Review: Multi-principal element alloys by additive manufacturing

Since processing defects such as porosity, cracks, and element segregation are the main reasons for inferior mechanical properties of AM fabricated parts, determining how to optimize the AM process to fabricate high quality and fully dense parts with no or minimal processing defects is a high priority for AM research. However, up until now most MPEAs fabricated by AM still suffer from various processing defects [99‐103]. In addition to MPEA powder feedstock, the AM processing parameters are considered as the main factors controlling the densification process of AM MPEAs. For instance, the relationship between the volume energy density (VED) and the relative density of LPBF FeCoNiCuAl was investigated and established by Ren et al. [104]. It was found that poor densification was generally caused by incomplete melting with low VED and keyhole formation with high VED, both leading to the formation of pores in the fabricated MPEA parts.

Kuzminova et al. [105] investigated CoCrNiFe MPEAs fabricated using a range of different VED, as shown in Fig. 1. The results showed that most porosity can be eliminated at a VED of 250 J/mm³ while the porosity can reach 11.8% using a high VED of 1150 J/mm³. Similarly, Niu et al. [106, 107] fabricated AlCoCrNiFe and CoCrNi MPEAs by LPBF and investigated their densification process during LPBF. It was shown that the highest relative density of AlCoCrNiFe and CoCrNi can reach 98.4% and 98.9% at a VED of 97.2 and 83.3 J/mm³, respectively. As for CoCrNiFeMn MPEAs, Jie et al. [108] fabricated a series of MPEA samples with more than 99.5% relative density at a VED in the range 62.5–115.6 J/mm³. It can be concluded that the optimized AM processing parameters for MPEAs with different alloy compositions vary greatly, even when there is only a minor difference in one element. Thus, careful work needs to be carried out for each individual MPEA to obtain its own and unique optimized AM processing window.

Even though VED to some extent can be an effective criterion for optimizing the LPBF process, due to the complexity of LPBF, other influential processing parameters should also be considered, e.g., the diameter of the laser beam and the scanning strategy. Lin et al. [109] fabricated CoCrNiFe MPEAs using different combinations of LPBF parameters with the same VED of 170 J/mm³ (Fig. 2a–d). The results showed very different microstructures as well as porosity level and confirmed that the densification is not simply correlated to VED. At the same time, they proposed a multiple regression model (Fig. 2f, g) based on polynomial regression analysis to accurately illustrate how the densification varies with multiple parameter combinations and validated its practicability by successfully fabricating a fully dense MPEAs part with a relative density of 99.71% (Fig. 2e). In addition, Zhou et al. [110] presented a model to predict the proper LPBF parameters for fabricating high quality MPEAs parts. The model uses the physical characteristics of the materials such as laser absorptivity, latent heat of melting, and specific heat capacity to calculate energy absorption (Q_a) and energy consumption (Q_c), and it predicts that an ideal part can be built when Q_a/Q_c has a value between 3 and 8. Based on this model, they successfully fabricated a CoCrNiFeMn sample with 99.9% relative density.

Cracks are another common processing defect in AM MPEAs [106]. While porosity in PBF is mainly caused by incomplete or excessive melting process, cracks are believed to be generally caused by thermal stresses and element segregation. For instance, Sun et al. [100] investigated the mechanisms for hot tearing in CoCrNiFe fabricated by various LPBF process parameters and scanning strategies. Intergranular hot cracks were found in all the samples with no element segregation, revealing that residual stresses caused by the coarse grain size were the major reason for hot crack formation.

MPEAs fabricated by DED technology are more commonly used in coating applications; thus, there are few studies reported on the actual densification process while most of previous studies are focused on the formation of processing defects during DED [86, 111, 112]. For example, Henrik et al. [113] fabricated TiZrNbTa MPEAs by DED and found that the proper feeding speed of nozzle (100 mm/min) can lead to high quality MPEAs samples. Moreover, Modupeola et al. [114] investigated the influence of processing parameters on the microstructure of DED fabricated AlTiCrFeCoNi and AlCoCrFeNiCu MPEAs. They indicate that using a laser power of 1200–1600 W and a scanning speed of 8–12 mm/s can result in reduced processing defects than other parameters. Meanwhile, they also applied a preheating strategy with a temperature of 400 °C and demonstrated its effect on the reduction of cracks through eliminating residual stresses during the DED process.

In general, the phase composition of AM produced MPEAs correspond well to the composition of the same MPEAs produced by conventional manufacturing processes, e.g., casting. For instance, Brif et al. [49] investigated the FeCoCrNi MPEA fabricated by LPBF and found that only FCC solid solution was produced, which is consistent with the phase formation in the previously reported FeCoCrNi MPEAs fabricated by induction and arc melting [115‐118].

However, for Al containing MPEAs, phase transitions have been found to be strongly related to the AM processes [51]. Prealloyed AlCoCrFeNi powder with A₂ (disordered BCC phase) + B₂ (ordered BCC phase) structures was used for LPBF processing by Niu et al. [107]. It was found that the as-built AlCoCrFeNi kept the A₂ + B₂ phase structure, and the B₂ phase was first formed at the boundary of the melt pools where a higher cooling rate existed. From the XRD profiles and phase composition maps shown in Fig. 3, the amount of A₂ phase decreased and the amount off B₂ phase increased with increasing VED [82, 119]. Additionally, composition variation also affects the phase formation and evolution in AM MPEAs. For instance, Sun et al. [120] produced Al_xCoCrFeNi with LPBF and found a FCC to BCC/B₂ + FCC and then to BCC/B₂ transition with increasing Al content from 0 to 1%, and all BCC/B₂ phases are emerging and located in the interdendritic and grain boundaries area. Li et al. [121] also reported a phase formation in DED fabricated (CrMnFeCoNi)_1-xFe_x, a BCC phase emerged when added 60% Fe into equal atomic ratio CrMnFeCoNi MPEAs. More interestingly, a phase transition from BCC to FCC was found in LPBF AlCrCuFeNi_x by Luo et al. [53]. It is reported that microstructure consisted of a B2 phase matrix and disordered BCC phase in AlCrCuFeNi_1.0 (Fig. 4a, b). Increasing the Ni ratio from 1.0 to 3.5 caused a lamellar FCC phase to form, while the relative fraction of basket wave and B2 matrix decreased and transformed into a spherical morphology (Fig. 4c, d).

It is known that the layout and morphology of the melt pool can affect the solidification microstructure in the AM MPEAs. Piglione et al. [122] carried out a detailed study of the solidification microstructure in LPBF fabricated CoCrNiFeMn MPEAs in both single layer and bulk samples. Solidification in the single layer occurred via the growth of grains and a fine cellular structure along the direction of maximum heat flux, i.e., from the existing grains in the substrate toward the center of the single layer top surface (Fig. 5a–c). For the bulk sample, however, columnar grains formed primarily from the substrate and developed via growth competition among different directions (Fig. 5d–f). It was concluded that highly directional heat flux promoted columnar growth of the solidifying grains that followed the thermal gradient and restrained other growth directions by multiple remelting processes.

AM processing parameters impact not only the processing defects but also the solidification microstructure [114, 123]. Tong et al. [124] uncovered the influence of DED processing parameters on the solidification of CoCrNiFeMn MPEAs. The microstructural feature in the vertical section is dendritic columnar grains which become coarser with increasing laser power (Fig. 6a–c). Since the dendritic columnar grains grow parallel with the build direction, the horizontal section of the columnar grains showed and equiaxed morphology with the dimensions increasing with higher laser power (Fig. 6d–f).

The influence of processing parameters on the microstructure of LPBF fabricated CoCrFeNiTi was reported by Takafumi et al. [93], with a focus on the morphology of a single track scan. The results showed that lower laser power, higher scanning speed and smaller laser spot diameter would cause a discontinuous melt track. Moreover, a surface texture parameter (S_a) was introduced to evaluate the correlation between the crystallographic texture and the LPBF processing parameters. The S_a value decreased with increased scanning speed and power and the surface roughness was improved for a given VED value, as shown in Fig. 7a. It was also shown that the solidification microstructure was largely dependent on the VED. As VED increased from 20.8 to 83.3 and then to 333 J/mm³, the depth and width of the melt pool trace observed from both planes showed a conspicuous difference (Fig. 7b–g). Regarding the grain morphology, higher VED allowed laser energy to transfer through more layers to promote the growth of columnar grains (Fig. 7h–j).

Cellular microstructures are a common feature in AM MPEAs, and they result from high-density dislocation tangling and/or element segregation during the rapid solidification process [51, 125, 126]. Yang et al. [127] reported a cellular microstructure in LPBF fabricated CrFeNiMn MPEAs. Equiaxed and elongated cells can be seen inside the grains in Fig. 8a, b. Moreover, an ultra-fine sub-cell structure formed inside the cells (Fig. 8c). The TEM results showed that high-density dislocations were concentrated on the cell walls, as shown in Fig. 8d. The formation of the ultra-fine sub-cells was mainly due to the dislocation tangling inside the cells, which is seen in Fig. 8f. EDS results in Fig. 8e–g revealed the segregation of Mn and Ni in the concentrated dislocations areas, as well as in the dendritic, cell boundaries and even sub-cell boundaries. The formation of cellular microstructure occurs in this case when elements with lower melting point (e.g., Mn and Ni) enrich in the solid–liquid interface which has a slower solidification rate than regions enriched with high melting point elements (e.g., Cr and Fe).

Anisotropic microstructures in AM MPEAs generally arise from the combined effects of directional global and local heat flux, build orientation, and scanning strategy [128‐130]. The microstructure of an LPBF fabricated part can be approximated as a continuous overlay of single melt pools. As such, an unsymmetrical melt pool can cause different solidification behavior/processes along different directions. For example, Jeong et al. [131] reported an LPBF produced (CoCrFeMnNi)₉₉C₁ MPEA. It has an anisotropic solidification microstructure in YZ, XZ, and XY planes caused by applying a bi-directional scanning strategy (Fig. 9a–c). Also, grains of the LPBF produced (CoCrFeMnNi)₉₉C₁ show anisotropic growth because of heat flux distribution in the melt pool (Fig. 9d, e).

Heat flux was demonstrated to play a key role in determining the growth direction of both grains and internal cells in a single melt pool [84, 92, 114]. It is known that by changing the scanning direction a different heat flux with different directions can be obtained in the melt pool of the fabricated part. Following this, the apparent anisotropy in both crystallographic texture and grain morphology as a function of different scanning strategies was extensively investigated in the past few years. For example, Bogdan et al. [132] assessed the microstructure of LPBF fabricated CoNiCrFeMn with various scanning strategies. One conclusion which was drawn from the study was that applying a scanning strategy with 67° rotation can contribute to higher relatively density (about 99.7%) than that with 0° and 90° (about 99.2%). It was also demonstrated that the grain morphology and crystallographic texture were highly consistent with meander and cheeseboard scanning strategies, especially for the equiaxed grain aligned in the Z direction (Fig. 10a–d). A sample with 67^o rotation between two consecutive layers using a chessboard strategy (Fig. 10e, f) produced a spiral grain structure (Fig. 10g, h), which further confirmed that the scan strategies could be used to design and control the crystallographic texture and morphology of the solidification grain structure.

Additionally, the chemical composition and distribution dramatically impact the solidification microstructure of AM MPEAs [53]. With the advantage of DED in producing compositional graded MPEAs, Borkar et al. [133] reported the microstructure of Al_xCrCuFeNi₂ by DED, as shown in Fig. 11. A typical dendritic microstructure was observed in the Al-free MPEA and particles distributed on the cellular boundaries were confirmed to be Cu segregation. A needle-like secondary phase structure with dark contrast was revealed in DED fabricated Al_xCrCuFeNi₂ with x = 0.8. Upon increasing the Al content, fine equiaxed grains with intragranular BCC/B₂ phase and intergranular FCC phase were formed and the relative amount of BCC/B₂ phase continued to increase until x = 1.5.

Some progress has been made in AM processing of refractory MPEAs [43, 68, 134] in the past five years. For instance, Takuya et al. [85] developed a structural TiNbTaZrMo BioHEA by LPBF (Fig. 12a) and validated its suitability for biomedical applications. A BCC solid solution structure was generated and found to have an anisotropic grain distribution corresponding to the melt pool tracks. Fine grains clustered along the boundary of the melt track while coarse grains centralized in the melt track on the XY plane (Fig. 12c). It was also shown that in the YZ plane, columnar grains along the cross section of the melt pool were caused by a 90° rotation scanning strategy (Fig. 12d). Fine equiaxed grains were formed at the bottom of melt pool and columnar grains grew toward the melt pool center (Fig. 12e–g). Compared with a conventionally cast sample, the elemental distribution in LPBF fabricated samples was relatively homogeneous, which was mainly attributed to the fact that a high cooling rate during AM can facilitate the formation of a solid solution structure with reduced element segregation (Fig. 12h–k).

During AM of MPEAs, high heating and cooling rates as well as cyclic thermal profiles (i.e., cyclic heating and cooling) can generate microstructures that are very distinct from those obtained via traditional manufacturing processes, in particular the heterogeneous and meta-stable microstructure in AM MPEAs. Unfortunately, the cyclic thermal profiles along with high heating and cooling rates during AM are quite complex phenomena to understand experimentally. This is mainly because in situ microstructural observation during AM is still quite challenging although more and more advanced in situ monitoring systems and techniques for AM have been developed recently. In recent years, great efforts have been made to uncover these above-mentioned unclear phenomena and some studies about the heterogeneity of microstructures of MPEAs produced by AM were recently reported [125, 128, 135]. For example, Wang et al. [136] investigated the microstructural evolution of LPBF fabricated CoCrFeMnNi with respect to cyclic and rapid thermal profiles. In their study, five specimens from the built part along the building direction (Fig. 13a) were selected. It was found that nanograins with average size ranging from several nanometers to tens of nanometers (i.e., 50 to 70 nm) were observed in the top region I and II, respectively (Fig. 13b–e). In region III, grains of several micrometers in size containing a random distribution of dislocations were observed (Fig. 13f, g). The microstructure in region IV contained elongated grains of ~ 3.4 μm in length and 1.3 μm in width. Cellular substructures with high dislocation density were also observed within the grains in region IV (Fig. 13h). Fine twins with 30 nm thickness were also observed to contain a random distribution of dislocations rather than a cellular substructure (Fig. 13i). In region V, there was an increase in the quantity of stacking faults and thinner nano-twinned structures while the cellular dislocation substructure in these regions decreased (Fig. 13j). Moreover, an HCP phase was observed in Fig. 13k, revealing that an FCC to HCP phase transition occurred at the bottom of the sample. The atom probe tomography (APT) data, as shown in Fig. 13l, exemplified the segregation of Mn and Ni in dislocation cell walls, which can be elucidated by precipitation order and extent of partitioning of elements during solidification [137].

In addition to the LPBF fabricated MPEAs, both Cai et al. [64, 138] and Shou et al. [64, 138] fabricated an FeCoCrNi part using DED, which exhibited a heterogeneous microstructure consisting of a solidification grain structure with columnar grains at the bottom and equiaxed grains at the top of the fabricated MPEAs, respectively. Moreover, the composition gradient generated by the DED process can also lead to a heterogeneous microstructure. Henrik et al. [113] fabricated refractory MPEAs by DED with a composition varying from Ti₂₃Zr₀Nb₄₂Ta₃₅ to Ti₂₃Zr₄₃Nb₀Ta₃₄ (Fig. 14a). This composition change was accompanied by a change in microstructure from dendrites to fine grains in the matrix (Fig. 14b–k), with EBSD mapping revealing a decrease in average grain size from 60 to ~ 2 μm (Fig. 14l). Overall, the heterogeneous microstructures exhibited by many MPEAs suggest that process parameter adjustment would be required during the build to achieve a homogeneous microstructure. Or, alternatively, there is a great opportunity to fabricate graded structures with location specific mechanical properties, for example harder at the surface and tougher at the subsurface.