6.

1/2 Sorption - introduction

Luuk van der Wielen, TU Delft and Bernal /
U Limerick & Hans Wesselingh, Groningen University

jan 2021/22/24
First Application .....
 Chromatographic separation at home

filtration
paper time

ball point
line
ethanol
 channel 0.01 mm pore 10 nm

adsorbent-
particle 0.1 mm

column
100 mm
protein
4 nm
Dimensions in
chromatography
Sorbent Pellet
most commercial sorbents volume fractions
are pelletized from smaller
units (microporous sorbents ) 0.4 0.1 0.5

adsorbates 

ds ~ 10-10 m
p
microporous selective material
sorbent (1   )(1   p )
dp
dm ~10-6 m
0.1-1 mm
 Four separation mechanisms

+ - + -+ -+ +
- -+ - + -
+- - + - + + -
+ + + +- -
+
- - +- -
- - +
size -+ + charge
(gel filtration) (ion exchange)

polarity chemistry
(hydrophobic interaction) (affinity adsorption)
 Distribution of a weak
solute over liquid q adsorption
and sorbent c

q
c

q q
c

c q strong
c adsorption
 Ion exchange capacity: determined by
number of fixed charged groups and
available space

+ +- -+- ++ - + -+
- - -- +-+
-+ +- - ++ +
- - - +
+
+ - capacity monolaag - +
0.1 x gelvolume
positive gel negative gel
(anion-exchanger) (cation-exchanger)
 Anion exchange capacity

capacity pI (isoelectric point)

-+----
+++-+

2 4 6 pH

capacity

increasing ionic strength

protein concentration
Chromatographic Separations by
...

... velocity differences

 pIA pIB which resin ?

3 5 7 pH 9
A stable
B stable
A capacity
B cation-exchanger

A capacity
B anion-exchanger

A is the desired product.

 1
Chromatography: separation by
wA wB velocity differences

adsorbens

concentration B A

time
eluent
1 2 3 4
Scale-up in
chromatography

1. ‘peak’, low
concentration

2. ‘band’, low
concentration

3. ‘band’, high
concentration
4. gradient in
the eluent
 Adsorption, wash and regeneration

feed regenerant
wash (eluent)
B A B A B A

A
B A A A
B

B B A A A
time
 Concentrate

q pH <

feed in eluent =
“weak” solvent stronger solvent
Practical scale-up of sorption processes

 same liquids
 same velocity
 same column length
 same particles

only the cross-sectional area is enlarged

Countercurrent contactor

fluidization stopped reverse refluidization

flow flow
Cloete Streat contactor

Pilot Plant at National Institute

for Metallurgy (S-Africa)

Full scale plant:

# stages 8
diameter 4.5 m
height 20 m

M. Streat
J Sep Proc Technol. 1(3):10, 1980
 Fresh Adsorbent Centrifugal Adsorption
Technology
Liquid Feed

Liquid Effluent

axis
Adsorbent
Effluent
Liquid effluent Liquid feed

Adsorbent feed Adsorbent

effluent
Centrifugal field: w2 R
w
Bischops, Van der Wielen and Luyben NL Patent 1 002 569 (1996)
 Simulated Moving Bed Chromatography

Typical Lysine recovery plant (under construction) with 30 m3 resin in each unit
 Chromatographic Processes
system process scale
ton/a

• glucose-fructose SMB >100 000

• lactic, citric acid FB/SMB 10 000
• amino acids, antibiotics SMB 1500
• peptides SMB 10
• milk proteins FB >1
• flavors FB 1

• food enzymes FB lab

• chiral compounds SMB 1-10

FB fixed bed; SMB Simulated Moving Bed

 Sorption – column design
Luuk van der Wielen, TU Delft and Bernal /
U Limerick & Hans Wesselingh, Groningen University

jan 2021
 Design Sequence

• mass balance
out in (overall, phase, local)

rate equilibrium • phase and reaction

equilibria (K’s)
in out
• hydrodynamics
mass balance • mass transfer and
reaction rate (k’s)
 Concentration-velocity
particles (1   ) z

A
v
liquid 
concentration z
concentration
c t t  t profile
concentration
velocity w
c
z
w
t
is a function of place
c, q, , v z
 Front velocity
Q  A v

balance over front

in A v c( z )t
accumulation A z c (liquid )
z
w A z (1   ) q ( sorbent )

out A v c( z  z ) t

(average) front velocity

q z v
c w 
c t    (1   ) q 
c  c
 Concentration velocity c  0
Q  A v

balance over differential element

in A v c(z )dt
accumulatie A dz dc (liquid )
dz
w A dz (1   ) dq ( sorbent )

out A v c( z  dz ) dt

velocity of a concentration:
q dz v
m   dq wC  
dc dt    (1   ) dq  derivative !
c  dc 
 Peaks and linear isotherms
H
m
q
v

c

all c’s at same velocity: cz

z=wt
(no peak deformation)
place
cuit increasing m
‘null-peak’ (m = 0)

residence time: O B A t time

cuit decreasing 
H  
t  1  m
v  1  
O B A
 Non-linear isotherm 1

sorbent has a lower capacity at higher

q
concentrations;
m’ = f(c)
high concentration velocities are larger
c

sharpening fronts cz ‘front’

place
non-linear
chromatography

adsorption A
A B
B
 Non-linear isotherm 2

sorbent has a lower capacity at higher

q
concentrations;
m’ = f(c)
high concentration velocities are larger
c

tails become cz ‘tail’

diffuse
place
non-linear
chromatography

regeneration A
A
eluent
 Regeneration is often difficult

low c high c
slow fast

w ~ 1/m c
q

m tail

c t
 Design of a chromatographic
separation

mixture,
feed time tM
wA wB
required
wAtS lenght
• maximum liquid velocity (restricted by wB(tS - tM)
pressure drop and mass transfer)
• column length and cycle time for
required purity
• increase column diameter
• rigorous but not (always) efficient eluent
approach....
 interface Combining Two
qi
q
Resistances
c MT  k aq  q  

 Gk f   k p1
1 1
film ci particle k
z
q overall or lumped coefficient
slope G-1
qi q resin composition that
q would be in equilibrium with
c the liquid
ci dc dc
G 
c dq q i dq q
 Linear Systems#
Two approaches possible:
• (numerical) solution of PDEs describing
pore diffusion and external mass transfer
• linear driving force (LDF) model
# linear isotherm
Fick diffusion
q
 ak (q *  q )  akK (c  c * )
t
with
1 Rp R 2p
 
akK 3k f 15 p KD p
 Mass Transfer: Correlations
0.357
Sh  Re 0.64 Sc 0.33 3  Re  2000 gas

1.903
Sh  Re 0.33 Sc 0.33 0.0015  Re  55 liquid

0.25
Sh  Re 0.69 Sc 0.33 55  Re  1050 liquid

 u od p 104
Re  liquid
Sh
 gas
 0.4
Rei 10
i
1
4
1 10

Sh g Re , Sc g , 
i
1000

100

10
superficial 10-3 Re 103
Sh Re , Sc L , 

velocity
1
i
0.1

0.01

0.001
0.001 0.01 0.1 1 10 100 1000
Re
i
 Analytical solution
G Carta Chem Eng Sci 43(10):2877, 1988
c
tE
tF t
kaz overall mass
N
c v transfer coefficient
ka(t F  t E )
t r
2 (1   )K

c(t ) tF 2  1  j 2N   j  tF 
    exp  2  * sin  
cF t F  t E  j 1 j  j r 
2
tF  tE 
 j  tF 2 j   z jrN  
 cos  t    2 2 
 F E
t  t t F  t E
 v  j  r 
 Summary
• front velocity and shape are crucial in design

equilibrium mass transfer

favorable + +/-
linear +/- -
unfavorable - =

• linear systems: analytical solution

• non-linear systems:
single component - approximations
multi components &
non-isothermal - numerical solutions
 Sorption – Simulated Moving Beds

Luuk van der Wielen, TU Delft and Bernal /

U Limerick & Hans Wesselingh, Groningen University

jan 2021
 The Problem
HSA salt
Human Serum Albumin:
protein
• DEAE Sepharose FF c
• productivity 1 kg HSA/hr
t
• 100 kg/year towards 100 ton/year

current process conventional scale-up

buffer 102 m3/a 105 m3/a
salt 103 kg/a 103 ton/a
V 0.15 m3 150 m3 .... ??
wA wB B
A
c
...
separate
location
“here” !!
withdraw Solution:
product
Fix “peaks” in
instead of ... time/place when
c sufficiently separated
B A

time
 Simulated Moving Bed
technology

Raffinate

Fix in time/place simulated

by (simulated) sorbent
motion of the Feed recycle
sorbent phase
Extract
place
Desorbent
recycle
Desorbent
make-up
Fixed Bed versus Countercurrent Process

fL fL fL

Elution
v
Carrousel System

Switch
 SMB Equipment

Raffinate

Desorbent
Feed recycle

Extract

actual SMB scheme equivalent true

moving bed
Bisschops, Pennings, Tysterman, 2007
WO2007043874A1 (Biopharma SMB)
 x0 x1 Equilibrium ‘stage’
L
‘0’ y1 1 y2 ‘2’
V

assumption: outgoing flows at equilibrium

ratio of transport capacities

Vy V
S K separation
Lx L (extraction)
factor
capacity of auxiliary phase “on top”

(this factor appears in adsorption, absorption, stripping, ...)

 Operating conditions
assume complete separation

V A
III II
B
L3 = L2 + F L2
zF F

A should not move in B should not move in

section III section II
V3 yA> L3 xA V2 yB < L2 xB
yA/xA ~ KA > L3 /V yB/xB ~ KB > L2 /V

or SA > 1 or SB < 1
 Flow Constraints
net flows separation factors

IV r IV SA > 1 SB > 1

f III III SA > 1 SB < 1

II e II SA > 1 SB < 1

d I I SA < 1 SB < 1

A strong key
B weak key
 Flowrate ratio selection:
from separation factor

m i= L i / S linear isotherms
m2 > KB m3 < KA m3 > m2
constraints:
KA
m3 < KA m3 > KB complete
separation 4
m2 < KA m2 > KB m3
3

m2 < KA 2
KB
not relevant; m3 > KB 1
covered by:
KB m2 KA
m3 > m2
“Classical” Columns