Journal of Membrane Science 292 (2007) 106115

Separation of xylose from glucose by nanofiltration from

concentrated monosaccharide solutions
Elina Sjoman a,b, , Mika Manttari a , Marianne Nystrom a , Hannu Koivikko b , Heikki Heikkila c
a

Laboratory of Membrane Technology and Technical Polymer Chemistry, Department of Chemical Technology, Lappeenranta University of Technology,
P.O. Box 20, FIN-53851 Lappeenranta, Finland
b Danisco Sugar and Sweeteners Development Center, Danisco Sugar Ltd., Sokeritehtaantie 20, 02460 Kantvik, Finland
c Danisco Sweeteners Ltd., Sokeritehtaantie 20, 02460 Kantvik, Finland
Received 8 November 2006; received in revised form 18 January 2007; accepted 22 January 2007
Available online 24 January 2007

Abstract
Complex separation of monosaccharides from each other is commercially carried out by chromatographic methods. The possibility of nanofiltration in a demanding separation of a pentose sugar, xylose, from a hexose sugar, glucose, is studied here. Xylose is an intermediate product in
xylitol production and glucose interferes in the process.
Feed solutions were made of xylose and glucose in different mass ratios and total monosaccharide concentrations. The mass ratios of xylose to
glucose in solutions were 1:9, 1:1 and 9:1 and the monosaccharide concentrations of the solutions were 2, 10 and 30 wt.%. Desal-5 DK, -DL and
NF270 membranes were used. Filtrations were done in total reflux mode (i.e. both permeate and retentate were recycled back to the feed tank) at
50 C and the applied pressures were from 2 to 40 bar.
The results indicate that the separation of xylose from glucose by nanofiltration is possible to a limited extent. The mass ratio of xylose to glucose
in the permeate was 1.53.0 times higher than their ratio in the feed. The observed monosaccharide retentions depend highly on permeate flux,
and retentions increase to certain reproducible level as pressure and consequently flux is increased. The observed xylose retentions were from 0 to
80% and the glucose retentions were from 10 to 90%. The effect of total monosaccharide concentration on the observed retention is smaller than
the effect of flux. The largest difference between xylose and glucose retentions was detected at permeate fluxes between 5 to 30 kg m2 h1 . The
ratio of xylose to glucose in the feed had an influence on permeate flux and on xylose retentions. Xylose retentions decreased as the proportion of
glucose increased in the feed. The higher the proportion of xylose in the feed the higher was the total permeate flux.
2007 Elsevier B.V. All rights reserved.
Keywords: Nanofiltration; Xylose; Glucose; Separation; Permeate flux

1. Introduction
Separation of monosaccharides from each other is rather
complex on an industrial scale. Many of the monosaccharides are
important ingredients in food and pharmaceutical industries and,
thus, pure fractions of a specific monosaccharide are needed.
Most often monosaccharide separations are done by chromatographic methods. Nanofiltration (NF) could offer cost-effective
Corresponding author at: Laboratory of Membrane Technology and Technical Polymer Chemistry, Department of Chemical Technology, Lappeenranta
University of Technology, P.O. Box 20, FIN-53851 Lappeenranta, Finland.
Tel.: +358 5 62111.
E-mail addresses: sjoman@lut.fi (E. Sjoman), mika.manttari@lut.fi
(M. Manttari), marianne.nystrom@lut.fi (M. Nystrom).

0376-7388/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2007.01.019

and easy-maintenance separation of pentoses from hexoses.

However, the possibilities of nanofiltration for finer separations
are not thoroughly studied. From the traditional point of view,
membrane filtration would require a decade difference in molar
mass or three times the difference in hydrodynamic radius to
be able to separate components from each other [1]. The molar
mass of a hexose sugar is only about 1.2 times the molar mass
of a pentose sugar. The difference in molecular radius is smaller
than a 10th of a nanometer, thus, the demand for selectivity
is great.
Separation of uncharged substances is mostly based on differences in molecular sizes and diffusivities. The possibility for
a partial separation of disaccharides (300360 g mol1 ) from
monosaccharides (pentoses and hexoses 150180 g mol1 ) is
seen in results of Pontalier et al. [2]. They filtered single solute

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

solutions of glucose and lactose (conc. 50 g L1 ) with two different membranes (cut-offs 100 g mol1 and 400 g mol1 ). The
100 g mol1 membrane retained both solutes almost completely
in the applied pressure range of 530 bar. The looser membrane
(cut-off 400 g mol1 ) retained lactose, and gave a retention of
99.2%. Glucose permeated partially and its reported retention
was 68%. Another example on the separation of saccharides is
the purification of oligosaccharides from contaminant monosaccharides by nanofiltration [3]. Goulas et al. [3] diafiltered a
commercial oligosaccharide mixture with the Desal-5 DL membrane and obtained a 98% yield of oligosaccharides, an 89%
yield of lactose and an 18% yield of glucose, i.e. oligosaccharides and lactose were retained almost completely, and most of
the glucose permeated.
Heikkila et al. [4] present a partial separation of pentose from
hexose by nanofiltration in a patent example. A 10 wt.% feed
solution of arabinose (60% of total dry solids) and rhamnose
(40% of total dry solids) was diafiltered at 50 C. In the end of
the diafiltration step 6869% of the permeate dry solids (DS) was
arabinose with the AFT-60 and the Desal-5 DK membranes. The
pentose sugar arabinose had a lower retention than the hexose
sugar rhamnose, thus, a partial separation of pentose from hexose
was possible by nanofiltration.
Factors affecting NF separation of saccharides, together with
membrane selectivity, are filtration pressure and temperature,
and total solution composition and concentration. Goulas et al.
[3] reported increasing retentions of mono-, di- and trisaccharides as pressure was increased from 6.9 to 27.6 bar. A pressure
increase leads to increased solvent flux and membrane compaction, and according to Goulas et al. [3] these effects together
lead to an overall increase in retentions, the monosaccharide
retentions being most affected. A stepwise increase in temperature from 25 to 60 C was reported to decrease retentions due
to reduced viscosity and increased diffusion [3]. In addition,
the results of Sharma et al. [5] exhibit an increase in polymeric membrane pore size and cut-off size (Desal-5 DL) due
to a temperature increase.
The size of a monosaccharide is equal or smaller than the
cut-off sizes of the NF membranes. The calculated diameters of
the monosaccharide molecules are approx. 0.60.8 nm and the
reported measured pore diameters of the popular commercial
NF membranes are from 0.6 to 2.0 nm, including the mean pore
diameter being approximately 0.80.9 nm [58]. Thus, the comparatively small monosaccharides are the most affected when the
total permeate flux changes due to the changes in pressure or in
temperature.
In nanofiltration of a commercial oligosaccharide mixture
Goulas et al. [3] report increasing retentions of oligosaccharides,
lactose and glucose as the feed concentration was decreased
from 80 to 16 g L1 . It should be noticed that their experiment
was done at constant filtration pressure (13.8 bar) and, evidently,
the total permeate flux was also increased as the concentration
was decreased. Bouchoux et al. [9] measured glucose retention
at two different glucose concentrations of 0.05 M (9 g L1 ) and
0.1 M (18 g L1 ). The presented intrinsic retentions were almost
the same at equal permeate flux values. There has not been
many studies published on NF retention results of monosaccha-

107

ride solutions at higher concentrations, >10 wt.% (i.e. approx.

>100 g L1 ). In addition, there is lack of information on influence of solution composition on retention and flux.
The aim of this study was to examine hexose-pentose separation by nanofiltration. Xylose and glucose were chosen for
pentose and hexose sugars, respectively. The separation of
xylose from glucose is important in commercial purification
of xylose for xylitol production. The most interesting sugar
solutions from industrial point of view are rather concentrated
(total dry solids up to 3060 wt.%) and, thus, demanding higher
operating pressures. Here the feed solutions had total monosaccharide concentrations from 2 to 30 wt.% and applied pressures
from 2 to 40 bar were used. The ratio of xylose to glucose in solution was varied and effects of different solution compositions
studied. The actual temperatures of industrial sugar solutions
are often high and, therefore, measurements were done at
50 C.
2. Experimental
2.1. Membranes and nanoltration set-up
The experiments were made with a DSS LabStak M20 (Alfa
Laval Copenhagen A/S, Denmark) plate and frame filtration
equipment. The membrane stack contained three different membranes (listed from bottom to top): Desal-5 DK, Desal-5 DL (GE
Osmonics, USA) and NF270 (Dow Liquid Separations, USA).
Each membrane had a filtration area of 0.18 m2 . Some chemical and physical characteristics of the membranes are listed in
Table 1. The feed tank was a 70 L reactor equipped with a mixing arm and a steam/water jacket. The liquid was pumped with
a piston pump from the feed tank to the filtration equipment.
A flow meter indicated retentate flow back to the feed tank and
permeate flux was measured on a scale.
The membranes used in the experiments have rather similar characteristics. The Desal-5 DK and DL and the NF270
are all thin-film composite membranes of hydrophilic character.
According to the literature data the Desal-5 DK membrane has
a small positive charge at pH 3.5 [18,19]. This pH is near to the
isoelectric points of the Desal-5 DL and the NF270 membranes
and the membrane charge is rather small at the measurement pH
[18,2022]. The temperature resistance of the NF270 membrane
is rather low [17], even so, the model solution filtrations were
run at 50 C, which is close to its resistance limit.
2.2. Pure water ux measurement
A new set of membranes was at first thoroughly flushed
with ion free water (conductivity <10 S cm1 ) at 2 bar and
45 C. Then the pressure was gradually raised to 3040 bar
and the membranes were treated at high pressure for 2 h. Next,
the pure water fluxes of the virgin membranes were measured
at 10 and 20 bar at 45 C with ion free water. After that the
membranes were cleaned with an alkaline cleaning agent, 0.2%
P3-Ultrasil-110 (Ecolab), at 2 bar and 45 C for 15 min. After
flushing the equipment and membranes with ion free water the
pure water flux was measured again. Before each filtration the

108

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

Table 1
Chemical and physical characteristics of the nanofiltration membranes used in the experiments

Support material
Surface material
Temperature resistance ( C)
pH resistance (20 C)
MgSO4 retention (%)
Isoelectric point (pH)
Contact angle (sessile drop method) ( )
Cut-off (g mol1 )
a
b

Desal-5 DK

Desal-5 DL

NF270

Polysulfone
Proprietary
90 [16]
211 [16]
98a [16]
4.1 [18], 4.0 [19]
31 [18]
150300

Polysulfone
Proprietary
90 [16]
211 [16]
96a [16]
3 [20], 4.1 [21]
41 [21], 44 [23], 54.4 [24]
150300, 250 [23] with PEGs

Polysulfone
Semi-aromatic piperazine based polyamide
45 [17]
310 [17]
>97b [17]
<3 [20], 3.3 [18], 3.5 [22], 5.2 [21]
30 [18], 26 [24]
150200

MgSO4 2000 mg L1 at 6.9 bar.

MgSO4 2000 mg L1 at 4.8 bar.

pure water flux was measured at 45 C to control the cleanliness

of the membranes.
The retention stability of the membranes was monitored with
a dilute MgSO4 -lactose-xylose solution for the cleaned membranes before each model solution filtration. This was seen to be
important as three different sets of membranes were used during
the measurement series and the same set of membranes was used
in several filtrations. The membranes were replaced with a set of
virgin membranes if deterioration of a membrane was suspected
because of poor separation performance or change in pure water
flux values. The measured retentions were reproducible and in
good comparison with the manufacturer information.
2.3. Xyloseglucose model solutions
2.3.1. Properties of xylose and glucose
The molar mass of a xylose molecule is 150 g mol1 and of a
glucose molecule it is 180 g mol1 . The diameter of a molecule
can be calculated and, e.g. the equivalent molar diameter for
xylose is 0.68 nm and for glucose 0.72 nm. These size measures
often imagine the molecule as a sphere. The other size measures
also show that the glucose molecule is only slightly bigger than
the xylose molecule, however, these molecules are very similar.
Table 2 lists the physical characteristics and some of the size
measures of the target molecules.
d-xylose and d-glucose have the same dominant conformation in water, pyranose ring in 1e2e3e4e conformation [13], i.e.
Table 2
Physical characteristics and size measures of d-xylose and d-glucose [1012]

(g mol1 )

Molar mass
pKa
Diffusion coefficient at 25 C [10]
(106 cm2 s1 )
Stokes diameter (nm)
Equivalent molar diameter (nm)
Molar volume at normal boiling point
[10] (cm3 mol1 )
Van der Waals volume [10] (cm3 mol1 )
Hydration number in aqueous solution at
298 K [11]
Solubility parameter [12]

d-Xylose

d-Glucose

150.3
12.26
7.495

180.6
12.43
6.728

0.65
0.68
155.0

0.73
0.72
189.2

73.6
6.8

88.4
8.4

31.0

32.0

the hydroxyl groups attached to carbons 14 are in equatorial

position to the six-membered carbon ring. In aqueous solution
the 4 C1 chair conformation prevails for both molecules. The only
difference in the dominant conformations between glucose and
xylose is that glucose has an arm, CH2 OH, reaching out of
the pyranose ring [13,14]. Cyclic -d-pyranose structures of glucose and xylose are depicted in Fig. 1. At equilibrium in aqueous
solution 38% of the glucose molecules are as -d-glucopyranose
and 62% as -d-glucopyranose (at 31 C, i.e. 304 K). The ratio
of - and -anomers is nearly the same with xylose, i.e. 36.5%
are as -d-xylopyranose and 63% as -d-xylopyranose at 31 C
[13]. The only difference between - and -anomers is the orientation of the hydroxyl group at the anomeric center. The other
possible structures are a five-membered furanose ring and an
open-chain free aldehyde. A minor proportion of glucose and
xylose is present in other possible configurations in aqueous
solutions at ambient temperatures. At higher temperatures proportion of the furanose forms have been seen to increase in
aqueous solution [13].
The water molecule fits well into the prevailing structure of
glucose and xylose and hydrogen bonds are formed [14]. Thus,
a hydration layer surrounds a monosaccharide. A parameter
describing this layer is the hydration number, which actually indicates the number of water molecules disturbed by the
presence of a monosaccharide [11]. The hydration number is
calculated at infinite dilution and this parameter cannot be used
as such for concentrated solutions. In addition, the variation in
hydration numbers obtained with different measurement methods is great, e.g. hydration numbers from approximately 319
have been reported for glucose [15]. The hydration layers of
carbohydrates in aqueous solutions cannot be exactly described
based upon present knowledge.

Fig. 1. -d-Pyranose forms of xylose and glucose. Anomeric centers marked

with asterisk. Redrawn from [13].

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

2.3.2. Filtrations of model solutions

The model solutions were mixed in ion free water (conductivity <10 S cm1 ) from pure crystalline substances. The dry
solids of the solutions were 2, 10 and 30 wt.%. At each dry
solids level three different solute compositions were made, i.e.
the mass ratio of xylose to glucose was 1:9, 1:1 and 9:1. These
mass ratios correspond on an average to molar ratios of 0.13,
1.2 and 10.8, respectively. The solution pH was adjusted to 3.5
with concentrated HCl. The HCl addition was a few milliliters
and it was assumed not to affect the separation of the sugars.
The chosen pH corresponds to the pH of an industrial solution where the separation of xylose from glucose would be
beneficial.
The filtrations were done in total reflux mode, i.e. permeate and retentate were circulated back to the feed tank. The
filtration pressures used were 2, 5, 10, 14, 20, 24 and 30 bar
for 2 and 10 wt.% solutions. For 30 wt.% solutions the pressures
were 10, 20, 30 and 40 bar. The filtration temperature was 50 C.
An advantage of the elevated temperature is the lower viscosity of the concentrated feed solution. Temperatures higher than
50 C would be beneficial from a microbiological point of view.
Here the chosen temperature was considered reasonably safe
according to membrane and monosaccharide stabilities.
The pressure of the feed stream was measured before the
inlet to the filtration equipment and at the outlet of the filtration equipment. The membrane stack caused a pressure loss
of 0.51 bar. The cross-flow velocity along the membrane surface was 0.71.0 m s1 depending on the membrane thickness.
The permeate flux was measured twice for each membrane
at the same pressure by weighing the permeate collected during a time lapse of 15 min. The stabilization time at each
pressure was 30 min before the first flux measurement and
the total filtration time at each pressure was from 1 to 2 h.
The retentate flow was measured by a flow meter. Samples
were taken once from permeates, retentate and feed at each
pressure. Xylose and glucose contents of the samples were analyzed by the HPLC method (Pb2+ column, 85 C, 0.6 mL min1 ).
In addition, pH and conductivity of the samples were
measured.
Membranes were thoroughly rinsed with ion free water
(conductivity <10 S cm1 ) and cleaned with acidic and basic
cleaning agent after each model solution filtration. Acetic acid
was used as acidic cleaning agent (5% acetic acid at 45 C for
15 min) and the basic cleaning was done in the same manner as
with a set of virgin membranes (see Section 2.2).

109

2.4. Calculations for separation evaluation

The observed retention, R, is a measure of membrane selectivity towards a solute as shown for xylose below:
Rxyl = 1

cp (xyl)
cf (xyl)

(1)

where Rxyl is the observed retention of xylose, cp (xyl) the concentration of xylose in permeate (g/100 g of solution), and cf (xyl)
is the concentration of xylose in feed (g/100 g of solution).
The xylose separation factor, Xxyl , is a measure of xylose
purification from glucose. This factor indicates the change in the
permeate composition compared to the original ratio of xylose
to glucose in the feed. The separation is achieved if the separation factor differs from unity. A value greater than one indicates
xylose enrichment in the permeate:
Xxyl =

1 Rxyl
(cp (xyl)/cp (glu))
=
(cf (xyl)/cf (glu))
1 Rglu

(2)

where cp (glu) is the concentration of glucose in permeate

(g/100 g of solution), cf (glu) the concentration of glucose in
feed (g/100 g of solution) and Rglu is the retention of glucose.
3. Results and discussion
3.1. Pure water permeabilities for virgin and used cleaned
membranes
The pure water permeability (PWP) was used as a measure of membrane cleanliness during the measurement series.
Table 3 summarizes the measured PWP values. The PWP is the
average value of the measurements at 10 and 20 bar. The pure
water permeabilities measured at 10 and 20 bar for used cleaned
membranes were similar (the difference in the permeabilities
0.2 L m2 h1 bar1 at maximum).
The PWP was stabilized to certain level after filtration and
cleaning cycles, e.g. for the DK membrane the PWP of a virgin
membrane after cleaning was on an average 9.4 L m2 h1 bar1
and after a few filtration and cleaning cycles it was stabilized to
5.87.1 L m2 h1 bar1 . A virgin membrane had a higher PWP
than a membrane that had been used and cleaned. This decline in
the PWP might be due to inefficient cleaning as the cleaning temperature (45 C) was lower than the filtration temperature of the
model solutions (50 C). The model solution was considered not
to foul the membranes heavily and milder cleaning conditions

Table 3
Pure water permeability (L m2 h1 bar1 ) of the virgin membranes, the virgin membranes after alkaline cleaning and pure water permeability of the used cleaned
membranes after several filtration and cleaning cycles. Permeability was measured at 45 C and at 10 and 20 bar with ion free water (conductivity <10 S cm1 )
(values shown as minimummaximum (average standard deviation))
Membrane

PWP of a virgin membranea

(L m2 h1 bar1 )

PWP of a virgin membrane after

cleaninga (L m2 h1 bar1 )

PWP of a used cleaned membraneb

(L m2 h1 bar1 )

Desal-5 DK
Desal-5 DL
NF270

6.69.6 (8.1 1.5)

7.110.9 (9.1 1.9)
15.217.1 (15.9 1.1)

8.710.4 (9.4 0.7)

9.811.3 (10.7 0.8)
16.618.6 (17.6 1.0)

5.87.1 (6.4 0.4)

6.68.8 (7.6 0.5)
12.617.5 (15.4 1.2)

a
b

Three measurements.
Seventeen measurements.

110

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

Fig. 2. Observed retentions of xylose and glucose with 2, 10 and 30% monosaccharide solution for the Desal-5 DK, Desal-5 DL and NF270 membranes. The mass
ratio of xylose to glucose in the feed solution was 1:1. The filtration temperature was 50 C and the applied pressures were from 2 to 40 bar, DSS LabStak M20-filter.

Fig. 3. Xylose separation factor for 2, 10 and 30 wt.% feed monosaccharide concentrations at total permeate fluxes below 100 kg m2 h1 . Mass ratio of xylose to
glucose in feed solution 9:1 for all the membranes and mass ratio of xylose to glucose 1:9 for the Desal-5 DL. Filtration temperature 50 C and applied pressures
from 2 to 40 bar, DSS LabStak M20-filter.

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

were chosen to spare the selective top layer of the membrane. It

might have been possible that some components stayed inside
the membrane structure and caused permanent decrease in water
flux. The PWP decline might be due to membrane compaction
or swelling during the measurements. The measured PWP values are comparable with the pure water permeabilities reported
recently by other authors [7,21,25].
3.2. Filtration results of xyloseglucose model solutions
The nanofiltration membranes used in this study are able to
separate xylose from glucose to some extent, e.g. filtration of
10 wt.% xyloseglucose solution with the Desal-5 DL membrane at 24 bar increased the xylose content in the total dry
substance from 50 to 67%. These results are in agreement with
the observations of Heikkila et al. [4]. In the following sections
the influence of permeate flux, total monosaccharide concentration and composition on the separation is studied.

111

3.2.1. Inuence of ux on xylose and glucose retentions

Xylose and glucose retentions are dependent on the effective
filtration pressure and, thus, on the total permeate flux. Fig. 2
shows the observed retentions measured for both monosaccharides with 2, 10 and 30 wt.% model solutions. Retentions of the
monosaccharides increase as the total permeate flux increases.
The increase in retentions due to the flux increase is nearly
the same at different feed concentrations. The observed retentions are stabilized to their maximum above fluxes of approx.
60 kg m2 h1 . Glucose retentions are up to 90% and xylose
retentions up to 80%. Apparently, the higher glucose retentions
are due to the larger size of the glucose molecule, which is in
accordance with the sieving effect [2,26].
The osmotic pressure and the viscosity effects are considerably stronger when a 30 wt.% feed solution is filtered compared
to a 2 wt.% feed solution. The rough estimates of osmotic pressures calculated according to the vant Hoff osmotic pressure
equation for 2, 10 and 30 wt.% feed solutions of different xylose

Fig. 4. Difference between observed glucose and xylose retentions for 2, 10 and 30 wt.% feed monosaccharide concentrations at total permeate fluxes below
100 kg m2 h1 . Mass ratios of xylose to glucose in feed solutions 9:1 and 1:9. Filtration temperature 50 C and applied pressures from 2 to 40 bar, DSS LabStak
M20-filter.

112

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

to glucose ratios were 33.3, 15.116.6 and 45.349.8 bar,

respectively. The mass ratio of xylose to glucose has a noticeable
influence on the osmotic pressure estimate as the total amount
of particles is increased at higher xylose ratios. The solutions cannot be considered as ideal, however, these estimates
give the magnitude of the feed-side osmotic pressure. The real
osmotic pressure barrier was considerably smaller as the osmotic
pressure on the permeate side was also significant.
The 2 wt.% model solution had a dynamic viscosity of
0.6 mPa s (at 50 C and atmospheric pressure). The dynamic
viscosity for water at similar circumstances is 0.5 mPa s. The
dynamic viscosities of 10 and 30 wt.% monosaccharide solutions at 50 C were on an average 0.7 and 1.2 mPa s, respectively.
The viscosities under the applied pressures on the membrane
surface and inside the pores are not easily determined. In fact,
the intrinsic viscosity inside the nanometer scale pores is disputable as a physical chemical parameter. Evidently, molecular
interactions inside the pores create some resistance to flow that
can be equated to viscous effects. The lower viscosity at 50 C
is important with concentrated solutions because the permeate
fluxes of 30 wt.% monosaccharide solutions would be negligible
at 25 C even at applied pressures of 3040 bar. Thus, comparatively high filtration temperatures (e.g. 50 C) have to be used
with concentrated sugar solutions.

glucoseethanolwater solutions at constant concentrations with

a dialysis membrane. The friction coefficients describing interactions between solution components with each other and with
the membrane depended on the hydration numbers of glucose
and ethanol [27]. The number of water molecules in the hydration layer is most probably somehow depending on the total
solute concentration, thus, the scheme of interactions in a concentrated multi-component solution is complex. Moreover, the
influence of monosaccharide concentration, and filtration pressure and temperature on the monosaccharide conformation is
vague. However, the size difference is playing a big role in retention as the two studied monosaccharides have presumably the
same dominant conformation in aqueous solution. Anyhow, the
hydration of glucose and xylose emphasizes the size difference.
The difference in observed glucose and xylose retentions,
Rglu Rxyl , are illustrated in Fig. 4. The difference in retentions
with the Desal-5 DK and the NF270 membranes have similar
characteristics, a strong increase in retention difference as the
permeate flux is increased to 10 kg m2 h1 . The largest difference in retentions is near a permeate flux of 10 kg m2 h1
and as the permeate flux increases the difference decreases and
levels off. The difference in retentions at maximum is somewhat larger for more concentrated solutions. This phenomenon
is especially observed with the Desal-5 DL membrane. How-

3.2.2. Inuence of total monosaccharide concentration on

xylose purication
The results indicate that higher xylose separation factors
are gained at higher feed concentrations (Fig. 3). However, a
possible impact of membrane compression and concentration
polarization should be noted as higher applied pressures were
used with the concentrated solutions. The xylose separation
factors for monosaccharide feed solutions at mass fraction 9:1
xylose to glucose at fluxes below 100 kg m2 h1 are shown in
Fig. 3. For comparison, results of the Desal-5 DL membrane are
shown at a xylose to glucose ratio 1:9. The xylose separation
factors were of the same magnitude with the mass ratios of 1:1
and 1:9 xylose to glucose as with the 9:1 ratio, Fig. 3.
Xylose separation factors measured with 30 wt.% feed solution show a nearly linear relationship from 10 to 40 bar (Fig. 3).
The Desal-5 DK gives the steepest increase in xylose separation factor versus flux when comparing with the other two
membranes. Xylose separation factors measured with 2 and
10 wt.% solutions increased until the permeate flux exceeded
2030 kg m2 h1 . Thus, a further increase in the applied pressure to gain better fluxes is not necessarily beneficial if xylose
purification is the purpose.
The separation mechanism of pentose from hexose is based
on their difference in sizes and diffusivities. However, knowledge of molecular interactions near the pore inlet and inside
the pores with concentrated solutions is sparse. A more sophisticated explanation could consider the equilibrium between
the dominant mutarotations of the monosaccharides. In addition, the role of the hydration layer of a carbohydrate in
ezak et al. [27] presented
nanofiltration is not well studied. Sl
SpieglerKedemKatchalsky model based equations for interactions of hydrated species. They modeled glucosewater and

Fig. 5. (A) Xylose and (B) glucose contents in feed solutions and in permeates
at various xylose to glucose ratios with 30 wt.% solutions. Mass ratio of xylose
to glucose in feed 9:1, 1:1, 1:9. Desal-5 DK, at 50 C and at applied pressures
of 10, 20, 30 and 40 bar, total reflux mode, DSS LabStak M20-filter.

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

113

Fig. 6. Total permeate flux, water flux, xylose and glucose flux at different mass ratios of xylose to glucose. The Desal-5 DL membrane at 50 C with 30 wt.%
monosaccharide solution, DSS LabStak M20-filter.

ever, the largest difference in xylose and glucose retention is at

fairly low total permeate flux (520 kg m2 h1 Desal-5 DK
and NF270, 530 kg m2 h1 Desal-5 DL). At higher fluxes
the water flow through the pores is strong and might partially
diminish the size difference between the xylose and the glucose
molecules.
3.2.3. Inuence of xylose to glucose ratio on xylose
separation
The results indicate that glucose pushes xylose through
the membrane in concentrated 30 wt.% monosaccharide solution
(Fig. 5). A high glucose content in the feed gave smaller xylose
retentions than a low glucose content, e.g. xylose retention at
a flux of 4 kg m2 h1 for a xylose to glucose ratio of 1:9 was
4% and for a ratio of 9:1 it was 23% (30 wt.% feed solution,
Desal-5 DK). This effect is clearly seen when monosaccharide
contents in permeate are compared to monosaccharide contents
in feed. Fig. 5(A) and (B) illustrate average xylose and glucose
contents in 30 wt.% monosaccharide feed solutions, and xylose
and glucose contents in the permeates of the Desal-5 DK. It is
seen that at a high glucose content (xylose to glucose mass ratio
1:9 in feed) practically all the xylose is permeated.
The ratio of xylose to glucose has an influence on the total permeate fluxes especially with concentrated solutions. The higher
the xylose content of the solution the higher is the permeate flux
measured at the same pressure. The effect of a high xylose con-

tent on the total permeate flux is shown in Fig. 6 for a 30 wt.%

feed solution with the Desal-5 DL. The same effect, but weaker,
is seen with 2 and 10 wt.% feed solutions. In addition, Fig. 6
illustrates water flux and xylose and glucose fluxes of the Desal5 DL membrane at different xylose to glucose mass ratios. The
ratio of xylose to glucose affects both the total sugar flux and the
solvent, i.e. the water flux. Interactions of monosaccharides with
water molecules are influenced on changes in solution composition. Thus, friction inside the pores grows as the ratio of glucose
increases in the solution and, sugar and water fluxes are reduced.
4. Conclusions
In commercial processes monosaccharides are separated
from each other by chromatographic methods. Here the separation of a pentose sugar from a hexose sugar by nanofiltration was
studied. Xylose, a pentose sugar, and glucose, a hexose sugar,
were chosen as model compounds because xylose purification
is an important step in commercial xylitol production. According to the results, nanofiltration has possibilities to enhance the
yield and partially replace chromatographic methods in xylose
production.
Monosaccharide solutions made of xylose and glucose
were filtered through tight nanofiltration membranes, cut-offs
approximately 150300 g mol1 . The results confirmed that
the separation of xylose from glucose is possible to certain

114

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

reproducible extent. The Desal-5 DK and DL and the NF270

membranes gave comparable fractionations. The results indicate that sieving is the main separation mechanism with these
small, uncharged organic molecules. Possible changes in the
membrane structure, like compaction and swelling, due to filtration temperature, pressure and solution pH might have an
influence on the separation.
The retentions of monosaccharides depend strongly on the
permeate flux, i.e. the effective filtration pressure. The total
solution concentration had minor effect on retentions. Higher
retentions were measured as permeate fluxes increased, e.g. with
the Desal-5 DK membrane the xylose retention was 27% at a flux
of 5.8 kg m2 h1 (10 bar) and 72% at a flux of 44 kg m2 h1
(30 bar) (10 wt.% monosaccharide solution with a mass ratio of
1:1 xylose to glucose). However, a total permeate flux increase
over 60 kg m2 h1 by increasing the applied pressure did not
increase retentions, i.e. at maximum, an 80% retention of xylose
and a 90% retention of glucose were reached.
All three membranes gave xylose separation factors over 2,
i.e. the mass ratio of xylose to glucose in the permeate was at least
two times higher compared to the original mass ratio of xylose to
glucose in the feed. At high feed concentrations (30 and 10 wt.%)
comparably high xylose separation factors were gained at reasonably low permeate fluxes but at high pressures (3040 bar).
Our results suggest that the most favorable xylose separation
from glucose is achieved with a concentrated monosaccharide
solution at rather high pressure.
Xylose retentions were lower than glucose retentions. The
difference between glucose and xylose retentions was the largest
between total permeate fluxes of 530 kg m2 h1 . At higher
fluxes the difference between retentions decreased and leveled
off. Glucose transport through the membrane is more restricted
than xylose due to larger size of the glucose molecule. The size
difference between these two monosaccharides is emphasized
if hydration numbers are included as the hydration numbers
reported for glucose are bigger than for xylose. The viscous
and osmotic pressure effects are important when concentrated
solutions are filtered. Although, knowledge on molecular interactions inside the nanometer scale pores is sparse.
The mass ratio of xylose to glucose in the feed had an
influence on the permeate fluxes and on the retentions. This
was clearly seen with the concentrated monosaccharide solution (30 wt.% monosaccharides). The mass ratio 1:9 xylose to
glucose solution gave a lower permeate flux of 2.6 kg m2 h1
and zero xylose retention at 30 bar and the 9:1 xylose to glucose
solution gave a flux of 6.3 kg m2 h1 and a higher 20% xylose
retention at the same applied pressure (Desal-5 DK). High xylose
concentrations in the feed gave higher total permeate fluxes and
higher xylose retentions than when the glucose concentrations
of the solution was high. In other words, a high glucose concentration in the feed enhances xylose permeation and reduces total
permeate flux.
Acknowledgement
Graduate School in Chemical Engineering (GSCE, The
Academy of Finland) is acknowledged for financial support.

References
[1] M. Cheryan, Ultrafiltration and Microfiltration Handbook, Technomic Publishing Company Inc., Lancaster, USA, 1998, p. 98.
[2] P.-Y. Pontalier, A. Ismail, M. Ghoul, Mechanisms for the selective rejection of solutes in nanofiltration membranes, Sep. Purif. Technol. 12 (1997)
175181.
[3] A.K. Goulas, P.G. Kapasakalidis, H.R. Sinclair, R.A. Rastall, A.S. Grandison, Purification of oligosaccharides by nanofiltration, J. Membr. Sci. 209
(2002) 321335.
[4] H. Heikkila, M. Manttari, M. Nystrom, M. Lindroos, H. Paananen, O.
Puuppo, H. Koivikko, Separation Process, WO02053781, 2002.
[5] R.R. Sharma, R. Agrawal, S. Chellam, Temperature effects on sieving characteristics of thin-film composite nanofiltration membranes: pore
size distributions and transport parameters, J. Membr. Sci. 223 (2003)
6987.
[6] W.R. Bowen, J.S. Welfoot, Modelling of membrane nanofiltrationpore
size distribution effects, Chem. Eng. Sci. 57 (2002) 13931407.
[7] G. Bargeman, J.M. Vollenbroek, J. Straatsma, C.G.P.H. Scroen, R.M.
Boom, Nanofiltration of multi-component feeds. Interactions between neutral and charged components and their effect on retention, J. Membr. Sci.
247 (2005) 1120.
[8] L.D. Nghiem, A.I. Schafer, M. Elimelech, Removal of natural hormones
by nanofiltration membranes: measurement, modelling and mechanisms,
Environ. Sci. Technol. 38 (2004) 18881896.
[9] A. Bouchoux, H. Roux-de Balmann, F. Lutin, Nanofiltration of glucose
and sodium lactate solutions. Variations of retention between single- and
mixed-solute solutions, J. Membr. Sci. 258 (2005) 123132.
[10] T. Shibusawa, Relation between molecular size and diffusivity for acid dyes
and their model compounds in aqueous solution, Seni Gakkaishi 43 (8)
(1987) 401415.
[11] S.A. Galema, H. Hiland, Stereochemical aspects of hydration of carbohydrates in aqueous solutions. 3. Density and ultrasound measurements, J.
Phys. Chem. 95 (1991) 53215326.
[12] F.W. Billmeyer, Textbook of Polymer Science, John Wiley & Sons, New
York, USA, 1984, p. 154.
[13] P.M. Collins, R.J. Ferrier, Monosaccharides Their Chemistry and Their
Roles in Natural Products, John Wiley & Sons, Chichester, England, 1995,
p. 18.
[14] H.S. El Khadem, Carbohydrate Chemistry. Monosaccharides and Their
Oligomers, Academic Press Inc., London, UK, 1988, pp. 1145.
[15] L. Ben Gada, C.G. Dussap, J.P. Gros, Variable hydration of small carbohydrates for predicting equilibrium properties in diluted and concentrated
solutions, Food Chem. 96 (2006) 387401.
[16] http://www.osmolabstore.com/OsmoLabPage.dll?BuildPage&1&1&1067;
web page visited on June 21, 2006.
[17] http://www.dow.com/PublishedLiterature/dh 0416/09002f1380416709.
pdf?filepath=liquidseps/pdfs/noreg/609-00346.pdf&fromPage=GetDoc;
web page visited on June 21, 2006.
[18] J. Tanninen, S. Platt, A. Weiss, M. Nystrom, Long-term acid resistance and
selectivity of NF membranes in very acidic conditions, J. Membr. Sci. 240
(2004) 1118.
[19] G. Hagmeyer, R. Gimbel, Modelling the salt rejection of nanofiltration
membranes for ternary ion mixtures and for single salts at different pH
values, Desalination 117 (1998) 247256.
[20] R. Liikanen, H. Kiuru, J. Peuravuori, M. Nystrom, Nanofiltration flux,
fouling and retention in filtering dilute model waters, Desalination 175
(2005) 97109.
[21] M. Manttari, A. Pihlajamaki, M. Nystrom, Effect of pH on hydrophilicity
and charge and their effect on the filtration efficiency of NF membranes at
different pH, J. Membr. Sci. 280 (2006) 311320.
[22] L.D. Nghiem, A.I. Schafer, M. Elimelech, Pharmaceutical retention mechanisms by nanofiltration membranes, Environ. Sci. Technol. 39 (2005)
76987705.
[23] K. Boussu, B. Van der Bruggen, A. Volodin, J. Snauwaert, C. Van Haesendonck, C. Vandecasteele, Roughness and hydrophobicity studies of
nanofiltration membranes using different modes of AFM, J. Colloid Interf.
Sci. 286 (2005) 632638.

E. Sjoman et al. / Journal of Membrane Science 292 (2007) 106115

[24] G. Cornelis, K. Boussu, B. Van der Bruggen, I. Devreese, C. Vandecasteele, Nanofiltration of nonionic surfactants: effect of the molecular weight
cutoff and contact angle on flux behavior, Ind. Eng. Chem. Res. 44 (2005)
76527658.
[25] M. Manttari, K. Viitikko, M. Nystrom, Nanofiltration of biologically treated
effluents from the pulp and paper industry, J. Membr. Sci. 272 (2006)
152160.

115

[26] B. Van der Bruggen, J. Schaep, D. Wilms, C. Vandecasteele, Influence of

molecular size, polarity and charge on the retention of organic molecules
by nanofiltration, J. Membr. Sci. 156 (1999) 2941.
ezak, S. Grzegorczyn, J. Wasik, Model equations for hydrated
[27] A. Sl
species in transmembrane transport, Desalination 163 (2004)
177192.