 1 millimeter are
usually spherical. The mechanisms of bubbling enhancement have been intensely studied
in MF and UF, but seldom in RO [10]. Two-phase flow was confirmed to increase mass
transfer during membrane filtration. The mechanisms for flux enhancement by bubbling
have been summarized as follows:
(1) Interruption of the concentration polarization boundary layer;
(2) Turbulence and secondary flow on the membrane surface;
(3) Structural modification of the cake layer (as observed in this study, Figure C.19).

The magnitude of these mechanisms depends on the membrane geometry (tubular, flat
sheet or hollow fiber) and membrane configuration (horizontal or vertical). In this brief
study the critical flux of the latex was confirmed to be enhanced in the presence of
bubbling. However, the degree of enhancement needs further study. With the help of a
limited flow of small bubbles, it is plausible that particle deposition in RO could be
reduced. (This observation is in line with that in section D ‘Effect of EMF and bubbles on
deposit fouling in RO tester’).

C.6. Conclusions and Recommendations

Experiments on a lab-scale DOTM membrane system equipped with the Grahamtek EMF
devices confirmed that under some conditions the AC field effect enhances critical flux
of particles by up to 30%. The enhancement was observed for yeast suspensions with the
EMF mounted around the membrane cell. It was not observed for yeast suspensions with
the EMF device located upstream of the membrane cell. The EMF effect was also not
observed for a suspension of latex particles with either EMF location. The different
responses could be due to differences in the dielectric constants of the particles and other
properties. The enhancement mechanism is not clear but is unlikely to be due to
aggregation and increased particle size. It is more likely to be due to a dielectrophoretic
back-transport velocity. Although the results were obtained in a low pressure membrane
cell it is reasonable to assume that the phenomena could occur under some conditions in
RO applications.

Air bubbling was found to be an efficient tool for preventing membrane fouling. Small
bubbles were formed and introduced into the DOTM membrane module. The particle
removal and deposition prevention were observed using DOTM. The bubbles were found
to be effective layer upon layer instead of lifting all the particles at the same time. This
could be an alternative to chemical cleaning which has environmental limitations. Small
bubbles could effectively prevent particles depositing permanently above the critical flux.
An equilibrium was observed at high flux where the particles were deposited and then
removed in rapid succession. Unfortunately, the critical flux with bubbling was not
identified due to instrument limitation.

Page 23
 C. Effect of EMF and Bubbles on Critical Flux of Particles Using DOTM

The EMF study was limited by the fixed frequency and power of the EMF devices
provided. A more thorough assessment, leading to a mechanistic understanding and
optimization requires a tunable EMF device and a wider range of feed suspensions. If
further studies are planned then these approaches are recommended.

The effect of microbubbles warrants further investigation. The benefits of bubble

introduction need to be balanced against the added complexity and potential energy
penalty. However it is possible that a properly optimized two phase flow system could
bring significant gains in fouling control. Further work on microbubble generation is
recommended.

C.7. References:

[1] R. W. Field, D. Wu, J. A. Howell and B. B. Gupta, Critical flux concept for
microfiltration fouling, J. Membr. Sci. 100 (1995) 259.

[2] P. Bacchin, A possible link between critical and limiting flux for colloidal system:
Consideration of critical deposit formation along a membrane, J. Membr. Sci. 228
(2004) 237.

[3] W. S. Ho and K. K. Sirkar, Membrane handbook, USA, Kluwer Academic

Publishers, 2001.

[4] A. L. Zydney and C. K. Colton, A concentration polarization model for the filtrate
flux in cross-flow microfiltration of particulate suspensions, Chem. Eng.
Conmmunications, 47 1986 1.

[5] H. Li, A. G. Fane, H. G. L. Coster and S. Vigneswaran, An assessment of

depolarization models of crossflow microfiltration by direct observation through the
membrane, J. Membr. Sci. 172 (2000) 135.

[6] T.H.Chong, A.G.Fane and F.S.Wong, Implications of critical flux and cake enhanced
osmotic pressure (CEOP) on colloidal fouling in RO: experimental observations,
(submitted to J.Membr.Sci.).

[7] H. Winters, Personal communication

[8] W. R. Bowen, and Q. Gan, Microfiltration of protein solutions at thin film

composite membranes, J. Membr. Sci. 80 (1993) 165.

[9] W. D. Mores and R. H. Davis, Direct visual observation of yeast deposition and
removal during microfiltration, J. Membr. Sci. 189 (2001) 217.

[10] Z.F.Cui, S., Chang and A.G. Fane, The use of gas bubbling to enhance membrane
processes, J.Memb.Sci., 221 ( 1-2 ), 1-35 ( 2003 ).

Page 24