http://www.paper.edu.

cn

Journal of Chromatography A, 1145 (2007) 5866

Evaluation of new high-density ion exchange adsorbents

for expanded bed adsorption chromatography
Hai-Feng Xia, Dong-Qiang Lin, Shan-Jing Yao
Department of Chemical and Biochemical Engineering, Zhejiang University, 310027 Hangzhou, China
Received 18 September 2006; received in revised form 13 December 2006; accepted 14 December 2006
Available online 10 January 2007

Abstract
New adsorbents Q HyperZ and CM HyperZ composed of hydrogel-filled porous zirconium oxide particles were evaluated for expanded bed
adsorption applications in the present work. The HyperZ adsorbents have wet density of 3.16 g ml1 , particle size of 44.5100.8 m and average
sphere diameter of 67 m. The bed expansion as the function of flow velocity and fluid viscosity was measured and correlated with RichardsonZaki
equation. The suitable expansion factor was considered less than 2.5, while the corresponding flow velocity was about 450 cm h1 . Liquid mixing
in the bed was determined to evaluate the stability of expanded bed. The Bodenstein numbers tested were higher than 40 and the axial mixing
coefficients (Dax ) were between 0.5 and 9.7 106 m2 s1 , which demonstrated that a stable expanded bed could be formed under suitable operation
conditions. Bovine serum albumin (BSA) and lysozyme were used as model proteins to estimate the adsorption capacities of Q and CM HyperZ,
respectively. The maximum equilibrium adsorption of Q and CM HyperZ could reach 45.7 and 27.2 mg g1 drained adsorbents, respectively. It
was found that yeast cells had little influence on the adsorption capacities of the two adsorbents tested. The dynamic adsorption capacity of BSA
at 10% breakthrough with Q HyperZ was 35.9 mg g1 drained adsorbent at flow velocity of 100 cm h1 for packed bed adsorption. The values for
expanded bed adsorption were 34.4 mg g1 drained adsorbent at flow velocity of 200 cm h1 , 33.6 mg g1 drained adsorbent at 300 cm h1 and
31.7 mg g1 drained adsorbent 400 cm h1 . The results demonstrated that Q HyperZ and CM HyperZ are suitable for expanded bed adsorption of
biomolecules.
2007 Published by Elsevier B.V.
Keywords: Expanded bed adsorption; Physical property; Expansion factor; Liquid mixing; Protein adsorption

1. Introduction
Expanded bed adsorption (EBA) has been proven as
an efficient technology to capture target proteins directly
from unclarified feedstock, such as culture suspensions, cell
homogenates or crude extracts [13]. Compared with classical packed bed adsorption chromatography, EBA processes use
the specially designed adsorbents which expand in the column
with an upward liquid phase, forming a perfectly classified fluidized bed (termed expanded bed) [4]. The enlarged bed voidage
allows small particulates in the mobile phase to pass through
the expanded bed and need not be removed earlier. Therefore,
EBA combines solidliquid separation with an adsorption step
in a single unit operation with the advantages of increased
overall yield and reduced operational time, as well as lower

Corresponding author. Fax: +86 571 87951015.

E-mail address: yaosj@zju.edu.cn (S.-J. Yao).

0021-9673/$ see front matter 2007 Published by Elsevier B.V.

doi:10.1016/j.chroma.2006.12.098

costs for capital investment and consumables. The solid adsorbent is essential to achieve the EBA process and should be
prepared specially. Unlike the packed bed adsorbents, a certain high density and appropriate size distribution of adsorbent
matrices are necessary for EBA applications. To the best of current knowledge, EBA adsorbents tend to be of high density,
small size, large pore and proper size distribution according
to the requirement of large-scale bioseparation process. High
density would suit high operation flow velocity and result in
reducing the processing time; small size and enlarged pore in
the particles would reduce the transport resistance and result
in increasing the dynamic adsorption capacity for high flow
velocity; a proper size distribution is the key point to form the
stable classified fluidized bed in the column to ensure adsorption
efficiency.
During the past 20 years, different concepts of matrices
design have been developed for EBA applications. In general, there are three types. First one is a high-density core
covered with the gel layer, such as agarosestainless steel [5]

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

and agaroseglass beads [6]. Second one is the type of highdensity powder dispersed uniformly in the solid phase, such
as agaroseNdFeB alloy [7], cellulosetitanium oxide [8,9],
cellulosestainless steel powder [10]. Third one is the type
of homogeneous beads using high-density porous materials
such as inorganic oxide, including zirconium oxide, titanium
oxide and hafnium oxide, et al. [1113]. Amersham Biosciences first introduced Streamline series adsorbents to the
market in 1990s. Streamline adsorbents with particle size of
100300 m and mean density of about 1.2 g ml1 are based
on 6% cross-linked agarose containing a crystalline quartz core
as the densifier. Streamline adsorbents can be used at suitable expansion factor of 23, while the corresponding flow
velocity is at 200400 cm h1 in water [14]. New adsorbent
named as Streamline Direct, which is densified with stainless steel core, was developed recently. With the particle size
of 80165 m and mean density of 1.8 g ml1 , the operating
flow velocity is relatively high, about 400800 cm h1 for the
expansion factor of 23 [15]. In addition, Fastline adsorbents
that are composed of agarosetungsten carbide were manufactured by UpFront Chromatography A/S. With high density of
2.53.5 g ml1 , Fastline adsorbents can be used for the flow
velocity as high as 600900 cm h1 [16]. Currently, new series
of adsorbents named HyperZ were developed by BioSepra, part
of Pall Corporation, which combine the desirable characteristic
of soft, high-capacity hydrogel with the dimensional stability of a rigid zirconium oxide [17]. It is the first commercial
gel-in-a-shell high-density beads for EBA application. Good
chemical resistance (12 M NaOH) of the zirconium oxide particles would facilitate the cleaning-in-place procedure. Up to
now the basic properties of HyperZ are quite limited in literature, which certainly hamper the real applications of these new
EBA adsorbents.
In the present work, some physical properties of two HyperZ
adsorbents, CM HyperZ and Q HyperZ, will be investigated,
including density, bead size and size distribution. The performance in the expanded bed mode will also be verified. With the
analysis of expansion characteristics of HyperZ adsorbents, a
suitable range of flow velocities used for EBA application will
be suggested and the bed stability will be studied with the pulse
response methods. In addition, the adsorption isotherms and protein breakthrough curves will be investigated in order to show
the potential abilities of protein adsorption for these new EBA
adsorbents.
2. Experimental
2.1. Materials
Q HyperZ and CM HyperZ adsorbents (50 g each package)
were kind gifts from Pall Corporation (PALL BioSepra, Cergy,
France). Bovine serum albumin (BSA, 67 kDa with a theoretical
pI of 4.9) and lysozyme (15 kDa and a theoretical pI of 11.2) were
purchased from Sigma (Milwaukee, WI, USA). Freezedried
bakers yeast was purchased from Dr. Oetker Nahrungsmittel
KG (Bielefeld, Germany). Prior to the experiments the yeast was
dispersed in the respective buffer and allowed to re-hydrate for

59

90 min under gentle stirring of the suspension. Other reagents

were of analytical reagent grade and obtained from the local
suppliers.
2.2. Measurement of physical properties
The shapes and structures of adsorbents were observed by
a scanning electron microscope, XL-30-ESEM (Philips Instruments Ltd., Japan). The size distribution and average size were
determined with the laser particle size analyzer, Mastersizer
2000 (Malvern Instruments Ltd., Worcestershire, UK). Usually,
measurement was carried out in triplicate for two samples, and
the average values were given.
HyperZ adsorbents are provided as dry particles. The settled density of dry adsorbent (s , g ml1 ) was measured by
contrasting the mass weight and the stacking volume. The material density of zirconium oxide (m , g ml1 ) was estimated by
contrasting the mass weight and real volume of dry zirconium
oxide particles. The wet density (p , g ml1 ) of particles was
determined by water replacement method in a 5 ml gravity bottle.
With m and p , the porosity (P, %) can be estimated as
follows:
P=

m p
100%
m w

(1)

where w (g ml1 ) is the density of water.

The voidage of the sedimented particles (0 ) can be determined as follows:
s
0 = 1
(2)
(p w )P
2.3. Determination of expansion characteristics
XK16/40 column (1.6 cm diameter, 40 cm length, from
Amersham Biosciences, Uppsala, Sweden) and Streamline 25
column (2.5 cm diameter, 1 m length, from Amersham Biosciences, Uppsala, Sweden) were used. The fluid distributor of
XK 16/40 column is composed of a support screen and a nylon
net ring for packed bed chromatography, and Streamline 25 column is specially designed for expanded bed adsorption with
metal distributor plate and metal net to allow the small particles
to go through. The details of the columns can be found in the
technical literature from Amersham Biosciences [18,19]. The
column vertical alignment was assured in all experiments. The
movable adapter was adjusted to control the position of liquid
outlet to the top of expanded bed. Distilled water and glycerol
solutions (10% and 20%, w/w) were used as the liquid phases
for XK16/40 column. Distilled water and bakers yeast solutions
(1%, 2% and 5%, w/w, dry weight in 20 mM pH 7.0 sodium phosphate buffer) were used as the liquid phases for Streamline 25
column. The fluid was transported by a peristaltic pump (Longer
Precision Pump, Baoding, China). The bed expansion factor E
is calculated as:
E=

H
H0

(3)


60

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

where H is the expanded bed height, and H0 is the sedimented

bed height. H0 was 12 cm for XK16/40 column and 15 cm for
Streamline 25 column in the present work.
The RichardsonZaki correlation equation [20] is used to
correlate the expansion characteristics as follows:
U = U t n

(4)

The voidage of expanded bed () and the superficial liquid

velocity (U) is correlated by two parameters, the terminal settling
velocity of particle (Ut ) and the expansion index (n). The value
of can be calculated as following:
= 1 (1 0 )

H
H0

(5)

where 0 is the voidage of sedimented bed, which could be

considered as the same as the voidage of the sedimented dry
particles.
The Stokes equation is used to calculate the theoretical value
of Ut :
g(p l )d 2
Ut =
18

(6)

where d is the particle diameter, l and are the density and

viscosity of the liquid phase.
2.4. Liquid mixing in expanded bed
The measurement and analysis of residence time distribution
(RTD) are normal methods to evaluate the liquid mixing in the
column. In each test, 0.5 ml acetone solution (10%, w/w) was
injected at the bottom inlet of the column as a tracer agent. The
output signal was detected by the UV detector (WellChrom fast
scanning spectrophotometer K-2600, KNAUER, Berlin, Germany) [21]. The theoretical plates number N can be calculated
as follows:

2
tR
N = 5.54
(7)
W1/2
where tR is the residence time and W1/2 is the half peak width
of RTD response.
The Bodenstein number, Bo, relates convective transport of
liquid to dispersion, defined as [22]:
Bo =

UH
Dax

(8)

where Dax is the axial mixing coefficient.

And Bo can be calculated with the following equations:
2
8
1
=
+
N
Bo Bo2
Thus, Dax can be determined from Eqs. (7)(9).

(9)

2.5. Protein adsorption equilibrium

BSA and lysozyme were used as model adsorbates for protein adsorption of Q HyperZ and CM HyperZ, respectively.
The 20 mM TrisHCl buffer (pH 8.0) was used for BSA

adsorption with Q HyperZ, and 20 mM acetic acid/sodium

acetate buffer (pH 5.0) was used for lysozyme adsorption
with CM HyperZ. For the experiments of protein isotherm
adsorption equilibrium, the drained HyperZ adsorbents were
added to 15 ml of 5 mg ml1 of protein solution (or with 1
and 2% yeast cells) in the buffer. The adsorption was kept
at 25 C for 10 h in a shaking incubator. After equilibrium,
the adsorbent was separated by centrifugation, and the supernatant was analyzed by a spectrophotometer (Ultrospec 3300
pro, Amersham Biosciences) for protein concentration. Thus,
the adsorbed protein could be calculated from the mass balance. For the isotherm adsorption equilibrium experiments for
yeast cells to adsorbents, 4 mg ml1 yeast cells suspension was
used and after equilibrium the concentration of cells was analyzed by a spectrophotometer (Ultrospec 3300 pro, Amersham
Biosciences).
The adsorption equilibrium is fitted by the well-known Langmuir equation:
Q =

Qm C
Kd + C

(10)

where Q* is the equilibrium adsorbed capacity, C* the protein or yeast cells concentration in liquid phase, Qm the
saturated adsorption capacity and Kd is the dissociation
constant.
2.6. Dynamic protein adsorption
Two methods were used to investigate the dynamic protein
adsorption: protein adsorption kinetics properties and protein
breakthrough behaviors. For measuring the protein adsorption
kinetics properties, 5 g Q HyperZ adsorbents was added to 50 ml
of 1 mg ml1 BSA solution in the appropriate buffer (without or with 1 and 2% yeast), and then 100 l of solution was
taken out at different time intervals to determine the protein
concentration.
For measuring the protein breakthrough curves, BSA, Q
HyperZ and 20 mM TrisHCl buffer (pH 8.0) were used. A small
column, Econo-Column (1 cm diameter, 30 cm length, from BioRad), the peristaltic pump and the UV detector (WellChrom
fast scanning spectrophotometer K-2600, KNAUER, Berlin,
Germany) were used to set up the chromatographic system.
The sedimented bed height was 9 cm. After the adsorbent was
expanded and kept stable for 15 min, 2 mg ml1 BSA solution
was pumped into the column, and the UV detector detected the
protein breakthrough at 280 nm.
As described by Griffith et al. [11], the dynamic adsorption
capacity at 10% breakthrough could be calculated as follows:
V
C0 0 10% (1 (C/C0 )) dV
Q10% =
(11)
M
where Q10% is the dynamic adsorption capacity, C and C0 the
protein concentration of outlet fluid and initial fluid, V10% the
loading volume corresponding to 10% breakthrough point and
M is the mass of the adsorbents.

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

61

Table 1
Physical properties of HyperZ adsorbents
p (g ml1 )

s (g ml1 )

m (g ml1 )

P (%)

Size rangea (m)

Ds (m)

3.16

1.97

5.15

48.0

0.265

44.5100.8

67.0

Size range counts from 10% to 90% volume content.

3. Results and discussion

3.1. Physical properties
The important physical properties of expanded bed adsorbents are considered, including sphericity, wet density, size
distribution and pore construction. Some physical properties of
HyperZ were measured and are summarized in Table 1. It can
be found that the wet density reaches 3.16 g ml1 . The particle
size ranges from 44.5 to 100.8 m, which counts from 10 to
90% volume content, and the average diameter is about 67 m.
Compared to the Streamline series adsorbents (1.2 g ml1 and
200 m), high density and small size of HyperZ series would
have a potential to reduce the mass transfer resistance and
result in high efficiency for EBA applications. The porosity estimated is about 48.0%, which is similar to the data (4753%)
in Griffiths work [11]. The appearance of HyperZ adsorbents
is shown in Fig. 1a. It can be seen that the adsorbent particles with different diameters show good sphericity. Fig. 1b
and c shows the surface and inner structure of HyperZ. The
dense and uniform pore structures in the particles could be
found clearly. In addition, as shown in Fig. 2, the logarithmic

Fig. 2. Size distributions of Q and CM HyperZ adsorbents. () Q HyperZ; ()

CM HyperZ.

symmetrical size distributions of adsorbent particles could be

found for Q HyperZ and CM HyperZ. The size distributions
are quite similar, which indicate same origin of basic material
for two HyperZ adsorbents tested. Proper size distribution is
the guarantee for forming a stable classified fluidized bed to

Fig. 1. Shapes and structures of HyperZ adsorbents. (a) Shapes of adsorbents, magnified by 100; (b) surface structure, magnified by 5000; (c) inner structure,
magnified by 5000.


62

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

Fig. 3. Bed expansion of HyperZ adsorbents with water and varying viscosity
fluids in XK16/40 column. () Q HyperZ with water; () Q HyperZ with 10%
glycerol; () Q HyperZ with 20% glycerol; () CM HyperZ with water; ()
Streamline DEAE with water (data from literature [9]).

ensure the adsorption efficiency. 0 was calculated as 0.265,

which is much lower than the data of 0.4 published for Streamline series [22] due to relative wide size distribution of HyperZ
particles.
3.2. Expansion characteristics
The expansion behavior in expanded bed is mainly determined by the properties of adsorbent particles and the mobile
phase. Small particle size would result in low operation flow
rate, and high wet density would increase the suitable flow rate.
Fig. 3 shows the expansion characteristics of Q and CM HyperZ
in XK16/40 column with different mobile phases, compared
with Streamline DEAE (data from literature [9]). Generally,
two adsorbents exhibit nearly same expansion behavior due
to same materials. The expansion factor of 22.5 for HyperZ
could be obtained at a flow velocity of 300450 cm h1 in
water, while that for Streamline DEAE was at a flow velocity
of 200300 cm h1 . This means that HyperZ adsorbents could
be used at higher operation flow than Streamline for the same
expansion factor, but lower than Streamline Direct and Fastline adsorbents [1416]. With the increase of fluid viscosity
with glycerol solutions, the expansion factor increased obviously under the same flow rate. For example, for 300 cm h1
flow velocity of Q HyperZ, the expansion factor was about 1.95
for water, 2.22 with 10% glycerol solution and 2.70 with 20%
glycerol solution.
It was found in the experiments that some part of adsorbents
on the top of the expanded bed became unstable and dispersed
when the expansion factor exceeded 2.5. Some particles began
to flow out of the bed when E reached 3. This might be mainly
due to the 12% adsorbents that particle size is in the range of
3035 m (as shown in Fig. 2). Based on the Stocks equation,
the terminal settling velocity of 32 m diameter particles with
the wet density of 3.16 g ml1 is about 480 cm h1 for water as
the mobile phase, while corresponding E is about 2.6. Therefore,
the result indicated that the suitable operation E should be less
than 2.5 for HyperZ series adsorbents.

Fig. 4. Bed expansion of CM HyperZ adsorbents with water and varying yeast
cell solution fluids in Streamline 25 column. () CM HyperZ with water; ()
CM HyperZ with 1% cell solution; () CM HyperZ with 2% cell solution; ()
CM HyperZ with 5% cell solution.

To investigate the biomass influences, the expansion characteristics of CM HyperZ in Streamline 25 column with yeast
cells suspension was also measured, and the results are shown
in Fig. 4. It could be found that the expansion factors in water
were quite similar for the two columns tested, XK 16/40 and
Streamline 25. The biomass concentration evidently affected
the expansion value. For example, at the flow velocity of
250 cm h1 , the expansion factor reached 2 with 1% cell suspension and almost 2.5 with 5% cell suspension. It indicated that the
physical properties of feedstock should be considered seriously
to determine the suitable flow velocity in the real application.
The expansion behaviors were correlated by the
RichardsonZaki equation (Eq. (4)). The results for XK16/40
column and Streamline 25 column are shown in Figs. 5 and 6,
respectively. All the data fitted well with RichardsonZaki
relations and the parameters of Ut and n are listed in Table 2.
The theoretical terminal settling velocity of particle was also
calculated with Eq. (6), which is based on the average particle
diameter (Ds ), and also shown in Table 2. It could be found
that the values of theoretical Ut based on Stocks equation
are significantly higher than that of experimental Ut . Similar

Fig. 5. RichardsonZaki correlation between flow velocity and bed voidage

(XK16/40 column). () Q HyperZ with water; () Q HyperZ with 10% glycerol;
() Q HyperZ with 20% glycerol; () CM HyperZ with water; () Streamline
DEAE with water (data from literature [9]).

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

Fig. 6. RichardsonZaki correlation between flow velocity and bed voidage

(Streamline 25 column). () CM HyperZ with water; () CM HyperZ with 1%
cell solution; () CM HyperZ with 2% cell solution; () CM HyperZ with 5%
cell solution.

63

Fig. 7. Bo number as the function of flow velocity and variety of fluid viscosity:
() Q HyperZ with water; () Q HyperZ with 10% glycerol; () Q HyperZ
with 20% glycerol; () Streamline DEAE with water (data from literature [26],
H0 = 15 cm).

results had been reported in many literatures [7,2325]. It is

considered that the theoretical Ut based on Stocks equation
is the average value based on the mean Ds , while the particles
actually have an obvious size distribution.
3.3. Liquid mixing in expanded bed
The Bodenstein number (Bo) and the axial mixing coefficient
(Dax ) were calculated to evaluate the liquid mixing in expanded
bed. The Bo numbers were investigated as the function of flow
velocity with different fluids. As shown in Fig. 7, for water as the
mobile phase, the Bo value of Q HyperZ could reach 200 at the
low flow velocity and significantly higher than that of Streamline
DEAE (data from literature [26]). All Bo values obtained were
higher than 40, of which the value considered was the critical
Bo value of plug flow performance in expanded bed [14]. It
was also found that the Bo decreased obviously with increasing
flow velocity and fluid viscosity. It indicated that the stability
of expanded bed would be disturbed with the increase of flow
velocity and viscosity of mobile phase. Fig. 8 shows the Bo as
the function of expansion factor.
The axial mixing coefficient as the function of flow velocity is
shown in Fig. 9. It was found that all the Dax values were below
105 m2 s1 . The Dax values of Q HyperZ in expanded bed
were in the range between 0.5 106 and 6.9 106 m2 s1 for

Fig. 8. Bo number as the function of expansion factor and variety of fluid

viscosity (symbols as Fig. 7).

water as the mobile phase, which were lower at low flow velocity
(<350 cm h1 ) but higher at high flow velocity (>350 cm h1 )
than that of Streamline DEAE (data from literature [26]). The
results indicated that for low flow velocity the expanded bed
formed with HyperZ was relatively stable than that with Streamline series. Dax values increased obviously with the increase
of flow velocity and fluid viscosity, which meant that the
axial mixing enhanced and the bed stability weakened. This

Table 2
Parameters for RichardsonZaki correlations and Ut calculated by Stokes equation
Column

Adsorbents/mobile phase

Ut (cm h1 )

Experimental

Stockesa

XK16/40

Q HyperZ/water
Q HyperZ/10% glycerol
Q HyperZ/20% glycerol
CM HyperZ/water

1366.5
1022.5
692.3
1394.1

2117.9
1251.7
772.8
2117.9

3.20
3.07
3.39
3.12

Streamline 25

CM HyperZ/water
CM HyperZ/1% yeast
CM HyperZ/2% yeast
CM HyperZ/5% yeast

1370. 5
1063.0
944.3
915.5

2117.9
1891.7
1569.1
928.2

3.37
3.17
3.17
3.66

Ut from Stockes equation is based on the average particle diameter (Ds ).


64

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

Fig. 9. Axial mixing coefficient (Dax ) as the function of flow velocity and variety
of fluids viscosity (symbols as Fig. 7).

result was consistent with the analysis of Bo. Dax values of

Q HyperZ with 10% and 20% glycerol as the mobile phase
were (0.98.3) 106 and (0.99.7) 106 m2 s1 , respectively. Dax as the function of expansion factor is shown in Fig. 10.
It could be found that Dax of Q HyperZ for different fluids
with varying viscosities were similar for same expansion factor, which meant that strong expansion under high flow velocity
is the main reason to cause the increase of Dax value. For same
expansion factor, the Dax of Streamline DEAE was a little lower
than that of HyperZ, mainly due to relative low flow velocity.
3.4. Protein adsorption equilibrium
The adsorption isotherm curves of Q HyperZ and CM HyperZ
are shown in Fig. 11. BSA was used as model target for protein
adsorption of Q HyperZ, while lysozyme was used for that of CM
HyperZ. In some cases, to simulate the influences of biomass,
yeast cells were added. According to the Langmuir correlation as
Eq. (10), the saturated adsorption capacity of BSA to Q HyperZ
was found as high as 45.7 mg g1 drained adsorbents (equal to
90 mg ml1 adsorbent), which is two-fold higher than Streamline DEAE (40 mg BSA ml1 adsorbent) and a little lower than
Streamline Q XL (about 110 mg BSA ml1 adsorbent) described
in literature [13]. Although the adsorption of yeast cells to Q

Fig. 10. Axial mixing coefficient (Dax ) as the function of expansion factor and
variety of fluids viscosity (symbols as Fig. 7).

Fig. 11. Adsorption isotherm curves of protein to HyperZ adsorbents. () BSA
to Q HyperZ; () BSA to Q HyperZ with 1% yeast; () BSA to Q HyperZ with
2% yeast; () lysozyme to CM HyperZ. The adsorption of BSA to Q HyperZ
was measured in 20 mM pH 8.0 TrisHCl buffer at 25 C and the adsorption
of lysozyme to CM HyperZ was in 20 mM pH 5.0 acetic acid/sodium acetate
buffer at 25 C.

HyperZ could reach 21 mg g1 drained adsorbents (as shown

in Fig. 12), the adsorption capacity of protein decreased little
in the yeast suspension and kept 41 mg g1 drained adsorbents
with 1% yeast and 39 mg g1 drained adsorbents with 2% yeast.
As shown in Fig. 12, almost no yeast cells were adsorbed on CM
HyperZ adsorbent due to the negative surface charge for both
yeast cells and CM HyperZ. The saturated adsorption capacity
of lysozyme to CM HyperZ was found as 27.2 mg g1 drained
adsorbents (equal to 53.6 mg ml1 adsorbent), which is similar to Streamline SP (about 60 mg lysozyme ml1 adsorbent)
described in literature [14].
3.5. Dynamic protein adsorption
The protein adsorption kinetics was used to evaluate the
adsorption efficiency of HyperZ adsorbents. The results are
shown in Fig. 13. In the buffer, BSA could be adsorbed by Q
HyperZ in a relative fast way. After 5 min, only 33% protein was
resident in the solution and about 13% remained after 60 min.
Protein adsorption was affected at first phase when yeast cells
were added. For 1% and 2% yeast cells in the solution, 44%

Fig. 12. Adsorption isotherm curves of yeast cells to HyperZ adsorbents. ()
Q HyperZ; () CM HyperZ. Buffer and temperature were same as shown in
Fig. 11.

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

65

drained adsorbent). At 200, 300 and 400 cm h1 flow velocities

in expanded bed, Q10% was about 34.4, 33.6 and 31.7 mg g1
drained adsorbents, equal to 95.8%, 93.6% and 88.3% of Q10% at
100 cm h1 in the packed bed (35.9 mg g1 drained adsorbents),
respectively. The results show that the adsorption efficiency in
expanded bed mode was almost as high as in packed bed mode,
which demonstrated that HyperZ adsorbents are suitable for
EBA applications.
4. Conclusions

Fig. 13. Protein adsorption kinetics curves of Q HyperZ adsorbents. () BSA
to Q HyperZ; () BSA to Q HyperZ with 1% yeast cell; () BSA to Q HyperZ
with 2% yeast cell. Buffer and temperature were same as shown in Fig. 11.

Fig. 14. Breakthrough curves of Q HyperZ at varying flow velocities and different operating modes. () 100 cm h1 in packed bed; () 200 cm h1 in
expanded bed; () 300 cm h1 in expanded bed. () 400 cm h1 in expanded
bed.

and 40% proteins were adsorbed after 15 min,and 18% and 25%
remained after 60 min. Therefore, for the EBA applications, the
residence of target protein in the column should be considered
according to the feedstock properties to ensure enough time.
The protein breakthrough behaviors of Q HyperZ under
varying flow velocity and different operating modes were investigated. As shown in Fig. 14, the breakthrough curves in
expanded bed at 200, 300 and 400 cm h1 were only a little
flatter than that in packed bed at 100 cm h1 . The dynamic
adsorption capacity at 10% breakthrough was calculated with
Eq. (11), and the results are listed in Table 3. It was found that
the dynamic adsorption capacities in packed bed could reach
78.6% of the saturated adsorption capacities Qm (45.7 mg g1

Q HyperZ and CM HyperZ as new series of high-density ion

exchange adsorbents for expanded bed adsorption were investigated in the present work. Two adsorbents have high wet density
and proper size distribution, which coincide with the requirements of EBA applications. The bed expansion characteristics as
the function of flow velocity and variety of fluids viscosity were
determined and corrected with RichardsonZaki equation. The
expansion factor of 2.5 was recommended not to exceed to keep
bed stable for the corresponding flow velocity of 450 cm h1 .
With RTD tests, the analyses of the Bo number and the axial
mixing coefficient demonstrated that HyperZ could form stable expanded bed. The bed stability decreased with the increase
of flow velocity and fluid viscosity. The saturated adsorption
capacity of BSA to Q HyperZ was found as high as 90 mg ml1
adsorbent and lysozyme to CM HyperZ for 53.6 mg ml1 adsorbent. Although the adsorption of yeast cells to Q HyperZ could
reach 21 mg g1 drained adsorbents, the adsorption capacity of
BSA with Q HyperZ was little affected. The dynamic adsorption at 10% breakthrough could reach 69.4% of the saturated
adsorption capacities even at high flow velocity of 400 cm h1 ,
which demonstrated that HyperZ could be used for EBA
processes.
Acknowledgments
The support of the Hi-tech Research and Development Program of China (863 Program, 2006AA02Z210 ) and National
Natural Science Foundation of China is gratefully acknowledged. The authors thank Mr. Xian-Jin Gu, Yan Dong and
Jing-Chang Wang for their kind help and Dr. Sylvio Bengio for
the useful discussion. The authors also want to thank the German Academic Exchange Service (DAAD) for the instrument
donation of Spectrophotometer K2600 and PALL BioSepra for
the kind gift of HyperZ adsorbents.
References

Table 3
Dynamic adsorption capacities of BSA to Q HyperZ under varying flow velocity
and different operating modes
U (cm h1 )

Operating mode

Q10%
(mg g1 )

Q10% /Qm

Qp10% /
Qe10%

100
200
300
400

Packed bed
Expanded bed
Expanded bed
Expanded bed

1
1.63
1.95
2.30

35.9
34.4
33.6
31.7

0.786
0.753
0.735
0.694

1
0.958
0.936
0.883

[1] R. Hjorth, Trends Biotechnol. 15 (1997) 230.

[2] J. Hubbuch, J. Thommes, M.-R. Kula, Adv. Biochem. Eng. Biotechnol. 92
(2005) 101.
[3] G.E. Hamilton, F. Luechau, S.C. Burton, A. Lyddiatt, J. Biotechnol. 79
(2000) 103.
[4] J. Thommes, Adv. Biochem Eng. 58 (1997) 185.
[5] E. Palsson, P.-E. Gustavsson, P.-O. Larsson, J. Chromatogr. A 878 (2000)
17.
[6] X. Zhou, Q.-H. Shi, S. Bai, Y. Sun, Biochem. Eng. J. 18 (2004) 81.
[7] X.-D. Tong, Y. Sun, J. Chromatogr. A 943 (2002) 63.


66

http://www.paper.edu.cn
H.-F. Xia et al. / J. Chromatogr. A 1145 (2007) 5866

[8] G.R. Gilchrist, M.T. Burns, A. Lyddiatt, in: D.L. Pyle (Ed.), Separations
for Biotechnology, vol. 3, Royal Society of Chemistry, London, 1994, p.
186.
[9] Y.-L. Lei, D.-Q. Lin, S.-J. Yao, Z.-Q. Zhu, J. Appl. Polym. Sci. 90 (2003)
2848.
[10] Z.-J. Miao, D.-Q. Lin, S.-J. Yao, Ind. Eng. Chem. Res. 44 (2005) 8218.
[11] C.M. Griffith, J. Morris, M. Robichaud, M.J. Annen, A.V. McCormick,
M.C. Flickinger, J. Chromatogr. A 776 (1997) 179.
[12] N. Voute, E. Boschetti, Bioseparation 8 (1999) 115.
[13] N. Voute, D. Bataille, P. Girot, E. Boschetti, Bioseparation 8 (1999) 121.
[14] Amersham Pharmacia Biotech, Expanded Bed AdsorptionPrinciples and
Methods, Code No. 18-1124-26, Uppsala, Sweden, 2000.
[15] Amersham Biosciences, Expanded Bed Adsorption, StreamlineTM Direct
CST-1: Adsorbent for Capture Proteins from Crude Feed, Uppsala, Sweden,
2005.
[16] UpFront Chromatography A/S, FastLine Columns and Adsorbents for
Expanded Bed Adsorption, Copenhagen, Denmark. http://www.upfrontdk.com.

[17] Pall Corporation, Q & CM HyperZTM Ion Exchange Sorbents High Productivity Expanded Bed Adsorption Chromatography, New York, USA.
http://www.pall.com.
[18] Amersham Biosciences, Column XK16, XK26, XK50 and Adaptors AK16,
AK26, AK50 Instructions, Code No. 56-1130-80.
[19] Amersham Biosciences, Streamline 25 Column User Manual, Code No.
18-1112-01.
[20] J.F. Richardson, W.N. Zaki, Trans. Inst. Chem. Eng. 32 (1954) 35.
[21] D.-Q. Lin, Z.-J. Miao, S.-J. Yao, J. Chromatogr. A 1107 (2006) 265.
[22] J. Thommes, A. Bader, M. Halfar, A. Karau, M.-R. Kula, J. Chromatogr.
A 752 (1996) 111.
[23] J. Thommes, M. Halfar, S. Lenz, M.-R. Kula, Biotechnol. Bioeng. 45 (1995)
205.
[24] Y.K. Chang, H.A. Chase, Biotechnol. Bioeng. 49 (1996) 512.
[25] A. Karau, C. Benken, J. Thommes, M.-R. Kula, Biotechnol. Bioeng. 55
(1997) 54.
[26] Y.-L. Lei, D.-Q. Lin, S.-J. Yao, Z.-Z. Liu, Z.-Q. Zhu, Chin. J. Chem. Eng.
11 (2003) 141.