82 acid and sulphuric acid •
2. Treatment 1 plus 6.0 hours at 800°C 3* Treatment plus 4.0 hours at 850°C
4. Treatment 1. plus 2.5 hours at 1000°C 5* Treatment 1. plus 4.0 hours at 1150°C
/* denotes apparently more intensive characteristic lines/ 7* * 7 + 6 * y + 0 * / \
e +{
k +o
) a*The corresponding x-ray diffraction data are presented in Table 4.2.2., together with respective ASTM data for comparison. Even under the altered firing programme indicated above, the broadening of characteristic lines among several aluminas could not be eliminated. This made the prospects for improving the accuracy and specifically identifying some polymorphs unsuccessful, particularly delta, delta + theta and theta + kappaaluminas which, in turn, would have been used as standards. The main difficulty lies in the large number of
simultaneous phases exhibited during firing, when crystallization of some polymorphs, apparently, either accelerates or retards other consecutive transformations. It is likely, however, that these trans formations follow characteristic sequences, which, . in turn, are time and temperature dependent.
4*2.3. X-ray analysis of products of reaction between liquid -pure
aluminium and aluminium alloys and vitreous silica.
4.2.3.1. Lower temperature reaction products - the diffraction data of representative specimens at various temperatures and times are given in Table 4*2.3., together with aluminium and silicon (standards) data and alumina polymorphs (standards) data for comparison. The analysis has shown that apart from silicon and aluminium, both always present in all specimens, the product layer resulting from reaction at temperatures within 760 to 860°C consisted of a mixture of alumina polymorphs, markedly of 0 and a-aluminas. It is possible, however, at these temperatures, other polymorphs such as gamma, delta, theta or a combination of these were involved as well, but due to considerable
overlapping of characteristic lines, these aluminas could not be clearly identified. In this regard, it should be added that line broadening was further aggravated due to the presence of residual aluminium (not totally eliminated by acid leaching, Table 3.5.) and/or silicon.
Despite the difficulty of identifying independent polymorphs of alumina, the d-spacings worked out show the tendency for the theta-alumina to transform into alpha-alumina with time of reaction. It is likely that other polymorphs, possibly involved but not identified, also will have been transformed to alpha via the theta-phase, during reaction. This will be discussed later.
4.2.3.2. Higher temperature reaction products - the x-ray diffraction analysis of specimens reacted at two specific temperatures, i.e. at 1110 and 1265°C and corresponding times, as indicated in Table 4.2.3., has shown, apart from silicon and aluminium, theta and alpha-aluminas as the only phases present in the reaction product layer. This is evident from d-spacings worked out from characteristic lines and presented in Table 4.2.3. A duplex structure, consisting of both aluminas, thus is to he: expected in the structure and this will be further demonstrated under view of both optical and electron micrographs. Similar results were obtained, when silica reacted with aluminium-manganese and aluminium-iron alloys with different compositions of manganese and iron at different times and temperatures. This is indicated in Tables 4.2.4.1 and 4.2.4.2. for both elements, respectively.
4.2.3.3. Intermediate temperature reaction products - aluminium, silicon and the alumina polymorphs theta, theta+kappa and alpha were identified as the products of the reaction carried out at 920 and 980°C. The d-spacings worked out from corresponding lines are presented in Table 4.2.3. It is possible, however, that similar to the lower temperature reactions, other polymorphs were involved as not all lines were eliminated.
The characteristic x - ray diffraction analysis of reaction systems has not indicated any silicates present in the product layer whatsoever. Thermodynamic calculations to confirm this evidence will be
introduced later, in Chapter 5.
4.3. Microstructural examination of the reaction product layer 4.3.1. The pure aluminium - silica reaction system
4.3.1.1. Lower temperature reaction - the reaction products
which developed when the silica rod reacted with liquid aluminium at temperatures betwwen 760 and 860° C were examined microscopically. The microstructures which developed in this temperature range did not differ significantly. A typical macrostructural change in the specimens with time at temperature is presented in Fig.4.3.1.1., where a sequence of continuous and symmetric growth of the product layer with increasing time at 860°C has been selected from the series of experiments with silica rods 5.0 mm in diameter. Spalling of the unreacted silica at first and then of the product layer did occur during cooling as a result of large difference in coefficients of thermal expansion between the silica and the product layer as a whole.
In view of the observed similarities among the reacting specimens, therefore, features of various specimens reacted at different temperatures and times are now described in order to identify:
a) The mechanism of the progressive coversion of silica into
product phases.
b) The correlation of this mechanism with a kinetic model
deduced to explain the reaction between vitreous silica and liquid aluminium.
(All specimens, here described, are identified in Tables 4.1.2.1. to 4.1.6.)
The beginning of the conversion of silica, when it first came into contact with liquid aluminium, in the low temperature range, is schematically idealized as follows:
product phases
Si Op
/iTfTMTTx
✓
tTTTT
k xtt^TTT
ix /rHTT\/<TTTT
k _____The reaction started in localized areas on the surface of the rod with the nucleation and growth of spherical segments, approx imately of same size, Fig. 4.3.1.2. It was observed in all specimens examined from the initial periods of reaction that some parts of the interface silica/aluminium did not react at first. With increasing temperature this effect became less evident. These observations suggest that the surface film of aluminium oxide on the melt is not normally drawn down into the liquid when the rod was immersed into it. However, the possibility that some portions of surface film have incidentally adhered to the rod at the moment of its immersion and, thus, acted as a barrier for reaction, cannot be excluded. Fig. 4.3.1.3 shows that later removal of the melt surface film of aluminium oxide adhered to the rod allowed reaction to start immediately and to progress normally
producing the feature shown in this micrograph.
As the reaction continues, the segments grow laterally,
Fig. 4.3.1.4, and impinge on each other to form a continuous reaction layer, Fig.4.3.1.5:
/
product phases S i0 2AL
The segments grow also in the radial sense, i.e. normal to the receding silica surface, whilst new segments continue to form. This growth did not progress noticeably until a continuous layer had been formed on the periphery of the rod.
The process of formation of segments and their lateral
growth is promoted by temperature, as previously mentioned. The growth of a continuous product layer is very fast at the higher temperatures in the range 760-860°C, occurring in only a few minutes, Fig.4.3.1.6. At this stage of reaction, the alumina formed is not the stable a-phase, but one of the metastable aluminas, probably ©-alumina. With increas
ing time and, thus, with increasing thickness of the product layer and
also favoured by the presence of a liquid phase throughout this layer,
polymorphic transformation of aluminas did occur. This will be introd
uced later in this section.
At intermediate stages of reaction, columnar cells of alumina are developed associated with the receding surface of silica, Fig.4.3.1.7. The texture of these columnar cells was beyond the re solution of the optical microscope. Therefore, they are shown in the scanning electron micrographs of different specimens presented in
Figs.4.3.1.8, a-c. It is clear from this evidence that the
columnar alumina growth is of a porous texture.
Some cracking occurred within the product layer during react ion, Fig.4.3.1.9. Large grains of recrystallizing product layer,
shrinkage and porosity, independently or collectively, appear to be responsible for the crack itself. By cracking, at advanced stages of reaction, Fig.4.3.1^10, the interface of unreacted silica/product layer loses its original symmetry and thus aluminium must flow to the vicinity to the affected area for completion to be achieved.
Further examination of the interface region between the unreacted silica and the product layer (interaction front) indicates that: