A high purity pig iron was melt by high frequency induction furnace. The C and Si contents were adjusted by adding carburizer at 1693 K and Fe-75%Si alloy at 1713 K, respectively. The melt for spheroidal graphite cast iron sample was spherodization-treated by adding Mg/RE/Ca/Si―spherodizer and Fe-45%Si- 8%Mg alloy. After inoculation with Fe-50%Si alloy the melts were poured into the CO₂ molds into Y-blocks as indicated in Figure 1. Their matrixes were modified by three kinds of heat treatment. These heat treatment methods were listed below in Figure 2.

The microstructures of these specimens were observed with magnification of 100 X. Graphite and matrix structures, such as mean diameter, counts, area fraction, nodularity of graphite nodules and austempered matrix structures were analyzed by using the image analysis software. The fracture surfaces of tensile and impact specimens were observed by a FE-SEM.

The hardness of the test pieces was measured with Brinell hardness (10/3000/30) and the matrix hardness (HV0.5) was measured by Micro-Vickers hardness tester with values averaged from ten measurements.

Tensile testing was performed with tensile test piece with 8 mm in diameter and 28 mm in gage length using an Instron-type universal testing machine with 8.33 × 10⁻³ mm/s in crosshead speed at room temperature. Ultimate tensile strength (UTS) and elongation were determined from the results of stress-strain curves and 0.2% proof stress was estimated by the offset method of ones.

The impact testing was carried out by using an instrumented Charpy testing machine. The test was performed in the temperature range between 173 K and 423 K. Figure 3 shows the configurations of the un-notched specimens and the 5 types of notched specimens used in this study. The notch with 8 mm, 2 mm, 1 mm and 45deg.-Vshaped in root radius (R5, R1, R0.5 and R0.25, respectively) was obtained by a milling machine, and the notch with 0.15 in root radius was obtained by a wet-type cutting machine with a blade 0.3 mm thickness. The notch depth of all specimens is 2.0 mm constant so that the area of section at the notch bottom is the same in all the notched specimens. According to the Nishida’s equation [11] , stress concentration factors (α) of these specimens are 1.0 (un-notched), 1.5 (R5), 2.3 (R1), 3.0 (R0.5), 3.9 (R0.25) and 4.8 (R0.15).

Table 1 and Figure 4 show the chemical compositions measured by emission spectro photometer using chilled sample and microstructures of specimens. The microstructure of the Austempered specimens (ADI) is upper bainite with

feather-like appearance.

The mechanical properties of these specimens were shown in Table 2. The sample graphite nodularity is 90%. The tensile strength shows 1066 MPa, and the elongation shows 9.0% in ADI specimen.

Figure 5 shows typical impact load-displacement curves of un-notched (α = 1.0) specimens in each ductile region. The features of the curves exhibit the characteristics of each matrix, FDI with lower load and larger displacement, PDI with higher load and smaller displacement and ADI with higher load and larger displacement, which is a similar tendency to that in tensile testing.

Figure 6 shows the ductile/brittle transition curves in Charpy impact value of ADI samples. This figure indicates that every specimen exhibits a clear transition curve, which is composed with upper shelf, transition region and lower shelf. They also indicate that addition of an external notch results in a considerable decreasing of impact value in ductile region and that the affection caused by increasing stress concentration factor (α) from 1.0 (un-notched) to 3.0.

Figure 7 shows the relation between impact transition temperature and stress

concentration factor (α) of ADI specimen obtained from transition curve based on the energy transition method. The Charpy impact transition temperature is obtained from the temperature to be a half value of absorbed energy in the impact transition curve. The transition temperature increases with increasing stress concentration factor of external notch for every specimen. In fact, increasing stress concentration factor makes the stress concentration at notch root strongly increases [12] . This increment in transition temperature caused by the increase of α from 1.0 to 4.8 is about 40 K [13] .

Figure 8 shows the relation between Charpy impact value and stress concentration factor (α) at the upper shelf (ductile region). This figure indicates that the increase of α makes the impact value in ductile region decrease considerably up to α = 3, then is less concerned for higher values.

In order to evaluate the influence of notches degrees to Charpy impact values,

we defined the external notch factor by the following equation [14] :

βn = Impact value of α = 1.0 specimen/impact value of notched specimen. (1)

Figure 9 shows the relation between external notch factor (βn) and α. This figure indicates that βn increases resolutely with increase in α. The figure shows that the increasing rate of βn becomes smaller in the range of α larger than 3.0. These results suggest that additional notch with α larger than some 5 may cause only a little influence to the impact value. Since the hardest samples present the larger external notch factors, it implies that sample with a harder matrix is easier to be affected by external notches.

The specimens of ADI specimen tested with a universal testing machine, and the affections of external notch to impact energy and the initiation and propagation behavior of crack were investigated.

Figure 10 shows the microscopic observation of side view of polished ADI α=1.0 and α= 3.0 impact specimens which observed several loading point during the bending test. From these results, the main macro crack initiates at the maximum load point through micro cracks starting from the graphite edges in the stage prior to the maximum load point [15] [16] . Figure 11 shows the relation between crack length and load-displacement curves of ADI α = 1.0 and α = 3.0 specimens in bending test. The crack propagation process is strongly observed in the bending load-displacement curves. It is observed in the case of ADI specimen when notched but it is not observed when without notch process. From this result, absorbed energy (En) in impact test could separate the crack initiation energy (Ei) and the crack propagation energy (Ep) for various matrix specimens [17] . Figure 12 shows the relation between the impact characteristics of these various matrix tested at room temperature. In the case of FDI sample [15] , absorbed energy consists of a high crack initiation energy and high propagation energy.

However, for PDI [16] and ADI sample, the absorbed energy is almost constituted of crack initiation energy, because crack propagation energy is negligible. In this way, if we want to obtain higher toughness samples, the crack initiation energy need to be improved.

Figure 13 shows the SEM observation photos of fracture surfaces of FDI [15] , PDI [16] and ADI α = 1.0 specimens in brittle, transition and ductile region of impact transition curves. In the upper shelf region, the fracture surfaces beneath the external notch root is of ductile which consists of the dimples nucleated by

spheroidal graphite, and increasing of α results in no significant difference in the fracture surface appearance. In the transition region, the fracture surface is a mixture of the dimples and cleavages, and the increase of α suggests an increasing of cleavage part.