Oxide-dispersion-strengthened (ODS) ferritic steel is known to be candidates for advanced fission and fusion materials that require excellent radiation resistance and high-temperature capabilities (Odette et al., 2008). We focus on the development of 9CrODS ferritic steel with a composition of Fe-9Cr-0.13C-2W-0.2Ti-0.35Y2O3 (mass%) (Ohtsuka et al., 2004, 2005a, 2005b, 2006, 2007; Ukai et al., 1998, 2002a, 2002b, 2003, 2007, 2009, 2011; Yamamoto et al., 2010, 2011). This steel contains nano-size Y-Ti complex oxide particles. Its microstructures can be easily controlled by a reversible α/γ phase transformation with a remarkably high driving force of a few hundred MJ/m3. The 9CrODS ferritic steel is recognized as a composite-like material composed of a residual ferrite and tempered martensite. The residual ferrite is a metastable phase, and high-temperature strength of 9CrODS steel is significantly ascribed to the presence of the metastable type of the residual ferrite. In this article, formation process of the composite-like structure and resultant superior high-temperature strength of the 9CrODS steel are reviewed with incorporating current result of manufacturing in terms of newly developed thermo-mechanical processing.
2. Formation of composite-like material
The composite-like structure of 9CrODS steel is controlled by the α/γ phase transformation of the steel, which yields the residual ferrite embedded in the tempered martensite. The nano-sized oxide particles are concomitantly distributed among both phases. This section describes the process governing a formation of composite-like structure.
2.1. Manufacturing process
Pure elemental powders of iron (99.5 mass%, 45–100 μm), carbon (99.7 mass%, 5 μm), chromium (99.9 mass%, under 250 μm), tungsten (99.9 mass%, 4.5–7.5 μm), and titanium (99.7 mass%, under 150 μm) were mechanically alloyed (MAed) together with Y2O3 powder (99.9 mass%, 20 nm) for 48 h in an argon gas atmosphere using a planetary-type ball mill (Fritsch P-6). Pots with a volume of 250cm3 were rotated at a speed of 420 rpm; the pots contained the powders (35 g) and balls (total weight: 350 g); the weight ratio of the balls to the powders was set at 10:1. The standard chemical composition of MAed powders produced is Fe–9Cr–0.13C–2W–0.2Ti–0.35Y2O3 (mass%). The excess oxygen (Ex.O) was measured to be 0.09 mass%, which was determined by subtracting the amount of oxygen contained in Y2O3 from the total oxygen content; this amount of oxygen is inevitable for the formation of a Y–Ti complex oxide.
The MAed powders were consolidated by spark plasma sintering (SPS) or hot-isostatic-pressing (HIP) at 1,150°C for 1 h. The final heat treatment was conducted by normazing at 1,050°C for 1 h and tempered at 800°C for 1 h. Microstructure of the produced 9CrODS steel is shown in Fig.1. The structure is of the dual-phase, composed of a smooth surface phase and surrounding tempered martensite containing black dots type of chromium carbide (Ukai, 2011). This smooth surface phase belongs to a residual ferrite, of which formation process is mentioned in the next section.
2.2. Residual ferrite formation
The computed phase diagram of the Fe–9Cr–0.13C–2W–0.2Ti system without Y2O3 is shown in Fig. 2 (a) with respect to the carbon content (Yamamoto et al., 2011). For a nominal carbon content of 0.13 mass%, the single austenite γ-phase containing TiC carbide is stable at the normalizing temperature of 1050°C. The equilibrium γ/γ+δ phase boundary at this temperature corresponds to a carbon content of 0.08 mass%, beyond which δ-ferrite is not stable. Fig. 3 shows SEM micrographs of the MAed powders for specimens with 0.1 mass%, 0.35 mass% and 0.7 mass% Y2O3 and without Y2O3 that were heat-treated by normalizing at 1050°C and quenching in water (Yamamoto et al., 2010). The specimens without and with 0.1 mass% Y2O3 have the single martensite structure, which were obtained by inducing the transformation from the γ-phase by water quenching. This result is consistent with the computed phase diagram shown in Fig. 2(a). Nevertheless, the specimens with 0.35 mass% and 0.7 mass% Y2O3 exhibit a dual phase comprising martensite and smooth surface phases.
Digital image analyses showed that the area fraction of the smooth surface phase was approximately 0.2 for both specimens with 0.35 mass% and 0.7 mass% Y2O3. No obvious difference was observed between X-ray diffraction peaks of the martensite-phase and smooth surface phases at room temperature.However, high-temperature X-ray diffraction measurement at 950 ºC showed a considerable difference, as shown in Fig. 4; the specimen without Y2O3 had diffraction peaks corresponding only to the austenite γ-phase, whereas specimens with 0.35 mass% and 0.7 mass% Y2O3 had diffraction peaks corresponding to the
austenite γ-phase and the ferrite phase (Yamamoto et al., 2010). The austenite γ-phase transformed to the martensite-phase, but the ferrite phase remained unchanged by water quenching. These results prove that the smooth surface phase is composed of ferrite. Considering that the ferrite phase is formed only in the specimens containing 0.35 mass% and 0.7 mass% Y2O3 and that four types of ODS steels shown in Fig. 3 have an identical chemical composition except for Y2O3 content, it seems that the presence of a Y2O3 particle favors the retention of the ferrite phase at 950 ºC and 1050 ºC.
Dilatometric measurement was also carried out from the room temperature to 1100 ºC with a heating rate and a cooling rate of 5 ºC/min. Fig. 5 shows the results of the dilatometric measurement when Fe–9Cr–0.13C–2W–0.2Ti was heated without and with 0.35 mass% Y2O3 (Yamamoto et al., 2010). In the case of the specimen without Y2O3, the linear thermal expansion coefficients started decreasing from AC1 point of 850 ºC to AC3 point of 880°C because of the reverse transformation of the martensite-phase to the γ-phase, which was in reasonably good agreement with the computed phase diagram shown in Fig. 2(a). The addition of 0.35 mass% Y2O3 induced an increase of up to the AC3 point of 930 ºC. By comparing both the curves, we found that the specimen with 0.35 mass% Y2O3 exhibited a smaller amount of reduction in the linear thermal expansion during the reverse transformation of the martensite-phase to the γ-phase; this observation indicates that the entire martensite-phase was not transformed to the γ-phase. The untransformed ferrite phase corresponds to that observed by the X-ray diffraction peaks in the specimen with 0.35 mass% Y2O3 at 950 ºC. This untransformed ferrite phase was designated as the residual ferrite. More detail consideration is discussed from a viewpoint of difference in the thermal expansion coefficient in reference (Yamamoto et al., 2010)
2.3. Thermodynamic analyses
Alinger et al. carried out a small angle neutron scattering (SANS) experiment for a U14YWT specimen (Fe–14Cr–0.4Ti–3W–0.25Y2O3). The neutron scattering cross-section and scattering vector were measured for the MAed powder after hot-isostatic-pressing (HIP) or thermal annealing at 700 ºC, 850 ºC, 1000 ºC, and 1150 ºC. Their results indicate that the MAed powder annealed at 700 ºC shows the smallest radius and highest density for the Y–Ti complex oxide particles (Alinger et al., 2004). Assuming that the Y2O3 particles were decomposed during mechanical alloying, the subsequent annealing resulted in the formation and precipitation of Y–Ti complex oxide particles at elevated temperatures of 700 ºC or higher. Since the reverse transformation of the ferrite α-phase to the austenite γ-phase took place at the elevated temperature of over 850 ºC, which is higher than the precipitation temperature of the Y–Ti complex oxide particles, it is possible to consider that the partial retention of the residual ferrite can be attributed to the presence of the Y–Ti complex oxide particles in the 9CrODS ferritic steels; these particles should block the motion of the α/γ interface, thereby partly suppressing the reverse transformation of the α-phase to the γ-phase.
The chemical driving force (ΔG) for the reverse transformation of the α-phase to the γ-phase in the Fe–0.13C–2W–0.2Ti system without Y2O3 can be evaluated in terms of Gibbs free energy vs. carbon content curves at each temperature; these curves were derived by using the Thermo-Calc code and TCFE6 database. The pinning force (
M is the mobility of the interface. ΔG and
When the chemical driving force (ΔG) for the reverse transformation of the α-phase to the γ-phase is derived, we should consider a change of carbon content in the matrix due to carbide dissolution, because carbon content greatly affects the Gibbs energy of γ-phase. At AC1 point, M23C6 carbide dissolves into matrix and carbon concentration at carbide-matrix interface becomes relatively high. For example, carbon content of 0.2 mass % leads to ΔG higher than pinning force F by oxide particle at 1050 ºC as shown in Fig. 6 (Yamamoto et al., 2011).
This result suggests that α to γ reverse transformation proceeds from the carbide-matrix interface, where M23C6 carbide dissolves into matrix. This situation is schematically illustrated in Fig. 2(c). When temperature rises up beyond AC3 point, M23C6 carbide completely dissolves into matrix and thereby average carbon concentration approaches to the initial content of 0.13 mass %, where pinning force
2.4. Oxide particle dispersion
The size distribution of oxide particles within the residual ferrite and the tempered martensite matrix was separately evaluated. In this study, both ODS fully ferritic steel and ODS fully martensitic steel were produced to simulate the residual ferrite and martensite phases, respectively, where the microstructure of the fully ferrite is completely the same as residual ferrite, because both phases are not transformed to γ-phase at the temperature above AC3 point. Fig. 7 shows the dispersed oxide particles in full martensite and full ferrite ODS steels, observed by TEM (Ukai, 2011). It is obvious that a few nm size oxide particles are finely dispersed in the fully ferrite ODS steel, whereas their size is coarser in the
martensite ODS steel. The image analysis of the TEM micrographs is shown in Fig. 8; the mean size of oxide particle in full ferrite and full martensite ODS steels is 3 nm and 7 nm, respectively (Yamamoto et al., 2011). From Fig. 8, the oxide particles dispersed in the ODS fully martensitic steel, i.e. martensite matrix, seems to shift to a coarser size compared to the dispersion in ODS fully ferritic steel, i.e. residual ferrite. One of the main reasons for this behavior could be associated with α/γ reverse transformation. Yazawa et al. reported that the coherent to incoherent transition at precipitation interfaces sometimes induces coarsening of precipitates (Yazawa et al., 2004). Similar behavior is expected in the ODS martensite matrix. Our consideration is schematically shown in Fig.9. The oxide particles
are formed by a process of precipitation during hot-pressing at 1100°C after their decomposition was induced by mechanical alloying. This precipitation leads to partially coherency between oxide particles and the α-ferrite matrix, since precipitation takes place at temperatures below the AC1 point (Alinger et al., 2004). Subsequent change of the matrix phase from α-bcc crystal structure to γ-fcc by the reverse transformation could disturb the interfacial coherency; the oxide particles coarsen to minimize the interfacial energy during staying at γ-phase. Therefore, martensite phase transformed from γ-phase by cooling contains larger size of oxide particles, whist the residual ferrite contains finer oxide particles due to keeping coherency.
3. Creep deformation characterization
The creep deformation of the 9CrODS steel is unique and so different from that of the existed heat-resistant steels. The packet and block boundaries also play an important role for deformation. This section characterizes a deformation mechanism of the 9CrODS steel from a viewpoint of composite-like behavior.
3.1. Nano-hardness of residual ferrite
With the aim of evaluating the mechanical property of the residual ferrite, nanoindentation measurements were conducted using a Berkovich tip-type indentator at a loading of 3 mN. The trace of a Berkovich tip of sub-micron size can be placed within the interiors of the residual ferrite regions, while the conventional micro-Vickers diamond tips using 100 mN loads cover 7 x 7 micron-meters. Considering that the plastic zone is only two to three times larger than the contact area of the Berkovich tip, the mechanical property of the residual ferrite can be adequately evaluated by nanoindentaion measurements. Fig.10 shows the hardness change in the individual phases as a parameter of the tempering conditions (Ukai et al., 2009). The tempering parameters are shown in equation (3) in in terms of a Larson-Miller parameter (LMP; C = 20);
where T is temperature in kelvin and t is time in hours. The time at different temperature can be derived so as to satisfy the same LMP value.. The hardness decrease is significantly restricted in the residual ferrite with increasing tempering conditions compared with that of the martensite phase. The overall hardness measured by the micro-Vickers tester is also shown by the broken line; this covers both the residual ferrite and martensite, thereby representing the average hardness of both phases. The hardness Hv is correlated with the yield stress σy using the following relationship (Tabor, 1951):
By tempering at 800°C for 58 h, which is equivalent to tempering at 700°C for 10,000 h based on the LMP, the hardness can be converted to the yield stress at room temperature for the individual phases: 1360 MPa for the residual ferrite and 930 MPa for the martensite. One of the main reasons for increasing hardness at the residual ferrite is come from the dispersion strengthening of oxide particles that are finely distributed in the residual ferrite as mentioned in the previous section. It is worth noting, therefore, that 9CrODS steel behaves like composite material comprising of hard and soft elements.
3.2. Creep strain curve
Fig. 11 shows the typical creep curve of 9CrODS steel after normalizing and tempering (NT) for stress loadings of 155 MPa and 230 MPa at 700°C (Ukai et al., 2009). The optical microstructure of this specimen is shown in the left upper of Fig. 12 (Ukai et al., 2009). This steel is composed of the residual ferrite and the tempered martensite. The secondary creep is obviously decelerated with strain accumulation, and the specimen ruptured without an accelerated creep region at 155 MPa. A stepwise increase in the primary creep strain can be observed with stress increase up to 230 MPa. These behaviors are completely different from those of conventional heat-resistant steels. In conventional steels, the rupture occurs after the accelerated creep deformation, where coarsening of carbides, movement of the lath and block boundaries and a local recovery of dislocations become dominant (Abe, 2004, 2008).
Such dynamic recovery and structure coarsening should be suppressed by the dispersed nano-size oxide particles in 9CrODS steel. Their dominant fracture mode could be deformation near block and packet boundaries. The stress induced by this deformation can’t be easily relaxed by the dislocation creep, since nano-size oxide particles can block the dislocation movement. This is the reason why the creep strain rate gradually decreases with increasing strain, and work hardening dominantly occurs during creep deformation.
3.3. Creep rupture property
The optical microstructures of 9CrODS steels with and without the residual ferrite are shown in Fig. 12 for NT and furnace cooling (FC) conditions. The 9CrODS steel containing the residual ferrite has a standard chemical composition, whilst higher content of excess oxygen was added for the free residual ferrite one, where the titanium combines with oxygen rather than carbon, thus increase of matrix carbon content leads to complete α to γ transformation. The detail is mention in reference (Ukai & Ohtsuka, 2007). Based on the continuous cooling transformation (CCT) diagram, NT treatment induces the tempered martensite due to a rapid cooling rate of 7000°C /h, while FC treatment causes a diffusional transformation to the α-phase by slow cooling at a rate of 30°C /h.
Fig.13 shows their creep rupture strength (Ukai et al.,2009). The following two characteristic features were observed in creep rupture properties of 9CrODS steels. The first is that the creep rupture strength of NT with residual ferrite (TM+RF) increases stepwise from that of NT without residual ferrite (TM). This is attributed to the effects of the harder residual ferrite strengthening the 9Cr-ODS steels, therefore. The second feature is that the furnace cooling (FC) heat treatment for with or without residual ferrite induces stepwise increases in the creep rupture strength in comparison with the NT heat treatment. This result means that the α-phase has superior strength to that of martensite. Martensitic phase transformation produces the packet and block boundaries, where grain boundary deformation and fracture can be preferably originated. The strength improvement in the α-phase induced by FC could be, therefore, due to the suppression of the deformation at packet and block boundaries.
3.4. Creep strain rate vs. stress relation
where, G is the shear modulus (50.6 GPa at 700°C); M, the Taylor factor (3.0) (Stoller & Zinkle, 2000); b, the magnitude of the Burgers vector (2.48 × 10–10 m). The parameter ro is the inner cut-off radius of a dislocation core, which was assumed to be the magnitude of the Burgers vector. In addition, λ is the average face-to-face distance between particles on a slip plane; it is given in terms of the average particle radius rs and the average center-to-center distance ls between the particles :
The values of ls and rs are determined using the particle radius r:
where f is the volume fraction of the dispersed particles, which is estimated from the oxide particle density (4.84 × 103 Kg/m3). The D corresponds to harmonic mean of 2rs and λ. At the elevated temperature of 700°C, the interface between the particles and matrix could tend to slide. Under such condition, an attractive interaction between a dislocation and the particles can be applied, as proposed by Srolovitz (Srolovitz et al., 1983). The A and B values are expressed by following formula.
where ψ is the critical angle at which the dislocation detaches from the particles: φ = 46 for a screw dislocation and ψ = 19 for an edge dislocation. ν is the Poisson’s ratio (0.334). Substituting these dispersion parameters into equation (5), the threshold stress σ for the average of edge and screw dislocations is derived to be 265 MPa. Fig. 14 shows the strain rate vs. stress relation (Ukai et al., 2009). The deformation occurs at a stress far below the threshold stress of 265 MPa. The deformation is, therefore, induced at weaker regions such as grain boundaries rather than being intra-grain themselves, which is strengthened by the dispersed oxide particles. The strain rate vs. stress relation is almost parallel, and the slope corresponding to a stress exponent n is estimated to be ranging from 10 to 14. It is well known that the typical value of the stress exponent is significantly higher (around 40) for deformation that is dominated by particle-dislocation interaction (Arzt, 1991). Therefore, the accelerated deformation at a lower stress level could have arisen from the local deformation associated with packet and block boundaries. The cause of this stress increase at which creep strain initiates by furnace cooling can be explained by a disappearance of packet and block boundaries, as described above. The residual ferrite also enlarges the critical stress for initiating creep strain. Concerning this phenomenon, the creep constitutive equation of 9CrODS steel should be formulated by analogy with a fiber composite material (Sakasegawa et al., 2008). From the stress distribution between the hardened residual ferrite and the softened tempered martensite, the stress loading on the tempered martensite is derived by the following equation:
where σt and σm are the total stress loading and stress onto the individual tempered martensite regions, respectively. The parameter fr is the volume fraction of the residual ferrite grains, and D is the stress distribution coefficient expressed as a ratio of the shear modulus of residual ferrite μr to that of the tempered martensite μm:
Assuming that the shear modulus ratio is approximately expressed as the hardness ratio of both phases, D becomes equal to 1.46 from Fig.10. Thus, based on Equation (12), σm/σt is equal to 0.88 using fr = 0.3, as determined by microstructure observation. This estimation implies that the stress loading onto the softened tempered martensite decreases to 0.88, which leads to increase in the stress corresponding to the initiation of creep strain.
As a summary of the creep deformation of 9CrODS steel, Fig. 15 schematically illustrates a configuration of the hardened residual ferrite and softened tempered martensite. Under stress loading, softened tempered martensite is preferentially deformed. Dislocations piled up at the packet and block boundaries could induce the localized deformation near these boundaries, since the intra-grain deformation is hardly taken place due to finely dispersed nano-sized oxide particles. The stress accumulated at such boundaries could lead to fracture. The hardened residual ferrite provides not only sharing the stress loading onto the softened tempered martensite, but also geometrically restricting the deformation of the softened tempered martensite. It can be said that deformation and fracture of the 9CrODS steel behaves like composite material.
4. Thermo-mechanical processing
Modification of a microstructure is required for the tempered martensite in 9CrODS steel, to suppress the localized deformation at packet and block boundaries. Hot-rolling was recently processed for this purpose. The results of currently developed thermo-mechanical processing and resultant high-temperature strength are assessed in this section.
The 9CrODS steel powders in the standard composition and high carbon content (0.41 mass%) were mechanically alloyed by the planetarium type boll mill, where high carbon specimen was prepared to produce the free residual ferrite 9CrODS steel. MAed powders were consolidated by spark plasma sintering (SPS) at 1100°C and 44 MPa for 1 h; the hot-rolling (HRing) was conducted in the total reduction rate of 84 % at the austenitic γ phase region. The detail condition is as follow;
keeping temperature at 1,100°C for 1 min in the furnace and taking out from the furnace.
hot-rolling at three times in the total reduction ratio of 52 %; temperature decreases to 930°C
back to furnace to keep temperature at 1,100°C for 10 min and taking out from the furnace
again hot-rolling at three times in the total reduction ratio of 67 %; temperature decreases to 840°C at the end of hot-rolling, and air-cooled (AC) to room temperature in the cooling rate of 4000°C/h.
As a parameter, normalizing (N) by water quench and tempering (T) heat-treatments were subsequently carried out at 1050°C × 1 h and 800°C × 1 h, respectively, to evaluate structure change of the HRed specimens.
Fig. 16 shows SEM micrographs of high carbon specimen (0.41mass%): (a) HR-AC and (b) subsequent NT heat treatment. Both structures are significantly different and the coarsened grains are formed just after HR-AC, whilst subsequent NT heat treatment induces finely homogeneous and the typical tempered martensite that is usual structure of 9CrODS steel without residual ferrite similar to the right upper optical photo of Fig. 12. The larger grains in HR-AC belong to ferrite that was transformed from the γ-phase during the air cooling after HR. A surrounding area is martensite structure. The microstructure of the standard 9CrODS steel is shown in Fig. 17. The larger grains are also observed in HR-AC specimen, and subsequent NT one has fine grains that involve the residual ferrite and tempered martensite similar to the left upper photo of Fig. 12. The coarsened grains in HR-AC specimen shown in Fig. 17(a) could be composed of the residual ferrite and transformed ferrite, however, it is difficult to separate them by SEM photograph, since both are the same ferrite phases.
Concerning a formation of the coarsened ferrite grains, it is obvious that these grains do not correspond to the coarsened residual ferrite, since coarsened grains are seen even in high carbon specimen without the residual ferrite as shown in Fig. 16(a). It is considered that the finer γ-grains made by hot rolling at 1000°C accelerate the ferrite formation rather than martensite transformation during air cooling, from a viewpoint of a continuous cooling transformation (CCT) diagram of 9CrODS steels (Ukai & Ohtsuka, 2007). Furthermore, accumulation of severe strain energy in γ-grains induced by hot-rolling could be responsible for coarsening of the transformed ferrite grains.
4.2. High-temperature mechanical properties
The results of the tensile test at 700°C are shown in Fig. 18. This test was conducted for the standard 9CrODS steel, varying with just HR-AC and subsequent NT heat treatment. The NT specimen exhibits the same level of strength and ductility as the conventional 9Cr-ODS steels 6). This result can be understood by considering that the structure induced by HR completely disappears by normalizing heat treatment and its structure is composed of the residual ferrite and tempered marutensite as shown in Fig. 17(b). For HR-AC condition, the tensile strength and rupture elongation are recorded more than 500 MPa and 20 %, respectively. It is to be noticed that these tensile properties are adequately improved by HR-AC treatment far beyond those of the conventional 9CrODS steels.The excellent tensile performance of HR-AC specimen is ascribed to an existing of the transformed ferrite grains in the coarser size, which is shown in Fig. 17(a). The block and packet boundaries in martensite structure sometimes give appropriate sites for fracture, and thus lead to premature rupture (Abe, 2004). The coarsened ferritic grains could suppress the grain boundary deformation, and thus improve the tensile strength and ductility.
The creep rupture data of the standard 9CrODS steel for HR-AC and subsequent NT specimens were superimposed by star symbols in Fig. 13, where data are shown in the stress-rupture time correlation at 700°C. The creep rupture strength of HR-AC specimen (solid star symbol) obtained in this test at 700°C and 200 MPa approximately lies on the extrapolated line of FC specimen, but it is slightly higher. This improved strength over FC could be owing to the coarsening of the ferrite grains made by hot-rolling. Subsequent NT specimen also shows slightly higher strength than NT condition of the previous specimen containing the residual ferrite. Based on those finding, it is to be noticed that the ferrite grains in coarser size adequately improve the creep strength. This is due to a suppression of the block and packet boundary deformation, as mentioned in the tensile test.
The 9CrODS steel is an attractive composite-like material consisted of the residual ferrite and tempered martensite. Formation of both phases can be controlled by the process of α-γ phase transformation. The residual ferrite is a metastable phase and involves extremely finer nano-size oxide particles that are responsible for harder hardness; on the contrary, the tempered martensite contains coarser oxide particles and thus being softer hardness. Mechanical response of 9CrODS steel is completely different from that of the existed heat-resistant steels. Under stress loading, softened tempered martensite is preferentially deformed, in particular at the packet and block boundaries. The stress accumulation could give rise to a localized fracture along the packet and block boundaries in the tempered martensite. The microstructure of the tempered martensite was modified in terms of thermo-mechanical processing, and high-temperature strength was significantly improved through changing microstructure from the martensite to the transformed ferrite. The hardened residual ferrite plays a key role that provides a geometrical restriction for the deformation of softened tempered martensite. The best microstructure of 9CrODS steel is considered to be composed of hardened residual ferrite and softened transformed ferrite with free block and packet boundaries.
The author wishes to acknowledge many colleagues: in particular Mr. Takeji Kaito and Dr. Satoshi Ohtsuka of Japan Atomic Energy Agency (JAEA) and Dr. Takanari Okuda of Kobelco Research Institute Co. and Dr. Tsukasa Azuma of Japan Steel Works (JSW). The author would like to also thank many graduate students of my laboratory.