Investigation of the orientation relationship between nano-sized G-phase precipitates and austenite with scanning nano-beam electron diffraction using a pixelated detector

Scanning nano-beam electron diffraction with a pixelated detector was employed to investigate the orientation relationship of nanometer sized, irradiation induced G-phase (M$_6$Ni$_{16}$Si$_{7}$) precipitates in an austenite matrix. Using this detector, the faint diffraction spots originating from the small G-phase particles could be resolved simultaneously as the intense matrix reflections. The diffraction patterns were analyzed using a two-stage template-matching scheme, whereby the matrix is indexed first and the precipitates are indexed second after subtraction of the matrix contribution to the diffraction patterns. The results show that G-phase forms with orientation relationships relative to austenite that are characteristic of face-centered cubic (FCC) to body-centered cubic (BCC) transformations. This work also demonstrates that nano-beam electron diffraction with a pixelated detector is a promising technique to investigate orientation relationships and variants of nano-sized precipitates with complex crystal structures in other material systems with relative ease.

G-phase is an intermetallic silicide with a complex cubic crystal structure (space group Fm3m, lattice parameter around 1.1 nm) [1] and a stoichiometry represented by the formula M 6 X 16 Si 7 with M and X representing multiple possible metals. G-phase can be precipitated in various steels, with M=Ti, Mn, Cr and X=Ni, Fe being the most common varieties. In duplex steels subjected to ageing heat treatment, the phase precipitates in the ferrite phase with the cube-on-cube orientation relationship [2,3,4]. Large G-phase domains on grain boundaries are associated with embrittlement [5,6], but precipitation of G-phase on dislocations can improve strength and creep resistance [7]. Hence G-phase is of high technological importance in ferritic and duplex stainless steels.
In austenitic stainless steels, G-phase (M 6 Ni 16 Si 7 ) does not form under any ageing conditions, but it readily forms under irradiation over a wide temperature range (∼300-700 • C) [8,9,10]. This is detrimental in nuclear applications, where the formation of G-phase in austenitic steels is associated with void swelling [11,12,13,14,9], which leads to unacceptable embrittlement and loss of strength [15]. G-phase grows in austenitic steel due to radiation induced segregation of Ni and Si [16]. However, the nucleation mechanism of G-phase from Ni-Si supersaturated regions is not well understood. Moreover, the orientation relationship of the phase in austenite has not been established but is critical for understanding its detrimental effects. The cube-on-cube orientation relationship [17] has been proposed but most authors suggest there is no preferential orienta-tion relationship [16,8]. Recent electron diffraction work has shown irradiation induced G-phase may have a preferential but complicated orientation relationship with austenite [18,19] ferrite or martensite (α). Since G-phase forms with the cube-on-cube orientation relationship in ferrite, it is plausible to assume that G-phase may form in austenite analogously to ferrite [3].
The difficulty in establishing the orientation relationship of nano-sized precipitates with complex crystal structures using classical methods like selected area electron diffraction (SAED) is that diffraction patterns tend to contain many closely spaced reflections from multiple particles. If the orientation relationships are non-trivial with multiple variants, indexing these patterns is extremely challenging, and it is impossible to tilt individual particles to different zone axis orientations. An additional layer of complexity is that irradiated austenite typically contains other phases with complex crystal structures and similar lattice parameter as G-phase, such as M 23 C 6 ; these phases are often misidentified.
In this paper, we attempted to elucidate the orientation relationship of nano-sized (10-20 nm) G-phase precipitates in ion-irradiated austenite using scanning nano-beam electron diffraction (NBED) inside a transmission electron microscope (TEM) utilizing a fast pixelated electron detector. Scanning nano-beam diffraction is a technique whereby an electron probe with small convergence angle (typically < 1 mrad) is scanned across the sample, and a diffraction pattern is collected at each scan position. If the distance between the particles is large relative to the thickness of the specimen, then diffraction patterns will contain signal from individual particles and the matrix only. By comparing the diffraction patterns to a library of simulated diffraction patterns of the crystal in different orientations (templates), local orientations in each scan position can be derived [20,21]. The advantages of this technique are that no specific sample orientation is required before data collection, and crystallographic information from diffraction patterns can be directly correlated to real space coordinates.
One challenge of this method is that diffraction patterns contain both (weak) signal from the small precipitates and (strong) signal from the matrix phase above and below the particles. Since some matrix reflections can correspond to some precipitate reflections and since they tend to be more intense than the precipitate reflections, this can confuse the template matching algorithm. Recently it was demonstrated that this problem can be overcome by indexing the patterns first with the matrix templates, masking the images based on the matrix template, then indexing the precipitate phase [22,23].
Our initial trials showed that the conventional method of NBED data collection utilizing the ASTAR system (NanoMegas) was insufficiently sensitive to detect the faint G-phase reflections. Reliable indexing the diffraction patterns of small precipitates with complex crystal structure requires electron detection with the highest possible resolution and dynamic range. Hence, diffraction pattern data was collected with the fast (up to 384 frames/s) TemCAM-XF416 pixelated CMOS detector (TVIPS). A comparison of data collected with the ASTAR and TVIPS systems is given in supplementary materials. The microscope used was a JEOL 2200FS TEM, equipped with a Schottky field-emission-gun and operating at 200 kV. In addition to NBED, the morphology of the precipitates was investigated using high resolution high angle annular dark field (HR-HAADF) scanning TEM (STEM) using a probe corrected Titan microscope (Thermo Fisher Scientific) operating at 300 kV.
The material used in the study was an austenitic stainless steel in the 15-15Ti family with composition 15 wt% Ni, 15 wt% Cr, 1.8 wt% Mn, 1.2 wt% Mo, 0.5 wt% Ti, 0.6 wt% Si, 0.1 wt% C, bal. Fe [24], with a grain size of 10-15 µm. The material was irradiated with 4.5 MeV Fe 2+ ions up to 40 dpa at 600 • C. Additional details about the material and its characterization were published elsewhere [25,26,27]. Figure 1 shows the G-phase precipitates as imaged in a) TEM dark field (DF), b) using scanning TEM (STEM) energy dispersive X-ray spectroscopy (EDX), and c) atom probe tomography (APT). The precipitates were 10-20 nm in diameter, enriched in Si, Ni, Ti and Mn, and depleted in all other alloying elements. No other irradiation induced phases were found in this steel, making this material ideally suited to study G-phase. Electron transparent samples for TEM investigations were prepared via focused ion beam cross-section lift-out in a SCIOS 2 HiVac (Thermo Fisher Scientific) using a final cleaning at 2 kV.
The NBED data was collected using a camera length of 80 cm and a convergence angle of approximately 0.5 mrad. A map of 260 × 200 scan positions was collected with a step size between 1-2 nm. The region of interest was tilted slightly away from a 110 zone axis so that only few matrix reflections would be strongly excited. The raw data from the camera was converted to the .BLO file format compatible with the ASTAR software (NanoMegas) using a custom tool [28] and analyzed with methods and tools described in refs. [22,23]. This conversion involved some information loss, since the TVIPS camera collects 16-bit images, but the BLO format only supports 8-bit images.
The two-stage process for indexing the G-phase is illustrated in Figure 2 (prepared with Hyperspy [29,30] and Pyxem [31]). First, the austenite matrix is indexed in each pattern. The orientation maps were processed with MTEX v5.4.0 [32]. Figure 3 summarizes the orientation results derived from the template matching procedure. Hence, the results suggest there is an analogy between the way G-phase forms in austenite and the way ferrite or martensite does. Despite the space groups of G-phase and austenite being the same (Fm3m), the G-phase lattice most closely resembles a superstructure of ferrite; only minor atomic shuffles are required to transform a 4 × 4 × 4 supercell of BCC ferrite into the G-phase unit cell [3]. If the G-phase structure is similar to a BCC su-percell, this implies the formation of the phase must be associated with a similar kind of tetragonal distortion and rotation of the FCC lattice as when austenite transforms to ferrite, which explains the similarity between the orientation relationships. For all but the smallest particles, these strains must be accommodated by interface dislocations or twinning in the precipitating phase, which to a large extent determine the precipitate morphology and habit plane [33,34].
Some evidence of twinning inside G-phase precipitates was found, one example shown in Figure 4 a). The two parts of the particle deviate 52 • over a tilt axis close to the [111] G direction (nearly parallel to the optical axis) which resembles a Σ3 twin of a 60 • tilt over a 111 tilt axis. The grains also appear to be joined on a plane parallel to {111} γ since the interface is parallel to the edge-on Frank loops in Figure 2. The deviation from the ideal twin rotation likely originates from a cumulation of errors in the indexing procedure; this could be the result of the relative weakness of the G-phase reflections, high concentration of defects and localized strains, the substantial image processing (including masking) prior to indexing the G-phase, and dynamic diffraction effects that are not accounted for in the template matching procedure. It may be possible to further improve the angular resolution by utilizing the full 16-bit image depth during indexation and optimizing template matching parameters; a solution along these lines is being developed within the Pyxem package [31].
FCC to BCC transformations tend to be associated with irrational in-terface planes containing an invariant line along an irrational lattice vector which can be predicted from lattice parameter ratios [33]. If it is assumed that the G-phase unit cell represents 4 × 4 × 4 BCC unit cells and the ratio In summary, nano-sized irradiation induced G-phase precipitates were found to follow orientation relationships in austenite characteristic of the FCC to BCC transformation through NBED using a pixelated detector. Only by collecting the data using a CMOS camera with high signal-to-noise and high dynamic range could the faint G-phase reflections be adequately resolved to index them with the template matching procedure. This case study also  Additional anomalous spots and the direct beam are also circled in cyan and green respectively. b) A diffraction pattern showing additional diffraction spots originating from edge-on Frank loops. The mask to select these features is superimposed. c) A composite VDF image. The yellow contribution represents the Frank loops as isolated by the mask shown in b). The red contribution represents the G-phase VDF obtained by masking the matrix and direct beam and Frank loops. d) and e) show diffraction patterns from locations containing G-phase particles. The best fit G-phase template is superimposed. Note that the background from inelastic scattering was removed from the diffraction patterns in a), b), d) and e) using a difference-of-Gaussians process.