Catalytic three-phase reactors

 Gas, liquid and solid catalyst

Function principle
 Some reactants and products in gas phase
 Diffusion to gas-liquid surface
 Gas dissolves in liquid
 Gas diffuses through the liquid film to the liquid
bulk
 Gas diffuses through the liquid film around the
catalyst particle to the catalyst, where the
reaction takes place
 Simultaneous reaction and diffusion in porous
particle
Three-phase reactors – catalyst

 Small particles (micrometer scale < 100

micrometer)
 Large particles (< 1cm)
Catalyst design
Reactors
Bubble column
Flow pattern in bubble column
Tank reactor

 Often called slurry reactor

Packed bed – trickle bed

 Trickle bed
Liquid downflow – trickling flow
 Packed bed, if liquid upflow
Packed bed- fixed bed – trickle bed
Flow chart: trickle bed
Trickle flow
Packed bed
Three-phase fluidized bed
Fluidized bed – flow chart
Monolith catalysts
Flow in monoliths
Monolith channel
Three-phase monolith reactor
Three-phase reactors
Mass balances
 Plug flow and axial dispersion
Columnr eactor
Tube reactor
Trickle bed
Monolith reactor

 Backmixing
Bubble column
Tank reactor
Three-phase reactor
Mass balances
 Mass transfer from gas to liquid, from
liquid to catalyst surface
 Reaction on the catalyst surface
 In gas- and liquid films only diffusion
transport
 Diffusion flow from gas to liquid
c b
 K c b
N Lib  Gi i Li
Ki 1

k Li kGi
Three-phase reactor
Mass transfer
Three-phase reactor
Mass balances
 For physical absorption the fluxes through
the gas- and liquid films are equal
N b
Li N s
Li N s
Gi N b
Gi

 Flux from liquid to catalyst particle =

component generation rate at steady state

N s
Li Ap  ri m p  0
Three-phase reactor
Mass balances
 Flux through the liquid film defined with
concentration difference and liquid-film
coefficient
s
N Li  k Li
s
 cLib  cLis 
 Catalyst bulk density defined by
mcat mcat
B  
VL  LVR
Three-phase reactor
Mass balances

Li Li  b
Li
s
Li 
N  k c  c a p   L  B ri
s s

 ap = total particle surface/reactor volume

Three-phase reactor
Mass balances
 If diffusion inside the particle affects the
rate, the concept of effectiviness factor is
used as for two-phase reactor (only liquid
in the pores of the particles)
 The same equations as for two-phase
systems can be used for porous particles

R j  ej R 'j (c B )
Three-phase reactor – plug flow
Three-phase reactor
- plug flow, liquid phase
 For volume element in liquid phase

 
n Li ,in  N A  n Li ,ut  N Lis Ap
b
Li

 Liquid phase 
d n Li
 N Lib av  N Lis a p
dVR


Three-phase reactor
Plug flow - gas phase
 For volume element in gas phase
 
n Gi ,in  n Gi ,ut  N Gi
b
A

d n Gi
 Gas phase dVR
  N Lib av

 - concurrent
 + countercurrent
Three-phase reactor
Plug flow
 Initial conditions
 Liquid phase
 
n Li  n 0, Li VR  0

 Gas phase, concurrent

 
n Gi  n 0,Gi VR  0

 Gas phase, countercurrent  

n Gi  n 0,Gi V  VR
Three-phase reactor
- plug flow model
 Good for trickle bed
 Rather good for a packed bed , in which liguid
flows upwards
 For bubble column plug flow is good for gas
phase but not for liquid phase which has a
higher degree of backmixing
Three-phase reactor
- complete backmixing
 Liquid phase
 
n Li  n 0 Li
 N Lb av  N Ls a p
VR

 
n Gi  n 0Gi
 Gas phase  N Lb av
VR
Three-phase reactor
- semibatch operation
 Liquid phase in batch
 Gas phase continuous

dnLi
dt
 
 N Lb av  N Ls a p VR

dnGi  
  n Gi  N Li avVR  n 0 Gi
b

dt

nLi  n0 Li t0
 Initial condition nGi  n0Gi t 0
Parameters in three-phase reactors
 Gas-liquid equilibrium ratio (Ki) from
Thermodynamic theories
Gas solubility in liquids (Henry’s constant)

 Mass transfer coefficients kLi, kGi

Correlation equations
DLi DGi
k Li  kGi 
L G
Numerical aspects
 CSTR – non-linear equations
Newton-Raphson method

 Reactors with plug flow (concurrent)

Runge-Kutta-, Backward difference -methods

 Reactors with plug flow (countercurrent) and

reactors with axial dispersion (BVP)
orthogonal collocation
Examples
 Production of Sitostanol
A cholesterol suppressing agent
Carried out through hydrogenation of Sitosterol
on Pd catalysts (Pd/C, Pd/Zeolite)

 Production of Xylitol
An anti-caries and anti-inflamatory component
Carried out through hydrogenation of Xylose on
Ni- and Ru-catalysts (Raney Ni, Ru/C)
Exemple: from cholesterol tol sitostanol
Reaction scheme

A superficially complicated scheme

From laboratory scale
to industrial scale

Slurry, three-phase reactor

Lab reactor, 1 liter, liquid amount 0.5 kg
Large scale reactor, liquid amount 8080 kg

Simulation of large-scalle reactor based on

laboratory reactor
Catalytic reactor

 Semi-batch stirred tank reactor

 Well agitated, no concentration differences appear in
the bulk of the liquid
 Gas-liquid and liquid-solid mass transfer resistances
can prevail
 The liquid phase is in batch, while gas is continuously
fed into the reactor.
 The gas pressure is maintained constant.
 The liquid and gas volumes inside the reactor vessel
can be regarded as constant
Mathematical
modelling

Reaction, diffusion and catalyst

deactivation in porous particles

dci 
  p1  ri  p  r S
dN i r S
 
Particle model dt dr 
 

Rates
ri   ij R j a j
j
Model implementation

 dc 
N i   Dei  i  , where Dei=(p/p)Dmi
 dr 

dci 1   d 2 ci s dci  
 p    p  ij R j a j  Dei  2   
dt   dr r dr  

Boundary conditions
dci
0 r0
dr

 dci 
Dei    k Li  ci  ci  R   rR
 dt  r  R
Catalytic reactor, mass balances

Liquid phase mass balance dci

 N i a p  N GLi aGL
dt

Liquid-solid flux N i  k Li  ci  ci  R  
b
cGi  K i c Lib
N GLi 
Ki 1
Gas-liquid flux 
k Li k Gi
Numerical approach

 PDEs discretizied with finite difference

formulae
 The ODEs created solved with a stiff
algorithm (BD, Hindmarsh)
Rate equations

Surface reaction, rate determining

Essentially non-competetive adsorption of
hydrogen and organics
Kinetics in laboratory scale

Concentrations as a function of reaction time

Kinetics in plant scale

Concentration of organics Hydrogen concentration in liquid phase

Comparison of lab and plant scale

Laboratory Factory
Hydrogen concentration in liquid
phase in plant scale
Hydrogenation of Xylose

D-Xylulose
O
OH hydrogenation CH2OH
CH2OH
HO HO
OH
OH
hydrogenation OH
Isomerization isomerization
CH2OH

D-arabinitol

CH2OH
=O
OH
OH hydrogenation HO
HO OH
OH CH2OH
OH
D -Xylose Xylitol
(aldehyde form)
Modelling results
Xylose hydrogenation

Heavy mass-transfer limitations

Heavy mass transfer Moderate mass-transfer limitations
Moderate mass transfer
50
60
45
40
35
D-xylose
xylitol
D-xylulose
Effect of external 50

D-arabitol
mass transfer 40 D-xylose
conc. / wt-%

30

conc. / wt-%
xylitol
25 30 D-xylulose
20 D-arabitol
20
15
10
10
5
0 0

0
23

69
92
115
138
161
184

230
253

299
46

207

276
23
46
69
0

92
115
138
161
184
207
230
253
276
299

t/min t/min

Light mass transfer

Light mass-transfer limitations

Heavy mass transfer and

Heavy mass-transfer limitations and moderately
deactivated catalyst 50

Moderatly deactivated 45
50 40
D-xylose D-xylose
45 35
xylitol
xylitol

conc. / wt-%
40 30
D-xylulose
35 D-xylulose 25
D-arabitol
conc. / wt-%

30 D-arabitol
20
25 15
20 10
15 5
10 0
5

46

230
23

69
92
115
138
161
184
207

253
276
299
0 t/m in
24

72
96
0

48

144

192

264
120

168

216
240

288

t/min
Gas-liquid reactors
Gas-liquid reactors
 Non-catalytic or homogeneously catalyzed
reactions
Gas phase
Liquid phase ( + homogeneous catalyst)
 Components i gas phase diffuse to the
gas-liquid boundary and dissolve in the
liquid phase
 Procukt molecules desorb from liquid to
gas or remain in liquid
Gas-liquid reactions

 Synthesis of chemicals
 Gas absorption, gas cleaning
 Very many reactor constructions used,
depending on the application
Gas-liquid reaction: basic principle
Gas-liquid reactor constructions
 Spray column
 Wetted wall column
 Packed column
 Plate column
 Bubble columns
 Continuous, semibatch and batch tank reactors
 Gas lift reactors
 Venturi scrubbers
Gas-liquid reactors - overview
Tank reactor
Gas-liquid reactors

 Packed column
Absorption of gases
Countercurrent principle: gas upwards, liquid
downwards
Column packings
 enable a large gas-liquid contact area
 made of ceramics, plastics or metal
 good gas distribution because of packings
 channeling can appear in liquid phase; can be
handled with distribution plates
Plug flow in gas and liquid phases
Gas-liquid reactors
 Plate column
 Absorption of gases
 Countercurrent
 Various plates used as in distillation, e.g.
 Bubble cap
 Plate column

 Packed column
 Absorption of gases
 Countercurrent
 A lot of column packings available; continuous development
Bubble column

Gas-lift -reactor
Bubble column – design examples
Bubble column
Packed column
Packings
Plate column
Gas-liquid reactors

 Gas scrubbers
Spray tower
 Gas is the continuous phase
 In shower !
 Venturi scrubber
 Liquid dispergation via a venturi neck

 For very rapid reactions

Spray tower
Venturi scrubber
Gas-liquid reactors

 Selection criteria
Bubble columns for slow reactions
Sckrubbers or spray towers for rapid reactions
Packed column or plate column if high reatant
conversion is desired

Mass balances

mol
Ni A    2   m2
m s
Gas-liquid reactors
Mass balances
 Plug flow
 
n Li ,in  N A  ri VL  n Li ,ut
b
Li

 Liquid phase d nLi

 N Lib av   L ri
dVR


d nGi
 Gas phase   N Gib av
dVR

 av =gas-liquid surface area/reactor volume

 L = liquid hold-up
Gas-liquid reactions
Mass balances
 Complete backmixing
 
n Li ,in  N A  riVL  n Li ,ut
b
Li

 
n Li  n 0 Li
 Liquid phase  N Lib av   L ri
VR

 
n Gi  n 0Gi
 Gas phase VR
  N Gib av

 av =gas-liquid surface area/reactor volume

 L = liquid hold-up
Gas-liquid reactors
Mass balances
 Batch reactor

 Liquid phase
dnLi
dt

 N Li
b

av   L ri VR

dnGi
 Gas phase   N Gi
b
avVR
dt
 av =interfacial area/reactor volume
 L = liquid hold-up
Gas-liquid reactors
- Gas-liquid film
 Fluxes in gas-liquid films
NbLi NbGi

 Two-film theory
Chemical reaction and molecular diffusion proceed
simultaneously in the liquid film with a thickness of
L
Only molecular diffusion in gas film, thickness G
 Fick’s law can be used: N Gib   DGi  dcGi   dc 
N Lib   DLi  Li 
 dz  z  G  dz  z  L
Gas-liquid reactors
Gas film
 Gas film, no reaction
 dcGi   dcGi 
 DGi  A   DGi  A
 dz  in  dz  ut

d 2 cGi
DGi
dz 2
0 b
N Gi 
 kGi cGi
b
 cGi
s

 Analytical solution possible
 The flux depends on the mass transfer
coefficient and concentration difference
Gas-liquid reactors
Liquid film
 Diffusion and reaction in liquid film:
 dcLi   dcLi 
  DLi  A  ri Az   Li
D  A
 dz  in  dz  ut

d 2 cLi
DLi 2
 ri  0
dz

 Boundary conditions:
N Gib  N Li vid z  0
cLi  cLib vid z   L
Gas-liquid reactors
Liquid film
 Liquid film
Equation can be solved analytically for
isothermas cases for few cases of linear
kinetics; in other case numerical solution should
be used
Reaction categories
 Physical absorption
 No reaction in liquid film, no reaction in liquid bulk

 Very slow reaction

 The same reaction rate in liquid film and liquid bulk –
no concentration gradients in the liquid film, a pseudo-
homogeneous system
 Slow reaction
 Reaction in the liquid film negligible, reactions in the
liquid bulk; linear concentration profiles in the liquid film
Reaction categories
 Moderate rates
 Reaction in liquid film and liquid bulk
 Rapid reaction
 Chemical reactions in liquid film, no reactions in bulk
 Instantaneous reaction
 Reaction in liquid film; totally diffusion-controlled
process
Concentration profiles in liquid film
Enhancement factor

 Real flux/flux in the presence of pure

physical absorption

s
N LA
EA  b
cGA  K A c LA
b

KA 1

k LA k GA

 EA  1
Gas-liquid reactors
- very slow reaction
 No concentration gradients in the liquid
film
 Depends on the role of diffusion resistance
in the gas film
s b
cGA cGA
KA  b KA  b
cLA cLA

b
N GA  N LA
b

 k GA cGA
b
 K A c LA
b

Gas-liquid reactors
- slow reaction
 Diffusion resistance both in gas- and liquid- film
retards the adsorption, but the role of reactions
is negligible in the liquid film
b
N GA  kGA cGA
b

 cGA
s

b
N LA  k LA c s
LA
b
 c LA 
b b
b cGA  K A c LA
N LA 
KA 1

k LA k GA
Gas-liquid reactors
- moderate rate in liquid film

 Chemical reactions in liquid film

b
N GA  N LA
b Reaction in liquid film
b
N GA  N GA
s
 N LA
s
No reaction in gas film

s
cGA d 2 cLA
KA  s
c LA
DLA 2
 rA  0
dz

 The transport equation should be solved numerically

Moderate rate in the liquid film

 Transport equation can be solved

analytically only for some special cases:

 isothermal liquid film – zero or first order

kinetics

Approximative solutions exist for rapid second

order kinetics
Moderate rate…

 Zero order kinetics

d 2 cLA  Ak
 
dz 2 DLA

s
N LA 
b
cGA b
 K A c LA  M  1
KA 1

k LA k GA

 A kDLA
M  2 b
2k LAcLA
Moderate rate…
 First order kinetics
d 2 cLA  A kcLA
 
dz 2 DLA
b
b K A c LA
cGA 
cosh( M )  A kDLA  Ak 2
s
N LA  M  2
  L
tanh M K A 1 k LA DLA

M k LA k GA

 Hatta number Ha=(compare with Thiele

modulus)
Rapid reactions

 Special case of reactions with finite rate

 All gas components totally consumed in
the film; bulk concentration is zero, cbLA=0
Instantaneous reactions

 Components react completely in the liquid

2
film D
LA
d c
dz
0 LA
2

 A reaction plane exists

 B L

 Reaction plane coordinate z' 

A
DLB cLB
B
s
 s
DLA cLA DLB cLB
Instantaneous reactions
 Enhancement factor: b
 A DLB c LB KA
EA  1 b
 B DLA cGA

 Flux at the interface: b

cGA 
 A DLB b
K A c LB
s  B DLA
N LA 
KA 1

k LA k GA

 Coordinate of the
interface:
d 2 cLA
DLA 0
dz 2
Instantaneous reactions
 Flux c 
b  A DLB b
K A cLB
GA
 B DLA
s
N LA 
KA 1

k LA kGA

 Only diffusion coeffcients affect !

 For simultaneous reactions can several
reaction planes appear in the film
Fluxes in reactor mass balances
 Fluxes are inserted in
mass balances
b s s
N Gi  N Gi  N Li
b s
 For reactants: Ν Li  Ν Li
 For slow and very
slow reactions: (no b s
reaction in liquid film) Ν Li  Ν Li
General approach
 We are left with the model for the liquid
film:

d 2 c Li
DLi 2
 ri  0
dz

 dcLi 
N   DLi 
b
Li 
 dz  z  L
Solution of mass balances

 Numerical strategy:
Algebraic equations
 Newton-Raphson method
Differential equations, initial value problem
(IVP)
 Backward difference- and SI Runge-Kutta-methods
Differential equations, BVP
 orthogonal collocation or finite differences
Number of equations
 N = number of components in the system

 N eqs for liquid phase; N eqs for gas phase

 N eqs for the liquid film
 Energy balances
1 for gas phase
1 for liquid phase

 3N+2 equations in total

Mass transfer coefficients

 Flux through the gas film

b
N GA  N GA
s

 kGA cGA
b
 cGA
s


 Partial pressures often used:

b
N GA  N GA
s

 k 'GA p A  p As 

 Ideal gas law gives the relation: k  k ' RT

GA GA
Gas-liquid equilibria
 Definition s
cGA
KA  s
cLA

 For sparingly soluble gases:

 Relation becomes ps p As
He A  A He' A  s
s
cLA xLA

 KA from thermodynamics; often Henry’s constant

is enough He A
KA 
RT
Simulation example

 Chlorination of p-kresol
p-cresol + Cl2 -> monocloro p-kresol + HCl
monocloro p-kresol + Cl2 -> dichloro p-kresol +
HCl
 CSTR
Newton-Raphson-iteration
 Liquid film
Orthogonal collocation
Chlorination of para-cresol in a CSTR
Fluid-solid reactions
 Three main types of
reactions:
 Reactions between gas
and solid
 Reactions between
liquid and solid
 Gas-liquid-solid
reactions
Fluid-solid reactions

 The size of the solid phase

Changes:
 Burning oc charcoal or wood
Does not change:
 oxidation av sulfides, e.g. zinc sulphide --> zinc oxide
Reactors for fluid-solid reactions

 Reactor configurations
Fluidized bed
Moving bed
Batch, semibatch and continuous tank reactors
(liquid and solid, e.g. CMC production, leaching
of minerals)
Processes and reactors
Fluid-solid reaction modelling
 Mathematical models used
 Porous particle model
 Simultaneous chemical reaction and diffusion throughout
the particle
 Shrinking particle model
 Reaction product continuously removed from the surface
 Product layer model (shrinking core model)
 A porous product layer is formed around the non-reacted
core of the solid particle
 Grain model
 The solid phase consists of smaller non-porous particles
(rasberry structure)
Fluid-solid reactions

 Solid particles react with gases in such a

way that a narrow reaction zone is formed

 Shrinking particle model can thus often be

used even for porous particles
 Grain model most rrealistic but
mathematically complicated
Product layer
Product layer

Concentration profiles in the

product layer
Shrinking particle
Grain model
Fluid-solid reactions

 Particle with a porous product layer

Gas or liquid film around the product layer
Porous product layer
The reaction proceeds on the surface of non-
reacted solid material
 Gas molecules diffuse through the gas film and
through the porous product layer to the surface of
fresh, non-reacted material
Fluid-solid reactions

 Reaction between A in fluid phase and B

in solid phase
mol
RA  () 2 ()m 2
m s
 R=reaction rate, A=particle surface area
 Generated B= Accumulated B
 B R c s 
dr M

dt xB  P
Fluid-solid reactions

 Diffusion through the porous product layer

(spherical particle)
d 2c A 2 dc
A
DEA (  )0
dr 2 r dr
 Solution gives NA=DeA(dcA/dr):

NA 

DEA c bA  c As    A R (c s )
 r 1  
r 1  1  

 R  BiM  
Fluid-solid reactions

 Fick’s law is applied for the diffusion in the

product layer gives the particle radius
dr M  B DeA  c bA  c As 

dt xB  p  r 
  A  r 1  1  1  
 R  BiAM 

 Surface concentration is obtained from


Dei cib  cis    i R (c s )
 r 1 
r 1  1   

 R  BiiM 
Fluid-solid reactions

 For first-order kinetics an analytical

solution is possible

 Four cases – rate limiting steps

Chemical reaction
Diffusion through product layer and fluid film
Diffusion through the product layer
Diffusion through the fluid film
Fluid-solid reactions
 Reaction time (t) and total reaction time
(t0 ) related to the particle radius (r)
 Limit cases
Chemical reaction controls the process –
Thiele modulus is small -> Thiele modulus
small
Diffusion through product layer and fluid film
rate limiting -> Thiele modulus large
Reaktorer med reaktiv fast fas
Diffusion through the product layer much slower
than diffusion through the fluid -> BiAM=

Diffusion through fluid film rate limiting ->

BiAM=0
Fluid-solid reactions

 Shrinking particle

Phase boundary
Fluid film around particles
 Product molecules (gas or liquid) disappear
directly from the particle surface
 Mass balance

In via diffusion through the fluid film + generated = 0

Fluid-solid reactions

 First order kinetics

Surface reaction rate limiting
Diffusion through fluid film rate limiting

 Arbitrary kinetics
A general solution possible, if diffusion through
the fluid film is rate limiting
Semibatch reactor

 An interesting special case

Semibatch reactor
 High throughflow of gas so that the concentrations in
the gas phase can be regarded as constant; used
e.g. in the investigation of gas-solid kinetics
(thermogravimetric equipment)
 Complete backmixing locally
 simple realtions between the reaction time and the
particle radius obtained
Reaction time and particle radius

 Thiele modulus, φ=-νAkR/DeA and Biot number,

BiM=kGAR/DeA
t 6(1  (r / R)  3 (1  (r / R ) 2 )  2 (1  (r / R ) 3 )(1  1 / BiM )

t0 6   (1  2 / BiM )

 Special cases – large Thiele modulus φ;

 control by product layer and fluid film

2 3
t 3(1  (r / R) )  2(1  (r / R) )(1  1 / BiM )

t0 1  2 / BiM
Fluid-solid reactions

 Product layer model

 Large Thiele modulus, φ=-νAkR/DeA and

large Bi - control by product layer
t
 1  3(r / R) 2 )  2(r / R) 3
t0

 Large Thiele modulus, φ=-νAkR/DeA and

small Bi - control by film
t
 1  (r / R) 3
t0
Fluid-solid reactions

 Product layer model

 Small Thiele modulus, φ=-νAkR/DeA and

large Bi - control by chemical reaction

t
 1  (r / R)
t0
Fluid-solid reactions

 Shrinking particle model

 Small Bi - control by film diffusion

t
 1  (r / R) 2
t0

 Large Bi - control by chemical reaction

t
 1  (r / R)
t0
Packed bed
Packed bed – operation principle
 Gas or liquid flows through a stagnant bed of
particles, e.g. combustion processes or ion
exchangers
 Plug flow often a sufficient description for the flow
pattern
 Radial and axial dispersion effects neglected
Simulation of a packed bed
Mechanistic modelling of kinetics and
mass transfer
for a solid-liquid system:
Leaching of zinc with ferric iron

Tapio Salmi, Henrik Grénman, Heidi Bernas,

Johan Wärnå, Dmitry Yu. Murzin

Laboratory of Industrial Chemistry and Reaction Engineering,

Process Chemistry Centre,
Åbo Akademi, FI-20500 Turku/Åbo, Finland
Reaction system
ZnS(s) + Fe2(SO4)3 ↔ S(s) + 2FeSO4 + ZnSO4

SEM
Experimental system

 Isothermal batch reactor

 Turbine impeller

 Ultrasound input

 SIA – analysis of Fe3+

Experimental data of Bernas (Markus) & Grénman

Markus et al, Hydrometallurgy 73 (2004) 269-282,

Grénman et al, Chemical Engineering and Processing 46 (2007) 862-869
Multi-transducer ultradound reactor

6 transducers

Generator (0-600W)
20 kHz
Reactor pot inserted
A time-variable
power input
Experimental results - Stirring speed

T = 85°C , Sphalerite : Fe3+ = 1.1:1

0.2
200 rpm
350 rpm
0.15 500 rpm
(mol/L)

700 rpm

0.1
3+
Fe

0.05

0
0 25 50 75 100 125 150
Time (min)

The effect of the stirring speed on the leaching kinetics.

Experimental results
T = 85°C, C0Fe(III) = 0.2 mol/L

0.2
Stoic. 0.5:1
Stoic. 0.9:1
0.15 Stoic. 1.1:1
(mol/L)

Stoic. 1.6:1
Stoic. 2.1:1
0.1
3+
Fe

0.05

0
0 50 100 150 200
Time (min)

The effect of the zinc sulphide concentration on the leaching kinetics.

Experimental results

T = 95°C, Sphalerite : Fe3+ = 1.1:1

0.3
(mol/L)

0.2
3+

0.1
Fe

0
0 25 50 75 100 125 150
Time (min)

The effect of the ferric ion concentration on the leaching kinetics.

Experimental results

T = 95°C, Sphalerite : Fe3+ = 1.1:1

0.2
0.2 mol/L
0.4 mol/L
0.15 0.6 mol/L
0.8 mol/L
(mol/L)

1.0 mol/L
0.1
3+
Fe

0.05

0
0 50 100 150 200
Time (min)

The effect of sulphuric acid on the leaching kinetics.

Experimental results - Temperature effect

Sphalerite : Fe3+ = 1.1:1

0.2
75°C
0.15 85°C
95°C
(mol/L)

0.1
3+
Fe

0.05

0
0 50 100 150 200
Time (min)

The effect of temperature on the leaching kinetics.

Experimental results - Ultrasound effect

T = 85°C Stirring rate 350 rpm

0.2
0W
60 W
0.15
120 W
(mol/L)

180 W
0.1
3+
Fe

0.05

0
0 25 50 75 100 125 150
Time (min)

The effect of ultrasound on the leaching kinetics.

Reaction mechanism and rate equations

 Surface reaction

 Stepwise process
( first reacts one Fe3+, then the second one! )

 Rough particles
Three-step surface reaction mechanism

ZnS(s) + Fe3+ ↔ I1 (I)

I1+ Fe3+ ↔ I2 (II)
I2 ↔ S(s) + 2 Fe2+ + Zn2+ (III)
ZnS(s) + 2Fe3+ ↔ S(s) + 2 Fe2+ + Zn2+ a1  k 1c FeIII
a 1  k 1
rates of steps (I-III) r1  a1  a1c I 1
a 2  k  2 c FeIII
r2  a2 c I 1  a2 c I 2 a2  k 2
a3  k 3
r3  a3c I 2  a 3
a 3  k 3 c FeIII
2
c ZnII

cI1, cI2 and cI3 = surface concentrations of the intermediates.

Development of rate equations

Pseudo-steady state hypothesis Rate equation

a1a2 a3  a1a 2 a 3
r1  r2  r3  r r
a1a 2  a1a3  a2 a3

Back-substitution of a1….a-3 gives

k 1k  2 k  3cFeIII 2  k 1k  2 k  3cFeII 2 cZnII D = k-1k-2+k-1k+3+k+2k+3cFeIII

r
D

k (c FeIII 2  c FeII 2 cZnII / K )

r
D
Rate equations

Final form

k1c FeIII 2
r where β = (k-1k-2+k-1k+3)/(k+2k+3)
  c FeIII

An alternative rate equation

k1c H c FeIII 2
r NOT VALID FOR THIS CASE!
c H  c FeIII
Area & Shape factor

Development of a general approach

The surface area (A) can be expressed with a generalized equation

aM n = amount of solid
A n0 x n1 x n0= initial amount of solid
 P R0
Shape factor (a=1/x)

A m0 P
A  Mn0 n11 / a
1/ a
a  P R0 a / R0    P
VP V0 P
Area & Shape factor

Geometry Shape factor (a) x 1-x

Slab 1 1 0
Cylinder 2 1/2 1/2
Sphere 3 1/3 2/3
Irregular,
high value 0 1
‘rough’ particle

Reaction order can vary between 0 and 1!

Mass balance for batch reactor

dcZnS  ZnS M dcFeIII  FeIII M

 cZnS r  cZnS r
dt x0 ZnS dt x0 ZnS

e  Ea / RT c ZnS c FeIII
2

r'  , where γ=(k1σM / x0ZnS)

  c FeIII

dci
  ir'
dt
Parameter estimation

Nonlinear regression applied on intrinsic kinetic data

Estimated Parameter Parameter value Est. Std. Error %

γ (L / mol min) 0.331 4.5
Ea (J / mol) 53200 4.8
β (mol / L) 0.2 24.9
Intrinsic kinetics - Model fit
T = 85°C

0.2 0.5:1
0.9:1
1.1:1
0.15
(mol/L)

1.6:1

0.1
3+
Fe

0.05

0
0 25 50 75 100
Time (min)

The effect of the ratio sphalerite : FeIII on the kinetics

Intrinsic kinetics - Model fit

0.2
75°C
85°C
0.15
95°C
(mol/L)

0.1
3+
Fe

0.05

0
0 25 50 75 100 125 150
Time (min)

Temperature effect on the kinetics.

Mass transfer limitations in Batch reactor

N Li s A   i rA  0

where ri=νir The mass transfer term (NLis) is described by Fick’s law

  i k1ci *2
k Li (ci  ci *)  β’=β/ci, γ’=(-νik1ci/kLi), y=ci*/ci
  ci *

The solution becomes

ci * / ci 
2 ' k1c FeIII *2
r
(  '1) 2  4( '1)  '   '1   c FeIII *
Liquid-solid mass transfer coefficient

General correlation

Sh  a  b Re1 / 2 Sc1 / 3  d 4 
1/ 3

Re   3  Sc   / Di
 
Sh  k Li d / Di

  4 
1/ 6 1/ 3 
 1/ 6 1/ 3 
Di  d       4 4/3 
 
k Li  

a  b 3   
 k Li 
Di 
a  b d 0 z     
d 
     Di   d 0 z1 / 3   3

  Di
  


 

where z=cZnS/c0ZnS. The index (i) refers to Fe(III) and Fe(II)

 Correlations in rate equation

k Li 
Di
d 0 z1 / 3
2  b ' z 2/9
 b’=b(ε d04/ υ 3)1/6(υ/Di)1/3

IF b’z2/9 >> 2 under stirring,

Di b' b' '
k Li  1/ 9
 1/ 9 γ’=-νFeIII x0ZnS γ cFeIIIz1/9/(σMb’’FeIII)
d0 z z

 '
 FeIII c FeIII    x0 ZnS d 01/ 3 /(Mb( /  )1/ 6 DFeIII
2/3
)
(c0 ZnS / cZnS )1 / 9

The surface concentration: The rate:

2 c FeIII cZnS cFeIII *2

c FeIII *  r' 
(   cFeIII ) 2  4( '1) cFeIII    cFeIII   cFeIII *
Determination of mass transfer parameter (ω)

Agitation rate & US effect ω (mol min / m3)

200 rpm 50
350 rpm 14
500 rpm 2.4
350 rpm US 0 W 14
350 rpm US 60 W 6.8
350 rpm US 120 W 5.8
350 rpm US 180 W 1.67
 Modelling of kinetics and mass transfer

0.2
200 rpm
350 rpm
0.15 500 rpm
(mol/L)

0.1
3+
Fe

0.05

0
0 25 50 75 100 125 150
Time (min)

External mass transfer limitations – modelling of individual mass transfer

parameters at different agitation rates.
Modelling of kinetics and mass transfer

0.2
0W
60 W
0.15
120 W
Fe (mol/L)

180 W
0.1
3+

0.05

0
0 25 50 75 100 125 150

Time (min)

External mass transfer limitations – modelling of individual mass transfer

parameters at different ultrasound inputs.
 60

50
Mass transfer
40
parameter
30
ω

20
Normal agitation
10

0
150 200 250 300 350 400 450 500 550
rpm (1/min)

16

14

12

10

8
Ultrasound
ω

0
0 50 100 150 200
US (W)
The real impact of mass transfer limitations

120
bulk )*100

100

80
3+
/ Fe

60
surface

40 500 rpm
350 rpm
3+

20 200 rpm
(Fe

0
0 25 50 75 100 125 150
Time (min)

The difference in the model based surface concentrations and measured

bulk concentrations of Fe3+ at different stirring rates.
 The real impact of mass transfer limitations

120
bulk )*100

100

80
3+
/ Fe

60 180 W
60 W
surface

40
0W
3+

20
(Fe

0
0 25 50 75 100 125 150
Time (min)

The difference in the model based surface concentrations and measured bulk
concentrations of Fe3+ at different ultrasound inputs.
Conclusions

 A new kinetic model was proposed

 A general treatment of smooth, rough and porous

surfaces was developed

 The theory of mass transfer was implemented in the

model

 Model parameters were estimated

 The model works

 Modelling and simulation of
porous, reactive particles in
liquids: delignification of wood

Tapio Salmi, Johan Wärnå, J.-P.

Mikkola, Mats Rönnholm
Åbo Akademi Process Chemistry
Centre,
Laboratory of Industrial Chemistry
FIN-20500 Turku / Åbo Finland,
[Link]@[Link]
Typical view of Finland

338000 km2 of
which 70% forest
Papermaking
 Wood chips
 This is where paper
making begins.
 A typical wood chip
measures 40 x 25 x
10 mm.
Wood
 Each chip comprises
water, cellulose wood
fibres and the binding
agent lignin.
.
Pulp
 To make paper, we need to first make pulp,
which is the process of breaking the wood
structure down into individual fibers

Chips Digester
Reactions
The reactions in chemical pulping are
numerous. Typical pulping chemicals Part of Lignin
are NaOH and NaHS molecule

cellulose

Overall process:

Lignin+Cellulose+Carbohydrates+Xylanes+OH+HS ->
Dissolved components
Kinetic modelling of wood delignification

 Purdue model (Smith [Link]. (1974) Christensen

et al. 1983), 5 pseudocomponents
 Gustafson et al. 1983, 2 wood components
Lignin and Carbohydrate, 3 stages
 Andersson 2003, 15 pseudocomponents

 Very few models available!

Wood chip structure
 Wood material is built
up of fibres
 We can expect
different diffusion
rates in the fibre
direction and in the
opposite direction to
the fibres.
Existing models

 The existing models for delignification of

wood consider a 1 dimensional case with
equal diffusion rates in all directions
 Is a 2- or 3-dimensional model needed ?
Characteristics of our model

 Time dependent dynamic model

 Complex reaction network included
 Mass transfer via diffusion in different
directions
 Structural changes of the wood chip
included
 All wood chips of equal size
 Perfectly mixed batch reactor assumed
Mathematical model,volume element

 3D –model for a wood

chip

y  N A  +  N A  +  N A  + r ΔV = '

z
ix yz in iy xz in iz xy in i

N A  +N A  +N A  +
ix yz out iy
dn
xz out iz xy out
i

dt

Dy
Dx Dz
x
Mass balance for a wood chip

dci Di  ' d 2 ci 2
' d ci
2
' d ci  ri'
  ε x + εy + εz  
dt  p  t   dx 2
dy 2
dz 2   p t

Porosity
x
 p  t   3  x'  y'  z'  
'
x
 px
Boundary conditions

The concentrations outside the wood chip are locally known

ci=cLi

at the centre of the chip (symmetry)

dci/dx=dci/dy=dci/dz=0
Reactor model

Batch reactor model, ideal flow

dc Li
= N ix a yx + N iy a xz + N iz a xy + ri
dt

Fluxes from wood chip

 dc   dc   dc 
N ix =  ε 'x Di  i  N ix = ε'y Di  i  N iz = ε'z Di  i 
 dx  x= Lx  dy  y= Ly  dz  z= Lz
 Structural changes of the wood
chip
Generally one can state that the porosity of the chip
increases during the process, since lignin and
hemicelluloses are dissolved

Change of porosity as a function

of the lignin conversion
c0l  cl c
ηl = =1 l
c 0l c0l
'
p 0p 0p 1 1 l
Kinetic models

Andersson model, 12 wood pseudocomponents


ri  k1 cOH 
c HS  k 2 Wi 
Purdue model (Christensen et al), 5 wood pseudocomponents
 dC
dt
 
 k i'   OH    k i"  (OH  ) a  ( HS  ) b  (C  C 0 )

Gustafsson model, 2 wood components, 3 stages dC 0.11 dL

dL
4807.69
 2.53   OH  
Initial stage, >22% Lignin,  36.2  T  e L 0.5 T
dt dt
dt
Bulk stage , 22% > Lignin > 2% dL 
17200 14400
0.4 
  OH   e   OH  S   L
35.19  29.23 0.5
 e T T
Residual stage < 2% Lignin dt  
10804
dL
  OH 
19.64  0.7
e T
L
dt
Diffusion models

2
9m
Del( 373.15)  2.13  10
s

McKibbins
4870
8
8 3 10
D NaOH , H 2O  5.667  10  T  e 1.9872T

Diffusivity [m^2/s]
Wilke-Chang 2 10
8

7.4  10 12    M B  T
D AB  0.6
 B VA 8
1 10

Nernst-Haskel (infinite dillution)

 0  0   z  z  0
D  8.931  10
o 14
 T   0     380 400 420 440 460
      z  z
0
 Temperature [K]
McKibbins
Wilke-Chang
Nernst-Haskell
Kappa number

The progress of delignification is by pulp

professionals described by the Kappa number

L
  500 5
L  CH

L = Lignin on wood, CH = Carbohydrates on wood

Numerical approach
 Discretizing the partial differential equations (PDEs) with
respect to the spatial coordinates (x, y, z).

 Central finite difference formulae were used to

approximate the spatial derivatives
 Thus the PDEs were transformed to ordinary differential
equations (ODEs) with respect to the reaction time with
the use of the powerful finite difference method.

 The created ODEs were solved with the backward

difference method with the software LSODES
Simulation results,
profiles inside wood chip

L1
Kappa

Kappa value
86
centre 19
centre
84
18.5
82

80 18

78
17.5
76
15

10
15 17
10 15
5
5 15
10
y 0 0 10
x Porosity y
5
5

y 0 0
x

0.55

0.54
Lignin content
0.53

0.52

0.51
T=170 ºC
centre
0.5
0
C0,NaOH=0.5 mol/l
5 15

10
5
10
Porosity
15 0
x y
The impact of 2-D model

19
19
18.8

18.6 18.8

18.4
18.6
y
18.2
Lignin (w%)

Lignin (w%)
18
18.4 x
17.8

17.6 18.2

17.4

18
17.2

17
0 2 4 6 8 10 12 17.8
surface x-node centre 0 2 4 6 8 10 12
surface y-node
centre

Red line, different diffusion rates in x and y directions

Blue line, same diffusion rates in x and y direction
(Andersson kinetic model)
Content of lignin on wood as a function of
reaction time

Lignin concentration (w-%) in wood chip as a function of

reaction time (min) with Andersson kinetic model (left) and
Purdue kinetic model (right).
Simulation software
 2-D model for a wood chip in a batch reactor
 Different kinetic and diffusion models available
 Structural change model included (porosity)
 Dynamic model
 all results can be presented as a function of reaction
time
 Temperature and alkali concentrationprofiles can
be programmed as a function of reaction time
Conclusions
 A general dynamic model and software for the
description of wood delignification
 Solved numerically for example cases, which concerned
delignification of wood chips in perfectly backmixed
batch reactors.
 Structural changes and anisotropies of wood chips are
included in the model.
 The software utilizes standard stiff ODE solvers
combined with a discretization algorithm for parabolic
partial differential equations.
 Example simulations indicated that the selected
approach is fruitful, and the software can be extended to
continuous delignification processes with more
complicated flow patterns.