Incorporation of Y₂O₃ Particles into 410L Stainless Steel by a Powder Metallurgy Route

The current most common commercialized nuclear fission reactors work at a temperature range of 250-300 °C, a pressure range of 7-15 MPa, and displacement per atom (dpa) in core structural components of 10-25. Future innovative nuclear systems and technologies have designs much more demanding on material performance. Some so-called “Generation IV” designs have operating requirements up to 1100 °C, 24 MPa, and dpa up to 150.

Efforts to develop new Fe-Cr-Mo-type heat-resistant steels were initiated for steam generators in the early 1970s. During the development of high-chromium (9 to 12%) steels for high-temperature application, the use of stainless steel for nuclear applications was demonstrated. This included attempts to replace Mo with W to reduce the radio-activation of the material (Ref 1). Since then, much experience has been gained with components fabricated with high-Cr martensitic steel (Ref 2). Austenitic stainless steels are not preferred for advanced nuclear reactor applications because of high swelling rates and high thermal stresses caused by low thermal conductivity and high thermal expansion coefficient.

Structural materials for future nuclear applications have to be stable at elevated temperatures and have good resistance to irradiation. This resulted in focussing structural materials research on ferritic-martensitic (F/M) steels. Presently, F/M steels are the primary candidate for components of advanced reactor systems such as fuel cladding and duct materials. Precipitates are a major source of strength in F/M steels. Processing procedures that decrease the size of precipitates and increase their number density could enhance mechanical properties. This is akin to the so-called oxide-dispersion-strengthened (ODS) steels, where particles are dispersed into the steel matrix resulting in an increase in strength at higher temperatures (higher creep resistance). ODS steels have been visualized as an alternative to increase the creep resistance of the ferritic, martensitic, or F/M steels (Ref 3). Powder metallurgy is the most commonly used method of producing ODS steels.

The dispersion particles are generally Ti₂O₃ and/or Y₂O₃. The first ODS steels were of high chromium (12 up to 20% Cr). Attempts to solve the anisotropy problems caused by grain structure and a strong deformation texture resulted in the use of 9-11% Cr, 2-3% W and by the austenite-to-martensite transformation (through cooling from the austenization treatment temperature). The Cr content is also optimized to achieve higher toughness and lower ductile-to-brittle transition temperatures. The resulting steels have outstanding tensile properties and reduced anisotropy, but the creep-rupture properties are reduced compared to the high-chromium ODS steels (Ref 3, 4).

Following initial work on ODS as potential materials for nuclear applications in the 1990s, there has recently been significant interest in the fabrication of fuel cladding based on ODS materials, as a replacement for zirconium alloys in pressurized water reactors, and also for sodium-cooled designs (Ref 5). For nuclear applications, steels chosen for base alloy are reduced-activation composition steels, so that mechanical properties at high temperatures such as resistance to creep and recrystallization are improved, without loss of the advantages of the ferritic or martensitic microstructures, especially the resistance to void swelling (Ref 6-8). The large interface area associated with a small volume (0.25%) of yttria addition provides “sinks” for radiation damage and implanted ions in a reactor core (Ref 9). The complex, ultrafine-grained microstructure consists of nanoclusters that are highly tolerant to high dose irradiation at elevated temperatures (Ref 10).

The chemistry of the base alloy can have an effect on the stability and dissolution of the ODS particles during processing. Titanium additions to the alloy have been found to promote yttria dissolution during processing and stabilize the particle size during subsequent heat treatment and forming (Ref 11).

The ODS particles effectively anchor dislocations, thus enhancing creep resistance, manifested as a decrease in creep ductility and an increase in creep-rupture life (Ref 12). However, studies trying to establish the upper operating temperature limit for these steels (Ref 13-15) are finding that the creep behavior is very sensitive to composition and microstructure. The fatigue properties have also been found to depend on the processing route and final microstructure (Ref 16).

Small-angle neutron scattering (SANS) has been found to be a valuable technique for the study of ODS steels, as it can reveal the distribution of the particles at the nanometre scale (Ref 17). Conventional neutron and x-ray diffraction methods can also be valuable in looking at the evolution of phases and dissolution in ODS materials (Ref 18, 19).

This work presents the results of a study on the powder metallurgy processing of an ODS 410L alloy steel. The microstructural evolution and mechanical properties were examined and measured. SANS was used to examine the dispersion and inclusion particles.

The base material was AISI 410L powder with nominal composition of 11.5-13 wt.% Cr, max 1 wt.% Si, max 1 wt.% Mn, max 0.5 wt.% S, and max 0.5 wt.% P. Two different particle sizes of yttria were used for incorporation into the material: 50 nm and 0.9 µm. These sizes were selected to provide a comparison with a particle size that is effectively comprised of single crystallites and can be incorporated into the matrix by dissolution (50 nm), and a larger size which offers greater scope for particle fracture during the mechanical milling process.

Scanning electron microscope (SEM) observations on the atomized 410L powder (Fig. 1) show particle sizes smaller than 100 µm, which is suitable for subsequent powder metallurgy operations. The morphology of the alloyed powder shows spherical particles. Some larger metal powder particles have many small ‘‘satellites’’ stuck to them, which ideally should be eliminated for good packing and flow attributes (Ref 20).

SEM observations on the yttria powders showed particles much larger than expected, with some particles up to 30 µm in size, with composition of pure yttrium rather than yttria. Both yttria powders, nominally of 0.9 µm and 50 nm in size, show these large particles, with an example shown in Fig. 2 for the 0.9-µm powder.

Materials were fabricated by Aerospace Metal Composites Ltd, Farnborough, by a mechanical alloying (MA), powder metallurgy route. The ODS material was fabricated by mechanical milling 0.25 wt.% Y₂O₃ with the alloy powders under inert argon atmosphere. Two different ODS variants were produced with Y₂O₃ particles of 0.9 µm and 50 nm sizes. After milling, the powders were canned and degassed. After degassing, the can was sealed. Finally, hot isostatic pressing (HIPping) was performed at a temperature of 1120 °C. Materials were also produced without oxide reinforcement, and one batch was prepared by simple consolidation of the 410 powder without the MA stage.

Heat treatment was performed on the final billets to achieve a tempered martensitic structure and remove features such as residual ferrites at grain boundaries arising due to the slow cooling rate (furnace cooling) after HIPping. The microstructures of as-received unreinforced 410 material without MA, before and after heat treatment, are shown in Fig. 3.

The heat treatment was austenitizing at 1000 °C for 30 min followed by oil quenching to achieve a fully martensitic microstructure and then tempering at 650 °C for 2 h and furnace cooling. The heat treatment was performed under ambient atmosphere. In Fig. 3b, it can be seen that a fully tempered martensitic microstructure was achieved by the heat treatment. No residual ferrite was observed.

Cylindrical tensile test specimens were extracted from the longitudinal direction of the HIP billets (though it must be noted that no direction-dependence in the properties would be expected). The tensile test specimens were 6.25 mm in diameter and 25 mm in gage length. The tensile samples were tested to failure at room temperature and at 625 °C at an extension rate of 0.01 mm/s.

Creep tests were carried out by applying a fixed tensile load at various temperatures. The “mini-creep” samples used had dimensions of 10 mm in gage length and 2 mm in diameter. Mini-creep tests were conducted at the Australian Nuclear Science and Technology Organisation (ANSTO) facilities. All tests were conducted in vacuum at 625 °C under loads of 75, 100, and 150 MPa.

The SANS experiment was performed on the KWS-1 instrument of JCNS at FRM-II, Munich. A magnetic field was applied at 0.4 and 1.2 T with a neutron wavelength of 7 Å. Rectangular samples were manufactured with dimension of 20 × 20 mm² with thickness of 2 mm. This thickness was chosen as higher thickness can cause multiple scattering (Ref 21). The instrument has an upper length scale limitation of 400 nm. A polydispersed spherical Schultz distribution/model was used for data analysis and fittings.

The PM and HIP process used here for the 410L martensitic stainless steels increased the material strength as compared to conventional processing. The ODS materials, however, did not show improved mechanical properties at room temperature. They have nearly the same mechanical properties than the non-ODS materials, yet they showed a lower fracture strain.

Ductility in these materials is controlled by microvoid nucleation and coalescence prior to fracture (Ref 24). Propagation of a fracture is the combination of nucleation of voids and growth and/or coalescence of these voids owing to plastic strain. One of the most important factors in the nucleation of voids is the interfacial bonds. As the inclusions in the materials tend to have very low-strength interface bonds, they are potential void nucleation points. Some inclusions like MnS can also be considered as virtual voids, i.e., with very little or no bond to the matrix surrounding them. In the case of oxides, void nucleation starts by particle cracking; while for sulfides, which have a lower interfacial energy between inclusion and the matrix, decohesion occurs at the inclusion/matrix interface, even at low strains.

After voids are nucleated at the inclusions owing to weak interfacial bonds, they tend to elongate along the tensile direction up to a certain limit by plastic extension. This limit is defined as the distance between neighbor inclusions. When they reach that limit, a slip plane is formed between voids, and they coalesce (Ref 25). This continues through all neighboring inclusions, and then, fracture propagates leading to complete rupture. The fractography analyses show inclusions inside the dimples where the microvoid nucleation has started. Figure 8 shows the mechanism of void elongation and coalescence under plastic extension followed by crack propagation and matrix separation leading to fracture (Ref 26). Thus, if the distance between these inclusions is small, the plastic strain needed to coalesce the nucleated voids will be less as well, which means the total fracture strain will be smaller and in consequence, the material has low ductility.

During creep, voids and cracks nucleate at grain boundaries and follow grain boundaries until rupture. The low creep lives observed are associated with the inclusion particles present at grain boundaries which ease the nucleation and propagation. Inclusion particles were clearly observed on the fracture surfaces following creep testing, as seen in Fig. 9. There was a high density of particles observed on the creep fracture surfaces. It was clear that the particles had initiated multiple voiding within the material and that this had led to an early onset of tertiary creep and a concomitantly low creep life.

Some of the particles are composed of two or more individual particles. Figure 10 indicates schematically why there are more particles seen on the fracture surface SEM image than are present as a volume fraction in the material.

Particles were observed in both non-ODS and ODS variants of the materials. In the non-ODS material, the particles are comprised of MnS and Si-O systems. In the ODS material, Si-O, Y-Si-O, and MnS were seen to form particle clusters.

The Y-Si-O particles are expected to be Y₂Si₂O₇ which is thermodynamically favorable (Ref 27). Figure 11 shows EDX data taken from the fracture surface of an ODS-0.9-µm sample after room temperature tensile testing. These EDX results show that the atom percentages of Y, Si, and O satisfy the condition of Y₂Si₂O₇ as 1 atom for each Y and Si and 3.5 atoms for O. Similar atomic calculations show that the other particles are SiO₂ and MnS particles.

Agglomeration of the inclusions, resulting in a much bigger overall particle size, has a negative effect when compared to particles of smaller sizes. Such larger particles tend to form voids at lower strains than smaller particles, which impairs rather than improves the mechanical properties.

There is an important point about the Y-Si-O interaction and formation of Y₂Si₂O₇ particles. It is quite common that yttria particles dissolve in the matrix during ball milling and precipitate again as pure yttria in the system at high temperature later in the manufacturing process: this is the main mechanism to disperse the matrix with oxide particles to make oxide-dispersion-strengthened alloys. This dissolution and precipitation are needed for a homogeneous microstructure and size control of oxide dispersions in the nanometre range (Ref 28). If yttria dissolved in the matrix and then precipitated, then there should be separate particles with composition Y₂O₃ or complex oxides such as Y₂Ti₂O₇ if Ti was added to the system. In our case, the yttria-containing particles studied by SEM and EDX analysis are all Y₂Si₂O₇. However, small-angle neutron scattering (SANS) experiment results showed that there are pure Y₂O₃ oxide particles in the system at the nanometre level which cannot be detected in the SEM. SANS results from a pore-free batch of the ODS-50-nm material are shown in Fig. 12. The size of the Y₂O₃ particles in the as-received condition was measured as 33 nm. So, in the system, there are both Y₂O₃ and Y₂Si₂O₇ particles together.

In the literature, it is reported that up to 15 wt.% yttria has been dissolved in ball milling in Fe-24 wt.% Cr steel (Ref 29). Since the material being investigated contains 0.25 wt.% yttria, it can be assumed that it is highly likely that yttria could be fully dissolved. On the other hand, some research on ODS manufacturing shows that yttria does not “dissolve” during ball milling, rather the effect is a reduction in size to a few nanometres due to fracturing or amorphization of the yttria which takes place at the interfaces of the matrix (Ref 18, 30). In many ODS analyses, the size of the oxide clusters was found to be around 1-4 nm after dissolution and precipitation (Ref 31-35). Even after heat treatment and coarsening of the particles, oxide clusters of size around 10 nm are observed (Ref 35). So, it is not likely that oxide particles of 33 nm are formed by dissolution and precipitation mechanisms, because even initially, the yttria particles were reported as 50 nm. It is more likely that the formations of oxide clusters seen here are a result of interaction between the yttria and SiO₂ particles during ball milling. The pure Y₂O₃ particles (33 nm) detected by SANS can be attributed to size reduction of the particles owing to fracturing in the MA process. The size of the yttria particles should decrease to some extent for dissolution to occur, as in theory Y and O are thought to enter the matrix lattice.

The presence of SiO₂ will, therefore, have a significant effect on the microstructure of ODS steels by preventing dissolution of the yttria and/or leading to nucleation (if there is dissolution and precipitation) of complex Y₂Si₂O₇ oxides. This complex oxide should not be mistaken with Y₂Ti₂O₇ or Y₂TiO₅ which are known to be better in strengthening ODS systems, as Ti is added to ODS systems purposely to improve mechanical performance. The sizes of the complex Y₂Si₂O₇ particles are relatively large due to their formation from the relatively large SiO₂ inclusions, with the size of the particles reaching 2-µm level in some cases, which is clearly not desirable. It has previously been observed that yttria particles in ODS steels have an affinity for Si (Ref 36).

As shown in Fig. 12, the size of the particles changed with heat treatment. The mean size of the Y₂O₃ increased from 33 to 52 nm after heat treatment, whereas the mean size of the complex oxide Y₂Si₂O₇ decreased from 261 to 197 nm. It is expected to have larger clusters after heat treatment as particles coarsen at high temperatures. However, the size of the complex oxide decreased. It can also be seen that the peak of pure yttria (Y₂O₃) broadened and the peak of the complex oxide (Y-Si-O) became narrower after heat treatment. The results also suggest that larger complex oxide particles dissolved prior to small ones because it is shown that minimum size of the complex particles increased in heat-treated condition from around 100 to 150 nm but the average size together with the maximum size decreased. So, this can be attributed to preferential dissolution of larger particles.

SANS results from the ODS 0.9 µm are shown in Fig. 13. As the maximum size of the particles that can be measured by the available SANS instrument was limited to 400 nm, only the start of the size distribution graph of Y₂O₃ particles is obtained. However, when compared to heat-treated and as-received results, it can be noted that the minimum size of the Y₂O₃ particles increases with heat treatment. Complex particles in this case are much larger than 400 nm; so, these are not shown in the results. An important outcome of the SANS experiment on ODS 0.9 µm is the effect of the initial yttria size on the size of the oxide particles after manufacturing. SANS results showed that the average size of the pure Y₂O₃ particles in ODS 0.9 µm is greater than 400 nm in both as-received and heat-treated conditions. This can contribute to the conclusion that dissolution of the yttria in ball milling did not occur, because if the yttria was dissolved in the matrix, then after precipitation during compaction nanometer-sized yttria particles would be observed (Ref 37). It appears that yttria particles fractured into smaller pieces in the ball milling and some of those particles interacted with SiO₂ and formed complex oxides and the remaining ones stayed as pure Y₂O₃ particles