3. The activity of a component in the slag may be
defined in terms of Temkin ionic fractions, i,e. ^ 0 = Nm2+ x N 2-M 0
A model was formulated for linear chains only, in which Masson considered polycondensatioii reactions where S i O ^ could dimerise and then react with higher members of the silicate species. For a binary basic oxide-silica melt, the following expression was obtaineds-
2
which gives the activity of MO as a function of composition if is known (Figure 17) • The model gave some agreement with experimentally obtained activities of MO and enabled values of
to be designated to some binary MO-SiO^ systems.
An expression was also derived to define the ionic fraction of a silicate species;
N =n
r
1<n"1f
11 + ®M0 j k h + 1
I
K11 “ ^0 ^ 1*140 1
"
^ 0 ^2.9
2 (n + 1)- where N is the ionic fraction of Si 0 ' '
n n 3n + 1
and a^Q is given by equation 2,8.
Figure 18 shows the variation of the calculated ionic fraction of linear silicate anions of chain length n = 1 to n = 6 with mole fraction of SiOg for FeO - SiO^ melts at 1257-1307°C.
Models hased on polymer theory that pertain to "branched 92
chain configurations have "been developed • The three fundamental assumptions used previously were retained. The activity of MO was obtained from the following expressions
2.10 K11
the resultant variation of a^Q with x„.SiOg being presented
graphically in Figure 19#
The branched chain model gave good agreement with
experimental data for the systems SnO-SiO^# FeO-SiO^, PbO-SiO^ and MnO-SiOg with recent work further confirming agreement with
93
the MnO-SiOg system . Figure 20 shows a comparison between experimental data for the CaO-SiO^ system at 1600°C with the theoretical prediction for K equal to 0.0016. Table 1* records
for various systems.
For branched chains the ionic fraction of a silicate species was given as:-
TCT 3nl n = (2n + 1) In! r 'I 1 n - 1 r 1 + 3aM0 ► L 1 2n+1 3aM0 2.11
SiOj^ anion is the most abundant species at all compositions up to * S1O2 = 0.£. For the CaO-SiC) system the monomer is the2
sole species present below equal to 0*3, but in the FeO-SiO^ system longer chains exist at silica compositions below the orthosilicate composition*
Masson has suggested that the magnitude of the
equilibrium constants etc. provide a quantitative measure of polymerisation in binary silicate melts and that the
tendency towards polymerisation is determined by the nature of the cation. The average chain length of a melt would, therefore, be governed by the'difference in magnitude between cation-
silicate and cation-oxygen ion attractions. Consequently, the extent of cation-oxygen attraction influences the degree of ionic bonding in silicon-oxygen linkages so that changes in bond strength can be reflected in standard free energies of formation of the various silicates from their constituent oxides. Masson showed that for the reaction
MCL. + M_liq 3 2 7 Sio0 = 2M 2 4 liqSiO. ...
that,
BT In K. . = 2 AG° , 11 ortho - A G°pyro
This equation indicates that a tendency for polymerisation always exists in silicate melts because K.^ may be small but can never be zero, unless the rigfct hand side of the equation takes a value of minus infinity. Application of the equation is
limited as available thermodynamic data pertains to simple molecules thus restricting determination of •
Criticisms of the Masson Models include doubts about the use of Temkin ionic fractions for a situation where different
size silicate anions are present. Also, the linear chain model predicts zero MO activity at the metasilicate composition which
go
is not observed experimentally • Since the development of the models, experimental evidence has indicated the existence of silicate rings in binary silicate melts and no provision has
73
been made for this situation .
Application of polymer theory to ternary silicate melts incurs difficulties due to the competitive interactions between the anions and various cations. An attempt has been made to develop a model for a ternary silicate melt where a common
9U
cation exists, e.g. MO-MF^-SiO^ • This work retained the
assumptions of the previous models and although a new treatment was given to this type of melt, it served mainly to demonstrate
the complexities of dealing with ternary systems.
2.3*3 Constitution of Silicate Melts
(a) Practical Attempts to Determine Silicate Melt Structures At present, no technique has been found to directly observe and quantify the amounts of different silicate species in a melt. Workers have depended upon measurements of physical properties to provide structural interpretation of melts but the presence of immiscibility gaps in some systems can limit the
information obtainable.
Density measurements of CaO-PeO-SiO^ melts have confirmed 73
predictions that greater polymerisation occurs in iron
silicate melts compared with calcium silicate melts and relates 2+
to the preferred association of Ca cations with silicate
2+ 2— 95
anions thus releasing Pe cations to associate with 0 anions , 96 a phenomenon which increases the PeO activity in such melts ,
High temperature x-ray diffraction techniques, Raman spectroscopy and Mossbauer spectroscopy have all been used to examine silicate melts but no quantitative assessments of ionic distributions have been achieved Mossbauer spectroscopy examinations of basic electric arc slags sampled
2+
during the oxygen blow has shown that Pe cations can be present in a silicate phase or a 1mixed oxide1 phase, the proportion taken up in the latter phase increasing with
101
basicity , This tends to concur with Gaskell’s comments on the preferred association of cations with silicate or oxygen
96
anions ,
The distribution of silicate ions in a melt has been
80
inferred from studies of phosphate glasses dissolved in water , The dissolution technique cannot be applied to silicates as
hydrolysis readily occurs in aqueous media producing indefinite polymers related to silica gel
A technique has been developed \diere the blocking of O" sites of the parent silicate structure of a mineral by
trimethylsilyl (IMS) groups, (CH^)^Si- , has been used to retain some of the original structure. Gas Liquid Chromato graphy (GLC) and Mass Spectrometry (MS) were used for
respective separation and identification of the silicate derivatives
The trimethylsilylation of silicates is not always amenable to quantitative interpretation and attempts to improve this method by the formation of methyl and ethyl
102
silicates has been unsuccessful , The trimethylsilylation of silicates has been developed by a number of workers and will be discussed in more detail,
(b) The Trimeihylsilylation of Silicates
Trimethylsilylation is a pretreatment commonly applied to 10li organic materials to improve their suitability for GLC analysis Inorganic anions other than silicates have been trimethylsilylated
105,
106Lentz trimethylsilylated various silicate minerals using a reaction mixture of crushed ice, propan-2-ol, hexamethyldisiloxane
103
and concentrated hydrochloric acid , The process was completed by use of an ion exchange resin, Amberlyst 15, A proposed reaction scheme commenced with leaching of the metal cation portions of the mineral to produce metal chloride and silicic acid:-
MoSi0. + I4HGL = H, SiO. + 2MGL
trimethylchlorosilane (TMCS) and trimethylsilanolj
[(CH^)^Si ] 2 0 + HC1 = (GH^)^SiCl + (CH ) SiOH
Either of the above products could be capable of reacting with silicic acid but Lentz considered
trimethylchlorosilane and the trimethysilylation reaction:- HjSiO, + U(CHJjS-CCl = [(CH ) Si]^ SiO^ + UHC1
where [(CH.) Sil SiO. is Hie IMS derivative of ihe 3 3 1 + 4
SiO. ^ anion,
4
Unfortunately, side reactions can occur and have been attributed to polymerisation and depolymerisation reactions taking place during trimethylsilylation and by hydrolysis of the
107
derivative itself , Success of trimethylsilylation is related 108
to both the type of metal cations and silicate anions 109
present . Leaching of the metal cation is undoubtedly 110 111
influenced by the cation type present 9 but the subject of acid attack on silicate minerals is still not fully understood 112• The presence of certain cations such as iron was
112
believed to aid leaching but increasing leachability by virtue of decreasing cation-oxygen bond strengih for a given
111
silicate group does not strictly confirm this . Complex silicate structures may render cations inaccessible to acids and so retard leaching For chain structures this appears
111