Complex formation, a competitive process

of the electrostatic eld should extend also outside the rst sphere. The qualitative con-
clusion from this model is that in a reaction, the di
erences in the properties between
the bulk water and the water in the rst and second solvation spheres will be larger the
larger the di
erences in electrostatic potential between reactants and products.
From a microscopic point of view we can describe Reaction (III.1) as a reaction
where\ordered" water in the rst and second coordination spheres is transferred to the
less \ordered" state in the bulk solvent. This process results in an increase in entropy for
the system, in the same way as when a gas is expanded from a small volume to a larger
one, or when solid ice is transformed to liquid water.
We would then expect the values of rSm between a set of di
erent metal ions with a
given ligand to increase with increasing charge and decreasing ionic radius of the central
ion. The entropy changes should also decrease between consecutive complex formation
reactions, because of the decrease in charge and solvation and the increase in size, when
more ligands are added to the central ion.
For ligands that contain di
erent number of donor atoms we expect the entropy change
to increase with the number of coordinated donor atoms. The data in Table III.2 indicates
that the experimental ndings conrm these expectations.
The data in the table indicates that complexes between hard donors and hard accep-
tors have large positive entropy contributions, while the enthalpy term usually is fairly
small. On the other hand, in complexes formed between soft donors and soft acceptors
the enthalpy term gives a major contribution to the stability of the complexes, while the
entropy contribution is smaller. Enthalpies of reaction involving the proton, e.g. proto-
nation of bases are usually fairly large but always less than the enthalpy of protonation
of hydroxide.
Information of the type discussed above is useful to estimate the temperature depen-
dence of a reaction when no experimental enthalpy of reaction is known. More detailed
discussion of the thermodynamics of complex formation reactions is given by Ahrland
67AHR, 68AHR]. The conclusions drawn above relate to aqueous solutions non-aqueous
systems behave in a di
erent way, as discussed also by Ahrland 79AHR, 82AHR].
III.5. Complex formation, a competitive process
Most experimental investigations of complex formation reactions have been made on bi-
nary systems, and information on ternary and higher systems is scarce. The one exception
is for ligands which are polyprotic acids, HnL, either organic or inorganic. For these lig-
ands, all the di
erent species Ln; , HL(n;1), H2L(n;2) , . . . are potential ligands, and in
any careful experimental study the investigators have to decide if complexes of the type
MHpLq are formed, or not. One also has to ascertain if a coordinated water molecule
can be deprotonated with the formation of ternary hydroxide complexes, M(OH)pLq , as
shown in Figures III.8 and III.9. The occurrence of ternary complexes in nature is proba-
bly more common than indicated by published equilibrium data. Experimentalists often
try to avoid them by selecting the experimental conditions in such a way that their for-
mation is minimised. For the user of equilibrium data it is essential to keep this in mind,
89
 Chemical Background for the Modelling of Reactions in Aqueous Systems

Table III.2: Thermodynamic data for complex formation reactions. The rst group of data
illustrates how the thermodynamic quantities depend on the charge of the positive ion.
The second and third groups show how the thermodynamic quantities vary for the stepwise
reactions, and for ligands containing a di
erent number of donor atoms, respectively. The
fourth group demonstrates how the thermodynamic quantities vary with the hardness of
the ligand for a typical soft acceptor.

Reaction r Gm rHm T rSm

kJ  mol;1 kJ  mol;1 kJ  mol;1

Charge dependence:
H+ + F; *) HF(aq) ;16.57 12.26 28.8
Be + F *
2+ ;
) BeF+ ;28.79 ;1.7 27.1
Al3+ + F; *) AlF2+ ;35.02 4.60 39.6
Stepwise reactions:
Tl3+ + Cl; *) TlCl2+ ;42.76 ;25.27 17.5
TlCl2+ + Cl; *) TlCl+2 ;32.97 ;16.95 16.0
TlCl+2 + Cl; *
) TlCl3(aq) ;19.37 ;4.52 14.9
TlCl3(aq) + Cl; *) TlCl;4 ;15.86 ;0.71 15.2
Chelate formation:
Cu2+ + ac; ) * Cu ac+ ;10.71 4.35 15.1
Cu + mal ; *
2+ 2 ) Cu mal(aq) ;32.17 11.92 44.1
Cu + edta *
2+ 4 ;
) Cu edta2; ;105.5 ;34.10 71.4
Hardness/softness:
Hg2+ + F; *) HgF+ ;5.82 3.8 9.6
Hg + Cl *
2+ ;
) HgCl+ ;38.49 ;23.0 15.5
Hg2+ + Br; *) HgBr+ ;51.46 ;42.68 8.8
Hg + I *
2+ ;
) HgI+ ;73.22 ;75.31 -2.1

90
 Complex formation, a competitive process
because outside the concentration ranges used in the laboratory studies, di
erent species
may form. If quantitative data on them are not given in the literature it is necessary to
estimate both possible stoichiometries, and the magnitude of the equilibrium constants.
This procedure is facilitated when one of the components is present in large excess over
the others, a situation typical for the modelling of the speciation of trace metals in aquatic
systems.
III.5.1. The pH dependence of complex formation reactions
For ligands that are anions of weak acids (hydroxide, uoride, carbonate, phosphate,
silicate, and the anions of most organic acids), there is always a competition between
metal ions and hydrogen ions for the given ligand, and the complex formation reactions
will be strongly pH dependent. This is one of the reasons why pH is an important
\master variable" for the description of equilibria in aqueous solution. Phosphate is the
most important of the inorganic ligands in this group complexes containing H2PO;4 ,
HPO24; and PO34; are known. The ligands with the largest negative charge always form
the most stable complexes. On the other hand these are also the strongest bases, with
the strongest competition from H+. For this reason, H2PO;4 and HPO24; are important
ligands in most aquatic systems. Figure III.10 shows the distribution of dioxouranium(VI)
phosphate complexes as a function of pH at a total concentration of phosphate of 510;6 M
(0.5 ppm).
III.5.2. Polynuclear complex formation
The metal ions in polynuclear complexes may be of the same type, or di
erent. However,
few complexes containing di
erent metal ions are known, one example 86GRE/RIG] is
the formation of (UO2)2(MO2)(CO3)66;, where M is Np or Pu. In this case we can describe
the complexes as formed by an isomorphic substitution of UO2+ 6;
2 in (UO2)3 (CO3 )6 , with
2+ 2+
NpO2 or PuO2 , cf. Figure III.11. This is a process analogous to isomorphic substitu-
tions in the solid state.
The relative importance of polynuclear, as compared to mononuclear complexes in-
creases with increasing total concentration of metal ions. For example, the equilibrium
3 UO2(CO3)22; *
) (UO2)3(CO3)66;
is shifted towards the right with increasing total concentration of uranium.
At trace concentration levels most complexes are mononuclear.
The chemistry of the metal hydroxides is dominated by polynuclear complexes. An
excellent review of the eld has been given by Baes and Mesmer 76BAE/MES].
III.5.3. The stoichiometry of hydroxide complexes
Most hydrolysis equilibria are studied experimentally by measuring the hydrogen ion
concentration, cf. Eq. (III.2). In this type of experiment one measures the concentration of
91
 Chemical Background for the Modelling of Reactions in Aqueous Systems
Figure III.10: The stabilities of the phosphate complexes of dioxouranium(VI). The dia-
gram has been calculated at PO34;]total = 5  10;6 M, and at UO2+
2 ]total = 10 M. Values
;8

for the equilibrium constants are those given in 92GRE/FUG] in the standard state
(I = 0).

1:0 UO2 (OH);3

UO2+
2
Fraction 0:8
of uranium(VI)
species
0:6
UO2 OH+
0:4 UO2 HPO4 (aq) UO2 PO4
;

UO2 (OH)2 (aq)

0:2

0:0
4 5 6 7 8 9
pH

hydrogen ions released during the hydrolysis reaction. This quantity gives no information
as to whether the released protons come from coordinated water or hydroxide. In this
particular type of experiment the following equations are therefore equivalent from a
thermodynamic and modelling point of view. However, the constitution of the complexes
and their chemical properties will be very di
erent:
p Men+ + q H2O(l) *
) Mep(OH)np q
;q
+ q H+ (III.2)
p Men+ + (q ; 1)H2O(l) *
) MepO(OH)np ;q
q ;2 + q H
+
p Men+ + (q ; 2)H2O(l) *
) MepO2(OH)np ;q
q;4 + q H
+
...
p Men+ + 2q H2O(l) * ) MepOnp ;q
q=2 + q H
+

It is important to realise that the stoichiometric formulas above are equivalent in order
to avoid the same complex being included more than once in the mass balance equations
used in speciation codes.
In some cases there are additional data that can be used to deduce the chemically
correct stoichiometry. This is of great importance for chemical discussions of structure
and bonding. One example is o
ered by (UO2)3(OH)+5 , which is a predominant hy-
drolysis complex in the pH range 4-7. This complex has been studied by X{ray scat-
92
 Complex formation, a competitive process

Figure III.11: Isomorphic substitution of UO2+ 2+ 2+

2 with NpO2 or PuO2 in (UO2 )3(CO3 )6 .
6;
One or possibly two uranyl ions may be replaced by another actinide(VI)yl-ion with the
overall geometry retained 83ABE/FER]. The lled circles denote MO2+ 2 ions and the
open circles the carbonate oxygen atoms.

tering in solution 70ABE] and found to have a planar geometry and the stoichiometry
(UO2)3(O)(OH)+3, cf. Figure III.12.
General information on the coordination may sometimes be used to ascertain
whether a proposed stoichiometry is chemically reasonable or not. An example
from Grenthe and Lagerman 91GRE/LAG] is given in Chemical Thermodynamics
of Uranium 92GRE/FUG], where the chemical composition of the ternary complex
(UO2)11(CO3)6(OH)212; proposed in a aqueous-solution study was justied through a
chemical-structure reasoning.
The composition of the complexes formed in polynuclear metal hydroxide systems de-
pends on the pH and the total concentration of the metal ion. This is shown in the
distribution diagrams (Figure III.13) for the hydrolysis of lead(II) at two di
erent total
concentrations, 0.10 M and 1  10;5 M, respectively.
In many modelling situations it is interesting to describe the behaviour of an element
present in trace concentrations. Under these circumstances we expect hydroxide com-
plexes to occur mainly in mononuclear form. However, when studying hydrolysis in the
laboratory it is often an experimental necessity to use much higher concentrations, where
the hydrolysis is dominated by the formation of polynuclear species, cf. Figure III.13. The
experimental determination of equilibrium constants for mononuclear species may then be
impossible, or accompanied by large uncertainties. As a result the database for mononu-
clear hydroxide complexes is often rather poor, and it may be necessary to estimate the
constants, we will come back to this in Section III.6.3.
93
 Chemical Background for the Modelling of Reactions in Aqueous Systems

Figure III.12: The structures of (UO2)3(O)(OH)+3 and (UO2)2(OH)2+ 2 in the solid state
(Uranium atoms are represented by the darker and larger spheres, Oxygen atoms by the
small light-gray spheres, and Chlorine atoms are shown by the spheres of intermediate
shade and size). In the left gure, the central oxygen atom represents an oxide ion and
the bridging oxygen atoms are hydroxide groups. The remaining coordination positions
are occupied by water oxygens and chloride ions. The bridging oxygen atoms in the
gure to the right belong to hydroxide ions. The X-ray technique used does not allow the
determination of hydrogen atom positions 70ABE].

III.5.4. Competition between di erent metal ions for the same ligand
We can illustrate this process by the following simple exchange reaction between two
metal ions M and N and a ligand L:
ML + N *
) M + NL
with the equilibrium constant
K = N=M = ML]N] NL]M]

The relative amounts of the species ML and NL depend both on the equilibrium constants,
and the total concentrations of M and N. As a result a metal ion like Ca2+ that is present
in fairly large amounts in many ground and surface water systems may have an important
in uence on the speciation of a trace element, even if the complex formation constants of
this are much larger than for Ca2+ .

94
 Complex formation, a competitive process

Figure III.13: Distribution diagrams for lead(II) hydrolysis at two total metal concen-
trations: 0.10 M (upper diagram) and 1  10;5 M (lower diagram). The source for the
hydrolysis equilibrium constants is Ref. 76BAE/MES].

2+ PbO(cr)
1 0 Pb
: Pb6 (OH)4+
8
Fraction 0 8
of lead(II)
:

species
06
Pb4 (OH)4+
:
4
04
:

02
:

Pb3 (OH)2+
4 Pb(OH);3
00
:
5 7 9 11 13
pH
2+ Pb(OH);3
1 0 Pb
:

Fraction 0 8 PbOH+
of lead(II)
:
Pb(OH)2 (aq)
species
06
:

04
:

02
:

00
:
5 7 9 11 13
pH

95
 Chemical Background for the Modelling of Reactions in Aqueous Systems

Example 5:
Calcium(II) and Cerium(III) are both present in ground and surface water systems. Both
elements form complexes with carbonate (a component in these systems), but of very
di
erent strength: log10 1(Ca2+ )  3:2, and log10 1(Ce3+ )  7:6, log10 2(Ce3+)  12:2,
for the formation of CaCO3 (aq), CeCO+3 and Ce(CO3);2 , respectively. Figure III.14 shows
the relative amounts of CaCO3(aq) and CeCO+3 as a function of log10CO23;] when the
total concentrations of Ca(II) and Ce(III) are 1  10;5 M and 1  10;8 M respectively.
III.6. Theoretical framework for the estimation of equilibrium constants
The varying chemical conditions in nature may result in the formation of totally new
species for which no laboratory data are available. When modelling chemical processes
it is essential to keep this possibility in mind. Thermodynamic databases may thus be
incomplete and it is necessary for the modeller to be able to judge if and where such
lack of data may occur. This requires knowledge both of the general chemical principles,
which often are qualitative, and methods to estimate the stoichiometry of complexes and
their corresponding thermodynamic data.
III.6.1. On the magnitude of equilibrium constants and the ratios between equilibrium
constants for successive complex formation reactions
In systems where more than one ligand is bonded to the metal ion we can dene equilib-
rium constants for the consecutive reactions, which are of the type:
M+L *
) ML 1 = K1
ML + L *
) ML2 K2
... ...
MLn;1 + L * ) MLn Kn
And we have n = K1K2 : : : Kn .
With few exceptions these equilibrium constants decrease in the order K1 > K2 >
: : : > Kn . The exceptions are found mainly among the mononuclear hydrolysis reactions
(where the quality of the data are not always as good as would be desired), for systems
containing -bonding ligands such as cyanide, and for some reactions involving Hg(II) and
Tl(III). This empirical nding is useful when estimating equilibrium constants for higher
complexes in situations where there is only experimental information for the formation
of the rst complex. We can go a step further and explore the expected ratios between
equilibrium constants for consecutive complex formation reactions in chemical systems
where we have a well dened coordination geometry, assuming that these are based on
statistical considerations alone.

96
 Theoretical framework for the estimation of equilibrium constants

Figure III.14: Carbonate complexes of calcium(II) and cerium(III). The pH-dependence of

the free carbonate concentration is shown in the upper plot under the same conditions as
the lower graphs (note that CO23;] changes by a factor > 106 in this pH-range). All three
diagrams have been calculated at CO23;]total = 0:01 M, and the two lower diagrams with
Ca(II)]total = 10;5 M, and Ce(III)]total = 10;8 M for both. Note that calcite (CaCO3 )
precipitates at pH > 9:2 under these conditions. Setting Ca(II)]total = 0 does not a
ect
the lower diagram because the carbonate/Ca(II) ratio is large. Equilibrium constants (at
I = 0 and 25 C) are from 78KRA/DEC, 90WOO, 93MOR/HER].

;2 H2 CO3 (aq) HCO;3 CO23;

;3
log10 Conc. OH;
;4
;5 H+
;6
;7
5 6 7 8 9 10 11 pH

2+
1 0 Ca
:

Fraction
of calcium(II) 0 8
:

species 06:
CaCO3(aq)
04
: CaCO3 (s)
02
CaHCO+3
:

00
:
5 6 7 8 9 10 11 pH

10 Ce(CO3 );2
CeCO+3
:

Fraction 0 8 Ce3+
of cerium(III) :

species 06:

04
:

02
: CeHCO2+
3 Ce(OH)3 (aq)
00
:
5 6 7 8 9 10 11 pH

97
 Chemical Background for the Modelling of Reactions in Aqueous Systems

Example 6:
Let us consider the case discussed by Beck and Nagyp!al 90BEC/NAG]: the coordination
of a bidentate ligand to a metal ion with octahedral coordination geometry (an octahedron
has six corners and twelve edges). The rst ligand molecule can bind along any of the
edges, and it can dissociate in only one way. The second ligand molecule can then bind
along any of the ve remaining edges, where no corner is occupied by the rst ligand, and
the dissociation can occur in two di
erent ways. If the complex formation is determined
by statistical factors alone we expect the ratio between the rst two constants to be:
K1=K2 = 121  52 = 4:8.
The third ligand can only be bonded in one way, but dissociate in three. This statement
must be modied for the case where the two ligands are bonded in trans-position to one
another. The two free positions in the coordination polyhedron are then opposite one
another and not along a common edge. This distance is so large that the ligand cannot
span it. One out of the ve possibilities of binding the second ligand gives rise to this trans-
bonding. Hence, the expected ratio between K2 and K3 based on statistical considerations
is: K2=K3 = 25  13  45 = 9:375.
III.6.2. Estimation of equilibrium constants for ternary complexes
It is also possible to calculate the statistical value of the equilibrium constant for a mixed
complex from the equilibrium constants of the corresponding binary complexes, e.g.,
MA2 + MB2 * ) 2 MAB
If we select ligands that give complexes with the same number of donor atoms we have
the following probabilities for the various reactions (adapted from 90BEC/NAG]):
MB2
1/2
>

MB
1/2
> Z
1/2ZZ
~
M MAB
Z 1/2
>
1/2ZZ
~
MA
Z
1/2ZZ
~
MA2
i.e., the probabilities of formation of MA2 and MB2 are both proportional to 12  21 = 41 ,
while the probability of formation of MAB is proportional to 2  14 = 12 . The statistical
value for the equilibrium constant is then ( 12 )2=( 14 )2 = 4, or MAB
2 = 4  MA2  MB2 .
Experimental observations indicate that the ratio between successive equilibrium con-
stants often deviates signicantly from the statistical value { the reason is that the chem-
ical interactions in general are speci
c .
98
 Theoretical framework for the estimation of equilibrium constants
In most cases the behaviour of the binary systems are experimentally known. In this
situation one can modify the statistical procedure for estimating the equilibrium constant
of the mixed complexes. The procedure for a three component system is outlined in the
following scheme (adapted from 90BEC/NAG]):
M
; @
K1B B] ; @ K1A A]
K1A A]+K1B B] ; @ K A A]+K B B]
; @ 1 1
; @
; @

; R
@
MB MA
KMB-A A] ; @ ; @
KMB-A A]+K2 B] ;; @ ; @ K A A]+ -
KMA B B]
B
@ ; @ 2 -
KMA B B]
; @ ; @
; K2A A] @ ; K2B B] @
; K2 A]+KMA-B B] @
A -
; KMB A A]+K2B B] @

; R
@ 
; R
@
MB2 MAB MA2
where the quantities KMA-B and KMB-A are equilibrium constants which describe the
anities of MA for B and MB for A, respectively. If MA and MB bind A and B equally
we have KMA-B = K2A and KMB-A = K2B. If B binds with the same strength to MA and
MB we have KMA-B = K2B and for A KqMB-A = K2A . If we do not know the real bond
strength we can use KMA-B = KMB-A = K2A  K2B as a reasonable estimate. K1 and K2
are the stepwise equilibrium constants in the binary systems.
Beck and Nagyp!al 90BEC/NAG] give an extensive discussion, and we will just present
their results for reactions of the type depicted in the previous scheme. The statistical
value for the equilibrium constant for the reaction
MA2 + MB2 *
) 2 MAB
is equal to
v
u v
uKA
K1A  u
t K2A + K1A  u
B B
t 2B
Kstat = 2 + K1B K2 K1 K2 (III.3)
q q
From this expression we nd that Kstat = 4 only holds if K1A= K2A = K1B= K2B. In all
other cases the value of Kstat is larger.
Figure III.15 shows the value of log10 Kstat if the formationqof the ternary complexes is
governed by statistical factors only and KMA-B = KMB-A = K2A  K2B. This curve could
be used as a guideline for estimating such equilibrium constants, when the behaviour of
the binary systems are known. The ternary system Hg(II)-Cl;-OH;, cf. Figure III.9, is
99
 Chemical Background for the Modelling of Reactions in Aqueous Systems
Figure III.15: The statistically expectedqequilibrium constant Kstat for MA2 + MB2 *
)
2MAB as a function of log10((K1 =K1 )  K2 =K2 ). Adapted from 90BEC/NAG].
A B B A

2.5

2.0
stat
K

1.5
10
log

1.0

0.5
0.0 0.5 1.0 1.5 2.0 2.5

log
10
( KA1 / KB1 ) + log10 √KB2 / KA2

an example where the magnitude of the equilibrium constant for the ternary complex
HgOHCl(aq) deviates strongly from the statistical value.
We can use data for the U(VI)-oxalate and carbonate systems to estimate the equilib-
rium constant for the formation of the complex UO2(ox)CO3]2;.
Example 7:
Estimate the equilibrium constants for the formation of UO2(oxalate)(CO3)2; from the
following equilibrium constants for the binary systems (data at I = 0, carbonate and
oxalate constants are taken from 92GRE/FUG] and 69HAV], respectively):

log10 1(UO2CO3) = 9:7  log10 2(UO2(CO3)22;) = 16:9 

log10 K2(UO2(CO3)22;) = 7:2
log10 1(UO2(oxalate)) = 7:2  log10 2(UO2(oxalate)22;) = 11:9 
log10 K2(UO2(oxalate)22;) = 4:7.

If we assume that the stepwise complexes in both the binary systems and in the
ternary complex are formed statistically we have
100
 Theoretical framework for the estimation of equilibrium constants

q
 stat(UO 2;
2(oxalate)CO3 ) = 2 2(UO2(oxalate)22;)2(UO2(CO3)22;)
log10  (UO2(oxalate)CO23;) = 14:7
If we use the experimental ratios of the stepwise constants we will have a di
erent
result because these are quite di
erent in the two systems. For the carbonate system
we have
log10 1(UO2CO3) ; 12 log10 K2(UO2(CO3)22;) = 6:1
and in the oxalate system
log10 1(UO2(oxalate)) ; 1 log10 K2(UO2(oxalate)22;) = 4:9
2
By using the curve in Figure III.15 and the experimental quantity
v
1(UO2CO3) u
log10  (UO (oxalate))
u
t K2(UO2(oxalate)2;22;)
1 2 K2 (UO2(CO3)2 )
we obtain log10 K stat = 1:25 and log10  (UO2(oxalate)CO23;) = 15:0.
This example may give an idea of the uncertainty in this type of estimations !
A \phase diagram", cf. Figure III.16, where the components, a metal ion and two
ligands are placed in the corners of an equilateral triangle, may be used to describe
di
erent ternary systems:
if only binary interactions occur in the system, all compositions fall along two of
the sides of the triangle. The properties of the three-component system are then
the sum of those of the constituent binary systems.
if the metal cation, Me, forms ternary complexes with the two ligands, X and Y,
there will be composition points inside the ternary diagram, and the system is
not additive. In the absence of experimental information, it is necessary to use
the general theories from coordination chemistry to estimate the stoichiometry and
equilibrium constants of the possible complexes that may form in these systems.
In systems where ternary complexes are formed, the properties of the ternary system
will be equal to the sum of the Me{X and Me{Y systems only when the concentration
of the ternary species is negligible in comparison with the binary complexes.
In general one nds that the system is additive when the bonding strength between the
metal cation and the two ligands di
ers considerably. The formation of predominating
amounts of ternary complexes (with the exception of H+ and OH; containing species)
101
 Chemical Background for the Modelling of Reactions in Aqueous Systems
Figure III.16: Compositions of complexes formed in the Al3+-Cl;-dimethylformamide
system. Adapted from 90BEC/NAG].
Al3+

AlCl (dmfa) AlCl (dmfa)+

3 2 2
AlCl − AlCl(dmfa)3+
4 4
Al(dmfa)3+
6
AlCl (dmfa)+ AlCl(dmfa)+
2 4 5
dmfa
Cl− (dimethyl-
formamide)

requires a close balance between ligand geometry, bonding strength and concentrations,
and this situation is likely to be rare.
In many modelling situations one has a rather large number of components, and there
is in general no information on the possible ternary, or higher interactions among all these
components. The discussion above indicates that to describe ternary interactions, only
a small number of the components need to be taken into account { the choice of these
components must be based on general chemical knowledge.
Example 8:
Estimate the possible stoichiometric compositions for a ternary system containing UO2+
2 ,
a bidentate ligand (oxalate, carbonate), and a strong binding monodentate ligand like
uoride or hydroxide. The coordination geometry around U(VI) is a pentagonal, or
hexagonal bipyramid. Figures III.17, III.18 and III.19 illustrate the geometry of some
binary and ternary complexes studied in the solid state. The uranium(VI) oxalate system
has been studied in solution by Havel 69HAV] who found that the equilibrium constant
for the reaction
UO2(oxalate)22; + oxalate2; ) * UO2(oxalate)43;
was only log10 K3 = 0:4, i.e., much smaller than those for the formation of the rst two
oxalate complexes (which are log10 K1 = 6:0 and log10 K2 = 4:7, all data at I = 1 M).
This nding ts nicely with the structural information (Figures III.17 to III.19) which
indicates that the third ligand can only use one of its carboxylate groups in bonding, with
102
 Theoretical framework for the estimation of equilibrium constants
Figure III.17: The coordination geometry of uranyl oxalate complexes. I.
UO2(oxalate)F33;. All ligands are located in a plane perpendicular to the linear O-U-
O axis, forming a pentagonal bipyramid.

a much lower equilibrium constant as a result. This is an example of steric interference.

There is not room for three oxalate groups bonded by both their carboxylate groups. The
ligand bite is at least 0.2 #
A smaller when the oxalate uses a carboxylate group that is
bonded \end-on" and this makes it possible to bind the third ligand.
Based on structural information alone, one might therefore expect the fol-
lowing ternary complexes to be formed: UO2(oxalate)F;, UO2(oxalate)F22;,
UO2(oxalate)F33;, UO2(oxalate)2F3;. F; can be replaced by hydroxide. For carbonate,
the same stoichiometries are possible. In the four component system U(VI)-carbonate-
oxalate- uoride, the ternary complex UO2(oxalate)CO23; might also be formed.

III.6.3. On the use of correlations for the prediction of equilibrium constants

The equilibrium constant is a characteristic of a reaction , nevertheless it is often possible
to correlate the values of log10 K for a series of reactions with some property(ies) of
the metal ion or the ligand. Correlations of this type are always based on a chemical
model/theory, and they are useful because they summarise the properties of a large group
of data in a concise way, at the same time as they provide a rationale for predicting the
values of equilibrium constants for which no experimental data are available. It must
be emphasised that the ecient use of correlations requires chemical information about
coordination geometry, donor-acceptor characteristics, information of the structure and
conformation freedom of the ligand, etc. It is within classes of reactions that share some
common characteristics that correlations are most precise. In the following section we
will give examples of some of the correlations used by coordination chemists.
103
 Chemical Background for the Modelling of Reactions in Aqueous Systems
Figure III.18: The coordination geometry of uranyl oxalate complexes. II.
UO2(oxalate)2H2O2;. All ligands are located in a plane perpendicular to the O-U-O
axis, forming a pentagonal bipyramid.

III.6.3.1. Correlations based on the size and charge of the metal ion
The theoretical idea behind these type of correlations is the ionic model of chemical
bonding, i.e., a model where the chemical forces between atoms are largely of electrostatic
type. We then expect that the bond energy, E , between a certain ligand and a series of
di
erent metal ions will be governed by the size and charge of the reactants as follows:
E / ZM ZL=dM;L
where ZM and ZL are the charges of metal ion and ligand, respectively, and dM;L is
the bond distance between them, cf. p.75. When comparing complexes between di
erent
metal ions and the same ligand we expect that: E / ZM =dM;L .
Correlations of this type should work best for chemical systems where the general chemi-
cal behaviour indicates a signicant electrostatic component in the chemical bonding, e.g.,
the bonding between hard acceptors and hard donors.
A theory based on this concept was developed by Kossiako
and Harker
38KOS/HAR] to describe the protolytic behaviour of inorganic acids of the type
MOp(OH)(2q p+q;z); . This theory was used and extended by Baes and Mesmer to de-
scribe the variations of the rst hydrolysis constants of metal-hydroxide complexes with
the \ion-potential", z=dM;O of the metal ion. Figure III.20 illustrates the results of this
type of correlations. It should be pointed out that the uncertainty in the experimental
104
 Theoretical framework for the estimation of equilibrium constants
Figure III.19: The coordination geometry of uranyl oxalate complexes. III.
UO2(oxalate)43;. The three oxalate ligands are placed in a plane approximately per-
pendicular to the O-U-O axis, forming a hexagonal bipyramid. The third ligand can only
use one of its carboxylate groups in bonding. This ligand is no longer planar as the other
two.

equilibrium constants may be large, because the predominant reactions in these systems
involve polynuclear species.
Fluoride complexes are also expected to show a similar behaviour and Figure III.21
shows the dependence of the equilibrium constant for the formation of MFn+ complexes
for several metal cations with z=dM;F .
Baes and Mesmer have made extensive use of \electrostatic" type of correlations in
their discussions of the chemistry of metal-ion hydrolysis and the solubility of hydrous
metal oxides.
For the reaction:
M(OH)z (cr) + (z ; 1) Mz+ *
) z MOH(z;1)+ (III.4)
we have
(z;1)+ ]z
K (III:4) = MOH z
Mz+]z;1 = K1 Ks
0
and
log10 K (III:4) = z log10 K1 + log10 Ks
0 (III.5)
105
 Chemical Background for the Modelling of Reactions in Aqueous Systems

Figure III.20: The linear dependence of log10 K1 for reaction: Mn+ +OH; * ) MOH(n;1)+,
on the ratio of the charge to the M-O distance for four groups of cations. The dashed lines
show the correlations given by Baes and Mesmer 76BAE/MES], cf. their Figure 18.4.
Zr4+
14 Tl3+ Hf4+
Bi3+ U4+
Fe3+
12 Ti3+ V3+
Pd2+ Ga3+
Th4+
10 Sn2+ Hg2+ In3+ Cr3+
Sc3+
Be2+ Al3+
8
1
K

Pb2+ Y3+
10

6 Cu2+ Ce3+ Lu3+

log

Gd3+
Zn2+ La 3+
Fe2+2+ Co2+
4 Cd2+ Ni
Mn2+
Ag+
2 Mg2+
Tl+ Sr2+
Li+ Ca2+
0 Na+ Ba2+
K+

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

z/d
M-O

where K1 and Ks
0 correspond to the following two equilibria respectively:

Mn+ + OH; ) * MOH(n;1)+ K1
*
M(OH)z (cr) ) M + z OHn + ;
Ks
0
The value of log10(K1z Ks
0) is ;5:6 for most metal cations, as indicated by the plot of
1 (log K + 5:6) vs. log K shown in Figure III.22.
z 10 s
0 10 1
Correlations of this type are very useful to predict log10 K1 (which often are dicult to
measure experimentally) from log10 Ks
0 (which often can be determined experimentally
with high precision).
For further details about these and other correlations for this important group of com-
plexes, the reader is referred to Chapter 18 in the classic book by Baes and Mesmer: \The
Hydrolysis of Cations" 76BAE/MES].
There are other correlations between the ionic properties of metal cations, e.g., be-
tween the ionic radius and the electronic conguration, the charge, the electronegativity,
and the ionisation potential. Hence, one also expects correlations between equilibrium
constants and these quantities, an example is the following correlation (van Panthaleon
106
 Theoretical framework for the estimation of equilibrium constants
Figure III.21: The dependence of log10 K1 for the formation of metal uoride com-
plexes, MFn+, with the ratio of the charge to the M-F distance. Ionic radii are
from 69SHA/PRE], and stability constants are from 76SMI/MAR, 82MAR/SMI,
89SMI/MAR] (circles: data at I = 0, triangles: data at I > 0).
10 Zr4+
U4+
Th4+ Hf4+
8
Sc3+
Al3+
Fe3+
6 Ga3+
Be2+ Y3+ Cr3+
1

In3+ V3+
K

4 Ln3+
Sn2+
10
log

Mg2+
2 Hg2+ Bi3+
Ca2+ Cu2+
Zn 2+
Ag+ Pb2+
Tl+ Ni2+ 2+
0 Li+ Cd 2+ Fe2+ Co
Na+ Sr2+ Mn2+
Ba2+
K+
-2
0.5 1.0 1.5 2.0
z/d
M-F

van Eck 53PAN]):

log10 K1 = p (In ; q)
where K1 is the equilibrium constant for the formation of the complex ML, In is the n:th
ionisation potential of the cation Mn+, and p and q are constants which depend on the
ligand and the experimental conditions. An example of this kind of correlation is given in
Figure III.23 for some complexes of glycine. The same equilibrium constants are plotted
as a function of the electronegativity of the metal cation in Figure III.24 as an example
of another type of correlation between equilibrium constants and ionic properties.
III.6.3.2. Ligand
eld theory and Irving and Williams series
The equilibrium constants for the 3{d transition elements from Mn(II) to Zn(II) vary in
a regular way as indicated in Figure III.25.
The observed variations can qualitatively be explained by using the ligand-eld the-
ory. According to this the observed variations are a result of the di
erent d-electron
congurations of the central ion. This theory is useful for describing the variation of
thermodynamic quantities for transition elements with partially lled d-orbitals. More
details may be found in standard textbooks 84GRE/EAR, p.1096 86COT/WIL, p.467].
The general increase in stability from Mn(II) to Zn(II) may be looked upon as a result
107
 Chemical Background for the Modelling of Reactions in Aqueous Systems
Figure III.22: The constancy of the equilibrium constant for Reaction (III.4):
M(OH)z (cr) + (z ; 1) Mz+ * ) z MOH(z;1)+ . The line corresponds to K (III:4) = K1z Ks
0 =
10;5:6, cf. Eq. (III.5). Data from 76BAE/MES] at I = 0 and 25 C.
Ca2+
0
Oxides
Oxy-hydroxides
-2 Ag+
Mg2+ Hydroxides
Mn2+
+ 5.6)

-4
Fe 2+ Pb2+
Cd2+
2+ La3+
Co
-6 2+ Ce3+Y3+
Ni2+ Zn
s,0

Gd3+3+
Lu
(log K

Cu2+ Be2+ Cr3+

-8
10

Sc3+
Al3+ 2+
Hg2+
-10 Sn
1/z

In3+ Ga3+ Bi3+

Th4+
-12 Fe3+
Hf4+Zr4+
U4+
-14 Tl3+
1 3 5 7 9 11 13 15

log K
10 1

of the decreasing ionic radius from Mn2+ to Zn2+, while the other variations are a result
of di
erences in the d-electron congurations among the metal ions. The variations ob-
served depend on the donor atoms in the ligands and the coordination geometry. Hence
interpolations may only be made for a series of complexes of the same stoichiometry and
coordination geometry. Correlations may be made not only for stability constants, but
also for other thermodynamic quantities 53IRV/WIL, 95JOH/NEL].

III.6.4. Correlations based on properties of the ligand

The most important of these correlations is between the log10 K values for the formation
of metal complexes between the same metal and di
erent ligands and the log10 K for
the protonation of the same ligands. The theoretical concept behind this is that both
the binding of a proton and a metal ion to the ligand are chemically similar processes,
a Lewis acid/base reaction involving the donation of an electron pair from a donor (the
ligand) to an acceptor (the proton, or a metal ion). This and other correlations assume
that one characteristic property is responsible for the variation of the log10 K -values.
This is seldom the case. However, the correlations are most likely to work for complexes
containing only few ligands that do not contain too many donor atoms, so that steric
interference is less important. We will now give examples of correlations of this type and
108