Journal of Hazardous Materials 244–245 (2013) 387–393

Contents lists available at SciVerse ScienceDirect

Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat

Study of fouling and scaling in capacitive deionisation by using dissolved

organic and inorganic salts
Mohamed Mossad, Linda Zou ∗
SA Water Centre for Water Management and Reuse, University of South Australia, Adelaide, SA 5095, Australia

h i g h l i g h t s

 Dissolved organic matter (humic acid) is the main cause of CDI electrode fouling.
 CDI removal efficiency and its production rate decreases by increasing feed TOC.
 CDI energy consumption (kWh/m3 ) increases by increasing feed TOC.
 Ferric ions seemed to be a great contributor to CDI scaling.
 Alkaline and acid cleaning solutions were able to completely restore the recovery of the CDI.

a r t i c l e i n f o a b s t r a c t

Article history: In this work, fouling, scaling and cleaning of the capacitive deionisation (CDI) with activated carbon elec-
Received 1 August 2012 trodes were systematically investigated for the first time. Electrode fouling caused by dissolved organic
Received in revised form matter using sodium salt of humic acid as a model foulant (measured by total organic carbon concentra-
28 November 2012
tion, TOC) and inorganic salt (NaCl, MgCl2 , CaCl2 and FeCl3 ) in the CDI feed solutions was investigated
Accepted 28 November 2012
in a series of controlled fouling experiments. After each CDI experiment, a series of cleaning steps was
Available online 7 December 2012
performed to understand the reversibility of fouling accumulated on the electrode surface by analysing
the cleaning solutions. The higher the TOC concentration in the CDI feed solution, the more the reduction
Keywords:
Capacitive deionization
of salt removal efficiency, declination in the production rate and energy consumption. Dissolved organic
Fouling matter is the main cause of electrode fouling, as it blocks the activated carbon pores and reduces their
Total organic carbon electrosorption capacitance. Ca and Mg have no noticeable effect on the CDI treatment performance.
Energy consumption However, Fe seemed to have a greater effect on CDI electrode fouling. Alkaline and acid cleaning solu-
Humic acid tions were able to restore the recovery of the CDI performance from fouling. Pre-treatment to reduce the
dissolved organic matter levels is recommended to achieve sustainable treatment performance.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction energy-efficient desalination technology because it operates at a

relatively low electrical voltage (typically 0.8–2.0 V) at which no
Water shortage has become one of the major problems in the electrolysis reaction occurs [5,6]. Other advantages of using CDI
world. The global water crisis has drawn much world attention on technology include no need for pressure-driven membranes or
finding sustainable ways to produce fresh drinking water. Many high-pressure pumps, simplicity of design, low cost, and operation
researchers have indicated that water will be the oil of the 21st at ambient conditions [7–9]. The CDI energy consumption is about
century [1–3]. Scarcity of fresh water may lead to warfare, as oil three times lower than that of the electrodialysis reversal process
did in the past. About 98 per cent of the planet’s water is either sea [10]. The production cost of CDI is about three times lower than that
water or brackish water. Therefore, we have to explore novel, more of the low-pressure RO process [11]. The CDI process cycle consists
efficient and energy effective ways to purify salty water. Currently, of two phases; namely, purification and regeneration. During the
most of the water desalination technologies are energy and cost purification phase, an electric potential is applied to electrodes,
intensive and can cause secondary pollution [4]. which forces charged ions in the electrolyte solution to migrate
Capacitive deionisation (CDI) technology stands out as one of towards oppositely charged electrodes, to be held in the electri-
the most useful tools for brackish water desalination. CDI is an cal double layer. Regeneration then takes place by reversing the
applied potential to expel the adsorbed ions into the waste stream
[12,13].
∗ Corresponding author. Tel.: +61 8 83202 5489; fax: +61 8 8302 3386. Although much research has been conducted on CDI, issues
E-mail address: [email protected] (L. Zou). such as fouling and scaling remain to be more thoroughly

0304-3894/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jhazmat.2012.11.062
388 M. Mossad, L. Zou / Journal of Hazardous Materials 244–245 (2013) 387–393

Solenoid valve
Sand cartridge Purified water
filter
Feed
water Waste water
reservoir
Pump
Solenoid valve
Two CDI cells connected in series

Fig. 1. Schematic diagram of the CDI pilot plant unit.

investigated before CDI can be widely accepted in applications. and the power supply is off, followed by another 30 s of flushing
Most research efforts have focused on developing novel electrode when the effluent waste valve and the influent valve are opened
materials and enhancing their properties [14–17]. CDI technology and the power is turned on with the opposite polarity of 1.5 VDC.
has also been used to remove inorganic salts from different types of In the purification stage, the influent valve and the effluent purifica-
water [7,8,11,18–21]. Until now, there has been a dearth of research tion valve are opened and the power is turned on with the polarity
reporting the effect of dissolved organic matter in brackish water of 1.5 VDC. The CDI unit included two electrode cells connected in
on CDI operation performance and treated water quality. Humic series (Fig. 1). Each cell assembly contained 100 pairs of activated
acid is one of the dissolved organic matter which is abundant in carbon electrodes with dimensions of 158 mm × 174 mm × 0.3 mm
groundwater, natural water and wastewater [22–24]. Humic acid and a total mass of 1354 grams of activated carbon. The CDI pro-
is widely used as a typical model compound to form organic foul- totype unit and the electrode material is reviewed in more detail
ing layers on membrane or other materials by many membrane and extended in Refs. [29,30]. The fouling and scaling of the CDI
filtration researchers [24–26]. No previous study has investigated cells were assessed by observing the decrease in permeate flow
the organic fouling of CDI electrodes and the effect of this on rate and electrosorption removal efficiency during the purification
CDI performance. The scaling problems commonly associated with cycle, and by comparing the CDI operational performance using
membrane technologies may be avoided using CDI [7,12]. The most various feed solution compositions. For each feed composition, the
common scalants for reverse osmosis technology are Ca, Mg and Fe CDI unit was operated continuously for 30 h using various feed
[27,28]. CDI scaling and fouling may result in a decline in the qual- solutions followed by as series of cleanings steps.
ity of purified water and a reduced production rate [5]. Therefore,
it is necessary to study the effect of dissolved organic and inorganic
compounds on the treatment efficiency of CDI caused by the accu- 2.2. Experimental methods
mulation of foulants on the activated carbon electrodes. Further,
it is essential to identify effective cleaning methods for the elec- A series of laboratory experiments was conducted to investi-
trodes, to remove fouling caused by different types of foulants. Such gate the effect of feed solution composition on CDI performance;
an improvement in the understanding of fouling issues for CDI’s is that is, the permeates flow rate, electrosorption removal efficiency
very timely, to employ CDI as a practical desalination technology and electrical energy consumption. Three different feed solutions
in fresh water augmentation. (F1, F2 and F3), as shown in Table 1, were used to identify the
The objectives of this study are to understand the effect of the organic fouling potential of the CDI, while other two feed solutions
different foulants in feed solution on CDI activated carbon elec- (F4 and F5) were used to investigate the scaling potential of the
trodes and to identify the composition of the foulants accumulated unit. F6 solution was used to investigate the effect of both organic
on the surface of the electrodes during the treatment of brackish and inorganic salts together on the CDI operational performance.
water. In this study, investigation was conducted using a proto- The chemicals used to prepare the CDI feed solutions with vari-
type CDI unit with the capacity of 1000–2000 L/day at laboratory. ous compositions were sodium chloride (NaCl), calcium chloride
To investigate the effective cleaning method for CDI electrodes, (CaCl2 ), magnesium chloride (MgCl2 ), ferric chloride (FeCl3 ) and
a series of specific cleaning protocols was conducted. The used sodium salt of humic acid. In this study, humic acid sodium salt as
cleaning solutions in each cleaning step were collected and ana- a model compound of dissolved organic matter was added in the
lysed to identify the major foulants of the CDI electrodes. The synthetic feed solution. In order to determine the concentration of
effect of CDI fouling on the treatment efficiency, process flow rate the dissolved organics in the treated water, instead of using other
and energy consumption was determined as well. This investiga- analytical method, such as HPLC, total organic carbon (TOC) anal-
tion provides valuable information on operation and maintenance yser was used to measure the concentration of dissolved organics.
regarding potential fouling of the CDI. Sodium salt humic acid was obtained from Sigma–Aldrich. Sodium
salt humic acid has an average molecular weight ranged from 4000
to 23,000 [31]. The total acidity is 5 m mol/g, with a carboxylic acid-
2. Experimental ity value of 3.4 m mol/g [24]. The solution chemistry of feed solution
of humic acid is described in more detail and in Refs. [24,31,32]. The
2.1. The CDI pilot plant pH of all the feed solutions was in the range of 6.7–7.1, which is
close to groundwater pH in previous studies [30,33]. No pH adjust-
Fig. 1 shows the combined pretreatment-CDI unit used in this ment for the feed synthetic solution was carried out through the
study. The CDI cells (Aqua EWP, USA) were comprised of porous whole study. The effect of organic fouling and scaling of CDI elec-
activated carbon electrodes with a specific area of 800 m2 /g; a flow trodes was reflected as a declining of flow rate and salt removal
rate of 2 L/min was used during all the experiments. The operat- efficiency. The flow rate of the CDI unit was continuously monitored
ing direct current voltage of the CDI process was 1.5 V. The process throughout the whole experiment period. Samples of both puri-
cycle was controlled by a programmable logic controller. Each cycle fied and waste stream were analysed periodically. The conductivity
consisted of a purification stage lasting 1.5 min, and a regeneration and total dissolved solids (TDS) were measured using an electri-
stage lasting 1 min. The regeneration step commences with 30 s of cal conductivity metre (HACH, HQ40d digital metre). pH values
flushing, when the effluent valve and the influent valve are closed were measured using a PHM201 device. A Cary 100 Bio ultraviolet
 M. Mossad, L. Zou / Journal of Hazardous Materials 244–245 (2013) 387–393 389

Table 1
Water quality of CDI feed solutions.

Parameters Feed solutions

F1 (Na+ F2 (Na+ + low F3 (Na+ + high F4 F5 F6

only) HA) HA) (Na+ + Ca2+ + Mg2+ ) (Na+ + Ca2+ + Mg2+ + Fe3+ ) (Na+ + Ca2+ + Mg2+ + Fe3+ + HA)

pH 7.1 6.9 6.8 6.9 6.9 6.7

Conductivity (␮S cm−1 ) 2016 2022 2031 2020 2026 2034
TDS (mg/L) 1000 1010 1016 1006 1012 1020
TOC (mg/L) – 0.9 3.1 – – 3.1
Na+ (mg/L) 392 395 402 246 244 243
Ca2+ (mg/L) – – – 75.2 75.8 73.1
Mg2+ (mg/L) – – – 57.8 58.1 55.2
Fe3+ (mg/L) – – – – 1.8 1.8
Cl− (mg/L) 605 608 606 620 625 628
HA salt (mg/L) – 3 10 – – 10

spectrophotometer (VARIAN) was used to analyse the peak absorp- the adsorbed TDS from the activated carbon electrodes. The results
tion of humic acid at a wavelength of 254 nm. TOC concentration indicate that sodium and chloride ions can be completely desorbed
was analysed using a TOC analyser (GE Sievers innovox TOC, USA). during the regeneration period and do not cause any accumulative
The concentrations of cation were measured by means of an Ana- effect on CDI treatment performance.
lyst 400 atomic absorption spectrophotometer (Perkin Elmer). The When salt of humic acid was added to the feed solution to
analytical method and procedure followed the standard methods maintain a TOC concentration of around 1 mg/L (F2), the water pro-
[34]. The energy consumption (kWh) of the CDI unit was measured duction rate and the electrosorption removal efficiency started to
using a power metre (Todae, MS6115). decline after about 17 h of operation. The CDI purified stream TDS
In our experiment, the electrosorption removal efficiency (Eq. concentration increased by 5.7 per cent after 30 h of operation. It
(1)) and water recovery rate (Eq. (2)) were calculated as follows: was noticed that the produced flow rate decreased gradually with
time from 2 L/min to 1.8 L/min. Compared with the previous results
(C0 − C)
Electrosorption removal efficiency (%) = × 100 (1) in the case of using F1 as the feed solution, dissolved organic matter
C0
had an adverse effect on CDI performance throughout the opera-
where C0 and C (mg/L) represent the feed and treated TDS concen- tion period. The reduction in the salt removal efficiency and CDI
trations, respectively.
VP
Water recovery rate (%) = × 100 (2) 100
VT
where VP and VT (L) represent the purified water volume and total 90
Removal efficiency (%)

water volume including both the purified and waste water volume,
respectively.
80
P
Electrical energy consumption (kWh per m3 ) = (3)
V
70 F1 F2
where P and V represent the electrical power energy consumed by
the CDI unit during 30 h of operation measured by the power metre
in kWh and the purified water volume produced during the 30 h in 60 F3 F6
m3 , respectively.
In this study, a cleaning operation of the fouled CDI unit was 50
conducted after each 30 h of continuous operation. Each cleaning 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
operation was conducted using the following procedure: firstly by Time (h)
deionised water, followed by 0.1 N hydrochloric acid solution, and
Fig. 2. Electrosorption removal efficiency of CDI unit with time using different feed
finally by 0.1 N sodium hydroxide solution. The CDI flow rate was solution compositions.
measured before and after each cleaning step to calculate the flow
rate recovery. The spent cleaning stream was analysed to determine
2.5
the main foulants and scalants accumulated on the CDI unit.

3. Results and discussion

Flow rate (L/min)

2
3.1. Organic fouling in capacitive deionisation

Figs. 2 and 3 show the electrosorption removal efficiency and

the permeate flow rate of the CDI unit with time for four differ- 1.5 F1 F2
ent feed solutions (F1, F2, F3 and F6). When the CDI unit was fed
with solution F1, containing NaCl only, no reduction in the TDS F3 F6
electrosorption removal efficiency or the produced flow rate was
observed at any point in the operation period. The electrosorption 1
removal efficiency averaged 87.7 per cent for the whole opera- 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Time (h)
tion cycle. The permeate flow rate remained 2 L/min for the 30-h
operational period (Fig. 3). Mass balance calculations show that the Fig. 3. Product flow rate of CDI unit with time when processing different feed solu-
regeneration cycle was able to remove more than 99.9 per cent of tions.
390 M. Mossad, L. Zou / Journal of Hazardous Materials 244–245 (2013) 387–393

treated water productivity was unnoticeable using a feed solution 100

with TOC around 1 mg/L. This is confirmed by the reduction of TOC
concentration in the CDI treated and waste stream, which indicates
the possible accumulation of some organic matter in the pores

Removal efficiency (%)

90
of the activated carbon electrodes, and which are not removed
during the regeneration period. These results show that dissolved
organic matter contained in feed water could cause organic foul- 80
ing of the CDI electrodes during long-term operation. However,
the brackish groundwater in inland areas often contains very low F1 F4
or non-detectable levels of TOC; as a result, the effect will not be 70
F5
significant.
When the CDI feed stream contains higher TOC concentra-
tion (humic acid salt) up to 3.1 mg/L (F3), the CDI electrosorption 60
removal efficiency and production rate were reduced significantly 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
(Figs. 2 and 3). The flow rate declined drastically after 10 h of oper- Time (h)
ation. The TDS removal efficiency decreased gradually from 85.7
Fig. 4. Electrosorption removal efficiency of CDI unit with time of operation when
per cent to 70.2 per cent after a 30-h operational period. The pro- processing different feed solutions.
cess flow rate was reduced by 35 per cent from the initial flow rate
over the course of the whole operation period. It was found that
during the regeneration step, no dissolved organic matter was des- removal efficiency of about 2 per cent was observed, as compared
orbed. It was noticed that the amount of dissolved organic matter with the initial removal efficiency. No declination in the produced
retained on the CDI electrodes during the purification and regen- flow rate was observed at any point during the operation period.
eration cycle was 35% and 51%, respectively. It is clear that the The regeneration cycle was able to remove 99.3 and 99.5 per cent
adsorption of the dissolved organic matter during the electrosorp- of the adsorbed Ca and Mg ions from the activated carbon elec-
tion stage was lower than that of the regeneration stage. During the trodes, respectively, according to the mass balance calculations. It is
purification stage, an electrical potential was applied between the clear that Ca and Mg (the two most prevalent divalent metal ions in
CDI electrodes, a competition between inorganic ions and humic groundwater) have no noticeable effect on CDI operational perfor-
acid to occupy the available adsorption sites on the electrodes mance with time. When ferric ions were added to the feed solution
surface occurred. This limits the adsorption of humic acid onto with a concentration of 1.8 mg/L (F6), the production flow rate and
the electrodes surface because of the organic matter often weakly the salt removal efficiency started to decline. The salt removal effi-
charged [18]. On the other hand during the first 30 s in the regen- ciency decreased gradually with time from 88.4 per cent to 81 per
eration stage no voltage was applied between the electrodes, the cent. Moreover, the permeate flow rate decreased by 200 mL/min
inorganic salts started to be desorbed from the electrodes, while throughout the 30-h operational period. Mass balance calculations
humic acid was still retained on the electrodes. In this stage more show that 30 per cent iron was accumulated on the activated car-
adsorption sites is available on the electrodes for humic acid to bon electrodes every CDI cycle. This indicates that ferric ions have
adsorbed because most of the inorganic ions are desorbed. In addi- a great impact on long-term CDI treatment efficiency and its pro-
tion, more humic acid was adsorbed on the electrodes during this ductivity throughout the operation period. Scaling might occur by
stage by hydrophobic interactions without any competition with precipitation of iron hydroxide on the activated carbon electrodes
other ions on the adsorption sites. The reduction of TDS removal instead of electrosorbed on the electrodes, since the pH of the feed
efficiency may be related to the initial surface area loss of the elec- solution encourages the formation of ferric hydroxide.
trodes covered with organic materials and the clogged the pores
of the activated carbon material [35]. It is clear that the accumu- 3.3. Combinations of organic fouling and scaling in capacitive
lated organic matter on the CDI electrodes has a great impact on the deionisation
electrosorption capacitance of the activated carbon electrodes [36].
One possible reason for this is that organic matter is adsorbed on the CDI operational performance was also affected by a feed solu-
CDI electrodes by chemical sorption and hydrophobic interactions tion containing NaCl, CaCl2 , MgCl2 and FeCl3 together with humic
into the activated carbon electrodes, rather than by electrosorption acid (F6). Interestingly, the change was similar to, but slightly more
because of the weakly charged characteristics of the organic mat- than, that caused by a feed solution of NaCl only with humic acid
ter [18]. Therefore, organic matter is not desorbed by reversing the
applied voltage during the regeneration cycle because the adsorp-
2.5
tion bond may be strong enough to avoid broken by reversing the
electric potential during the regeneration stage. It is clear that com-
petition occurred between the inorganic salts and the humic acid 2
to be adsorbed on the activated carbon electrodes; this depended
Flow rate (L/min)

mainly on the concentration of dissolved organic matter in the feed 1.5

solution. It can be concluded that the higher the concentration of
organic matter in the feed stream of CDI, the more pronounced the no Fe
1
reduction of TDS removal efficiency and the produced flow rate 1.8 mg/L Fe
occurring throughout the operation period.
0.5
3.2. Scaling in capacitive deionisation
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
The dependence of salt removal efficiency and permeate flow
Time (h)
rate reduction of the CDI unit with time on the composition of feed
stream (F1, F4 and F5) are shown in Figs. 4 and 5, respectively. Fig. 5. Product flow rate of CDI unit with time when processing feed solutions with
When the CDI unit was fed with F5, a slight reduction in the TDS or without Fe.
 M. Mossad, L. Zou / Journal of Hazardous Materials 244–245 (2013) 387–393 391

100 2.5
F3
90 2 F4
Removal efficiency (%)

Flow rate (L/min)

80
1.5
70
Na Ca
1
60
Mg Fe
0.5
50

40 0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Fouled Hydraulic cleaning Acid cleaning Alkaline cleaning
Time (h)
Fig. 8. Process flow rate of CDI unit after cleaning sequence in case of using feed
Fig. 6. Cations electrosorption removal efficiency of the CDI unit with time of oper- solutions F3 and F4.
ation using feed solution F6.

F4, F5 and F6 was 3.25, 3.68, 4.52, 3.28, 3.33 and 4.56 kWh/m3 ,
(F3). The TDS removal efficiency decreased gradually from 86.8 per respectively. It can be seen that as the dissolved organic matter
cent to 67 per cent after the 30-h operational period (Fig. 2). It was concentration increases in the feed solution, the energy consump-
also noticed that the produced flow rate decreased gradually with tion also increases. By comparing the results obtained using F3
time from 2 L/min to 1.2 L/min (Fig. 3). As compared to the reduc- and F4 as the feed, it is clear that the presence of organic mat-
tion that occurred in the experiment using F3 as a feed solution, ter in the feed stream, rather than other ions, is the main reason
the permeate flow rate and TDS removal efficiency were further for this increase in CDI energy consumption. A possible reason for
reduced by 5 and 2.5 per cent, respectively. This indicates that Ca, this is the accumulation of organic matter on the activated car-
Mg and Fe have a slight additional impact on CDI treatment effi- bon electrode surface, which increases electrode resistance, thus
ciency and its productivity. It was noticed that Na, Ca and Mg were increasing CDI energy consumption. The reduction of flow rate with
effectively desorbed from the activated carbon electrodes during time affected the energy consumption as measured in kWh/m3 , as
the CDI regeneration cycle. It can be concluded that the presence of the treated amount of water per hour was lower at low flow rates
dissolved organic matter in the feed water is the main factor con- [29]. In conclusion, for feed streams with high concentrations of dis-
tributing to the deterioration of the activated carbon electrodes. solved organic matter, pre-treatment is recommended to remove
The results indicated that CDI feed water with dissolved organic dissolved organic matter. This will increase the sustainability of
matter appears to reduce the inorganic sorption efficiency of the the operational performance and treatment efficiency of the CDI
activated carbon electrodes, as shown in Fig. 6. The electrosorption process.
removal efficiency was reduced by 9.6 per cent for Na, 8.1 per cent
for Ca, 9.6 per cent for Mg and 9.3 per cent for Fe over the course 3.5. CDI process flow recovery by cleaning
of the operation period. The decrease in the inorganic salts adsorp-
tion could be attributable to pore blockage, competition with humic A series of cleaning steps (hydraulic, acid and alkaline solution
acid, or a combination of the two. These results are consistent with cleaning) was employed after each fouling experiment. The flow
the results reported by Narbaitz and Benedek [37] and Lee et al. rate recovery was evaluated after each cleaning step. Fig. 8 shows
[19]. The recovery ratio for all previous experiments was observed the CDI process flow rate recovery after each cleaning step for two
to be unchanged at 75 per cent throughout the operation period. cleaning experiments after using F3 and F6 as a feed solution. In the
This may be related to the weakly charged nature of organic matter case of using F3 as a feed solution, hydraulic and acid cleaning could
[18]. not recover the flow rate significantly. However, alkaline cleaning
was able to recover the flow rate to the initial flow rate of 2 L/min. It
3.4. Effect of CDI fouling and scaling on energy consumption was noticed that hydraulic, acid and alkaline cleaning were able to
recover the flow rate by 14.3, 0 and 85.7 per cent, respectively, by
Fig. 7 shows the effect of feed solution composition and concen- removing the accumulated organic matter on the activated carbon
trations on the electrical energy consumption of CDI in kWh/m3 . electrodes. These experimental results suggest that, as organic mat-
The energy consumption of CDI using a feed solution of F1, F2, F3, ter was the main foulant for the activated carbon electrode found in
this study, the process flow can be completely recovered by alkaline
cleaning. It can be concluded that, in the case of treating brackish
6
water with high dissolved organic matter concentrations, alkaline
Energy consumption (Kwh/m3 )

5 cleaning may be required after some time of operation, so as to

maintain the operational performance and treatment efficiency of
4 the CDI unit.
In the case of feeding the CDI unit with F6, the flow rate was
3 more effectively recovered after acid cleaning, as comparing to the
previous experiment. It can be seen that hydraulic, acid and alka-
2
line cleaning were capable of increasing the flow rate by 12.5, 25
1 and 62.5 per cent, respectively, as a result of removing the accu-
mulated organic and inorganic materials on the electrode surface.
0 These results indicate that acid cleaning is effective in removing
F1 F2 F3 F4 F5 F6
inorganic salts retained on the CDI electrodes. It can be concluded
Fig. 7. Energy consumption of the CDI unit when processing different feed solutions that the major cause of severe fouling in the CDI was organic fouling
for 30 h of operation. rather than inorganic fouling; although the latter also contributed
392 M. Mossad, L. Zou / Journal of Hazardous Materials 244–245 (2013) 387–393

Table 2
Concentrations and percentage of organic and inorganic materials in waste cleaning streams.

Cleaning sample Parameter (mg/L (%))

Ca Mg Fe TOC

Hydraulic cleaning 36.6 (35.9) 31.9 (36.2) 10.6 (12.4) 6 (2.4)

Acid cleaning 59 (57.8) 41.8 (47.4) 71.9 (83.9) 34 (13.6)
Alkaline cleaning 6.4 (6.3) 14.5 (16.4) 3.2 (3.7) 210 (84)
Total amount 102 (100) 88.2(100) 85.7 (100) 250 (100)

to the fouling with a small percentage. The results also indicate kWh/m3 . Alkaline cleaning was able to remove most of the organic
that activated carbon electrodes are not susceptible to irreversible matter accumulated on the activated carbon electrodes. Ca and Mg
fouling. (the most prevalent divalent metal ions in groundwater) had no
significant effect on the CDI operational performance in the 30-h
3.6. Analysis of CDI cleaning waste stream operation time, as they desorbed effectively from the electrodes
during the regeneration cycle. Fe seemed to be a great contributor
Table 2 shows the concentration of organic and inorganic mat- to CDI scaling by precipitation on the electrode surface among other
ter and its ratio in each cleaning disposal after using F6 as a feed inorganic salts. Acid cleaning was effective in removing inorganic
solution for the CDI unit. A trace amount of Na was detected salts retained in the CDI cells and enhanced the treatment perfor-
in the hydraulic waste stream in this test. The alkaline cleaning mance. It was found that activated carbon CDI electrodes were not
was the most effective at removing the dissolved organic matter susceptible to irreversible organic fouling and inorganic scaling in
from the activated carbon electrodes, while the acid cleaning was this study. Biofouling was not included in the scope of this study.
more effective for removing Fe, Ca and Mg. This indicates that acid Pre-treatment steps to remove organic matter from brackish water
cleaning and alkaline cleaning are recommended in cases of inor- streams may be required to reduce the fouling potential of the
ganic salts and organic matter accumulating on the CDI electrodes, CDI, especially when there is a high organic matter concentration.
respectively. This is because inorganic salts dissociate easily in acid CDI technology could be a promising alternative for brackish water
solutions, while organic matter is better dissolved in alkaline solu- desalination, especially in cases of no or low dissolved organic mat-
tions. Hydraulic cleaning was able to remove an average of 16.2 ter concentrations.
per cent of the accumulated inorganic and organic matter in the
CDI cells. This indicates that a percent of the ions retained on the Acknowledgements
electrodes may be caused by slow kinetics of ion desorption from
the activated carbon surface especially in the case of Ca and Mg. The authors acknowledge the financial support of the National
It can be seen that organic matter represents the most accumula- Centre of Excellence in Desalination, which is funded by the Aus-
tive material on the carbon surface as compared to inorganic salts; tralian Government through the Water for the Future initiative and
this ensures that organic matter is the main foulant for the CDI Water Quality Research, Australia. The author, Mohamed Mossad,
electrodes. Further, the amount of Ca and Mg accumulated on the also appreciates the PhD scholarship provided by the University of
activated carbon electrode surface was slightly greater than the South Australia.
accumulated Fe, which is consistent with the inorganic composi-
tion in the feed stream (see Table 1). Although the composition References
ratio of inorganic and organic matter in the feed solution was 54.9
per cent Ca, 41.4 per cent Mg and 1.3 per cent Fe and 2.4 per cent [1] P. Simon, Tapped Out: The Coming World Crisis in Water and What We Can Do
TOC, the composition ratio in the cleaning stream was 19.4 per cent About it, Welcome Rain, New York, 1998.
[2] C.A.S. Hall, J.W. Day, Revisiting the limits to growth after peak oil, Am. Sci. 97
Ca, 16.8 per cent Mg, 16.3 per cent Fe and 47.5 per cent TOC. From (2009) 230–237.
this, it is clear that organic matter comprises the bulk of the mate- [3] M.A. Anderson, A.L. Cudero, J. Palma, Capacitive deionization as an electro-
rial accumulated on the carbon surface and that, of the inorganic chemical means of saving energy and delivering clean water. Comparison to
present desalination practices: will it compete? Electrochim. Acta 55 (2010)
matter, Fe was the most accumulated on the CDI cell electrodes.
3845–3856.
[4] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.M.
4. Conclusions Mayes, Science and technology for water purification in the coming decades,
Nature 452 (2008) 301–310.
[5] Y. Oren, Capacitive deionization (CDI) for desalination and water
This research has investigated fouling and scaling of a small CDI treatment—past, present and future (a review), Desalination 228 (2008)
prototype unit using various feed solutions with different compo- 10–29.
[6] P.M. Biesheuvel, B. van Limpt, A. van der Wal, Dynamic adsorption/desorption
sitions. The CDI unit was able to remove inorganic salts from water process model for capacitive deionization, J. Phys. Chem. C 113 (2009)
sufficiently. The CDI operational and treatment efficiency through- 5636–5640.
out the operation period depended mainly on the composition and [7] P. Xu, J.E. Drewes, D. Heil, G. Wang, Treatment of brackish produced water using
carbon aerogel based capacitive deionization technology, Water Res. 42 (2008)
concentration of the dissolved organic matter in the feed stream. 2605–2617.
The CDI achieved a water recovery ratio of 75 per cent over the [8] S.J. Seo, H. Jeon, J.K. Lee, G.Y. Kim, D. Park, H. Nojima, J. Lee, S.H. Moon, Investi-
course of the operation period. CDI show a low fouling potential gation on removal of hardness ions by capacitive deionization (CDI) for water
softening applications, Water Res. 44 (2010) 2267–2275.
in cases of feed solution with no or low dissolved organic matter
[9] Y. Kim, J. Choi, Enhanced desalination efficiency in capacitive deionization with
concentrations. A high concentration of dissolved organic matter an ion-selective membrane, Sep. Purif. Technol. 71 (2010) 70–75.
in brackish water had a great effect on CDI treatment performance, [10] American Water Works Association (AWWA) Manual, Reverse Osmosis and
Nanofiltration, AWWA, Colorado, 1999.
as it reduced the salt removal efficiency and the permeate flow rate
[11] T.J. Welgemoed, C.F. Schutte, Capacitive deionization technologyTM : an alter-
with operation time. Organic matter accumulated in the pores of native desalination solution, Desalination 183 (2005) 237–340.
the activated carbon electrodes of the CDI cells did not respond to [12] L. Zou, G. Morris, D. Qi, Using activated carbon electrode in electrosorptive
the desorption by reversing the applied voltage during the regener- deionisation of brackish water, Desalination 225 (2008) 329–340.
[13] Z. Chen, C. Song, X. Sun, H. Guo, G. Zhu, Kinetic and isotherm studies on the
ation cycle. The higher the dissolved organic matter concentration electrosorption of NaCl from aqueous solutions by activated carbon electrodes,
in the CDI feed solution, the higher the CDI energy consumption in Desalination 267 (2011) 239–243.
 M. Mossad, L. Zou / Journal of Hazardous Materials 244–245 (2013) 387–393 393

[14] Y. Gao, L. Pan, Y. Zhang, Y. Chen, Z. Sun, Electrosorption of FeCl3 solutions with [26] A. Seidel, M. Elimelech, Coupling between chemical and physical inter-
carbon nanotubes and nanofibers film electrodes grown on graphite substrates, actions in natural organic matter (NOM) fouling of nanofiltration mem-
Surf. Rev. Lett. 14 (2007) 1033–1037. branes: implications for fouling control, J. Membr. Sci. 203 (2002)
[15] H. Li, Y. Gao, L. Pan, Y. Zhang, Y. Chen, Z. Sun, Electrosorptive desalination by 245–255.
carbon nanotubes and nanofibers electrodes and ion-exchange membranes, [27] H.L. Yang, C. Huang, J.R. Pan, Characteristics of RO foulants in a brackish water
Water Res. 42 (2008) 4923–4928. desalination plant, Desalination 220 (2008) 353–358.
[16] L. Zou, L. Li, H. Song, G. Morris, Using mesoporous carbon electrodes for brackish [28] E. Gwona, M. Yub, H. Ohb, Y. Yleea, Fouling characteristics of NF and RO oper-
water desalination, Water Res. 42 (2008) 2340–2348. ated for removal of dissolved matter from groundwater, Water Res. 37 (2003)
[17] L. Li, L. Zou, H. Song, G. Morris, Ordered mesoporous carbons synthesized by 2989–2997.
a modified sol–gel process for electrosorptive removal of sodium chloride, [29] M. Mossad, L. Zou, A study of the capacitive deionisation performance
Carbon 47 (2009) 775–781. under various operational conditions, J. Hazard. Mater. 213–214 (2012)
[18] C.J. Gabelich, T. Tran, I.H. Suffet, Electrosorption of inorganic salts from aqueous 491–497.
solution using carbon aerogels, Environ. Sci. Technol. 36 (2002) 3010–3019. [30] M. Mossad, W. Zhang, L. Zou, Using capacitive deionisation for inland brack-
[19] Y. Lee, H.Y. Ng, S.L. Ong, G. Tao, K. Kekre, B. Viswanath, W. Lay, H. Seah, ish groundwater desalination in a remote location, Desalination 308 (2013)
Integrated pre-treatment with capacitive deionization for reverse osmosis 154–160.
reject recovery from water reclamation plant, Water Res. 43 (2009) 4769– [31] Y.P. Chin, G. Aiken, E. Oloughlin, Molecular weight, polydispersity, and spectro-
4777. scopic properties of aquatic humic substances, Environ. Sci. Technol. 28 (1994)
[20] H. Li, L. Zou, L. Pan, Z. Sun, Using graphene nano-flakes as electrodes to remove 1853–1858.
ferric ions by capacitive deionization, Sep. Purif. Technol. 75 (2010) 8–14. [32] A.W.P. Vermeer, W.H. van Riemsdijk, L.K. Koopal, Adsorption of humic acid to
[21] G. Tao, B. Viswanath, K. Kekre, L.Y. Lee, H.Y. Ng, S.L. Ong, H. Seah, RO brine treat- mineral particles: 1. Specific and electrostatic interactions, Langmuir 14 (1998)
ment and recovery by biological activated carbon and capacitive deionization 2810–2819.
process, Water Sci. Technol. 64 (2011) 77–82. [33] D.C. Koh, L.N. Plummer, D.K. Solomon, E. Busenberg, Y.J. Kim, H.W. Chang,
[22] J.J. Alberts, Z. Filip, N. Hertkorn, Fulvic and humic acids isolated from ground- Application of environmental tracers to mixing, evolution, and nitrate con-
water: compositional characteristics and cation binding, J. Contam. Hydrol. 11 tamination of ground water in Jeju Island, Korea, J. Hydrol. 327 (2006)
(1992) 317–330. 258–275.
[23] G.G. Leppard, D. Mavrocordatos, D. Perret, Electron-optical characterization of [34] APHA–AWWA–WPCF, Standard Methods for the Examination of Water and
nano- and micro-particles in raw and treated waters: an overview, Water Sci. Wastewater, 20th ed., APHA–AWWA–WPCF, 1998.
Technol. 50 (2004) 1–8. [35] G. Newcombe, M. Drikas, S. Assemi, R. Beckett, The influence of characterised
[24] C.Y. Tang, Y.N. Kwon, J.O. Leckie, Fouling of reverse osmosis and nanofiltration natural organic material on activated carbon adsorption: I. Characterisation of
membranes by humic acid—effects of solution composition and hydrodynamic concentrated reservoir water, Water Res. 31 (1997) 965–972.
conditions, J. Membr. Sci. 290 (2007) 86–94. [36] I.N. Najm, V.L. Snoeyink, Y. Richard, Effect of initial concentration of a SOC in
[25] A.I. Schafer, A. Pihlajamaki, A.G. Fane, T.D. Waite, M. Nystrom, Natural organic natural water on its adsorption by activated carbon, J. Am. Water Works Assoc.
matter removal by nanofiltration: effects of solution chemistry on retention 83 (1991) 57–63.
of low molar mass acids versus bulk organic matter, J. Membr. Sci. 242 (2004) [37] R.M. Narbaitz, A. Benedek, Adsorption of 1,1,2-trichloroethane from river
73–85. water, J. Environ. Eng. 120 (1994) 1400–1415.