food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120

Contents lists available at ScienceDirect

Food and Bioproducts Processing

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

Production of colorless liquid sugar by

ultraltration coupled with ion exchange

Heru Susanto a, , Anis Roihatin b , I. Nyoman Widiasa a

a Department of Chemical Engineering, Faculty of Engineering, Diponegoro University Jl., Prof. Soedarto-Tembalang,
Semarang, Indonesia
b Convertion Energy Study Program, Semarang State Polytechnic, Indonesia Jl., Prof. Soedarto-Tembalang,

Semarang, Indonesia

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

Article history: Liquid sugar is an alternative product to crystal sugar, which is currently being considered by
Received 26 June 2015 many sugar companies. However, production of colorless liquid sugar is still a big challenge.
Received in revised form 20 This work aimed to study the performance of ultraltration membrane and ion exchange
November 2015 (IE) both as single and integrated processes for decoloring cane sugar solution. The exper-
Accepted 2 December 2015 iments were performed by using 10, 20, 50, and 100 kDa commercial UF membranes. Two
Available online 13 December 2015 commercial resins were used in ion exchange experiments. The results of decoloring sugar
solution using UF membranes showed that the 20 kDa UF membrane made of polysulfone
Keywords: displays the best performance in term of ux and color removal. Integrating ion exchange
Sugar decolorization process into UF membrane improves the color reduction resulting colorless liquid sugar.
Fouling The UFIE conguration shows a better performance than both the IEUF and UF/IE alone
Ion exchange congurations. These results provide important information about technologies that should
Liquid sugar be used for liquid sugar production.
Ultraltration 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Process conguration

1. Introduction and a low concentration of non-sugar. It is not easy to ful-

Sugar is one of the commodities that have a strategic role,
because it is one of the basic needs of domestic and indus-
tries for food, beverage and pharmaceutical. Due to energy
saving, liquid sugar is currently being considered to replace
partly crystal sugar products. It is known that in most of
the food or beverage industries crystal sugar is dissolved in
water before use. The energy for dissolution can be avoided
if liquid sugar is used. Further energy used for crystalliza-
tion in crystal sugar industry can also be avoided if liquid
sugar is produced. It was reported that crystallization is one of
the most energy-intensives stages in crystal sugar industries
(Chen and Chou, 1993; Maravic et al., 2015; Wang, 2009). How-
ever, production of liquid sugar needs technologies that can
produce colorless sugar solution with high content of sugar
l those requirements because (i) sugar cane juice contains
various components, of which only sugar is desirable, (ii) the
juice extracted from sugarcane are naturally yellowbrown,
(iii) many macromolecules, small molecules and minerals
present in sugar cane juice that should be removed, (iv) the
possibility of color increasing during processing as a result of
browning reactions, and (v) the increase in viscosity of sugar
solution as increase in sugar concentration causing a prob-
lems in purication (Balakrishnan et al., 2000a,b; Chen and
Chou, 1993; Hinkova et al., 2002; Jegatheesan et al., 2009; Saha
et al., 2006).
Many technologies have been developed to increase the
performance of crystal sugar industry or cane sugar renery,
i.e., either to increase purity of sugar or to decrease production
cost. One promising alternative is the use of ultraltration (UF)

Corresponding author. Tel.: +62 247460058; fax: +62 247480675.
E-mail addresses: heru.susanto@undip.ac.id, susanto.heru@gmail.com (H. Susanto).
http://dx.doi.org/10.1016/j.fbp.2015.12.002
0960-3085/ 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
12 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120
membrane for sugar cane juice clarication either stand alone
or combined with the coagulation process (Balakrishnan et al.,
2001; Bhattacharya et al., 2001; Donovan et al., 2002; Fechter
et al., 2004; Ghosh et al., 2000; Ghosh and Balakrishnan, 2003).
In general, UF membranes offer (i) increasing the transmission
of sucrose, (ii) improving the quality of sugar, (iii) increasing
the yield of conventional process, and (iv) reducing the chem-
ical and energy used. However, most of previous publications
of application UF membranes are dedicated to production of
crystal sugar, where permeate of ultraltration membrane
will further be processed by crystallization (e.g., Balakrishnan
et al., 2000a,b; Chen and Chou, 1993; Jacob and Jaffrin, 2000;
Mak, 1991). In this context, the color of sugar solution result-
ing from UF permeate is not requested to be at a low level.
Besides its function to form sugar crystal, crystallization pro-
cess can also play as a decolorization process via excluding
colorants from sugar crystals during crystal growth as well as
crystal washing after being separated from the liquor. By con-
trast, no crystallization process is used in production of liquid
sugar.
To produce high quality of liquid sugar, we need technolo-
gies that can eliminate soluble non-sugar macromolecular
and colorants. Removal of colorants from liquid sugar is very
important because color is the most signicant quality deter-
minant. Colorants in sugar cane juice can originate from
native of sugarcane plant such as avonoids and melanins
and those developed during cane juice processing such as
melanoidins and caramels (Flood and Flood, 2006; Hamachi
et al., 2003). This colorization increases during processing as
a result of enzymatic reactions and heat induced procedures
involved. Sugarcane juice contains 35% of soluble solids in
the form of colorants, color precursors, etc. and an additional
0.851.45% of soluble solids in the form of organic non-sugars
such as protein, polysaccharides and waxes (Chen and Chou,
1999). Several techniques can be used for removing color from
the sugar solution including ion exchange (IE), activated car-
bon and UF. Among them, UF and ion exchange seem to be of
highest relevance from practical point of view.
UF membrane has been proposed as an attractive tech-
nology for color removal in production of crystal sugar and
others (Decloux et al., 2000; Hamachi et al., 2003; Jacob and
Jaffrin, 2000; Mak, 1991). However, this technology has low
efciency for removing color originating from very small
macromolecules or minerals. The use of nanoltration mem-
brane may be able to remove such small colorants but it can
also reduce sugar content. Ion exchange (IE) has been applied
in many food industries (Achaerandio et al., 2002; Lyndon,
1996). In principle, this technology can be used for soften-
ing, demineralization and decolorization. Application of ion
exchange for sugar cane juice decolorization was reported in
previous publications (Bento, 1990; Lin and Chen, 1991). How-
ever, the use of IE technology is limited by its high operating
cost due to resin regeneration. In addition, high amount of
wastewater will be resulted from regeneration process.
In summary, UF and IE have been proposed for decolor-
ing cane sugar solution or sugarcane juice. However, most of
previous studies on decolorization of sugar solution were per-
formed using single process, i.e. either UF or IE alone. The
objective of the current study was to examine the performance
of ultraltration membranes and ion exchange for the decol-
orization of sugar solution. Measurements of the permeate
ux and other key parameters of the separation were also
performed to determine the viability of the UF membrane
process. The important investigation was the performance of
combination of both technologies. The process conguration
of UF and IE for colorant removal to obtain colorless sugar
solution was investigated.
2. Materials and methods
2.1. Materials
For identication of suitable membranes, commercial poly-
sulfone (PS) and polyethersulfone (PES) UF membranes with
different pore sizes (represented by MWCO, molecular weight
cut off) and one UF membrane made from uoro polymer
were used in this study. The membranes tested were GR81PP
(10 kDa, polyethersulfone, called PES-10), GR61PP (20 kDa,
polysulfone, called PS-20), GR51PP (50 kDa, polysulfone, called
PS-50), GR40PP (100 kDa, polysulfone, called PS-100), and
FS61PP (20 kDa, uoro polymer, called FP-20). All membranes
were supplied by Alfa Laval, Denmark. The membranes were
initially washed by soaking overnight in water to remove
impurities or additives left from the manufacturing process
before use. Fresh membranes were used in all experiments.
Brown sugar crystal was purchased from the local sugar com-
pany in Indonesia. Sodium hydroxide and hydrochloric acid
were purchased from Merck. Lewatit adsorber resins, i.e., S
6368 (R1) and OC 1074 (R2), were used for ion exchange pro-
cess. S 6368 (R1) is strongly basic macroporous anion exchange
resin (type I) with beads of uniform size (monodisperse) based
on polystyrene. OC 1074 (R2) is a strongly basic macroporous
type I anion exchange resin based on polyacrylamide. It is
bead-shaped and has a special bead size distribution. Dem-
ineralized water from home-made RO-ion exchange system
was used for all experiments.
2.2. Methods
2.2.1. Ultraltration experiments
Ultraltration experiments were performed by crossow l-
tration using a self-home-made laboratory scale similar with
our previous publication (Susanto and Widiasa, 2009). The
set-up consisted of a feed tank, a pump, a pressure indica-
tor connected to feed side and a at-sheet membrane cell
made of stainless steel. The cell has a diameter of 10 cm with
a 4 cm diameter membrane holder. A simplied diagram of
the set-up is given in Fig. 1. In each experiment, a new circu-
lar membrane disk with the effective area of 0.00125 m2 was
used and was rstly compacted by ltering pure water for at
least 0.5 h at a pressure of 3 bars. In order to maintain con-
stant feed concentration, the volume of feed was much larger
than the volume taken as sample for the analysis. In addition,
the retentate and permeate were returned back to the feed
tank. During UF experiments, the ux prole over time was
gravimetrically monitored. All experiments were conducted
at room temperature (28 2 C) and at a constant transmem-
brane pressure (300 kPa for PES-10 and 100 kPa for PS-20, PS-50,
PS-100 and FS-20). The transmembrane pressure drop (the dif-
ference between feed pressure and retentate pressure) was
less than 0.1 bar. The sugar content, color, and absorbance of
feed and permeate were analyzed. The apparent color rejec-
tion was calculated using Eq. (1).
Cd
R=1 (1)
Cu
 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120
Fig. 1 The schematic diagram of the cross-ow ltration set-up used in this study.
Fig. 2 Schematic diagram of the ion exchange
experimental set-up.
where, Cd (IU) and Cu (IU) are color intensity of downstream
(permeate) and upstream (feed) respectively.
2.2.2. Ion exchange experiments
Ion exchange (IE) experiments included batch and continuous
experiments. The batch experiment was conducted by mixing
a volume of sugar solution and an amount of resin in a closed
glass container. The mixture was then stirred at 300 rpm at
a certain time. First, the adsorption capacity (in terms of color
reduction) of three different resins from Lewatit, namely S
6368 (called R1), OC 1074 (called R2) and mixed bead resins
comprising of S 6368 and OC 1074 with the ratio of 1:1 (called
R3) was compared. After contacting the resin with sugar solu-
tion for 24 h, the solution was ltered using a lter paper and
then analyzed. Second, adsorption kinetics were investigated
by measuring adsorption capacity at different times using the
best resin obtained from the adsorption capacity experiments.
Third, the maximum capacity was determined by conducting
experiments using various color concentrations at the opti-
mum time obtained from the kinetic experiment.
The continuous experiment was performed using equip-
ment presented in Fig. 2. The resin was placed in the column
with inside diameter of 25 mm. The height of resin bed was
25 cm. The adsorption experiment was carried out at room
temperature (28 2 C). A sugar solution from the feed con-
tainer was pumped to the IE column at a constant volumetric
feed ow rate.
13
2.2.3. Determination of process sequence
In order to know the best process sequence for decoloriza-
tion of sugar solution, experiments using various process
sequences were conducted. The process sequences included
UF (alone), IE (alone), UFIE and IEUF were investigated. The
operation condition used in each process was determined
from the previous experiments showing the best performance.
Thereafter, the best process sequence was examined for decol-
orization of real cane juice. Nevertheless, in this experiment,
microltration using a 0.45 m membrane was applied for pre-
treatment of real cane juice before it was used as UF feed.
2.2.4. Analyses
The sugar content was measured as brix using digital refrac-
tometer (ATC-1E, ATAGO). The color of the juice was measured
according to GS17 method, where the sugar solution was
adjusted to pH 7 by 0.1 N HCl or 0.1 N NaOH and ltered
through a 0.45 m lter (ICUMSA, 1994). Thereafter, the
absorbance of sugar solution was measure at wavelength
420 nm using a spectrophotometer (Shimadzu UVmini 1240
UV). Beside the pH of sugar solution, the viscosity of sugar
solution was measured by using viscometer DV-II + Pro (Brook-
eld, USA). The color intensity was calculated using Eq. (2).
108 As
Color(IU) = (2)
b RDS 
where, As is the absorbance of the sugar solution at 420 nm,
b is the cell length of the absorbing path (cm), RDS is the
refractometric dry substance (the sugar content determined
by measuring the refractive index of the solution) and  is the
density of the solution (kg/m3 ).
The membrane surface chemistry was analyzed by using
the instrument IR Prestige-21 (Shimadzu, Japan). A total of 32
scans were performed at a resolution of 4 cm1 and tempera-
ture of 21 1 C over the wavelength range of 5004000 cm1 .
A software program, IR solution 1.5, was used to record the
sample spectra versus the corresponding background spectra.
The top surface morphology of the membranes was visual-
ized by using a Supra 35 VP eld emission scanning electron
microscope (FESEM) with applied voltage was 10 kV. The outer
surface of the sample was coated with gold/palladium and
sputtered for 1 min before analysis.
14 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120
1.1
PS-100
1 macromolecules (c.f., signicant color reduction after ultral-
PS-20
0.9
FP-20
0.8
0.7
0.6
J / Jo
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90
Filtration time (min)
Fig. 3 Normalized ux prole during ultraltration of
sugar solution (15 g/L.) using various UF membranes.
Fluxes after stopping and restarting the ltration relative to
initial water ux are also included.
3. Results and discussion
3.1. Effect of membrane pore size on ux and color
reduction
Various UF membranes were examined to lter sugar solution
(15 g/L). The performance was expressed in term of ux and
color rejection. Fig. 3 shows the ux as a function of ltra-
tion time performed with constant transmembrane pressure
(300 kPa for PES-10 and 100 kPa for PS-20, PS-50, PS-100 and FS-
20). Table 1 shows the initial water permeability of membranes
used, color and sugar content of feed and permeate and color
rejection.
As shown in Fig. 3, all UF membranes displayed a ux
decline at the early stage of ltration and after about 30 min
they reached a condition approaching steady state ux. The
ux after stopping the ltration for about 5 min showed a
higher value compared to the ux before stopping the l-
tration, but still much lower compared to the initial water
ux (Jo). The increase in ux after stopping the ltration was
within the range 510% depending on the pore size. This phe-
nomenon indicates that concentration polarization is taken
place and contributes to the ux decline. However, irreversible
ux decline or fouling was the dominant reason for these ux
decreases as noticed by the values of J/Jo after stopping the
ltration are still within the range 3060%.
It was clearly observed that the extent of ux decline was
inuenced by the membrane pore size. For the same mem-
brane material (PS), it was seen that the pore size inuences
the resulting ux behavior. The PS-100 showed the lowest ux
ratio indicating the highest fouling occurred in this mem-
brane. The PS-50 and PS-20 showed comparable ux ratios
but their color rejection was signicantly different, i.e., 74.4%
for PS-50 and 85.6% for PS-20 (Table 1). The PES-10 showed
a lower ux ratio compared to both PS-50 and PS-20 mem-
branes. In term of nominal product permeance, the PES-10,
PS-20, PS-50, PS-100 and FP-20 membranes showed 2.8, 25.5,
36.5, 33.2, and 48.2 L m 2 h 1 bar1 , respectively. FP-20 had the
highest product permeance. Thus, a higher MWCO membrane
did not automatically produce higher permeate ux. Simi-
lar phenomenon was reported by other authors (e.g., Chhaya
et al., 2012; Fukumoto et al., 1998; Susanto and Ulbricht, 2008).
Table 1 shows also brix decrease in the permeate after ultral-
tration. It is believed that this decrease in brix is mostly caused
by removal of non-sugar component such as colorants and
PS-50
PES-10
tration).
No systematic trend was observed with respect to the effect
of pore size on fouling behavior. This result suggests that multi
components with different size (larger and smaller than mem-
brane pores) and different chemistry were present in sugar
solution. It has been reported that the sugar cane contains
many potential foulants such as proteins, polysaccharides
Relative flux after
stopping and restarting and other macromolecules as well as inorganic soluble com-
the filtration ponents with different sizes and charges (Chen and Chou,
1993; Saha et al., 2007). These components lead to fouling
100 110 120 130 140 by accessing both membrane pore and membrane surface.
If the pore size is larger than the solute(s) size, the possi-
bility of solute(s) to access membrane pores will be larger
leading to higher pore narrowing than pore blocking. If the
solute(s) size is larger than the pore size, gel/cake layer foul-
ing will probably be formed. Readers, who are interested in
developing models that describe ux decline, are directed to
the references (Bhattacharya et al., 2001; Mondal et al., 2013;
Verma and Sarkar, 2015). Comparing the membranes having
the same poresize but different material (PS-20 and FP-20)
showed that those membranes have signicantly different ini-
tial water ux, even though they had same MWCO determined
by the company but their pore sizes were practically differ-
ent. This is common in the membrane eld, and has already
been explained in previous publications (Susanto and Ulbricht,
2005; Susanto et al., 2007). The FP-20 showed a slightly higher
ux ratio than PS-20 (0.65 vs. 0.57) but it had signicant lower
color rejection (68.0 vs. 85.7%). These results suggest that
membrane material has signicant inuence on the results
of both membrane characterization and fouling behavior.
Overall, study of the effect of pore size on ux behavior and
color rejection demonstrated that the separation mechanism
of colorants is determined not only by sieving mechanism but
also membrane-solute(s) interactions (membrane material).
These results are different from the results of the previously
reported study by Hamachi et al. (2003). They stated that the
color removal of sugar solution by using UF membrane is
determined by pore size exclusion. Evaluation of the mem-
brane performance, which was based on both the ux ratio
and the color reduction, suggests that the PS-20 should be
chosen for the further experiments.
3.2. Effect of feed characteristics on ux and color
reduction
In this part, the effects of feed concentration and pH on ux
and color reduction were investigated using PS-20 kDa mem-
brane. The results are presented in Figs. 4 and 5. The pH of
15 g/L sugar solution (without adjustment) was 5.5. As was
also found in the previous experiments, a rapid ux decline
followed by a steady ux was observed for all feed concentra-
tions. The increase in feed concentration decreased the ux
ratio indicating higher ux decline. It was clearly observed
that concentration polarization has contributed to the ux
decline as indicated by the increase in ux ratio measured
after stopping the ltration. Furthermore, it was seen that the
higher feed cocentration the higher increase in ux ratio mea-
sured after stopping the ltration (the increases in ux ratio
are 3.9%, 7.8%, and 13.7% for the feed concentration of 10, 15,
and 20 g/L, respectively). This suggests that the concentration
polarization is increased by increasing feed concentration.
 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120 15

Table 1 The initial water permeance of membranes used and the resulting color rejection.
No. Membrane Pure water permeancea (L m2 h1 bar1 ) Feed Permeate Color rejection (%)

Brix Color Brix Color

1 PS-100 97 19 13.7 1497 8.4 449 70.0

2 PS-50 67 16 13.5 1511 8.3 387 74.4
3 PS-20 44 14 13.6 1490 8.2 213 85.6
4 PES-10 6 3 13.5 1505 7.8 203 86.5
5 FP-20 78 15 13.6 1492 8.3 478 68.0

a
Average value standard deviation obtained from at least three measurements.

1.1 1.1

1 C = 10% C = 15% C = 20% 1 pH 4 pH 7 pH 9

0.9 0.9

0.8 0.8

0.7 0.7

0.6 0.6

J / Jo
J / Jo

0.5 0.5

0.4 0.4
Relative flux after
0.3 stopping and restarting 0.3 Relative flux after
the filtration stopping and restarting
0.2 0.2 the filtration

0.1 0.1

0 0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140

Filtration time (min) Filtration time (min)

Fig. 4 Normalized ux prole during ultraltration of
various concentrations of sugar solution using PS-20 kDa
membrane. Fluxes after stopping and restarting the
ltration relative to initial water ux are also included.
In addition to concentration polarization the effect of feed
concentration on irreversible ux decline (fouling) was also
very clear as indicated by the values of J/Jo after stopping the
ltration are still within the range 3040%. An increase in feed
concentration increased concentration polarization meaning
that more suspended or dissolved substances on the mem-
brane surface. This phenomenon facilitates the formation of
fouling by altering interaction between membrane and solutes
(= suspended or dissolved substances) as well as penetrating
of solutes into membrane pores. Further, the increase in feed
concentration increased the feed viscosity (Table 2) that will
add a resistance of feed ow through membrane pores.
Beside membrane pore size, the color reduction is inu-
enced by feed concentration. Increases in feed concentration
were found to increase color rejection. However, further
increase in feed concentration would not increase color rejec-
tion anymore. As feed concentration increases the possibility
of solids to be deposited on the membrane surface and/or
membrane pores (via solute adsorption and pore blocking)
increases resulting in higher color rejection. At feed concen-
tration is high enough, the accumulated solids rejected by the
membrane increases and is able to form a gel layer. This gel
Fig. 5 Normalized ux prole during ultraltration of
various pH of sugar solution using PS-20 kDa membrane
(sugar concentration 15 g/L). Fluxes after stopping and
restarting the ltration relative to initial water ux are also
included.
layer can promote the diffusion of solid already deposited on
the membrane surface and/or membrane pores.
Fig. 5 shows that the ux at acidic condition (pH = 4)
decreased more signicantly compared to the ux at both neu-
tral (pH = 7) and alkaline (pH = 9) conditions indicating more
signicant fouling was taken place at the acidic condition.
This observation is analogous to the results in previous pub-
lications (Balakrishnan et al., 2000a,b; Kishihara et al., 1989).
The higher ux at alkaline condition can be due to the change
in the conformation of the protein impurities. In addition,
by decreasing the pH of the solution, protonation will occur
and consequently, the electrostatic repulsion between the feed
component and the membrane surface is reduced. The pH
of the feed solution also inuences color rejection. The color
rejection were 72.1%, 87.2%, and 80.3% for the feed pH solution
of 4, 7, and 9, respectively; whereas the permeate concentra-
tions were 8.2, 8.4, and 8.0 brix for the pH feed solution of
4, 7, and 9, respectively. These results indicate that the feed
solution comprises of many components, which have differ-
ent charge. The highest ux rejection was observed at neutral

Table 2 The feed and permeate characteristics, ux ratio and color rejection.
No. Csugar (g/L) Feed J/Jo Permeate Color rejection (%)

Brix Color (IU) Viscosity (mPa s) Brix Color (IU)

1 5 4.1 605 0.902 0.835 3.8 196 67.6

2 10 8.9 1052 1.081 0.682 6.8 197 81.3
3 15 13.6 1490 1.263 0.573 8.3 214 85.6
4 20 18.3 2104 1.516 0.447 12.6 317 84.9
5 25 23.4 2499 1.932 0.331 18.9 370 85.2
16
1.2
Fresh membrane
Fouled membrane
1 First of all, three different resins namely R1 (S 6368), R2 (OC
0.8
Absorbance
0.6
0.4
0.2
0 Decolorization of permeate of the PS-20 UF membrane was
500 1000 1500 2000 2500
Wavenumber (cm-1)
Fig. 6 FTIR spectra of fresh and fouled PS-20 UF
membranes (fresh membrane was washed with water).
condition indicating there is an inuence of charge interaction
on separation mechanism.
3.3. Fouling characterization on membrane surface
The ultraltration fouling after the membrane was rinsed with
pure water was characterized by using FTIR. In addition, depo-
sition of foulant on the membrane surface was visualized by
using SEM. The results are presented in Figs. 6 and 7, respec-
tively.
Fig. 6 shows the spectra of fresh and fouled membranes.
Aromatic bands at 1485 and 1580 cm1 and stretching
bands at 1240 cm1 were observed for both pristine and
fouled membranes. These bands indicate benzene ring and
phenylene ether, respectively. In addition, absorption for
asymmetric band at 1325 cm1 was also observed. All those
spectra are typical IR spectra of PS material. In general, the
spectrum of the fouled membrane was similar to the fresh
membrane but the intensity of many peaks was higher. The
higher peaks for fouled membrane were observed for exam-
ples at 910, 990, 1043, and 2360 cm1 . In addition, new peaks
at 990 and 1732 cm1 were observed. Further, the peak at
2360 cm1 attributed to the O H stretching of carbohydrate
structure was also observed. These peaks can indicate the
presence of carbohydrate on the membrane surface as also
reported in previous publications (Anjos et al., 2015; Bureau
et al., 2009; Tewari and Irudayaraj, 2004; Velzquez et al., 2009).
The increase in IR spectra within the range 30003700 cm1
(due to OH group) after ultraltration indicates the presence
of polyphenolic compounds on the outer membrane surface.
In summary, these results suggest that foulant deposition on
the membrane surface is conrmed.
Visualization of membrane surface morphology supports
the previous explanation presented in Section 3.1. Membranes
with the same material but different MWCO (PS-20 and PS-100)
and membranes with the same MWCO but different mate-
rial (PS-20 vs. FP-20) demonstrated different pore structure.
SEM images of the PS-20 fouled membranes showed that the
entire membrane surface is clearly covered by foulant layers
(Fig. 7(E)(H)) indicating fouling indeed occurred during ultra-
ltration of sugar solution. Further, it is seen that the foulant
layer created by sugar solution at acidic condition is denser
than at alkaline condition. These results support the previous
ux behavior results.
food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120
3.4. Decolorization using ion exchange resins
1074) and R3 (mixture of S 6368 and OC 1074, 1:1) were com-
pared for color removal in batch experiment. The results are
presented in Table 3. Decolorization of sugar solution, which
has high color intensity, demonstrated that all resins show
relatively low color reduction. The R1 resin showed a lower
color reduction (25.0%) than both the R2 resin (52.1%) and R3
resin (39.6%). The increase in resin dose did not increase color
reduction signicantly for all three resins (the increase in color
reduction was within the range 815%).
3000 3500 4000
more effective. Regardless the property of resin, it is believed
that at higher color concentration the possibility of resin plug-
ging by high molecular organics will be higher. For the sugar
solution previously treated by UF membrane, the R3 resin
showed the highest color reduction followed by R1 and R2
resins, respectively. As also observed for the feed having high
level color intensity, the increase in resin dose did not increase
color reduction signicantly. These results suggest that the
50 g/L dose of resin may be already in the saturation condi-
tion. Further study on the effect of resin dose is presented in
the following section.
Overall these results suggest that for the feed having high
level color intensity (1500 IU), ion exchange is not a suit-
able process for color reduction. The explanation could be
that the contribution of large macromolecular or organic com-
pound, which causes fast saturation of resins, to color of
sugar solution or plug IE resins, is signicant. Thus, only some
of the colorants were adsorbed by ion exchange. Regarding
the performance of resins used, it can be explained that the
binding colorant by ion exchange resins occurs via ionic and
hydrophobic interactions. Both resins used were strongly basic
macroporous anion exchange resins with monodisperse for
R1 and heterodisperse for R2. In addition, R1 was based on
polystyrene, whereas R2 was based on polyacrylamide. On this
basis, the R2 resin was more effective to adsorb high molecular
hydrophilic anionic organic colorants. By contrast the R1 resin
will adsorb organic substances, organic acids and mineral acid
more effectively. This explanation supports the results pre-
sented in Table 3. Nevertheless, the best result was shown
by the mixed resins indicating that the UF permeate consists
of macromolecular organic colorants and organic acids and
mineral acids.
3.4.1. Kinetics study
The kinetic study was performed to determine the rate of col-
orants uptake by ion exchange resin. Thus the time required
for the decolorization to reach equilibrium will be known.
The experiment was performed by using UF permeate as feed
and R3 resin. Fig. 8 shows that the color reduction is signi-
cantly increased by increasing the contact time. However, the
increase in color reduction from the period of 0 to 5 h was
lower than at the period from 10 to 20 h; apparently, after
20 h slight increase in color reduction was observed indicat-
ing the adsorption of colorants by R3 resin has approached an
equilibrium condition. The experiment was performed with
stirring in order to minimize the boundary layer mass trans-
fer resistance from the solution to the lm around the resin
beads. Thus, adsorption on the outer surface of resin and
diffusion into the resin pore should be the rate controlling
step as indicated by relatively long time to reach a plateau
region. In addition, a slight increase in color reduction with
 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120 17

Fig. 7 Visualization of membrane surface morphology by using SEM: (A) PES-10 fresh membrane, (B) PS-20 fresh
membrane, (C) PS-100 fresh membrane, (D) FP-20 fresh membrane, (E) PS-20 membrane fouled by 15 g/L sugar solution
without pH adjustment, (F) PS-20 membrane fouled by 15 g/L sugar solution at pH 4, (G) PS-20 membrane fouled by 15 g/L
sugar solution at pH 7 and (H) PS-20 membrane fouled by 15 g/L sugar solution at pH 9.

Table 3 Decolorization of sugar solution in batch experiment using different resins.

Resin Dosea (g/L) Contact time (h) Color before (IU) Color after (IU) Reduction (%)
b
R1 50 24 1493 1112 25.5
R2 50 24 1493b 716 52.1
R3 50 24 1493b 902 39.6
R1 50 24 224c 81 63.8
R2 50 24 224c 102 54.5
R3 50 24 224c 47 79.0
R1 100 24 1507b 1065 29.3
R2 100 24 1507b 659 56.2
R3 100 24 1507b 847 43.8
R1 100 24 235c 73 68.8
R2 100 24 235c 98 58.2
R3 100 24 235c 39 83.3

a
Resin to sugar solution ratio.
b
15 g/L sugar solution.
c
Obtained from the permeate of PS-20 UF membrane.

increasing long adsorption time supports that the diffusion
into the resin pore was taken place.
3.4.2. Equilibrium experiment
The amount of resin is critical for the application of ion
exchange at industrial scale. In order to know the optimum
resin required, adsorption equilibrium experiments were
performed using UF permeate as feed. The ratio of resin to
sugar solution was varied (by varying the amount of resin) and
color reduction was determined after 30 h of contact. As can
be seen in Fig. 9, the increase in resin dose within the range
025 g/L signicantly increased color reduction, presumably
by increasing the surface area. Further increases will cause
some resins are not in maximum loading (has not reached
18 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120

90 90

80 80

70 70
Color reduction (%)

Color reduction (%)

60 60

50 50

40 40

30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25
Time (h) Feed flow rate (L/h)

Fig. 8 The effect of contact time on color reduction. The Fig. 10 The effect of feed ow rate on color reduction.
experiment used the UF permeate as feed and R3 resin
(resin dose was 25 g/L). R3 is mixed bead resins comprising
intensity 230.65 IU) was 25 g/L. This means that 1 L of sugar
of S 6368 and OC 1074 with the ratio of 1/1.
solution obtained from UF permeate (20 kDa), needs 25 g of
90 resin resulting in color reduction is 80%. This optimum dose
is higher than the result obtained by Achaerandio et al. (2002)
80
who reported that the optimum dose of resin for white vinegar
70 decolorization was 20 g/L.
Color reduction (%)

60

50 3.4.3. Continuous experiment

40
30
20
10
0 removal was observed with increasing feed ow rate from 0.1
0 10 20 30 40
Resin to sugar solution ratio (g/L)
Fig. 9 The effect of resin dose on color reduction. Resin
dose 0 g/L as a control experiment means sugar solution
without resin was stirred for the same contact time.
saturation condition). Rapid color reduction was observed
within the rst 25 g/L dose. However, slight increase was
observed for the rise dose higher than 25 g/L. This indicates
that some color compounds were not adsorbed. Further, the
results suggest that the optimum resin dose for color removal
of sugar solution resulted from the UF permeate (with color
A continuous experiment was performed to investigate the
effect of feed ow rate on color reduction. Based on the results
obtained from batch experiment (Sections 3.4.1 and 3.4.2), R3
was selected for continuous experiment. The results are pre-
R1 R3
sented in Fig. 10. It is shown that the increase in feed ow rate
decreased color reduction. Initially, the slow decrease in color
50 60 70 80 90 100 110
to 0.3 L/h, while fast decrease was observed with increasing
feed ow rate from 0.3 to 0.5 L/h. Further slow decrease was
observed at the increase in feed ow rate higher than 0.5.
These results suggest that the best ow rate for color removal
should be 0.2 to 0.3 L/h, which corresponds to the specic ow
rate of 1.62.4 bed volumes per hour.
3.5. Process sequences
In this experiment the performance of process conguration,
which included UF (alone), IE (alone), UFIE and IEUF were
investigated. The results are presented in Table 4. It is shown

Table 4 Performance comparison of different process sequences for decolorization.

No Process conguration Color of feed (IU) Color of product (IU) Color reduction (%) J/Jo for UF

1 UF alone 1490 213 85.7 0.57

2 IE alone (R3) 1502 915 39.1
3 IE alone (R2) 1498 693 53.8
5 UFIE (R3) 1499 222239 97.4 0.61
5 IE(R2)UF 1515 703190 87.5 0.72
#6 (MF-UF)IE (R3) 10,694a 2768b 356c 357d 99.5e /97.9f 0.41

(15): Feed = Sugar solution (15 g/L), IE ow rate = 0.3 L/h, color removal based on nal product and initial feed; (6): Feed = real sugar cane juice,
IE ow rate = 0.3 L/h, 0.45 mm MF membrane was used as pre-treatment before UF was applied.
a
Sugar cane juice.
b
Permeate of 0.45 MF membrane.
c
Permeate of UF membrane.
d
Product of IE using R3.
e
Color reduction based on sugar cane juice.
f
Color removal based on permeate of MF membrane.
 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120
that both UF and IE used as single process for decolorization
could not result in liquid sugar with color intensity lower than
100 IU. For single process, UF showed a better performance
than IE process using either R2 or R3 resin. As predicted, a
combination of UF and IE resulted in sugar solution with lower
color intensity. However, the process sequence had a signi-
cant inuence on color reduction. The UFIE using R3 resin
showed the best performance with total color reduction is
97.4%. As discussed earlier, the reason behind this nding can
be that ultraltration removes colorants from large macro-
molecules, which will also avoid plugging of ion exchange
beds. The remaining colorants are then removed by IE process
using R3 resin.
Examination of the best process sequence (UFIE) for decol-
orization of real cane juice showed that the color of liquid
sugar product is higher (57 IU) than using sugar solution as
feed (39 IU) but still signicantly lower than the target (less
than 100 IU). Even though a MF membrane was applied for
pre-treatment of cane juice, the color of permeate, which was
used as feed of UF was signicantly higher than sugar solu-
tion used in experiment no. 4 (2768 vs. 1499 IU). The ux ratio
(J/Jo) observed for real cane juice was also signicantly lower
than for sugar solution (0.46 vs. 0.61). Overall these results sug-
gest that application of UFIE for decolorization of real cane
juice can produce colorless sugar solution, but to be practically
implemented proper pre-treatment before UFIE is needed to
increase both color reduction and ux ratio.
4. Conclusion
Ultraltration experiments for cane sugar solution show that
the membrane pore size (MWCO) of UF membrane has a sig-
nicant inuence on both ux behavior and color reduction.
A ux decline at the early ltration followed by constant
ux after about 30 min of ltration was observed for all UF
membranes. The color reduction is within the range 6887%
depending on the MWCO and material of UF membrane used.
The 20 kDa PS UF membrane shows the best performance
in term of ux and color reduction. Further, concentration
and pH of sugar solution inuence signicantly both ux
and color reduction. Fouling phenomena is clearly observed
and conrmed by FTIR spectra as well as visualization using
SEM.
Ion exchange experiments conducted by using feed with
high color intensity demonstrate that all resins used show
relatively low color reduction. However, the color reduction
is signicantly increased when permeate of ultraltration is
used as feed. In this context, the mixed resin (R3) showed the
highest color reduction. It was observed that the color reduc-
tion is inuenced by time of adsorption and the amount of
resin. It is observed that 30 h seems to be the optimum time
for adsorption, whereas the optimum resin to sugar solution
ration (resin dose) is 25 g/L. Continuous experiments suggest
that the best specic ow rate for color removal using R3 resin
should be 1.62.4 bed volumes per hour.
Process evaluation shows that either UF or IE used as a sin-
gle process is not suitable for decoloring sugar solution. These
process have to be used in combination with the conguration
is UF followed by IE. The UFIE using R3 resin shows the best
performance with total color reduction is 97.4%. Examination
of UFIE for decolorization of real cane juice can produce col-
orless sugar solution. However, to be practically implemented
a proper pre-treatment before UFIE is needed.
19
Acknowledgments
The authors acknowledge the support of the Directorate Gen-
eral of Higher Education, the Republic of Indonesia for funding
this project. The authors thank Alfa Laval (Denmark) for dona-
tion of the membranes.
References
Achaerandio, I., Guell, G., Lopez, F., 2002. Continuous vinegar
decolorization with exchange resins. J. Food Eng. 51, 311317.
Anjos, O., Campos, M.G., Ruiz, P.C., Antune, P., 2015. Application
of FTIRATR spectroscopy to the quantication of sugar in
honey. Food Chem. 169, 218223.
Balakrishnan, M., Dua, M., Bhagat, J.J., 2000a. Evaluation of
ultraltration for juice purication in plantation white sugar
manufacture. Int. Sugar J. 1213, 2124.
Balakrishnan, M., Dua, M., Bhagat, J.J., 2000b. Effect of operating
parameters on sugarcane juice ultraltration: results of a eld
experience. Sep. Purif. Technol. 19, 209220.
Balakrishnan, M., Dua, M., Khairnar, P.N., 2001. Signicance of
membrane type and feed stream in the ultraltration of
sugarcane juice. Sep. Sci. Technol. 36, 619637.
Bento, L.S.M., 1990. Sugar decolorization by ion-exchange resins
with regenerant recovery. Inter. Sugar J. 92, 116125.
Bhattacharya, P.K., Agarwal, S., De, S., Gopal, U.V.S.R., 2001.
Ultraltration of sugar cane juice for recovery of sugar:
analysis of ux and retention. Sep. Purif. Technol. 21, 247259.
Bureau, S., Ruiz, D., Reich, M., Gouble, B., Bertrand, D., Audergon,
J.M., Renard, C.M.G.C., 2009. Application of ATR-FTIR for a
rapid and simultaneous determination of sugars and organic
acids in apricot fruit. Food Chem. 115, 11331140.
Chen, J.C.P., Chou, C.C., 1993. Cane Sugar Handbook: A Manual for
Cane Sugar Manufacturers and their Chemists, 12th ed. John
Wiley & Sons, New York.
Sharma, C.C., Mondal, S., Majumdar, G.C., De, S., 2012.
Clarication of Stevia extract by ultraltration: selection
criteria of the membrane and effects of operating conditions.
Food Bioprod. Process. 90, 525532.
Decloux, M., Tatoud, L., Mersad, A., 2000. Removal of colorants
and polysaccharides from raw cane sugar remelts by
ultraltration. Zuckerindustrie 125, 106113.
Donovan, M., Jansen, R.P., Reisig, R.C., Hlavacek, M., Walker, G.,
Williams, J.C., 2002. Sugar cane membrane ltration process.
US Patent Nr. 6406548.
Fechter, W.L., Brewer, P.A., Pypekamp, V.D., Smith, I.A., 2004.
Treatment of sugar juice, US Patent Nr. 6709527.
Flood, C., Flood, A.E., 2006. Removal of color from the raw sugar
manufacturing process by membrane treatment. Suranaree J.
Sci. Technol. 13, 331342.
Fukumoto, L.R., Delaquis, P., Girard, B., 1998. Microltration and
ultraltration ceramic membranes for apple juice
clarication. J. Food Sci. 63, 845850.
Ghosh, A.M., Balakrishnan, M., Dua, M., Bhagat, J.J., 2000.
Ultraltration of sugarcane juice with spiral wound modules:
on-site pilot trials. J. Membr. Sci. 174, 205216.
Ghosh, A.M., Balakrishnan, M., 2003. Pilot demonstration of
sugarcane juice ultraltration in an Indian sugar factory. J.
Food Eng. 58, 143153.
Hamachi, M., Gupta, B.B., Aim, R.B., 2003. Ultraltration: a means
for decolorization of cane sugar solution. Sep. Purif. Technol.
30, 229239.
Hinkova, A., Bubnik, Z., Kadlec, P., Pridal, J., 2002. Potentials of
separation membranes in the sugar industry. Sep. Purif.
Technol. 26, 101110.
ICUMSA, 1994. Methods Book. International Commission for
Uniform methods of Sugar Analysis, Norwhich.
Jacob, S., Jaffrin, M.Y., 2000. Purication of brown cane sugar
solutions by ultraltration with ceramic membranes:
Investigation of membrane fouling. Sep. Sci. Technol. 35,
9891010.
20 food and bioproducts processing 9 8 ( 2 0 1 6 ) 1120
Jegatheesan, V., Phong, D.D., Shu, L., Aim, R.B., 2009. Performance
of ceramic micro- and ultraltration membranes treating
limed and partially claried sugarcane juice. J. Membr. Sci. 27,
6977.
Kishihara, S., Tamaki, H., Fujii, S., Komoto, M., 1989. Clarication
of technical sugar solutions through a dynamic membrane
formed on a porous ceramic tube. J. Membr. Sci. 41, 103114.
Lin, W.F., Chen, C.H., 1991. Clarication of cane juice by ion
exchange resins. Taiwan Sugar 38, 1722.
Lyndon, R., 1996. Commercialization of adsorber technology in
the fruit juice industry. Fruit Process. 6, 130134.
Mak, F.K., 1991. Removal of colour impurities in raw sugar by
ultraltration. Int. Sugar J. 93, 263265.
Maravic, N., Kiss, F., Seres, L., Bogdanovic, B., Bogdanovic, B.,
Seres, Z., 2015. Economic analysis and LCA of an advanced
industrial-scale raw sugar juice purication procedure. Food
Bioprod. Process 95, 1926.
Mondal, S., Chhaya Rai, C., De, S., 2013. Identication of
foulingmechanism during ultraltration of stevia extract.
Food Bioprocess Technol. 6, 931940.
Saha, N.K., Balakrishnan, M., Ulbricht, M., 2007. Sugarcane juice
ultraltration: FTIR and SEM analysis of polysaccaride fouling.
J. Membr. Sci. 306, 287297.
Saha, N.K., Balakrishnan, M., Ulbricht, M., 2006. Polymeric
membrane fouling in sugarcane juice ultraltration: role of
juice polysaccharides. Desalination 189, 5970.
Susanto, H., Widiasa, I.N., 2009. Ultraltration fouling of amylose
solution: behavior, characterization and mechanism. J. Food
Eng. 95, 423430.
Susanto, H., Ulbricht, M., 2008. High-performance thin-layer
hydrogel composite membranes for ultraltration of natural
organic matter. Water Res. 42, 28272835.
Susanto, H., Franzka, S., Ulbricht, M., 2007. Dextran fouling of
polyethersulfone ultraltration membranesCauses, extent
and consequences. J. Membr. Sci. 296, 147155.
Susanto, H., Ulbricht, M., 2005. Inuence of ultraltration
membrane characteristics on adsorptive fouling with
dextrans. J. Membr. Sci. 266, 132142.
Tewari, J., Irudayaraj, J., 2004. Quantication of saccharides in
multiple oral honeys using transform infrared micro
attenuated total reectance spectroscopy. J. Agr. Food Chem.
52, 32373243.
Velzquez, T.G., Revilla, G.O., Loa, M.Z., Espinoza, Y.R., 2009.
Application of FTIR-HATR spectroscopy and multivariate
analysis to the quantication of adulterants in Mexican
honeys. Food Res. Int. 42, 313318.
Verma, S.P., Sarkar, B., 2015. Analysis of ux decline during
ultraltration of apple juice in a batch cell. Food Bioprod.
Process. 94, 147157.
Wang, L., 2009. Energy Efciency and Management in Food
Processing Facilities. CRC Press, Boca Raton.