Microstructure evolution during nitridation. The FeNi nanopowders (NPs) were synthesised using a novel low oxygen induction thermal plasma (LO-ITP) system. The as ITP synthesized NPs were then reduced under a hydrogen gas flow to remove the possible oxide surface. Finally, the FeNi NPs were nitrided under an ammonia gas flow (see Methods). To investigate the formation mechanism of the FeNiN precursor phase in the FeNi NPs during nitridation, we conducted a detailed microstructure analysis of NPs in an intermediate state, that is, a mixture of the Fe2Ni2N parent and FeNiN product phases. Figure 2a,b shows the morphology with scanning electron microscopy (SEM) images of the as ITP processed and nitrided FeNi NPs, respectively. Over 500 NPs were counted for the particle size distribution (insets in Fig. 2a,b), and the mean particle size is approximately 116.9 ± 72.6 and 99.8 ± 53.2 nm for the as ITP processed and nitrided NPs, respectively. Figure 2c presents the X-ray diffraction (XRD) patterns of (i) as ITP processed and (ii) nitrided FeNi NPs. As indicated, the as ITP processed FeNi NPs possess a single A1-FeNi phase, whereas the nitrided FeNi NPs show a mixture of Fe2Ni2N and FeNiN phases. Rietveld analysis using the RIETAN-FP software13 revealed approximately 80% of the Fe2Ni2N phase and 20% of the FeNiN phase in the nitrided FeNi NPs. Such an intermediate state with mixed phases during nitridation can serve as a suitable material platform for investigating the formation mechanism of the FeNiN precursor phases.
Nucleation of the FeNiN phase in the nanotwinned region. To investigate the formation mechanism of the FeNiN precursor phase, it is important to first identify where and how the FeNiN phase is distributed in the nitrided FeNi NPs. Since Fe2Ni2N and FeNiN phase belongs to space groups with distinct lattice symmetry (Pm-3m and P4/mmm for the Fe2Ni2N and FeNiN phase, respectively) and have different lattice constants (a = 0.377 nm for the Fe2Ni2N phase vs. a = 0.400 nm, c = 0.371 nm for the FeNiN phase). Thus, it is possible to distinguish between the two phases using transmission Kikuchi diffraction (TKD) analysis. Figure 3a,b show the TKD image quality and phase maps, respectively. The clear colour contrast in the TKD phase map (Fig. 3b) confirms a mixture of the FeNiN (red) and Fe2Ni2N (green) phases in the nitrided FeNi NPs, which is consistent with the XRD results (Fig. 2c). Interestingly, as indicated by the white arrows in Fig. 3b, the FeNiN and Fe2Ni2N phases in the nitrided NPs generally share sharp/straight boundaries.
Another approach considered to distinguish between the FeNiN and Fe2Ni2N phases is the different nitrogen concentration in the Fe2Ni2N (20.0%) and FeNiN (33.3%) phases. Specifically, the visible contrast between the electron energy loss spectroscopy (EELS) and energy dispersive X-ray spectroscopy (EDS) mapping of nitrogen can help accurately identify where and how the FeNiN phase is distributed in the nitrided FeNi NPs. Figure 3c,d show the representative EELS maps of the nitrided FeNi NPs, marked as “B” in Fig. 3a. With a high sensitivity to the light elements, the EELS map of the nitrogen distribution in Fig. 3d shows that the left part of “B” has a relatively higher nitrogen concentration than the remaining right part, sharing a boundary which is consistent with the TKD phase map in Fig. 3b. Interestingly, the nitrogen distribution feature detected here is different from the conventional core/shell structure reported in the nitrided Ti and Fe NPs14,15, implying a distinct nitrogen diffusion process and special attention to the twin boundary in the current FeNi NPs.
Figure 3e shows the annular bright-field scanning transmission electron microscopy (ABF-STEM) image of the FeNi NP marked as “C” in Fig. 3a. The multiple bright/dark stripe diffraction contrasts in the ABF-STEM image, which is due to the lattice direction change across the twin boundaries (TBs), reveals a high-density nanotwin substructure in the nitrided FeNi NPs. The low-magnification TKD grain boundary maps (refined rotation angle: 55–65°, shown in Supplementary Fig. S1) reveal that nearly 49.5% of the as ITP processed and ~38.1% of the nitrided FeNi NPs contained nanotwin substructure. Such intensive nanotwins in the FeNi NPs may have formed during the quenching of the ITP process (quenching twins or mechanical twins16), induced by the thermal strain energy field relief or nitridation process (annealing twins17), a consequence of the recrystallisation/phase transformation of face-centred cubic (fcc) metals with low stacking fault energies. Surprisingly, we observe significant solutes segregation at the TBs detected in the corresponding EDS maps (Fig. 3f-i). The line scan profiles across the nanotwinned regions (Fig. 3j) indicate a clear enrichment of Fe, and N while depletion of Ni occurs at the TBs (indicated by the arrows in Fig. 3j). More solutes segregation examples (EDS maps) can be found in Supplementary Fig. S2. To confirm whether such solutes segregation initially occurs during the thermal plasma process or the nitridation process, EDS mapping was also checked for the as ITP processed FeNi NPs. Figure 3k-n shows the representative ABF-STEM images and corresponding EDS maps of the as ITP processed FeNi NPs. Although the typical nanotwins were observed in Fig. 3k, there was no visible solute segregation in the corresponding EDS maps (Fig. 3l-n) (see additional example in Supplementary Fig. S3), which indicates that solute segregation mostly arised during the nitridation process. Grain boundary segregation and solute diffusional phenomena are well reported interface characteristics18,19. In contrast, coherent TBs are dislocation-free and have perfectly aligned atomic stacks, resulting in low interfacial energies compared to grain boundaries; thus, strong solute segregation is not initially expected at coherent TBs. However, several recent reports on different alloy systems have shown solute segregation along TBs and proved it is associated to the migrating heterophase interface boundary, driven by minimizing the intrinsic stacking fault, elastic strain, and/or total energy in the system20-22. On the other hand, it is also worth notice that solutes accumulation at the TBs can also act as Cottrell clouds surrounding the twins and thereby restrict further slip of the dislocations which was well discussed as the strength enhancement mechanism for the twinned soft materials23-25. Since the strong twin boundary-dislocation pinning interaction, the twin growth and/or detwinning process which determined by the dislocation motion will become time dependent. It could be one of possible reasons why the synthesis of the FeNiN precursor phase is the most time-consuming part of the nitrogen topotactic reaction route.
Figure 4a,c presents the atomic-resolution angle annular dark-field (ADF)-STEM images viewed along the [1
10] zone axis to elucidate the detailed features of the atomic stacking and TBs of the as ITP processed and nitrided FeNi NPs, respectively. From the ADF-STEM images, the stacking faults parallel to the (111
) plane creates nanotwin substructures in the matrix with a width varying from 2 (~0.5 nm) to 16 atomic layers (~3.5 nm). The nanotwin substructures are also confirmed by the typical two-set patterns presented in the fast Fourier transform (FFT) images in Supplementary Fig. S4. For the as ITP processed A1-FeNi NP in Fig. 4a, the (110) interplanar spacing in the nanotwinned region (marked as “B” in Fig. 4a, = 0.255 nm) seems to expand slightly compared to the matrix region (marked as “A” in Fig. 4a, = 0.253 nm). More remarkable (110) interplanar spacing expansion is detected in the nitrided FeNi NP with = 0.278 nm in the nanotwinned region (marked as “B′” in Fig. 4c), while = 0.243 nm in the nanotwinned region (marked as “A′” in Fig. 4c). Furthermore, the (111) planes in both the matrix and nanotwinned regions, which are supposed to be related by mirror symmetry with respect to the (111
) twining plane, possess relatively different tilted angles, as indicated in Fig. 4a (70.5 vs. 72.8º). The degradation in the mirror symmetry becomes more serious in the nitrided FeNi NPs, as indicated in Fig. 4c (66.2 vs. 76.5º) which finally result in an “incoherent” twin boundaries with much higher boundary energy than a coherent twin boundaries26. To estimate the lattice distortion/strain distribution at the nanotwinned regions, a geometric phase analysis (GPA) was conducted to reconstruct strain maps from the corresponding atomic-resolution ADF-STEM images (see Methods). Figure 4b,d shows the shear strain (εxy) field maps of the as ITP processed and nitrided FeNi NPs, respectively. Although the atomic strain analysis is affected by camera resolution, the difference in the shear strain maps is substantial, indicating the qualitative difference in the lattice distortion between the twin and matrix regions of both the as ITP processed and nitrided FeNi NPs. The bright contrast in the shear strain maps which indicating positive shear strain suggests that the crystal units are significantly stretched along the [110] direction (which is close to xy direction) and compressed in the [002] direction for both the as ITP processed and nitrided FeNi NPs. Furthermore, a comparison of the interplanar spacing of the (110) and (002) planes between the matrix and nanotwinned regions of the nitride FeNi NPs (spacing calculated from Fig. 4c) reveals a +12.8% expansion along the [110] direction and –10.2% shrinkage along the [002] direction in the nanotwinned region (the bright region in Fig. 4d). The severe shear strain triggers the dislocation accumulations in the nanotwinned region, which eventually lead to significant lattice distortions as well as the detected (111
) plane expansion. Similar phenomena has been reported by Wu et al. in the FeCoCrNi high-entropy alloy27.
Thus, we have experimentally demonstrated that: a) twins are typical features present in both the as ITP processed (~49%) and nitrided FeNi NPs (~38%); b) in some of the nanotwinned regions, there is a visible enrichment of Fe and N solutes while a depletion of Ni at the TBs; c) the nanotwinned region suffers from severe shear strain (εxy ~ 10%) which consequently induces a significant lattice distortion as well as a lattice expansion. The expanded lattice planes in the nanotwinned region will open a channel and favours the diffusion of nitrogen atoms into the core of the FeNi NPs; meanwhile, the enriched Fe at the TBs can attract more nitrogen due to its relatively smaller iron-nitride formation energy than nickel-nitride28. Thus, it is believed that such nanotwinned regions with relatively higher nitrogen concentration and low-symmetry crystal will be energy favourable to promote the nucleation of the FeNiN precursor phase.
Massive transformation of Fe2Ni2N to FeNiN phase. As mentioned previously, the FeNiN product phase possesses different nitrogen concentrations and crystal structures compared to the Fe2Ni2N parent phase; thus, one can detect sharp and sometimes straight phase boundaries in nitrided FeNi NPs (as illustrated in Fig. 3b,d) in which the phase transformation is not fully complete. Therefore, after the preferential nucleation of the FeNiN product phase in the nanotwinned region, it is important to understand how this new product phase grow into the Fe2Ni2N parent matrix. To clarify this critical question, a detailed microstructure analysis focusing on the FeNiN/Fe2Ni2N phase boundaries was performed.
Figure 5a,b,c shows the ABF-STEM images of the nitrided FeNi NPs in which the distinct bright/dark phase contrast indicates clear FeNiN/Fe2Ni2N phase boundaries. Figure 5d,e,f shows the corresponding EDS composite (Fe (red)-Ni (blue)-N (green)) maps. The light yellow/pink contrast illustrates a relatively N-rich/-poor zone, which was later identified corresponding to the FeNiN product/Fe2Ni2N matrix phase by matching the lattice spacing and indexing in the FFT patterns and the atomic-resolution ADF-STEM image (see Supplementary Fig. S5). The clear, sharp, and sometimes straight phase boundaries are consistent with the phase/elements distribution characteristics observed in Fig. 3b,d. With detailed microstructure characterisation (Fig. 5g,h,i) of the FeNiN/Fe2Ni2N phase boundaries, it is surprising that the FeNiN product phase usually develops a high-index/irrational orientation relationship (OR) with the Fe2Ni2N parent matrix. The FeNiN/Fe2Ni2N phase boundary displayed in Fig. 5g (corresponding to the dash squared region in Fig. 5d) seems to show a high-index/irrational OR at the interface with the Fe2Ni2N phase on a [110] zone axis, while the FeNiN phase is tilted slightly away from the exact zone axes. In Fig. 5h (corresponding to the dash squared region in Fig. 5e), the FeNiN/Fe2Ni2N phase boundary demonstrates a sharp/fully epitaxial coherent interface with the Fe2Ni2N phase on a [100] zone axis, while the FeNiN phase is tilted slightly away from the exact zone axes. Finally, in Fig. 5i (corresponding to the dash squared region in Fig. 5f), the FeNiN/Fe2Ni2N phase boundary demonstrates an atomic rough incoherent interface with the FeNiN phase on a [113] zone axis, while the Fe2Ni2N phase is tilted slightly away from the [001] zone axes. The possible high-index OR developed at the FeNiN/Fe2Ni2N interface is illustrated in Fig. 5j: (121
)FeNiN [113]FeNiN || (100)Fe2Ni2N [001]Fe2Ni2N. Such a high-index OR results in a small lattice misfit (1.8%) at the interface which favours a lower strain/interfacial energy. The results indicate that it is possible for two phases with a high-index OR, and to obtain some degree of atomic matching across the high-index interface planes by selecting a suitable interface plane in the nitrided FeNi NPs. However, some occasional planes apparently have no matching planes on the opposing side of the boundary, makes such an interface partly coherent in the plane-to-plane geometry. The absence of a lattice correspondence defines such a FeNiN/Fe2Ni2N interface in Fig. 5i as structurally incoherent. Thus, the FeNiN/Fe2Ni2N phase boundaries in the nitrided FeNi NPs present both a structurally commensurate (coherent) and incommensurate (incoherent) interface, with a variable degree of commensurability depending on the orientation and planarity of a particular interface.
Figure 5k,l shows the relative low-magnification ABF-STEM images of the FeNiN/Fe2Ni2N interfaces displayed in Fig. 5a and 5c, respectively. It demonstrates the typical terraces/ledges featured at the interfaces, implying that the growth of the FeNiN product phase, accompanied by the prorogation of the interfaces, follows a ledge mechanism29,30. This kind of ledgewise growth motion of the interface has been reported in the FeNi31,32, MnAl21,33, and TiAl34,35 alloy systems associated with the massive transformation process. It has been experimentally demonstrated that massive transformations involve a special crystallographic OR between the parent and the product phases, leading to coherent nucleation. This is followed by growth, in which the interface changes from coherent to partly coherent, or even to a high-index irrational interface, and migrates by a ledge mechanism29,30, which is principally accomplished by short-range atom jumps that occur almost continuously across the structurally incoherent interfaces21,34. The massive γ to α transformation was previously reported in the FeNi binary alloy system (with a Ni concentration ≤15%)36-38. The present results of elements segregated at the TBs (Fig. 3i,j), the high-index OR developed at the FeNiN/Fe2Ni2N interface (Fig. 5g-j), and the ledgewise growth motion of the FeNiN/Fe2Ni2N interface (Fig. 5k,l) agrees well with the typical characteristics of a massive transformation, indicating that the FeNiN phase is most likely formed through a massive transformation from the Fe2Ni2N matrix phase during the nitridation process. And this is the first experimental demonstration of a massive transformation in the Fe-Ni-N system. Based on our experimental results, one possible formation route for the FeNiN precursor phase during the nitridation process is proposed as follows: First, the FeNiN product phase initially energetically favours to nucleation at the nanotwinned region in the Fe2Ni2N parent matrix. Then the FeNiN/Fe2Ni2N interface develops a high-index OR through the short-range atoms jump across the interface to minimise the strain/interface energy. Finally, the massive Fe2Ni2N to FeNiN phase transformation is accomplished through the migration of the high-index OR interfaces following a ledge mechanism. In some nanotwinned regions, solutes segregation at the TBs during the massive transformation may occur and the block the dislocation motion which will restrict the FeNiN/Fe2Ni2N interface migration and finally leads to a time-consuming phase transformation process.