Co-existence of nanoprecipitates with solute nitrogen evades the strength-ductility trade-off in metastable high entropy alloy

N and Ti co-doped Fe45Mn35Co10Cr10 alloys were prepared by Laser-Directed Energy Deposition (L-DED). We find that precipitate reinforcement of TiN particles introduced by harnessing N solid solution can result in a microstructure featuring attached MnO and TiN composite nanoparticles. Co-doped samples exhibit a notable 69% increase in yield strength compared with that of the free-doped, rising from 310 MPa to 533 MPa, maintaining 44.5% elongation. Even at 77 K, the co-doped alloy displays ultra-high strength, approximately 1400 MPa. These findings on metastable high entropy alloys through additive manufacturing open new avenues for their application in various fields. GRAPHICAL ABSTRACT

to the pre-consumption of γ -transformation-induced plasticity and the inevitable formation of MnO with monoclinic structure.On the other hand, HEAs composed of face-centered cubic (FCC) phase, encounter limitations of insufficient yield strength (less than 500 MPa).Recent efforts to enhance the yield strength of these alloys have primarily centered on the introduction of interstitial solutes (boron [8], carbon [9], nitrogen [10]) and the incorporation of strengthening particles.In all these cases, the common factor contributing to reduced plasticity is the accumulation of particles at grain boundaries caused by the affinity of interstitial elements for grain boundaries and low-melting-point particles [11][12][13].Strength-ductility trade-off is still a key issue for non-metallic element interstitial-strengthened high entropy alloys.
Titanium nitride (TiN) is recognized for its high melting point of 2950°C, outstanding chemical stability at elevated temperatures, and strong affinity for metals, which makes it a promising candidate for reinforcing metal matrix composites with particles [14].However, a common challenge encountered when using mechanically mixed method to prepare particles-reinforced metals is agglomeration tendency and limited size [14,15].Conversely, the in-situ reaction can potentially obtain fine size reinforcement and composites with clean interface and high interface bonding strength [16,17].
In this study, nitrogen (N) and titanium (Ti) co-doped Fe 45 Mn 35 Co 10 Cr 10 powders by vaporization in gas atomization.Single element metal (Fe, Mn, Co, Cr, Ti)with purity ≥ 99.9% is used as the raw material.The atomization process involved a mixture of nitrogen and argon gases to introduce nitrogen into the powder which shown in Figure 1(a).Then in-situ TiN particles could be synthesized in the powder because Ti atom has strong affinity with N atoms.The atomic percentage of N (0.74%) is higher than that of Ti (0.09%), so that excess nitrogen acts as solute.Powders with particle sizes ranging from 53 to 150 μm were carefully selected for Laser-Directed Energy Deposition (L-DED), the 60mm × 30mm × 30 mm sample were obtained.For comparison purposes, we also produced undoped Fe 45 Mn 35 Co 10 Cr 10 powder using the same method, with pure argon as the atomization gas.The precise chemical composition of these two powders was determined using Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES, PerkinElmer 8300), and the results are detailed in Table 1.
Sample preparation and testing are provided by supplementary materials.The mechanical properties at 293 K are depicted in the form of engineering stress-strain curves (Figure 1c).The co-doped samples exhibit an ultimate tensile strength of 752 MPa, a yield strength of 523 MPa, and an elongation of 44.5%.In comparison, the free-doped samples exhibit a yield strength of approximately 310 MPa.This represents a 69% increase in yield strength for the co-doped samples while maintaining an equivalent level of elongation.Notably, the co-doped samples exhibit a superior combination of strength and ductility when compared to the reference FeMnCoCr alloys [7,[18][19][20], which is listed in Table 2.At lower temperatures of 77 K, the co-doped alloy exhibits ultra-high strength, reaching approximately 1400 MPa, with a yield strength of 1021 MPa.Compared to the continuous work hardening process observed in the free-doped material, the co-doped material exhibits a distinct yield point.To thoroughly comprehend the distinction in mechanical properties between the free-doped and codoped alloys, an extensive and meticulous microscopic analysis was conducted(experimental methods in supplementary materials).Figure 1(e) illustrates the average GND density in different phases for Co-doped and Freedoped samples under various conditions.The lattice's net curvature reflects its ability to store random dislocations.In the undeformed samples, Co-doped exhibits a higher density of stored random dislocations compared to Free-doped.This may be attributed to the formation of new precipitate phases, leading to the generation of more dislocations during the shrinkage process under residual stress.For the room temperature fractured samples, the higher GND density in Co-doped indicates its superior ability to nucleate and propagate dislocations, implying the presence of more mobile dislocations for enhanced strain hardening.This higher density of dislocations contributes to its increased strength at room temperature.However, under 77 K conditions, the GND density in the fractured Co-doped shows an abnormal decrease, significantly lower than that of the Free-doped.This indicates a significant reduction in the deformation capacity of Co-doped at low temperatures, leading to premature failure.The decrease in GND density suggests a difficulty in accommodating large strain distributions through dislocation nucleation and slip, with plastic deformation behavior primarily associated with martensite/twin phase transformation.
The initial HCP phase fraction in the co-doped alloy is remarkably low, amounting to just 1%, indicating minimal phase transition during the L-DED process (Figure 1f 1 ).However, upon reaching the fracture strain of 44.5%, there is a slight increase in the fraction of HCP phase (28%), which is significantly lower than the 80% observed in the fractured free-doped HEAs (Figure 1f 2 and g 2 ).This discrepancy can be attributed to the solid solution of N, which enhances the stacking fault energy (SFE) of the matrix, thus improving phase stability and inhibiting the transition from FCC to HCP.The SFE can be calculated using the relationship between SFE and stacking fault probability established by Reed [21] et al.The calculated average SFE of the co-doped material is 20.2 mJ/m 2 (see supplementary materials).Previous researches [22][23][24] have indicated that the value of SFE determines the dominant plastic deformation mechanism, whether it is transformation-induced plasticity (TRIP, SFE < 20 mJ/m 2 ) or twinning-induced plasticity (TWIP, SFE > 20 mJ/m 2 ).In contrast, the SFE of the free-doped material is only 3.1 mJ/m 2 , suggesting that its plastic deformation mechanism is predominantly TRIP.It is worth noting that the TRIP effect at 77 K is lower than 293 K.The main reason is that the existence of short-range ordered structure (SRO, in supplementary materials) reduces the martensite nucleation rate.
Further ECCI analysis is conducted on the microstructure evolution of the co-doped during deformation process.The presence of slip bands, as observed in Figure 2(a), suggests that plane slip becomes the predominant mode of dislocation movement in the current alloy.Figure 2(a) also reveals Taylor lattice (TL), which is a kind of low energy dislocation structure.It is reported that this structure occurs when a single slip system is activated [25][26][27].By contrast, Taylor lattice was not observed in the base sample.The presence of TL restricts the increment of dislocations, resulting in minimal growth of dislocation.Furthermore, at a strain of 44.5%, the sample exhibits primary twins and secondary twins (Figure 2b).This observation aligns with a previous report [28], where the fraction of primary twins was found to increase almost linearly with decreasing SFE (from 30 mJ/m 2 to 20 mJ/m 2 ), while secondary nano-twins were more frequently at SFE levels less than 20 mJ/m 2 .For comparison, the free-doped sample with 3.1 mJ/m 2 stacking fault energy is more prone to martensite and martensite variants.This increase in twin density results in a significant rise in dislocation.Consequently, twins and secondary twins become the primary mechanisms of plastic deformation, while dislocation slip and phase transformation are auxiliary.Accordingly, the codoped alloy keeps the ductility to 44.5%.Concurrently, the dynamic Hall-Petch effect caused by secondary twins and TL contributes to the strength.Grain size and cellular structure information are depicted in Figure 3. Low magnification EBSD images reveal distinct grain growth tendencies in the horizontal cross-section (Figure 3a) along the [110] direction [29].This phenomenon can be attributed to the heat flow spreading horizontally in the as-built structure, creating a higher temperature gradient.The average grain size, statistically determined from Figure 3(a), measures 53.3 μm.Notably, clear molten pool traces are evident in Figure 3(b), illustrating the typical heterogeneous microstructure characteristics associated with L-DED in metal processing.The heat-affected zone exhibits fine crystals due to the high energy density at the center of the laser beam and the thermal cycle effect, resulting in recrystallization.Conversely, the high temperature gradient in the BD direction leads to the formation of columnar crystals extending from the molten boundary to the molten central zone.Interestingly, the consistent heterogeneous microstructure features, with an approximate grain size of 56.4 μm, are also observed in the free-doped sample.This demonstrates that the introduction of N and Ti does not induce any microstructural changes in the alloy.Consequently, the grain size effect is ruled out as the reason behind the improved mechanical properties.
Furthermore, Figure 3(c) provides an intricate view of the internal grain regions, exposing the sub-structure morphology induced by cellular features.EDS maps disclosed that the sub-structure's origin lies in the enrichment of Fe within the intracellular region and Mn within the intercellular region, with chemical compositions of Fe 47.2 Mn 32.5 Co 9.8 Cr 10.5 and Fe 41.5 Mn 37.8 Co 10.5 Cr 10.5 , respectively.This discrepancy in Mn content causes stacking fault energy heterogeneity.Mn-rich regions situated cell boundary possess high stacking fault energy, while Fe-rich regions inside cell exhibit low stacking fault energy.The average equivalent circle diameter of the cellinduced sub-structure measures 5.7 ± 1.1 μm.Moreover, a multitude of spherical particles with uniform distribution is observed in the co-doped sample.Initially identified as MnO with average diameter of 225 ± 17 nm, similar to the free-doped HEAs.The area fraction measured by software is 0.8 ± 0.1%.A common hierarchical structure ranging from nanometers to micrometers was observed in both the co-doped and free-doped HEAs (Figure 3c and d).However, it is worth noting that particles in the co-doped were characterized as nano-scale Ti-rich particles.The details comparison of black spots between the co-doped and free-doped is presented in Fig. S3 and Fig. S4.
Details of particles in the co-doped are presented by TEM characterizations (Figure 4).Two distinct nanoparticles are visible, one cubic and the other spherical, as evidenced by the HAADF images (Figure 4a).Based on qualitative assessments, the corresponding EDS maps confirm that these nano-particles are composite structures composed of Mn-rich oxides and Ti-rich nitrides.Further detailed TEM analysis was conducted to ascertain the structure and interfacial relationship of the composite nanoparticles.The Ti-rich nitride was identified as TiN with FCC structure based on the SAED patterns with the beam direction parallel to the [110], [114] and [112] zone axes (Figure 4d-f).On the other hand, the Mn-rich oxides were characterized as MnO with an HCP structure using SAED patterns with the [81 92 ], [10 10] and [12 32] zone axes.It's important to note that MnO in the free-doped alloy exhibits a monoclinic structure.The three sets of SAED patterns confirm that MnO and TiN are in perfect coherence, and the crystal plane spacing of MnO is twice that of TiN, as demonstrated in Figure 4(d).Additionally, Figure 4(c 1 and e 1 ) display WBDF images corresponding to (020) of TiN and (02 21) of MnO, respectively.Due to the significantly higher melting point of TiN compared to MnO, the nucleation of TiN primarily occurs during the powder melting and solidification process.Subsequently, heterogeneous nucleation of MnO takes place on the surface of TiN, resulting in a mixture of composite nanoparticles with MnO attached to TiN.The interface between TiN and matrix is incoherent (Fig. S5), which confirms TiN precipitated directly from the melt during solidification rather than from solid-state phase transformations.
It is worth paying attention to interaction between co-exist nanoprecipitates with twins and martensite during deformation.Figure 5(a) illustrates lath structures composed of 15-70 nm lamellae, which consist of multiple twins and martensite (T/M), as depicted in Figure 2(a).The width of an individual twin measures around 7.7 ± 1.4 nm, with twin boundaries measuring 1.5 ± 0.5 nm.Nano-composite T/M lamellar structure limited the width of martensite, while twinning introduced additional fresh slip systems.There is a noticeable increase of the spindle-like shape (Figure 2a and b), which is a result of void formation due to the disintegration of composite particles from the matrix [30,31], as illustrated in Figure 5  stacking faults and substantial local distortion stemming from the stress field around the particles are observed.Figure 5 demonstrates that the lamellae consisting of T/M spread over the particles along the length direction, indicating that the particles effectively transfer the applied load.
An interesting statistic is that the TiN-matrix interface with a small lattice mismatch of ∼ 0.14% keeps semicoherent.Moreover, the incoherent MnO have orientation relationship with matrix: [10 10] MnO // [110] FCC (Figure 5b) .To sum up, such semicoherent and incoherent co-exist nanoprecipitates offer several advantages in enhancing alloy properties: (I) MnO precipitation reduces oxygen (O) content in the matrix, while high O content have detrimental effects on mechanical properties [32].(II) N preferentially combines with Ti to form TiN, preventing the formation of Cr 2 N which results in low elongation [13].(III) the crystal structure of MnO changes, enabling MnO to maintain its orientation relationship with the matrix.As Ref. [33], specific relationship formation of precipitate with matrix could minimize the strain energy of the system.The interface misfit between MnO and matrix is benefit to storing dislocations.(IV) spherical MnO and square TiN co-exist precipitates strengthen the alloy through the Orowan mechanism.It is proved that the Orowan increments in yield stress produced by prismatic precipitate are invariably larger than spherical particles [34].
The improvement of yield strength is attributed to coexist nanoprecipitates and solute nitrogen.It can be seen from Ref. [35] that the yield strength increase originating from per at.%N is 143.8 MPa in FeMnCoCr.Hence, the calculated increment caused by solid-solution N(0.6 at.%) is ∼ 86 MPa in the co-doped.The total increment of the co-doped is 213 MPa, which comes from the subtraction of 523 MPa in the co-doped sample and 310 MPa in the free-doped sample (Engineering stress-strain curves in Figure 1c).The extra contribution of co-exist nanoprecipitates to yield strength is then estimated to be ∼ 127 MPa.
In conclusion, this study focused on the development of a high-strength and high-ductility FeMnCoCr alloy fabricated by LAAM.Co-doping of N and Ti in Fe 45 Mn 35 Co 10 Cr 10 alloy facilitated the in-situ formation of thermally stable TiN particles through heterogeneous

Figure 3 .
Figure 3.The horizontal cross-section microstructures of the current HEAs.(a) EBSD inverse pole figure (IPF) maps display grain orientation, (b) corresponding grain boundary contrast maps highlight the heterogenous microstructure composed of heat affect zone (HAZ) and molten central zone (MCZ); ECCI images of (c) the co-doped and (d) the free-doped and corresponding EDS maps of Fe, Mn, Ti show sub-structure and particles distribution comparison.BD: build direction, SD: scanning direction.

Figure 4 .
Figure 4. TEM characterizations of the composite nanoparticles.(a) HADDF images and corresponding EDS maps; (b)BF image; (c) SAED pattern with [001] TiN axis in the yellow region marked in (b), and (c 1 ) the corresponding weak beam dark field(WBDF) image of (020) TiN plane; (d, e, f) SAED patterns in the red region marked in (b), here, (d) contains HRTEM microstructure observation of the MnO and TiN interface; (e 1 ) the corresponding WBDF image of (02 21) MnO plan; (g) the HCP structure MnO from the International Centre for Diffraction Data.

Table 2 .
Comparison of yield strength versus elongation obtained in the current materials with the previously reported FeMnCoCr alloys.