(50 solidify An alternative technique has been developed by Rolls-Royce

in which newly drawn fibres are pre-coated by passing them continuously

a bead of molten aluminium, sometimes containing small additions of antimony or bismuth to control the chemical reaction between the

bonding at the fibre-matrix interface is a compromise. There must be sufficient affinity between fibre and matrix to ensure at least some mechanical keying so that load transfer can occur. Too great an affinity between the components, as is the case with aluminium and silica, can

lead to thermal degradation. Chemical interaction of the aluminium and .silica at elevated temperatures produces severe deterioration in the

mechanical properties, probably as a result of volume changes occurring in the reaction zone leading to crack formation and thereby initiating premature failure.

throughout service life of the component, it is imperative to minimise chemical interaction between the fibre and the matrix both during fab­ rication and during service. Clearly important practical conclusions can be drawn from the present work. The existence of a reaction rate minima

liquid interaction during the fabrication process. In addition it would silica and the aluminium. Fabrication is by;;:hotpressing of suitably aligned coated fibres.

In all fibre refinforced systems, the achievement of adequate

If optimum mechanical properties are to be achieved and retained

appear that dilution of the aluminium by alloying elements such as mang-

V

,

anese, iron and silicon would certainly further reduce liquid-fibre reaction and might also improve the resistance to solid state chemical reaction between the fibre and the matrix. Unfortunately it would appear unlikely that there will be any practical opportunity to test the useful­ ness of these suggestions since commercial interest in aluminium-silica Kfibre composite materials as an engineering material has waned in recent

6. CONCLUSIONS AND SUGGESTIONS FOR FURTHER INVESTIGATIONS

The experimental technique developed to determine the rates

of the reaction between liquid aluminium and vitreous silica at_*

temperatures of practical importance was very satisfactory. The results show that aluminium as a strong deoxidant readily reacts with silica allowing it to be postulated that a similar reaction must occur during ^ deoxidation practice of steelmaking, where secondary reactions take

place between pre-existing silicate inclusions and dissolved aluminium. The mechanism for these reactions has been clearly identified at temperatures below 860°C. The reaction is almost entirely diffusion controlled and maintained by transport of aluminium and silicon across the porous matrix. The porosity of the matrix is derived from the volume changes occurring during reaction.

Above ~880°C the process is abruptly discontinued and further progress of reaction is conditioned by a mechanism not fully elucidated but the polymorphs of alumina and the formation of a spinel type phase contribute. It is only as high as ~1050°C that the rate of reaction is gradually recovered by fissure formation as a result of sintering of the alumina phase and a consequent variation in the mechanism of reaction. The process is probably controlled by the transport of

aluminium to the interaction front and limited by the saturated liquidus alloy present at the interface.

Additions of silicon up to 15 atomic percent to the liquid aluminium does not affect the rate of reaction whatsoever. Local equil-

ibrium of iron and manganese concentrations effectively reduces the solubility of silicon in the liquid metal phase, thereby reducing rates of counter-diffusion of silicon and aluminium and brings about a prop­ ortional reduction in the rate of reaction.

This wSrk has not been exhaustive in its treatment of the reaction between liquid aluminium and vitreous silica. For instance, .it would be of interest, if the intercellular alumina spacing along the

reaction product layer were measured for temperature and time conditions. With increasing temperature, the transport of silicon along the coarse matrix will become more important than its passage through the inter­ action front. A detailed study of the porosity and tortuosity factors would give means for solving exact equations in order to prescribe the flux of aluminium in porous cells of alumina. The mechanism in the

i

intermediate temperature range 860-1060 C must be evaluated conclusively. The composition gradient of silicon, gradually dissolved in the liquid phase, for temperature and time conditions, needs more study to elucidate more fullythe aluminium and silicon counter-diffusion fluc­ tuations in the reaction front. A more detailed interpretation of the behaviour of the interaction front at various times and temperatures could lead to a model describing the progress of reaction. This is notably absent from the literature at present.

The inhibiting effect of the increasing concentrations of both manganese or iron on aluminium may be attributed to the diminishing

thermo-dynamic driving force behind the reaction, when aluminium is no longer the pure reacting agent. Available information concerning the effect of elements such as manganese and iron on the thermo-dynamic

activity of liquid aluminium, when reacting with silica (and/or silicates), should be further investigated by more experimental work.