Alternativmaterialien zu Andalusit in Feuerbetonen

Andalusite based refractories are generally used for applications where a superior high-temperature volume stability, high thermal shock resistance and corrosion resistance against alkali rich slags and atmospheres is necessary. Depending on the exact requirements of the respective application, high-temperature volume stability, for example, may be more important than corrosion resistance and the material must be adapted accordingly.

The role of andalusite in andalusite based castables depends on the grain size, because the reaction is strongly kinetically hampered. Due to the small surface to volume ratio, the andalusite coarse grain decomposes at higher firing temperatures than the small andalusite grains in the fine matrix. Therefore, the major temperature interval of the andalusite transformation in RR, obtained from the results of XRD is between 1400 and 1500 °C, whereas the andalusite decomposition in RE already starts at 1200 °C and significantly at 1300 °C. The influence of the dwell time is also represented by the temperature difference for the andalusite decomposition from the firing without dwell time and the firing for 8 hours, respectively. Without dwell time, the andalusite decomposition in RR starts in the test-piece fired to 1400 °C, whereas it already starts in the test-pieces fired to 1300 °C for 8 hours.

The superior high-temperature volume stability of andalusite based refractories is claimed to be the result of skeletal mullite crystals with interstitial amorphous silica, sometimes designated as “felted mullite” aggregates originated from slowly converting andalusite. The microstructure of the fired andalusite based reference castables (between 1000 and 1500 °C) showed no interaction between the andalusite coarse grain and the matrix that could be the due for the high-temperature volume stability. The volume expansion of 4-5 % that is related to the andalusite decomposition has a positive influence on the volume stability. The volume expansion has a counteracting effect to the shrinkage due to ceramic sintering. The slow decomposition of andalusite is particularly favourable in a refractory material that is exposed to a thermal gradient. In the first millimetres of the hot side of the lining, the andalusite transforms, while it is still stable on the cold side. The andalusite based castables demonstrated high creep resistance during the first 25 h, while the model castables containing andalusite, kyanite and luting sand exhibited lower creep resistance. Only the model castable containing SiC exhibits superior volume stability. A superior high-temperature volume stability of andalusite based refractories could also be achieved using the modelled concrete with kyanite.

In the matrix of the andalusite based castables no mullite was found after thermal treatment. The mullite and interstitial amorphous silica that was formed within the andalusite decomposition has formed a phase of anorthitic composition with the existing CaO from the matrix, which was probably liquid at peak temperature, but anorthite crystals formed on cooling. The melt formation leads to a plastic behaviour at application temperature of a refractory castable and therefore prevents the spread of cracks induced by thermal shock.

The high degree of melting of the reference concretes also becomes clear when the results of the wedge splitting test are used. Significant proportions of melting phase must already be present from 1250 °C, as neither the notch tensile strength nor the specific fracture energy hardly show any strength. In the context of this work, however, it can be shown that the strengths for the model castables also collapse at 1250 °C but are better than for the reference castables. The three investigated andalusite based castables have a good corrosion resistance against alkali rich slags and atmospheres.

Anorthite could facilitate melt phase formation in high alumina low cement castable. Kyanite on the other hand as explained before produces mullite during its transformation. A combination of kyanite and anorthite in the model castable were incorporated. The aim here was to examine their interactions with other matrix components and to observe the resulting phase formations and to determine if this combination could serve as a viable substitute for andalusite in refractory castables, providing valuable insights into its potential suitability.

In principle, it can therefore be said that the current opinion on andalusite-containing castables must be questioned. In the reference castables analysed, it could not be proven that the formation of mullite during the decomposition of the andalusite causes a solidification of the structure, which results in the much-described refractory stability. In general, it should be noted here that this statement is based on the results of RuL and Creep. It must be taken into account here that in both cases the loss in height of the test specimens is compensated for by the increase in volume caused by the decomposition of the andalusite and kyanite, and ultimately an apparently high refractoriness is measured. Andalusite therefore appears to act primarily as a shrinkage compensator, which acts in the refractory lining to inhibit the formation of cracks on the hot side, although a high melt phase content is already demonstrably present. The melt phase on the hot side, in turn, seals the surface by significantly reducing the open porosity as in a glaze, which, unlike glazes, is in the liquid phase at operating temperature and thus greatly hinders the infiltration of volatile corrosive media. This may be the reason for the high corrosion resistance that is commonly attributed to andalusite castables. Despite the verified high melt phase content, andalusite castables are described as extremely resistant to thermal shock, although the thermomechanical strength (wedge splitting test) is no longer present from around 1250 °C. However, the melt phase in turn opens up the possibility of reacting ductile to thermal stresses, which means that thermomechanically induced cracks can no longer be initiated and consequently no spalling occurs.

An alternative must therefore be found for a target-oriented substitution of andalusite that is capable of expansion on the one hand and forms a dedicated amount of smelt phase on the other. The development of the model castables was aimed at this with a useful result, but the composition must be further adapted in all cases. In particular, additions with luting sand appear to be a highly interesting solution here, as the kaolinite forms melt on the one hand and the deep quartz expands strongly during the transformation to cristobalite on the other. Unfortunately, the post-expansion only sets in at around 1400 °C, which is actually too late to inhibit shrinkage induced cracks. The targeted use of small quantities of kyanite as an additional component could be helpful here.