1 250'C were observed using a high-voltage X-ray fluoroscope and a high-speed video camera. Thefrequency of bubble formation at the nozzle exit, the mean bubble diameter and the mean bubble rising velocity were obtained for a wide range of injected argon gas f]ow rate. Empirical correlations of these quantities were proposed and comparedwith previously published experimental data and empirica] coTrelations, It has been commonly believed that the frequency of bubble formation at a nozzle placed in a molten metal bath dependssolely on the gas flow rate and the outer diameter of the nozzle when the gas flow rate is relatively high and the wettability betweenthe nozzle material and the molten metal is bad. The present experimental results, however, revealed that the frequency of bubble formation has a close relationship not with the outer diameter but with the inner diameter of the nozzle for a higher gas flow rate. FuTthermore, the bubble frequency depended on the gas flow rate and the physical properties of gas and molten metal, The critical gas flow rate for the initiation of small bubbles due to disintegration of large bubbles coming from the nozzle exit was approximately 60cm'/s under the present experimental conditions. The bubble behavior near the bath surface also was made clear.

Mechanism of surface crack formation of steel induced by residual copper (Cu) is investigated using a new technique of Greeble test. Two kinds of experiments were carried out, to clarify the effect of temperature and to understand the behavior of crack growth. Crack is caused by liquid Cu, which precipitates at steel-scale interface during oxidation. However, no crack formed at higher temperature. Micro analysis indicates that it is due to the formation of liquid scale above eutectic temperature of FeO-2FeO¥SiO2. Liquid Cu-precipitates are trapped in the liquid scale area, and they cannot penetrate into austenite grain boundaries. The fact that silicon addition reduces the crack formation also supports this mechanism. Deformation test with various strains reveals that there exist two stages in the behavior of crack growth. At the first stage, crack grows deeper, because liquid Cu penetrates into the boundary. The crack stops growing along the depth direction and opens its width in the second stage, because of the lack of liquid Cu. That means the amount of Cuprecipitates decides the crack depth.

l/R1=ec/as and 1/R2=sinc/x ........,. (7) Thus, the gas/slag interface is described by the system: ac/as = -sin c/x+ ((pG-Ps) Schematic drawing of the interface between two phases. interface has an opposite sign compared to that of gas/slag interface as shown in Fig. 3 o(1=(1+fr4J/W*)/2 and oc2=1-0(1 ...... (13) where J is the suD* erficial velocity of gas phase (ratio gas flow rate to cross section area). Wallis proposed a diagram as shown in Fig. 4 (1-a)J/W* 33 (1993)

In the smelting reduction process of