3.1. Phase Composition and Grain Structure
The XRD patterns of the studied samples are shown in
Figure 1a. Only the typical diffraction peaks representative of FCC matrix can be detected. Of note that the peaks of the HA-30 sample shows a little asymmetry, generally indicating the presence of ordered L1
2 phase [
17]. Through deconvolving the (111) diffraction peak (
Figure 1b), the lattice constants of FCC and L1
2 phases were calculated to be 0.3582 nm and 0.3578 nm, respectively. Such the phenomenon of peak overlap is not obvious in the HA-3 sample (
Figure 1c), so whether the presence of L1
2 phase needs to be further verified.
The backscatter electron (BSE) technique was employed to examine the microstructure. From the low magnification BSE images (
Figure 2a, b), it is observed that the black particles are uniformly distributed in both the samples studied. Upon magnification, some fine particles distributed at grain boundaries (GBs) can be found in the HA-3 sample (
Figure 2c). With the annealing time prolonging to 30 min (HA-30), the intergranular particles grew and could be clearly identified as grey and white, as shown in
Figure 2d. The SEM-EDS point-scanning results (
Figure 2e) reveal that the black particles are enriched in the Ti element, while the grey and white ones are enriched in the Cr and the Ni, Al, Ti elements, respectively.
In order to identify the precipitates accurately, detailed microstructure analyses of the HA-30 sample based on TEM technology were carried out. The bright-field (BF) image in
Figure 3a shows a micron-sized particle, which can be identified as MC carbide by selected-area electron diffraction (SAED). The corresponding TEM-EDS mapping (
Figure 3b) reveals that the MC carbide is rich in Ti, and depleted in Co, Cr, Ni, and Al, consistent with the black particles in BSE images. As reported that the higher affinity and negative mixing enthalpies of C-Ti atom pairs is favorable for the formation of MC carbides [
18]. According to the calculated phase diagram in our previous study [
7], it is speculated that the MC carbides were mainly inherited from the solidification phase and grow further during the solid solution and hot rolling stages, thus reaching the sizes of micrometer-scale.
Figure 2.
BSE images showing the microstructures of (a, b) HA-3 and (c, d) HA-30 samples. (e) Elemental distributions of the precipitates present in the alloy studied.
Figure 2.
BSE images showing the microstructures of (a, b) HA-3 and (c, d) HA-30 samples. (e) Elemental distributions of the precipitates present in the alloy studied.
Figure 4a shows a cluster of particles distributed at GB regions. Through combining the SAED (
Figure 4b
1 and b
2) and TEM-EDS mapping (
Figure 4c), two types of precipitates of M
23C
6 carbide and L1
2 phase were identified. The former is enriched in Cr, while the latter in Ni, Al, Ti, which is consistent with the grey and white particles in the BSE images, respectively. Upon inspection of many BSE and BF-TEM images, it is noted that the M
23C
6 carbide and L1
2 phase were almost always distributed adjacently. Moreover, we found that the M
23C
6 carbide, FCC matrix, and L1
2 phase show a cubic-cubic orientation relationship, i.e., [110]
M23C6∥[110]
FCC∥[110]
L12 (
Figure 4b
3). These may indicate an interactive correlation between the formation of the M
23C
6 carbide and L1
2 phase. Specifically, the rapid diffusion of C atoms at high temperatures favors the preferential precipitation of M
23C
6 carbides at GB regions; this can not only deplete Cr to create in turn a favorable elemental environment locally enriched in Ni, Al, and Ti, but also provide favorable nucleation sites, thus contributing to the precipitation of L1
2 phase adjacent to the M
23C
6 carbides.
Previous reports [
19,
20] have shown that in the similar Al/Ti-doping 3d transition H/MEAs, the nano-scaled L1
2 particles are typically precipitated in the grain interior uniformly upon medium-temperature annealing. Therefore, we conducted a further observation for the intragranular microstructure of both the samples studied.
Figure 5a is a BF image of the HA-30 sample, showing the residual dislocations. The SAED pattern shows the weak spots in addition to the main diffraction spots representative of the FCC matrix, indicative of the presence of ordered L1
2 phase. The dark-field (DF) image in
Figure 5b shows the diffuse distribution of nano-scaled L1
2 particles.
Figure 5c is the high-resolution (HR) TEM image and corresponding Fast Fourier transform (FFT) images. By measuring the crystal plane spacing directly, the lattice constants of FCC and L1
2 phases were calculated to be 0.3581 nm and 0.3577 nm, respectively, almost in agreement with that from the XRD result. Inserting the values into the equation of
yielded the lattice mismatch value (
) of 0.11%. Such the lower lattice mismatch contributed to a lower elastic-misfit energy barrier for high-density nucleation [
21]. Meanwhile, the sluggish diffusion rates of solutes in multicomponent matrix and the lower interfacial energy (coherent interfaces of FCC/L1
2 phases) suppressed the kinetic and thermodynamic driving forces of Ostwald ripening, respectively [
22,
23]. These were jointly responsible for the high-density yet nanoscale of the intragranular L1
2 precipitates. In contrast to the HA-30 sample, almost none of the intragranular L1
2 precipitates were detected in the HA-3 sample, probably due to the very short annealing duration. This rationalizes the phenomenon of insignificant peak overlap in the XRD pattern of this sample.
Figure 4.
(a) The BF image showing a cluster of precipitates, which were identified as M23C6 carbides and L12 phase, respectively, by the corresponding SAED patterns (b). (c) Elemental distributions of the M23C6 carbides and L12 phase.
Figure 4.
(a) The BF image showing a cluster of precipitates, which were identified as M23C6 carbides and L12 phase, respectively, by the corresponding SAED patterns (b). (c) Elemental distributions of the M23C6 carbides and L12 phase.
Figure 5.
The TEM observations showing the intragranular L12 precipitates of the HA-30 sample. (a) BF and SAED images. (b) DF pattern. (c) HRTEM and corresponding FFT images.
Figure 5.
The TEM observations showing the intragranular L12 precipitates of the HA-30 sample. (a) BF and SAED images. (b) DF pattern. (c) HRTEM and corresponding FFT images.
EBSD examinations were conducted to uncover the recrystallization microstructure. As shown in
Figure 6a and c, the inverse pole figure (IPF) maps of the present samples both show the {111} texture orientation, corresponding to the orientation peak amplitude of (111) plane in XRD patterns. The HA-3 sample presented a relatively homogenous grain structure consisting of uniformly sized equiaxed-grains (
Figure 6a). Corresponding kernel average misorientation (KAM) distribution mapping (
Figure 6b) shows a high density of residual dislocations almost throughout the microstructure, indicative of the lower degree of recrystallisation. As the annealing time increased to 30 min, the density of residual dislocations decreased significantly, while part of the grains were coarsened and some recrystallized fine-grains formed (
Figure 6c, d). Thus, the HA-30 sample presented a heterogeneous grain structure. In this regard, in addition to the possibly uneven distribution of deformation storage energy, the effect of intergranular precipitation behavior may be the main reason for the inhomogeneous recrystallization [
16,
24]. On the one hand, the particles preferentially precipitated at GB regions provide heterogeneous nucleation sites for recrystallisation. On the other hand, the intergranular precipitates produced Zener pinning forces that inhibit the homogeneous expansion of GBs.
3.3. Deformation Mechanisms
The engineering stress-strain curves of the present samples Detailed deformation substructures were examined via TEM analyses of the fractured samples to reveal the deformation mechanisms. As shown in
Figure 8a, the typical (111) plane slip trace and the high-density dislocation walls (HDDWs) can be observed in the HA-3 sample. This generally indicates the planar slip mechanism, prevailed in FCC alloys with lower stacking fault energy (SFE) [
10].
Figure 8b shows the dislocation tangles in a BF image, where the inserted SAED pattern presents the streaky lines. The deformation-induced stacking faults (SFs) are clearly shown in the corresponding DF image, as marked in
Figure 8c.
Figure 8d and e show the deformation twins (DTs) in the BF and DF images, respectively. The twin-related spots were marked in the corresponding SAED pattern (as shown in the insert in
Figure 8d) taken along the [110] zone axis. The HRTEM image shows the Lomer-Cottrell (L-C) lock that formed by the interactions of SFs and twin boundaries (TBs), as marked in
Figure 8f. The corresponding FFT image shown in the insert presents both the streaky lines and twin-related spots.
Similar to the HA-3 sample, the parallel plane slip trace and HDDWs substructures were also observed in the HA-30 sample (
Figure 9a). A detailed examination of the slip region revealed almost non-existent dislocation pile-up or dislocation loop substructures, although the inserted SAED pattern indicated the presence of L1
2 precipitates (
Figure 9b). From the DF image (
Figure 9c), it seems to observe that there were new interfaces produced in the L1
2 precipitates after deformation making their shapes irregular. Previous reports [
11,
25,
26] have shown that the slip dislocations are usually shearing rather than bypassing the coherent L1
2 precipitates because of the coherent phase interfaces with lower lattice misfit. This is effective in avoiding dislocation accumulation and thus the local stress concentrations.
Figure 9d and e show the SFs in the BF and DF images, respectively. The SFs were further characterized through the HRTEM technology as shown in
Figure 9f. The inserted FFT pattern taken from the marked region presents both the streaky lines and faint superlattice reflection spots. This indicates that the SFs can also shear the L1
2 precipitates. Noticeably, in contrast to the HA-3 sample, none of the DTs were detected upon the detailed SAED examinations for the HA-30 sample.
Based on the above observations, it is revealed that the deformation of both the HA-3 and HA-30 samples is mainly dominated by the plane slip mechanism. One major difference is that the SFs and DTs play a mediated role in the former, while only the SFs in the latter. Previous reports [
4] have shown that the lower SFE can facilitate the full dislocations to tend to plane slip or dissociate into two Shockley partial dislocations, reducing the propensity of cross slip. Referring to other studies [
10,
11], the SFEs of equiatomic CoCrNi-base MEAs have generally been evaluated as 20~30 mJ/m
2. It is predicted that the present (CoCrNi)
93.5Al
3Ti
3C
0.5 MEA in this work has also a lower SFE, contributing to the plane slip mechanism. In addition, it has been shown that the solute of C element can also promote the plane slip mechanism, as the enhanced lattice friction results in the dislocation movement slowing down [
12]. For the HA-30 sample, there was an additional factor, i.e., the high density of intergranular L1
2 precipitates inducing the slip plane softening, contributing to the plane slip mechanism [
34]. Specifically, the L1
2 precipitates acting as obstacles to dislocation slip that is energetically favorable for the formation of dislocation pairs [
35]. As the first dislocation shears through the L1
2 ordered domain, that will produce high-energy bonds triggering the increased system energy; with the entry of the second dislocation with a same Burgers vector to the first one, the bonding within the ordered region will be restored and thus the system energy reduced, so it is energetically favorable [
36]. After several of dislocation pairs shear through the L1
2 ordered domain, the resistance to dislocation movement will be reduced, and the subsequent shearing will not need the slip of dislocation pairs. As a result, the slip planes will be softened and thus promoting the plane slip mechanism.
Another point worth noting is that the absence of DTs in the HA-30 sample is also related to the intergranular L1
2 precipitates. It is established that the formation of DTs is closely associated with the SFs extension behavior. Previous studies [
37] reported that with the SFs shearing the ordered L1
2 precipitates, the complex SFs (CSFs) with higher energy will be produced in their wake. Eliminating the CSFs to form twins requires the diffusion-mediated reordering within the L1
2 ordered domain, as revealed in the Ni-base superalloys [
38]. However, this was apparently almost impossible to achieve at room temperature, and thus the formation of DTs was inhibited. In addition, it has been shown that the high density of L1
2 precipitates narrowing the matrix channels and thus enhancing the critical twinning stress (
) is also an important factor in suppressing the twinning behavior. The corresponding theorical equation as expressed below [
10,
36]:
Where
is constant reflecting the dislocation character (0.5 [
36]),
is the shear modulus (88.7 GPa [
37]),
(measured from the XRD result) is the Burgers vector of Shockley partial dislocations,
is the stacking fault energy (~22 mJ/m
2 [
10], taken from the value of CoCrNi MEA for simplicity),
is the lattice friction stress (51 MPa [
37]),
is the effective length of a twinning source,
and
represent the volume fraction and average diameter of the intragranular L1
2 precipitates (measured on average from multiple DF-TEM images). The critical shear stress and normal stress for twinning in the HA-30 sample were calculated to be 724 MPa and 2413 MPa, respectively. The latter is far higher than the sample’s UTS (1070 MPa), which can rationalize the absence of DTs upon deformation to fracture.