Modeling Ion
Modeling Ion
Modeling Ion
Erkki Räsänen
VTT Processes
3
4
Räsänen, Erkki. Modelling ion exchange and flow in pulp suspensions. Espoo 2003. VTT
Publications 495. 62 p. + app. 111 p.
Keywords ion exchange, paper and pulp, pulp suspensions, cellulosic fibers, aqueous
solutions, models, modelling, transition metals, bleaching, chelants, metal
complexes, theses, dissertations
Abstract
This thesis summarizes the results of four papers dealing with modelling of ion
exchange between cellulosic fibers and their surrounding aqueous solution.
An ion exchange model to describe interaction of ionic species with kraft pulps
has been developed in the present work. The model can be applied in pulp
washing and chelation. The present model takes into account overall ionic
composition of pulp suspensions, i.e. the cations and simple anions as well as the
ligands and the complexes. The model facilitates calculation of pH,
complexation and distribution of ionic species between the aqueous phase
confined to fibers and the external solution. The present model also accounts for
the ion exchange kinetics and flow of water through the immobile fiber network,
i.e. the ”pulp bed”.
The present model is based on the formerly developed ion exchange model,
which is based on so-called Donnan equilibrium theory. The development of the
new model required the inclusion of the different anionic species to model and
the combination of the Donnan model with a complexation model. The ion
exchange kinetics is described as a two-step phenomenon. The first rate-limiting
step can be due to the dissolution of solid metal compounds in the fiber phase as
5
well as the diffusion of ions in the fiber wall, while the second step is the
transfer of the ionic species between the fiber phase and the external solution,
which is governed by the concentration gradient between the phases. In the
model, the transfer of ions between the phases is directly related to the Donnan
theory.
In the testing of the elements of the ion exchange model, it was found that the
combination of the Donnan model with a complexation model quantitatively
predicts the pH, the ionic distribution as well as the complexation equilibria in
batch-type systems when the time to achieve full equilibrium is not limited. In
the pulp beds subject to flow the ion exchange kinetics reduces the metal
removal efficiency and selectivity. This can be considered to be due to the
repulsion of anionic species (simple anions, ligands and complexes) from the
negatively charged fiber phase. It can also be considered that the rate of change
of fiber charge determines the overall ion exchange kinetics.
6
Preface
I am grateful to my supervisor, Professor Emeritus Per Stenius for his support,
guidance and patience during this study. I am also grateful to Professor Tapani
Vuorinen and to Professor Emeritus Johan Gullichsen, at the Helsinki University
of Technology, to Professor Adriaan van Heiningen, at the University of Maine,
and to Dr. Pertti Koukkari, at the Technical Research Centre of Finland, for
sharing their enthusiastic attitude towards scientific work. I wish to express my
gratitude to all of my co-authors who have contributed this work. I am especially
grateful to Mr. Pekka Tervola for his help in mathematics.
Finally, I wish to thank my close friend Timo Pääkkönen and my dear Nina.
7
List of original publications
This thesis is mainly based on the papers listed below:
8
Contents
Abstract................................................................................................................. 5
Preface .................................................................................................................. 7
Abbreviations...................................................................................................... 11
Nomenclature...................................................................................................... 12
1 Introduction...................................................................................................... 15
1.1 Towards better understanding of the chemistry of pulp suspensions ..... 15
1.2 Hypothesis .............................................................................................. 16
1.3 The objectives of the work...................................................................... 17
2 Theory.............................................................................................................. 19
2.1 Cellulose fibers in ionic media ............................................................... 19
2.2 The Donnan equilibrium model.............................................................. 21
2.3 Extensions of equilibrium model to include kinetics.............................. 27
2.4 Mobile phase model................................................................................ 32
2.5 Parallel Plug Flow model........................................................................ 34
9
5 Conclusions...................................................................................................... 53
References........................................................................................................... 57
Appendices
Papers I–IV
10
Abbreviations
BCA6 Bis-N-[2-(1,2-dicarboxylethoxy)]-aspartic acid
BTC Breakthrough Curve
COD Chemical Oxygen Demand (required for oxidation of organic
carbon)
CTMP Chemi-Thermo-Mechanical Pulp
CSTR Continuously Stirred Tank Reactor
D Chlorine Dioxide Bleaching Stage
DTPA diethylenetriaminepentaacetic acid
EDTA ethylenediaminetetraacetic acid
EOP Alkaline Extraction Stage with Oxygen and Peroxide
FSP Fiber Saturation Point
PPF Parallel Plug Flow
TOC Total Organic Carbon
WRV Water Retention Value
11
Nomenclature
Latin letters
12
M' uncomplexated metal
n molar amount, mol
13
Greek letters and symbols
14
1 Introduction
Chemically delignified and fully bleached kraft pulp is the most important raw
material of paper and many other fiber products. Due to the quality requirements
of pulp, environmental reasons, and the cost of the chemicals, energy and
process equipment, it is essential to understand and manage the chemistry of
pulping and the fiberline processes as well as the chemistry of papermaking
processes. The quality of kraft pulp determines the value and the technical
properties of products and the reusability of fiber material.
The chemistry of pulping and bleaching process is highly complex. Due to the
heterogenous chemical composition of wood raw material, its reaction products,
the chemicals as well as the so-called non-process elements, the many different
phenomena related to interactions of these components have been difficult to
understand and characterize. Fundamental characterization of the chemical
structure of wood material and its reactions with different chemicals has
significantly increased knowledge of the chemistry of pulping processes during
the last two decades. Another focus in the research of pulping chemistry has
been the characterization of composition of process effluents and filtrates as well
as their effects and equilibrium behavior in pulping processes. This study deals
with the latter aspect.
Due to these advances, the chemistry of pulping processes can be managed quite
well nowadays. However, much of this practice is strictly on an empirical basis.
Understanding of fundamental phenomena is the premise for continuous
advance; it has been demonstrated quite recently that fundamental
characterization of basic chemical interactions and process chemistry can result
in significant breakthroughs in the development of pulping process technology
/1/.
15
was begun in the first half of the 1990’s by Sjöström /2/, Laine et al. /3/ and by
Vuorinen et al. /1/.
The research was carried out in close collaboration with research groups in
Sweden, The Department of Inorganic Chemistry, University of Umeå, The
Institute for Surface Chemistry, Stockhom and STFI, and with other research
communities sharing the same interests.
1.2 Hypothesis
The ion exchange kinetics and ionic equilibria in pulp – water systems can be
described quantitatively when the overall ionic composition accounting for
anions, cations, ionic species and fiber charge properties are known.
16
suspensions, some of these mechanisms can be defined on a phenomenological
basis only.
The objectives of this work were to develop equilibrium and kinetic models for
the characterization of ion interactions with pulp fibers in water suspensions
subject to varying amounts of ions and flow of water. The models developed in
this work are extensions of formerly developed equilibrium model /4/, which
was based on the use of the so-called Donnan equilibrium theory.
1. The first model development is a method to account for overall anionic and
cationic composition of pulp suspensions, which is needed in many applied-
oriented purposes where the calculation of pH is required when amounts of
ionic components in the system vary. Such a system is, for example, a
reactor with inputs and outputs of fibers, water and chemicals.
These models were developed and evaluated in Papers I–IV. Papers I and II
describe the formulation and evaluation of the first and the second extensions,
17
i.e. the inclusion of anionic species and a complexation model into the Donnan
equilibrium model. Paper III is focused on the experimental characterization of
the behavior of metals in pulp beds subject to displacement flow. The kinetic
models as well as the displacement model were formulated and tested in Paper
IV.
18
2 Theory
2.1 Cellulose fibers in ionic media
19
Interactions of ions with the functional groups bound to fibers can be divided
into non-specific and specific interactions. Non-specific interactions are
electrostatic attraction or repulsion between ions and the charged matrix of the
fiber cell wall. Specific interactions include e.g. complexation of metal cations
with functional groups as well as adsorption of ions to polarized surfaces. The
interactions of ions with the cellulose fibers depend largely on the charge
properties of fibers. As shown in Table 1, the types and amounts of ionizable
functional groups in cellulose fibers depend largely on the grade of pulp.
20
Native wood contains carboxylic groups in uronic, fatty and rosin acids ( -log K
= 3.5 – 5), phenolic hydroxyl groups in lignin ( -log K = 6 – 11) and hydroxyl
groups in carbohydrates (-log K = 11 – 12.5) /5, 6/. The chemical composition
of functional groups in mechanical, thermomechanical and chemimechanical
pulps is quite similar to native wood and thus, the charge properties of
mechanical pulps depend largely on the wood raw material. Peroxide bleaching
of these pulps generates new carboxyl groups and increases the total anionic
charge. Dithionite used in the bleaching of chemimechanical pulps generates
sulphonic acid groups that are strongly acidic. The wood raw material
determines also the charge characterisitics of chemically delignified pulps.
Chemical delignification of wood in the alkaline sulphate (kraft pulping) process
decreases the total charge of wood due to chemical reactions and dissolution of
wood extractives, lignin and hemicelluloses. The kraft pulping process also
creates carboxylic groups, i.e. hexenuronic acids that do not exist in native
wood. These groups are generated from methylglucuronic acids in xylan. The
xylan dissolved in kraft pulping conditions can reprecipitate on the surface of
pulp fibers, which causes non-homogenous distribution of the charge in the fiber
cell wall /7/.
The original Donnan theory was first used in the description of the so-called sol-
concentration effect of electrochemical cells /15/. Later, the theory was applied
in the characterization of colloids, i.e. the molecular weight and charge of
colloidal particles /16/. Donnan theory was extended to polyelectrolyte systems
by Procter and Wilson /17/. The theory was used to characterize electrolyte
interactions with cellulose fibers by Neale /18/ in 1929 and in 1952 by Farrar
and Neale /19/.
21
The original Donnan theory assumes ideal behavior in both aqueous phases. To
include non-ideal behavior, the theory was later extended e.g. by Donnan /20/,
Hückel /21/ and Kameyama /22/.
When applied to fiber chemistry, the Donnan theory describes the distribution of
ions between the aqueous phase in fibers and the external solution according to:
[X i ] f = λzi [X i ] s (1)
Towers’ and Scallan’s ion exchange model /3/ cleverly combined the conditions
of Donnan distribution, the mass balances of ions between the aqueous solution
in fibers and the external solution surrounding the fibers, as well as the
electroneutrality conditions into a single equation. Their model is expressed by:
∑ [M ] zi
⋅V
(λ )[ ] ( ) [ ]
i T
2
−1 ⋅ H +
+ zi ⋅ λ zi +1
−1 ⋅ i
− ∑ Af-, i =0 (2)
s
(
V +F λ − 1zi
) i
T
The inputs of the model of Towers and Scallan are the total amounts and
valences of metals initially present in fibers, the total volume of water, the
volume of water in the fiber cell wall, the mass of pulp fibers, the pH of the fiber
suspension and the total charge of fibers i.e. the total amounts of fiber-bound
anionic groups and their dissociation constants. From these inputs one can obtain
the value of Donnan distribution coefficient as a numerical solution of Eq. (2).
Using this coefficient, the concentrations of ions in the external phase can be
calculated from:
22
nXi
[X i ] s = (3)
(V + F (λzi − 1))
Although based on the ideal behavior of aqueous phases, Towers and Scallan
demonstrated that the model is able to describe ion exchange of Na+, Mg++, Ca++
and Mn++ ions between pulp fibers and the external aqueous solution realistically
at least under laboratory conditions /4/. Later, their model was used in many
studies describing the interactions of fibers with ions in solution and other
related phenomena.
Laivins and Scallan /13/ studied the effect of ionic strength on the removal of
metal ions from kraft and CTMP pulps. They demonstrated that the model is
able to predict the ion exchange in a wide ionic strength range and that the
model can be used to describe charge and ion exchange properties of CTMP
pulps with a high total charge.
Laivins and Scallan /24/ also studied the effect of the ionic form of the pulp
fibers on the swelling of fibers and on the strength properties of handsheets
prepared from the pulps. They qualitatively related these properties to the
Donnan theory.
Lindgren /11, 25/ studied the applicability of the Donnan model in the
description of the association of Al+++, Fe+++ and Cu++ species with pulp fibers.
Lindgren demonstrated that the species of these metals also interact with fibers
according to the Donnan theory, but specific interactions of these species with
fibers and complexation with OH- ions must be taken into account.
Athley /27/ studied the interactions of Na, Ca and Mg ions with pulp fibers,
comparing the Donnan ion exchange model with a surface complexation model.
23
A quite similar comparison was made by Lindgren /11/. Athley demonstrated
that the experimentally measured ionic distributions deviate to quite a significant
degree from the theoretically predicted ones. The deviations may be caused by
the assumption of ideal behavior of aqueous solution in the Donnan model.
Andersson et al. /28/ studied the washing of ionic components from the pulp. In
their study the Donnan theory was qualitatively related to removal of anionic
macromolecular components from pulp. It was demonstrated that the anionic
macromolecules are easily removable from the pulp fibers due to their repulsive
interaction with the negatively charged fiber wall.
All of the studies described above were made using the original model of
Towers and Scallan. However, when considering practical or industrial
applications, the Towers’ and Scallan’s model has several limitations. The most
fundamental limitations are that the input of the ionic components is restricted to
metal cations and the complexation equilibria in solution are not included in the
model.
The pH is the main factor that governs the ion exchange and the complexation
equilibria in fiber suspension. The electroneutrality conditions define the
dependence of pH on the ionic composition of the suspension. Thus, if the
overall composition of cations, simple anions and chelants in the suspension are
to be accounted for in the modelling, the pH of suspension cannot be regarded as
a ‘known’ input parameter.
In the present work, the anionic components were included in the Donnan model
(Papers I and II) by using electroneutrality equations that account for the
presence of different anionic species, e.g. simple anions and multivalent acids or
ligands. These conditions, i.e. Eqs (4) and (5) are written for the whole
suspension and for the external solution:
[ ] − ∑ [A ]
z i ∑ X iz i T
−
f, i T
=0 (4)
i i
zi ∑ X iz i
i
[ ] s
=0 (5)
24
This definition differs from Towers’ and Scallan’s model that used
electroneutrality conditions for the external solution and fiber phase, which were
combined in Eq. (2). Two electroneutrality conditions and inputs for different
anionic species allow the calculation of pH of the suspension and the Donnan
distribution coefficient λ as a numerical solution of Eqs (4) and (5). The
concentrations of ionic species in Eqs (2) and (4) are given by:
KA A f [ ]
[A ]
−
f T = f
K Af + λ ⋅ 10
T
− pH S
(6)
10 − pHS ⋅ (V + F (λ − 1))
[H ]
+
T =
V
(7)
nX i
[X i ] T = (9)
V
[H ]
+
s = 10 − pHS (10)
[OH ] −
s = 10 −14+ pHS (11)
nXi
[X i ] s = (12)
(V + F (λzi − 1))
In particular, for k-protic acids and ligands in Eqs (9) and (12) the molar amount
of acid that is present as an anion formed by dissociation of p protons ( z i = p )
can be calculated from:
25
n A− p =
i
n Ai ⋅ K A1K A 2 ...K Ap H + ([ ] ) s
k −p
(13)
( [H ] )
+
s
k
([ ] )
+ K A1 H + s
k −1
+ K A1K A 2 ( [H ] )
+
s
k −2
+ ... + K A1K A 2 ..K Ak
As can be seen from the electroneutrality equations the immobile fiber charge is
accounted for in the overall electroneutrality condition only, i.e. in Eq. (4). The
freely mobile anionic species are accounted for in both electroneutrality
conditions, Eqs (4) and (5). The charge of fiber with varying ionic
concentrations is calculated from the determined total charge of pulp and the
dissociation constants of fiber-bound groups. The dissociation constants for
ligands and many other known small-molecular-weight anionic species are
available from literature /e.g. 29/.
When compared to the model of Towers and Scallan /4/ the Donnan model, as
defined by eqs (1)–(13), is not fundamentally different but the formalism is more
general. In earlier Donnan models, the electroneutrality conditions of the two-
phase Donnan system are written separately for both phases. The way the
electroneutrality condition is used in this work is consistent with the general
two-phase formalism; if the whole system and the external phases are
electroneutral, the fiber phase must also be electroneutral.
K MY
M zi + Y − p ⇔ MY ( zi − p ) (14)
K MY =
[MY ] (15)
[M '][Y ']
26
and, using the following mass balances of metals and complexes:
YT =
[Y ] + [HY ] + ... + [ H Y ] + ∑ [M Y
−p −( p −1)
p i
( zi − p )
]= (16)
i
[Y ] + ∑ [M Y ]
'
i
( zi − p )
[
M T = [M '] + MY ( z − p ) ] (17)
MT
[M '] = (18)
1 + K MY [Y ']
YT
[Y '] = (19)
1 + ∑ K MiY [M ' i ]
i
In Eqs (15)–(19) [M '] and [Y '] denotes the forms of metals and chelant
available for complexation, which depend on pH, see Paper II. Using Equations
(15) and (19) the pH-dependent complexation of metals can be calculated. With
a simple algorithm, Equations (16)–(19) can be combined with the
electroneutrality Equations (4) and (5), which facilitate the calculation of the pH,
the complexation equilibria and the Donnan distribution of all ionic species in
the pulp suspension. An experimental validation of the Donnan model combined
with the complexation equilibria is presented in Paper II.
27
! The reaction kinetics in general depends on the thermodynamic gradient,
i.e. the extent of chemical reaction, required to achieve complete
equilibrium of a non-equilibrium system /30/.
The first factor generally describes all kinds of chemical systems; when a
transient change in the system occurs, the system composition changes
according to defined chemical laws until the system is in equilibrium. However,
the overall kinetics usually depends on individual rate-determining steps or
reactions, which may be difficult to define separately due to complicated
reaction pathways or the occurrence of many competing reactions with similar
kinetics. When applied to fiber suspensions, the initial and final equilibrium
compositions are known or can be calculated by using the Donnan theory.
However, internal factors, i.e. diffusion as well as the specific reactions
occurring in aqueous phases determine the overall ion exchange rate.
Molecular diffusion has an important role in the mass transfer kinetics between
cellulose fibers and process water. Mass transfer between two phases have been
studied and modelled e.g. by Sherwood and Wei /33/ as well as by Ala-Kaila
/34/ who applied mass transfer models in the description of ion exchange in
cellulose fiber suspensions. Both studies applied mass transfer film theory,
which accounts for diffusitivity of ions through films with defined thickness and
diffusion resistance as a mechanism that limits the mass transfer kinetics. By
analogy to physical fiber structure, the films are confined to the aqueous
immobile solution in the fiber phase /34/. Ala-Kaila defined three mass transfer
films with different diffusitivity resistances. The number of mass transfer films,
28
however, was used as a curve-fitting parameter; Ala-Kaila did not relate the
number of films to any topochemical characteristics of fibers, except that the
total volume of liquid in the mass transfer films was equal to the volume of
solution confined in the fiber phase.
When considering molecular and ionic diffusion, the structure of the fiber cell
wall has characteristics that can be accounted for as factors limiting the mass
transfer rate. In water suspensions the fibers swell so that the whole internal
surface area of the fibers is associated with water. Highly swollen fibers have
structure freely accessible for simple ions and even for macromolecules. The
swelling, however, depends on the history of pulp; never-dried pulps usually
have a higher surface area than the dried and re-wetted pulps, since drying
causes irreversible morphological changes in the fiber wall that reduce the
specific internal surface area, modify the pore structure of fibers and reduce the
accessibility and transfer of solute components between the phases. Thus,
phenomenological relationships between the ionic diffusitivity and the
topochemical characteristics of the fibers could be defined. However, other
mechanisms that do not directly depend on the structure of the fibers can be
more significant as factors limiting the overall ion exchange kinetics.
Due to the highly complex compositions of pulp suspensions, the third factor,
i.e. the dissolution kinetics and the other specific kinetic reactions occurring in
the solution phase are difficult to account for in the modelling. Sjölander
(Norberg) et al. /26/ experimentally demonstrated that many metals are present
in oxygen-delignified pulp as precipitated metal hydroxides and carbonates.
Various other solid forms of metals can also exist in pulps, e.g. as oxalates and
sulphates. The main factor determining the solubility of different solid forms of
metals is the pH. The effect of other ions in aqueous multicomponent systems on
the solubility of solid metal compounds can be described theoretically when all
the concentrations, solubility products and other thermodynamic properties of
the components are known.
29
solubility limit depend on the particle size of solids, temperature, concentrations
of dissolved organics (as measured by COD and TOC) as well as on the ionic
strength.
Thus, due to the highly complex connection between the ion exchange kinetics
and the ionic composition of pulp suspension, the internal factors affecting ion
exchange (i.e. the diffusitivity as well as solubility and other specific reaction
mechanisms) in this work were replaced by simple ‘internal’ kinetics, which is
phenomenological without theoretical relations to specific kinetically limited
reactions and fiber properties. However, the overall ion exchange kinetics
describing the transfer of ions from the fiber phase to the external phase is
directly related to Donnan theory. Thus, when compared to other studies, e.g.
/34/, the relations to specific mechanisms determining the ion exchange kinetics
remain empirical, but the Donnan theory facilitates easier calculation of the ion
exchange kinetics of complex multicomponent systems.
(A) The total concentration of species X i in the fiber is the sum of a fraction
that can not transfer between the phases due to the kinetic restrictions and
another fraction that can transfer between the fiber and the external phase:
30
[X i ] f = [X i∗ ] f + [X i ,f ↔s ] f (20)
X i∗,f →
ki
X i ,f →s (21)
(C) The rate of transfer of species X i between fiber and solution is described by
the rate constant k λ :
X i ,f ↔s ←→
kλ
X i ,s (22)
The rate equations describing the internal and overall ion exchange kinetics are:
d n X i∗,f k i n X i∗,f
−
dt Vf
=
Vf
[ ]
= k i X i∗ f
(23)
d [X i ] f
−
dt
{ } [ ]
= k λ [X i ] f ↔s − [X i ] f , eq − k i X i∗ f
(24)
Eq. (24) relates the amount of ions in the non-transferable fraction (*) to first-
order kinetics with respect to the concentration of ions i remaining in the
fraction. Conceptually, it is assumed that the non-transferable fraction is present
in a separate phase, which is electroneutral within the fiber phase. Eq. (24) is a
mass balance equation that accounts for the transport of ions from the non-
transferable fraction and the transfer of ions from the fiber phase to the external
solution. This kinetics is first-order with respect to the theoretical difference
between the transient and equilibrium concentrations in the fiber phase as
described in Fig. 2.
31
concentration
[X ] i f
[X ] i f, eq
time
Figure 2. The relationship between transient and theoretical equilibrium
concentrations in the fiber phase. The concentrations can either decrease or
increase in the fiber phase depending on the overall concentration transient in
the system.
The present kinetic ion exchange model is similar to Ala-Kaila’s model /34/ in
the sense that the driving force for the diffusion of freely mobile ionic species is
the concentration gradient between the fiber and the external phases. However,
the main difference between the present model and the model of Ala-Kaila is
that the multiple mass transfer films have been replaced by separate fraction(s)
of non-transferable ions that can be present as separate phases in fibers. Mass
transfer film models invoke diffusion as the only mechanism that limits the mass
transfer kinetics. Multiple phase formalism, however, can easily be modified to
account for e.g. the presence and dissolution kinetics of solid compounds that
can be taken into account as separate, electroneutral phases.
32
pulp fibers, we can thus assume that the fibers contain a solution that is
immobilized in the bed, while the external solution flows through the bed. It is
thus easy to account for input and output of solute components with the external
solution; a mass balance for such a system is shown in Fig. 3. A single control
volume with input and output of water and solute components represents a
continuously stirred tank reactor, CSTR.
[X i
∗
ki
→ X i ,f ↔ s ] f
←
kλ
→ [X i ] s
[X ]i out
[X i
∗
ki
→ X i ,f ↔ s ] f
←
kλ
→ [X i ] s
[X i
∗
ki
→ X i ,f ↔ s ] f
←
kλ
→ [X i ] s l = L
33
the layer above, as shown in Fig. 3. The model consisting of many CSTRs in
series describes flow that approaches ideal plug flow as the number of layers
increases.
d [X i ] s,l
dt
=
F
V −F
{ }
⋅ k λ [X i ] f ↔ s,l − [X i ] f , eq,l + [X i ] in,l − [X i ] out ,l (25)
where the layer number l = L =1 for a system consisting of one control volume.
For a series of control volumes, l = 1…L with l = 1 being the top layer and l = L
the bottom layer.
The mobile phase model, i.e. Eq. (25) does not account for the flow-through
properties of real pulp beds. Results from the model, however, can be used as
input for the so-called parallel plug flow model which allows adjustment of flow
dispersion between ideal plug flow and fully dispersed flow, i.e. the one
approaching the CSTR condition. In the parallel plug flow model the exit
concentration of the pulp bed is calculated as the ‘mixing cup’ average of exit
concentrations of many ideal plugs with different residence times.
Qj J
C (T ) = ∑ C j (T ) (26)
Qd j =1
34
T2
Qj
= ∫ E (T ) dT (27)
Qd T1
Tr j ≅ (T1 + T2 ) 2 (28)
1 (1 − T )2
E (T ) = exp − (29)
2 π / Pe 4 / Pe
The Peclet number, Pe, describes the extent of flow dispersion as:
qH
Pe = (30)
D
where q is the superficial velocity of fluid, H is the bed height and D is the
dispersion coefficient.
35
3 Materials and methods
3.1 Pulp properties
Oxygen delignified hardwood (birch) kraft pulp was chosen for the experiments.
The hardwood kraft pulp fibers have relatively high charge (i.e. high ion
exchange capacity) and high hexenuronic acid content which made it easy to
modify the charge in the experiments of Paper I.
The experiments described in Papers I–IV were carried out using batch
displacement cooked hardwood kraft pulp. The pulp used in the experiments of
Papers I and II was the same, the analyses of fiber properties were repeated
when the experiments of Paper II were carried out. The pulp used in the
experiments of Paper III was obtained from the same mill, production line and
similar cook. As shown in Table 2 ranges of variation of some fiber properties
were determined by replicate experiments from the pulp used in the experiments
of Paper III.
Between the experimental periods, the batch pulp samples were stored in a
freezer. During the experiments, the pulps were stored in a refrigerator. It is well
known that freezing changes the fiber properties by causing irreversible changes
in the fiber structure. However, when considering the aim of the study, the
changes in the fiber properties were not significant whenever there was no direct
comparison between the experimental results. The charge of fibers and the metal
contents of fibers were always determined anew when the pulp had been stored
for a longer period. In particular, the metal analysis of fiber by AAS is subject to
quite significant experimental errors, up to 20% for sodium. This depends on the
homogeneity of pulps as well as on the weighing and determination of dry
matter content during the metal analyses.
36
Table 2. Pulp properties.
The metal amounts in the analysis results are quite similar with other studies e.g.
those reported by Vuorinen et al; Na 123 mmol kg-1, Mg 8.1 mmol kg-1, Ca 42
mmol kg-1, Mn 1.7 mmol kg-1 for oxygen delignified birch kraft pulp /37, 38/.
The equilibrium experiments were carried out as titrations and batch equilibrium
experiments as described in Papers I and II. Fiber charge was usually measured
by potentiometric titration with commercial apparatus (Metrohm GPD 751) but
also by methylene blue adsorption /39/ as well as more specific uronic acid
analyses of fibers, as discussed in Paper I. All the chemicals used in the titrations
and other experiments were analytical grade, except the BCA6 chelant (Paper
IV) which was obtained as technical grade containing impurities from the
37
synthesis (monocarboxylic acids). Calculated chemical concentrations were used
in the experiments and modelling.
Separate experiments not reported in Papers I–IV for the determination of ion
exchange kinetics of pulp suspensions were also carried out. The experimental
procedure given in the following describes one such experiment.
38
the pulp to Ca++ form. The excess solution was removed from the pulp by
centrifuging and the fibers were then suspended in a 0.7 mmol dm-3 solution of
CaCO3 at 1% consistency in orded to precipitate solid calcium carbonate into the
fibers. The total amount of calcium was 117 mmol kg-1 o.d. fibers after the
saturation treatment.
After the conversion of the pulp to Ca++/CaCO3 form, the pulp was diluted to 1%
consistency in distilled water. The pulp was acidified with H2SO4 to pH = 4.75,
after which samples of the suspension were taken with a syringe over a 5 hours.
The suspension samples were filtered using a quantitative filter, on the
assumption that most of the solid calcium carbonate remains in the pulp. Filtrate
samples were acidified to pH = 1 by the addition of 60 % HNO3, after which the
Ca++ concentration in filtrate was determined by AAS. The pH of the suspension
was monitored during the experiment.
39
4 Results and discussion
4.1 The influence of ionic composition on the
Donnan distribution
The equations formulated and tested in Papers I and II account for the anionic
composition of pulp suspension and complexation equilibria in the aqueous
phases. The Donnan distribution coefficient, λ , depends on the total amounts of
all ionic species in the pulp suspension. The pH of the suspension is varied by
the addition of different acids or bases, for example with H2SO4 and NaOH. The
value of the Donnan distribution coefiicient depends mainly on the pH-fiber
charge relationship but the ionic strength and the amounts and valencies of
freely mobile cations are also of importance.
Thus, the addition of chemicals and specific complexation reactions cause sharp
changes in the behavior of λ as a function of pH. This effect is very clearly
demonstrated in the calculation result presented in Fig. 4 that represents the
calculated value of λ as a function of pH in fiber suspension in the presence of
EDTA and DTPA.
The calculation results suggest that the total charge of the fibers and the ionic
strength of the system determine the maximum value of λ . The concentrations
of free cations vary due to the specific complexation with chelants in the pH
range 4–7 due to the different complexation abilities of EDTA and DTPA. This
has a clear effect on the calculated Donnan distribution coefficient. The fiber
charge reaches its maximum at pH = 10, i.e. the fiber charge does not increase
with an increasing amount of NaOH. Thus the value of the Donnan distribution
coefficient decreases due to the increase of electrolyte in the fiber phase above
pH 10.
40
25
Addition of H2SO4 Addition of NaOH
20
15
EDTA
λ
DTPA
10
0
1 2 3 4 5 6 7 8 9 10 11 12
pH of the external solution
41
fiber phase in the same molar proportions and are much more strongly enriched
than the monovalent ions.
As described in Paper II, the amount of chelant was limited for the chelation of
all divalent metal ions. The complex stabilities of divalent metals decrease in
order of Mn:DTPA > Ca:DTPA > Mg:DTPA. The amount of chelant was
sufficient for the complexation of most of the manganese and about 50 % of the
calcium, while the complexation of magnesium (and sodium) was almost
insignificant.
In the experimental validation of the model (Paper II) it was also seen that it
correctly predicts the pH ranges where the metal complex concentrations in the
external phase start to increase due to complexation, i.e. the inflection points in
the concentration curves of manganese and calcium.
42
1.2
1
Na
Mg
0.8
Ca
Mn
0.6
Na
Mg
0.4
Ca
0.2 Mn
0
0 2 4 6 8 10 12
pH
43
0.00055
0.00050
Me(++)
0.00045 Me(++) *
Ca(++), exp
0.00040
0.00035
0 0.2 0.4 0.6 0.8 1
Dimensionless time
44
fractions of metals. However, even the rapid kinetics must be accounted for
when considering the flow of a solution in an immobile pulp bed.
The experiments described in Paper III were carried out using strong acid (1.5
mmol dm-3 HNO3) and 1 mmol dm-3 solutions of BCA6 and EDTA chelants.
The predominant factor determining the overall mass balance of removed metals
as well as the removal kinetics was the pH of the solution.
In the displacement with nitric acid solution without chelant, the mechanism of
removal of metals from the bed can be described as follows: When acid is added
to an initially alkaline bed the acid will be first neutralized by OH- ions in free
solution. Excess electrolyte, i.e. metal cations with free anions, transfers in the
fiber phase. Due to the increased electrolyte concentration in fiber the phase the
concentration differences of all ions between the phases reduce. When the
displacement solution moves through the bed there will ideally be a neutralized
‘front’ of displacement solution that has increasing ionic strength with
increasing distance from the top of the bed. This displacement front with
increasing ionic strength also promotes the transport of free anions to the fiber
phase and the charge neutralization of fiber-bound anionic groups.
The polyanionic ligands and free complexes have a higher negative valency than
simple anions and thus they are excluded to a significantly larger extent from the
fiber phase. The experiments with chelants described in Paper III were carried
out at pH that only partially neutralized the charge of fibers. As demonstrated
experimentally and by model calculations in Papers III and IV the removal of
metals decreases with the higher pH of the chelant solution, which stresses the
significance of reactions in the anionic fiber phase as a factor that limits the
kinetics of ion exchange between the fiber and the external phases.
45
In the results represented in Papers III and IV, the overall removal of metals
from the bed with BCA6 and EDTA chelants was significantly lower than
removal with nitric acid. The low equilibrium concentrations of ligands with
high ionic charge in the fiber should also have a negative effect on metal
removal selectivity. Indeed, it was observed in the experiments that the removal
of metal with relatively high complex stability, i.e. Mn, was reduced so that
similar removals of metals with lower complex stability, i.e. Ca and Mg, was
achieved. When compared to batch ion exchange experiments (Paper II and Fig.
5), the chelants are able to remove metals from fibers quantitatively and
selectively when the pH is in the correct range and the reaction time is sufficient.
The transfer of metal cations from the fiber phase can be seen to be caused by
charge neutralization of anionic groups bound in fibers. There is a ‘competition’
between the negative charge of fibers that keeps the metal cation in fibers, and
ligands in external solution that bind the cations in anionic complexes that are
excluded from the fiber phase. Thus the ion exchange kinetics is of high
importance when considering the flow of chemical solutions in the pulp bed or
suspensions. In comparison to other displacement studies /44, 45, 46/ it can also
be deduced that the chelant concentration as it determines the pH is a critical
factor determining the metal removal efficiency in displacement chelation.
The kinetic mass balance equations can easily be extended to account for input
and output of solute components in the external phase. The principle shown in
Fig. 3 and the derived equation, i.e. the Eq. (25) can be used to model a single
CSTR reactor as well as the flow of water approaching the plug flow in the fiber
bed by combining many CSTR model units in series.
Because a single model ‘unit’, i.e. l = L = 1 represents a CSTR and many units
combined in series a plug flow reactor, it is possible to model different
46
displacement or washing efficiencies by varying the number of model units.
However, as it can be quite difficult to describe the displacement efficiencies of
such a unit combination, this approach is not robust. In fact, this kind of
modelling attempt of washing has been carried out /47/.
The earlier washer model development /47/ consisted of many CSTR units in a
two-dimensional arrangement. The ion exchange kinetics was not taken into
account, i.e. it was assumed that the pulp suspension is always in equilibrium. In
order to model varying washing and displacement efficiencies, the number of
model units in series or in parallel had to be changed, which required extensive
programming. Due to the neglect of kinetics in the model the control of overall
mass balances of ionic species was only approximate /47/. However, the
modelling attempt described in ref. /47/ was a preliminary proof for the
applicability of the Donnan model and its extensions to modelling of washing
and other fiberline processes.
47
The curve denoted as “ideal plug flow” denotes output of the mobile phase
model only. The output of the mobile phase model was then used as input for the
PPF model (Eqs 26–30) with a number of channels, J = 1000 and by varying the
Peclet number in range 5–100. Actually, the PPF model reproduces a
breakthrough curve that is exactly similar to the breakthrough curve from the
mobile phase model (i.e. input of PPF model) when the Peclet number is very
high, >1000.
1.2
100 IDEAL PLUG FLOW
1.0
50
0.8
CSTR 5
0.6
0.4
0.2
0.0
0 1 2 3 4 5 6
DIMENSIONLESS DISPLACEMENT VOLUME
48
Modelling of removal of calcium by BCA6 and EDTA chelants and by nitric
acid from pulp beds is shown in Fig. 8. Two BTC’s for displacement with BCA6
are shown in Fig. 8; the thin solid curve denoted as “IDEAL PLUG FLOW”
represents the output from an almost ideal plug flow model, i.e. the exit
concentration from a series of ten CSTR model units. The line denoted as “PPF
MODEL” represents the modelling result when the output of an ideal plug flow
model has been used as an input of a PPF model at Pe = 50. The other BTC
curves represent the modelling results of displacement experiments with nitric
acid (the highest concentrations) and EDTA (the lowest concentrations).
2.0
1.5
1.0
0.0
0 1 2 3 4 5 6
DIMENSIONLESS DISPLACEMENT VOLUME
As can be seen from the dotted BTC in Fig. 8, the concentration of manganese
steeply increases immediately after displacement of one bed volume of chelant
solution. In the PPF model, the same ideal BTC curve is normalized to different
49
residence times of pulp bed ‘channels’, weighed i.e. multiplied by flow fractions
of each channel and integrated over a range of average residence times of the
bed. The result from the PPF model is the solid model curve shown in Fig. 8.
The effect of kinetics on calcium removal can be seen from the secondary
maximum in the dotted BTC and as a tailing behavior of both BTC’s. The
secondary concentration maximum occurs in modelling when the pH of the bed
becomes acidic and the internal kinetics become significant. As can be seen by
comparison of the PPF model curve with the experimental results in Fig. 8 and
in the other results represented in Papers III and IV the combination of the
mobile phase model with the PPF model describes quite well the removal of
metals from the pulp bed.
The basic input to all modelling examples are the fiber charge properties, the
FSP or WRV of fibers, the total amounts of metals, anions and chemicals as well
as the suspension and liquid volumes and mass of pulp. The determination of all
of these parameters requires quite a lot of analytical work. It can also be argued
that a model with many inputs is subject to inaccuracy and experimental
precision when determining the model inputs. Fortunately the basic Donnan
equilibrium model, as well as its extensions described in this work, is quite
robust and not too sensitive to the small variations in conditions.
The total anionic charge of fibers and the composition of acid structures in fibers
are quite critical parameters. The volume of water in the fibers can be assumed
50
to be constant although it varies with pH and can become quite a critical
parameter in higher ionic strengths (as discussed in Paper I). As shown by the
conceptual calculation example shown in Fig. 9 the fiber charge as a function of
the external solution pH has a very significant effect when compared to the
variation in swelling.
1.0
0.8
0.4
0.2
DIVALENT IONS
0.0
1 2 3 4 5 6 7 8 9 10
pH of the external solution
Thus, estimates of the fiber properties could be used in modelling when only
qualitative information of the system is required. Collection of data of fiber
properties from different processes and pulp grades would facilitate the rapid
estimation of input parameters of models and qualitative analysis of the system.
Quantitative characterization requires precise determination of fiber properties.
51
The ionic composition of pulp suspensions, i.e. the total amounts of metal
cations and different anions, is a critical input in the Donnan model. It is an
essential requirement that the total amounts of the cations dominating the
cationic composition (e.g Na, K, Ca, Mg, Mn) are known. It is also essential to
know the mass balances of chemicals used e.g. in the pH adjustment or what
chemicals are used and how much. As described in Paper II, the ionic conditions
in pulp suspension can be predicted quite well by calculating the theoretical
metal concentration and pH curves when the additions of acids, bases or chelants
are known.
52
5 Conclusions
This study describes development and validation of several extensions to a pulp
suspension equilibrium model based on the Donnan equilibrium theory. The
extensions account for the overall ionic composition of pulp suspensions, the
complexation equilibria, ion exchange kinetics and flow of external phase with
respect to immobile fibers. The flow model was extended to a batch
displacement model.
The incorporation of the different anionic species in the Donnan model can be
considered as a generalization of earlier models based on Donnan theory.
However, the general formalism of the Donnan model presented in this study
requires a new way of utilizing the electroneutrality conditions of the Donnan
model and more precise knowledge of the overall ionic composition of pulp
suspensions as input parameters. An important finding of this study is that the
ion exchange kinetics of pulp suspensions plays a major role in determining the
selectivity and efficiency of the chelation of pulp suspensions. The main factor
determining the ion exchange kinetics is the pH through the reactions taking
place within the fiber phase when the external pH is changed.
Kinetic ion exchange models developed in this work, in particular the ‘internal’
kinetic model, require further refinement. Better relations of ionic conditions of
suspension to actual forms of metals in fibers and to structural characteristics of
fibers need to be defined. The connection of a simple kinetic model to Donnan
equilibria, however, has proven to be a quite easily manageable approach in the
modelling of kinetics of complex ionic composition of pulp suspensions.
53
The ionic distribution model for fiber suspension and its extensions, as generally
represented by Eqs (1)–(30), facilitates quantitative modelling of equilibrium
and kinetic chemical phenomena occurring in pulp suspension in systems subject
to varying amounts of ions and flow of water with respect to immobile fibers in
the bed. As shown in Papers I and II the Donnan equilibrium model is otherwise
similar to earlier Donnan equilibrium models but it is more generally applicable
for the modelling of varying ionic compositions and direct calculation of the pH
of pulp suspensions. As demonstrated in Papers III and IV, realistic modelling
results have been obtained by using model inputs that are measurable properties
of pulp and bed and just a few adjustable parameters that describe the kinetics.
The model calculations and the analysis of experimental data obtained in Papers
I–IV suggest the following:
• The pH of the suspension is the most important factor determining the ion
exchange rate between the fiber and the external phases. Thus, it can be
understood that the kinetics of ion transfer from the fiber phase to the
external phase (or vice versa) is largely governed by the rate of change of
fiber charge.
• The ion exchange kinetics can be described by dividing it into two main
processes. The internal process includes sub-mechanisms that can be
54
dissolution of solid metal compounds and the effect of fiber structure on ion
diffusion rates. The overall ion exchange of the suspension is governed by
the difference between the initial concentrations and the final equilibrium
state that the system is approaching.
• The metal ions occurring in the pulp fibers in highest amounts e.g. Na+, K+,
Mg++, Ca++, and Mn++ interact non-specifically with the fibers. The
interactions of Mn++, Ca++ and Mg++ with chelants, however, are specific.
The competition between these non-specific and specific interactions is won
by fibers whenever the mass transfer kinetics of metals from the fibers is
limited in the time scale of reaction. The result will be reduced chelation
efficiency and selectivity. It also seems that the chelation process can be
optimized and new techniques further developed by optimizing the process
conditions in terms of pH, chemical concentrations and reaction time.
55
6 Ideas for future research
The equilibrium models used in this work are based on ideal Donnan theory. The
models describe laboratory experiments quantitatively but they may be
unsuitable for the modelling of industrial pulp suspensions. This problem can be
overcome by applying nonideal solution models in the Donnan theory. Work is
currently being done in the area of theoretical modelling /52/.
56
References
1. Vuorinen, T., Buchert, J., Teleman, A., Tenkanen, M. and Fagerström, P.
‘Selective hydrolysis and its application in ECF and TCF bleaching of kraft
pulps’, Intl. Pulp Bleaching Conference, April 14–18, 1996, Washington
D.C., p. 39.
3. Laine, J. ‘The effect of cooking and bleaching on the surface chemistry and
charge properties of kraft pulp fibers’ Doctoral thesis, Helsinki University of
Technology, Espoo, Finland, 1996.
6. Ragnar, M., Lindgren, C.T. and Nilvebrandt, N.-O. ‘On the dissociation
constants of phenolic groups in lignin structures’ 10th Intl. Symposium on
Wood and Pulping Chemistry, Vol. II, June 7–10, 1999, Yokohama, Japan,
p. 154.
57
10. Sundberg, A., Pranowich, A. and Holmbom, B. ‘The origin and distribution
of anionic groups in TMP’, Pre-Symposium of the 10th Intl. Symposium on
Wood and Pulping Chemistry, Seoul, Korea, June 2–4, 1999, p. 254.
11. Lindgren, J. ‘Experimental studies of the acid/base properties and metal ion
affinities of wood fibers’, Doctoral thesis, Umeå University, Dept. of
Inorganic Chemistry, Umeå, Sweden, (2000).
13. Laivins, G.V. and Scallan, A.M. ‘The exchange and removal of the metal
cations in pulps’ Fundamentals of papermaking materials – 11th
Fundamental Research Symposium, Vol. 2. Sept. 21–26, 1997, Cambridge,
UK, p. 837.
14. Teleman, A., Harjunpää, V., Tenkanen, M., Buchert, J., Hausalo, T.,
Drakenberg, T. and Vuorinen, T. ‘Characterization of 4-deoxy-B—L-threo-
hex-4-enopyranosyluronic acid attached to xylan in pine kraft pulp and
pulping liquor by 1H and 13CNMR spectroscopy’, Carbohydr. Res.
275(1995):55.
15. Donnan, F.G. and Harris, A.B. ‘The osmotic pressure and conductivity of
aqueous solutions of Congo-red, and reversible membrane equilibria’, J.
Chem. Soc. 99(1911):1554.
17. Procter, H.R. and Wilson, J.A. ‘The acid-gelatin equilibria’, J. Chem. Soc.
109(1916):307.
18. Neale, S.M. ‘The swelling of cellulose and its affinity relations with aqueous
solutions’, J. Textile Inst. 20(1929):373.
19. Farrar, J. and Neale, S.M. ‘The distribution of ions between cellulose and
solutions of electrolyte’, J. Colloid Interface Sci. 7(1952):186.
58
20. Donnan, F.G. and Allmand, A.J. ‘Ionic equilibria across semi-permeable
membranes’, J. Chem. Soc. 105(1914):1.
23. Scallan, A.M. and Grignon, J. ‘The effect of cations on pulp and paper
properties’, Svensk Papperstidn. 82(1979):40.
24. Laivins, G. and Scallan, T. ‘Acidic versus alkaline beating of pulp’, J. Pulp
Paper Sci. 26(2001)2:228.
26. Sjölander, C., Lidén, J. and Öhman, L.-O. ‘Modelling the distribution of
“free”, complexed and precipitated metal ions in a pulp suspension using
Donnan equilibria’, 2000 Intl. Pulp Bleaching Conference, June 27–30,
2000, Halifax, Canada, Oral Presentations, p. 273.
27. Athley, K. and Ulmgren, P. ‘Interaction between divalent metal ions and
oxygen-delignified kraft pulp’, Nordic Pulp Paper Res. J. 16(2001)3:204.
28. Andersson, R., Lidén, J. and Öhman, L.-O. ‘The Donnan theory applied to
pulp washing’ 7th Intl Conference on New Available Technologies, June 4–
6, 2002, Stockholm, Sweden, p. 42.
29. Martell, A.E. and Smith, R.M. ‘Critical stability constants Volume 1–4’,
Plenum Press, 1974.
30. Laidler, K.J. and Meiser, J.H. ‘Physical Chemistry’ The Benjamin &
Cummings Publishing Company, Menlo Park, California, 1982, 919 p.
59
31. Herrington, T.M. and Petzold, J.C. ‘Surface area of papermaking woodpulps
used by the British paper industry’, Cellulose 2(1995):83.
33. Sherwood, T.K. and Wei, J.C. ‘Ion diffusion in mass transfer between the
phases’, A.I.Ch.E.J., 1(1955)4:522.
36. Levenspiel, O. ‘Chemical Reaction Engineering’ 2nd ed. John Wiley & Sons,
New York, 1976.
37. Vuorinen, T. ‘TCF system closure’. The 1st Johan Gullichsen Colloquium,
4–5 September 1997, Proceedings, The Finnish Paper Engineer’s
Association, p. 123.
39. Davidson, G.F. ‘The acidic properties of cotton cellulose and derived
oxycelluloses, Part II: The adsorption of methylene blue’, J. Textile Inst.
39(1948):65.
40. Sjöström, E., Janson, J., Haglund, P. and Enström, B. ‘The acidic groups in
wood and pulp as measured by ion exchange’, J. Polym. Sci. (1965)11:221.
60
41. Mäenpää, H. ‘Simulation of ionic equilibria in the wet-end of
papermachines’, Master’s Thesis, Helsinki University of Technology,
Espoo, Finland, 2001 (in Finnish).
43. Yantasee, W. ‘Kinetic and equilibrium analysis of metal ion adsorption onto
bleached and unbleached kraft pulps’, PhD Thesis, Oregon State University,
USA, 2001.
44. Ni, Y., Krishnan, S. and van Heiningen, A.R.P. ‘Modelling manganese
removal from pulp fibers during displacement chelation’, J.Pulp Paper Sci.
27(2001)12:412.
46. Ni, Y., Li, Z. and van Heiningen, A.R.P. ‘Pulse chelation with DTPA’, Pulp
Paper Can. 100(1999)12:T382.
47. Räsänen, E., Tervola, P., Fiskari, J., Stenius, P. and Vuorinen, T. ‘Modelling
of non-process element balances in washing and bleaching of kraft pulp’,
2000 Intl. Pulp Bleaching Conference, June 27–30, 2000, Halifax, Canada,
Oral Presentations, p. 267.
48. Nordén, H.V. and Pekkanen, M. ‘Semibatch models of filter cake washing’,
Kemia-Kemi 2(1975)6:247.
49. Ali, O. ‘Dynamic chlorination of kraft pulp’ PhD thesis, McGill University,
Montréal, Canada, 1982.
50. Ni, Y., Li, Z., and van Heiningen, A.R.P. ‘Flowthrough characterization of
mechanical pulp pads’, J. Pulp Paper Sci. 25(1999A)8:294.
61
51. Räsänen, E., Kärkkäinen, L., Tervola, P., Gullichsen, J., Stenius, P. and
Vuorinen, T. ‘Characterization of acid-base properties of cellulose and
lignin-originated reaction products by potentiometric titration’, Grenoble
Workshop on Advanced Methods for Lignocellulosics and Paper Products
Characterization, Post-Symposium of the 11th Intl. Symposium on Wood and
Pulping Chemistry, June 18–19, 2001, Grenoble, France, p. 86.
52. Koukkari, P., Pajarre, R. and Pakarinen, H. ‘Modelling of the ion exchange
in pulp suspensions by Gibbs energy minimization’, J. Solution Chem.
42(2002)10:623.
62
Published by Series title, number and
report code of publication
VTT Publications 495
VTT–PUBS–495
Author(s)
Räsänen, Erkki
Title
Modelling ion exchange and flow in pulp suspensions
Abstract
This thesis summarizes the results of four papers dealing with modelling of ion exchange between cellulosic fibers
and their surrounding aqueous solution.
Transition metals, in particular manganese, iron and copper, induce fiber-deteriorating radical reactions in oxygen-
chemical-based bleaching, while magnesium inhibits such reactions. Thus the removal of the harmful transition
metals from a pulp is required while the beneficial magnesium should be retained in a pulp. Selective control of
metal contents of pulps is a key factor in the control of bleaching chemistry. Thus, ability to predict the extent of ion
exchange and amounts of metals are of great practical interest.
An ion exchange model to describe interaction of ionic species with kraft pulps has been developed in the present
work. The model can be applied in pulp washing and chelation. The present model takes into account overall ionic
composition of pulp suspensions, i.e. the cations and simple anions as well as the ligands and the complexes. The
model facilitates calculation of pH, complexation and distribution of ionic species between the aqueous phase
confined to fibers and the external solution. The present model also accounts for the ion exchange kinetics and flow
of water through the immobile fiber network, i.e. the ”pulp bed”.
The present model is based on the formerly developed ion exchange model, which is based on so-called Donnan
equilibrium theory. The development of the new model required the inclusion of the different anionic species to
model and the combination of the Donnan model with a complexation model. The ion exchange kinetics is described
as a two-step phenomenon. The first rate-limiting step can be due to the dissolution of solid metal compounds in the
fiber phase as well as the diffusion of ions in the fiber wall, while the second step is the transfer of the ionic species
between the fiber phase and the external solution, which is governed by the concentration gradient between the
phases. In the model, the transfer of ions between the phases is directly related to the Donnan theory.
In the testing of the elements of the ion exchange model, it was found that the combination of the Donnan model
with a complexation model quantitatively predicts the pH, the ionic distribution as well as the complexation
equilibria in batch-type systems when the time to achieve full equilibrium is not limited. In the pulp beds subject to
flow the ion exchange kinetics reduces the metal removal efficiency and selectivity. This can be considered to be
due to the repulsion of anionic species (simple anions, ligands and complexes) from the negatively charged fiber
phase. It can also be considered that the rate of change of fiber charge determines the overall ion exchange kinetics.
It was found in the experiments that a low pH of displacement solution favors rapid metal removal from pulp beds.
However, a low pH reduces the metal removal selectivity, i.e. harmful manganese and the beneficial magnesium is
also removed from the pulp. The metal removal process in laboratory or in industrial scale could be optimized in
terms of chemical concentrations, pH, reaction time and wash flow conditions. The present model facilitates realistic
modelling of these conditions.
Keywords
ion exchange, paper and pulp, pulp suspensions, cellulosic fibers, aqueous solutions, models, modelling,
transition metals, bleaching, chelants, metal complexes, theses, dissertations
Activity unit
VTT Processes, Biologinkuja 7, P.O.Box 1602, FIN–02044 VTT, Finland
ISBN Project number
951–38–6044–2 (soft back ed.)
951–38–6045–0 (URL:http://www.vtt.fi/inf/pdf/)
Date Language Pages Price
May 2003 English 62 p. + app. 111 p. D
Series title and ISSN Sold by
VTT Publications VTT Information Service
1235–0621 (soft back ed.) P.O.Box 2000, FIN–02044 VTT, Finland
1455–0849 (URL: http://www.vtt.fi/inf/pdf/) Phone internat. +358 9 456 4404
Fax +358 9 456 4374
ESPOO 2003 VTT PUBLICATIONS 495