Microsoft Word - szolcs_eloszo.doc


 
 HUNGARIAN JOURNAL  
 OF INDUSTRIAL CHEMISTRY 
 VESZPRÉM 
 Vol. 32. pp. 23-31 (2004) 

 
 
 

INVESTIGATION OF REVERSE PHASE SMB-CHROMATOGRAPHIC 
BIOSEPARATIONS OF AMINO ACID AQUEOUS SOLUTIONS 

 
Z. MOLNÁR1, M. NAGY1, A. ARANYI2, L. HANÁK1, T. SZÁNYA1 and J. ARGYELÁN1 

 
 

1University of Veszprém, Department of Chemical Engineering, 
H-8201 Veszprém, P.O.Box 158, HUNGARY; 

Fax number: +36 88 421 905 e-mail: hmolnar78@freemail.hu 
2Gedeon Richter Pharmaceutical Works, H-1475 Budapest, 

P.O.Box. 27, HUNGARY 
 
The authors investigated amino acid aqueous solutions as model system for the purpose of studying reverse 
phase chromatographic bioseparation. Desalting of DL-β-phenylalanine was studied on a small-laboratory 
scale simulated moving bed (SMB) preparative liquid chromatograph (number of columns=6, column 
length=125mm, column I.D.=13mm). DIAION HP20 polymeric adsorbent resin was used for the reverse 
phase chromatography. The feed (sample) of the SMB equipment is 3.5g DL-β-phenylalanine/dm3 and 58.5g 
sodium-chloride/dm3 aqueous solution. With the three zones opened loop SMB with 2-2-2 column 
configuration can amino acid product be achieved with less than 50ppm NaCl in crystal form after 
evaporation. Later a large laboratory scale (number of columns=4, column length=500mm, column 
ID=50mm) automatized SMB equipment was constructed. The applied model system for bioseparation 
contains glycine (1.5g/dm3), L-phenylalanine (3.3g/dm3) in water and SEPABEADS SP825 adsorbent. Both 
L-phenylalanine and glycine were produced in more than 99.9% m/m purity and 99% yield at productivity 
3.7-9.5mg/(g adsorbent h) in case of three zones open loop 2-1-1 column configuration. The SMB 
experiments were simulated with the help of equilibrium cascade model. The measured and calculated data 
agreed well. 
 
Keywords: simulated moving bed (SMB), column liquid chromatography, polymeric adsorbent resin, amino acid 
aqueous solution 
 
 

Introduction 
 
 
It is typical in biotechnology and pharmaceutical 
industry that water phase mixture for processing 
contains end-product or active ingredient in small 
concentration beside the contaminant or polluting 
components. Simulated moving bed (SMB) 
preparative chromatography can be advantageous 
among the end-product recovery methods for 
continuous processing of high purity products, or 
products being difficult to isolate. 

Basically the SMB chromatograph works 
similarly to the true moving bed (hereafter 
abbreviated as TMB). The TMB works in the 
following way by Figure 1a: in a column the 
mobile phase moves upwards through the 
adsorbent that moves downwards simultaneously. 

The column is fitted with a feed inlet in the middle 
of the side-wall, a raffinate outlet on the upper part 
and an extract outlet on the lower part. Choosing 
the appropriate adsorbent moving velocity value 
and suitable eluent, feed, extract and raffinate flow 
rates, a stationary state can be obtained in the 
column with constant concentration profiles. The 
more binding component occurs in the lower (I-II) 
part of the column and the less binding one is 
located on the upper part (III-IV) of the column. 
This way two pure components can be obtained 
simultaneously in the extract and in the raffinate. 
The chromatographic quality of TMB is difficult to 
realize technically, therefore an SMB was 
employed in the preparative chromatography. 

SMB liquid chromatograph (hereafter-
abbreviated SMB-LC) (Fig. 1b) is a multi-column 
system with two inputs and outputs (products), in 



 

 

24 

which liquid phase moves in counter-current of 
adsorbent phase. The counter-current stream is not 
real, but simulated, since the packed 
chromatographic stationary phase moves 
periodically after each switching time. The shorter 
the switching time and the more the number of 
columns are in the SMB-LC, the better it 
converges to the TMB-LC. 

The SMB technique is basically a two product 
preparative chromatographic operation. It is 
suitable for mixtures to be separated having two 
components or can be produced two product 
fractions. 

In case of continuous system the two input 
streams are the fresh eluent and feed, the two 
outputs are the extract and raffinate. Above all in 
basic case regenerated eluent of re-circulating 
stream is added to the fresh eluent. 

Similarly to the TMB the lower part (I-II) of 
the column is rich in the more binding component 
and at the upper part (III-IV) of the column 
contains the less binding component. The inlet and 
the outlet fluid flow streams divide the column 
system into four zones (Fig.1.) 
 

I

II

III

IV

Eluens

F

E

R

A Rec   D+D Rec

D Rec

D

CM

IV

III

D+D Rec

F

R

E

I

II

D Rec

A; B;  
Figure 1. The scheme of  a) true moving bed (TMB) and b) 

simulated moving bed (SMB). True moving bed (TMB) 
adsorber: I, II, III, IV – zones; D – desorbent (solvent, eluent); 
D Rec – recirculated eluent; A Rec – adsorbent recirculation; 
E – extract stream with the better adsorbed component A; F – 

feed stream with the components A and B; R – raffinate 
stream with the less adsorbed component B. Simulated 

moving bed (SMB) liquid chromatograph – I, II, III, IV – 
zones, respectively HPLC columns; CM – direction of 

simulated moving of HPLC  columns. 
 

The inlet liquid stream of the first zone is the 
mixture of the fresh and recirculated eluent. The 
first column of the first zone has to be regenerated 
till the end of each switching time period to protect 
carrying strongly adsorbed components by 
adsorbent phase. The inlet liquid stream of the 
second zone is the mobile phase from the first zone 
minus the flow stream of the extract. This flow 

stream must be determined so that the less binding 
component could leave the first column of the 
second zone till the end of the switching period 
avoiding to get into the extract. The inlet liquid 
stream of the third zone is the mobile phase from 
the second zone plus the feed stream to be 
separated. The function of this zone is holding the 
more binding component in the adsorbent phase, 
since the less binding component is taken away as 
a raffinate at the end of the zone. Four-zone SMB 
is favourable when retarding the less binding 
component – regenerating the solvent. The 
regenerated solvent can be recycled and added to 
the fresh eluent. In case of recycled solvent the 
system is called closed loop SMB. This version is 
more favourable compared to the open loop system 
from economic and environmental point of view. 
Three-zone SMB is preferred in systems with high 
selectivity factor, when the less binding component 
has low capacity factor flowing nearly together 
with the mobile phase [1,2]. 

According to the above facts the operational 
parameters of the process are the switching time, 
the flow rates of the mobile phase in each zones 
determined by the external flow rates (fresh and 
recycled eluent, extract, feed, raffinate). 

Summing up the possibilities for amino acid 
preparative separation the following 
chromatographic methods were applied in practice: 
ion-exchange column liquid chromatography, ion-
exchange parametric pumping, size-exclusion 
chromatography, reverse phase adsorption 
chromatography. 

In case of reverse phase adsorption 
chromatography for separation of amino acids 
solved in water: the styrene-divynilbenzene 
copolymers with non-polar surface and the 
polymethacrylate resins with weakly polar surface 
can be used in the presence of electrolytes or polar 
solvent. In such systems the adsorption equilibrium 
depends on the temperature, the solvent strength, 
the pH [3,4] and on the electrolyte concentration of 
aqueous solution [2]. 

The design of industrial scale SMB-LC process 
requires numerous preliminary experiments. At the 
selection of the packing we can reduce the number 
of the possible alternatives if we consider the 
chemical character of the adsorbents. The most 
frequently used technique is the determination of 
adsorption selectivity with an analytical HPLC 
instrument with a given adsorbent by injection 
method. The adsorbent is giving the best selectivity 
to be examined further on within small-scale or 
large-scale laboratory circumstances. We examined 
the model samples by frontal adsorption-desorption 



 

 

25

method on the small scale lab size column packed 
with polymer adsorbent. The advantage of this 
method is that frontal adsorption and desorption 
processes of the SMB-LC can be investigated. 

After determining the equilibrium data of the 
selected systems and the column packing 
characteristics the initial operating parameters of 
the SMB can be calculated. The initial operating 
parameters for a three zone open loop SMB was 
calculated by the method of Morbidelli et al. [5].  
The first zone regeneration is appropriate when:  

)L(1

LT
A
D

mK IA ε

ε

−

−
=<     (1) 

The less binding component must be removed 
from the second zone till the end of the switching 
time. The function of the third zone is the retarding 
of the better-adsorbed component, namely this 
component must not break through the third zone: 

AIIB K)L(1

LT
A

ED

m<K <
−

−
−

=
ε

ε
   (2) 

 

AIIIB K)L(1

LT
A

F+ED

m<K <
−

−
−

=
ε

ε
   (3) 

In a two component-system a mII-mIII area can 
be determined from the geometric data of the SMB 
equipment, the equilibrium parameters of 
Langmuir-type isotherms and the concentrations of 
the mixture to be separated. 

The given operating conditions determined a 
point in the mII-mIII diagram. This point must be 
placed in the Morbidelli-area (or Morbidelli-
triangle), then both products stream purity are 
theoretically 100%. When the optimal 
chromatographic packing is selected, the following 
operating variable can have influence on the SMB-
LC operation: fresh eluent, recirculated eluent, 
feed, extract and raffinate flow rates, column 
switching time. During calculation the influence of 
a given parameter can be investigated while the 
others must be considered as constants.  

Better productivity, product purity, yield, 
eluent consumption can be achieved by optimizing 
operating conditions by computer.  

The Morbidelli-area changes with the change 
of the following parameters during the planning: 
geometrics of SMB equipment, temperature and 
exchange of adsorbent, composition of fresh eluent 
and feed. 

During the planning of a preparative 
chromatography it must be considered, that the 
models of linear chromatography described is not 
appropriate the system. Many research schools are 

dealing with the mathematical modeling of the 
nonlinear chromatography [6,7]. 
 
 

Experiments 
 
 
The examined model systems during research are 
the aqueous solution of DL-β-phenylalanine and 
sodium-chloride, and the aqueous solution of 
glycine and DL-β-phenylalanine. 
 
 

Select reversed phase packing 
 
 
In case of L-phenylalanine and glycine amino 
acids separation in water the next reversed phase 
non-polar polymer adsorbents were studied by 
frontal adsorption-desorption methods:  DIAION 
HP20, SEPABEADS SP825. For desalting DL-β-
phenylalanine DIAION HP2MG, Amberlite XAD7 
polymethacrylate, Amberlite XAD1180, DIAION 
HP20 styrene-divynilbenzene copolymer resins 
were investigated. 

Parameters of research: measurement went on 
glass column in which resin bed length was 122 mm, 
I.D. 13 mm. L-phenylalanine concentration 0.005 
mol L-phenylalanine/dm3 aqueous solution, glycine 
concentration 0.0025 mol glycine/dm3 aqueous 
solution, volumetric stream 2.9 ml/min. L-
phenylalanine was detected on-line by GILSON116 
UV-spectrophotometer at 271 nm wavelength. 
Glycine and phenylalanine analyses went on 
afterwards, samples fractionally taken in each 4 ml 
by OE914 amino acid analyzer. While desalting 0.02 
mol/dm3 DL-β-phenylalanine aqueous solution and 
1mol sodium-chloride/dm3 aqueous solution, the next 
detection was used during adsorbent selection. The 
phenylalanine was detected on-line UV at 271nm, 
NaCl by conductivity meter with flowing through 
cuvette. Ion exchanged water was used for all 
experiments as desorbents. 
 
 

Measurement of packing properties 
 
 
Frontal adsorption–desorption method for 
adsorbent selection gave the best packing filled by 
slurry technique into SMB-LC column. After 
packing, the number of theoretical plate (NTP) was 
measured in function of volume stream for 
L=500mm I.D.=50mm preparative SMB-column 
in 30-180 cm3/min range using Na2SO4  detecting 
compound. In case of L=125mm ID=13mm small-



 

 

26 

laboratory SMB column NTP and column porosity 
(ε) were determined in 3.0-22 cm3/min range by 
injection method with 0.25 mol Na2SO4/dm3 water 
solution at 20°C. The answer function of the 
injection was detected by a RADELKIS OK102/1 
conductivity meter with flow-through cuvette. This 
date is necessary for modeling, too. Measurement 
of bulk density: resin being packed in a column 
was quantitatively removed and placed in max 5 
mm thin layer on a plate in known mass and dried 
at 85-90°C till constant mass, then re-measured.  
 
 

Measurement of equilibrium isotherms 
 
 
Multi-step frontal adsorption method was used in a 
glass column (L=122 mm, ID=13 mm) packed 
with DIAION HP20 respectively SEPABEADS 
SP825 at 20°C in 0.002 mol/dm3…0.02 mol/dm3 
aqueous solution concentration range with 0.002, 
0.005, 0.01, 0.015, and 0.02 mol amino acid /dm3 
water solutions, cc. 2 cm3/min volumetric stream. 
The 60°C isotherm measurement went on jacketed 
glass column equipped with plunger with 147 mm 
packing length and ID 12 mm similarly to the 
measurement at 20°C. L-phenylalanine was 
detected at 271 nm, glycine at 230 nm by UV-
spectrophotometer. The equilibrium isotherms of 
DL-β-phenylalanine were measured in ion-
exchanged water, further in 0.25 and 1 mol NaCl / 
dm3 aqueous solutions. Equilibrium data were used 
for planning processing parameters and 
mathematical models. 
 
 

Description of production-scale SMB-LC and 
small-laboratory SMB-LC equipments 

 
 
There are geometric and other parameters of the 4-
column production scale SMB-LC equipment 
being constructed by the Central Workshop of 
University of Veszprém: column number is four 
and made of stainless steel. Useful column length: 
500 mm, ID 50 mm. Special hole-channel system 
and 50 µm stainless steel fritts assure liquid 
distribution and collection on the full cross-section. 
Columns are supplied with stainless steel 
thermostable jacket. Joining pipes are 1/8’’ made 
of stainless steel. For switching time controlling of 
input and output sites 4 pieces of 4-way 5 set and 4 
pieces of 2 way 3 set cocks were built in. Above 
all, there are 2 pieces of 2 ways 3 set deairation 
cocks built in for removing air from pump outputs. 

The two input streams are assured by membrane 
pumps in 0-250 ml/min region; output streams are 
forwarded by two plunger type pumps. Pumps are 
placed on separate scaffolds. The third output 
point, where regenerated solvent is taken away in 
the four-zone running, is open to the atmosphere 
without recirculation. There are the working 
parameters of the 4-column SMB-LC: a four-zone 
1-1-1-1-column configuration and a 3-zone 1-1-2-
0, 2-1-1-0 and 1-2-1-0 column configuration. The 
listed working methods have been built in the 
program of the PLC controlled automation.  

The six-column, three-zone small-laboratory 
SMB-LC was built of six glass columns of equal 
geometry, with packing length of 127 mm and with 
inner diameter of 13 mm, a six-position low 
pressure multifunction valve, a time actuator and 
Teflon pipes. The upper part of the valve is fixed 
through which the inlet of the sample and the eluent 
is carried out. The lower part is the rotating one 
which revolves along with the columns by a column 
position in every given time interval (Fig. 2.) 
 

 
 

Figure 2. Photography of a) production scale SMB and b) 
small-lab SMB equipment. 

 
Parameters of SMB-LC measurements 

 
 
The L-phenylalanine – glycine – water separation 
was carried out on the production scale SMB 
equipment (AA03, AA06). The initial experiment at 
20°C, later on at 60°C was improved with the help 
of the equilibrium cascade model. The 
desalinization process of DL-β-phenylalanine was 
carried out on the small-lab scale SMB (SMB 3, 4, 
5) unit. We varied the flow streams of eluent, feed, 
raffinate, extract and the switching time (Table I.) 

a) b) 



27 
Table I. Operating conditions of desalting of phenylalanine with the small-laboratory scale six-column SMB (SMB3, SMB4, 

SMB5), and of phenylalanine-glycine separation with the preparative four-column SMB (AA03, AA06). 
 

Column configuration Temperature T D E F R
Phe NaCl (min) (cm3/min) (cm3/min) (cm3/min) (cm3/min)

Column configuration Temperature T D E F LROUT
Phe NaCl (min) (cm3/min) (cm3/min) (cm3/min) (cm3/min)

AA06 3.3 1.5 2:1:1:0 43.5 78.8

cF (g/dm
3)

20.3 60.6

60°C 30 237.4 202.1

AA03 3.3 1.5 2:1:1:0 20°C 45 241.9 201.6

1.21 5.11

Prep. 
SMB

0.83 3.93

SMB5 3.5 58.5 2:2:2 20°C 12.5 6.95 3.06

20°C 15 5.55 2.45SMB4 3.5 58.5 2:2:2

1.23 0.33 1.962:2:2 20°C 26 2.86

cF (g/dm
3)

SMB3 3.5 58.5

Small lab 
SMB

 
 

 
SMB product stream analysis 

 
 
Production scale SMB unit – phenylalanine – 
glycine separation: Both extract and raffinate were 
collected in separate reservoir by switching times. 
After shaking it, sample was taken out of the 
reservoirs and analyzed by AminoChromII OE914 
amino acid analyzer. Over the average sample, part 
sample was taken in the last switching time by a 
given periodicity to examine concentration 
transient within cycle [quasi-stationary state]. Half 
a minute sample was taken in every 7 minutes at 
the 20°C measurement, at 60°C first in the second 
minute, then by 5 minutes.  

Summarizing the amino acid analysis: glycine 
and L-phenylalanine were separated on high 
efficiency Durrum DC-4A cation exchange column 
at constant pH and ion strength. Components 
leaving the column were mixed with ninhydrine 
reagent and reacted in capillary pipe reactor at 80 
°C for ca. 10 min, the forming colorful products 
were detected by spectrophotometer at 570 nm. 
After proper dilution the samples were injected by 
automatic injector from the 30 µl loop (extract was 
not diluted, raffinate and feed in 5 times dilution). 
Used eluent pH is 4.25, ion strength is 0.2, volume 
stream is 20 ml/hour. Ninhydrine volume stream is 
10 ml/hour. Amino acid content of the sample is 
proportional to the area beneath the 
chromatographic peak. Amino acid mixture was 
used as calibrating standard with 0.0005 mol/l 
amino acid concentration. The analysis time was 
ca. 25 min/sample. 

Small-lab SMB unit – phenylalanine desalting: 
both in extract and raffinate stream on-line UV 
detection of amino acid at 271 nm wavelength. Salt 
detection: on-line by conductivity meter with 
flowing-through cuvette. 
 
 

Results 
 
 

Packing selection 
 
 
Frontal measurements were evaluated by the next 
capacity relation: 

NaCl

NaClaminoacid'

V
VV

k
−

=      (4) 

Dead volume (VNaCl) was defined by 1 mol 
NaCl/dm3 water injection (VNaCl). The following 
capacity factors were given at 20°C for the selected 
resins: 

HP20 for desalinization: DL-phenylalanine 
k’=4.20 

SP825 for glycine-phenylalanine separation: L-
phenylalanine k’=12.99; glycine k’=0.73 
(Table II.) 

 
 
 
 
 
 
 
 
 
 
 
 



28 

 

Table II. Capacity factors of the investigated resins. 
 

NaCl Phe NaCl Phe

XAD7 123.00 3.00 13.80 29.30 1.12

58.40 3.00 13.80 23.40 0.70

58.40 3.00 13.80 23.80 0.72

HP2MG 58.50 2.70 12.10 33.20 1.74

58.50 3.22 12.10 21.00 0.74

XAD1180 58.50 2.70 14.00 62.90 3.49

58.50 3.22 14.00 56.00 3.00

HP20 58.50 2.70 12.50 75.30 5.02

58.50 3.22 12.50 65.00 4.20

Gly Phe Gly Phe

HP20 1.50 3.30 14.85 62.70 0.19  4.03

SP825 1.50 3.30 16.50 133.65 0.73  12.99

Adsorbent for Gly-
Phe separation

Sample composition 
[g dm-3]

Volume/breaktrough 
curve inflection point 

[cm3]

Capacity 
factor  k'     
Gly   Phe

Capacity 
factor       

k'

Adsorbent for Phe 
desalinization

Sample composition 
[g dm-3]

Volume/breaktrough 
curve inflection point 

[cm3]

 
 
 

Equilibrium measurement 
 
 

L-phenylalanine and glycine adsorption on SP825 
resin can be written by Langmuir equilibrium 
isotherms. The amino acid isotherms at 20°C (Fig. 
3.a):  

Gly

Gly
Gly 0.9291c1

0.5344c
q

+
=   and

 
Phe

Phe
Phe 0.1621c1

34.01c
q

+
=    (5) 

At 60°C: 

Gly

Gly
Gly 0.9535c1

0.353c
q

+
=   and

 
Phe

Phe
Phe 0.671c1

23.47c
q

+
=    (6) 

The unit of liquid phase concentrations (cPhe, 
cGly) is mg amino acid/cm3 solution, the solid phase 
concentrations (qPhe, qGly) are in mg amino acid/ g 
dry resin. Rising temperature solid phase amino 
acid quantity decreases. 

0

10

20

30

40

50

60

70

80

0 0,5 1 1,5 2 2,5 3 3,5 4
c [mg amino acid/ml solution]

q 
[m

g 
am

in
o 

ac
id

/g
 d

ry
 a

ds
or

be
nt

]

0

0,5

1

1,5

2

2,5

3

3,5

4

4,5

5

Phe t=20°C
Phe t=60°C
Gly t=20°C
Gly t=60°C

SEPABEADS

 
(a) 

 

0

0,05

0,1

0,15

0,2

0,25

0,3

0,35

0 0,005 0,01 0,015 0,02 0,025
c phenylalanine [mol/dm3]

q 
ph

en
yl

al
an

in
e 

[m
ol

/d
m

3 ]

DIAION HP20

 
(b) 

 

Figure 3. a) Adsorption isotherms of glycine and of L-
Phenylalanine in aqueous solution at temperature 20°C and 

60°C on SP825 resin. c: liquid phase amino acid concentration 
in mg amino acid/ cm3 liquid, q: solid phase amino acid 

concentration in mg amino acid/ g dry adsorbent. b) Isotherms 
of DL-β-phenylalanine on HP20 resin. Langmuir parameters 

depends on the NaCl concentration of the liquid phase: in 
water (marker *): a=20.86, b=33.5; in 0.25 M NaCl solution 
(marker +): a=22.37, b=32.1; in 1 M NaCl solution (marker 

x): a=26.94, b= 32.8. The dimension of Langmuir-b 
parameter: cm3 mmol-1. Langmuir-a is dimensionless. 
 
 
The investigated adsorption of the DL-β-

phenylalanine in aqueous phase and in salt solution 
on HP20 adsorbent can be characterized also by 
Langmuir isotherms (Fig. 3.b). Increase of the 
concentration of the salt solution means the DL-β-
phenylalanine concentration increase in the resin 
phase. Taking into consideration the dependence of 
the “a” parameter of the Langmuir equation on the 
salt concentration it can be written as: 

Phe

PhePheNaCl
Phe 32.8c1

20.86cc6.08c
q

+
+

=   (7) 

The concentration units are in mmol amino 
acid/ ml solution (cNaCl, cPhe), respectively in mmol 
amino acid/ ml packing total volume (qPhe).  
 



29 

Packing characteristics 
 
 
SMB-LC columns were characterized by Na2SO4 – 
water solution injection at different volumetric 
streams. The production-scale SMB-columns 
I.D.=50mm L=500mm were packed with 0.3 mm 
particle diameter SP825 resin by water slurry 
technique. Each column was filled with 265.0 g 
SP825 adsorbent air dried till constant mass value. 
For example the NTP of the column No.2. can be 
determined with the following equation: 

1.20350.0043u
50

NTP
+

=     (8) 

Unit of “u” : cm3/min mobile phase flow rate. 
The NTP is defined with this equation for 500 mm 
packing length. Overall porosity of columns were 
determined by salt-solution injection and ε=0.59 
value was received for all the four columns. The 
bulk density value is 0.27 g dry resin/cm3 column 
volume. 
The NTP of the small-lab scale SMB packed with 
16.0 g HP20 resin per column can be determined: 

0.02910.00225v
125

NTP
0 +

=    (9) 

Unit of „v0” : mm/min mobile phase velocity. 
Overall porosity: ε=0.71 . 
 
 

Results of SMB-LC measurements 
 
 
Results of measurement at 20°C on 2-1-1-0 
column configuration of the production-scale four-
columns SMB unit: more than 99.9% m/m amino 
acid purity in extract for L-phenylalanine and in 
raffinate for glycine. The yields are over 99% for 
both amino acids. Productivity depending on the 
feed, volumetric velocity and concentration is 
relatively low. L-phenylalanine productivity is 
3.76 mg/g SP825/h, glycine productivity is 1.72 
mg/g SP825/h (AA03 measurement).  

In case of measurement at 60°C (AA06 
measurement) on 2-1-1-0 column configuration the 
feed volumetric velocity was increased from 25 to 
45 ml/min, so productivity was increased 1.8 times 
and less diluted product output was given. Values 
of purity are over 99.9% m/m and the yields over 
99% m/m. Desalting DL-β-phenylalanine (SMB 3, 
4, 5 measurements): productivity for phenylalanine 
was 0.67-2.3 mg/g HP20/h, yield 89-96% and 
amino acid purity 98.5% m/m (Table III.) 
 

Discussion 
 
 

Planning of SMB-LC processing parameters 
 
 

The task of the first zone in the 4 zone 
recirculation SMB-LC is the regeneration of 
adsorbent, namely the washing down and 
desorption of the better binding component, till the 
end of switching time. The second and third zones 
are the separation zones. The less binding 
component must be removed from the second zone 
till the end of the switching time. The function of 
the third zone is the retarding of the better binding 
component on the stationary phase. In the fourth 
zone a portion of the eluent is recovered purely, so 
here the less adsorbing component is to be 
adsorbed, retained. Because of the very weak 
glycine respectively NaCl adsorption in this water 
phase system, very low recirculation is possible – 
not much more than empty volume liquid quantity 
calculated from column total porosity. Thus, the 
fourth zone was neglected and the work was 
continued in the open three-zone SMB-LC system.  

 

0

5

10

15

20

25

0 5 10 15 20 25

mII

m
II

I

Morbidelli triangle t=20°C
AA03 measurement t=20°C
Morbidelli triangle t=60°C
AA06 measurement t=60°C

 
(a) 

0

5

10

15

20

0 5 10 15 20

mII

m
II

I

KPhe
mII=mIII
SMB3
SMB4
SMB5

 
 

(b) 
 

Figure 4.The measurements points placed in the Morbidelli 
triangle: a) phenylalanine-glycine separation at temperature 

20°C (AA03) and 60°C (AA06)  b) desalting of phenylalanine 
(SMB 3, 4, 5). 



30 
Table III. Operation parameters of the SMB separations. 

 

SMB3

SMB4

SMB5

Phenylalanine Glycine Phenylalanine Glycine Phenylalanine Glycine

AA03, t=20°C 3.760 1.724 >99.9 >99.9 >99.9 >99.9

AA06, t=60°C 8.130 3.697 >99.9 >99.9 >99.9 >99.9

Small lab SMB

Preparative SMB

Phenylalanine
95.58

93

89.072.302

Phenylalanine
98.34

98.5

98.21
Productivity [mg/gh] Purity [%] Yield [%]

Productivity [mg/gh] Purity [%] Yield [%]
Phenylalanine

0.667

1.645

 
 
 
 

Beside the prescribed purity and yield in 
industrial production the productivity must be the 
highest, the solvent use the less and adsorbent 
utility is the best. The initial operating conditions 
were planned with the help of Morbidelli’s 
equilibrium method (Fig. 4.). The first parameters 
can be improved, while increasing the feed value 
or concentration. An obvious possibility is the 
increase of all flow rates (eluent, feed, extract, 
raffinate) proportionally and decrease the 
switching time. We used this method for desalting 
of DL-β-phenylalanine on the small-lab SMB. 
There are kinetic limits of the flow rate increase 
(Fig. 5.). Other possibility is to improve the 
regeneration of the first zone for example by 
increasing temperature. With this method less fresh 
eluent is necessary, thus we can increase the feed 
flow rate and so the productivity improves. By the 
phenylalanine–glycine separation lower selectivity 
was measured at higher temperature. The initial 
steep of adsorption equilibrium isotherm 
decreased, therefore L-phenylalanine desorption 
went on easier. Thus the switching time could be 
reduced from 45 min to 30 min, so feed stream was 
increased from 20.3 ml/min to max. 43.5 ml/min. 
 
 

Conclusions 
 
 
We planned initial parameters with the help of 
Morbidelli’s equilibrium method for SMB 
separation in water of phenylalanine-glycine, 
respectively phenylalanine–sodium-chloride model 
systems on polymer adsorbents. We investigated 
two ways to improve the productivity. The increase 
of all flow streams and the decrease of period time 
in the desalting of DL-β-phenylalanine is limited 
by adsorption and desorption kinetics of the amino 
acid: the productivity was increased 3 times, but  

the yield decreased from 95.6% to 89% (SMB 3, 4, 
5 measurements).  
 

0

2

4

6

8

10

12

14

0 12.5 25 37.5 50 62.5 75 87.5

measurement time [min]

U
V

, c
on

du
ct

iv
ity

 s
ig

na
l [

m
V

]

 
(a) 

 

0

5

10

15

20

25

0 12.5 25 37.5 50 62.5 75 87.5

measurement time [min]

U
V

, c
on

du
ct

iv
ity

 s
ig

na
l [

m
V

]

 
(b) 

 
Figure 5. Measured concentration commensurable signals by 

SMB5 experiment in a) raffinate and b) extract. Markers: 
signal of phenylalanine (+) and signal of NaCl (*). 

 
The phenylalanine–glycine SMB-LC system 

temperature was risen from 20°C to 60°C (AA03, 
AA06 measurements), thus productivity was 
increased 1.8 times. Rising temperature gives 
solution only for an optimal value because 
application of ventiles, cocks, fittings, etc. is 
limited by temperature.  
 
 



31 

 

SYMBOLS 
 
 

D – flow rate of eluent (cm3/min) 
E – flow rate of extract (cm3/min) 
F – flow rate of feed to be separated (cm3/min) 
T – period time or column switching time (min) 
A – cross-section of the SMB-column (cm2) 
L – column length (cm) 
ε – overall porosity 
mI, mII, mIII, mIV – Morbidelli’s parameters 
KA, KB – equilibrium distribution coefficient 
k’ – capacity factor 
Vaminoacid, VNaCl – inflection point of the 

breakthrough curve (cm3) 
qPhe – phenylalanine concentration in stationary 

phase (mg/g) 
cPhe – phenylalanine concentration in mobile phase 

(mg/cm3) 
cNaCl – NaCl concentration in mobile phase 

(mg/cm3) 
 

REFERENCES 
 
 

1. HASHIMOTO K., YAMADA M. and SHIRAI Y.:  
J. of Chem. Eng. (1987), Vol. 20 No. 4, 405-409 

2. MOLNAR Z., NAGY M., HANAK L., SZANYA T. 
and ARGYELAN J.: J. Chromatogr. (2004), Vol. 
60, 75-80 

3. GRZEGORCZYK D. S. and CARTA G.: Chem. Eng. 
Sci., 51 (1996) 807 

4. GRZEGORCZYK D. S. and CARTA G.: Chem. 
Eng. Sci., 51 (1996) 819 

5. MIGLIORINI C., MAZZOTTI M. and MORBIDELLI 
M.: J. Chromatogr. A, 827 (1998) 161 

6. GUIOCHON G.: J. CHROM. A. 965 (2002) 129-
161 

7. HEUER C., KÜSTENS E., PLATTNER T. and 
SEIDEL-MORGENSTERN A.: J. Chromatogr. A, 
827 (1998) 175 

8. SZANYA T., ARGYELAN J., KOVATS S. and 
HANAK L.: J. Chromatogr. A, 908 (2001) 265