The solubility of carbon in Fe-Mn-C-P melts containing high manganese Was determined at 1400℃.The solubility data can be expressed as: N_C＝0.1961＋0．1183N_(Mn)- 0.8525 Np The interaction coefficients between Mn and C,P and C in this melt were obtained as follows: ε_C ~（Mn）＝－1．67，ρ_C ~(Mn)＝0．29，ε_C ~P＝11．00，ρ_C ~P＝10．82 The activity coefficients of Mn,P and Solvent Fe in melts containing high Mn were estimated.

The high precision and good repeatability were carried out by the Padday cone method to determine the surface tension of the titaniferous slag foaming intensely towards high temperature. An ion-cluster structure model was deduced. The surface tension seems to be correlated with the parameters which are calculated from the model of titaniferous slag. The results may be used both to discuss the difference among the components and to predict the surface tension of the system.

A modified form has been proposed for Margules expansion, Redlich-Kister expansion and Legendre expansion to represent thermodynamic properties of binary systems with stronger convergency and independent characters for the coefficients.

Influence of dissolved carbon content on recrystallization γ-fibre texture in ultra-low carbon deep drawing steel sheet with high purity was investigated and the relationships between dissolved carbons and recrystallization γ-fibre texture and {111} <112> and {111} < 110 > textures were analysed by 3-dimensional orientation distribution function.

DSC result indicated that the thermal stability of amorphous (Ni0.53,Fe0.33,Co0.14)73B17Si8Nb2 alloy could be increased further with the pre-denucleation of the melts. Denucleation enables the thicker amorphous stripes to possess higher degree of thermal stability. The rapidly quenching of the denucleated melts is effective for preparing bulk amorphous alloy and improving its property of application.

The mathmatical and physical model has been adopt to compute the fluid flow distribution in case of local gas holdup in mould. The photography was used to show fluid field. The predicated flow patterns show reasonable agreement with exprimental observations using actual water model.

The effects of particle type and size of SiC, Al2O3 or SiO2 particle reinforced A1-4%Mg composites on the pushing behaviour during solidification were investigated. The intensity of pushing during solidification was determined by the mass ratio of equiaxed cell to particle. The theoretical predictions were in fair accordance with experimental results.

Kohonen′s self-organization neural network is presented to discriminate D88 structure of inter-metallic compounds. After the research for 65 compounds, the results indicated that the performance of the network is good, and therefore it might be referred as an effective assistant technique for analysis of structure of compounds.

The relation among the MgO-B2O3-SiO2 slag compositions and the efficiencies of extraction of B has been fitted and predicated by artificial neural network. The optimum composition corresponding to the highest efficiency of extraction of B was obtained using genetic algorithm. It is believed that the artificial neural network and genetic algorithm may provide a new and effective way for fitting and optimizing the process of extraction of B.