Escolar Documentos
Profissional Documentos
Cultura Documentos
by
August 2005
THE UNIVERSITY OF NEW SOUTH WALES
Thesis/Project Report Sheet
Abstract
Ion exchange equilibria are presented for Au(CN) 2 [ ] − −
[
/ Cl , Au(CN) 2 ]− / SCN − − −
and SCN / Cl in aqueous solution, and in
various mixed solvents, at 303K using Purolite A500 as the ion-exchanger. The mixed solvents investigated include water-acetone,
water-dimethylsulfoxide (DMSO) and water-N-methyl-2-pyrrolidone (NMP). In aqueous solution, the selectivity of Purolite A500 for
[
exchange resin for the Au(CN) 2 ]− species. In mixed solvents, however, the selectivity of Purolite A500 for [Au(CN) 2 ]− decreases
with an increase in the composition of the organic solvent in the external solution. Mixed solvents containing greater than 60 mol%
organic solvent are preferred for the displacement of Au(CN) 2 [ ]− from the resin. The effectiveness of a given type of mixed solvent
generally increases in the following order: DMSO < acetone < NMP.
The ion exchange equilibria are correlated using the Law of Mass Action, modified with activity coefficients, to determine the
equilibrium constant for each binary system. The fitted values of the equilibrium constants are consistent with the trends observed in
the ion exchange isotherms. The accuracy of the correlation results in the mixed solvent systems range from 1 to 10% and this is
similar to the level of accuracy obtained for the ion exchange equilibria in aqueous solution. From these results it can be concluded that
the Law of Mass Action is equally valid in mixed solvent systems.
The variation in the equilibrium constant with mixed solvent composition, for a given binary system, correlates well with the dielectric
constant of the mixed solvent. For a given value of the dielectric constant, however, the equilibrium constant, however, the equilibrium
constant is dependent on the type of mixed solvent. A fundamental relationship is derived between the equilibrium constants and the
Gibbs energies of transfer associated with the solvation of the ions in the mixed solvents. Based on this relationship, the redistribution
of ions between the pore solution and the bulk mixed solvent, appears to be the most significant factor that governs the selectivity of
the resin in mixed solvent systems.
I am fully aware of the policy of the University relating to the retention and use of higher degree project reports and thesis, namely that the University
retains the copies submitted for examination and is free to allow them to be consulted or borrowed. Subject to the provisions of the Copyright Act
1968, the University may issue a project report or thesis in whole or in part, in Photostat or microfilm or other copying medium.
I also authorise the publication of University Microfilms of a 350 word abstract in Dissertation Abstracts International (applicable to doctorates only).
The University recognizes that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for
restrictions for a period of up to 2 years must be made in writing to the Registrar. Requests for longer period of restriction may be considered in
exceptional circumstances if accompanied by a letter of support form the Supervisor or Head of School. Such request must be submitted with the
thesis/project report.
Ion exchange equilibria are presented for [Au(CN) 2 ] / Cl − , [Au(CN) 2 ] / SCN − and
− −
SCN − / Cl − in aqueous solution, and in various mixed solvents, at 303K using Purolite
A500 as the ion-exchanger. The mixed solvents investigated include water-acetone,
water-dimethylsulfoxide (DMSO) and water-N-methyl-2-pyrrolidone (NMP). In
aqueous solution, the selectivity of Purolite A500 for a given anion increases in the
order: Cl − > SCN − > [Au(CN)2 ] . This selectivity sequence confirms the high affinity of
−
the ion exchange resin for the [Au(CN)2 ] species. In mixed solvents, however, the
−
from the resin. The effectiveness of a given type of mixed solvent generally increases
in the following order: DMSO < acetone < NMP.
The ion exchange equilibria are correlated using the Law of Mass Action, modified with
activity coefficients, to determine the equilibrium constant for each binary system. The
fitted values of the equilibrium constants are consistent with the trends observed in the
ion exchange isotherms. The accuracy of the correlation results in the mixed solvent
systems range from 1 to 10% and this is similar to the level of accuracy obtained for the
ion exchange equilibria in aqueous solution. From these results it can be concluded that
the Law of Mass Action is equally valid in mixed solvent systems.
The variation in the equilibrium constant with mixed solvent composition, for a given
binary system, correlates well with the dielectric constant of the mixed solvent. For a
given value of the dielectric constant, however, the equilibrium constant is dependent on
the type of mixed solvent. A fundamental relationship is derived between the
equilibrium constants and the Gibbs energies of transfer associated with the solvation of
the ions in the mixed solvents. Based on this relationship, the redistribution of ions
between the pore solution and the bulk mixed solvent, appears to be the most significant
factor that governs the selectivity of the resin in mixed solvent systems.
ii
ACKNOWLEDGMENTS
First and foremost I would like to thank Dr Frank Lucien, for giving me the privilege to
work under his supervision. His knowledge, perceptiveness and attention to detail have
taught me a great deal more than the thesis itself. I have reached the end of this road,
only because of his enthusiasm, dedication and friendship.
This work would not exist without the strong foundation set by Rueben Rajasingam:
thank-you for your guidance and friendship. I would also like to thank my colleagues
that have made the past four years a most memorable experience: Kenneth Lee, Hon-
Sang, Grzegorz, Yun Lei, Ringo, Robert and Assem.
A special thank-you to Raf and Danielle for sharing a home with me for the duration of
my PhD and for a friendship that was indispensable in getting through the past four
years.
My parents Rekha and Nimal Jayasinghe gave me the freedom to find my own way. I
am indebted to the sacrifices they made to fulfill my education. Also, to my sister
Shivani, for being by my side, and for her constant love. A special thank-you to the
Wickramanayake family who kept their doors open and extended their warmth
unreservedly.
Finally to Rasika for making me believe that the end was achievable. And most of all
for being my best friend, without which this would not have been possible.
iii
TABLE OF CONTENTS
CHAPTER 1
INTRODUCTION……………………………………………………………. 1
1.1 REFERENCES………………………………………………………………5
CHAPTER 2
ION EXCHANGE IN GOLD PROCESSING………………………………….... 6
2.1 INTRODUCTION…………………………………………………………… 6
2.2 PROPERTIES OF GOLD……………………………………………………. 6
2.2.1 Gold ore deposits ………………………………………………………..7
2.3 THE EXTRACTION OF GOLD FROM ORES ……………………………….9
2.3.1 Cyanidation ……………………………………………………………9
2.3.2 Carbon technology …………………………………………………….14
2.4 ION EXCHANGE TECHNOLOGY ………………………………………….20
2.4.1 Properties of ion exchange resin ……………………………………….20
2.4.2 The use of ion exchange resins in the gold industry …………………...26
2.5 SELECTIVITY ……………………………………………………………28
2.5.1 Copper-gold ores ………………………………………………………32
2.5.2 Loading ………………………………………………………………34
2.5.3 Elution …………………………………………………………………36
2.6 CYANIDE MANAGEMENT ………………………………………………43
2.6.1 The AugMENT process ………………………………………………..45
2.6.2 The Vitrokele process …………………………………………………45
2.6.3 Elutech ………………………………………………………………..46
2.7 REFERENCES …………………………………………………………….48
CHAPTER 3
ION EXCHANGE EQUILIBRIA: SELECTIVITY AND PREDICTIVE MODEL… 54
iv
3.5 PREDICTIVE MODELS FOR ION EXCHANGE EQUILIBRIA ……………….73
3.5.1 Quantitative treatment of selectivity …………………………………...73
3.5.2 Adsorption models ……………………………………………………..77
3.5.3 The Law of Mass Action ………………………………………………80
3.6 REFERENCES …………………………………………………………….91
CHAPTER 4
[2
] 2
[ ]
ION EXCHANGE EQUILIBRIA OF Au(CN) − /Cl − , Au(CN) − /SCN − and
− −
SCN /Cl BINARY SYSTEMS IN AQUEOUS SOLUTIONS ………………..96
CHAPTER 5
[2
] 2
[ ]
ION EXCHANGE EQUILIBRIA OF Au(CN) − /Cl − , Au(CN) − /SCN − and
− −
SCN /Cl BINARY SYSTEMS IN MIXED SOLVENTS …………………..124
v
CHAPTER 6
[
MODELING ION EXCHANGE EQUILIBRIA OF Au(CN) − /Cl − ,
2
]
[Au(CN)
2
]
− − − −
/SCN and SCN /Cl BINARY SYSTEMS IN MIXED
SOLVENTS ………………………………………………………………157
CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS ………………………………188
vi
LIST OF FIGURES
Figure 2-1: Eh (potential)-pH diagram for the Au-CN-H2O system, with initial
concentrations of 0.001 and 0.0001M for CN- and Au respectively.….…12
Figure 2-2: A comparison of the elution profiles of gold in the Murdoch, AARL and
Zadra processes (adapted from Ruane, 1982).……………………………17
Figure 2-3: Schematic cross-sectional view of an ion exchange resin bead (A). Fixed
charges (-) are balanced by counter ions (+) diffused throughout the
framework of the resin (B). Typical anion exchange resins have a styrene
divinyl benzene matrix, with a tri-methyl amine functional group (C).….24
Figure 2-4: A typical RIP or RIC flowsheet for treating gold ores…………………...29
Figure 2-5: Equilibrium loading for various metal cyanide complexes from a mine
leach solution (adapted from Fleming and Cromberge, 1984a.…………..42
Figure 2-6: The effect of organic solvent on the elution of gold from resin (adapted
from Law et al., 1985). …………………………………………………42
Figure 3-1: Schematic diagram of the formation of an ion-pair ………………………58
Figure 3-2: Ion exchange isotherms for the Cu 2+ / Na + binary system as a function of
total solution normality (adapted from Subba Rao and David, 1957) …...58
Figure 3-3: The effect of hydration on counterion-functional group interaction.……..61
Figure 3-4: The loading of metal ions on to anion exchange resin from an aqueous
solution adjusted with 4M LiCl (adapted from Fleming and Monhemius,
1979). ……………………………………………………………………71
Figure 3-5: The loading of metal ions on to anion exchange resin from an acetonitrile
solution adjusted with 4M LiCl (adapted from Fleming and Monhemius,
1979). ……………………………………………………………………71
Figure 3-6: A summary of major factors influencing the selectivity of an ion exchange
resin for representative ions X − and Y − . ………………………………72
Figure 3-7: Ion exchange isotherms for the exchange involving ion A and ion B. …...75
Figure 3-8: The variation of the activity coefficients of Mn 2+ and Na+ in the resin
phase as a function of the equivalent fractions of the ions in the resin for
the Mn 2+ / Na + binary system (adapted from Bajpai et al., 1973)………83
Figure 4-1: Representative kinetic data depicting the length of time required for the
loading process to reach equilibrium for 1. [Au(CN) 2 ] (B) / Cl − (A) , 2.
−
[Au(CN) 2 ]− (B) / SCN − (A) and 3. SCN − (B) / Cl − (A) binary systems...101
Figure 4-2: Cross sectional views of resin beads loaded with [Au(CN)2 ] originally in
−
the chloride form in, a partially loaded resin (A), and a completely loaded
resin bead (B). The wt % is represented by the colour scheme shown on
the legend. ………………………………………………………………103
vii
Figure 4-3: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in aqueous solution at
−
303 K and for various total solution concentrations. The solid curve
represents the correlation of the data with equation 4-10 (3-parameter
regression). (B) Expanded view of the ion-exchange isotherm shown in
(A).………………………………………………………………………110
Figure 4-4: Ion exchange isotherms for [Au(CN) 2 ] (A) / SCN − (B) in aqueous solution
−
at 303 K and for various total solution concentrations. The solid curve
represents the correlation of the data with equation 4-10 (3-parameter
regression). ……………………………………………………………..111
Figure 4-5: Ion exchange isotherms for SCN − (B) / Cl − (A ) in aqueous solution at 303 K
and for various total solution concentrations. The solid curve represents
the correlation of the data with equation 4-10 (3-parameter regression). 111
Figure 4-6: The selectivity of resin towards the gold cyanide complex as a function of
ionic refractivity of ion, X − displaced from the resin by the gold cyanide
complex (adapted from Aveston et al., 1955). ………………………….114
Figure 4-7: The spread of data points over the range of resin phase equivalent fraction
for the SCN − (B) / Cl − (A) binary system for various total solution
concentrations (2-parameter regression). ……………………………….120
Figure 4-8: Variation of the activity coefficients of [Au(CN) 2 ] and Cl − in the resin
−
Figure 4-9: Variation of the activity coefficients of [Au(CN)2 ] and SCN − in the resin
−
viii
Figure 5-5: The effect of the type of counterion on the rate of ion exchange in acetone–
water mixtures of 5.73 mol% (A); 68.6 mol% (B). ……………………134
Figure 5-6: Capacities of the cation exchange resin Amberlite IR-120 in various pure
solvents (adapted from de Lucas et al., 2001).………………………….136
Figure 5-7: The amount of [Au(CN) 2 ] in the resin in the [Au(CN ) 2 ] / Cl − binary
− −
13.9 mol% (B); 26.7 mol% (C); 68.6 mol% (D) .………………………150
Figure 5-17: Comparisons of the effect of the type of counterion on the ion exchange
isotherms of [Au(CN) 2 ] in DMSO-water of 5.91 mol% (A); 14.3 mol%
−
(B). ……………………………………………………………………...152
Figure 5-19: The effect of type of solvent on the ion exchange isotherms for
[Au(CN ) 2 ]− / Cl − in solvent-water compositions of, 4-6 mol% (A); 13-15
mol% (B); 26-28 mol% (C); 62-70 mol% (D). …………………………153
Figure 5-20: The effect of type of solvent on the ion exchange isotherms for
[Au(CN) 2 ]− / SCN − in solvent-water mixtures of, 4-6 mol% (A); 13-15
mol% (B); 26-28 mol% (C); 62-70 mol% (D). …………………………154
ix
Figure 6-1: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in mixed acetone
−
x
LIST OF TABLES
Table 4-5: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria obtained
at 303 K. ………………………………………………………………….116
Table 4-6: The variation of KAB as a function of the total solution concentration using
the 2-parameter regression method. ………………………………………116
Table 4-7: The variation of KAB as a function of the total solution concentration using
the 3-parameter regression method. ………………………………………117
Table 5-1: Properties of selected solvents (adapted from Marcus, 1997).…………...125
Table 5-2: Compositions of acetone, DMSO and NMP investigated in the study of
binary ion exchange equilibria. …………………………………………...127
Table 5-3: Maximum loadings of Purolite A500 in water and in various mixed
solvents……………………………………………………………………139
xi
Table 6-1: Debye-Hückel parameters for water mixtures of acetone, DMSO and NMP
at 303 K a.…………………………………………………………………159
Table 6-2: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in water
mixtures of acetone at 303 K. …………………………………………….161
Table 6-3: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in water
mixtures of DMSO at 303 K. ……………………………………………..162
Table 6-4: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in water
mixtures of NMP at 303 K. ……………………………………………….162
Table 6-5: The reciprocity constraint applied to ion exchange equilibria in various
mixed solvents.……………………………………………………………168
Table 6-6: Gibbs energies of transfer for selected ions (from Marcus, 1997, unless
otherwise stated).………………………………………………………….177
xii
1. INTRODUCTION...................................................................................... 1
1.1 REFERENCES .................................................................................................5
1
1. INTRODUCTION
The definition for gold extends beyond its physical description. The concurrent role of
gold as an alternative currency or a store of wealth has undeniably superseded its
physical functions. The current state of world affairs exemplifies the role of gold in the
financial world. Since the attacks in New York on September 11, 2001, political unrest
and economic instability have affected many parts of the world. Consequently,
investors have increasingly turned to gold as an insurance against political and fiscal
instability. By December 2004 the price of gold has surged to a 16 year high of USD
430 per ounce (US dollars). Many economists predict a continued increase in the value
of gold as a result of a weaker US currency and due to prevailing political conflicts
throughout the world.
These unprecedented events have swiftly changed the fate of the gold industry. For
example, the net profits of Newmont Mining Corp., one the world's largest miners of
gold, have increased to USD 128.7 million in 2004. This represents a 7-fold increase
from profits gained in 2001 (Salt Lake Tribune, 2004; Journal Sentinel Milwaukee,
2003). Furthermore, gold mining companies have sharply expanded their exploration
and development activities. Existing mines that were abandoned because of low gold
prices in the mid 1990s are slowly reopening in the wake of increased confidence in the
gold market. The positive outlook for gold in the world economy has poised the gold
industry to face a number of challenges that have seriously affected the economic
viability of many mining operations in the recent past.
1
One of the most serious issues affecting the gold industry today is the depletion of high-
grade gold ores. At present, the gold content of ore deposits typically varies from 1-2
g/tonne. The majority of the gold in these ores is associated with other minerals
including sulfide or carbonaceous components. Liberating gold from these ores often
requires complex and expensive pre-treatment procedures in order to make the ore
amenable to subsequent leaching and extraction stages. The processing of copper-gold
ores is a particular problem in Australia. The concentration of copper in these ores is
often of an order of magnitude greater than the concentration of gold.
During leaching, copper minerals and other base metals are dissolved by cyanide. The
resulting increase in the consumption of cyanide further inflates operating costs. The
relatively high concentration of base metal cyanide complexes competes with the gold
cyanide complex during the extraction stage. As a result, the efficiency of gold
extraction is significantly reduced. Furthermore, at the end of extraction cycles, most of
the base metal cyanide complexes are released into tailing dams, exacerbating the
environmental problems associated with cyanide.
The gold industry has come under intense international scrutiny over the utilisation of
cyanide following the tailing spill at Baia Mare, Romania, in January 2000. Some
100,000 m3 of cyanide contaminated tailings were released into the Tisza River system
in central Europe. This incident instigated the development of an international code of
practice to regulate the use of cyanide in the industry. Public upheaval has also caused
some governments to completely ban the use of cyanide. In the state of Montana in the
United States of America, a complete cyanide ban has been in place since 1998. In
Australia, the specification for cyanide discharge into tailing dams is 30 ppm (New
South Wales - EPA). Such strict regulations mean that the treatment of cyanide waste
in many cases is more expensive than recovering gold itself.
The most important technical development in the gold industry in the last 50 years has
been the introduction of the carbon-in-pulp process. However, the dominance of carbon
technology for the recovery of the gold cyanide complex from leach solutions is set to
decline in view of the issues described above. With the prevailing favourable economic
conditions, the gold industry is presented with a window of opportunity to welcome the
2
next technical revolution. In this context, ion exchange has been the most promising
technology to emerge since the introduction of the carbon-in-pulp process.
Outside of the former USSR, the Golden Jubilee mine in South Africa is the only
example where a large-scale ion exchange process has been used for the recovery of the
gold cyanide complex. The reluctance to adopt ion exchange technology elsewhere is
attributed, in part, to the non-selectivity of ion exchange resins. Whilst this feature is
advantageous from the point of view of cyanide management, the separation of gold
from base metals in subsequent elution procedures is a major technical obstacle
hindering the widespread commercialisation of ion exchange resins in the gold industry.
The conventional method for the elution of the gold cyanide complex from the resin
involves contacting the resin with an aqueous solution containing another cyanide
complex, such as the zinc cyanide complex, as the counterion. The problem with this
process is that it leads to the formation of the toxic HCN species during the subsequent
regeneration of the resin, necessitating further processing to eliminate the presence of
this species in waste streams. Simple counterions, such as chloride, are normally
inadequate because a large excess of counterion is required to recover the gold cyanide
complex successfully.
Nonaqueous solvents have been explored in the past as a possible means of facilitating
the elution of the gold cyanide complex from ion exchange resin using simple
counterions. Nonaqueous solvents are known to alter the distribution of ions between
the resin phase and the external solution phase. Whilst the potential of such solvents
has been established, no study has been devoted to understanding the selectivity of the
resin for the gold cyanide complex in mixtures of water and organic solvents (mixed
3
solvents). The principle means of assessing the selectivity of the resin for a given
species is through the measurement of binary ion exchange equilibria. As such, the
main objective of this study was to measure binary ion exchange equilibria involving
the gold cyanide complex in aqueous and mixed solvent systems.
The design and scale-up of ion exchange processes requires reliable thermodynamic
models for the prediction of ion exchange equilibria. The measurement of ion exchange
equilibria provides a way of testing existing models and is integral to the development
of new ones. An additional objective of this work was to extend the application of the
Law of Mass Action (LMA) for the correlation of ion exchange equilibria in mixed
solvents. The main output of this activity is the determination of the equilibrium
constant associated with the ion exchange process in a given binary system.
Prior to the discussion and correlation of the experimental ion exchange equilibria, two
separate reviews are presented in this thesis. The use of ion exchange in gold
processing is reviewed in Chapter 2. This chapter also includes descriptions of the
mineralogy of gold ore deposits as well as current methods used for the leaching and
extraction of gold. Extraction processes used in the gold industry are detailed in a
chronological fashion, even though many of these methods continue to co-exist within
the industry. Of these, carbon technology is dealt with in detail. Finally, this chapter
considers the role of ion exchange in the important area of cyanide management.
Chapter 3 reviews the concept of ion exchange selectivity and discusses specific
characteristics of selectivity in aqueous and nonaqueous systems. A detailed description
of nonaqueous solvents and their impact on the selectivity of resins is included. In
addition, this chapter introduces the quantitative methods of predicting ion exchange
equilibria. The development and historical significance of a number of these theories
are explained. However, it becomes apparent that much of the refined theories of today,
such as the Law of Mass Action, are the result of ideas that have converged from many
contrasting ideas about equilibria. The LMA is discussed in detail as the preferred
means of correlating ion exchange equilibria.
Experimental ion exchange equilibria involving the gold cyanide complex with chloride
and thiocyanate in aqueous solution are presented in Chapter 4. The trends are
4
discussed in terms of the causes of selectivity described in Chapter 3. The experimental
data are modeled using the LMA to determine equilibrium constants for each of the
binary systems. In chapter 5, binary ion exchange equilibria for the gold cyanide
complex with chloride and thiocyanate are obtained in mixed solvent systems. A
qualitative discussion of the selectivity trends is included in this chapter and direct
comparisons are made with the aqueous ion exchange equilibria.
In Chapter 6, the experimental data from Chapter 5 are correlated using the LMA to
obtain equilibrium constants for each of the binary systems in mixed solvents. The
equilibrium constants are compared with the isotherms to establish consistencies in the
observed trends of selectivity. Furthermore, the equilibrium constants are correlated
with properties such as the dielectric constant of the mixed solvent and the Gibbs energy
of transfer associated with the solvation of the ions in mixed solvents.
Chapter 7 summarises the major conclusions of this work, and draws attention to future
work in the area of ion exchange equilibria and the extraction of the gold cyanide
complex.
1.1 REFERENCES
Fleming, C. A., The potential role of anion exchange resins in the gold industry, in EPD
Congress, Proceedings of Sessions and Symposia held at the TMS Annual
Meeting, San Antonio, 1998. 16-19: p. 95-117.
Green, B. R., Kotze, M. H., and Wyethe, J. P., Developments in ion exchange: the
Mintek perspective. JOM, 2002. 54(10): p. 37-43.
Journal Sentinel Milwaukee, Gold Price Jump Sends Newmont Profits Up, October
2003 (online edition).
The Salt Lake Tribune, Earnings Snapshots, October, 2004 (online edition).
5
CHAPTER 2
2. ION EXCHANGE IN GOLD PROCESSING ................................................ 6
2.1 INTRODUCTION .............................................................................................6
2.2 PROPERTIES OF GOLD ..................................................................................6
2.2.1 Gold ore deposits ...................................................................................7
2.3 THE EXTRACTION OF GOLD FROM ORES ....................................................9
2.3.1 Cyanidation ............................................................................................9
2.3.2 Carbon technology ...............................................................................14
2.4 ION EXCHANGE TECHNOLOGY ..................................................................19
2.4.1 Properties of ion exchange resin ..........................................................20
2.4.2 The use of ion exchange resins in the gold industry ............................26
2.5 SELECTIVITY...............................................................................................28
2.5.1 Copper-gold ores..................................................................................32
2.5.2 Loading ................................................................................................34
2.5.3 Elution ..................................................................................................36
2.6 CYANIDE MANAGEMENT ............................................................................42
2.6.1 The AugMENT process .......................................................................44
2.6.2 The Vitrokele process ..........................................................................45
2.6.3 Elutech .................................................................................................45
2.7 REFERENCES ...............................................................................................47
Figure 2-1: Eh (potential)-pH diagram for the Au-CN-H2O system, with initial
concentrations of 0.001 and 0.0001M for CN- and Au respectively............12
Figure 2-2: A comparison of the elution profiles of gold in the Murdoch, AARL and
Zadra processes (adapted from Ruane, 1982). ............Error! Bookmark not
defined.
Figure 2-3: Schematic cross-sectional view of an ion exchange resin bead (A). Fixed
charges (-) are balanced by counter ions (+) diffused throughout the
framework of the resin (B). Typical anion exchange resins have a styrene
divinyl benzene matrix, with a tri-methyl amine functional group (C). ......23
Figure 2-4: A typical RIP or RIC flowsheet for treating gold ores................................29
Figure 2-5: Equilibrium loading for various metal cyanide complexes from a mine leach
solution (adapted from Fleming and Cromberge, 1984a). ...........................42
Figure 2-6: The effect of organic solvent on the elution of gold from resin (adapted
from Law et al., 1985)..................................................................................42
1
2. ION EXCHANGE IN GOLD PROCESSING
2.1 INTRODUCTION
The processing of gold refers to a set of procedures required to liberate gold found in
the Earth’s crust into a form that is pure and marketable. Foremost in this operation is
the mining of the gold-containing earth known as gold ore. Once the ore is mined, the
large bodies of rock undergo a series of mechanical processes known as milling.
Milling operations are aimed at reducing the size of the ore to finer parts amenable for
further processing. The crushed ore is then ready to commence a series of
hydrometallurgical processes. These are a set of chemical processes that involve
dissolving the gold from the ore, extracting the dissolved gold to a concentrated form,
and finally recovering a pure gold product. This chapter reviews these
hydrometallurgical processes with an emphasis on the extraction of gold using ion
exchange resin.
6
Table 2-1: Physical properties of gold (Hermann and Johns, 1993)
dilute solutions and at temperatures up to its boiling point (Hermann and Johns, 1993).
Gold can be dissolved in relatively mild oxidising conditions with certain complexing
agents. A typical example is the dissolution of gold in cyanide in the presence of
[ ]−
oxygen forming the highly stable gold cyanide complex, Au(CN) 2 . Gold complexes
are stable owing to the relatively high electronegativity of gold. While gold can be
found complexed to a limited number of ligands, it exists mostly in its native state.
The unique physical and chemical properties of gold are attributed in theory to the
electronic configuration of the gold atom (Cotton, 1997). For atoms of high atomic
number, such as the lanthanide series of elements, the s shell approaches the nucleus
more closely than would otherwise be expected. The resulting contraction is known as
the relativistic effect. Of all elements, this effect is most pronounced for gold and leads
to the formation of stronger and shorter covalent bonds (Schmidbaur, 1992). The
relativistic effect in gold is likely to be responsible for a range of physical and chemical
properties. These include a higher than expected boiling point and density as well as
high ionization energies and electron affinity (Cotton, 1997).
7
the ore in downstream processes such as concentration and refining. Based on mineral
complexity, the two fundamental classes of gold ore deposits are simple ores and
refractory ores. Knowledge of these ore deposits is necessary for the design of process
schemes for the extraction of gold.
Placer and free milling ores can be classified as simple gold ore deposits. Simple ores
characteristically require minimal effort to liberate the gold from the ore body. Placer
ores are the easiest to recognise in the category of simple ores due to their clearly visible
deposits of gold enriched veins (La Brooy et al., 1994). These ores usually require
minimal chemical treatment as the gold in its native form is partially liberated. Due to
the simplicity of the mineral make up of the ore, other metal values are minimal if not
non-existent. Since the 20th century, placer ores have become significantly depleted in
abundance and only contribute between 2% and 5% of the total annual world production
of gold (Marsden and House, 1992). Free milling or oxide ores require more intense
treatment than placer ores to liberate gold. Cyanide can extract 95% of the gold from
free milling ore when the ore is ground to a size of 80% of the particles sized under 75
μm. These ores are relatively free of base metals and cyanide consuming materials and
hence provide easy and economic recovery of gold (La Brooy et al., 1994). Currently,
free milling ores are the largest contributor to gold production. However, these reserves
are likely to be exhausted rapidly leaving mining operations to deal with more complex
ore bodies.
Ores that do not provide economic gold recovery with standard cyanide concentrations
are classified as refractory ores (La Brooy et al., 1994). Refractory ores are
characterised by a complex assortment of minerals, which complicate subsequent
processing options for these ores. Refractory ores can be further divided into classes
based on reagent consumption. Cyanide consuming ores require high levels of cyanide
due to the reactions of some sulfidic and oxidic ores containing a range of sulfur
minerals such as pyrrhotite (Fe (1- x) S) , mascasite (FeS 2 ) , covelite (CuS) , chalcocite
(Cu 2 S) and sulfides of arsenic, and antimony (Marsden and House, 1992). The high
abundance of cyanide consuming copper minerals is a significant problem in Australia.
In these ores, copper is usually leached at a higher rate than gold, and the highly soluble
copper cyanide complexes hinder the subsequent extraction of the gold cyanide
8
complex (Tran et al., 1997). Oxygen consuming ores contain reactive sulfides, such as
pyrrhotite which can lead to high oxygen requirements, as Fe(II) is oxidised to Fe(III)
and sulfides are oxidized to sulfate. Oxygen requirements are met by the addition of
pure oxygen, hydrogen peroxide, or calcium peroxide. Finally, preg-robbing ores are
natural carbonaceous materials in the ore that can adsorb the gold cyanide complex in
downstream processes such as leaching (La Brooy et al., 1994). Adsorption of the gold
cyanide complex by these materials leads to a reduction in the gold concentration in the
leach liquor. Generally, refractory ores require an additional set of treatment processes
to make the ore constituents more amenable during the leaching stage. These pre-
treatment options include aeration, roasting, pressure oxidation, biological oxidation,
chlorination and nitric acid based processes (Marsden and House, 1992).
2.3.1 Cyanidation
Gold deposited in low-grade ores is either in a fine powder form distributed throughout
the ore body or refractory in nature where the complex mineralogy of the ore hinders its
liberation by physical means. In such cases it is necessary to convert the gold locked in
the ore body to a more soluble form. The process of dissolving gold from its ore is
9
known as leaching. The use of cyanide to dissolve gold in ore bodies is known as
cyanidation. The ability of cyanide to dissolve gold was recognized as early as 1783 by
Scheele. The cyanidation process as it is known today was patented by MacArthur and
the Forest brothers in 1888. Cyanide, as the universal lixiviant for gold has dominated
the gold industry in the twentieth century and remains undisputed to the present day.
In 1846, Elsner first described the reaction mechanism for the formation of the gold
cyanide complex (Marsden and House, 1992). More recently, a number of studies have
improved the understanding of the dissolution reaction of gold (Cathro and Walkley,
1961; Nicol, 1980; Covenjo and Spottiswood, 1984). Essentially, the reaction involves
the oxidation of elemental gold to the Au(I) oxidation state by the formation of the gold
cyanide complex, [Au(CN) 2 ] , accompanied by the reduction of oxygen. Sodium
−
cyanide is the most commonly used salt in the cyanidation process, as it readily
dissolves in water. The series of anodic and cathodic reactions involved in the
dissolution of gold in aerated cyanide is summarized in the following set of equations
(Nicol, 1980):
Anodic reactions
Au → Au + + e − (2-1)
Au + + CN − → [AuCN]ads (2-2)
Cathode reactions
O 2 + 2H 2 O + 2e − → 2OH − + H 2 O 2 (2-6)
Overall reaction
4Au + 8CN − + O 2 + 2H 2 O ⇔ 4Au(CN) 2− + 4OH − (2-8)
10
The successful use of cyanide for dissolving gold is largely attributed to the stability of
Due to the well known environmental problems caused by cyanide, a number of other
ligands such as thiourea (Syna and Valix, 2003; Tatarua, 1973), thiosulfate (Nicol and
O’Malley, 2002), and thiocyanate (Wan et al., 2003) have been considered. In contrast
to cyanide, the gold complexes produced using these alternative leaching agents are
considerably less stable than the gold cyanide complex (Table 2-2). Thiocyanate
provides comparable gold recoveries to that of cyanide, at the expense of relatively high
reagent consumption (Wan et al., 2003). Thiosulfate has been tested rigorously in
laboratories and pilot plants. An advantage of the thiosulfate system is that it does not
leach copper from any of its common mineral forms. Therefore, the leaching reaction is
particularly selective for the gold cyanide complex. (Nicol and O’Malley, 2002).
The process stream that enters the leaching phase is known as a slurry or a pulp. A
slurry is a viscous liquid mixture of water and fine ore particles with a total solid
content between 35% and 50%. Initially, the pH of the slurry is adjusted to suit
operating conditions to a range between 9.5 and 11.5. Cyanide is introduced into the air
agitated tanks, and the concentrations are monitored frequently to maximise the
dissolution of gold. The total residence time of the slurry in the leaching phase depends
on the number of stages and the efficiency of each stage. The leach solution then
proceeds to an extraction stage, where the gold cyanide complex is concentrated and
separated from other ore constituents. The Merill-Crowe process was one of the earliest
extraction procedures to be coupled to cyanidation, increasing gold recoveries from
70% to 95%.
11
Figure 2-1: Eh (potential)-pH diagram for the Au-CN-H2O system, with initial
concentrations of 0.001 and 0.0001M for CN- and Au respectively.
3
Reduction potential (volts)
Au(OH)3
2
Au(Cn)2-
Au
-1
0 2 4 6 8 10 12 14
pH
Table 2-2: Stability constants for Au(I) complexes with selected ligands (Marsden and
House, 1992)
CN − 2 × 1038
S2 O 32− 5 × 1028
Cl − 109
Br − 1012
I− 4 × 1019
CS(NH 2 ) +2 2 × 1023
12
2.3.1.1 The Merill-Crowe process
From the early 1900s until approximately 1970, the Merill-Crowe process was the
method of choice for recovering gold cyanide complex from leached ore. The Merill-
Crowe process generally involves the precipitation of gold from the leach solution using
zinc. A clear solution is required in order for precipitation to take place. As such, leach
pulps undergo a solid-liquid separation process, followed by further clarification to
produce an acceptable feed for precipitation. The zinc cementation reaction is sensitive
to oxygen concentration, pH and cyanide concentration (Nicol et al., 1979; Tran, 1992).
These parameters are usually optimized and the reaction usually takes place in a
vacuum tower. Zinc is added in the form of chips or dust along with lead nitrate to
catalyse the reaction. The cementation reaction of gold with zinc can be represented by
the following equation,
The simplicity and the relative efficiency with which the Merill-Crowe process is able
to recover the gold cyanide complex are its advantages. The major drawback of the
cementation process is its sensitivity to impurities present in the clarified liquor (Muir,
1982). High cyanide consumption by base metals present in the liquor can prevent the
dissolution of zinc. Impurities such as arsenates and sulfide cause the formation of
films on the zinc surface preventing further cementation. In addition, ores with a high
content of clay complicate the filtration and clarification process, usually resulting in
the loss of soluble gold. Due to the problems associated with treating ores with high
impurities, the Merill-Crowe process was only economic when treating high-grade gold
ores. By the 1970’s the carbon-in-pulp (CIP) process was gaining momentum as an
alternative to conventional zinc cementation practices.
13
2.3.2 Carbon technology
The ability of carbon to adsorb cyanide and chloride complexes of gold from leached
solutions was first realized in Australia as early as 1880 (McDougall and Hancock,
1981). At the time, however, there were no incentives to further establish carbon as a
viable technology, due to the acceptance of the Merill-Crowe process as a simple and
economical solution for the extraction of the gold cyanide complex. It was not until the
mid 1950s that carbon technology gained worldwide attention with the inception of the
first carbon-in-pulp (CIP) plant at the Homestake mine in the United States of America.
Furthermore, the Merill-Crowe process could not handle low-grade ores, which were
increasingly discovered during this time. Cyanidation coupled with carbon in the CIP
process allowed exploration of previously untouched gold deposits.
2.3.2.1 Adsorption
Carbon has a maximum capacity for the gold cyanide complex of 7% by mass
(McDougall and Hancock, 1981). Typically, gold loading of 10000-12000 ppm (parts
per million) are obtained from plant leach solutions (Muir, 1982). The most important
factors that affect the rate of loading include mixing efficiency, and particle size of
carbon granules (Fleming and Nicol, 1984; Fleming 1982). On the other hand, the
equilibrium loading of carbon is affected by the ionic strength, pH, temperature,
(Fleming and Nicol, 1984) and oxygen content (Van Der Merwe and Van Deventer,
1988; Tsuchida and Muir, 1986).
14
The mechanism by which the gold cyanide complex is adsorbed onto carbon remains
unclear. This is mostly due to the poor knowledge of the properties of the carbon
material itself. From the extensive work done on the subject, ion-pair formation appears
[
to be the most plausible mechanism. An ion-pair in the form M+-- Au(CN)2 ]−
can be
adsorbed onto the carbon surface, where M+ represents any spectator cation in the leach
solution such as Na+ and Ca2+ (Davidson, 1974). McDougall and Hancock (1981) have
proposed a two-step process in which the partial reduction of the gold cyanide complex
follows the initial ion-pair adsorption. Furthermore, X-ray Spectrophotometric studies
(XPS) have revealed that the graphite-like structure of carbon, with its delocalised
electrons, contributes to the adsorption of the gold cyanide complex (Klauber, 1991). It
has been proposed that oxygen plays a pivotal role in the adsorption process by creating
charged moieties on the carbon surface (Tsuchida and Muir, 1986; Van Der Merwe and
Van Deventer, 1988). However, this has been disputed by studies, using techniques
such as XPS, in which oxygen is said to play a minor if not negligible role (Ibrado and
Fuerstenau, 1992). The inability to deduce a definitive mechanism for the adsorption
process on carbon is due in part to a lack of understanding of the nature of carbon itself,
which is not amenable to direct analytical techniques.
2.3.2.2 Elution
The elution procedures most commonly employed for carbon are the Zadra and Anglo
American Research Laboratory (AARL) processes. In the Zadra process, a solution of
NaOH and NaCN between 95-100˚C is circulated through a column containing loaded
carbon (Zadra et al., 1952). In order to achieve shorter elution cycles, the Zadra process
is operated at pressures between 500-600 KPa and temperatures of up to 140˚C (Ruane,
1982). The AARL process involves pretreatment of the loaded carbon with a solution
of NaCN and NaOH, followed by elution with pure water (Davidson and Veronese,
1979; Davidson and Duncanson, 1977). Both processes use an electrowinning circuit to
reduce the gold cyanide complex to metallic gold. An advantage of the AARL
procedure is that elution is complete within 8-10 hours, whereas elution using the Zadra
process takes up to 36 hours at atmospheric pressure. A drawback of the AARL
procedure is its sensitivity to water quality. Both processes are costly due to the energy
input required to maintain high pressures and temperatures for efficient gold cyanide
recovery (Ruane, 1982).
15
An alternative to the conventional Zadra and AARL elution of the gold cyanide
complex involves the use of organic solvents. The earliest success at using nonaqueous
solvents for the elution of the gold cyanide complex from activated carbon was the work
by Heinen, et al. (1976). In this work, ethanol (20% by volume) was added to the
conventional Zadra eluent to facilitate the displacement of gold and silver cyanide
complexes from resin. A refined version of this scheme was later commercialized as
the Duval process in the United States.
Since then, many variations of the elution procedure for carbon involving organic
solvents have been studied. For example, in Australia two notable processes are the
Micron Solvent Distillation and the Murdoch processes. Micron Solvent Distillation,
involves the use of the vapour condensate of an organic solvent to elute the gold
cyanide complex (Muir et al., 1985). In this process, the gold-loaded carbon is initially
presoaked in a mixture containing 10% NaCN and 2% NaOH solutions. Hot methanol
vapours are then refluxed and recycled through the carbon bed. This procedure offers
the advantages of the economical use of solvent, as well as producing highly
concentrated gold solutions suitable for subsequent recovery using precipitation or
electrowinning. The Murdoch process uses acetonitrile in a mixture of sodium cyanide
and sodium hydroxide at ambient temperature to elute gold from carbon (Ruane, 1982).
Figure 2-2 compares the performance of the Murdoch process with those of the Zadra
and AARL elution procedures. The sharp elution profile for gold using the Murdoch
process demonstrates that concentrated gold solutions are obtained with the addition of
organic solvent compared to the conventional elution methods.
Muir (1982) concluded that the effect of an organic solvent varies widely according to
the structure and polarity of the solvent. Solvents are most effective in the order:
ketones> higher alcohols> dipolar aprotics> simple alcohols >> water. The success of
organic solvents in eluting the gold cyanide complex was attributed to their effect on the
activity of the counterions, to an extent that causes the reversal in selectivity of carbon
for [Au(CN) 2 ] (Muir, 1987, Tsuchida, et al., 1984).
−
From these studies, it was
suggested that, for example, on transfer from water to organic mixtures, the activity
coefficient of CN - increases and that of [Au(CN) 2 ] decreases. Thus, with cyanide as
−
16
Figure 2-2: A comparison of the elution profiles of gold in the Murdoch, AARL and
Zadra processes (adapted from Ruane, 1982).
a counterion, the desorption of the gold cyanide complex from carbon to the solution is
favoured.
Despite the efficiency of organic solvents for eluting the gold cyanide complex from
carbon, the technology has never been used at an industrial scale. A problem identified
at a laboratory level is the absorption of organics solvents onto carbon (Heinen et al.,
1976). The adsorbed organic constituents, diminish the ability of the carbon to further
adsorb gold. For solvents such as acetonitrile, this problem has been overcome by a
stripping and fractionation procedure which quantitatively removes solvent from the
carbon. The gold adsorption activity of the resulting carbon is largely unaffected in
17
subsequent loading cycles. Other major draw backs of organic solvents are the high
cost of reagents and potential likelihood of the reagents being a fire hazard.
The CIP process generally comprises the following unit operations: adsorption, elution,
recovery, and carbon reactivation. Prior to the adsorption stage, the cyanide pulp is
screened to remove coarse material using hydrocyclones (Laxen et al., 1994). Large ore
particles from the pulp are sent back to the leaching tanks. The pulp then moves
through 6-8 adsorption tanks in series. The carbon is introduced to the last tank of the
series and flows counter-current to the pulp. The counter-current movement allows the
most favourable adsorption kinetics in the adsorption tanks with the lowest gold
concentration. The residence time of the pulp in each tank is approximately one hour.
The concentration of the gold cyanide complex on the carbon increases as it moves
counter-current from each adsorption tank. The loaded carbon is removed from the first
tank periodically. The elution stage involves the stripping of the gold cyanide complex
from the carbon, either via the Zadra process, or the AARL process. These elution
procedures are usually coupled to an electrowinning cell to recover the gold.
Alternatively, the gold can also be recovered by cementation onto zinc. The stripped
18
carbon then goes through a thermal reactivation treatment, where adsorbed organic
materials are removed. The reactivated carbon is then recycled back to the adsorption
stage.
Variations of the CIP process are the carbon-in-leach (CIL) process and the carbon-in-
column (CIC) process (Fleming, 1998a). The CIP and CIL processes are applied
directly to leach pulps, whereas the CIC process is applied to clarified liquors. When
preg-robbing constituents are present in the ore the CIL process is used, where
cyanidation and carbon adsorption occur in the same reactor. The presence of carbon in
the leaching stage prevents the preg-robbing constituents from adsorbing gold. A factor
that influenced the prominence of carbon processes was the versatility and robustness of
the process itself. This meant that it could be applied to different mine sites with
different conditions. For example, heap leaching is a technique adopted to treat low-
grade ores, where setting up of a conventional milling circuit is uneconomical. The
process consists of piling ore to a given height on an impermeable bed and spraying
dilute cyanide solution onto the top of the heap. After percolating through the heap, the
leach solution is directed to a collection pond. The pregnant solution is pumped through
carbon columns (CIC) to recover the gold (Fleming, 1998a).
For the past 30 years carbon has been the undisputed choice for the extraction of gold
from ore. From a technical perspective, the biggest drawback of carbon is the complex
elution and regeneration procedures. The problems associated with carbon are further
exacerbated by the gold industry having to deal with the depletion of high-grades of
gold ore, and strict environmental regulations governing mine output. Ion exchange
resins offer the versatility to handle the increasingly diverse demands placed on a
modern gold plant.
19
nuclear reactors, agriculture and pharmaceuticals (Streat, 1999). Lately, ion exchangers
have been considered in the recovery of precious metals from leach liquors (Fleming,
1998b; Lukey et al., 1998).
The basic framework of ion exchange resins consists of an inert matrix with functional
groups attached to the surface. Resins are commonly spherical in nature, however,
fibrous resins have been produced recently that appear to have some advantages
compared to granular beads (Kautzmann et al., 2002). The charge of the functional
group is neutralized by an ion of opposite charge known as the counterion. Ion
exchange resins that displace a positively charged counterion with an ion of the same
charge are known as cation exchange resins. Similarly, anion exchange resins are those
that displace a negatively charged counterion for an anion of equal charge. As anions
are pertinent to this work, the following discussion is mainly focused on anion exchange
resins.
The resin matrix can be formed from a number of polymeric compounds including
styrene, acrylics, phenol-formaldehyde and polyalkamines (de Dardel and Arden, 1993).
Polystyrene-divinylbenzene (DVB) polymers are the most common matrix for resins.
The current method of choice for the preparation of the polymeric resin matrix is via
addition polymerization (Pietryzk, 1990). In this process, styrene is polymerized under
the influence of a catalyst such as benzoyl-peroxide, yielding a linear polystyrene
molecule. Mixing a proportion of DVB with polystyrene causes cross-linking in the
polymer matrix. Usually, resins have an average DVB content of 8%. Once a cross-
linked product is obtained the polymer is treated with chloromethyl ether to form
20
chloromethylated polystyrene. The chlorine in the chloromethyl group can be replaced
by an amine or ammonia depending on the functional group desired for the anion
exchange resin.
Ion exchange resins can be manufactured into one of two physical structures; gel or
macroporous. Macroporous resins are prepared by mixing an organic solvent, which is
soluble in the monomer (styrene and DVB monomers), but poorly soluble in the
polymer. Typical organic solvents used for this purpose are toluene, ethylbenzene and
diethylbenzene (Abrams and Miller, 1997). As polymerization progresses, the organic
solvent is forced out of the growing copolymer regions, forming a network of pathways
throughout the resin bead. Macroporous resins are preferred for the extraction of gold,
as they possess a high surface area for the adsorption of the gold cyanide complex,
whilst providing better mechanical strength for plant processes (Marsden and House,
1992). On the other hand, gel resins are prepared without the addition of any substances
that increase its porosity. As a result the porosity of gel resins varies depending on the
type of exchanging ion and solvent, whereas macroporous resins maintain a permanent
porosity (Abrams and Millar, 1997).
The functional group attached to the resin matrix determines ion exchange behaviour of
the resin. The number of functional groups determines the capacity of ions that the
resin can exchange. An important factor is the base strength of the functional group of
the resin. At high values of pH weak base groups such as –NH3+ lose a proton, forming
uncharged –NH2. Strong base groups such as –N+(CH3)3 remain ionized even at high
pH. The capacities of weak base resins are therefore determined by the solution pH. In
contrast, the capacities of strong base resins are unaffected by solution pH as their
charges are maintained across the pH spectrum. The strong base quaternary ammonium
type structure is the most common type of anion exchange resin. The charge of the
functional group is balanced by counterions such as Cl - , OH - , and SO 24− . Figure 2-3 is
a schematic representation of a resin bead and its chemical structure. The resin is a
strong base anion exchange resin with quaternary ammonium functional group attached
to a polystyrene-divinylbenzene matrix back bone.
21
2.4.1.1 Physical properties of resin
The two most important physical characteristics of anion exchange resins are: (1)
chemical and thermal stability and (2) mechanical strength. Generally, resins degrade
under strongly oxidizing and reducing conditions. For example, oxidants such as
chlorine and chromic acid attack the resin matrix and destroy the cross linking. Strong
base anion exchange resins are comparatively more susceptible to chemical and thermal
degradation than the equivalent cation exchange resins. The quaternary ammonium
group configuration is unstable especially in the hydroxyl ion form (Helfferich, 1995).
At elevated temperatures, the quaternary ammonium group is converted into a tertiary
amine group resulting in the loss of strong base groups.
The mechanical strength of a resin bead is determined by its resistance to osmotic shock
and physical abrasion. The cause of osmotic shock involves the concept of swelling
equilibrium in ion exchange resins (Helfferich, 1995). Solvent molecules surrounding
the ionic polar constituents of the ion exchanger causes the matrix to stretch. On the
other hand, structural properties of the matrix such as cross-linking DVB groups resist
the process of expansion. Therefore, swelling equilibrium results when the elastic
forces of the matrix balance the tendency towards dissolution due to solvation (Pietryzk,
1990).
During the cycles of adsorption and elution the configuration around each functional
group in the resin changes causing abrupt shifts in the swelling equilibrium (Fleming,
1998b). The resin swells and contracts during these cycles as the size of the adsorbed
and displaced ions are different during an exchange reaction. The stress endured by the
beads is called osmotic shock. The osmotic shock causes cracks in the resin beads,
which leads to its eventual degradation. The correct choice of resin and gradual
changes to conditions which the resin is exposed to, can minimize the effect of osmotic
shock. Nonetheless, compared to carbon, resins are more resistant to the potential
damage caused by abrasion from fine ore particles in the pulp, and to damage caused by
moving mechanical parts during the course of the processing cycle (Bolinski and
Shirley, 1996; Hosking, 1984).
22
2.4.1.2 The chemistry of ion exchange
Anion exchange can be described by the following generic equation
β A α(r)− + αB (s)
β−
⇔ βA α(s)− + αB (r)
β−
(2-10)
where the replacement of ion A on the resin phase (r) occurs with ion B from the
solution phase (s). The terms α and β denote valencies of ion A and B respectively.
The ion exchange process as described by equation 2-10 underestimates the complexity
of the ion exchanger itself and the ion exchange process. Firstly, the ion exchange
represented by equation 2-10 can easily be interpreted as a chemical reaction. However,
the heat evolved during an ion exchange process is approximately 2 kcal/mol
(Helfferich, 1995). The rather low heat of reaction suggests that the exchange process is
diffusion controlled without the formation of any chemical bonds. From equation 2-10
it can also be inferred that the functional groups are strongly bound to the counterion of
opposite charge. Contrary to this perception, Helfferich (1995) and Pietzyrk (1990)
both suggest counter ions as being diffuse, freely moving ions that neutralize the
charges of the functional groups. When the resin is placed in an electrolytic solution,
the counterions are free to move into the solution that enters the pores providing
electroneutrality is satisfied i.e. the solution provides simultaneously an equivalent
amount of counterion to replace those that are leaving from the resin.
Figure 2-3: Schematic cross-sectional view of an ion exchange resin bead (A). Fixed
charges (-) are balanced by counter ions (+) diffused throughout the framework of the
resin (B). Typical anion exchange resins have a styrene divinyl benzene matrix, with a
tri-methyl amine functional group (C).
23
-
+
- + -
+
(A) +
-
-
+
Resin Matrix
CH2
Functional +
CH3 N CH3
Group
CH3
(C)
24
The generic ion exchange in equation 2-10 can be used to deduce schemes by which the
gold cyanide complex loads on to anion exchange resin. Strong base and weak base
anion exchange resins have different chemistries by which they undergo ion exchange
with the gold cyanide complex. Strong base resins possess quaternary ammonium
functionality with a permanent positive charge independent of solution pH. The loading
of gold cyanide complex on a strong base resin can be represented by the following
equation,
−
Cl (r) (
+ [Au(CN) 2 ]
−
)
(s)
−
⇔ Cl (s) ([
+ Au(CN) 2 ])
−
(r)
(2-11)
Here, the resin (r) is originally in the form of the chloride anion, and is displaced by the
[Au(CN) ] 2
−
ion from the solution (s).
Weak base resins contain primary, secondary and tertiary amine functional groups or a
mixture of these (Hosking, 1984). Properties of weak base resins are governed by the
solution pH. In an acidic solution, weak base resins undergo the following reaction,
In equation 2-12, the R groups represent alkyl chains attached to the functional group of
the resin matrix. In acidic solution, the equilibrium lies to the right hand side of
equation 2-12 resulting in the protonation the functional group. Hence, at low pH the
resin behaves similar to a strong base resin (Fleming and Cromberge, 1984a). The
protonated functional groups of the weak base resin can then undergo ion exchange with
[Au(CN) ] 2
−
according to,
(R +
2 HX - )
(r)
([
+ Au(CN) 2 ])
−
(s )
( [
⇔ R +2 H Au(CN) 2 ])
−
(r)
+ X − (s ) (2-13)
Due to the dependence of the above reaction on solution pH, the pKa of the functional
group determines the nature of a weak base resin (Fleming and Cromberge, 1984a).
The pKa of a substance is indicative of the degree to which an acid dissociates. For
most commercial weak base resins this value lies between pH 6 and 8. However, in
25
cyanide leach liquors with a pH between 10 and 11 the equilibrium in equation 2-12 lies
predominantly to the left hand side. As a consequence, functional groups are mostly in
the unprotonated free base form and are unable to undergo ion exchange with
[Au(CN) ] . Usually, the solution pH is adjusted to match the pKa value of the resin.
2
−
Outside the Soviet Union, research conducted in South Africa demonstrated the
potential of the resin-in-pulp process, provided that the concentration of gold in the
cyanide liquor was reasonably high and the concentrations of base metals were low
(Davison et al., 1961). This work led to the commissioning of the first RIP plant in the
Western World at the Golden Jubilee Mine in South Africa in 1988. The carbon-in-
leach (CIL) process originally in operation at the site was beleaguered by a high clay
and organic content of the ore as well as metallurgical problems associated with carbon
reactivation. The plant could only achieve gold loading of approximately 1000 g/t and
the overall gold recovery was poor (60-70%). The problem was exacerbated by high
elution and reactivation costs, as well as carbon replacement costs.
26
Initially, the RIP plant at Golden Jubilee used thiourea to remove metal cyanide
complexes from the resin. However, the acidic reagent caused the precipitation of base
metal salts, (Zn2[Fe(CN)6] and Fe4[Fe(CN)6]3 ) into the resin pores (Fleming and
Seymore, 1990). As a result, the solution used for elution was changed to alkaline zinc
cyanide, and all metal cyanide complexes were eluted successfully from the resin. After
the inception of RIP plant at Golden Jubilee, the mill capacity was almost doubled and
overall gold recovery improved from 65 to over 85% (Fleming and Seymore, 1990). A
schematic diagram of a RIP process based on the Golden Jubilee RIP plant is presented
in Figure 2-4.
Experiences at the Golden Jubilee mine and numerous laboratory investigations have
demonstrated the many advantages of the application of ion exchange resin over
conventional carbon technology (Fleming, 1998b; Fleming, 1982; Fleming and
Cromberge, 1984a). Among these are fast loading rates, potentially higher loading
capacities, the lack of need for thermal activation, and a versatility provided by the
synthetic substrate. Resins are also less affected by the presence of naturally occurring
carbonaceous matter or organic reagents compared to carbon (Petersen and Van
Deventer, 1997). The ability of resins to tolerate significant amounts of calcium salts
and organic matter provides a wider scope for their application.
Resins can be used for cyanide solutions emanating from the heap leaching of low-
grade gold ores, the leaching of calcines and floating tailings (Riveros and Cooper,
1987). Synthetic resins are more versatile than carbon as the functional groups can be
manipulated during synthesis to alter their sorptive properties. Resins do not require
periodic re-activation treatment or high temperature stripping which are standard
operations with activated carbon. Although there are insufficient examples from
industry, the experience at the Golden Jubilee mine demonstrated that significant capital
cost savings can be realized by using ion exchange in place of the conventional carbon
process (Green et al., 2002; Satalic et al., 1996; Fleming and Seymore., 1990).
Apart from the Golden Jubilee mine in South Africa, the gold-mine operators have been
reluctant to adopt ion exchange at an industrial scale. Information is limited on the
engineering complexities associated with resins exposed to pulps containing ore
particles. The low density of resins compared to carbon causes them to accumulate at
27
the surface of the pulp. Continuous agitation is required to ensure that the resin is
properly contacted with the leached pulp. No quantitative data are available on the
effect of adsorption and elution cycles on matrix degradation. A summary of the major
advantages and disadvantages of the Merill-Crowe process, carbon technology and ion
exchange technology are presented in Table 2-3.
If ion exchange is to be adopted by the gold industry, it needs to meet the demand of a
critical problem facing the industry. Complex ores with high levels of cyanide soluble
minerals are the cause of two significant issues facing the gold industry. Firstly, the
leaching of base metals from the ore depletes the cyanide available for gold dissolution.
Therefore, increased cyanide concentrations are required to maintain the rate of gold
dissolution in the presence of base metals. As a consequence, mine operations are faced
with higher reagent costs, as well as increased levels of cyanide discharged into tailing
dams. Secondly, the presence of base metal cyanide complexes has a detrimental effect
on the extraction of the gold cyanide complex using resin. Base metal cyanide
complexes reduce the selectivity of the resin for the gold cyanide complex. The
following section is a discussion of the impact of cyanide soluble minerals during the
gold leaching process and its impact on the selectivity of resins for the gold cyanide
complex.
2.5 SELECTIVITY
Ores in which gold is not amenable to conventional cyanidation are the complex ores
and highly refractory ores. As discussed in section 2.1.1, gold is usually found locked
within minerals in highly refractory ores preventing contact with the leach reagent.
Complex ores on the other hand possess reactive minerals that consume cyanide and
oxygen and therefore deplete the concentrations of these reagents for the dissolution of
gold. These minerals that have the capacity to consume cyanide are known as
cyanicides. Cyanicides include cyanide soluble salts such as sulfates and arsenates,
ferrous salts and oxidation compounds of copper, zinc, arsenic, antimony and sulfur
compounds that react to form thiocyanate. The abundance of gold in these ores is
significantly lower than the concentrations of cyanicides.
28
Figure 2-4: A typical RIP or RIC flowsheet for treating gold ores.
Ore
Loading
Resin-in- Loaded Resin-in-
pulp Resins column
Elution/ Regeneration
Regeneration Elution
Zn(SO)4 HCN
[Zn(CN)4]2-
Caustic
Recovery Bath
NaCN Electrowinning
Gold bullion
29
Table 2-3: A comparison of the advantages and disadvantages of the Merill-Crowe
process, carbon technology and ion exchange technology (adapted from Muir, 1982 and
Fleming, 1998b).
Advantages Disadvantages
Merill-Crowe • Efficient • Sensitive to pH and
(zinc • Simple cyanide concentration.
cementation) • Inexpensive • Sensitive to oxygen
concentration
• Sensitive to S2-, As, Sb,
• Non-selective
• Labour intensive
• Requires filtration and
clarification steps.
Carbon • Efficient • Partially selective
Technology • Simple • Elution requires high
• Cheap provided that carbon is temperatures and pressures
recycled. • Carbon requires
• Suitable for low-grade ores. periodic reactivation
• Can operate in pulp
Ion Exchange • Superior kinetics and • Physical strength of
Technology equilibrium capacities. resins in industrial
• Elution at ambient temperatures situations not understood.
• Does not require reactivation • Low density of resins
• Not effected by organic foulants. causes mixing problems.
• Other metal cyanides can be co- • More expensive than
extracted to produce cleaner effluent carbon
streams. • Lack of complete gold
• Simple stripping procedures selectivity.
• Can be tailor made during
synthesis
• Can operate in pulp
30
Sulfides are one of the most common mineral groups that readily decompose in
solutions of alkaline cyanide in the presence of oxygen. For example, the general
oxidation reaction for a sulfide containing a divalent metal cation, MS, is given by
(Marsden and House, 1992),
Metal cyanide complexes formed from the oxidation reactions of iron, zinc and copper
[Cu(CN) ]3
2−
(Hedley and Tabachnick, 1968). The type and proportions of base metal
cyanides present in the leach liquor are important considerations in designing gold
processing schemes. Pyrrhotite is the most reactive iron mineral in alkaline cyanide
solutions that produce the hexavalent iron cyanide complex. Zinc containing minerals
are less common in gold ores. The behaviour of the tetravalent zinc cyanide complex is
of considerable interest in gold operations using the Merill-Crowe process or anion
exchange resin for the extraction of the gold cyanide complex. Zinc in its native and
complexed forms are used in the Merill-Crowe process and during elution of anion
exchange resin. Metal cyanides, Ni(CN) 24− and Co(CN) 36− formed from the oxidation of
minerals such as pentlandite and cobaltite, respectively, are present in leach solutions to
a lesser extent than other common base metal cyanides. Copper minerals that form
complexes such as Cu(CN) 32− are an important class of cyanicides and have caused
significant problems in the gold extraction process.
The presence of copper minerals in gold ore bodies is a major challenge to the gold
industry and its operators. The problem is especially important in Australia where
copper-gold ores are frequently mined. Two of the most recent copper-gold deposits to
be discovered in Australia include Red Dome in Queensland and Rothsay in Western
Australia. The Red Dome deposit was complicated by varying amounts of copper
mineral present throughout the primary and oxide ore zones (Ryan, 1991). On the other
hand, only 0.15% of the copper present in the Rothsay ore was cyanide soluble
(Keating, 1991). However, it is generally accepted that ores with a copper content
greater than 0.1% are uneconomical to process by conventional cyanidation.
31
2.5.1 Copper-gold ores
The dissolution of copper minerals in alkaline cyanide solution varies depending on the
type of mineral. For example, minerals such as chalcocite, bornite, enargite, cuprite,
malachite and azurite completely dissolve copper, where as chrysocolla and
chalcopyrite only dissolve up to 10% copper. Cu(I) is the most dominant oxidation
state of copper found in leach liquors. Any Cu(II) from the ore is found to be converted
to Cu(I) in the presence of cyanide, whilst cyanide is oxidized to cyanate (Stramboliades
et al., 1978). The chemistry of Cu(I) complexes has been investigated extensively
(Hedley and Kentro, 1945; Hedley and Tabachnick, 1968). The copper cyanide
complexes that are formed in the presence of excess free cyanide are summarized by the
following set of equations (Jay, 2001):
CuCN + CN − ⇔ [Cu(CN) 2 ]
−
(2-16)
An Eh-pH diagram for the Cu-CN-H2O system shows the large area of predominance of
the Cu(CN) 32− under gold dissolution conditions (Osseo-Asare et al., 1984). However,
the stability constants of the three complexes are quite close in value, and as such all
three complexes are present in solution to some extent (Hefter et al., 1993).
The predominant copper cyanide complexes formed and their distributions are governed
by factors including the concentration of free cyanide in solution, the ionic strength of
solution and the pH. As the free cyanide concentration increases the tricyano complex
forms, and further increases in free cyanide concentration leads to the formation of the
[
tetracyano complex, Cu(CN) 4 ]
3−
. These higher copper complexes are also favoured in
32
Hefter and co-workers (1993) showed that higher copper cyanide complexes are more
stable in solution containing 1M NaCl. The work was supported by Lukey et al., (1999)
[
who demonstrated that Cu(CN) 4 ]
3−
was the predominant species in saline waters
containing 0.004M free cyanide at pH 8. Furthermore, the formation constant for
[Cu(CN) ]4
3−
is increased in 1M NaCl (k=102.38) compared with non-saline solution
(k=101.1). The presence of the higher copper cyanide complexes Cu(CN) 3 [ ]2−
and
[Cu(CN) ]4
3−
is advantageous in the CIP and RIP processes as these species are not
readily adsorbed onto either carbon or resin and hence gold is selectively adsorbed. The
maintenance of these complexes requires a high concentration of cyanide, which is
undesirable as the subsequent levels of cyanide in effluent streams also increase.
Copper cyanide cannot dissolve gold, therefore a sufficient amount of cyanide must be
present to dissolve gold as well as the copper present in the ore. If the ore contains a
significant amount of soluble copper, free cyanide used to leach gold is significantly
depleted. When no copper is present in the ore, maximum gold dissolution is achieved
with a free cyanide concentration of approximately 2.2 mM. However, copper-gold
ores require 5 moles of cyanide per mole of copper present in the ore to maintain the
dissolution of gold at an acceptable level (Hedley and Tabachnick, 1968). Therefore a
high ratio of CN to Cu is maintained in order to reduce the losses of gold, and to
maintain higher complexes of copper cyanide that would compete less favourably with
the gold cyanide complex in subsequent extraction processes.
The concentrations of base metal cyanide complexes in the leach liquor are significantly
higher than the concentration of the gold cyanide complex. Base metal cyanide
complexes therefore compete strongly with the gold cyanide complex for resin sites and
the capacity of the resin for gold decreases dramatically. Furthermore, the affinity of
33
2.5.2 Loading
Anion exchange resins generally have a high affinity for the gold cyanide complex
(Fleming and Cromberge, 1984a; Hosking, 1984). Anion exchange resins can load up
to 600 grams of gold per kilogram of resin (Hosking, 1984; Burstall et al., 1953). This
is significantly higher than the capacities for metal cyanide complexes such as silver
and copper with loadings of 341 and 82 g/ kg resin respectively (Burstall, et al., 1953).
However, in the presence of base metal cyanides, the loading capacity of anion
exchange resin for gold decreases to less than 10 g/kg resin (Bolinski and Shirley, 1996;
Davison et al., 1961). Figure 2-5 demonstrates the effect of competing ions on capacity
of the resin for the gold cyanide complex. Compared to strong base resins, the capacity
of weak base resins for the gold cyanide complex is almost one quarter of that of strong
base resins. The loading of the gold cyanide complex on weak bases resins is mostly
affected by the pKa of the resin, the type of resin and the strong base content of the
resin (Fleming and Cromberge, 1984a; Palmer, 1986).
Resins have a number of features that can be tailored to achieve characteristics selective
for the adsorption of gold cyanide complex. Substituents, with various alkyl chain
lengths, can be attached to functional groups. Riveros (1993) studied the effect
substituents on the selective extraction of the gold cyanide complex in the presence of
competing metal cyanide complexes. The study found that the larger and longer tri
ethyl groups favoured the loading of the gold cyanide complex compared to those resins
with trimethyl substituents in their functional groups. Multivalent complexes such as
[Fe(CN) ]
6
4−
[
and Cu(CN) 4 ]
3−
require multiple active sites in close proximity to balance
their charges. This is hindered in the case of the triethyl resin whose bulky functional
groups reduce the accessibility of adjacent sites on the resin. However, mono and
bivalent complexes such as those of Au, Zn and Ni are less affected by steric
constraints, as they only require one or two adjacent sites to satisfy their charge
requirements.
The gold cyanide complex is classed as relatively hydrophobic compared to other metal
cyanides (Riveros, 1993; Hosking, 1984). Therefore, the gold cyanide complex has a
strong affinity to resins that are also relatively hydrophobic compared to the external
solution. The hydrophilicity of the resin matrix is determined by the structural
34
composition of the resin matrix. In the study by Riveros (1993), the weak base resins
Dowex WGR and WGR-2 did not display the selective attributes towards the gold
cyanide complex associated with normal polystyrene resins. Dowex WGR and WGR-2
are condensation resins that do not have the typical structure of polystyrene weak base
resins. The structure of a condensation resin is shown below.
The hydroxyl group increases the hydrophilicity of the resin matrix. Therefore,
condensation resins present a less attractive environment for the relatively hydrophobic
gold cyanide complex. Other strong base resins known for their hydrophilicity are
Amberlite IRA-458 (Riveros, 1993) and Purolite A860S (Leao and Ciminelli, 2000).
Both resins are polyacrylic resins with a carbonyl group and secondary amine groups
which are also less selective for gold than polystyrene resins.
Weak base resins are known to be more selective for the gold cyanide complex than
strong base resins. Fleming and Cromberge (1984a) compared the loading of strong
base and weak base resins exposed to a mixture of metal cyanides. The distribution
coefficient for metal cyanides were determined for strong and weak base resin at two
different pH values. For the weak base resin, gold and silver cyanide distributions were
similar at both pH values. However, the cyanide complexes of Co, Cu, Ni and Fe were
adsorbed 10-20 times less at high pH values. The strong base resin showed no
selectivity for gold cyanide at either pH. Riveros (1993) explained this observation in
terms of the ionic density of the resin. Ionic density refers to the number of active sites
per unit volume of resins (Riveros 1993). Low ionic density resins have widely spaced
functional groups. In alkaline solution, weak base resin functional groups exist in the
free base form. In these cases, the strong base content determines the capacity of the
resin. As a result, weak base resin can be viewed as strong base resins of low ionic
density. The low ionic density of weak base resins at a high pH, enables it to selectively
extract gold cyanide complex.
Researchers have attempted to manipulate these properties to obtain a resin selective for
the gold cyanide complex. Ideal properties required of a selective resin are fast kinetics,
35
high equilibrium loading capacity, and compatibility with a simple and effective elution
procedure. The only commercial resin produced thus far claiming selectivity for the
gold cyanide complex is Minix®. Minix® is a strong base resin developed by
researchers in South Africa, and has been tested at laboratory and pilot plant scales
(Green et al., 2002). According to laboratory tests the resin capacity for gold was
approximately 36 g/kg resin after multiple adsorption cycles. The loading values for
cyanide complexes of zinc, nickel and copper on Minix were 10, 7.3 and 1.2 g/kg resin,
respectively. Elution was carried out with an acid thiourea solution followed by an
electrowinning circuit, resulting in 99% of the gold recovered during a complete cycle
(Conradie et al., 1995).
2.5.3 Elution
For some advocates of ion exchange, the non-selective loading properties of resins are
seen as an advantage. Co-extraction of base metal cyanide complexes with the gold
cyanide complex decreases the levels of total cyanide in the tailings. This aspect of
resins is discussed further in Section 2.6. However, after non-selective adsorption, the
gold cyanide complex must be separated from other base metal cyanide complexes. To
date, most industrial elution circuits rely on the initial elution of base metal cyanide
complexes followed by the elution of the gold cyanide complex.
In the elution sequence used in the former U.S.S.R, some or all of the base metals are
initially eluted with an acid such as sulfuric acid (Bolinski and Shirley, 1996).
However, some gold cyanide complex is also known to be lost during the elution of
base metal cyanides. Furthermore, the use of an acid eluent causes the precipitation of
salts such as copper cyanide, which decreases the capacity of the resin over time. As
the gold cyanide complex is bound strongly to the resin, elution requires severe
conditions or strong reagents. To date, most industrial elution circuits are based on one
of two methods: eluting the gold cyanide complex by the addition of an anion more
strongly bound to the resin than gold cyanide complex, or via the chemical destruction
of the gold cyanide complex.
A number of anion complexes can displace the gold cyanide complex from resin if
sufficiently high concentrations are utilized in eluent solutions. Of these, the zinc
36
[
cyanide complex, Zn(CN)4 ]2−
is the most universal counterion used in the elution of
gold cyanide complex from resin (Fleming, 1998b; Riveros et al., 1993; Wan and
[
Miller, 1990). The Zn(CN) 4 ]
2−
is prepared from ZnO and NaCN. Initially, base
[
metals are selectively eluted from the loaded resin with a cold, dilute Zn(CN) 4 ]2−
solution, which is then returned to the loading system (Riveros et al., 1993; Tran et al.,
2000). Meanwhile, the concentration of gold on the resin is allowed to build up.
Periodically, a bleed of loaded resin is subjected to a complete elution cycle using a
more concentrated and warm zinc cyanide solution, in order to elute the gold cyanide
complex. Treatment of the resin with sulfuric acid regenerates the resin to the sulfate
form whist producing zinc sulfate and HCN. Zinc cyanide was the chosen method of
[Au(CN) ]2
−
elution at the Golden Jubilee mine in South Africa (Fleming and
Cromberge, 1984b).
The thiocyanate complex is also known to compete favourably with the gold cyanide
complex on resin and has thus been under constant investigation as an alternative eluent
to the zinc cyanide complex (Wan et al., 2003; Stamboliadis et al., 1978). A
shortcoming of thiocyanate elution is its lack of selectivity towards the gold cyanide
complex during elution. Thiocyanate causes the simultaneous elution of base metal
cyanides with the gold cyanide complex (Burstall and Wells 1955). Furthermore,
thiocyanate is not used in practice because of the complexity of the regeneration process
(Fleming, 1998b). Following elution, the resin requires regeneration and this is mostly
conducted with an iron salt. The cations FeSCN 2+ and Fe(SCN) +2 are formed and are
readily displaced from the resin (Wan and Miller, 1990). Thiocyanate recovered from
the solution is recycled by precipitation of the iron from solution using ferric hydroxide.
The alternative elution technique to ion exchange mass action is elution by chemical
reaction. The most prominent method in this category is thiourea-acid elution. This
was the preferred method of elution in the former Soviet Union, where a thiourea acid
solution was used after sequential elution of base metals (Bolinski and Shirley, 1996).
The chemical reaction between the gold cyanide complex and thiourea involves the
formation of the gold-thiourea cation complex. The gold-thiourea complex is easily
displaced from the resin. The method is deemed effective because of the irreversible
37
nature of the reaction resulting in fast kinetics and the ease by which gold can be
recovered from the thiourea eluent using cementation (Tataru, 1973).
The major drawback of the thiourea method arises due to the presence of impurities in
the pregnant solution. Other metal cyanides that are co-extracted with gold cyanide are
poorly eluted with acidic thiourea. Hence, multi elution stages are required to elute all
metal cyanides from the resin (Bolinski and Shirley, 1996). The situation is particularly
problematic when cobalt cyanide and iron cyanide complexes are loaded on the resin.
In acidic solution these complexes are converted to species that are difficult to strip
from the resin (Fleming and Cromberge, 1984b). Acidic thiourea elution trialed at the
Golden Jubilee mine caused massive iron precipitation in the resin matrix. As a
possible solution, researchers at the Canadian Metallurgy Research Center (CANMET)
have proposed a technique where concentrated sodium nitrate is used to scrub the iron
cyanide complex prior to acid exposure (Riveros et al., 1993). As mentioned
previously, Conradie and co-workers (1995) trialed the acidic thiourea elution method
on the gold selective strong base resin Minix® (Rohm and Haas). Whilst exact
percentages of other metal cyanides eluted were not reported, the authors claimed that
all elements loaded on the resin in cyanided gold pulp were completely eluted with the
acid thiourea. Minimal adsorption of cobalt and iron by the selective resin was able to
resolve the problem of subsequent metal precipitation. Due to the high gold
concentration of the eluent, the method was coupled to an electrowinning circuit to
recover the gold.
A novel approach proposed by Lukey and co-workers (2000a) involves the use of saline
solutions for the selective elution of base metal cyanide complexes from strong base
anion exchange resin. The resin was initially loaded with a solution containing a
mixture of gold, silver, zinc, copper and iron cyanide complexes (Lukey et al., 2000a).
A 2M potassium chloride eluent was able to selectively elute 100% of the iron and 76%
of the copper cyanide complexes from a trimethylamine strong base resin. Meanwhile
less than 1% of the gold and zinc cyanide complexes were eluted.
[
Raman spectroscopic studies have confirmed Cu(CN) 3 ]
2−
as the copper cyanide species
predominantly loaded onto anion exchange resin (Lukey et al., 2000b). Previous work,
38
has also found that Cu(CN) 4[ ]3−
is the most predominant copper species in saline
solutions (Lukey et al., 1999). Lukey et al., (2000a) have suggested that saline solutions
[
Moreover, as Cu(CN) 4 ]
3−
is a more hydrated complex, it is less favoured by the
Weak base resins are preferred to strong base resins because of their more favourable
elution characteristics. At pH values higher than the resin pKa, the equilibrium of the
ion exchange reaction given by equation 2-12, lies to the left and the functional groups
are no longer able to retain their adsorbed species. Elution merely involves a simple
acid-base hydrolysis, where the functional group of the resin is converted to the free
base form. The rate of elution of weak base resins increases with temperature and with
NaOH concentration up to an optimum concentration of 0.5M (Fleming and Cromberge,
1984b). The equilibrium presented in equation 2-13 is reversed by the addition of an
alkaline solution such as NaOH to release gold cyanide as follows,
Despite the fact that weak base resins offer a simple methodology for elution, the low
capacities of these resins are a major drawback preventing their more widespread use.
As a result, studies on elution have mostly focused on strong base resins. A summary
of the major methods of eluting the gold cyanide complex from anion exchange resins is
presented in Table 2-4. Generally, none of the elution procedures discussed above can
satisfactorily separate the gold cyanide complex from other metal cyanide complexes.
A promising approach that has been investigated at the laboratory level for the selective
elution of the gold cyanide complex is elution via the use of organic solvents.
39
Table 2-4: Methods used for the elution of metal cyanide complexes from anion
exchange resins.
40
2.5.3.1 Organic solvents
Presented with cumbersome and complex elution procedures, researchers have focused
on organic solvents as a means of facilitating the removal of the gold cyanide complex
from resin. In comparison to studies on carbon fewer research efforts have been
devoted to the use of mixed solvents in ion exchange systems for the recovery of gold.
[
Selective elution of Au(CN)2 ]
−
from ion exchange resin using organic solvents was
first studied by Burstall et al., (1953) and Burstall and Wells (1955). Initially,
hydrochloric acid was used to elute base metal complexes from the resin. The gold
concentration was built up through a series of base metal adsorption elution cycles. In a
final step, the gold cyanide complex was stripped from the resin using a mixture of
acetone in 5% HCl and 5% water. All of the gold cyanide was eluted with the acetone /
HCl mixture and only 6.6% and 5.6% of the Cu and Fe were simultaneously eluted.
Furthermore, the build up of the gold concentration through a number of
adsorption/elution cycles was able to offset the high price of the organic eluting agent
(Burstall et al., 1953). Tests have also been conducted using ethanol to replace acetone
(Burstall and Wells, 1955; Stamboliadis et al., 1978; Riveros et al., 1993).
Law and co-workers (1985) investigated a range of solvents for the elution of the gold
cyanide complex including dimethylacetamide (DMA), acetone, N-methyl-2-
pyrrolidone (NMP), dimethylsulfoxide (DMSO), hexamethylphosphramide (HMPA)
and tetrahydrofuran (THF). The solvents were investigated with SCN- as the counter
ion. The results of this investigation are presented in Figure 2-6. The performance of
each solvent in eluting gold is compared on the basis of solvent composition. It is
evident that compared to water, all mixed solvents improve the elution of the gold
cyanide complex. DMF, DMA, NMP and acetone perform better than DMSO, HMPA
and THF. From this work the recommended eluent for gold cyanide complex was a
solution containing 50% DMF and 4.88M KSCN.
41
Figure 2-5: Equilibrium loading for various metal cyanide complexes from a mine
leach solution (adapted from Fleming and Cromberge, 1984a).
Figure 2-6: The effect of organic solvent on the elution of gold from resin (adapted
from Law et al., 1985).
Two proposals were put forward to explain the observed selectivity of the resin for the
gold cyanide complex in nonaqueous solvents. First, it was proposed that organic
solvents may directly weaken the bond between the quaternary ammonium group on the
resin and the Au(CN) −2 ion so that another anion could exchange with the Au(CN) −2 ion
(Burstall et al., 1953). Alternatively, the change in the activity of the counterions in
nonaqueous solvents was thought to affect the solution equilibrium and thus alter the
distribution of Au(CN) −2 between the resin phase and the solution (Muir and Parker,
1982).
42
international cyanide management code was established by the United Nations in
cooperation with mining companies, governments and environmental groups. The code
has set an upper limit of 50 mg/L for weak-acid dissociable cyanide in open pond areas
such as tailings impoundments. The two solutions to the problem of cyanide in the gold
industry are either to destroy or recycle the cyanide.
The alternative option of recycling cyanide for re-use in the leaching process produces
environmentally acceptable effluent solutions and offsets the cost of remediation. The
Acidification-Volatilisation-Regeneration (AVR) process was one of the first cyanide
recovery processes and is based on the use of acid to recover as HCN gas (Riveros et
al., 1993). A caustic or lime solution is used to adsorb the HCN gas, producing sodium
cyanide or calcium cyanide, respectively. The AVR circuit in its original form is only
suitable for treating clarified, concentrated liquors in small volumes. Direct application
of the AVR process to a pulp is hindered by difficulties associated with controlling the
pulp pH, to avoid the decomposition of metal cyanides such as copper (Ciminelli, 2002;
Tran et al., 2000).
43
The AVR technique is most useful for handling low volumes of concentrated cyanide
streams, produced via solvent extraction or via ion exchange (Tran et al., 2000; Riveros
et al., 1993). Ion exchange resins are an attractive option due to the possibility of
recovering cyanide species directly from the pulps and concentrating the cyanide. For
this purpose the non-selective nature of ion exchange resins in the extraction of metal
cyanides is an advantage, as the extraction of these base metal cyanides from process
streams presents the possibility of recycling cyanide (Ciminelli, 2002; Leao and
Ciminelli, 2000; Fleming, 1998; Riveros et al., 1993). Coupling ion exchange
technology with the AVR circuit can produce environmentally acceptable effluent
solutions and presents the opportunity of recycling excess cyanide to the leaching
circuit. A number of ion exchange processes has been developed for this purpose,
including the Augment, Vitrokele and Elutech processes.
The maximum loading of copper and cyanide are achieved when all the resin sites are
[ ]−
occupied by Cu(CN) 2 . The copper is recovered by contacting the loaded resin with a
concentrated copper cyanide solution with a high cyanide to copper ratio (4:1). The
high concentration of cyanide converts the copper cyanide on the resin from
[Cu(CN) ]2
−
[
to Cu(CN) 3 ]
2−
. As the latter species requires two resin sites, the loading
of copper on the resin is halved. The liberated copper is recovered via electrolysis. The
44
AugMENT process was successfully tested by Lakefield Research to treat a Merill
Crowe barren solution from a Canadian gold mine. The tests produced effluent
solutions containing less than 1 mg/L copper and cyanide (Fleming, 1998b).
Vitrokele technology was later adopted by Imperial Mining at the May Day mine in
New South Wales, Australia (Tran et al., 2000). The treated ore contained 2.2 g/t of
gold and 0.12% copper. A non-selective adsorption circuit was followed by elution of
the copper cyanide species using a caustic mixture containing 100g/L NaCN and 10 g/L
NaOH. This eluant was ineffective at selectively eluting copper from the resin. The
gold cyanide complex also co-eluted with the copper cyanide resulting in a 2-5% loss of
gold for each elution cycle. The incomplete elution of copper left more than 50% of
copper on the resin. The residual copper on the resin caused the precipitation of CuCN
in subsequent acid regeneration processes. From the experiences at the May Day mine,
Vitrokele technology appears to be better suited to ores with low levels of copper, as
was the case at Connemara, Zimbabwe. The Vitrokele process at May Day Mine was
abandoned and a new process (Elutech technology) was implemented as a replacement.
2.6.3 Elutech
The Elutech process was successfully implemented at the May Day Mine in Australia
following the failure of the Vitrokele process to treat copper-gold ores (Tran et al.,
2000). The Elutech process scheme at the May Day mine initially involved the non-
selective loading of all metal cyanide complexes onto strong base anion exchange resin.
45
Subsequently, a water-wash was used to remove non-adsorbed base metal cyanides.
Base metals adsorbed on the resin were then eluted using an oxidative-acid eluent that
converts base metal cyanides to their respective sulfate forms. Hence, copper cyanide
species were converted from the Cu(I) oxidation state to the Cu(II) oxidation state,
producing copper sulfate salt. The cyanide released during elution was converted to
HCN and subsequently dissolved in the eluent. The concentrated HCN solution was
then treated in an AVR circuit to recover cyanide. The HCN was removed in a
stripping column and adsorbed in dilute NaOH. This effectively recycled cyanide back
to the NaCN form.
A number of adsorption and base metal elution cycles were undertaken to allow the
build up of the gold cyanide complex on the resin. The gold cyanide complex was later
eluted with zinc cyanide, as practiced at the Golden Jubilee mine in South Africa.
Sulfuric acid was used to regenerate the resin from the zinc cyanide form to the sulfate
form. Up to 99% of the Au was recovered from pregnant liquors, even though the
loading was non-selective. Base metal elution was achieved with no accompanying
release of the gold cyanide complex. The disadvantage of the Elutech process was
associated with the formation of double salts on the resin (Tran et al., 2000). A double
salt is formed when iron, solubilised in the leach liquor, reacts with Cu(II). The
resulting salt blocks the resin pores preventing further adsorption. Furthermore, large
fluctuations of pH during elution and regeneration cycles were speculated to cause resin
degradation.
46
2.7 REFERENCES
Abrams, I. A. and Millar, J. R. A history of the origin and development of macroporous
ion-exchange resins. Reactive and Functional Polymers, 1997. 35: p. 7-22.
Bolinski, L. and Shirley, L., Russian resin-in-pulp technology, current status and recent
developments. in Conference proceedings of the Randol gold forum, 1996: p. 419-
423.
Burstall, F. H., Forrest, P. J., Kember, N. F., and Wells, R. A., Ion-exchange process for
the recovery of gold from cyanide solution. Ind. Eng. Chem., 1953. 45: p. 1648-
1658.
Burstall, F. H. and Wells, R. A., Recovery of gold from cyanide solution by ion
exchange. in Ion exchange and its applications, London. Society of Chemical
Industry, 1955: p. 83-92.
Cathro, K. J. and Walkley, A., The cyanidation of gold, 1961, CSIRO Publication,
Melbourne, Australia.
Ciminelli, V. S. T., Ion exchange resins in the gold industry. JOM, 2002. 54(10): p. 35-
36.
Conradie, P. J., Johns, M. W., and Fowles, R. J., Elution and electrowinning of gold
from gold-selective strong-base resins. Hydrometallurgy, 1995. 37(3): p. 349-66.
Cotton, S. A., Chemistry of precious metals. 1997, Blackie Academic and Professional,
London.
Davidson, R. J. and Duncanson, E., Desorption of gold from activated carbon with
deionized water. Journal of the South African Institute of Mining and Metallurgy,
1977. 77(12): p. 254.
Davidson, R. J. and Veronese, V., Further studies on the elution of gold from activated
carbon using water as the eluant. Journal of the South African Institute of Mining
and Metallurgy, 1979. 79(15): p. 437-495.
Davison, J., Read, F. O., Noakes, F. D. L., and Arden, T. V., Ion exchange for gold
recovery. Bulletin - Institution of Mining and Metallurgy, 1961. 651: p. 247-90.
47
DeMent, E. R. and King, N. D., Merill-crowe/ Carbon-in-pulp an economic evaluation.
in Carbon-in-pulp seminar. 1982. The Aus.I.M.M. Perth and Kalgoorlie branches
and Murdoch University, Perth.
Fleming, C. A. , Thirty years of turbulent change in the gold industry. CIM Bulletin,
1998a: p. 55-67.
Fleming, C. A., The potential role of anion exchange resins in the gold industry, in EPD
Congress, Proceedings of Sessions and Symposia held at the TMS Annual
Meeting, San Antonio, 1998b. 16-19: p. 95-117.
Fleming, C. A. and Cromberge, G., The extraction of gold from cyanide solutions by
strong- and weak-base anion-exchange resins. Journal of the South African
Institute of Mining and Metallurgy, 1984a. 84(5): p. 125-37.
Fleming, C. A. and Cromberge, G., The elution of aurocyanide from strong- and weak-
base resins. Journal of the South African Institute of Mining and Metallurgy,
1984b. 84(9): p. 269-280
Fleming, C. A. and Nicol, M. J., The adsorption of gold cyanide onto activated carbon,
III: Factors influencing the rate of loading and equilibrium capacity. Journal of
the South African Institute of Mining and Metallurgy, 1984. 84(4): p. 85-93.
Fleming, C. A. and Seymore, D., Golden Jubilee RIP plant – process economics and
recent changes and improvements in plant performance, in Conference
proceedings of the Randol gold forum, 1990: p. 237-241.
Green, B. R., Kotze, M. H., and Wyethe, J. P., Developments in ion exchange: the
Mintek perspective. JOM, 2002. 54(10): p. 37-43.
Hedley, N and Kentro, D. M., Trans. Can Inst. Min. Metall., 1945. 48: p. 237.
Hefter, G., May, P. M., and Sipos, P., A general method for the determination of copper
(I) equilibria in aqueous solution. Journal of the Chemical Society, Chemical
Communications, 1993. 22: p 1704-1706.
Heinen, H. J., Peterson, D. G., and Lindstrom, R. E., Gold desorption from activated
carbon with alkaline alcohol solutions. in World Mining and Metals Technology.
1976: AIME.
48
Hermann, R. and Johns, M.W., Gold, gold alloys and gold compounds, in Ullmann’s
encyclopedia of Industrial Chemistry, Schuluz, G., (Ed). 1993. p. 620.
Hosking, J. W., The recovery of gold from ores by ion-exchange resins. Symposia
Series - Australasian Institute of Mining and Metallurgy 39(Gold-Min., Metall.
Geol.), 1984: p. 127-39.
Ibrado, A. S. and Fuerstenau, D. W., Effect of the structure of carbon adsorbents on the
adsorption of gold cyanide. Hydrometallurgy, 1992. 30: p. 243-256.
Jay, W. H., Copper-gold cyanide recovery systems. Cyanide: Social, Industrial and
Economic Aspects, Proceedings of Symposium held at Annual Meeting of TMS,
New Orleans, LA, United States, Feb. 12-15, 2001: p. 317-340.
Kautzmann, R. M., Sampaio, C. H., Cortina, J. L., Soldatov, V., and Shunkevich, A.,
The use of fibrous ion exchangers in gold hydrometallurgy. JOM, 2002. 54(10): p.
47-51.
Keating, R., Rothsay gold mine. in The processing of gold-copper ores colloquium.
1991. Perth: AMMTEC Pty. Ltd.
Laxen, P. A., Becker, G. S. M., and Rubin, R., Developments in the application of
carbon-in-pulp to the recovery of gold from South African ores. Journal of the
South African Institute of Mining and Metallurgy, 1994. 3: p. 189-203.
La Brooy, S. R., Linge, H. G., and Walker, G. S., Review of gold extraction from ores.
Minerals Engineering, 1994. 7(10): p. 1213-41.
Law, H. H., Wilson, W. L., and Gabriel, N. E., Separation of gold cyanide ion from
anion-exchange resins. Ind. Eng. Chem. Process Des. Dev., 1985. 24(2): p. 236-
238.
Lukey, G. C., Van Deventer, J. S. J., Chowdhury, R. L., and Shallcross, D. C., The
effect of salinity on the capacity and selectivity of ion exchange resins for gold
cyanide. Minerals Engineering, 1999. 12(7): p. 769-785.
Lukey, G. C., Van Deventer, J. S. J., and Shallcross, D. C., Selective elution of copper
and iron cyanide complexes from ion exchange resins using saline solutions.
Hydrometallurgy, 2000a. 56(2): p. 217-236.
Lukey, G. C., Van Deventer, J. S. J., Chowdhury, R. L., Shallcross, D. C., Huntington,
S. T., and Morton, C. J., The speciation of gold and copper cyanide complexes on
49
ion-exchange resins containing different functional groups. Reactive and
Functional Polymers, 2000b. 44(2): p. 121-143.
Lukey, G. C., Van Deventer, J. S. J., and Shallcross, D. C., Is ion-exchange technology
for gold extraction ready for commercialization? Publications of the Australasian
Institute of Mining and Metallurgy 2/98(AusIMM '98: The Mining Cycle), 1998:
p. 349-354.
Marsden, J. and House, I., The Chemistry of Gold Extraction. 1992, Ellis Horwood,
New York.
Mishra, R. K., Cyanide destruction and gold recovery - a review. Precious Metals 2002:
p. 44-65.
Muir, D., Recovery of gold from cyanide solutions using activated carbon: A review. in
Carbon-in-pulp seminar. 1982. The Aus.I.M.M. Perth and Kalgoorlie branches and
Murdoch University, Perth.
Muir, D. M., Aziz, M., and Hoecker, W. Cyanide losses under CIP conditions and the
effect of carbon on cyanide oxidation. Proceedings of the International
Hydrometallurgy Conference, Beijing, China, 1988.
Muir, D. M., Hinchliffe, W. D., and Griffin, A., Elution of gold from carbon by the
Micron Solvent Distillation procedure. Hydrometallurgy, 1985. 14(2): p. 151-169.
Nicol, M. J., The anodic behaviour of gold. Part II. Oxidation in alkaline solutions.
Gold Bulletin, 1980. 13: p. 105-111.
Nicol, M. J. and O'Malley, G., Recovering gold from thiosulfate leach pulps via ion
exchange. JOM, 2002. 54(10): p. 44-46.
Nicol, M. J., Schalch, E., Balestra, P., and Hedegus, H., A modern study of the kinetics
and mechanism of the cementation of gold. Journal of the South African Institute
of Mining and Metallurgy, 1979. p. 191-198.
Osse-Asare, K., Xue, T., and Ciminelli, V. S. T., Solution chemistry of cyanide
leaching systems. in: Precious Metals: Min. Extr. and Prc., Kudryk, V. (Ed),
AIME. 1984: p. 173-197.
50
Palmer, G. R., Ion-exchange research in precious metals recovery, Salt Lake City Res.
Cent.,Bur. Mines,Salt Lake City, USA, 1986: p. 2-9.
Petersen, F. W. and Van Deventer, J. S. J., Competitive adsorption of gold cyanide and
organic compounds onto porous adsorbents. Separation Science and Technology,
1997. 32(13): p. 2087-2103.
Pietrzyk, D. J., Ion Exchangers, in Chromatographic Science Series, 1990. 47: p. 585
Riveros, P. A., Selectivity aspects of the extraction of gold from cyanide solutions with
ion exchange resins. Hydrometallurgy, 1993. 33(1-2): p. 43-58.
Riveros, P. A. and Cooper, W. C., Ion exchange recovery of gold and silver from
cyanide solutions. in Proceedings of the Metallurgical Society of the Canadian
Institute of Mining and Metallurgy, 1987. 1: p. 379-393.
Ruane, M., Comparison of the Zadra, Anglo-American and organic procedures for
desorption of gold from activated carbon. in Carbon-in-pulp seminar. 1982. The
Aus.I.M.M. Perth and Kalgoorlie branches and Murdoch University, Perth.
Ryan, M. W., Processing of gold-copper ore at Red Dome gold. In The processing of
gold-copper ores colloquium. 1991. Perth: AMMTEC Pty. Ltd.
Satalic, D. M., Spencer, P. A., and Paterson, M. R., Vitrokele - Commercial application
in the gold industry. in Publications of the Australasian Institute of Mining and
Metallurgy 1/96(Diversity: The Key to Prosperity), 1996: p. 167-171.
Streat, M., Ion Exchange - A technologist's perspective of the 20th century. In Advances
in Ion Exchange for Industry and Research; Williams, P.A., Dyer, A., (Editors).
The Royal Society of Chemistry: Cambridge, 1999: p. 3-11.
Stramboliades, E., McHardy, J., and Salman, T., Ion exchange techniques for the
recovery of gold from cyanide solutions. Can. Min. Metall. Bull., 1978. 71(796): p.
124.
Syna, N. and Valilx M., Assessing the potential of activated bagasse as gold adsorbent
for gold-thiourea. Minerals Engineering, 2003. 16: p. 511-518.
Tataru, S. A., Recovery of gold from cyanide solutions by ion exchange. in proceedings
of the Tenth International Mineral Processing Congress, 1973. London: The
institution of mining and metallurgy: p. 1135-45.
51
Tran, T., Lee, K., Fernando, K., and Rayner, S., Use of ion exchange resin for cyanide
management during the processing of copper-gold ores. Publications of the
Australasian Institute of Mining and Metallurgy 5/2000(MINPREX 2000), 2000:
p. 207-215.
Tran, T., Nguyen, H. H., Hsu, Y. J., and Wong, P. L. M., Copper-gold interaction
during the processing of copper-gold ores. Publications of the Australasian
Institute of Mining and Metallurgy 2/97(World Gold '97), 1997: 95-98.
Tsuchida, N. and Muir, D., Studies on role of oxygen in the adsorption of Au(CN) −2 and
Ag(CN) −2 onto activated carbon. Metallurgical Transactions B, 1986. 17B: p.
529-533.
Tsuchida, N., Ruane, M., and Muir, D. M., Studies n the mechanism of gold adsorption
on carbon. In MINTEK 50, 1984: p. 647-656.
Van Der Merwe, P. F. and Van Deventer, J. S. J., The influence of oxygen on the
adsorption of metal cyanides on activated carbon. Chemical Engineering
Communications, 1988. 65: p. 121-138.
Wan, R.Y., Brierley J.A., Acar, S., and LeVier, K. M., Using thiocyanate as lixivant for
gold recovery in acidic environment. in Volume 1: Leaching and Solution
Purification, Hydrometallurgy 2003 – Fifth International conference in Honor of
Professor Ian Ritchie, 2003. TMS: p. 105-121.
Wan, R. Y. and Miller, J. D., Research and development activities for the recovery of
gold from alkaline cyanide solutions. Minerals Processing and Extractive
Metallurgy Review, 1990. 6: p. 143-190.
Zadra, J. B., Engel, A. K., and Heinen, H. J., A process for recovery of gold from
activated carbon by leaching and electrolysis. USMB reports of investigations,
1952. 4843.
52
CHAPTER 3
3. ION EXCHANGE EQUILIBRIA: SELECTIVITY AND PREDICTIVE
MODELS ...................................................................................................... 54
3.1 INTRODUCTION ...........................................................................................54
3.2 THE SIGNIFICANCE OF ION EXCHANGE EQUILIBRIA ................................54
3.3 ION EXCHANGE SELECTIVITY ....................................................................55
3.3.1 Properties of counterions ........................................................................57
3.3.2 Hydration ................................................................................................60
3.4 ION EXCHANGE SELECTIVITY IN NONAQUEOUS AND MIXED SOLVENTS .63
3.4.1 Properties of organic solvents .................................................................64
3.4.2 The effect of nonaqueous solvents on ion-pair formation ......................67
3.4.3 The stability of ions in nonaqueous solvents ..........................................68
3.5 PREDICTIVE MODELS FOR ION EXCHANGE EQUILIBRIA ..........................73
3.5.1 Quantitative treatment of selectivity .......................................................73
3.5.1.1 Isotherms .............................................................................................73
3.5.1.2 The equilibrium constant ....................................................................76
3.5.1.3 The selectivity coefficient ...................................................................76
3.5.2 Adsorption models ..................................................................................77
3.5.3 The Law of Mass Action.........................................................................80
3.5.3.1 Solution phase activity coefficient ......................................................84
3.5.3.2 Resin phase activity coefficient ..........................................................88
3.6 REFERENCES ...............................................................................................91
1
Table 3-4: A summary of the various forms of the Law of Mass Action (LMA) and
methods used for calculating solution phase and resin phase activity
coefficients. ..................................................................................................81
Table 3-5: Comparison between component objective functions determined using two
different methods to evaluate the solution phase activity coefficients
(Shallcross et al., 1988)................................................................................87
Table 3-6: The reciprocity relation proposed by Allen et al., (1989) applied to a number
of binary systems..........................................................................................90
2
3. ION EXCHANGE EQUILIBRIA: SELECTIVITY AND
PREDICTIVE MODELS
3.1 INTRODUCTION
The non-selectivity of anion exchange resins for metal cyanide complexes is a principal
obstacle hindering their use for the extraction of the gold cyanide complex. As
reviewed in Chapter 2, a considerable amount of data has been accumulated on the
extraction of gold using ion exchange resin. By and large, these research approaches
have been motivated by a need to optimise the industrial output of gold. Therefore, less
success has been achieved in building a comprehensive theoretical framework for the
fundamental factors that influence selectivity. In particular, the potential of nonaqueous
solvents to address the problem of selectivity of ion exchange resins in the recovery of
metal cyanide complexes is yet to be fully explored. This chapter reviews the factors
that influence selectivity in ion exchange processes in aqueous and nonaqueous
solvents. In addition, the Law of Mass Action (LMA) is examined in detail as the
model of choice for predicting ion exchange equilibria.
54
basis of which technical and economic feasibilities can be estimated. In addition,
equilibrium data provide a foundation for understanding the selectivity of anion
exchange resins for the gold cyanide complex during loading and elution.
Similar studies in the area of gold hydrometallurgy are lacking. Except for a few
studies by Riveros (1993), and Lukey et al., (2000 and 1999), there is relatively little
literature considering the physical significance of selectivity or rigorous models for
predicting equilibria for ion exchange systems involving the gold cyanide complex.
The extensive literature on ion exchange equilibria of other ionic systems serves as a
starting point for understanding ion exchange equilibria involving the gold cyanide
complex.
55
βA αr − + αB βs − ⇔ βA sα − + αBβr − (3-1)
In equation 3-1, α and β represent the valence of ions A and B. At equilibrium, both
ions A and B in equation 3-1 are present in the resin and solution phases. The
distribution of these ions between the two phases is governed by the selectivity of the
resin. Most studies to date have focused on investigating the effect of a single factor on
ion exchange selectivity. The fact that no single investigation has attempted to confront
all the facets linked to selectivity is testimonial to the complexity of the problem.
Furthermore, very little knowledge exists on the resin itself, as no direct methods are
available for analysing the resin during an exchange process. Hence it should be stated
at the outset that the currently accepted ideas on ion exchange selectivity are insufficient
to explain all the observed variations of selectivity.
Ion exchangers prefer counterions that can interact most strongly with functional groups
on the resin matrix (Helfferich, 1995). Therefore the mechanism by which a counterion
associates with the functional group is important in identifying the factors that
determine the selectivity of a resin for a given ion. While the exact mechanism of ion
exchange is still to be elucidated, earlier theories and recent findings point to
electrostatic interactions dominated by ion-pair formation as the most likely mode of
interaction between counterions and functional groups. Helfferich (1995) and Pietryzk
(1990) suggest that provided that electroneutrality is maintained, counterions are free to
move within the charged matrix of the resin, with no counterion permanently bound to a
particular functional group. In fact, counterions exist in the vicinity of the resin surface
where the potential energy between the two ions of opposite charges (counterion and the
charged site on the resin) is likely to be at a minimum. Such conditions favour the
formation of ion-pairs (Okada, 1997).
56
molecules, provided that the intermolecular forces between the ions are sufficiently
strong. Such ion-pairs are referred to as solvent-separated ion-pairs (C). However,
once solvent effects overcome intermolecular forces, the ion-pair dissociates into
constituent ions (D).
The following discussion on ion exchange selectivity assumes that the predominant
mode of ion interaction with resins is through the formation of ion-pairs. Hence the
primary properties of ions that influence ion-pair formation are first outlined. These
factors are best understood when they are isolated from the solvent environment of the
resin. Next, factors that influence ion-pair formation are re-considered by introducing
the solvent environment. The effect of hydration in an aqueous ion exchange
environment are first examined, while nonaqueous and mixed solvent systems are
discussed in a later section of this chapter.
57
Figure 3-1: Schematic diagram of the formation of an ion-pair.
R X R+ X- R+ X- R+ X-
A B C D
Ionic bond in a Ion-pair Solvent-separated Dissociated ions
crystal lattice ion-pair
Figure 3-2: Ion exchange isotherms for the Cu 2+ / Na + binary system as a function of
total solution normality (adapted from Subba Rao and David, 1957).
According to Figure 3-2 the affinity of the resin for Cu 2+ increases with dilution of the
external solution. This observation is rationalized by the Donnan theory, which
suggests that the driving force for ion exchange is governed by concentration
differences between the solution and resin phases (Helfferich, 1995). Generally, the
resin has a higher concentration of charged ions compared to the external solution.
Consequently, a potential difference arises between the two phases which leads to ion
exchange. The strength of the potential difference between the solution and resin
phases is likely to increase with dilution of the external solution. Therefore, as
illustrated in Figure 3-2, the resin phase composition of Cu 2+ is highest for the lowest
total concentration in the external solution.
58
Apart from the valance of an ion, the electrostatic field strength exerted by an ion is also
influenced by its ionic radius, as well as the distribution of the charge amongst atoms in
the molecule. Generally, the strength of an electric field exerted by an ion is inversely
proportional to the size of the ion. Table 3-1, lists the ionic radii of selected anions.
The relationship between the charge of an ion and its size can be quantified by defining
a quantity known as the charge density. The charge density is determined by dividing
the charge on an ion by its ionic radius, and is indicative of the distribution of the charge
[ ]
−
as Au(CN)2 , which has a relatively low charge density. Hence in the absence of
solvent contributions, ions with a small radius or high charge density are likely to be
more successful at forming ion-pairs with charged groups on a resin.
Cl − 181 5.53 2
[Fe(CN) ] 6
4− 450 8.89 3.2
a
Data from Marcus (1997)
59
b
Ionic radii of ions isolated from solvent.
c
charge density = charge/ionic radius
3.3.2 Hydration
Predictions of selectivity based purely on the primary properties of ions are seldom
observed. For example, based on ionic radius and charge density, ion exchange resins
should prefer chloride compared to the gold cyanide complex. In fact, ion exchange
resins have a stronger affinity for the gold cyanide complex in aqueous solutions. These
deviations can be understood by accounting for the solvent environment of an ion
exchange process. Generally, solvation refers to the interaction occurring between an
ion and any solvent, including organic solvents. In an aqueous environment, solvation
by water is termed hydration. Hydration influences ion-pair formation between
counterions and functional groups on the resin by weakening electrostatic interactions
caused by primary ionic properties. Thus the selectivity displayed by a resin for a given
ion is altered.
When an electrolyte is placed in water two processes take place. First, bonds are broken
between ions in the crystal structure of the electrolyte. Second, the detached ions are
enveloped by water molecules and dispersed homogeneously throughout the aqueous
solution (Popoviych and Tomkins, 1981). The latter process, where water molecules
surround ions in solution, is known as hydration. The average number of water
molecules bound to an ion is given by the hydration number.
Hydration numbers for individual ions are determined from different thermodynamic
properties of solutions and ions. The hydration numbers presented in Table 3-1 are
calculated based on the charge and ionic radius of the ions (Marcus, 1997). Values
based on this method of calculation are readily available for most ions including
[Au(CN) 2 ]− . The absolute hydration number varies considerably, depending on the
method of calculation. Hence the values listed in Table 3-1 should only be used for
comparative purposes. The hydration numbers in Table 3-1 increase with increasing
charge density of the ions. Ions with a high charge density can exert a stronger
electrostatic field to attract surrounding water molecules.
60
Water molecules involved in ion-hydration orientate themselves around the surface of
ions, causing the formation of solvent shells (Popoviych and Tomkins, 1981). Solvent
shells of a counterion are likely to weaken the electrostatic interactions between the
counterion and the functional group on the resin. Figure 3-3 is a graphical depiction of
the effect of hydration on the interactions between counterions and functional groups.
The resin matrix and the functional group are represented by R and N+, respectively.
The ionic radius of anion, X − is less than that of Y − . In essence, ion X − is expected to
have a stronger localized charge and is therefore more hydrated in an aqueous solution
compared to Y − . Consequently, in the case of ion X − , the interaction with the
functional group is compromised by hydration. On the contrary, the less hydrated ion,
Y − is able to interact more closely with the functional group of the resin.
As an example, Okada (1997) investigated the ion exchange behaviour of ClO −4 and
An interesting finding in the work by Okada (1997) was that the preference for ClO −4
over Cl − was larger in the quaternary ammonium resin compared to the primary
ammonium resin. Okada explained the more pronounced preference for the perchlorate
61
ion in the former resin by considering the process of desolvation. According to Okada
(1997), hydrated counterions undergo a process of desolvation at the charged site of the
resin functional group. Desolvation is an energetically unfavourable process involving
the elimination of water molecules associated with the solvent shells of an ion.
Elimination of water from solvent shells enables closer interaction between the
functional group and the counterion, thus enabling the formation of stronger ion-pairs.
Apart from the effect of hydration on the strength of electrostatic interactions between
counterions and functional groups on the resin, hydration also affects the relative
hydrophilicity of counterions. Molecules such as [Au(CN) 2 ] and ClO −4 are more
−
hydrophobic in character compared to ions such as Cl − , due to the relatively low states
of hydration in the former ions (Riveros, 1993). The ion exchange resin which mostly
consists of a carbon matrix is relatively more hydrophobic compared to the external
aqueous phase. Based on these conditions alone, ions that are less hydrated and
therefore more hydrophobic are likely to prefer the resin phase compared to the external
[
solution. The affinity of the resin for the Au(CN)2 ]
−
complex in the presence of base-
metal cyanide complexes has been explained on the basis of its relative hydrophobicity
in a number of studies as previously detailed in Chapter 2.
62
surrounding water molecules better than large, low-charged ions. In fact low-charged
ions break up a large volume of water structure but do not have a sufficiently high
charge density to tightly bind the nearest water molecules.
Chu et al., (1962) proposed that the solution in the resin phase has a highly disrupted
water structure compared to the external solution. As a result the resin phase offers
much less resistance to the entrance of a large ion. Larger ions are more easily
transferred out of the dilute aqueous solution into the less ordered water structure of the
resin phase. The theory is limited by the size of the resin pores. According to Chu et
al., (1962), the selectivity trends of small ions are based on hydration factors, whereas
for larger ions it is dependent on disrupting the water structure of the external solution.
Many hydrometallurgical processes employing ion exchange resins have benefited from
the incorporation of nonaqueous solvents. The selectivity sequences of both anions and
cations are altered in pure and mixed solvent systems. For example Fessler and Strobel
(1963) found that the cations Na + , K + , and Cs + , displayed a greater affinity for cation
exchange resin in solvents containing high compositions of methanol. Strelow et al.,
63
(1974) have shown that uranium can be quantitatively separated from copper due to the
selective elution of copper from resin with mixtures of hydrobromic acid and acetone.
Particular improvements are seen in the application of nonaqueous solvents in chloride
hydrometallurgy. The separation of cations is vastly improved, as a result of the relative
stabilities of the respective chloro-complexes in mixed solvent systems containing a
mineral acid such as hydrochloric acid (Fleming and Monhemius, 1979, Korkisch and
Ahluwalia, 1966). A selection of studies using nonaqueous solvents in ion exchange
systems is summarized in Table 3-2.
Table 3-2: A summary of research on ion exchange resins using nonaqueous solvents
for the separation of ions.
64
An important characteristic of a solvent molecule is its polarity. Polar molecules have a
permanent electric dipole due to the unequal distribution of electrons amongst atoms in
the molecule. A measure of the electric dipole of a molecule is the dipole moment.
Dipole moments for selected solvents are listed in Table 3-3. Solvents such as benzene
have dipole moments close to zero indicating the absence of polarity in the molecule.
These solvents are classed as non-polar. On the other hand water, aliphatic alcohols,
acetone, dimethylsulfoxide (DMSO), N-methyl pyrrolidone (NMP) and acetonitrile
(AN) are classed as polar due to the relatively high dipole moments of their constituent
molecules.
Polar solvents solvate (or hydrate in the case of water) ions in solution by clustering
around ions so that the negative poles of the solvent molecules surround positive
charges, while the positive poles of the solvent molecules surround negative charges.
Therefore an ion is stabilised in solution by dispersion of the charge over the partial
charges of polar solvent molecules. The dielectric constant of a solvent is a measure of
the ability of a solvent to insulate charged species in solution from one another. Table
3-3 lists the dielectric constants for the selected solvents. Polar solvents generally have
a higher dielectric constant compared to nonpolar solvents, which are poor solvators of
ions in solution.
Polar solvents are further categorized as protic or aprotic (Popovych and Tomkins,
1981). The division of solvents as protic or aprotic is based on the hydrogen bonding
ability of the solvents. When a hydrogen atom is bound to an electron withdrawing
group such as oxygen or nitrogen a partial positive charge is established in the hydrogen
atom. As a consequence, the hydrogen atom is able to bond with an electron donating
Table 3-3: Important properties of selected solvents (adapted from Marcus, 1985).
65
AN Aprotic 41.1 0.777 3.44 37.5 46.0
Acetone Aprotic 58.1 0.784 2.69 20.7 42.2
DMSO Aprotic 78.1 1.096 3.90 46.7 45.1
NMP Aprotic 99.1 1.028 4.09 32.0 42.2
Benzene Non-polar 78.1 0.874 0.00 2.28 34.5
Pertinent to the preceding discussion are the aprotic solvents acetone, DMSO and NMP.
Acetone is a typical ketone with a carbonyl attached to two alkyl groups. The oxygen
on the carbonyl gives rise to the polarity of the molecule. However the partial positive
charge on the carbon is stabilised by the electron donating ability of the two alkyl
groups. As a result of the stabilised carbonyl group only 0.2% of acetone is hydrated in
water. An important function of ketones in organic chemistry is in the formation of new
carbon-carbon bonds. These include the reaction of acetone with cyanide to form
acetone cyanohydrin, which is synthetically useful for subsequent reactions such as in
the formation of carboxylic acids (Covington and Dickinson, 1973).
DMSO is a liquid of similar density to water. X-ray scattering studies indicate that
DMSO hydrogen bonds to water molecules to form H2O-DMSO clusters (Covington
and Dickinson, 1973). According to the Linus-Pauling scale, the electronegativity of
66
the sulfur atom is equivalent to that of carbon. Hence the partial charge on the sulfur
should be similar to that of the carbon on acetone. On the other hand the dielectric
constant of DMSO is higher than that of acetone. Both NMP and DMSO are excellent
solvents for nucleophilic substitution reactions (Parker, 1969). This is because polar
aprotic solvents have atoms with nonbonding electrons that can stabilise positive
charges. Thus the rate of an SN2 reaction involving a negatively charged nucleophile
will be greater in a polar aprotic solvent than in a polar protic solvent.
Aprotic solvents are weak hydrogen bond donors, and are less successful at interacting
with anions. As a result anions are less solvated in aprotic solvents compared to water.
Less solvation decreases the effective solvated diameter of counterions, enabling closer
interaction with charged sites on the resin. Consequently, the counterions and the
charged sites on the resin are likely to form stronger ion-pairs in aprotic solvents
compared to protic solvents. In fact ion-pair formation between primary ammonium
type charged sites and the anions, Cl − , Br − , I − and ClO −4 are significantly higher in
aprotic solvents such as AN and dimethylformamide (DMF), compared to protic
solvents such as methanol (Okada, 1997).
67
Generally, the selectivity sequence for anions in protic solvents is reversed upon the
transfer of the anions to an aprotic solvent system. Phipps (1968) studied binary anion
exchange equilibria in DMSO. In the solutions of DMSO, the selectivity of anions for
the resin increased in the order ClO −4 < SCN − < NO 3− < I − < Cl − < Br − . Due to
decreased solvation in aprotic solvents, ion-pair formation is likely to be influenced by
primary counterion properties. To this effect, the resin is likely to be more selective for
ions with localized charge densities, such as Cl − , as opposed to ions with dispersed
charges such as ClO −4 . This premise supports the observed selectivity sequence for
anions in DMSO.
AgCl + Cl - ⇔ [AgCl 2 ]
−
(3-2)
In an aqueous system, the equilibrium in equation 3-2 strongly favours the formation of
insoluble silver chloride. However silver chloride is known to be highly soluble in
aprotic solvents such as DMSO. The high solubility of silver chloride in DMSO is
attributed to the stronger solvation of [Ag(Cl)2 ] and the very much weaker solvation of
−
Cl - by aprotic solvents relative to water. For example the activity of Cl - can increase
by a factor of 108 in acetone, DMSO or AN compared to water (Muir and Parker, 1982).
As a result of the activity changes in dipolar solvents, the equilibrium in equation 3-2
lies strongly to the right and results in the dissolution of AgCl. A novel process for
recovering silver is based on the changes to the equilibrium distribution of silver
68
chloride in water and DMSO (Parker et al., 1979). Initially DMSO is used to selectively
solubilise silver chloride from impure sources of silver. Water is then added to reverse
the equilibrium, and pure silver chloride is recovered.
Similarly, the relative stabilities of ions in nonaqueous and mixed solvents are
suggested to be the cause of various separations using ion exchange resin. In particular,
cation separation processes have been improved by the addition of a complexing agent
to the nonaqueous solvent or mixed solvent system. Such complexing agents include
hydrochloric acid (Hassan, et al., 1981, Fleming and Monhemius, 1979, Simek 1974),
hydrobromic acid (Strelow et al., 1974), perchloric acid (Hassan et al., 1981), and
glycine (Brajter and Miazek, 1981). As illustrated in the example of the effect of
DMSO on the dissolution of AgCl, complexes formed between complexing agents and
cations have different stabilities in various nonaqueous solvents compared to water.
69
exchange resin. Figure 3-5 illustrates the same systems in the presence of acetonitrile
containing 4M lithium chloride. It can be seen that the elements nickel, cobalt and
copper are all strongly adsorbed on the anion exchange resin. The observed change in
selectivity of cations in the mixture of acetonitrile and LiCl was attributed to the relative
stability of the chloro-complexes in the organic solvent.
Figure 3-6 is a summary of the factors that contribute to the selectivity displayed by an
ion exchanger for ion X − and Y − . Generally ion exchangers prefer the counterion that
can form the strongest ion-pair with the functional group of the resin. Free from solvent
effects, ion-pair formation is influenced by electrostatic properties of the counterion and
the fixed ion of the functional group. These characteristics include valence, ionic radius
and charge density of an ion. However, selectivity trends predicted purely on the basis
of electrostatic influences are seldom observed. Figure 3-6 demonstrates that the
solvent environment has a substantial influence on ion-pair formation. Generally, in an
aqueous environment, ion hydration can significantly alter the strength of electrostatic
interaction between counterions and functional groups. The situation is further
complicated in nonaqueous and mixed solvent systems where both hydration and
solvation can take place depending on the composition of the solution. Furthermore, in
nonaqueous and mixed solvent systems, the stability of ions in the solution phase and
resin phase becomes an additional driving force which contributes to the ultimate
selectivity of a resin for a given ion.
70
Figure 3-4: The loading of metal ions onto anion exchange resin from an aqueous
solution adjusted with 4M LiCl (adapted from Fleming and Monhemius, 1979).
25
10
0
0 5 10 15 20 25
2
Metal concentration in solution (10 mg/L)
Zn Pb Fe(III) Fe(II)
Cu(II) Ni Co
Figure 3-5: The loading of metal ions onto anion exchange resin from an acetonitrile
solution adjusted with 4M LiCl (adapted from Fleming and Monhemius, 1979).
25
Metal concentration in resin
20
(mg/g resin)
15
10
0
0 2 4 2 6
Metal concentration in solution (10 mg/L)
71
Figure 3-6: A summary of the major factors influencing the selectivity of an ion
exchange resin for representative ions X − and Y − .
Hydration Solvation
Ion
ce
ic
X- Y-
len
rad
Va
ius
RN+ N+R
Charge Density
Selectivity
OR
72
3.5 PREDICTIVE MODELS FOR ION EXCHANGE EQUILIBRIA
Most of the early theories on ion exchange equilibria originated from empirical
observations, based on thermodynamic analogies. The interested reader is directed to a
monograph by Helfferich (1995) for detailed comment on these theories. Over time, a
flourish of data has not only evolved the understanding of ion exchange equilibria but
also improved the models used for their prediction. Models proposed in the literature
for the prediction of ion exchange equilibria can be divided into theories based on: (1)
physical adsorption and (2) chemical exchange according to the Law of Mass Action
(LMA). The former category considers ion exchange as a physical process whereas
mass action theory depicts ion exchange as a chemical reaction.
Whilst the following discussion is based on the division of theories based on adsorption
and LMA, the reader should keep in mind that the distinction between the two
categories is debatable. Strictly speaking adsorption can be expressed as a mass action
process. The surface coverage of a molecule on an adsorbate is dependent on the rates
of adsorption and desorption. Furthermore the concepts in the two categories often
overlap in many theories on ion exchange equilibria and often result in the prediction of
similar trends and patterns of selectivity. The discussion on predictive models for ion
exchange equilibria is preceded by a discussion on terminology used to describe ion
exchange equilibria.
3.5.1.1 Isotherms
An isotherm is a graphical representation of ion exchange equilibria. An ion exchange
isotherm presents all possible experimental conditions of a system at a given
temperature. Each point on an isotherm corresponds to a particular set of experimental
conditions such as solution concentration or resin volume. In an isotherm, the
composition of the resin phase is plotted as a function of the solution phase
composition. A number of methods can be used to represent the resin and solution
compositions. The most suitable approach involves the use of equivalent ionic
73
fractions. Equivalence is an alternative concentration term that takes the charge of an
ion into account. Therefore the equivalent ionic fraction of counterion i is defined by,
z i ci
xi = (3-3)
N
z i qi
yi = (3-4)
Q
where xi and yi are the equivalent ionic fractions in the solution phase and resin phase,
respectively, zi is the valence of ion i, N is the total normality of exchanging ions in the
solution phase and Q is the capacity of the resin (equivalents/L). The summation for the
total concentrations in the solution and resin phases are carried out over all counterion
species involved in the ion exchange process. Generalised ion exchange isotherms are
presented in Figure 3-7, for the binary ion exchange involving ion A and ion B. The
isotherms are plotted in terms the ionic fraction of ion A in the resin phase (yA) as a
function of the ionic fraction of ion A in the solution phase (xA). The selectivity of the
resin can be ascertained from the shape of the isotherm. For the isotherm that lies above
the diagonal (I), the resin is considered to be selective for ion A. On the other hand, the
isotherm below the diagonal (II) indicates that the resin is selective for ion B. In
essence, the diagonal on Figure 3-7 represents a hypothetical situation in which the ion
exchanger shows no preference for either ion. Selectivity thus necessitates a nonlinear
isotherm.
74
Figure 3-7: Ion exchange isotherms for the exchange involving ion A and ion B.
(I)
yA
(II)
0 1
xA
75
3.5.1.2 The equilibrium constant
The equilibrium constant is a quantitative expression of the selectivity exhibited by an
ion exchanger. Equilibrium constants characterise the whole isotherm and can be
thought of as true constants dependent only on temperature. The equilibrium constant,
KAB for B entering the resin phase and displacing A (equation 3-1) is expressed by
(Valverde et al., 2002; de Lucas Martinez et al., 1994; Shallcross et al., 1988; Sengupta
and Paul, 1985)
( γ A cA ) β ( γ B q B ) α
K AB = (3-5)
( γ A q A ) β ( γ B cB ) α
where ci is the molar concentration of ion i in the solution phase and qi is the molar
concentration of ion i in the resin phase. The activity coefficients of ion i in the solution
phase and resin phase are denoted by γ i and γ i , respectively. The activity coefficients
are integral quantities in equation 3-5 as they characterise the deviation of a system
from ideality. The extent to which a calculated equilibrium constant is a ‘true’
thermodynamic equilibrium constant is dependent on the accuracy with which the
activity coefficients, particularly in the resin phase, are calculated. The use of equation
13-5 to describe the equilibrium constant assumes that swelling-pressure effects are
negligible (Helfferich, 1995).
(c A ) β ( q B )α
′ =
K AB (3-6)
( q A ) β (c B )α
76
Selectivity coefficients are easily measured, and present a more practical alternative to
the equilibrium constant. Furthermore, for sufficiently dilute systems, averaged values
of the selectivity coefficient give a reasonable representation of the overall equilibrium
constant of the system. Nonetheless, the selectivity coefficient varies with experimental
conditions, and any specific value only corresponds to one point on the isotherm
surface. A corrected form of equation 3-6, termed the equilibrium quotient, is defined
by Smith and Woodburn, (1978). The equilibrium quotient is a quantity corrected with
the solution phase activity coefficients, and as such it can be determined from
experimental binary equilibrium data. A number of other studies have also used the
equilibrium quotient as an experimental parameter (Shehata et al., 2000; Mehablia, et
al., 1994; Shallcross et al., 1988). The equilibrium quotient, λAB , is stated as follows,
(γ A cA ) β (qB )α
λAB = (3-7)
(γ B cB )α (qA ) β
In their work Myers and Byington (1986) noted that experimental procedures could not
distinguish bulk liquid from pore liquid. Ideally the solid phase should only contain
fixed ionic groups and bound counter ions. However because of solvent adsorption, the
solid phase also contains both solvent molecules and unbound ions or dissolved
electrolyte. In order to mitigate this anomaly Myers and Byington (1986) introduced
the concept of excess surface, nie , as follows,
77
y i − xi
nie = (3-8)
Q
where yi and xi are the equivalent ionic fractions in the resin phase and solution phase
respectively, and Q is the capacity of the resin (equivalents/L). According to equation
3-8 the excess surface of an ion i refer to the difference between the amount of an ion
which is adsorbed and the amount which would have been adsorbed had the equivalent
ionic fraction in the resin been the same as its value in the bulk liquid phase. Myers and
Byington (1986) suggested that the depiction of ion exchange isotherms according to
the dependence of the excess surface of ion i on the equivalent fraction of ion i in
solution, was more accurate for adsorption theory, than isotherms based on equivalent
fractions on the resin and solution phases.
Isotherms depicting the surface excess of ion A are predicted by considering the
energetic heterogeneity of the ion exchanger. In particular, statistical distributions of
the adsorption energies are considered, while assuming that each adsorption site
behaves ideally, i.e. according to a Langmuir type isotherm. The model involves three
adjustable parameters per binary system: the average adsorption energy of each site, the
standard deviation of the average energy and a parameter defining the skewness of the
energy distribution. As such the number of parameters is the same as current models
based on the Law of Mass Action. However, the main difference in the Myers and
Byington (1986) approach is that it accounts for solute-adsorbent rather than solute-
solute interactions, and takes them as being responsible for the deviations from ideal
behaviour. The consequence is that for a multicomponent system, there is no need to
investigate all of the binary systems formed by all possible pairs of NC counterions
present in the system, i.e. NC (NC-1)/2. Only (NC-1) binary systems formed for each
one of the NC counterions needs to be considered. This significantly reduces the
experimental effort required to develop a multicomponent equilibrium model, as well as
reducing the overall number of adjustable parameters involved.
Recently, the model proposed by Myers and Byington (1986) has been advanced by
Melis et al., (1995) by incorporating parts of the LMA into the existing model. Thus the
application of the model has widened to include mono and bivalent ions at various
values of solution normality. However the model becomes identical to the Myers and
78
Byington model in the case of exchange between homovalent ions. In their approach
Melis et al., (1995) maintain the assumption of Myers and Byington (1986) that both the
fluid and solid phases behave ideally. The non-ideality of the system is taken into
account by the heterogeneity of the adsorption sites on the ion exchanger. In other
words the resin is considered as having two different types of exchange sites.
Accordingly the total capacity of the resin Q may be expressed as,
where q0.1 and q0.2 are the capacities of the resin with respect to the functional groups of
type 1 and 2, respectively. Whereas Myers and Byington (1986) treated the behaviour
on each site according to the ideal Langmuir model, the model of Melis et al., (1995)
considers the two different types of exchange sites as having their own equilibrium
constants. Accordingly the equilibrium constant for the exchange process occurring on
the jth functional group is given as follows,
y j , B x A ⎛ q0. j ⎞
Kj = ⎜ ⎟⎟ (3-10)
y j , A xB ⎜⎝ N ⎠
where, y j ,i represents the ionic fraction of ion i on the jth functional group according to,
q j .i
y j ,i = (3-11)
q 0. j
For a resin containing sites of type 1 and 2, two equilibrium constant K1 and K2 are
obtained for each system by regressing the experimental binary data. The difference
between the two equilibrium constants is an indication of the non-ideality of the
exchanger phase. The model has been successfully applied by Melis et al., (1995 and
1996), and Valverde et al., (1999). Recently, Vo and Shallcross, (2003) extended the
model to include activity coefficients for the solution phase. This extended model was
applied to several binary cationic systems and was found to predict equilibrium
behaviour more accurately than the original model of Melis et al., (1995).
79
3.5.3 The Law of Mass Action
Models for ion exchange equilibria based on the Law of Mass Action (LMA) have been
reviewed by Valverde et al., 1999, Mehablia et al., 1994, and Shallcross et al., 1988.
These models can be divided into two main categories based on their complexities.
Models of the first category assume that all or part of the system behaves ideally, i.e.,
the activity coefficients in either the solution phase or solid phase are equal to unity. In
the second category of LMA models, ideal assumptions are eliminated to provide a
more realistic representation of an ion exchange system. In this respect, the latter
category is more accurate in predicting ion exchange equilibria. Nonetheless, the two
categories represent the historical progression of the LMA model, from its simple
formulations in the early years to the complex robust model in modern times. A
summary of the most important contributions in developing the LMA model for ion
exchange equilibria is presented in Table 3-4.
The LMA in its simplest form was first attempted by Gans in 1913 (Helfferich, 1995).
Gans defined equilibrium according to the selectivity coefficient depicted in equation 3-
5. The same approach was later taken by Subba Rao and David, (1957), Peironi and
Dranoff, (1963) and Klein et al., (1967) for ion exchange resins. Furthermore Subba
Rao and David (1957) found that the selectivity coefficients calculated for a number of
solution compositions varied by up to 34% from an average value. The largest
deviations were noted for the lowest and highest solution normalities (Subba Rao and
David, 1957).
Kielland (1935) made the first attempt to describe ion exchange equilibria by a
theoretical equation which accounts for the deviations from ideal behaviour. This work
was one of the first to introduce a method for determining the solid-phase activity
coefficients. Kielland (1935) followed the method of Vanselow (1932) in treating the
heteroionic form of the ion exchanger (containing both competing species A and B) as a
solid solution of the components AZ and BZ (Z symbolises a structural unit of the
zeolite). In this work changes in the water content of the resin and electrolyte sorption
were neglected. These two simplifying assumptions hold fairly well for the zeolites
then in use. Kielland (1935) assumed a van der Waals type of equation of state and
obtained a relation for the rational activity coefficients in the solid (relationship only
80
Table 3-4: A summary of the various forms of the Law of Mass Action (LMA) and
methods used for calculating solution phase and resin phase activity coefficients.
81
holds for univalent ions). The relationship obtained was able to present most
experimental data obtained with zeolites. For ion exchangers however it proves
inadequate due to the assumptions leading to the equations (Helfferich 1995).
Bajpai and co-workers (1973) noted that the selectivity coefficient as defined by
equation 3-6 varied widely with solution phase composition. The variation with
external solution composition was less in the case of homovalent ions compared to
heterovalent systems. Like the work of Elprince and Babcock (1975), Bajpai et al.,
(1973) assumed an ideal solution phase and only accounted for deviation in the solid
phase. Bajpai et al., (1973) used an expression derived by Hogfeldt et al., (1950) to
determine an expression for the resin-phase activity coefficients. This method was
based on correlating the equilibrium constant and the selectivity coefficients of the resin
by integrating experimentally derived values for the selectivity coefficient (Hogfeldt et
al., 1950). Elprince and Babcock (1975) used the Wilson equation to determine the
resin phase activity coefficients. From these studies it was concluded that there was a
strong dependence of resin phase activity coefficients on ionic composition and that the
ratio γ B / γ A does not remain constant over the entire composition range.
The dependence of the resin phase activity coefficients on ionic composition is depicted
in Figure 3-8 for the Mn2+/Na+ binary system. The activity coefficients on the resin
phase for both Mn2+ and Na+ are plotted as a function of the ionic fraction Mn2+ on the
resin phase. For both ions, the activity coefficients vary considerably at the extremes of
resin phase composition. Similar trends have been observed in other studies (Valverde
et al., 2001 and 2002), and highlight the importance of evaluating the resin phase
activity coefficients.
Unlike studies that modified the selectivity coefficient to account for resin phase non-
idealities, Kataoka and Yoshida, (1980) proposed a model that corrects the selectivity
coefficient for non-idealities in the liquid phase. The model gave rise to two important
limitations. First, the predictions made by the model may be inaccurate when the initial
fraction of one of the counterions in the liquid phase is much less than those of the other
counterions. Second, the model implies that these corrected selectivity coefficients for
each binary pair must be independent of composition.
82
2+
Figure 3-8: The variation of the activity coefficients of Mn and Na+ in the resin
phase as a function of the equivalent fractions of the ions in the resin for the
Mn 2+ / Na + binary system (adapted from Bajpai et al., 1973).
One of the most comprehensive efforts of determining equilibrium constants for ion
exchange systems was that of Smith and Woodburn (1978). In their work, non-
idealities in the solution and resin phases were taken into account. The Debye- Hückel
equation was used to evaluate the activity coefficients of ions in the solution phase
while the Wilson equation was applied to correlate activity coefficients of the resin
phase. Similarly the model proposed by Sengupta and Paul (1985) also recognizes that
both phases in an ion exchange system may be non-ideal. In the Sengupta and Paul
model the solution phase activity coefficients were determined by evaluating specific
ionic interactions and, the resin phase activity coefficients were determined according to
the method proposed by Gaines and Thomas (1953).
83
The developments that have occurred since then have mostly been refinements to the
existing definition of the equilibrium constant defined in equation 3-5. Shallcross and
co-workers (1988), introduced the Pitzer equation as a more accurate method of
estimating the solution phase activity coefficients. Furthermore they determined the
exact concentration of free ions in solution according to a method proposed by Kester
and Pytowicks (1969). The Wilson parameters and the equilibrium constant were
estimated simultaneously by regressing experimental equilibrium data. This approach
for modeling equilibrium data has since been used by other workers in the field such as
Valverde et al., (2002 and 2001), Shehata et al., (2000) and de Lucas Martinez et al.,
(1994).
As observed by Allen et al., (1989) and confirmed by the work of Mehablia et al.,
(1992), the values for the equilibrium constant and the two binary interaction
parameters, whilst fitting the experimental data, are highly sensitive to slight changes in
the experimental data. Mehablia et al., (1994) developed a model, where the
equilibrium constant is calculated independently from the binary interaction parameters.
In this model the equilibrium constant was first calculated using the Gaines and Thomas
(1953) approach. Decoupling these parameters yields a value for the equilibrium
constant that is independent of the solution concentration. A drawback of the Mehablia
model (Mehablia et al., 1994) is that the method requires data extending over the entire
isotherm. Ioannidis et al., (2000) proposed a method by which the Gaines and Thomas
approach is modified to represent cases where the data points on an isotherm do not
extend over the whole range of the liquid phase mole fractions, or the data are
inaccurate at the limits of the range. The model is based on the assumptions of constant
exchange capacity, at constant normality, and constant volume of pore water within the
resin.
84
interactions or ion-solvent interactions. In the Debye-Hückel model, departures from
ideality are accounted for by the long range and strength of the coulombic interaction
between ions. The fundamental Debye-Hückel expression derived from this theory,
known as the limiting law, is expressed as,
log γ i = − A z + z − I (3-12)
1
I = ∑
2 i
z i2 ci (3-13)
A z+ z− I
log γ i = − (3-14)
1 + Bai I
where A and B are the Debye-Hückel parameters, ai is the effective diameter of the
hydrated ion, and I is the ionic strength of the solution phase in molar units. Equation
3-14 is suitable for moderately dilute solutions of up to 0.1 mol/kg (Hovath, 1985). The
numerator of equation 3-14 gives the effect of the long range coulomb forces while the
denominator shows how these are modified by the short-range interactions between
ions. This assumes that the ions in question are non-deformable spheres of equal radii
(Robinson and Stokes, 1965). Furthermore it neglects the short-range interactions
between ions and solvent molecules as well as the short range interaction between ions.
Guggenheim (1935) included an additional term to the Debye-Hückel extended law to
85
incorporate these additional interactions. The Guggenheim extended model is
represented by the following equation,
A z+ z− I
log γ i = − + bI (3-15)
1 + Bai I
Although they assumed ideal solution behaviour in their model, Elprince and Babcock
(1975) were the first to suggest the use of the Pitzer theory for determining solution
phase activity coefficients. The Pitzer theory for estimating activity coefficients for
single ions in aqueous solutions incorporates terms that relate to both binary and ternary
interaction between the ions, as well as incorporating the Debye-Hückel electrostatic
term (Pitzer 1973, 1979). Shallcross et al., (1988) were the first to use the Pitzer theory
to determine solution phase activity coefficients for several cation exchange systems.
The constants required in the Pitzer equation are tabulated for a number of ions (Pitzer,
[
1979). However for ions such as Au(CN)2 ] −
where such values are not available, an
estimate must be made from similar ions. Obviously, such a process will counter any
advantage that the Pitzer model may have over other simpler models such as Debye-
Hückel. Shallcross, et al., (1988) compared the objective function for each component
in a ternary system (Na, Ca, Mg) calculated by using either the Debye-Hückel extended
law or the Pitzer equation to estimate the activity coefficients in the liquid phase. The
objective function expresses the sum of deviations between experimental data and
calculated values, for a set of data. As shown in Table 3-5, the Pitzer theory does not
appear to yield results with a statistically significant improvement over the extended
Debye-Hückel relation used by previous workers.
86
Table 3-5: Comparison between component objective functions determined using two
different methods to evaluate the solution phase activity coefficients (Shallcross et al.,
1988).
87
3.5.3.2 Resin phase activity coefficient
The Wilson equations (Wilson, 1964) are commonly used for correlating the resin phase
activity coefficients. The original derivation of the Wilson equation was based on
athermal solution theory and was successfully applied to vapour-liquid equilibria (Smith
et al., 2001). Analogous to an athermal system, the heats of mixing in an ion exchange
system are small, making the Wilson equation particularly appropriate. The Wilson
equation is defined as,
ΔG M ⎧M ⎫
= ∑ y mi ln ⎨∑ Λ ij y mi ⎬ (3-16)
RT i =1 ⎩ j =1 ⎭
where ΔG refers to the excess Gibbs free energy, R is the gas constant, T is the absolute
temperature, ymi is the mole fraction of species i in the solid phase, M is the number of
counterion species in the solid phase, and Λ ij is the Wilson binary interaction parameter
resin phase, Λ ij = 1 = Λ ji for all i and j. Therefore deviation of these parameters from
Equation 3-17 can be used to calculate the resin phase activity coefficient, γ i , provided
that the binary interaction parameters are known. One of the earliest studies on the
application of equation 3-17 on an ion exchange system was that of Elprince and
Babcock (1975). In their work equilibrium composition of the ternary systems
Na + / Cs + / Rb + and Na + / K + / Cs + were determined by correlating resin phase activity
coefficients using the Wilson equation and by assuming that the values for the
equilibrium constants were known independently. Later, Smith and Woodburn (1978)
used the Wilson equation to correlate activity coefficients for SO 24- / NO3- , SO 24- / Cl - and
88
Cl - / NO 3- binary systems. The model proposed by Smith and Woodburn (1978) allows
the equilibrium constant to be estimated along with the Wilson binary interaction
parameters.
A number of attempts have been made to reduce the number of parameters required to
model ion exchange equilibria. Allen et al., (1989) postulated a relationship between
the Wilson interaction parameters. The response surface produced by plotting the
values of Λ ij as a function of Λ ji produced a long, curved valley of approximately
Λ ij Λ ji = 1 (3-18)
Equation 3-18 was found to be a good first approximation of the observed response
curve for the binary ion exchange system between sodium and potassium (Allen et al.,
1989). The reciprocal relation in equation 3-18 between the Wilson parameters allows
the characterisation of binary equilibria with only two parameters: an equilibrium
constant and a single Wilson parameter. Whilst Allen et al., (1989) found that data
obtained with only two parameters did not cause a statistically significant reduction in
the quality of the fit, the validity of the reciprocity relation is questionable. As
illustrated in Table 3-6, the product of the interaction parameters obtained for a variety
of binary systems deviate widely from unity. Ioannidis et al., (2000) concluded that the
assumptions made in the derivation of equation 3-18 limits the flexibility of the model
for the representation of experimental data.
Λ ij Λ jk Λ ki = Λ ji Λ kj Λ ik (3-19)
89
Table 3-6: The reciprocity relation proposed by Allen et al., (1989) applied to a
number of binary systems.
90
3.6 REFERENCES
Allen, R. M., Addison, P. A., and Dechapunya, A. H., The characterization of binary
and ternary ion exchange equilibria. Chem. Eng. J., 1989. 40: p. 151-158.
Bajpai, R. K., Gupta, A. K., and Rao, G. M., Binary and ternary ion-exchange
equilibria. Sodium-cesium-manganese-Dowex 50w-x8 Systems. The Journal of
Physical Chemistry, 1973. 77: p. 1288-1293.
Brajter, K. and Miazek, I., Investigations on the effect of aqueous acetone medium on
separation of metal ions on Chelex 100 ion-exchanger. Talanta, 1981. 28: p. 759-
764.
Burstall, F. H. and Wells, R. A., Recovery of gold from cyanide solution by ion
exchange. in Ion exchange and its applications, London. Society of Chemical
Industry, 1955: p. 83-92.
Chu, B., Whitney, D. C., and Diamond, R. M., On anion-exchange resin selectivities.
Journal of Inorg. Nucl. Chem., 1962. 24: p. 1405-1415.
de Lucas Martinez, A., Diaz, J. Z., and Canizares, P. C., Ion-exchange equilibrium in a
binary mixture. Models for its characterization. Int. Chem. Eng., 1994. 34: p. 486-
497.
de Lucas Martinez, A., Canizares, P., and Diaz, J. Z., Binary ion exchange equilibrium
for Ca2+, Mg2+, K+, Na+ and H+ ions on Amberlite IR-120. Chem. Eng. Technol.,
1993. 16: p. 35-39.
de Lucas Martinez, A., Zarca, J., and Canizares, P., Ion-exchange equilibrium of Ca2+,
Mg2+, K+, Na+ and H+ Ions on Amberlite IR-120: Experimental determination and
theoretical prediction of the ternary and quaternary equilibrium data. Separation
Science and Technology, 1992. 27: p. 823-841.
Fessler, R. G. and Strobel, H. A., Nonaqueous ion exchange. II. Univalent cation
exchange in alcohols and methanol–water, ethanol–water, and methanol–ethanol
mixtures. Journal of Physical Chemistry, 1963. 67: p. 2562-2568.
Fleming, C. A. and Monhemius, A. J., On the extraction of various base metal chlorides
from polar organic solvents into cation and anion exchange resins.
Hydrometallurgy, 1979. 4: p. 159-167.
91
Gaines, G. L. and Thomas, H. C., Adsorption studies on clay minerals. II. A formulation
of the thermodynamics of exchange adsorption. The Journal of Chemical Physics,
1953. 21: p. 714-718.
Hassan, E. A., Abo El-Magd, A. S., Kamal, F H., El-Hady, M. F., and Ismail, N. M.,
Cation exchange distribution studies of Ca(II), Mg(II), Co(II), Cu(II), and Fe(III)
ions in organic solvent-HCl media. Indian J. Chem., 1981. 20A: p. 598-599.
Hogfeldt, E., Ekedhal, E., and Sillen, L. G., Acta. Chem. Scan., 1950. 2: p. 1471.
Hovath, A. L., Handbook of aqueous electrolyte solutions. 1985, Ellis Horwood Ltd.,
London.
Ioannidis, S., Anderko, A., and Sanders, S. J., Internally consistent representation of
binary ion exchange equilibria. Chemical Engineering Science, 2000. 55: p. 2687-
2698.
Kataoka, T. and Yoshida, H., Ion exchange equilibria in ternary systems. J. of Chem.
Eng. Of Japan, 1980. 13: p. 328-330.
Kester, D. R. and Pytkowicz, R M., Sodium, magnesium and calcium sulfate ion-pairs
in seawater at 25C. Liminology and Oceanography, 1969. 14: p. 686-692.
Klein, G., Tondeur, D., and Vermeulen, T., Multicomponent ion exchange in fixed beds.
Ind. Eng. Chem. Fundam., 1967. 6: p. 339-351.
Korkisch, J. and Ahluwalia, S. S., Separation of large amounts of iron(III) from cobalt,
nickel, and aluminium by combined ion exchange-solvent extraction. Analytica
Chimica Acta., 1966. 34: p. 308-313.
Lukey, G. C., Van Deventer, J. S. J., and Shallcross, D. C., Selective elution of copper
and iron cyanide complexes from ion exchange resins using saline solutions.
Hydrometallurgy, 2000. 56(2): p. 217-236.
Lukey, G. C., Van Deventer, J. S. J., Chowdhury, R. L., and Shallcross, D. C., The
effect of salinity on the capacity and selectivity of ion exchange resins for gold
cyanide. Minerals Engineering, 1999. 12(7): p. 769-785.
Marcus, Y., Ion Properties. 1997, Marcel Dekker Inc., New York.
Marcus, Y., Ion Solvation. 1985, John Wiley and Sons, New York.
92
Martin, M. W., Selectivity of some ion chromatography stationary phases for small
anions in solvent-water mixtures with hydroxide. Microchemical Journal, 1999.
62: p. 203-222.
Mehablia, M., Shallcross, D. C., and Stevens, G. W., Ternary and quaternary ion
exchange equilibria. Solvent Extraction and Ion Exchange, 1996. 14: p. 309-322.
Mehablia, M., Shallcross, D. C., and Stevens, G. W., Prediction of multicomponent ion
exchange equilibria. Chemical Engineering Science, 1994. 49: p. 2277-2286.
Mehablia, M., Shallcross, D. C., and Stevens, G. W., The Wilson equation applied to the
non-idealities of the resin phase of multicomponent ion exchange equilibria. In,
Proceedings of IEX ’92, International Conference on Ion Exchange Advances,
Cambridge, Slater, M. J., Ed, 1992. United Kingdom.
Melis, S., Markos, J., Cao, G., and Morbidelli, M., Multicomponent equilibria on ion-
exchange resins. Fluid Phase Equilibria, 1996. 117: p. 281-288.
Melis, S., Cao, G., and Morbidelli, M., A new model for the simulation of ion exchange
equilibria. Ind. Eng. Chem. Res., 1995. 34: p. 3916-3924.
Parker, A. J., Clare, B. W., and Smith, R. P., Solvation of ions. Some applications IV A
novel process for the recovery of pure silver from impure silver chloride.
Hydrometallurgy, 1979. 4: p. 233-245.
Parker, A. J., Protic – dipolar aprotic solvent effects on rates of bimolecular reactions.
Chemical Reviews, 1969. 69(1): p. 1 – 32.
Pieroni, L. J. and Dranoff, J. S., Ion exchange equilibria in a ternary system. A.I.Ch.E.
Journal, 1963. 9: p. 42-45.
Pietrzyk, D. J., Ion Exchangers, in Chromatographic Science Series, 1990. 47: p. 585
93
Pitzer, K. S., Theory: ion interaction approach, In Activity Coefficients in Electrolyte
Solutions, Pytkowicz (Ed)., CRC Press, Florida. 1979. 1: p. 157-208.
Riveros, P. A., Selectivity aspects of the extraction of gold from cyanide solutions with
ion exchange resins. Hydrometallurgy, 1993. 33(1-2): p. 43-58.
Sengupta, M. and Paul, T. B., Multicomponent ion exchange equilibria. I. Zn2+- Cd2+-
H+ and Cu2+-Ag+-H+ on Amberlite IR 120. Reactive Polymers, 1985. 3: p. 217-
229.
Shallcross, D. D., Herrmann, C. C., and McCoy, B. J., An Improved Model for the
Prediction of Multicomponent Ion Exchange Equilibria. Chem. Eng. Sci., 1988.
43: p. 279-288.
Shehata, F. A., El-Kamash, A. M., El-Sorougy, M. R., and Aly, H. F., Prediction of
multicomponent ion-exchange equilibria for a ternary system from data of binary
systems. Separation Science and Technology, 2000. 35: p. 1887-2000.
Smith, J. M., Van Ness, H. C., and Abbott, M. M., Introduction to chemical engineering
thermodynamics, 6th ed.; 2001, McGraw-Hill, New York.
Strelow, F. W. E., Victor, A. H., and Weinert, C. H., Separation of copper(II) from
uranium(VI) and many other elements by cation-exchange chromatography in
acetone-hydrobromic acid media. Analytica Chimica Acta, 1974. 69: p. 105-110.
94
Valverde, J. L., de Lucas, A., Gonzalez, M., and Rodriguez, J. F., Equilibrium data for
the exchange of Cu2+, Cd2+, and Zn2+ ions for H+ on the cationic exchanger
Amberlite IR-120. J. Chem. Eng. Data, 2002. 47: p. 613-617.
Valverde, J. L., de Lucas, A., Gonzalez, M., and Rodriguez, J. F., Ion-exchange
equilibria of Cu2+, Cd2+, Zn2+, and Na+ ions on the cationic exchanger Amberlite
IR-120. J. Chem. Eng. Data, 2001. 46: p. 1404-1409.
Valverde, J. L., de Lucas, A., and Rodriguez, J. F., Comparison between heterogeneous
and homogenous MASS action models in the prediction of ternary ion-exchange
equilibria. Ind. Eng. Chem. Res., 1999. 38: p. 251-259.
Wilson, G. M., Vapor-liquid equilibrium. XI. A new expression for the excess free
energy of mixing. Journal of the American Chemical Society, 1964. 86(2): p. 127-
30.
95
CHAPTER 4
4. ION EXCHANGE EQUILIBRIA FOR Au(CN) 2 − /Cl − , [ ]
[Au(CN) ]− /SCN − AND SCN − /Cl− IN AQUEOUS SOLUTION ............ 96
2
Figure 4-1: Representative kinetic data depicting the length of time required for the
loading process to reach equilibrium: 1, [Au(CN) 2 ] (B) / Cl − (A) ; 2,
−
303 K and for various total solution concentrations: (A), The solid curve
represents the correlation of the data with equation 4-10 (3-parameter
regression); (B), Expanded view of the ion-exchange isotherm shown in
(A). .............................................................................................................110
Figure 4-4: Ion exchange isotherms for [Au(CN) 2 ] (A) / SCN − (B) in aqueous solution
−
at 303 K and for various total solution concentrations. The solid curve
represents the correlation of the data with equation 4-10 (3-parameter
regression). .................................................................................................111
Figure 4-5: Ion exchange isotherms for SCN − (B) / Cl − (A ) in aqueous solution at 303 K
and for various total solution concentrations. The solid curve represents the
correlation of the data with equation 4-10 (3-parameter regression). ........111
Figure 4-6: The selectivity of ion exchange resin for the gold cyanide complex as a
function of the ionic refractivity of the displaced halide anion (adapted from
Aveston et al., 1958). .................................................................................114
Figure 4-7: The integration method for the determination of KAB as a function of the
total solution concentration for the SCN − (B) / Cl − (A ) binary system (2-
parameter regression). ................................................................................120
1
Figure 4-8: Variation of the activity coefficients of [Au(CN)2 ] and Cl − in the resin
−
Figure 4-9: Variation of the activity coefficients of [Au(CN)2 ] and SCN − in the resin
−
Figure 4-10: Variation of activity coefficients of SCN − and Cl − in the resin phase as
a function of equivalent ionic fraction of SCN − in the resin phase for the
SCN − / Cl − binary system. ........................................................................120
Table 4-5: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria obtained
at 303 K. .....................................................................................................116
Table 4-6: The variation of KAB as a function of the total solution concentration using
the 2-parameter regression method. ...........................................................116
Table 4-7: The variation of KAB as a function of the total solution concentration using
the 3-parameter regression method. ...........................................................117
2
4. ION EXCHANGE EQUILIBRIA FOR [Au(CN) ]− /Cl− , 2
SOLUTION
4.1 INTRODUCTION
In this chapter, binary ion exchange equilibria for [Au(CN) 2 ]− / Cl − ,
[Au(CN) 2 ]− / SCN − and SCN − / Cl − in aqueous media at 303 K using commercially
available Purolite A500 as the ion exchanger are presented. These systems are relevant
to an ion exchange process in which [Au(CN)2 ] is loaded onto the Cl − -form of the ion
−
exchanger, followed by elution of [Au(CN)2 ] using SCN − as the counterion, and the
−
subsequent regeneration of the resin to the Cl − -form. The observed selectivity of the
resin for the three counterions [Au(CN)2 ] , SCN − and Cl − is considered in terms of
−
factors that effect the interaction of the counterions with the functional group on the
resin, including the influence of hydration. The experimental data are correlated using
the Law of Mass Action modified with activity coefficients to determine the equilibrium
constant for each binary system.
96
4.2 EXPERIMENTAL SECTION
4.2.1 Materials
Potassium aurocyanide (99%) and potassium thiocyanate (99%) were supplied by EBS
and Associates and Sigma-Aldrich, respectively. The anion exchange resin, Purolite
A500, was supplied by Purolite International. Purolite A500 is a Type 1 strong base,
divinylbenzene, macroporous resin in chloride form. The principal properties of the ion
exchange resin are listed in Table 4-1. Prior to use, the resin was soaked in distilled
water for 24 hours to stabilize resin swelling. Hydrated ferric nitrate (> 98%), used in
the determination of SCN − in solution, was supplied by Asia Pacific Specialty
Chemicals. All stock solutions were prepared by volume (1000 ± 0.4 mL) with distilled
water. The accuracy of weighing was ± 0.0001 g.
Ion exchange equilibria for the [Au(CN) 2 ] / Cl − and SCN − / Cl − binary systems were
−
determined by contacting the Cl − -form of the resin with a stock solution containing a
fixed concentration of [Au(CN)2 ] or SCN − . A 5 ± 0.2 mL (wet and settled volume)
−
sample of the resin was measured and filtered under vacuum to remove excess moisture.
The resin was placed in a 250 mL glass container followed by the addition of 100 ± 1
mL of the relevant stock solution. The glass container was sealed and placed in a
temperature-controlled orbital shaker (Paton Scientific, 01-2116) for which the
temperature was maintained to ± 0.5 K. The rotation speed of the orbital shaker was set
at 200 orbits per min in all experiments. In all experiments a minimum duration of 8
hours was used for determining equilibrium concentrations in the solution phase. The
experimental procedure used to establish the time to reach equilibrium is outlined in
section 4.2.2.1.
97
Table 4-1: Properties of Purolite A500.
Property Description
Appearance Spherical particles
Type of matrix Styrene Divinylbenzene
Functional group structure R(CH)3N+
Effective size 0.50 mm (average)
Moisture content, as supplied 53-58%
pH range 0-14
Maximum operating temperature 100°C
Total exchange capacity 1.3 eq/L
After equilibrium was established, the solution phase was separated from the resin
phase and analysed for [Au(CN)2 ] or SCN − . From knowledge of the change in
−
concentration of the anion in the solution phase, it was possible to calculate the
composition of [Au(CN)2 ] or SCN − on the resin phase at equilibrium. A mass
−
balance was used to determine the equilibrium compositions of the resin and solution
The concentration of [Au(CN)2 ]
−
phases with respect to Cl − . in solution was
determined by analysis for Au using inductively coupled plasma - atomic emission
spectroscopy. The standard uncertainty of the analysis was ± 1%. The analysis for
SCN − in solution was carried out in accordance with ASTM D 4193-95. This standard
method is based on the reaction of SCN − with the ferric ion at pH < 2 to form a
coloured complex. The intensity of the solution is determined by measuring its
absorbance at 460 nm using a UV-Visible spectrophotometer. The uncertainty of the
analysis for SCN − was ± 5%.
The partially loaded resin was then contacted with a fresh 100 mL quantity of the stock
solution and allowed to reach equilibrium. The process of draining the solution phase at
equilibrium, followed by the addition of a fresh quantity of stock solution to the same
sample of resin, was repeated until no further loading of [Au(CN)2 ] or SCN − occurred
−
on the resin. This condition was confirmed when the change in the concentration of the
anion in the solution phase was found to be negligible.
98
Ion exchange equilibria for the [Au(CN) 2 ] / SCN − binary system were determined
−
using resin which had been fully loaded with SCN − . The resin was prepared by
contacting 100 mL of Purolite A500 resin originally in the Cl − form with an excess
quantity of potassium thiocyanate. The resin and solution mixture were agitated in the
shaker incubator for over 48 hours. Samples were taken before and after to ensure
loading of SCN − from solution. For the [Au(CN) 2 ] / SCN − binary system, a 5 mL
−
sample of the SCN − -form of the resin was repeatedly contacted, in the manner
described above, with 100 mL of a stock solution containing a fixed concentration of
[Au(CN)2 ]− .
By gradually loading the resin to capacity in the manner described it was possible to
construct an ion-exchange isotherm encompassing a wide range of composition for each
phase. Ion exchange equilibria are expressed in terms of equivalent ionic fractions for
the solution and resin phases (see equations 4-3 and 4-4). The experimental ionic
fraction data that are reported represent the average of duplicate runs, with an
uncertainty of ± 5%.
The effect of the total solution concentration on the ion-exchange isotherms was
determined by contacting the resin with stock solutions containing different total
concentrations of the loading anion. For the [Au(CN) 2 ] / Cl − and [Au(CN) 2 ] / SCN −
− −
binary systems, ion-exchange equilibria were obtained with stock solutions containing
[Au(CN)2 ]− at concentrations of 0.0025 M, 0.0051 M and 0.0066 M. In the case of the
The duration of time required to establish equilibrium for the [Au(CN) 2 ] / Cl − and
−
SCN − / Cl − binary systems was determined by contacting the Cl − -form of the resin
with a stock solution containing a fixed concentration of [Au(CN)2 ] or SCN − for 24
−
hours. Similarly, resin in the SCN − -form was used for the [Au(CN) 2 ] / SCN − system.
−
99
Each ion exchange system was exposed to a 5 ± 0.2 mL (wet and settled volume)
sample of the resin, measured and filtered under vacuum to remove excess moisture. 1
mL samples of the solution phase were taken at regular intervals to confirm the
attainment of equilibrium between the resin and solution phases for each of the three
binary systems. For the [Au(CN) 2 ] / Cl − and [Au(CN) 2 ] / SCN − binary systems, time
− −
of 0.01 M. In the case of the SCN − / Cl − binary system, a stock solution containing
SCN − at a concentration of 0.03 M was used. The concentrations of Au and SCN −
were analysed according to methods discussed above.
Representative kinetic data for the three binary systems are presented in Figure 4-1.
The data are plotted in terms of the concentration of the loading ion (B) in the resin
phase as a function of time. For the [Au(CN) 2 ] / Cl − and [Au(CN) 2 ] / SCN − binary
− −
systems the loading ion is [Au(CN)2 ] , while in the SCN − / Cl − binary system the
−
loading ion is SCN − . It can be seen that equilibrium is established within a period of
four hours for all three binary systems. Similar kinetic experiments were performed at
different concentrations of the loading ion, encompassing the range of concentrations
used in the determination of the isotherms. Furthermore, kinetic studies were also
conducted on partially loaded resins. From these studies it was confirmed that the
duration of time for equilibrium did not vary significantly with the solution
concentration or the resin loading. Nonetheless, equilibrium samples from solution
were taken after 8 hours for each binary system to ensure that equilibrium was
established.
from the [Au(CN) 2 ] / Cl − binary system, was analysed for Au content with an electron
−
microprobe (Cameca SX50). Samples of the partially and completely loaded resin
beads were set in an epoxy resin and sliced along their cross-section. The electron
100
Figure 4-1: Representative kinetic data depicting the length of time required for the
loading process to reach equilibrium: 1, [Au(CN) 2 ]− (B) / Cl − (A) ; 2,
101
probe microscope was then used to map the cross-sectional face of a resin bead for gold
content.
The microprobe was initially calibrated with a sample of pure gold and the
corresponding intensity was fixed at 100 weight percent gold. The instrument then
calculated the weight percent of gold in the sample resin bead. In this way, the
microprobe was able to map the cross-sectional surface of a resin bead for gold content.
Figure 4-2A presents the surface map of a representative resin bead, initially in the Cl −
form, partially loaded with [Au(CN)2 ] . Figure 4-2B presents a fully loaded resin bead
−
for the same binary system. The colours on the map are representative of the weight
percent (wt %) of gold, with red corresponding to the highest concentration and blue
corresponding to the lowest concentration. The distances on the x and y axis are shown
in micrometers (μm).
As expected the partially loaded resin in Figure 4-2A has a low concentration of gold
within its structure compared to the fully loaded resin shown in Figure 4-2B. For the
partially loaded resin, the concentration of gold is significantly higher at the center of
the bead compared to the edges closer to the surface. This suggests that the initial
loading within a resin bead occurs from the centre of the resin. The fully loaded resin
bead in Figure 4-2B shows that gold is evenly distributed throughout the resin bead.
These maps provide qualitative confirmation of the complete loading of the resin.
Similar results were obtained for the [Au(CN) 2 ] / SCN − binary system. Unfortunately
−
an analysis could not be conducted on the SCN − / Cl − system, as the electron probe
microscope is only sensitive to the concentration of metallic elements.
102
Cross-sectional views of resin beads loaded with [Au(CN)2 ] : (A),
−
Figure 4-2:
partially loaded resin; (B) completely loaded resin. The wt % of Au is represented by
the colour scheme shown on the legend.
μm
μm Wt % Au
103
4.3 DATA CORRELATION
As introduced in Chapter 3, the replacement of ion A on the resin phase (r) with ion B
from the solution phase (s) is governed by a reversible ion-exchange equation as
follows:
βA αr − + αB βs − ⇔ βA sα − + αB βr − (4-1)
At equilibrium, the solution and resin phases are composed of both ions in proportions
that reflect the selectivity of the resin. An ion exchange process can be defined by an
equilibrium constant in accordance with the Law of Mass Action modified with activity
coefficients (de Lucas Martinez et al., 1994; Shallcross et al., 1988):
( γ A cA ) β ( γ B q B ) α
K AB = (4-2)
( γ A q A ) β ( γ B cB ) α
where ci is the molar concentration of ion i in the solution phase, qi is the molar
concentration of ion i in the resin phase and KAB refers to the equilibrium constant for B
entering the resin phase and displacing A. The activity coefficients of ion i in the
solution phase and resin phase are denoted by γ i and γ i , respectively. Molar
concentrations in the resin phase are here calculated with respect to the wet and settled
volume of the ion-exchange resin. The use of equation 4-2 to describe the equilibrium
constant assumes that swelling-pressure effects are negligible (Helfferich, 1995).
As a rule, the equilibrium concentrations of ions in the resin and solution phases are
expressed in terms of equivalent ionic fractions (Helfferich, 1995). Equivalent
concentrations take in to account the charge of the exchanging species, and therefore
better represent an ion exchange system. However for competing ions of equal charge,
the equivalent ionic fraction is equal in value to the mole fraction. The molar
concentration terms in equation 4-2 can be replaced by equivalent ionic fractions by
employing the following relations:
z i ci
xi = (4-3)
N
104
z i qi
yi = (4-4)
Q
where xi and yi are the equivalent ionic fractions in the solution phase and resin phase,
respectively, zi is the valence of ion i, N is the total normality of exchanging ions in the
solution phase and Q is the capacity of the resin (equivalents/L). Substitution of
equations 4-3 and 4-4 into equation 4-2, and noting that α = β = 1 for all species in this
study, leads to the following expression for the equilibrium constant:
( γ A xA )( γ B yB )
K AB = (4-5)
( γ A yA )( γ B xB )
In the calculation of values of yi using equation 4-4, the capacity of the resin was set at a
value corresponding to the maximum amount of anion loaded onto a given form of the
resin, as determined experimentally using the procedure described previously. The
experimental values of the resin capacity obtained in this way are presented in Table 4-
2. As such, these values represent the apparent capacity of the resin and agree closely
with the manufacturer's specification of 1.3 eq/L.
Strictly speaking, the individual ionic activity coefficients appearing in equation 4-5
cannot be determined experimentally. Kielland (1937) has pointed out, however, that
individual ionic activity coefficients in sufficiently dilute solutions may be calculated
from an extended form of the Debye-Hückel limiting law as follows:
A z i2 I
log γ i = − (4-6)
1 + Bai I
1
I = ∑
2 i
z i2 ci (4-7)
where A and B are the Debye-Hückel parameters, ai is the effective diameter of the
hydrated ion and I is the ionic strength of the solution phase in molar units. This semi-
empirical approach is often used for estimating mean ionic activity coefficients which
are measurable. The Debye-Hückel parameters appearing in equation 4-6 are equivalent
105
Table 4-2: Apparent capacities of the ion exchange resin.
System Apparent Capacity/(eq/L)a
[Au(CN) 2 ]− / Cl − 1.23
SCN − / Cl − 1.32
Table 4-3: Debye-Hückel parameters for aqueous solutions at 303 K and effective
hydrated diameters of [Au(CN)2 ] , SCN − and Cl − .
−
108ai/cm
-1/2 -7 -1/2 -1
A/(mol/L) 10 B/(mol/L) cm
[Au(CN)2 ]− SCN − Cl −
0.5161a 3.301a 4.5b 3.5b 3.0b
a b
Data are from Robinson and Stokes (1965). Data are from Kielland (1937). The value for
[Au(CN)2 ]− is the lower limit of the range (4.5 to 7) suggested for organic ions.
to those in the corresponding equation for the mean ionic activity coefficient and
depend on temperature and the dielectric constant of the solvent (Robinson and Stokes,
1965). The values of A, B and ai required in this study are listed in Table 4-3. Note,
cations are also included in the summation in equation 4-7.
The individual ionic activity coefficients in the resin phase were also calculated using a
semi-empirical approach involving the Wilson equations formulated for vapour-liquid
equilibria (Smith et al., 2001; Wilson, 1964). As discussed in Chapter 3, this method
has been applied successfully in the correlation of binary ion exchange equilibria for
cationic systems (Valverde et al., 2001; de Lucas Martinez et al., 1993; Shallcross et al.,
1988). The relevant equations for the exchange of ions A and B are:
yA y B Λ BA
ln γ A = 1 − ln ( y A + y B Λ AB ) − − (4-8)
y A + y B Λ AB y B + y A Λ BA
yB y A Λ AB
ln γ B = 1 − ln ( y B + y A Λ BA ) − − (4-9)
y B + y A Λ BA y A + y B Λ AB
106
where Λ AB and Λ BA are the Wilson parameters.
In view of the fact that Λ AB and Λ BA are adjustable parameters, it was not possible to
calculate the equilibrium constant directly from equation 4-5. Instead, KAB, Λ AB and
Λ BA for each binary system were regressed from the experimental data (3-parameter
regression). The working equation for this calculation procedure is obtained by
rearranging equation 4-5 as follows:
K AB γ A xB γ B
yB = (4-10)
K AB γ A xB γ B + (1 − xB ) γ A γ B
Equation 4-10 was used to generate calculated values of yB from the experimental
values of xB and yB and for given values of KAB, Λ AB and Λ BA . The optimum values of
KAB, Λ AB and Λ BA , for a given isotherm, were obtained by minimising the sum of
squared relative deviations (SSRD) with respect to the resin phase composition:
2
⎛ y calc − y exp
M
⎞
SSRD = ∑ ⎜⎜ B exp B ⎟⎟ (4-11)
i =1 ⎝ yB ⎠i
where M is the number of data points and y Bcalc and y Bexp are the calculated and
experimental values of the equivalent ionic fraction of ion B in the resin phase,
respectively. In the discussion which follows, the average absolute relative deviation
(AARD) is defined as:
Alternatively, Mehablia et al., (1994) have described a method in which the equilibrium
constant is first calculated independently and then used in the regression of Λ AB and
Λ BA from the experimental data (2-parameter regression). The equilibrium constant is
calculated using,
107
1
ln K AB = ( z A − z B ) + ∫ ln(λ AB ) dy B (4-13)
0
β α
⎛γ c ⎞ ⎛ yB ⎞
λ AB = ⎜⎜ A A ⎟⎟ ⎜⎜ ⎟⎟ (4-14)
⎝ yA ⎠ ⎝ γ B cB ⎠
where λ AB is the equilibrium quotient defined in relation to equation 4-1. Thus, the
equilibrium constant can be estimated by integrating the area under a plot of ln( λ AB )
versus yB. The Wilson parameters are then regressed using the procedure described
above in which equation 4-10 is used to generate calculated values of yB, followed by
the minimisation of the objective function in equation 4-11. The advantage of this
method is that it decouples the intercorrelation between the equilibrium constant and the
pair of Wilson parameters.
Calculated values of the equivalent ionic fraction of ion B, y Bcalc , were obtained from a
program designed using FORTRAN-77. The input values for the program are the
experimental equivalent ionic fractions of ion B for the solution and resin phase (xB and
yB), the ionic strength (I), and the constants for the Debye-Hückel equation (A, B and
ai). Initially, the program assumes trial values for KAB, Λ AB and Λ BA in the 3-
parameter regression. Using the trial values for Λ AB and Λ BA , activity coefficients are
determined for the resin phase using Wilson equations 4-8 and 4-9. The solution phase
activity coefficients are determined using I, as an input value. Values of the activity
coefficients for the resin phase and the solution phase, along with the initial guess for
KAB, are used in equation 4-10 to determine y Bcalc .
In the case of the two parameter model, KAB values, previously calculated using
equation 4-13 are used as constants. The program then calculates the SSRD between
y Bcalc and yBexp using equation 4-11. If the SSRD is not a minimum, the program repeats
the above steps by changing the initial guesses. The search method for determining
KAB, Λ AB and Λ BA is a nonlinear regression method based on the Marquardt algorithm.
Once a minimum value for the SSRD is calculated, it is assumed that the best set of
108
values for the three parameters KAB, Λ AB and Λ BA is obtained. As an example, the
program code used for the 3-parameter regression of the [Au(CN) 2 ] / Cl − binary
−
aqueous solution at 303 K are listed in Appendix 2 and presented graphically in Figures
4-3 to 4-5. The equivalent ionic fractions in Figures 4-3 to 4-5 are plotted with respect
to ion B, the loading anion, in accordance with the ion exchange process shown in
equation 4-1. The selectivity of the resin can be ascertained from the shape of the
isotherm. The resin is considered to be selective towards ion B if the isotherm lies
above the diagonal.
For the [Au(CN) 2 ] / Cl − system (Figure 4-3A) it can be seen that the resin is highly
−
selective for [Au(CN)2 ] (ion B). An expanded view of the data in Figure 4-3A is
−
provided in Figure 4-3B. The suitability of this type of ion exchange resin for the
recovery of [Au(CN)2 ] from leach solutions is particularly evident from the relatively
−
high loading that is attained when the resin is in contact with a dilute solution of
[Au(CN)2 ]− . The high affinity of the resin towards the gold cyanide complex can be
understood qualitatively in terms of the factors influencing selectivity as discussed in
Chapter 3.
Ion exchangers are generally selective for ions that can interact most strongly with their
charged sites. Ion-pair formation is the most likely mode of interaction between the
ions and these sites (Okada, 1997). In the absence of solvent effects, ion-pair formation
generally favours ions that can exert the strongest electrostatic field. Therefore, ions
with a small ionic radius or high charge density are likely to lead to strong ion-ion
interactions. Based on electrostatic properties alone, Cl − is expected to form the
strongest ion-pair with the functional group on the resin. However, this behaviour is not
109
Figure 4-3: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in aqueous solution at
−
303 K and for various total solution concentrations: (A), The solid curve represents the
correlation of the data with equation 4-10 (3-parameter regression); (B), Expanded view
of the ion-exchange isotherm shown in (A).
A
Error! Not a valid link.
B
Error! Not a valid link.
110
Figure 4-4: Ion exchange isotherms for [Au(CN) 2 ] (A) / SCN − (B) in aqueous solution
−
at 303 K and for various total solution concentrations. The solid curve represents the
correlation of the data with equation 4-10 (3-parameter regression).
Figure 4-5: Ion exchange isotherms for SCN − (B) / Cl − (A ) in aqueous solution at 303
K and for various total solution concentrations. The solid curve represents the
correlation of the data with equation 4-10 (3-parameter regression).
111
observed in the [Au(CN) 2 ] / Cl − binary system. This can be explained by considering
−
The water molecules contained in the hydration shells of a counterion increase the
distance between the resin functional group and the counterion. In the case of the
[Au(CN) 2 ]− / Cl − binary system, [Au(CN)2 ] is less hydrated compared to Cl − (Table
−
3-1). As a result [Au(CN)2 ] is better able to interact with the functional groups on the
−
resin. In addition, Okada (1997) suggests that the successful formation of an ion-pair
between a counterion and a charged site on the resin requires the removal of some water
molecules from the hydration shells of the counterion. This process, known as
desolvation, is energetically unfavourable and is less likely to occur for highly hydrated
ions such as Cl − . Thus, in aqueous environments, large or less hydrated ions are more
strongly retained by the resin phase (Okada, 1997; Martin, 1999).
For the [Au(CN) 2 ] / SCN − system (Figure 4-4), the resin is again more selective
−
towards [Au(CN)2 ] (ion B). The same reasoning considered for the [Au(CN) 2 ] / Cl −
− −
binary system can be applied here to explain the selectivity of the resin for [Au(CN)2 ]
−
over SCN − . Comparing Figures 4-3A and 4-4, it appears that SCN − is more successful
than Cl − as a counterion for the displacement of the gold cyanide complex. The
hydration numbers of thiocyanate and chloride indicate that the average number of
water molecules in the hydration shells of chloride is higher than that of thiocyanate.
An additional thermodynamic property which quantifies the degree of hydration of an
ion is the enthalpy of hydration, Δ hyd H . A comparison of the Δ hyd H values for Cl −
and SCN − in Table 4-4 also confirms that Cl − is more hydrated than SCN − .
Based on the idea that the selectivity of the resin is primarily influenced by the degree
of hydration of the counterions, it is therefore expected that the resin is selective for
SCN − in the SCN − / Cl − binary system. The isotherm for the SCN − / Cl − system
(Figure 4-5) clearly confirms this result. Overall, it can be generalised that the
selectivity of the resin for a given ion increases in the order Cl − < SCN − <
[Au(CN)2 ]− .
112
The selectivity sequence of the anion exchange resin can also be explained in terms of
the polarisability of the exchanging counterion. Aveston et al., (1958) studied ion
exchange between the gold cyanide complex and a number of counterions in aqueous
media. The selectivity coefficients calculated for the exchange processes were plotted
against the ionic refractivity of each counterion (Figure 4-6). The ionic refractivity is a
measure of the polarisability of an ion. For the series of halide anions, the selectivity of
the resin for the gold cyanide complex appears to decrease with increasing polarisability
of the counterion. Based on polarisability alone, F − appears to be the most suitable
[
anion for loading Au(CN)2 ]
−
into the resin while I − is the least suitable. Ionic
113
Table 4-4: Thermo-chemical properties of [Au(CN)2 ] , SCN − and Cl − .
−
KCl - -702.78 -
KSCN - -591.77 -
Cl − 1 -381 8.6
[Au(CN)2 ]− 0.20 - 37
a
Enthalpies of hydration data calculated from the lattice energies and enthalpy of solvation data for
potassium salts of the respective ions from Lide 1995. The enthalpies of hydration for the anions were
calculated by subtracting the value for potassium (-322 kJ/mol) according to Bockris and Reddy, 1970.
b
Ionic refractivity data obtained from Marcus 1997
Figure 4-6: The selectivity of ion exchange resin for the gold cyanide complex as a
function of the ionic refractivity of the displaced halide anion (adapted from Aveston et
al., 1958).
A comparison of the two methods used for determining KAB, Λ AB and Λ BA is presented
in Table 4-5. It can be seen that the independently calculated values of KAB are
comparable to those derived from the 3-parameter regression. Also, the correlation of
the data with the 3-parameter regression is slightly more accurate than that obtained
with the 2-parameter regression. The AARD with respect to yB, using either method,
range from 2 to 10% and this highlights that the calculated values of yB agree well with
the experimental values for all three systems. The deviations between experimental and
calculated values of yB are also presented in Appendix 2.
114
Optimised KAB values for binary systems were obtained by assuming that the isotherms
were independent of the total solution concentration. To verify this assumption, KAB
values were determined for each concentration. These values are presented in Tables 4-
6 and 4-7. The relative standard deviation (RSD) of the mean KAB, for the given range
of solution concentration, is also shown for each binary system. In the 2-parameter
regression, the equilibrium constant decreases with increasing concentration in all three
binary systems. This can be attributed to the integration method used in the
determination of KAB.
The integration method is dependent on having equilibrium data that cover the entire
range of resin composition (0 < yB < 1). However, as can be seen in Appendix 2, the
range of values of yB decreases as the total solution concentration increases, particularly
for the SCN − / Cl − system (Figure 4-7). Some underestimation of the equilibrium
constant occurs in this situation and the error increases as the total solution
concentration increases. Thus, the variation in the value of KAB does not truly reflect
the effect of concentration on this parameter.
On the other hand, the values of the equilibrium constant in the 3-parameter regression
exhibit less variation. For the [Au(CN ) 2 ] / Cl − and [Au(CN) 2 ] / SCN − systems, KAB
− −
115
Table 4-5: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria obtained
at 303 K.
System (B/A) KABa ΛAB ΛBA AARD (%)
3-parameter regression
[Au(CN) 2 ]− / Cl − 224.3 2.571 4 × 10-5 8.7
Table 4-6: The variation of KAB as a function of the total solution concentration using
the 2-parameter regression method.
Concentration a KAB
(mol/L) [Au(CN) 2 ]− / Cl − [Au(CN) 2 ]− / SCN − SCN − / Cl −
0.0025 207.0 13.60 11.48
(0.0086)
0.0051 169.1 11.84 7.880
(0.0172)
0.0066 134.7 11.00 6.661
(0.0224)
RSDb (%) 21.24 10.92 28.89
a − −
Values in parentheses represent the total solution concentrations in the SCN / Cl binary system.
b
The value represents the relative standard deviation (RSD) of the mean KAB for the three concentrations,
where RSD (%) = (standard deviation of KAB /mean KAB)
116
Table 4-7: The variation of KAB as a function of the total solution concentration using
the 3-parameter regression method.
Concentration a KAB
(mol/L) [Au(CN) 2 ]− / Cl − [Au(CN) 2 ]− / SCN − SCN − / Cl −
0.0025 210.5 14.53 14.51
(0.0086)
0.0051 265.1 14.63 15.55
(0.0172)
0.0066 252.2 14.41 22.23
(0.0224)
RSDb (%) 11.76 0.76 23.98
a − −
Values in parentheses represent the total solution concentrations in the SCN / Cl binary system.
b
The value represents the relative standard deviation (RSD) of the mean KAB for the three concentrations,
where RSD (%) = (standard deviation of KAB /mean KAB)
concentration compared to any other binary system. The RSD of the mean KAB for this
system (< 11%), in both regression methods, is significantly lower than those for the
other systems (12 - 29%).
According to the triangle rule (Mehablia et al., 1994), the equilibrium constants of two
of the three constituent binary systems of a ternary system can be used to predict the
equilibrium constant of the third binary system. For the [Au(CN)2 ] (C)/ SCN − (B)/
−
K AC = K AB K BC (4-15)
In the 3-parameter regression, the triangle rule is obeyed to within less than 4% (KAC =
224.3 and KABKBC = 232.5). In the 2-parameter regression, the product on the right
hand side of equation 4-15 is 16% lower than the experimental value of KAC. The
underestimation of the equilibrium constant in the latter method is partly attributed to
the value of the equilibrium constant for the SCN − / Cl − system. As noted above, some
underestimation of the equilibrium constant occurs with the integration method and this
is particularly significant for the SCN − / Cl − system.
117
The values for Λ AB and Λ BA obtained in this study fall generally within the range of 0
to 3. A similar range of values for the Wilson parameters has been reported for cation-
exchange systems (Valverde et al., 2001; de Lucas Martinez et al., 1993; Shallcross et
al., 1988). The dependence of the resin phase activity coefficients on the composition
of the loading ion, B, in the resin phase is shown in Figure 4-8 to 4-10. Generally, the
activity coefficient for each ion in a given binary system does not vary significantly
with increasing composition of the loading ion in the resin phase. However, the activity
coefficients of [Au(CN)2 ] in the [Au(CN) 2 ] / Cl − binary system and SCN − in the
− −
SCN − / Cl − binary system increase considerably at low ionic fractions of the loading
ion in the resin phase.
In the 2-parameter regression, the optimisation routine yielded some slightly negative
values for the Wilson parameters, even though positive values are required to maintain
consistency with the Wilson model. Allen et al., (1989) have noted that the product of
the two parameters is approximately equal to unity for a range of binary systems. This
relation is only evident for the parameters derived for the [Au(CN) 2 ] / SCN − system.
−
Significant deviations from unity have also been observed in other studies (Valverde et
al., 2001; Mehablia et al., 1994). The Hála constraint (Hala, 1972), which relates the
three pairs of Wilson parameters required in a ternary system, is not obeyed in either the
3-parameter or the 2-parameter regression.
The large value of KAB for the [Au(CN) 2 ] / Cl − system is consistent with the strong
−
that for the [Au(CN) 2 ]− / Cl − system. This reflects the reduced ability of [Au(CN)2 ] to
−
118
SCN − . Similarly, the regeneration of the resin from the SCN − -form to the Cl − -form
requires an excess concentration of Cl − (see Figure 4-5).
4.5 CONCLUSIONS
Ion exchange equilibria obtained for [Au(CN) 2 ]− / Cl − , [Au(CN) 2 ]− / SCN − and
SCN − / Cl − in aqueous solution at 303 K confirm the high affinity of Purolite A500 for
the [Au(CN)2 ] species. The selectivity of the resin for a given ion increases in the
−
order Cl − < SCN − < [Au(CN)2 ] . The ion exchange isotherm for each binary system
−
119
Figure 4-7: The integration method for the determination of KAB as a function of the
total solution concentration for the SCN − (B) / Cl − (A ) binary system (2-parameter
regression).
Figure 4-8: Variation of the activity coefficients of [Au(CN)2 ] and Cl − in the resin
−
phase as a function of the equivalent ionic fraction of [Au(CN)2 ] in the resin phase for
−
Figure 4-9: Variation of the activity coefficients of [Au(CN)2 ] and SCN − in the resin
−
phase as a function of the equivalent ionic fraction of [Au(CN) 2 ] in the resin phase for
−
Figure 4-10: Variation of activity coefficients of SCN − and Cl − in the resin phase as
a function of equivalent ionic fraction of SCN − in the resin phase for the SCN − / Cl −
binary system.
120
4.6 REFERENCES
Allen, R. M., Addison, P. A., and Dechapunya, A. H., The characterization of binary
and ternary ion exchange equilibria. Chem. Eng. J., 1989. 40: p. 151-158.
Aveston, J., Everest, D. A., and Wells, R. A., Adsorption of gold from cyanide solutions
by anionic resins. Journal of the Chemical Society, 1958: p. 231-239.
de Lucas Martinez, A., Diaz, J. Z., and Canizares, P. C., Ion-exchange equilibrium in a
binary mixture. Models for its characterization. Int. Chem. Eng., 1994. 34: p.
486-497.
de Lucas Martinez, A., Canizares, P., and Diaz, J. Z., Binary ion exchange equilibrium
for Ca2+, Mg2+, K+, Na+ and H+ ions on Amberlite IR-120. Chem. Eng.
Technol., 1993. 16: p. 35-39.
Kielland, J., Individual activity coefficients of ions in aqueous solutions. J. Am. Chem.
Soc., 1937. 59: p. 1675-1678.
Lide, D. R., Handbook of Chemistry and Physics. 76 ed. 1995, CRC Press, New York.
Marcus, Y., Ion Properties. 1997, Marcel Dekker Inc., New York.
Martin, M. W., Selectivity of some ion chromatography stationary phases for small
anions in solvent-water mixtures with hydroxide. Microchemical Journal, 1999.
62: p. 203-222.
Shallcross, D. D., Herrmann, C. C., and McCoy, B. J., An improved model for the
prediction of multicomponent ion exchange equilibria. Chem. Eng. Sci., 1988.
43: p. 279-288.
121
Smith, J. M., Van Ness, H. C., and Abbott, M. M., Introduction to chemical engineering
thermodynamics, 6th ed.; 2001, McGraw-Hill, New York.
Valverde, J. L., de Lucas, A., Gonzalez, M., and Rodriguez, J. F., Ion-exchange
equilibria of Cu2+, Cd2+, Zn2+, and Na+ ions on the cationic exchanger Amberlite
IR-120. J. Chem. Eng. Data, 2001. 46: p. 1404-1409.
Wilson, G. M., Vapor-liquid equilibrium. XI. A new expression for the excess free
energy of mixing. Journal of the American Chemical Society, 1964. 86(2): p. 127-
30.
122
CHAPTER 5
5. ION EXCHANGE EQUILIBRIA FOR Au(CN) 2 − /Cl − [ ]
[Au(CN) ]− /SCN − AND SCN − /Cl− IN MIXED SOLVENTS............... 124
2
Figure 5-1: The effect of the initial resin loading on the rate of ion exchange in the
SCN − / Cl − binary system in acetone-water mixtures: (A), 5.73 mol%
acetone; (B), 68.6 mol% acetone. ..............................................................131
Figure 5-2: Viscosities of mixed solvents containing acetone, DMSO and NMP
(adapted from Assarsson and Eirich, 1968; LeBel and Goring, 1962;
Washburn, 1929). .......................................................................................132
Figure 5-3: Dielectric constants of mixed solvents containing acetone, DMSO and
NMP (adapted from Marcus, 1985). ..........................................................132
Figure 5-4: The effect of the type of mixed solvent on the rate of ion exchange in the
[Au(CN) 2 ]− / Cl − binary system: (A), 4-6 mol% organic solvent; (B), 62-70
mol% organic solvent.................................................................................133
Figure 5-5: The effect of the type of counterion on the rate of ion exchange in acetone–
water mixtures: (A), 5.73 mol% acetone; (B), 68.6 mol% acetone. ..........134
Figure 5-6: Capacities of the cation exchange resin Amberlite IR-120 in various pure
solvents (adapted from de Lucas et al., 2001)............................................135
Figure 5-7: The loading of [Au(CN) 2 ] onto the Cl − -form of the resin in various
−
1
Figure 5-12: Ion exchange isotherms for [Au(CN) 2 ] (B) / SCN − (A) DMSO-water
−
2
5. ION EXCHANGE EQUILIBRIA FOR [Au(CN)2 ]− /Cl −
5.1 INTRODUCTION
The data presented in Chapter 4 established that [Au(CN) 2 ] loads onto ion exchange
−
elution of this species from the resin. In aqueous solutions, an excess concentration of
the counter is required to displace [Au(CN) 2 ]
−
from the resin. In this chapter,
nonaqueous solvents are investigated as a means of facilitating the elution of
[Au(CN) 2 ]− from anion exchange resin. Binary ion exchange equilibria are presented
303 K, using commercially available Purolite A500 as the ion exchanger. The mixed
solvents investigated include mixtures of water and acetone, dimethylsulfoxide (DMSO)
and N-methyl-2-pyrrolidone (NMP).
124
5.2 EXPERIMENTAL SECTION
5.2.1 Materials
Potassium aurocyanide (99%) and potassium thiocyanate (99%) were supplied by EBS
and Associates and Sigma-Aldrich, respectively. The anion exchange resin, Purolite
A500, was supplied by Purolite International. Purolite A500 is a Type 1 strong base,
divinylbenzene, macroporous resin in chloride form. The principal properties of the ion
exchange resin were previously presented in Table 4-1. Prior to usage, the resin was
soaked in distilled water for 24 hours to stabilize resin swelling. Hydrated ferric nitrate
(> 98%), used in the determination of SCN − in solution, was supplied by Asia Pacific
Specialty Chemicals.
125
5.2.2 Preparation of mixed solvent stock solutions
The general procedure for determining ion exchange equilibria in mixed solvent media
involved contacting the resin with a mixed solvent stock solution containing the relevant
loading anion. The effect of an organic solvent on the ion exchange isotherm for a
given binary system was examined by varying the amount of organic solvent in the
stock solution. In order to simplify the preparation process, stock solutions were
initially prepared on the basis of an unmixed volume of water and an organic solvent.
The nominal volume of all stock solutions was 1000 mL. For example, a stock solution
containing 20 vol% acetone was prepared by combining 200 mL of acetone with 800
mL of water. The compositions of the stock solutions were subsequently converted to
mole percent (mol%). Table 5-2 presents the range of compositions of acetone, DMSO
and NMP used in the determination of ion exchange isotherms for the
[Au(CN) 2 ]− / Cl − , [Au(CN) 2 ]− / SCN − and SCN − / Cl − binary systems. It should be
noted that the SCN − / Cl − binary system was only investigated in acetone-water
mixtures.
The loading anion was incorporated into the stock solution as follows. A known mass
of KAu(CN)2 or KSCN was weighed (± 0.0001g) into a volumetric flask. The salt was
initially dissolved in distilled water. Once the salt had completely dissolved, the
required amount of organic solvent was added to the flask. During the addition of the
organic solvent, some heat was released from the mixture, followed by a contraction in
the volume of the solution. The volume reduction varied between 1% and 2%
depending on the type and composition of the mixed solvent. The reduction in volume
of the solution was taken into account in the calculation of the concentration of the
loading anion in each stock solution.
In the previous chapter, it was established that the ion exchange isotherms for the given
binary systems in aqueous solution are independent of the total solution concentration,
within the ranges of concentration considered. Only a single value of the total solution
concentration was therefore employed in the determination of ion exchange isotherms in
mixed solvents. For the [Au(CN) 2 ] / Cl − and [Au(CN) 2 ] / SCN − binary systems, ion
− −
126
Table 5-2: Compositions of acetone, DMSO and NMP employed in the determination
of binary ion exchange equilibria in mixed solvents.
contacted with a stock solution containing [Au(CN)2 ] and SCN − , respectively. For
−
the [Au(CN) 2 ] / SCN − binary system, the SCN − -form of the resin was contacted with
−
From a knowledge of the change in concentration of the loading anion in the solution
phase, it was possible to calculate the composition of [Au(CN)2 ] or SCN − on the resin
−
127
compositions of the resin and solution phases with respect to Cl − . The concentration of
[Au(CN)2 ]− in solution was determined by analysis for Au using inductively coupled
plasma - atomic emission spectroscopy (ICP-AES). The standard uncertainty of the
analysis was ± 1%. The concentration of SCN − in solution was determined using UV-
Visible spectrophotometry, in accordance with ASTM D 4193-95. The uncertainty of
the analysis for SCN − was ± 5%. In both the ICP-AES and UV-Visible analysis,
calibration curves were established by preparing standards representative of the mixed
solvent composition in the samples. Ion exchange equilibria are expressed in terms of
equivalent ionic fractions for the solution and resin phases (see equations 4-3 and 4-4).
The experimental ionic fraction data that are reported represent the average of duplicate
runs, with an uncertainty of ± 5%.
Representative kinetic data for the loading of SCN − onto the Cl − -form of the resin in
acetone-water mixtures are presented in Figures 5-1A and 5-1B. The data are plotted in
128
terms of the concentration of SCN − in the resin phase as a function of time. Three
different values of the initial loading of SCN − on the resin are considered. In 5.73
mol% acetone (Figure 5-1A), equilibrium is attained in less than 1 day, although the rate
of ion exchange decreases as the initial loading of SCN − increases. In 68.6 mol%
acetone (Figure 5-1B), the rate of ion exchange also decreases with an increase in the
initial loading of SCN − on the resin. Around 2 days are required to establish
equilibrium when the resin is initially loaded with 0.42 eq/L SCN − . At the two
extremes of mixed solvent composition, however, it can be seen that the rate of ion
exchange is relatively rapid when fresh resin ( Cl − -form) is contacted with the stock
solution. In this situation, equilibrium is attained within a period of several hours.
The viscosity and dielectric constant are the two most significant liquid properties
which influence the rate of ion exchange in polar organic solvents. The rate is
particularly low when the solvent is very viscous or when the dielectric constant is
relatively low. The viscosities and dielectric constants of the three types of mixed
solvents considered in this study are shown in Figures 5-2 and 5-3, as a function of
composition. In terms of viscosity, the mixed solvents exhibit positive deviations from
ideality, resulting in viscosities which are significantly higher than the pure component
values, over certain ranges of composition. In contrast, the dielectric constants of the
mixed solvents are intermediate to the pure component values over the entire range of
composition.
In comparing Figures 5-1A and 5-1B, it is evident that an increase in the composition of
acetone in the mixed solvent results in a decrease in the rate of ion exchange. Although
the 68.6 mol% acetone mixture has a lower viscosity than the 5.73 mol% acetone
mixture, the effect of the lower dielectric constant of the former appears to exert the
greater influence on the rate of ion exchange. A comparison between the three types of
mixed solvents for the loading of [Au(CN) 2 ] onto fresh resin, in the Cl − -form, is
−
shown in Figures 5-4A and 5-4B. The data in Figure 5-4A are for mixed solvents
containing comparable levels of organic solvent in the range of 4-6 mol% while Figure
5-4B conveys the same information with respect to mixed solvents containing 62-70
mol% organic solvent. Overall, it is clear once again that an increase in the composition
129
of the organic solvent results in a decrease in the rate of ion exchange, due to the
reduction in the dielectric constant of the mixed solvent.
In the case of DMSO and NMP, the effect of viscosity on the rate of ion exchange is
masked to some degree because the decrease in the dielectric constant of the mixed
solvent is accompanied by an increase in the viscosity of the mixed solvent. The effect
of viscosity on the rate of ion exchange can be ascertained by considering the average
rate of loading of [Au(CN) 2 ] onto fresh resin over the first day of loading. Based on
−
the dielectric constant alone, the rate of loading in a given mixed solvent (at fixed
composition) would be expected to increase in the following order: acetone < NMP <
DMSO. With reference to Figure 5-4B, DMSO exhibits the highest average rate of
loading as expected. However, the rate of loading for NMP is substantially lower than
that for acetone. This is attributed to the much higher viscosity exhibited by the mixed
solvent containing NMP.
The effect of the type of counterion on the rate of ion exchange is illustrated in Figures
5-5A and 5-5B. These figures contain representative kinetic data for the loading of
[Au(CN) 2 ]− or SCN − onto the Cl − -form of the resin in acetone-water mixtures. For
each mixed solvent composition, the resin is initially loaded with around 0.2 eq/L of
counterion. In 5.73 mol% acetone, the average rate of loading of SCN − , over the first
day of loading, is higher than that for [Au(CN) 2 ] . The same observation can be made
−
with respect to the loading of SCN − onto the resin in 68.6 mol% acetone. This result is
attributed to the smaller ionic radius of the SCN − species. In general, the rate-
determining step in ion exchange is the diffusion of the counterions within the resin
particle or across a liquid film adherent to the resin. Small ions are considered to be
more mobile than large ions, leading to higher rates of ion exchange in systems
containing the former.
130
Figure 5-1: The effect of the initial resin loading on the rate of ion exchange in the
SCN − / Cl − binary system in acetone-water mixtures: (A), 5.73 mol% acetone; (B), 68.6
mol% acetone.
A
Error! Not a valid link.
B
Error! Not a valid link.
131
Figure 5-2: Viscosities of mixed solvents containing acetone, DMSO and NMP
(adapted from Assarsson and Eirich, 1968; LeBel and Goring, 1962; Washburn, 1929).
Figure 5-3: Dielectric constants of mixed solvents containing acetone, DMSO and
NMP (adapted from Marcus, 1985).
132
Figure 5-4: The effect of the type of mixed solvent on the rate of ion exchange in the
[Au(CN) 2 ]− / Cl − binary system: (A), 4-6 mol% organic solvent; (B), 62-70 mol%
organic solvent.
A
Error! Not a valid link.
B
Error! Not a valid link.
133
Figure 5-5: The effect of the type of counterion on the rate of ion exchange in acetone–
water mixtures: (A), 5.73 mol% acetone; (B), 68.6 mol% acetone.
A
Error! Not a valid link.
B
Error! Not a valid link.
de Lucas et al. (2001) have recently reported capacities for the cation exchange resin,
Amberlite IR-120, in various pure alcohols. In the same study they also present ion
exchange equilibria for Na+/K+ in the various solvents in which equilibrium was
achieved in 3 days and with vigorous stirring of the resin/solution mixture. A summary
of the capacities obtained from their work is shown in Figure 5-6. Generally, the
capacity of the resin decreases as the polarity of the solvent decreases. It is noteworthy,
however, that in their work there is no clear indication that these capacities were
obtained under true equilibrium conditions. Furthermore, the specific procedure used
for determining the capacity of the resin, as opposed to the procedure used for
measuring ion exchange equilibria, is not described.
134
In the present study, the capacity of the resin was determined by repeatedly contacting
the same sample of resin with a fresh quantity of stock solution until no further loading
of counterion occurred. This condition was confirmed when the change in the
concentration of the anion in the solution phase was found to be negligible. For mixed
solvent systems containing in excess of 25 mol% organic solvent, this procedure was
not satisfactory for determining the capacity of the resin. In these systems, the total
loading of the anion on the resin increased relatively slowly in response to the repeated
exposure of the resin to the stock solution. This characteristic is illustrated in Figures 5-
7A to 5-7C for the loading of [Au(CN) 2 ] onto the Cl − -form of the resin in various
−
mixed solvents.
Figure 5-6: Capacities of the cation exchange resin Amberlite IR-120 in various pure
solvents (adapted from de Lucas et al., 2001).
Figure 5-7: The loading of [Au(CN) 2 ] onto the Cl − -form of the resin in various
−
B
Error! Not a valid link.
C
Error! Not a valid link.
135
For mixed solvents containing 4-6 mol% organic solvent, it can be clearly seen that the
loading of [Au(CN) 2 ] becomes invariant after around 15 contacts between the resin
−
and the stock solution. The maximum loading of [Au(CN) 2 ] achieved is also very
−
close to the theoretical exchange capacity of the resin (1.3 eq/L). As the composition of
the organic solvent increases, the loading of [Au(CN) 2 ] decreases for a given number
−
of contacts, i.e, the resin becomes less selective for this species. However, it is not clear
from the data whether the loading of [Au(CN) 2 ] in each case approaches a limit which
−
is lower than the theoretical exchange capacity. In addition, the projected trends for
mixed solvents containing in excess of 25 mol% organic solvent suggest that an
exceedingly large number of contacts would be required to achieve the theoretical
exchange capacity.
The maximum loadings of the resin found in this way, for each composition of organic
solvent, are presented in Table 5-3. For mixed solvents containing less than 25 mol%
organic solvent, the maximum loadings in the three binary systems correspond closely
with the theoretical exchange capacity. However, for mixed solvents containing greater
than 60 mol% organic solvent, the maximum loadings are well below 1.3 eq/L. For
these systems, the number of contacts with the stock solution was insufficient to ensure
complete resin loading. In view of the uncertainty associated with the loading
procedure, the exchange capacity of the resin was set to 1.3 eq/L for the construction of
ion exchange isotherms in all of the mixed solvents.
136
Table 5-3: Maximum loadings achieved on Purolite A500 in water and in various
mixed solvents.
the various mixed solvents at 303 K are presented in Figures 5-8 to 5-14. The
numerical data pertaining to these figures are presented in tabular form in Appendix 3.
As a reference, the ion exchange isotherm for each binary system in aqueous solution is
also shown in the respective figures. The equivalent ionic fractions are plotted with
respect to ion B, the loading anion, in accordance with the ion exchange process
presented in equation 4-1.
137
Figure 5-8: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in acetone-water
−
mixtures at 303 K.
Figure 5-9: Ion exchange isotherms for [Au(CN) 2 ] (B) / SCN − (A) in acetone-water
−
mixtures at 303 K.
138
Figure 5-10: Ion exchange isotherms for SCN − (B) / Cl − (A) in acetone-water
mixtures at 303 K.
139
Figure 5-11: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in DMSO-water
−
mixtures at 303 K.
Figure 5-12: Ion exchange isotherms for [Au(CN) 2 ] (B) / SCN − (A) DMSO-water
−
mixtures at 303 K.
140
Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in NMP-water
−
Figure 5-13:
mixtures at 303 K.
Figure 5-14: Ion exchange isotherms for [Au(CN) 2 ] (B) / SCN − (A) in NMP-water
−
mixtures at 303 K.
141
Beginning with the [Au(CN) 2 ] / Cl − binary system, it can be seen that the resin is
−
generally selective towards [Au(CN) 2 ] (ion B) in mixed solvents containing less than
−
25 mol% organic solvent. The selectivity for this species decreases as the composition
of the organic solvent increases. For mixed solvents containing greater than 60 mol%
organic solvent, the selectivity of the resin changes dramatically in favour of Cl − . The
qualitative features of the ion exchange isotherms for [Au(CN) 2 ] / Cl − are also evident
−
in the other binary systems. In particular, a reversal in the selectivity of the resin occurs
in mixed solvents containing relatively high levels of organic solvent. Thus, for mixed
solvents containing greater than 60 mol% organic solvent, the selectivity of the resin for
a given anion increases in the following order: [Au(CN) 2 ] < SCN − < Cl − . Note, this
−
The extent to which a reversal in the selectivity occurs depends, in part, on the value
used for the exchange capacity in the calculation of the equivalent ionic fractions in the
resin phase (equation 4-4). This is illustrated in Figure 5-15 for [Au(CN) 2 ] / Cl − in
−
acetone-water mixtures. As noted earlier, the exchange capacity of the resin in the
various mixed solvents was set at 1.3 eq/L. However, if the exchange capacity of the
resin is set to the value corresponding to the maximum loading (see Table 5-3), at a
given composition of acetone, the selectivity for Cl − is only observed in the mixed
solvent containing 68.6 mol% acetone. Regardless of the basis used for constructing the
isotherms, it is no doubt clear that the composition of the mixed solvent significantly
influences the selectivity of the resin.
The anions considered in this study exhibit large differences in terms of their charge
densities and polarisabilities. The effect of mixed solvent composition on selectivity
can be explained qualitatively in terms of the degree of solvation of the anions in the
mixed solvents. Generally, the solvation of ions in mixed solvents is achieved by both
ion-water and ion-solvent interactions. The ion-water interactions refer simply to the
process by which ions become hydrated in the form of hydration shells. The ion-solvent
interactions refer to the corresponding process with respect to the organic solvent.
There are a number of ways in which anions can interact with water and organic
142
solvents. Hydrogen bonding and interaction via the polarisabilities of anions and
solvent molecules (dispersion forces) are only considered in the following discussion.
The organic solvents employed in this work are all examples of dipolar aprotic solvents.
Hydrogen-bonding interactions between anions and such solvents are normally absent.
Small weakly polarisable anions, such as Cl − , are strong hydrogen-bond acceptors and
are therefore much more solvated by water than by dipolar aprotic solvents. This type
of interaction decreases as the charge density of the anion decreases. Thus, large
polarisable anions, such as [Au(CN) 2 ] , are less solvated than Cl − in water (Parker,
−
1969; Muir et al., 1985). Dipolar aprotic solvents are much more polarisable than water
and interact strongly with large polarisable anions. In most cases, however, anion
solvation is more significant in water than in dipolar aprotic solvents. An interesting
exception to this is the solvation of I 3− in DMSO (Parker, 1969).
The previous considerations suggest that, in water, the degree of solvation of the anions
considered here increases in the order of: [Au(CN) 2 ] < SCN − < Cl − . The opposite
−
binary system, it was noted previously that the resin is generally selective towards
[Au(CN) 2 ]− in mixed solvents containing relatively low levels of organic solvent. In
this situation, it is expected that the solvation of the anions occurs mainly via hydration
and that [Au(CN) 2 ]
−
is less solvated than Cl − . Ion-pair formation between the
functional groups of the resin and the anions is therefore energetically more favourable
for [Au(CN) 2 ] .
−
In mixed solvents containing high levels of organic solvent, the solvation of the anions
occurs mainly via ion-solvent interactions. The relatively weak solvation of Cl − in
these mixtures significantly enhances the activity of this species and promotes ion-pair
formation between Cl − and the resin. This accounts for the reversal in the selectivity of
the resin. The transition to a system in which anion solvation is achieved principally by
ion-solvent interactions, rather than by hydration, can also be used to explain the
143
Figure 5-15: The effect of exchange capacity on the ion exchange isotherms of the
[Au(CN ) 2 ]− / Cl − binary system in acetone-water mixtures: (A), 5.73 mol%; (B), 13.9
mol%; (C), 26.7 mol%; (D), 68.6 mol%.
A
Error! Not a valid link.
B
Error! Not a valid link.
C
Error! Not a valid link.
D
Error! Not a valid link.
reversal in the selectivity of the resin in the [Au(CN) 2 ] / SCN − and SCN − / Cl −
−
resin. This effect is particularly significant at low ionic fractions of [Au(CN)2 ] in the
−
solution phase (xB<0.5) and suggests that SCN − is less solvated than Cl − in this range
of mixed solvent composition.
At higher ionic fractions of [Au(CN)2 ] (xB>0.5), the isotherms intersect such that Cl −
−
becomes a slightly more effective counterion than SCN − . The interpretation of this
result is that an increase in the concentration of [Au(CN)2 ] in solution reduces the
−
interaction between the organic solvent and the counterions. The effect is more
144
pronounced for Cl − in view of its lower polarisability. As the composition of acetone
increases, the contribution of ion-solvent interactions to the solvation of the anions
increases and so the isotherms intersect at a lower concentration of [Au(CN)2 ] . This
−
trend is particularly evident for an increase from 13.9 to 26.7 mol% acetone (Figures 5-
16B and 5-16C). In 68.6 mol% acetone, the contribution of hydrogen bonding to the
solvation of the anions is minimal and thus Cl − becomes a more effective counterion
than SCN − .
The effect of the type of counterion on the selectivity of the resin in acetone-water
mixtures is also observed in DMSO-water (Figure 5-17) and NMP-water mixtures
(Figure 5-18). These findings suggest that small weakly polarisable anions are the
preferred counterions for the elution of [Au(CN) 2 ] from ion exchange resin in mixed
−
respectively.
In mixed solvents containing 4-6 mol% organic solvent, the contribution of ion-solvent
interactions to the solvation of the anions is minimal. In this case, it is expected that the
isotherms for the various mixed solvents, in a given binary system, are very similar.
This result is only observed for the isotherms in mixed solvents containing acetone or
DMSO. The isotherms for the NMP-water mixtures highlight that this type of mixed
solvent is more effective for displacing [Au(CN) 2 ] from the resin.
−
As the mixed solvent composition increases, the differences between the various mixed
solvents become more pronounced. Over the whole range of mixed solvent
145
composition, the effectiveness of a mixed solvent, for displacing [Au(CN) 2 ] , generally
−
increases in the following order: DMSO < acetone < NMP. The unusual feature of this
trend is that it does not correlate with the dielectric constants of the mixed solvents. In
section 5.3.3, it was observed that the selectivity of the resin for [Au(CN) 2 ] , in a given
−
type of mixed solvent, decreases with an increase in mixed solvent composition. This
trend also correlates with the dielectric constant of the mixed solvent, i.e. the selectivity
of the resin for [Au(CN) 2 ] decreases as the dielectric constant of the mixed solvent
−
decreases.
Based on these observations, the expected trend in the effectiveness of a mixed solvent
for displacing [Au(CN) 2 ] is: DMSO < NMP < acetone. Although this sequence
−
differs from that obtained experimentally it is clear that dipolar aprotic solvents with
relatively low dielectric constants are more effective in reducing the selectivity of the
resin for [Au(CN) 2 ] . In particular, a mixed solvent containing in excess of 60 mol%
−
NMP, in combination with Cl − as the counterion, is remarkably effective for the elution
of [Au(CN) 2 ] from anion exchange resin. (see Figure 5-19D).
−
5.4 CONCLUSIONS
The selectivity of the anion exchange resin for [Au(CN) 2 ] decreases with an increase
−
in the composition of the organic solvent in the external solution. Generally, mixed
solvents containing greater than 60 mol% organic solvent are preferred for the elution of
[Au(CN) 2 ]− . Small weakly polarisable anions, such as Cl − , are the preferred
counterions in this range of mixed solvent composition. Dipolar aprotic solvents with
relatively low dielectric constants are more effective in reducing the selectivity of the
resin for [Au(CN) 2 ] .
−
146
Figure 5-16: The effect of the type of counterion on the selectivity of the resin for
[Au(CN) 2 ]− in acetone-water mixtures: (A), 5.73 mol%; (B), 13.9 mol%; (C), 26.7
mol%; (D), 68.6 mol%.
A
Error! Not a valid link.
B
Error! Not a valid link.
C
Error! Not a valid link.
D
Error! Not a valid link.
147
Figure 5-17: The effect of the type of counterion on the selectivity of the resin for
[Au(CN) 2 ]− in DMSO-water mixtures: (A), 5.91 mol%; (B), 14.3 mol%; (C), 27.4
mol%; (D), 69.5 mol%.
A
Error! Not a valid link.
B
Error! Not a valid link.
C
Error! Not a valid link.
D
Error! Not a valid link.
148
Figure 5-18: The effect of the type of counterion on the selectivity of the resin for
[Au(CN) 2 ]− in NMP-water mixtures: (A), 4.46 mol%; (B), 62.7 mol%.
A
Error! Not a valid link.
B
Error! Not a valid link.
149
Figure 5-19: The effect of the type of solvent on the ion exchange isotherms for
[Au(CN ) 2 ]− / Cl − in various mixed solvents: (A), 4-6 mol%; (B), 13-15 mol%; (C), 26-
28 mol%; (D), 62-70 mol%.
A
Error! Not a valid link.
B
Error! Not a valid link.
C
Error! Not a valid link.
D
Error! Not a valid link.
150
Figure 5-20: The effect of the type of solvent on the ion exchange isotherms for
[Au(CN) 2 ]− / SCN − in various mixed solvents: (A), 4-6 mol%; (B), 13-15 mol%; (C),
26-28 mol%; (D), 62-70 mol%.
A
Error! Not a valid link.
B
Error! Not a valid link.
C
Error! Not a valid link.
D
Error! Not a valid link.
151
5.5 REFERENCES
Anderson, R. E. and Hansen, R. D., Phenol sorption on ion-exchange resins. Journal of
Industrial and Engineering Chemistry, 1955. 47: p. 71-75.
Assarsson, P. and Eirich, F. R., Properties of amides in aqueous solution. IA. Viscosity
and density changes of amide-water systems. B. An analysis of volume deficiencies
of mixtures based on molecular size differences (mixing of hard spheres). The
Journal of Physical Chemistry, 1968. 72(8): p. 2710-2719.
Bodamer, G. W. and Kunin R., Behaviour of ion exchange resins in solvents other than
water. Industrial and Engineering Chemistry, 1953. 45(11): p. 2577-2581.
de Lucas, A., Valverde, J. L., Romero, M. C., Gomez, J., and Rodriguez, J. F., Ion
exchange equilibria in nonaqueous and mixed solvents on the cationic exchanger
Amberlite IR-120. Chem. Eng. Data., 2001. 46: p. 73-78.
Fleming, C. A. and Monhemius, A. J., On the extraction of various base metal chlorides
from polar organic solvents into cation and anion exchange resins.
Hydrometallurgy, 1979. 4: p. 159-167.
Gable, R. W. and Strobel, H. A., Non-aqueous ion exchange. I. Some cation equilibrium
studies in methanol. Journal of Physical Chemistry, 1956. 60(5): p. 513-517.
Hassan, E. A., Abo El-Magd, A. S., Kamal, F H., El-Hady, M. F., and Ismail, N. M.,
Cation exchange distribution studies of Ca(II), Mg(II), Co(II), Cu(II), and Fe(III)
ions in organic solvent-HCl media. Indian J. Chem., 1981. 20A: p. 598-599.
Katzin, L. I. and Gebert, E., Absorption of inorganic salts from organic solvents onto
anion-exchange resins. Journal of the American Chemical Society, 1953. 75: p.
801-803.
Marcus, Y., Ion Properties. 1997, Marcel Dekker Inc, New York.
Marcus, Y., Ion Solvation. 1985, John Wiley and Sons, New York.
Muir, D. M., Singh, P., Kenna, C. C., Tsuchida, N., and Benari, M. D.,
Hydrometallurgical thermodynamics. II. Solvent effects on the activity and free
energies of transfer of cyanide (CN-), silver cyanide (Ag(CN)2-) and gold cyanide
(Au(CN)2-) in ethanol-water and acetonitrile-water mixtures. Australian Journal
of Chemistry, 1985. 38(7): p. 1079-1090.
Nandan, D. and Gupta, A. R., Solvent sorption isotherms, swelling pressures, and free
energies of swelling of polystyrenesulfonic acid type cation exchangers in water
and methanol. J. Phys. Chem., 1977. 81: p. 1174-1178.
152
Parker, A. J., Protic – dipolar aprotic solvent effects on rates of bimolecular reactions.
Chemical Reviews, 1969. 69(1): p. 1 – 32.
153
CHAPTER 6
6. MODELING ION EXCHANGE EQUILIBRIA OF Au(CN) 2 − /Cl − [ ]
[Au(CN) ]− /SCN − AND SCN − /Cl− BINARY SYSTEMS IN MIXED
2
SOLVENTS.................................................................................................. 157
6.1 INTRODUCTION .........................................................................................157
6.2 DATA CORRELATION ................................................................................157
6.3 RESULTS AND DISCUSSION .......................................................................160
6.3.1 Modeling results....................................................................................160
6.3.2 Correlation of KAB with the dielectric constant.....................................170
6.3.3 The Gibbs energy of transfer ................................................................174
6.3.4 Correlation of KAB with the Gibbs energy of transfer...........................178
6.4 CONCLUSIONS ...........................................................................................185
6.5 REFERENCES .............................................................................................186
Figure 6-1: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in acetone-water
−
157
Figure 6-14: A schematic representation of an ion exchange process occurring between
[Au(CN) 2 ]− and Cl − in a mixed solvent system.......................................179
Figure 6-15: The composition of [Au(CN) 2 ] in the resin phase as a function of the net
−
Table 6-1: Debye-Hückel parameters for mixed solvents containing acetone, DMSO
and NMP at 303 K a....................................................................................159
Table 6-2: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in
acetone-water mixtures at 303 K................................................................161
Table 6-3: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in
DMSO-water mixtures at 303 K. ...............................................................162
Table 6-4: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in NMP-
water mixtures at 303 K. ............................................................................162
Table 6-5: The reciprocity relation applied to the Wilson parameters derived from ion
exchange equilibria in various mixed solvents. .........................................168
Table 6-6: Gibbs transfer energies for selected ions (from Marcus, 1997, unless
otherwise stated).........................................................................................177
158
6. MODELING ION EXCHANGE EQUILIBRIA OF
[Au(CN) ]− /Cl− [Au(CN) ]− /SCN − AND SCN − /Cl−
2 2
6.1 INTRODUCTION
In this chapter, binary ion exchange equilibria for [Au(CN) 2 ]− / Cl − ,
[Au(CN) 2 ]− / SCN − and SCN − / Cl − in mixed solvent media are correlated using the
Law of Mass Action, modified with activity coefficients, to determine the equilibrium
constant for each binary system. The relationship between the equilibrium constant and
the dielectric constant of the mixed solvent is examined. Finally, the equilibrium
constants are correlated with the Gibbs energies of transfer associated with the solvation
of the ions in mixed solvents.
157
ionic fractions in the resin phase (yi), the exchange capacity of the resin was set at a
value corresponding to the theoretical exchange capacity of the resin (1.3 eq/L), as
noted in Section 5.3.2.
The equation for the equilibrium constant was modified with activity coefficients for the
solution phase and resin phase. The individual ionic activity coefficients in the resin
phase were calculated using the semi-empirical approach involving the Wilson
equations. An extended form of the Debye-Hückel limiting law was used in Chapter 4
for the determination of solution phase activity coefficients. Alternatively, Shallcross et
al., (1988) have used the Pitzer (1979) method to estimate solution phase activity
coefficients. However, the required parameters for the pure and mixed electrolytes
relevant to this study are not available in the literature. Furthermore, Hovath (1985)
highlights that the Pitzer equation is incapable of dealing with electrolytes in mixed
solvent systems. The solution phase activity coefficients in mixed solvents were
therefore calculated with the extended Debye-Hückel limiting law.
1.8246 × 10 6
A= (6-1)
(εT ) 3 / 2
50.29 × 10 8
B= (6-2)
(εT )1 / 2
Calculated values of yB were generated from the experimental values of xB and yB and
for given values of KAB, Λ AB and Λ BA , using equation 4-10 (3-parameter regression).
The optimum values of KAB, Λ AB and Λ BA , for a given isotherm, were obtained by
158
Table 6-1: Debye-Hückel parameters for mixed solvents containing acetone, DMSO
and NMP at 303 K a.
minimising the sum of squared relative deviations (SSRD) with respect to the resin
phase composition (equation 4-11).
In Chapter 4, apart from the 3-parameter regression method, equilibrium constants for
aqueous ion exchange systems were also determined using a 2-parameter regression
method. In the 2-parameter regression, the equilibrium constant was obtained
independently using the Gaines and Thomas equation. For the aqueous ion exchange
equilibria, no significant differences were observed between the 3-parameter and 2-
parameter regression methods. The 2-parameter regression is also less reliable when
equilibrium data do not extend over the whole range of resin composition (0 < yB < 1).
This point is particularly relevant to the ion exchange equilibria obtained in mixed
solvents containing high levels of organic solvent (see Section 5.3.2). Accordingly, the
3-parameter regression method was used exclusively to obtain equilibrium parameters
for mixed solvent systems.
159
6.3 RESULTS AND DISCUSSION
According to the triangle rule (Mehablia et al., 1994), the equilibrium constants of two
of the three constituent binary systems of a ternary system can be used to predict the
equilibrium constant of the third binary system. For the [Au(CN) 2 ]−
(C)/ SCN − (B)/ Cl − (A) ternary system, this rule may be stated as
K AC = K AB K BC (6-3)
The equilibrium constants derived from equation 6-3 for [Au(CN ) 2 ] / Cl − in acetone-
−
water mixtures are plotted in Figure 6-8. The equilibrium constants predicted from the
triangle rule are qualitatively in agreement with the optimised values, although the
former are consistently overestimated relative to the latter. The relative deviation
between the two sets of data is more significant for acetone-water mixtures containing
less than 25 mol% acetone. Note, the triangle rule is only applied to equilibrium
constants in acetone-water mixtures since ion exchange equilibria for SCN − / Cl − were
not measured in the other mixed solvent systems.
160
Table 6-2: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in
acetone-water mixtures at 303 K.
Au(CN) −2 / Cl −
5.73 55.71 3.073 0.325 2.9
13.9 5.573 3.277 0.306 4.6
26.7 0.642 3.028 0.330 5.3
68.6 0.020 3.461 0.289 6.2
Au(CN) −2 / SCN −
5.73 10.50 1 × 10-4 2.123 0.9
13.9 2.932 0.084 1.708 2.5
26.7 0.772 2.303 0.138 4.3
68.6 0.157 3.044 0.063 2.7
SCN − / Cl −
5.73 10.36 1.109 1.127 1.2
13.9 3.732 2.477 0.404 2.6
26.7 1.339 2 × 10-5 2.902 2.8
68.6 0.091 4 × 10-6 5.012 8.8
a
KAB refers to the equilibrium constant for ion B entering the resin phase and displacing ion A.
161
Table 6-3: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in
DMSO-water mixtures at 303 K.
Au(CN) −2 / Cl −
5.91 81.36 1.923 0.761 3.7
14.3 21.58 3.887 0.258 4.7
27.4 2.977 3.641 0.274 5.0
69.3 0.040 4.910 0.204 9.2
Au(CN) −2 / SCN −
5.91 16.05 6 × 10-3 2.170 0.8
14.3 9.284 0.136 1.775 1.3
27.4 3.592 0.058 2.059 0.9
69.3 0.417 1.829 0.546 4.5
a
KAB refers to the equilibrium constant for ion B entering the resin phase and displacing ion A.
Table 6-4: Optimised values of KAB, ΛAB and ΛBA for ion exchange equilibria in NMP-
water mixtures at 303 K.
Au(CN) −2 / Cl −
4.46 23.76 3.751 0.266 4.6
62.7 3 × 10-3 3.635 -0.014 7.5
Au(CN) −2 / SCN −
4.46 4.133 1.194 0.918 3.3
62.7 0.013 1.092 4.349 9.7
a
KAB refers to the equilibrium constant for ion B entering the resin phase and displacing ion A.
162
Figure 6-1: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in acetone-water
−
mixtures at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
Figure 6-2: Ion exchange isotherms for [Au(CN) 2 ] (A) / SCN − (B) in acetone-water
−
mixtures at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
163
Figure 6-3: Ion exchange isotherms for SCN − (B) / Cl − (A) in acetone-water mixtures
at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
164
Figure 6-4: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in DMSO-water
−
mixtures at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
Figure 6-5: Ion exchange isotherms for [Au(CN) 2 ] (A) / SCN − (B) in DMSO-water
−
mixtures at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
5.91mol% 14.3mol%
27.4mol% 69.3mol%
165
Figure 6-6: Ion exchange isotherms for [Au(CN) 2 ] (B) / Cl − (A) in NMP-water
−
mixtures at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
4.46mol% 62.7mol%
Figure 6-7: Ion exchange isotherms for [Au(CN) 2 ] (A) / SCN − (B) in NMP-water
−
solutions at 303 K.
0.8
0.6
yB
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
xB
4.46mol% 62.7mol%
166
Figure 6-8: A comparison of the equilibrium constants predicted from the triangle rule
(equation 6-3) with those of the 3-parameter regression method for the
[Au(CN) 2 ]− / Cl − binary system in acetone-water mixtures.
100
K AB
1
0 20 40
Acetone (mol%)
167
The values for Λ AB and Λ BA in the mixed solvent systems fall generally within the
range of 0 to 5. This range of values is consistent, although slightly larger, than that
observed for the ion exchange equilibria in aqueous solution. The optimisation routine
produced a negative parameter in one instance: Λ BA for [Au(CN ) 2 ] / Cl − in 62.7 mol%
−
NMP. The products of the Wilson parameters are presented in Table 6-5 to assess the
validity of the reciprocity relation proposed by Allen et al. (1989).
In many cases, one of the Wilson parameters is significantly less than 0.2, which
indicates that only a single Wilson parameter is required to accurately correlate the ion
exchange equilibria. In this situation, it is expected that the reciprocity relation is
generally not applicable. This explains why the reciprocity relation is not observed for
[Au(CN) 2 ]− / SCN − in most of the mixed solvents. The same explanation can be used
containing greater than 25 mol% acetone. Furthermore, the Hála constraint (Hála,
1972), which relates the three pairs of Wilson parameters required in a ternary system,
is not obeyed in any of the acetone-water mixtures.
Table 6-5: The reciprocity relation applied to the Wilson parameters derived from ion
exchange equilibria in various mixed solvents.
168
The reciprocity relation is most consistently observed for the [Au(CN ) 2 ] / Cl − binary
−
system. The only significant deviation from the reciprocity relation is that for
[Au(CN) 2 ]− / SCN − in 62.7 mol% NMP. The correlation of this set of ion exchange
equilibria produced the least accurate results (AARD = 9.7%). Both of the Wilson
parameters for this system are relatively large suggesting that this system exhibits much
more nonideal behaviour.
example, the equilibrium constant in 5.73 mol% acetone is around 4 times lower than
that in aqueous solution. The equilibrium constant is further reduced by an order of
magnitude for a relatively modest increase in the mixed solvent composition from 5.73
to 13.9 mol% acetone.
less than 25 mol% organic solvent, the KAB values are lower for SCN − as the
counterion. The opposite trend is evident in mixed solvents containing more than 25
169
mol% organic solvent. These results are consistent with the observation made in
Chapter 5 that Cl − is a better counterion than SCN − for displacing [Au(CN) 2 ] from
−
the resin in mixed solvents containing high levels of dipolar aprotic solvents. The
equilibrium constant for SCN − / Cl − in 68.6 mol% acetone further supports this
observation.
The equilibrium constants for [Au(CN) 2 ]− / Cl − and [Au(CN) 2 ]− / SCN − are also
strongly dependent on the type of mixed solvent, as shown in Figures 6-9 and 6-10. At
a given composition of organic solvent, the lowest value of KAB is obtained for NMP-
water mixtures. Figures 6-9 and 6-10 are consistent with Figures 5-19 and 5-20 and
confirm that the effectiveness of a given dipolar aprotic solvent, for displacing
[Au(CN) 2 ]− from the resin, increases in the following order: DMSO < acetone < NMP.
The term, effectiveness, is used here to draw attention to the composition of the mixed
solvent. Thus, higher levels of DMSO or acetone are required, in comparison to NMP,
to achieve a similar reduction in the loading of [Au(CN) 2 ] on the resin.
−
For a given type of mixed solvent, it was also observed that the general decrease in the
selectivity of the resin for [Au(CN) 2 ] correlates with the dielectric constant of the
−
mixed solvent. The inference here is that the dielectric constant provides, to a first
approximation, a measure of the relative degree of solvation of the anions. As the
composition of the organic solvent increases, the dielectric constant of the mixed
solvent decreases and the solvation of Cl − or SCN − decreases relative to [Au(CN) 2 ] .
−
170
Figure 6-9: A graphical depiction of the variation of the equilibrium constant of the
[Au(CN) 2 ]− / Cl − binary system with mixed solvent composition.
1000
100
10
K AB
1
0 20 40 60
0.1
0.01
0.001
organic solvent (mol%)
Figure 6-10: A graphical depiction of the variation of the equilibrium constant of the
[Au(CN) 2 ]− / SCN − binary system with mixed solvent composition.
100
10
K AB
1
0 20 cv 40 60
0.1
0.01
organic solvent (mol%)
171
On this basis, the equilibrium constants can also be correlated with the dielectric
constant of the mixed solvent, as shown in Figures 6-11 and 6-12.
For a given value of the dielectric constant, however, the equilibrium constant is
dependent on the type of mixed solvent. In fact, with Cl − as the counterion, the
displacement of [Au(CN) 2 ] from the resin increases in the order of acetone < NMP <
−
DMSO (Figure 6-11). On the other hand, with SCN − as the counterion, the
displacement of [Au(CN) 2 ] increases in the order of acetone < DMSO < NMP (Figure
−
6-12). These trends are in contrast to the selectivity sequence, based on mixed solvent
composition, in which KAB decreases in the order of DMSO < acetone < NMP, for both
counterions (Figures 6-9 and 6-10).
As such, the dielectric constant alone cannot be used to explain the differences in the
equilibrium constants for the various mixed solvents. Ion exchange selectivity in mixed
solvents, and the role of ion-solvent interactions, can be understood more
comprehensively by considering the change in the total solvation energy of an anion
when it is transferred from one solvent to another. This change in the total solvation
energy of an anion is often referred to as the solvent medium effect and is directly
related to the Gibbs energy of transfer of the anion (Marcus, 1985).
172
Figure 6-11: The equilibrium constant of the [Au(CN) 2 ] / Cl − binary system as a
−
1000
100
10
log(KAB )
1
0 20 40 60 80
0.1
0.01
0.001
ε
Figure 6-12: The equilibrium constant of the [Au(CN) 2 ] / SCN − binary system as a
−
100
10
log(KAB )
1
0 20 40 60 80
0.1
0.01
ε
173
6.3.3 The Gibbs energy of transfer
The stability of ions in solvents is most readily quantified with the Gibbs energy of
transfer, Δ t G o (i, w → s) . The transfer energy represents the change in the total
solvation energy of a solute (i) when it is transferred from a reference solvent, usually
water (w), to another solvent (s). As this study involves mixed solvents, the energy
associated in the transfer of a solute from water to a mixed solvent, Δ t G o (i, w → ms) ,
will be referred to in the following discussion. The symbol ‘ms’ denotes the mixed
solvent system. The magnitude of Δ t G o (i, w → ms) reflects the stability of the ion in
the mixed solvent relative to water. The Gibbs energy of transfer of solute i is
essentially the difference in the standard state chemical potentials of the solute in the
two solvents expressed as follows,
where μ o (i, w ) and μ o (i, ms) are the standard chemical potentials of solute i in water
and the mixed solvent, respectively. The chemical potential, μ (i ) , of solute i in the two
solutions is related to the activity coefficients in the respective solvents according to,
In equation 6-5 and 6-6, R and T refer to the gas constant and temperature respectively,
while m(i) is the molality of species i. The relative numerical values of the two activity
coefficients for species i, γ (i, w ) and γ (i, ms) , in water and mixed solvent respectively,
are dependent only on the two reference states. The Gibbs energy of transfer can then
be expressed in terms of the activity coefficients as:
γ (i, w )
Δ t G o (i, w → ms) = RT ln = RT ln γ (i, w → ms) (6-7)
γ (i, ms)
174
The quantity, γ (i, w → ms) is known as the solvent medium effect and reflects the
change in the activity coefficient of a solute on transfer from water to a mixed solvent.
The transfer energy is related to the logarithm of the change in the activity coefficient.
Therefore, a small change in the transfer energy corresponds to a significant change in
the stability of the solute.
The transfer energy described thus far refers to a neutral solute or a salt and as a result it
can be related to the standard solubility product, K spo (i ) of the electrolyte in the two
⎡ K spo (i, w ) ⎤
Δ t G (i, w → ms) = RT ln ⎢ o
o
⎥ (6-8)
⎣⎢ K sp (i, ms) ⎦⎥
The transfer energies of unknown ions can be determined by using the values obtained
for the reference ions. A detailed description of this method is given elsewhere (Kalidas
et al., 2000; Marcus 1985). As an example, the transfer energy of K + can be estimated
as follows:
175
Thus by initially employing the TATB assumption to evaluate Δ t G o (K + ) , the
( )
Δ t G o [Au(CN) 2 ] = Δ t G o (KAuCN 2 ) − Δ t G o (KPh 4 B) + 1 / 2Δ t G o (TATB)
−
(6-11)
comparison purposes. Note that the values in Table 6-6 apply to pure solvents. A
positive value for the transfer energy implies that an anion is less solvated in the given
solvent in comparison to water. All anions presented in Table 6-6, expect for ClO −4 , are
more solvated in water (hydrated) than in any of the three solvents. For ClO −4 , the
decrease in the free energy upon transfer from water to DMSO or NMP suggests that it
is more stable in these solvents compared to water. Transfer energies of [Au(CN) 2 ]
−
and SCN − from water to acetone are unavailable. However, the respective values in
DMSO and NMP suggest that Δ t G(oi, w →ms ) from water to acetone should lie in the
Unfortunately, the transfer energies of the anions of interest in this study are unavailable
for the range of mixed solvent compositions considered. Muir et al., (1985) have
determined the transfer energies of the cyanide complexes of gold and silver from water
to mixed solvents containing acetonitrile (AN). The relationship between the transfer
energies and the composition of the mixed solvent is shown in Figure 6-13. Transfer
energies for Cl − and CN − are also depicted in the figure. All of the anions presented
in Figure 6-13 are more solvated in water than in AN-water mixtures containing more
than 30 mol% AN.
176
Figure 6-13: Gibbs transfer energies for various anions from water to acetonitrile-
mixtures (adapted from Muir et al., 1985).
35
15
-5 0 20 40 60 80 100
-15
AN composition (mol%)
gold cyanide silver cyanide
cyanide chloride
Table 6-6: Gibbs transfer energies for selected ions (from Marcus, 1997, unless
otherwise stated).
Cl − 57 40 51
SCN − - 10 18
ClO −4 6 -1 -12
177
The differences in the transfer energies of anions provides a measure of the relative
degrees of solvation of the anions in the solvent of interest. For example, the cyanide
complexes of gold and silver are generally more solvated than Cl − and CN − in AN-
water mixtures. The difference between the cyanide complexes and the small anions
also becomes more pronounced as the composition of AN increases. For an ion
exchange process conducted in a mixed solvent, this suggests that the equilibrium
constant, and therefore the selectivity of the resin, is related to the difference in the
transfer energies of the exchanging ions. A relative or net Gibbs energy of transfer can
be defined for a pair of exchanging anions in a given mixed solvent as follows:
Δ t G net
0
(A − B, ms) = Δ t G 0 (A, w → ms) − Δ t G 0 (B, w → ms) (6-12)
where A and B correspond to the anions used in the definition for KAB. When Δ t Gnet
0
>
0, A is less solvated than B in the mixed solvent. In this situation, one would expect
that KAB in the mixed solvent decreases in relation to that observed in aqueous solution.
If KAB is initially much greater than 1 in aqueous solution, a positive value for Δ t Gnet
0
implies that the resin is less selective for anion B in the mixed solvent. For a
sufficiently large and positive value of Δ t Gnet
0
, it is possible that KAB < 1 in the mixed
solvent and that the resin is therefore selective for anion A.
represented by R4N+. The pore solution (ps) within the resin is far less organic in nature
compared to the mixed solvent (ms) in the external solution. As a result, the
counterions are presented with essentially two different solvent environments in a
mixed solvent system.
178
Figure 6-14: A schematic representation of an ion exchange process occurring between
[Au(CN) 2 ]− and Cl − in a mixed solvent system.
Given the two alternative environments in the pore and external solutions, counterions
would prefer the phase that provides the most stability, or in other words, the solvent
environment which results in the least ion activity. In the [Au(CN) 2 ] / Cl − binary
−
system, the decrease in the selectivity for [Au(CN) 2 ] in mixed solvent systems can be
−
explained by considering the transfer energies of the counterions. The Δ t G(oi, w →ms )
value for Cl − in Table 6-6 indicates that chloride is more solvated in water rather than
in acetone, DMSO or NMP. Relative to chloride, the gold cyanide complex is more
solvated or more stable in the organic solvents. A higher proportion of chloride ions
will therefore transfer to the resin phase, forcing the gold cyanide complex into the
solution phase, in comparison to the exchange process in aqueous solution.
The two-step ion exchange process can be quantitated as follows. The first step
involves the ion exchange between the resin (r) and the pore solution (ps). The
equilibrium constant associated with [Au(CN) 2 ]
−
(B) entering the resin phase and
[Au(CN) 2− ] r [Cl − ] ps
K ps
= (6-13)
[Au(CN) 2− ] ps [Cl − ] r
AB
179
The second step involves the distribution of the exchanging anions between the pore
solution and the mixed solvent. While an equilibrium constant can be assigned to the
distribution of each anion between the two liquid phases, it is more convenient to
consider the overall equilibrium constant of the ion exchange process:
[Au(CN) 2− ] r [Cl − ] ms
ms
K AB = (6-14)
[Au(CN) 2− ] ms [Cl − ] r
From equations 6-13 and 6-14, the ratio of the equilibrium constants is given by:
ps
K AB [Au(CN) 2− ] ms [Cl − ] ps
= (6-15)
ms
K AB [Au(CN) 2− ] ps [Cl − ] ms
It is assumed here that the distribution of the anions between the pore solution and the
mixed solvent is governed by the solubility products of the corresponding salts of the
anions. For example, the ratio of the chloride concentrations in equation 6-15 is
assumed to be comparable to the ratio of the solubility products of the corresponding
salts. Since only potassium salts were used in this study, the solubility products of the
salts, K spo (i ) in the respective solutions are as follows:
Note, the activity coefficients have been omitted from equation 6-13 to 6-19 to simplify
the presentation of the equations. Equation 6-15 can therefore be expressed in terms of
solubility products as follows:
ps
K AB K sp1 K sp4
ms
= (6-20)
K AB K sp2 K sp3
180
Since water is preferentially absorbed into the pores of the resin, the equilibrium
ps
constant in the pore solution, K AB , is assumed to be comparable to that in aqueous
w
solution ( K AB ). The same approximation can be made with respect to the solubility
products in equations 6-17 and 6-19. Equation 6-20 can now be written as
w
K AB K sp1 K sp4
ms
= (6-21)
K AB K sp2 K sp3
or
By considering equation 6-8, it is evident that the right hand side of equation 6-22 can
be expressed in terms of the Gibbs energies of transfer of the salts from water to the
mixed solvent:
⎛Kw ⎞
RT ln⎜⎜ AB ⎟ = Δ t G 0 (KCl, w → ms) − Δ t G 0 (KAu(CN) 2 , w → ms)
ms ⎟
(6-23)
⎝ K AB ⎠
The Gibbs energy of transfer of an electrolyte can be expressed in terms of the sum of
the transfer energies of its constituent ions (see equation 6-9). Thus equation 6-23 can
be simplified to:
⎛Kw ⎞
RT ln⎜⎜ AB ⎟ = Δ t G o (Cl − , w → ms) − Δ t G o (Au(CN) −2 , w → ms)
ms ⎟
(6-24)
⎝ AB ⎠
K
or
⎛ K AB
w
⎞
RT ln⎜ ms ⎟⎟ = Δ t Gnet
⎜ o
(A − B, ms) (6-25)
⎝ K AB ⎠
181
The significance of the net Gibbs energy of transfer can now be further demonstrated.
w
Firstly, given the value of K AB , the equilibrium constant in a mixed solvent can be
estimated from the transfer energies of the exchanging anions. Secondly, the mixed
solvent composition at which there is a reversal in the selectivity of the resin can be
predicted using equation 6-25. For example, rearrangement of equation 6-25 leads to:
ms
RT ln K AB = RT ln K AB
w
− Δ t G net
o
(A − B, ms) (6-26)
If Δ t Gnet
o w
< RT ln K AB , the equilibrium constant in the mixed solvent is greater than 1
and the resin is selective for anion B. The resin is selective for anion A if Δ t Gnet
o
>
w
RT ln K AB . The composition of the mixed solvent at which Δ t Gnet
o w
= RT ln K AB marks
the point at which there is a reversal in the selectivity of the resin.
w
The generality of equation 6-25 suggests that a plot of RT ln( K AB ms
/ K AB ) versus Δ t Gnet
o
yields a straight line, with a gradient of 1, which is common to all mixed solvent
systems and for all combinations of the exchanging anions. As noted before, however,
the transfer energies required to calculate Δ t Gnet
o
for the exchanging anions in this study
w ms
are unavailable. Alternatively, the optimised values of K AB (Chapter 4) and K AB
In section 6.2.2, it was observed that KAB correlates with the dielectric constant of the
mixed solvent, although the value of KAB varies with the type of mixed solvent (see
Figures 6-11 and 6-12). The relationship between the loading of [Au(CN) 2 ] on the
−
presented in Figure 6-15. The loadings of [Au(CN)2 ] shown correspond to the resin in
−
equilibrium with a solution in which xB = 0.5. The same trends described below were
observed for xB values of 0.2 and 0.9. For both binary systems, an increase Δ t Gnet
o
is
associated with a decrease in the loading of [Au(CN)2 ] , i.e. KAB decreases with an
−
increase in mixed solvent composition. In contrast to Figure 6-11 and 6-12, it can be
182
Figure 6-15: The composition of [Au(CN) 2 ] in the resin phase as a function of the net
−
0.8
0.6
y Au
0.4
0.2
0
0 5 10 15 20 25 30
Net transfer energy
Au/Cl Ace Au/Cl DMSO
Au/Cl NMP Au/SCN Ace
Au/SCN DMSO Au/SCN NMP
183
seen that the variation in the loading of [Au(CN)2 ] is independent of the type of mixed
−
more solvated than ion A in the mixed solvent. In Figure 6-15, the values of Δ t Gnet
o
, for
both binary systems, are positive over the whole range of mixed solvent composition.
This appears to contradict some of the ideas about ion solvation introduced in Chapter
5. For example, in mixed solvents containing relatively low levels of organic solvent, it
was proposed that the solvation of the anions occurs mainly via hydration and that
[Au(CN) 2 ]− is less solvated than Cl − or SCN − . As such, the resin is selective for
Δ t Gnet
o
values, a more complex process appears to be controlling the selectivity of the
resin in mixed solvents.
indicates that ion A accumulates in the pore solution while ion B accumulates in the
mixed solvent. The higher concentration of ion A in the pore solution leads to an
increase in the loading of ion A on the resin and a corresponding decrease in the loading
of ion B. Note that the equilibrium constant for ion exchange between the resin and the
pore solution remains unchanged. However, the overall equilibrium constant for the ion
exchange process (equation 6-14) decreases to reflect the change in the selectivity of the
resin in the mixed solvent.
In the case of the Au(CN) 2− / Cl − and Au(CN) 2− / SCN − binary systems, the positive
values of Δ t Gnet
o
indicate that Cl − / SCN − ions gradually accumulate in the pore solution
as the mixed solvent composition increases. In mixed solvents containing relatively low
levels of organic solvent, the resin is still selective for [Au(CN) 2 ]− but the
redistribution of the ions leads to a reduction in the selectivity of the resin, relative to
that in aqueous solution. In mixed solvents containing greater than 60 mol% organic
184
solvent, the redistribution of the ions is such that the resin becomes selective for
Cl − / SCN − .
Undoubtedly, equation 6-25 is subject to the constraints of the assumptions made in its
derivation. These assumptions include: the pore solution is composed mainly of water
and the distribution of the counterions between the pore solution and the bulk mixed
solvent is governed by the solubility products of the corresponding salts of the
counterions. Nonetheless, this correlation highlights the importance of ion solvation in
ion exchange processes conducted in mixed solvents. In addition, equation 6-25
provides a direct relation between the selectivity of the resin in mixed solvents and the
Gibbs energies of transfer of the ions.
6.4 CONCLUSIONS
Optimised values of KAB for [Au(CN) 2 ] / Cl − , [Au(CN) 2 ] / SCN − and SCN − / Cl − in
− −
mixed solvents are consistent with the trends observed in the isotherms of these ion
exchange systems. The accuracy of the correlation results is similar to that obtained for
the ion exchange equilibria of the same species in aqueous solution. The fitted values of
the Wilson parameters generally conform to the reciprocity relation where the two
parameters are required to accurately correlate ion exchange equilibria. From these
results it can be concluded that the Law of Mass Action is equally valid in mixed
solvent systems.
The equilibrium constants for the [Au(CN) 2 ] / Cl − and [Au(CN) 2 ] / SCN − binary
− −
systems, in a given mixed solvent, correlate well with the dielectric constant of the
mixed solvent. For a given value of the dielectric constant, however, the equilibrium
constant is dependent on the type of mixed solvent. A fundamental relationship
between the equilibrium constant and the Gibbs transfer energies of the counterions has
been established in this work. The redistribution of ions between the pore solution and
the bulk mixed solvent, appears to be the most significant factor that governs the
selectivity of the resin in mixed solvent systems.
185
6.5 REFERENCES
Allen, R. M., Addison, P. A., and Dechapunya, A. H., The characterization of binary
and ternary ion exchange equilibria. Chem. Eng. J., 1989. 40: p. 151-158.
Frankel, L. S., Anion exchange resin solvent selectivity and anion solvation: a nuclear
magnetic resonance study. Canadian Journal of chemistry, 1970. 48(15): p. 2432-
2435.
Hovath, A. L., Handbook of aqueous electrolyte solutions. 1985, Ellis Horwood Ltd.,
London.
Kalidas, C., Hefter, G., and Marcus, Y., Gibbs energies of transfer of cations from
water to mixed aqueous organic solvents. Chemical Reviews, 2000. 100(3):
p.820-852.
Marcus, Y., Ion Properties. 1997, Marcel Dekker Inc, New York.
Marcus, Y., Ion Solvation. 1985, John Wiley and Sons, New York.
Marcus, Y. and Naveh, J., Anion exchange of metal complexes. XVII. Selective swelling
of the exchanger in mixed aqueous-organic solvents. Journal of Physical
Chemistry, 1969. 73(3): p. 591-596.
Muir, D. M., Singh, P., Kenna, C. C., Tsuchida, N., and Benari, M. D.,
Hydrometallurgical thermodynamics. II. Solvent effects on the activity and free
energies of transfer of cyanide (CN-), silver cyanide (Ag(CN)2-) and gold cyanide
(Au(CN)2-) in ethanol-water and acetonitrile-water mixtures. Aust. J. Chem.,
1985. 38(7): p. 1079-1090.
Mehablia, M., Shallcross, D. C., and Stevens, G. W., Prediction of multicomponent ion
exchange equilibria. Chemical Engineering Science, 1994. 49: p. 2277-2286.
Nandan, D., Gupta, A. R., and Shankar, J., Swelling and solvent fractionation
characteristics of PSS type ion exchange resins in methanol-water media. Indian
Journal of Chemsitry, 1972. 10: p. 83-87.
Parker, A. J., Protic – dipolar aprotic solvent effects on rates of bimolecular reactions.
Chemical Reviews, 1969. 69(1): p. 1 – 32.
186
Pitzer, K. S., Theory: ion interaction approach, in Activity Coefficients in Electrolyte
Solutions, Pytkowicz. (Ed), 1979. CRC Press, Florida. 1: p. 157-208.
Rajasingam, R., A novel hybrid process for the recovery of gold from ion exchange
resin. 2003, The University of New South Wales.
Shallcross, D. D., Herrmann, C. C., and McCoy, B. J., An improved model for the
prediction of multicomponent ion exchange equilibria. Chem. Eng. Sci., 1988. 43:
p. 279-288.
187
CHAPTER 7
7. CONCLUSIONS AND RECOMMENDATIONS ........................................ 188
7.1 MAJOR CONCLUSIONS FROM THIS WORK ...............................................188
7.2 RECOMMENDATIONS FOR FUTURE WORK ..............................................189
188
7. CONCLUSIONS AND RECOMMENDATIONS
binary systems in aqueous solution were obtained by exposing a volume of resin in the
chloride or thiocyanate form to a solution of the appropriate counterion. The resin was
contacted with the counterion until no further counterion was adsorbed. Ion exchange
equilibria obtained in this way for [Au(CN) 2 ] / Cl − and [Au(CN) 2 ] / SCN − binary
− −
systems in aqueous solution confirm the high affinity of Purolite A500 for the
[Au(CN)2 ]− species. Overall, the selectivity of the resin for a given ion increases in the
order Cl − < SCN − < [Au(CN)2 ]− . Furthermore, ion exchange equilibria are
independent of the total solution concentration within the range of concentrations
considered. The Law of Mass Action modified with activity coefficients provided a
satisfactory correlation of the ion exchange equilibria for all systems. The fitted values
of the equilibrium constants are in qualitative agreement with the selectivity of the resin
for the various ionic species.
The high affinity of anion exchange resin towards [Au(CN) 2 ] complicates subsequent
−
elution of this species from the resin. Mixed solvents were investigated in Chapter 5 as
a means of facilitating the elution of [Au(CN) 2 ] with simple counterions such as Cl −
−
188
SCN − / Cl − in mixed solvents were obtained by re-exposing resin in a given form to
counterion solution of the mixed solvent. The organic solvents investigated include
acetone, DMSO and NMP. In mixed solvents, the selectivity of the resin towards
[Au(CN) 2 ]− decreases with increasing composition of organic solvent in the external
solution. For all practical purposes, mixed solvents containing greater than 60 mol%
organic solvent are preferred for the elution of [Au(CN) 2 ] . Small weakly polarisable
−
anions, such as Cl − , are the preferred counterions in this range of mixed solvent
composition. Dipolar aprotic solvents with relatively low dielectric constants are more
effective in reducing the selectivity of the resin for [Au(CN) 2 ] .
−
In Chapter 6, the Law of Mass Action was used to obtain optimised values of KAB for
[Au(CN) 2 ]− / Cl − , [Au(CN) 2 ]− / SCN − and SCN − / Cl − in mixed solvents. Values of
KAB are consistent with the trends observed in the isotherms of these ion exchange
systems. The resilience of the modeling process has been confirmed by the fact that
predictions of KAB for mixed solvents were equally successful as those of aqueous
systems. From these results it can be concluded that the Law of Mass Action is equally
valid in mixed solvent systems.
The dielectric constant of a mixed solvent cannot alone be used to explain the dramatic
reversal in selectivity of the resin for the gold cyanide complex in mixed solvents. A
fundamental relationship between the equilibrium constant and the Gibbs transfer
energies of the counterions has been established in this work. The redistribution of ions
between the pore solution and the bulk mixed solvent appears to be the most significant
factor that governs the selectivity of the resin in mixed solvent systems.
189
cyanide complex requires further understanding of a multitude of factors affecting the
selectivity of the resin.
The method by which isotherms were obtained in the current work involved re-
equilibrating the same sample of resin to a constant volume and concentration of stock
solution over a number of exposure cycles. It is possible that these conditions may
inhibit the resin from attaining a high level of loading of counterion, particularly in
mixed solvents containing high levels of organic solvent. In future experiments, higher
loadings of [Au(CN) 2 ]
−
in the resin phase could be achieved by increasing the
resin in contact with the stock solution. In addition, the experimental methodology
could be modified to reduce the time required to establish an isotherm. Instead of re-
equilibrating the same sample of resin, a number of resin samples could be exposed to
various concentrations of the counterions representing the range of solution phase
compositions of an isotherm.
190
for the individual counterions need to be determined using a rigorous experimental
method such as solubility or potentiometry.
Ion exchange equilibria in mixed solvents also need to be determined for other ions
relevant to the extraction of the gold cyanide complex from cyanide solutions. Of these,
the cyanide complexes of copper are of particular interest in view of the problems
associated with the processing of copper-gold ores in Australia. The main difficulty
with copper is that the cyanide complexes exist in three different forms and this
introduces additional complexities in the modelling of ion exchange equilibria.
A pseudo-component approach could be used in this case in which the three copper
cyanide complexes are represented by a single species with an average charge number.
Ion exchange equilibria between the copper cyanide complex and a number of simple
counterions could then be determined. The selectivity of the resin for the copper
cyanide complex in the presence of the gold cyanide complex is also of particular
interest. The binary ion exchange equilibria could be used as a basis for predicting the
behaviour of more complex ternary and quaternary systems which are more
representative of the leach liquors produced in mining operations.
Three dipolar aprotic solvents were considered in this study: acetone, DMSO and NMP.
Of these, NMP was the most successful at facilitating the elution of [Au(CN) 2 ] . The
−
191
resemble NMP in terms of their dielectric constants and polarisabilities. Comparisons
could also be made with protic solvents.
Finally, there is some evidence in the literature that dipolar aprotic solvents form
complexes with large polarisable anions such as [Au(CN) 2 ] . The effect of these
−
192
A. APPENDIX 1
A-1
APPENDIX 1
This appendix contains the program code used to calculate the equivalent ionic fraction
of ion B, y Bcalc , using FORTRAN-77. Ion B represents the loading ion in a binary ion
exchange system. In the case of the [Au(CN ) 2 ] / Cl − and [Au(CN ) 2 ] / SCN − binary
− −
systems ion B is [Au(CN) 2 ] . In the case of the SCN − / Cl − binary system ion B is
−
SCN − . Generally, for all aqueous systems the program code for the 3-parameter
regression is identical. The main difference between binary systems is the number of
data points and the Debye-Hückel parameter, a, which varies depending on the size of
the exchanging ions (Table 4-3). For the 2-parameter regression, the equilibrium
constant previously determined using the Gaines and Thomas approach, is used as a
constant in the program.
In the case of mixed solvent systems, the Debye-Hückel constants A and B vary for each
composition of solvent. These values can be found in Table 6-1. As an example the
program code for the [Au(CN ) 2 ] / Cl − binary system in aqueous solution is presented.
−
In this instance, 58 data points are use in the regression. XB and YB refer to the
experimental equivalent ionic fractions in the solution and resin phase respectively. SI
refers to the ionic strength in solution determined according to equation 4-7. Comments
explaining the program are preceded by the symbol ‘!’.
! Program for fitting KAB and Wilson coefficients from ion exchange
! equilibrium data. Ion 'A' is initially loaded on the resin
! while ion 'B' is the counter ion.
! The program reads the experimental values of XB(I), YB(I) and SI(I)
! from the file specified in the OPEN statement (unit=2).
! The fitted values of KAB and the Wilson coefficients are obtained
A-1
! by minimising the SSRD with respect to YB(I).
OPEN(2,FILE='faucldata.dat',FORM='FORMATTED',STATUS='OLD')
OPEN(3,FILE='raucl3.dat',FORM='FORMATTED',STATUS='NEW')
DO 10 I=1,NDP
READ(2,*) YB(I)
10 CONTINUE
DO 12 I=1,NDP
READ(2,*) SI(I)
12 CONTINUE
DO 15 I=1,M
YBCALC(I) = YB(I)*(F(I) + 1.0D0)
WRITE (3,50) "YBCALC(",I,") , ",YBCALC(I)
15 CONTINUE
SUMARD = 0.0D0
DO 20 I=1,M
A-2
SUMARD = SUMARD + ABS(F(I))
20 CONTINUE
AARD = 100.0D0*SUMARD/ANDP
! ----------------------------------------------------------------------
SUBROUTINE RESID (X,M,N,F)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
DIMENSION X(N),F(M)
DIMENSION XB(M),YB(M),SI(M),PHI(M),YBCALC(M), &
GAMAL(M),GAMBL(M),GAMAR(M),GAMBR(M),DENOM(M)
OPEN(4,FILE='aucldata.dat',FORM='FORMATTED',STATUS='OLD')
DO 5 I=1,M
READ(4,*) XB(I)
5 CONTINUE
DO 10 I=1,M
READ(4,*) YB(I)
10 CONTINUE
DO 15 I=1,M
READ(4,*) SI(I)
15 CONTINUE
CLOSE (UNIT=4)
A-3
GAMBR(I) = GAMR(2,YB(I),X(1),X(2))
20 CONTINUE
RETURN
END
! ----------------------------------------------------------------------
DOUBLE PRECISION FUNCTION GAML (A,B,D,S)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
GAML = 10**(-1.0*A*SQRT(S)/(1.0+B*D*SQRT(S)))
RETURN
END
! ----------------------------------------------------------------------
DOUBLE PRECISION FUNCTION GAMR (N,Y,PAB,PBA)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
RETURN
END
! ----------------------------------------------------------------------
SUBROUTINE ZXSSQ (FUNC,M,N,NSIG,EPS,DELTA,MAXFN,IOPT, &
PARM,X,SSQ,F,XJAC,IXJAC,XJTJ,WORK,INFER,IER)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
DIMENSION X(45),F(58),PARM(4),XJAC(58),XJTJ(6),WORK(137)
INTEGER U1,U5,U4
COMMON/UNITA/U1,U5,U4
DATA SIG/9.3D0/
DATA AX/0.1D0/
DATA I0,I1,I2,I3,I4,I5/0,1,2,3,4,5/
DATA P01,TENTH,HALF,ZERO,ONE,ONEP5,TWO, &
TEN,HUNTW,ONEP10/0.01D0,0.1D0,0.5D0,0.0D0, &
A-4
1.D0,1.5D0,2.D0,10.0D0,1.2D2,1.D10/
! ERROR CHECKS
IER = I0
IF (M.LE.0.OR.M.GT.IXJAC.OR.N.LE.0.OR.IOPT.LT.0.OR.IOPT.GT.2) &
GOTO 305
IMJC=IXJAC-M
IF (IOPT.NE.2) GOTO 5
IF (PARM(I2).LE.ONE.OR.PARM(I1).LE.ZERO) GOTO 305
! MACHINE DEPENDENT CONSTANTS
5 PREC = TEN**(-SIG-ONE)
REL = TEN**(-SIG*HALF)
RELCON = TEN**(-NSIG)
! WORK VECTOR IS ONCATENATION OF
! SCALED HESSIAN,GRADIENT,DELX,SCALE,
! XNEW,XBAD,F(X+DEL),F(X-DEL)
IGRAD1 = ((N+1)*N)/2
IGRADL = IGRAD1+1
IGRADU = IGRAD1+N
IDELX1 = IGRADU
IDELXL = IDELX1+1
IDELXU = IDELX1+N
ISCAL1 = IDELXU
ISCALL = ISCAL1+1
ISCALU = ISCAL1+N
IXNEW1 = ISCALU
IXNEWL = IXNEW1+1
IXBAD1 = IXNEW1+N
IFPL1 = IXBAD1+N
IFPL = IFPL1+1
IFPU = IFPL1+M
IFML1 = IFPU
IFML = IFML1+1
IMJC = IXJAC - M
! INITIALIZE VARIABLES
AL = ONE
CONS2 = TENTH
IF (IOPT.EQ.0) GOTO 20
IF (IOPT.EQ.1) GOTO 10
AL = PARM(I1)
F0 = PARM(I2)
UP = PARM(I3)
CONS2 = PARM(I4)*HALF
GOTO 15
10 AL = P01
F0 = TWO
UP = HUNTW
15 ONESF0 = ONE/F0
F0SQ = F0*F0
F0SQS4 = F0SQ**4
20 IEVAL = 0
DELTA2 = DELTA*HALF
ERL2 = ONEP10
IBAD = -99
ISW = 1
ITER = -1
INFER = 0
IER = 0
DO 25 J=IDELXL,IDELXU
WORK(J) = ZERO
A-5
25 CONTINUE
GOTO 165
! MAIN LOOP
30 SSQOLD = SSQ
! CALCULATE JACOBIAN
IF (INFER.GT.0.OR.IJAC.GE.N.OR.IOPT.EQ.0.OR.ICOUNT.GT.0) GOTO 55
! RANK ONE UPDATE TO JACOBIAN
IJAC = IJAC+1
DSQ = ZERO
DO 35 J=IDELXL,IDELXU
DSQ = DSQ+WORK(J)*WORK(J)
35 CONTINUE
IF (DSQ.LE.ZERO) GOTO 55
DO 50 I=1,M
II=IFML1+I
G = F(I)-WORK(II)
K=I
DO 40 J=IDELXL,IDELXU
G = G+XJAC(K)*WORK(J)
K = K+IXJAC
40 CONTINUE
G = G/DSQ
K=I
DO 45 J=IDELXL,IDELXU
XJAC(K) = XJAC(K)-G*WORK(J)
K = K+IXJAC
45 CONTINUE
50 CONTINUE
GOTO 80
! JACOBIAN BY INCREMENTING X
55 IJAC = 0
K = -IMJC
DO 75 J=1,N
K = K+IMJC
XDABS =DABS(X(J))
HH =REL*(DMAX1(XDABS,AX))
XHOLD = X(J)
X(J) = X(J)+HH
! print*, "about to call func - 1", ",",n
CALL FUNC (X,M,N,WORK(IFPL))
! print*, "call to func - 1 completed", ",",n
IEVAL = IEVAL+1
X(J) = XHOLD
! print*, "isw=",isw
IF (ISW.EQ.1) GOTO 65
! CENTRAL DIFFERENCES
X(J) = XHOLD-HH
! print*, "about to call func - 2", n
CALL FUNC (X,M,N,WORK(IFML))
IEVAL = IEVAL+1
X(J) = XHOLD
RHH = HALF/HH
DO 60 I=IFPL,IFPU
II=I+M
K = K+1
XJAC(K) = (WORK(I)-WORK(II))*RHH
60 CONTINUE
GOTO 75
A-6
! FORWARD DIFFERENCES
65 RHH = ONE/HH
DO 70 I=1,M
K = K+1
II=IFPL1+I
XJAC(K) = (WORK(II)-F(I))*RHH
! print*, "xjac(",k,")=", xjac(k)
70 CONTINUE
75 CONTINUE
! CALCULATE GRADIENT
80 ERL2X = ERL2
ERL2 = ZERO
K = -IMJC
DO 90 J=IGRADL,IGRADU
K = K+IMJC
SUM = ZERO
DO 85 I=1,M
K = K+1
SUM = SUM+XJAC(K)*F(I)
85 CONTINUE
! print*, "loop 85 completed"
WORK(J) = SUM
ERL2 = ERL2+SUM*SUM
90 CONTINUE
! print*, "loop 90 completed"
ERL2 = SQRT(ERL2)
! CONVERGENCE TEST FOR NORM OF GRADIENT
! print*, "ijac=",ijac
IF(IJAC.GT.0) GOTO 95
IF (ERL2.LE.DELTA2) INFER = INFER+4
IF (ERL2.LE.CONS2) ISW = 2
! CALCULATE THE LOWER SUPER TRIANGE OF
! JACOBIAN (TRANSPOSED) * JACOBIAN
95 L = 0
IS = -IXJAC
DO 110 I=1,N
IS = IS+IXJAC
JS = -IXJAC
DO 105 J=1,I
JS = JS+IXJAC
L = L+1
SUM = ZERO
DO 100 K=1,M
LI = IS+K
LJ = JS+K
SUM = SUM+XJAC(LI)*XJAC(LJ)
100 CONTINUE
XJTJ(L) = SUM
105 CONTINUE
110 CONTINUE
! print*, "loop 110 completed"
! ONVERGENCE CHECKs
! print*, "infer=", infer
IF (INFER.GT.0) GOTO 315
! print*, "ieval=", ieval
IF (IEVAL.GE.MAXFN) GOTO 290
! COMPUTE SCALING VECTOR
! print*, "iopt=",iopt
IF (IOPT.EQ.0) GOTO 120
K=0
A-7
DO 115 J=1,N
K = K+J
II=ISCAL1+J
WORK(II) = XJTJ(K)
115 CONTINUE
! print*, "loop 115 completed"
GOTO 135
! COMPUTE SCALING VECTOR AND NORM
120 DNORM = ZERO
K=0
DO 125 J=1,N
K = K+J
II=ISCAL1+J
WORK(II) = SQRT(XJTJ(K))
DNORM = DNORM+XJTJ(K)*XJTJ(K)
125 CONTINUE
DNORM = ONE/SQRT(DNORM)
! NORMALIZE SCALING VECTOR
DO 130 J=ISCALL,ISCALU
WORK(J) = WORK(J)*DNORM*ERL2
130 CONTINUE
! ADD L-M FACTOR TO DIAGONAL
135 ICOUNT = 0
140 K = 0
DO 150 I=1,N
DO 145 J=1,I
K = K+1
WORK(K) = XJTJ(K)
145 CONTINUE
II=ISCAL1+I
IJ=IGRAD1+I
WORK(K) = WORK(K)+WORK(II)*AL
II=IDELX1+I
WORK(II) = WORK(IJ)
150 CONTINUE
! print*, "loop 150 completed"
! CHOLESKY DECOMPOSITION
! print*, "about to call leq in subr zxssq"
155 CALL LEQ (WORK,1,N,WORK(IDELXL),N,0,G,XHOLD,IER)
! print*, "call to leq done"
IF (IER.EQ.0) GOTO 160
IER = 0
IF (IJAC.GT.0) GOTO 55
IF (IBAD.LE.0) GOTO 240
IF (IBAD.GE.2) GOTO 310
GOTO 190
160 IF (IBAD.NE.-99) IBAD = 0
! CALCULATE SUM OF SQUARES
165 DO 170 J=1,N
II=IXNEW1+J
IJ=IDELX1+J
WORK(II) = X(J)-WORK(IJ)
170 CONTINUE
! print*, "about to call func - 3"
CALL FUNC (WORK(IXNEWL),M,N,WORK(IFPL))
IEVAL = IEVAL+1
SSQ = ZERO
DO 175 I=IFPL,IFPU
SSQ = SSQ+WORK(I)*WORK(I)
A-8
175 CONTINUE
IF (ITER.GE.0) GOTO 185
! SSQ FOR INITIAL ESTIMATES OF X
ITER = 0
SSQOLD = SSQ
DO 180 I=1,M
II=IFPL1+I
F(I) = WORK(II)
180 CONTINUE
GOTO 55
185 IF (IOPT.EQ.0) GOTO 215
! CHECK DESCENT PROPERTY
IF (SSQ.LE.SSQOLD) GOTO 205
! INREASE PAAMETER AND TRY AGAIN
190 ICOUNT = ICOUNT+1
AL = AL*F0SQ
IF (IJAC.EQ.0) GOTO 195
IF (ICOUNT.GE.4.OR.AL.GT.UP) GOTO 200
195 IF (AL.LE.UP) GOTO 140
IF (IBAD.EQ.1) GOTO 310
GOTO 300
200 AL = AL/F0SQS4
GOTO 55
! ADJUST MARQUARDT PARAMETER
205 IF (ICOUNT.EQ.0) AL = AL/F0
IF (ERL2X.LE.ZERO) GOTO 210
G = ERL2/ERL2X
IF (ERL2.LT.ERL2X) AL = AL*DMAX1(ONESF0,G)
IF (ERL2.GT.ERL2X) AL = AL*DMIN1(F0,G)
210 AL = DMAX1(AL,PREC)
! ONE ITERATION CYCLE COMPLETED
215 ITER = ITER+1
DO 220 J=1,N
IJ=IXNEW1+J
X(J) = WORK(IJ)
220 CONTINUE
DO 225 I=1,M
II=IFML1+I
WORK(II) = F(I)
II =IFPL1+I
F(I) = WORK(II)
225 CONTINUE
! RELATIVE CONVERGENCE TEST FOR X
IF (AL .GT. 5.0D0) GOTO 30
DO 230 J=1,N
IJ=IDELX1+J
XDIF =DABS(WORK(IJ))/DMAX1(DABS(X(J)),AX)
IF (XDIF.GT.RELCON) GOTO 235
230 CONTINUE
INFER = 1
! RELATIVE CONVERGENCE TEST FOR SSQ
235 SQDIF = DABS(SSQ-SSQOLD)/DMAX1(SSQOLD,AX)
IF (SQDIF.LE.EPS) INFER = INFER+2
GOTO 30
! SINGULAR DECOMPOSITION
240 IF (IBAD) 255,245,265
! CHECK TO SEE IF CURRENT
! ITERATE HAS CYCLED BACK TO
! THE LAST SINGULAR POINT
245 DO 250 J=1,N
A-9
IJ=IXBAD1+J
XHOLD = WORK(IJ)
IF (DABS(X(J)-XHOLD).GT.RELCON*DMAX1(AX,DABS(XHOLD))) GOTO 255
250 CONTINUE
GOTO 295
! UPDATE THE BAD X VALUES
255 DO 260 J=1,N
IJ=IXBAD1+J
WORK(IJ) = X(J)
260 CONTINUE
IBAD = 1
! INCREASE DIAGONAL OF HESSIAN
265 IF (IOPT.NE.0) GOTO 280
K=0
DO 275 I=1,N
DO 270 J=1,I
K = K+1
WORK(K) = XJTJ(K)
270 CONTINUE
II=ISCAL1+I
WORK(K) = ONEP5*(XJTJ(K)+AL*ERL2*WORK(II))+REL
275 CONTINUE
IBAD = 2
GOTO 155
! REPLACE ZEROES ON HESSIAN DIAGONAL
280 IZERO = 0
DO 285 J=ISCALL,ISCALU
IF (WORK(J).GT.ZERO) GOTO 285
IZERO = IZERO+1
WORK(J) = ONE
285 CONTINUE
IF (IZERO.LT.N) GOTO 140
IER = 38
GOTO 315
! TEMINAL ERROR
290 IER = IER+1
295 IER = IER+1
300 IER = IER+1
305 IER = IER+1
310 IER = IER+129
IF (IER.EQ.130) GOTO 9000
! OUTPUT ERL2,IEVAL,NSIG,AL, AND ITER
315 G = SIG
DO 320 J=1,N
IJ=IDELX1+J
XHOLD = DABS(WORK(IJ))
IF (XHOLD.LE.ZERO) GOTO 320
G = DMIN1(G,-DLOG10(XHOLD/DMAX1(AX,DABS(X(J)))))
320 CONTINUE
IF(N.GT.2) GOTO 330
DO 325 J = 1,N
II=IGRAD1+J
325 WORK(J+5) = WORK(II)
330 WORK(I1) = ERL2+ERL2
WORK(I2) = IEVAL
SSQ = SSQOLD
WORK(I3) = G
WORK(I4) = AL
WORK(I5) = ITER
A-10
IS = G
IF(IS .GE. NSIG) IER=0
IF (IER.EQ.0) GOTO 9005
9000 CONTINUE
WRITE(U1,9001) IER
IF(IER.EQ.129.OR.IER.EQ.132) WRITE(U1,9129)
IF(IER.EQ.130) WRITE(U1,9130) M,N,IOPT,(PARM(I),I=1,2)
IF(IER.EQ.131) WRITE(U1,9131) WORK(I4),PARM(I3)
IF(IER.EQ.133) WRITE(U1,9133) MAXFN
IF(IER.EQ.38) WRITE(U1,9038)
WRITE(U1,9050) WORK(I1),IS,WORK(I4),ITER,IEVAL
9005 CONTINUE
!
! FORMAT STATEMENTS
!
9001 FORMAT(1H ,' ZXSSQ ERROR NO:',I3)
9038 FORMAT(1H ,'TRIVIAL SOLUTION')
9050 FORMAT(2X,'NORM.GRAD=',E10.4,2X,'NO.OF SIGNIFICANT DIGITS:', &
I5,/,3X,'MARQUARDT PARAMETER:',E10.4,/ &
2X,I4,' ITERATIONS;',2X,I4,' FUNCTION EVALUATIONS.')
9129 FORMAT(1H ,'SINGULAR JACOBIAN')
9130 FORMAT(1H ,'INCORRECT: M=',I5,' N=',I5,' IOPT=',I5, &
' PARM(1)=',E10.4,' PARM(2)=',E10.4)
9131 FORMAT(1H ,'MARQUARDT PARAMETER ',E10.4, &
' EXCEED PARM(3)',E10.4)
9133 FORMAT(1H ,'FUNCTION EVALUATIONS EXCEED MAXFN',I5)
RETURN
END
A-11
IP = 1
IER=0
DO 45 I = 1,N
IQ =IP
IR = 1
DO 40 J=1,I
X = A(IP)
IF (J .EQ. 1) GOTO 10
DO 5 K=IQ,IP1
X= X - UL(K) * UL(IR)
IR = IR+1
5 CONTINUE
10 IF (I.NE.J) GOTO 30
D1 = D1*X
IF (A(IP) + X*RN .LE. A(IP)) GOTO 50
15 IF (DABS(D1).LE.ONE) GOTO 20
D1 = D1 * SIXTH
D2 = D2 + FOUR
GOTO 15
20 IF (DABS(D1) .GE. SIXTH) GOTO 25
D1 = D1 * SIXTN
D2 = D2 - FOUR
GOTO 20
25 UL(IP) = ONE/DSQRT(X)
GOTO 35
30 UL(IP) = X * UL(IR)
35 IP1 = IP
IP = IP+1
IR = IR+1
40 CONTINUE
45 CONTINUE
GOTO 9005
50 IER = 129
9005 RETURN
END
SUBROUTINE ELM (A,B,N,X)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
DIMENSION A(1),B(1),X(1)
DATA ZERO/0.0d0/
! SOLUTION OF LY = B
IP=1
IW = 0
DO 15 I=1,N
T=B(I)
IM1 = I-1
IF (IW .EQ. 0) GOTO 9
IP=IP+IW-1
DO 5 K=IW,IM1
T = T-A(IP)*X(K)
IP=IP+1
5 CONTINUE
GOTO 10
9 IF (T .NE. ZERO) IW = I
IP = IP+IM1
10 X(I)=T*A(IP)
IP=IP+1
15 CONTINUE
! SOLUTION OF UX = Y
N1 = N+1
DO 30 I = 1,N
A-12
II = N1-I
IP=IP-1
IS=IP
IQ=II+1
T=X(II)
IF (N.LT.IQ) GOTO 25
KK = N
DO 20 K=IQ,N
T = T - A(IS) * X(KK)
KK = KK-1
IS = IS-KK
20 CONTINUE
25 X(II)=T*A(IS)
30 CONTINUE
RETURN
END
A-13
A. APPENDIX 2
A-14
APPENDIX 2
Ion exchange isotherms in this study were plotted with respect to equivalent ionic
fraction data of the loading ion. The equivalent ionic fractions in the solution phase (xB)
and resin phase (yB) were determined according to equations 4-3 and 4-4. The
following tables present experimental equivalent ionic faction data for the binary
systems [Au(CN ) 2 ] / Cl − , [Au(CN ) 2 ] / SCN − and SCN − / Cl − in aqueous solution.
− −
Predicted resin phase compositions, y Bcalc were determined using the law of mass action
as outlined in Appendix 1. The following tables also present the deviations between
y Bexp and y Bcalc for each binary system.
A-14
0.0051 0.5395 0.0353 0.9229 0.9982 0.0015
0.0060 0.5779 0.0345
total solution concentration = 0.0051 M
0.0002 0.0776 -0.0477 0.0099 0.6966 0.0219
0.0005 0.1553 -0.0555 0.0129 0.7656 0.0008
0.0011 0.2328 -0.0157 0.0201 0.8379 -0.0029
0.0015 0.3104 -0.0263 0.0423 0.9123 0.0023
0.0023 0.3879 -0.0063 0.2280 0.9722 0.0138
0.0034 0.4653 0.0113 0.6633 0.9984 -0.0005
0.0047 0.5426 0.0121 0.9592 1.0000 -0.0002
0.0067 0.6197 0.0169
total solution concentration = 0.0066 M
0.0003 0.1029 -0.0510 0.0201 0.8183 0.0175
0.0009 0.2057 -0.0254 0.0441 0.9167 0.0012
0.0016 0.3084 -0.0110 0.3376 0.9848 0.0071
0.0027 0.4111 0.0094 0.8978 0.9947 0.0048
0.0044 0.5135 0.0262 0.9587 0.9973 0.0025
0.0071 0.6157 0.0347 0.9652 1.0000 -0.0001
0.0115 0.7174 0.0298
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
Table A2-2: Ion exchange equilibria for Au(CN) −2 (B) / SCN − (A ) in aqueous solution
at 303 K and for various total solution concentrations
xB yB δyBa xB yB δyBa
total solution concentration = 0.0025 M
0.0020 0.0349 -0.0070 0.1343 0.6952 0.0009
0.0051 0.0725 -0.0042 0.1460 0.7250 -0.0062
0.0077 0.1086 -0.0089 0.1778 0.7538 0.0126
0.0122 0.1446 0.0049 0.1974 0.7804 0.0109
0.0147 0.1804 -0.0056 0.2249 0.8075 0.0126
0.0179 0.2162 -0.0109 0.2569 0.8335 0.0138
0.0228 0.2517 -0.0035 0.2835 0.8585 0.0083
0.0246 0.2759 -0.0130 0.3024 0.8829 -0.0029
A-15
0.0339 0.3097 0.0217 0.3465 0.9058 -0.0034
0.0375 0.3414 0.0134 0.4405 0.9244 0.0095
0.0408 0.3750 0.0001 0.5232 0.9411 0.0119
0.0477 0.4083 0.0060 0.6321 0.9540 0.0162
0.0545 0.4413 0.0076 0.6832 0.9650 0.0116
0.0586 0.4742 -0.0058 0.8016 0.9720 0.0156
0.0643 0.5069 -0.0131 0.7626 0.9803 0.0044
0.0673 0.5395 -0.0327 0.7916 0.9876 -0.0003
0.0780 0.5718 -0.0242 0.8327 0.9934 -0.0030
0.0871 0.6037 -0.0251 0.9279 0.9959 0.0004
0.0966 0.6352 -0.0274 0.9546 0.9975 0.0002
0.1059 0.6649 -0.0311 0.9888 1.0000 -0.0005
total solution concentration = 0.0051 M
0.0056 0.0716 0.0030 0.1503 0.7382 -0.0112
0.0137 0.1426 0.0226 0.2021 0.7957 0.0022
0.0196 0.2132 0.0078 0.2677 0.8484 0.0078
0.0279 0.2832 0.0056 0.3486 0.8953 0.0067
0.0424 0.3521 0.0327 0.4692 0.9327 0.0088
0.0519 0.4204 0.0158 0.5924 0.9615 0.0040
0.0621 0.4879 -0.0036 0.7499 0.9791 0.0045
0.0756 0.5544 -0.0157 0.8422 0.9903 0.0007
0.0944 0.6196 -0.0186 0.9147 0.9963 -0.0007
0.1158 0.6801 -0.0220 0.9429 1.0000 -0.0029
total solution concentration = 0.0066 M
0.0068 0.0897 -0.0005 0.1712 0.7577 0.0010
0.0158 0.1785 0.0074 0.2409 0.8262 0.0090
0.0250 0.2666 -0.0002 0.3468 0.8818 0.0180
0.0364 0.3536 -0.0055 0.4765 0.9291 0.0137
0.0590 0.4385 0.0316 0.6269 0.9628 0.0073
0.0750 0.5220 0.0137 0.7759 0.9830 0.0029
0.0958 0.6037 0.0005 0.8904 0.9929 0.0012
0.1230 0.6829 -0.0092 0.9128 1.0000 -0.0045
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
A-16
Table A2-3: Ion exchange equilibria for SCN − (B) / Cl − (A ) in aqueous solution at 303
K and for various total solution concentrations
xB yB δyBa xB yB δyBa
total solution concentration = 0.0086 M
0.0175 0.1352 0.0134 0.1454 0.7747 0.0173
0.0254 0.2694 0.0155 0.2517 0.8777 0.0070
0.0351 0.4022 0.0002 0.4872 0.9483 0.0080
0.0571 0.5320 0.0213 0.7149 0.9876 -0.0046
0.0910 0.6571 0.0272 0.9050 1.0000 -0.0045
total solution concentration = 0.0172 M
0.0195 0.2570 -0.0282 0.7414 0.9648 0.0203
0.0516 0.5057 0.0173 0.9157 0.9869 0.0091
0.1321 0.7332 0.0383 0.9714 0.9944 0.0043
0.3750 0.8970 0.0352 0.9786 1.0000 -0.0009
total solution concentration = 0.0224 M
0.0186 0.3397 -0.0936 0.9520 0.9890 0.0088
0.0914 0.6542 0.0313 0.9769 0.9970 0.0020
0.3441 0.8813 0.0418 0.9937 1.0000 -0.0003
0.7368 0.9724 0.0124
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
A-17
A. APPENDIX 3
A-18
APPENDIX 3
Ion exchange isotherms in this study were plotted with respect to equivalent ionic
fraction data of the loading ion. The equivalent ionic fraction in the solution phase (xB)
and resin phase (yB) were determined according to equations 4-3 and 4-4. The
following tables present experimental equivalent ionic faction data for the
[Au(CN) 2 ]− / Cl − , [Au(CN) 2 ]− / SCN − and SCN − / Cl − binary systems in various
mixed solvents. Predicted resin phase compositions, y Bcalc were determined using the
law of mass action as outlined in Appendix 1. The following tables also present the
deviations between y Bexp and y Bcalc for each binary system.
A-18
0.1395 0.4451 0.0166 0.6808 0.9221 -0.0385
0.1799 0.5100 0.0078 0.7404 0.9425 -0.0331
0.2227 0.5716 -0.0035 0.8107 0.9574 -0.0200
0.2687 0.6290 -0.0137 0.8619 0.9682 -0.0123
0.3136 0.6829 -0.0276 0.9023 0.9758 -0.0062
organic solvent composition = 26.7 mol%
0.0574 0.0755 -0.0052 0.5943 0.4580 0.0143
0.1406 0.1442 -0.0006 0.6097 0.4889 -0.0089
0.2316 0.2058 0.0063 0.6402 0.5175 -0.0121
0.3137 0.2607 0.0081 0.6475 0.5454 -0.0387
0.4069 0.3082 0.0296 0.6736 0.5713 -0.0414
0.4530 0.3520 0.0140 0.6939 0.5956 -0.0475
0.5146 0.3908 0.0215 0.6974 0.6196 -0.0724
0.5632 0.4258 0.0239
organic solvent composition = 68.6 mol%
0.4413 0.0438 -0.0034 0.7816 0.1270 -0.0081
0.6303 0.0728 0.0008 0.8248 0.1407 0.0040
0.7284 0.0898 0.0143 0.8325 0.1537 -0.0075
0.7449 0.1098 -0.0050
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
Table A3-2: Ion exchange equilibria for Au(CN) −2 (B) / SCN − (A ) in acetone-water
mixtures at 303 K.
xB yB δyBa xB yB δyBa
organic solvent composition = 5.73 mol%
0.0096 0.0777 -0.0040 0.1636 0.6514 -0.0103
0.0218 0.1544 -0.0007 0.1991 0.7137 -0.0059
0.0371 0.2299 0.0092 0.2378 0.7729 -0.0032
0.0524 0.3042 0.0072 0.2781 0.8290 -0.0026
0.0696 0.3771 0.0040 0.3150 0.8828 -0.0054
0.0918 0.4484 0.0087 0.3774 0.9316 0.0016
0.1133 0.5179 0.0024 0.4318 0.9762 0.0013
A-19
0.1333 0.5859 -0.0124 0.5167 1.0000 0.0000
organic solvent composition = 13.9 mol%
0.0441 0.0757 0.0101 0.4474 0.7084 0.0077
0.0733 0.1491 -0.0083 0.4869 0.7490 0.0133
0.1070 0.2199 -0.0178 0.5138 0.7876 0.0095
0.1497 0.2873 -0.0115 0.5556 0.8228 0.0153
0.1819 0.3456 -0.0155 0.5849 0.8558 0.0129
0.2239 0.4070 -0.0097 0.6064 0.8870 0.0066
0.2624 0.4655 -0.0081 0.6282 0.9165 0.0003
0.3159 0.5197 0.0129 0.6683 0.9428 -0.0010
0.3469 0.5710 0.0077 0.6969 0.9668 -0.0073
0.3797 0.6197 0.0059 0.7181 0.9892 -0.0162
0.4284 0.6645 0.0203 0.7597 1.0000 -0.0181
organic solvent composition = 26.7 mol%
0.1187 0.0705 0.0020 0.4942 0.4364 0.0155
0.1750 0.1366 -0.0084 0.5408 0.4729 0.0261
0.2317 0.1981 -0.0118 0.5437 0.5091 -0.0072
0.3033 0.2539 0.0038 0.5622 0.5438 -0.0235
0.3698 0.3043 0.0204 0.5909 0.5766 -0.0275
0.4059 0.3519 0.0107 0.6053 0.6082 -0.0450
0.4493 0.3959 0.0107
organic solvent composition = 68.6 mol%
0.2876 0.0559 0.0001 0.6381 0.2416 0.0097
0.3945 0.1033 -0.0010 0.6642 0.2677 0.0044
0.4771 0.1443 0.0008 0.6750 0.2930 -0.0130
0.5324 0.1810 -0.0030 0.6934 0.3170 -0.0219
0.5891 0.2132 0.0018 0.7509 0.3364 0.0196
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
Table A3-3: Ion exchange equilibria for SCN − (B) / Cl − (A ) in acetone-water mixtures
at 303 K.
xB yB δyBa xB yB δyBa
A-20
organic solvent composition = 5.73 mol%
0.0148 0.1545 -0.0004 0.3509 0.8014 0.0287
0.0378 0.3054 0.0028 0.4830 0.8825 0.0080
0.0710 0.4512 -0.0032 0.5840 0.9477 -0.0256
0.1228 0.5888 -0.0062 0.7781 0.9825 -0.0157
0.1967 0.6996 -0.0009 0.8550 1.0000 -0.0202
organic solvent composition = 13.9 mol%
0.0331 0.1532 -0.0060 0.6382 0.8624 -0.0263
0.0962 0.2964 0.0137 0.6988 0.9052 -0.0368
0.1788 0.4265 0.0223 0.7943 0.9378 -0.0222
0.2621 0.5434 0.0056 0.8647 0.9592 -0.0124
0.3453 0.6472 -0.0181 0.8857 0.9773 -0.0219
0.4590 0.7329 -0.0103 0.8987 0.9934 -0.0328
0.5441 0.8051 -0.0253 0.9457 1.0000 -0.0205
organic solvent composition = 26.7 mol%
0.1005 0.1440 -0.0114 0.7360 0.7558 0.0003
0.2267 0.2678 0.0092 0.7608 0.7849 -0.0035
0.3557 0.3710 0.0379 0.8343 0.8114 0.0407
0.4056 0.4661 -0.0153 0.8380 0.8676 -0.0014
0.5028 0.5272 0.0151 0.8534 0.8911 -0.0042
0.5461 0.5999 -0.0217 0.8611 0.9087 -0.0083
0.6292 0.6593 -0.0044 0.8945 0.9256 0.0067
0.6609 0.7136 -0.0300
organic solvent composition = 68.6 mol%
0.3612 0.1002 -0.0175 0.7692 0.2903 0.0215
0.5703 0.1676 -0.0001 0.7906 0.3232 0.0085
0.6807 0.2176 0.0190 0.7918 0.3558 -0.0304
0.7467 0.2574 0.0352 0.8075 0.3860 -0.0467
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
A-21
organic solvent composition = 5.91 mol%
0.0006 0.0765 0.0037 0.0292 0.6790 -0.0133
0.0014 0.1529 -0.0101 0.0407 0.7521 -0.0253
0.0025 0.2289 -0.0175 0.0649 0.8233 -0.0176
0.0048 0.3047 0.0111 0.1157 0.8907 -0.0083
0.0076 0.3802 0.0221 0.3760 0.9384 0.0327
0.0106 0.4556 0.0071 0.8330 0.9512 0.0452
0.0158 0.5305 0.0159 0.9588 0.9543 0.0449
0.0208 0.6051 -0.0059
organic solvent composition = 14.3 mol%
0.0013 0.0772 -0.0068 0.1794 0.8002 -0.0560
0.0046 0.1542 0.0057 0.2580 0.8575 -0.0414
0.0106 0.2307 0.0211 0.3862 0.9050 -0.0189
0.0185 0.3065 0.0205 0.5896 0.9367 0.0090
0.0292 0.3816 0.0166 0.8177 0.9508 0.0310
0.0421 0.4557 0.0039 0.9168 0.9573 0.0352
0.0567 0.5286 -0.0168 0.9618 0.9602 0.0365
0.0732 0.6002 -0.0418 0.9717 0.9624 0.0352
0.1025 0.6696 -0.0400 0.9770 0.9642 0.0338
0.1321 0.7367 -0.0560
organic solvent composition = 27.4 mol%
0.0096 0.0766 -0.0098 0.4945 0.7073 -0.0390
0.0344 0.1512 0.0103 0.5449 0.7425 -0.0370
0.0688 0.2232 0.0186 0.5904 0.7742 -0.0363
0.1118 0.2919 0.0225 0.6327 0.8026 -0.0355
0.1585 0.3570 0.0186 0.6749 0.8277 -0.0318
0.2132 0.4178 0.0204 0.6953 0.8513 -0.0428
0.2463 0.4761 -0.0127 0.7495 0.8706 -0.0253
0.2982 0.5303 -0.0188 0.7991 0.8862 -0.0082
0.3430 0.5811 -0.0328 0.8727 0.8960 0.0289
0.4077 0.6269 -0.0228 0.9088 0.9031 0.0438
0.4652 0.6683 -0.0186 0.9312 0.9084 0.0518
organic solvent composition = 69.5 mol%
A-22
0.2254 0.0590 -0.0051 0.8115 0.2300 -0.0143
0.4727 0.0991 0.0099 0.8228 0.2435 -0.0244
0.6005 0.1296 0.0115 0.8606 0.2676 -0.0137
0.6726 0.1545 0.0064 0.8863 0.2763 0.0178
0.7294 0.1994 -0.0342 0.9054 0.2835 0.0493
0.7858 0.2157 -0.0160
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
Table A3-5: Ion exchange equilibria for Au(CN) −2 (B) / SCN − (A ) in DMSO-water
mixtures at 303 K.
xB yB δyBa xB yB δyBa
organic solvent composition = 5.91 mol%
0.0063 0.0764 -0.0016 0.1159 0.6547 -0.0071
0.0135 0.1523 -0.0036 0.1403 0.7208 -0.0113
0.0231 0.2275 0.0039 0.1706 0.7846 -0.0111
0.0337 0.3018 0.0061 0.2139 0.8451 -0.0028
0.0465 0.3752 0.0091 0.2600 0.9020 0.0000
0.0599 0.4475 0.0038 0.3544 0.9504 0.0099
0.0773 0.5184 0.0047 0.4210 0.9950 0.0003
0.0950 0.5881 -0.0026
organic solvent composition = 14.3 mol%
0.0125 0.0771 0.0043 0.2004 0.7027 0.0065
0.0237 0.1534 -0.0060 0.2426 0.7619 0.0127
0.0390 0.2284 -0.0018 0.2924 0.8171 0.0164
0.0537 0.3023 -0.0075 0.3351 0.8691 0.0081
0.0710 0.3676 -0.0023 0.4010 0.9158 0.0043
0.0918 0.4385 0.0012 0.4738 0.9569 -0.0053
0.1100 0.5080 -0.0087 0.5240 0.9941 -0.0238
0.1374 0.5754 -0.0010 0.6260 1.0000 -0.0186
0.1689 0.6403 0.0064
organic solvent composition = 27.4 mol%
0.0273 0.0760 -0.0012 0.4330 0.7318 0.0019
0.0578 0.1495 0.0008 0.4766 0.7727 0.0060
A-23
0.0879 0.2208 -0.0030 0.5250 0.8098 0.0128
0.1280 0.2889 0.0098 0.5426 0.8455 0.0014
0.1634 0.3542 0.0091 0.5892 0.8776 0.0061
0.1952 0.4171 0.0003 0.6081 0.9082 -0.0017
0.2252 0.4776 -0.0118 0.6476 0.9357 -0.0018
0.2681 0.5347 -0.0059 0.6963 0.9594 -0.0017
0.3052 0.5890 -0.0086 0.7697 0.9774 0.0000
0.3470 0.6400 -0.0058 0.8154 0.9918 -0.0047
0.3911 0.6875 -0.0009 0.8497 0.9979 -0.0066
organic solvent composition = 69.5 mol%
0.1139 0.0675 -0.0070 0.5706 0.3724 -0.0086
0.2370 0.1262 0.0044 0.5951 0.4035 -0.0190
0.3310 0.1776 0.0112 0.6105 0.4329 -0.0357
0.4158 0.2226 0.0240 0.6731 0.4581 0.0036
0.4429 0.2637 0.0002 0.7006 0.4811 0.0104
0.4899 0.3029 -0.0045 0.7255 0.5022 0.0179
0.5266 0.3394 -0.0127
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
Table A3-6: Ion exchange equilibria for Au(CN) −2 (B) / Cl − (A ) in NMP-water mixtures
at 303 K.
xB yB δyBa xB yB δyBa
organic solvent composition = 4.46 mol%
0.0013 0.0768 -0.0043 0.0883 0.6714 -0.0535
0.0043 0.1538 0.0047 0.1163 0.7401 -0.0659
0.0089 0.2304 0.0038 0.1572 0.8049 -0.0688
0.0164 0.3065 0.0136 0.2218 0.8654 -0.0620
0.0277 0.3817 0.0271 0.3666 0.9143 -0.0249
0.0415 0.4558 0.0268 0.5820 0.9468 0.0031
0.0563 0.5287 0.0089 0.8092 0.9616 0.0212
0.0706 0.6006 -0.0241 0.9237 0.9676 0.0263
organic solvent composition = 62.7 mol%
A-24
0.7455 0.0196 0.0003 0.9355 0.0499 -0.0016
0.8816 0.0287 -0.0057 0.9384 0.0547 -0.0025
0.9157 0.0352 -0.0021 0.9428 0.0592 -0.0015
0.9321 0.0449 -0.0007 0.9459 0.0634 -0.0011
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
Table A3-7: Ion exchange equilibria for Au(CN) −2 (B) / SCN − (A ) in NMP-water
mixtures at 303 K.
xB yB δyBa xB yB δyBa
organic solvent composition = 4.46 mol%
0.0193 0.0762 0.0057 0.3487 0.6571 0.0247
0.0346 0.1512 -0.0138 0.4016 0.7036 0.0233
0.0592 0.2243 -0.0083 0.4506 0.7463 0.0172
0.0860 0.2953 -0.0068 0.5007 0.7851 0.0116
0.1186 0.3638 -0.0003 0.5158 0.8227 -0.0175
0.1546 0.4295 0.0041 0.5777 0.8555 -0.0148
0.2046 0.4913 0.0241 0.6297 0.8843 -0.0171
0.2318 0.5510 0.0011 0.6632 0.9102 -0.0275
0.2852 0.6065 0.0107
organic solvent composition = 62.7 mol%
0.4615 0.0418 0.0045 0.7904 0.1686 -0.0254
0.5655 0.0756 -0.0107 0.8045 0.1838 -0.0336
0.6897 0.0997 0.0020 0.8429 0.1960 -0.0127
0.7551 0.1187 0.0127 0.8646 0.2066 0.0004
0.7705 0.1351 0.0024 0.9013 0.2322 0.0285
0.7782 0.1524 -0.0138
a
δy B = y Bcalc − y Bexp . The y Bcalc values are those calculated in the 3-parameter regression.
A-25
A. APPENDIX 4
A-28
APPENDIX 4
(mol%) (kJ/mol)
Water 0 224.3 0.000
Acetone-water 5.73 55.71 3.509
13.9 5.573 9.308
26.7 0.642 14.75
68.6 0.020 23.55
DMSO-water 5.91 81.36 2.555
14.3 21.58 6.898
27.4 2.977 10.89
69.3 0.040 21.75
NMP-water 4.46 23.76 5.655
62.7 0.003 28.27
Table A4-2: Thermodynamic data for the Au(CN) 2− / SCN − binary system in mixed
solvents at 303 K.
Solvent Composition KAB Δ t Gnet
o
(mol%)
Water 0 14.94 0.000
Acetone-water 5.73 10.5 0.888
13.9 2.932 4.102
26.7 0.772 7.463
68.6 0.157 11.48
DMSO-water 5.91 16.05 -0.181
14.3 9.284 1.198
27.4 3.592 3.590
A-26
69.3 0.417 9.018
NMP-water 4.46 4.133 3.237
62.7 0.013 17.75
A-27