Mechanical properties of pulsed electric current sintered CrFeNiMn equiatomic alloy

An equiatomic CrFeNiMn alloy was consolidated using pulsed electric current sintering (PECS) from gas atom-ized (GA) powder. A range of sintering temperatures was applied to determine its impact on the received microstructure and material properties. It was found that the phase structure of the disc shape sintered samples varied greatly depending on the sintering temperature. According to X-ray diffraction (XRD) the surface of the samples contained a noticeable amount of BCC phase while in the middle of the sample cross sections no BCC phase was found. Ball milling of the gas atomized powder prior to sintering for 25 h increased the density and hardness of the sintered sample from 98.3% to 99% and from 200 HV to 300 HV, respectively. At the same time, the ultimate tensile strength increased from 700 MPa to 1000 MPa and the elongation at fracture decreased from 40% to 25%. The enhanced hardness and tensile strength were attributed to the grain refinement caused by milling the powder prior to sintering. Milling of the powder resulted in a reduction in grain size of the sintered material from 5.7 μ m to 1.9 μ m when sintered at 1100 ◦ C for 5 min. The grain refinement was also found to affect the deformation mechanisms. While the deformation of the samples prepared directly from GA powder was accommodated by the cell-forming process and deformation twinning, milling refined the initial microstructure to such an extent that deformation twinning was suppressed. In summary, a single phase equiatomic CrFeNiMn alloy can be achieved by pulsed electric current sintering of gas atomized powder. Sintering at temperatures above 1100 ◦ C results in a single-phase FCC structure excluding the surface of the samples. Milling of the starting powder increases the density and strength and decreases the grain size of the sintered material. Decrease in grain size also suppresses deformation twinning and promotes dislocation cell formation. As the CrFeNiMn alloy system does not contain cobalt as an alloying element it shows great promise to be used in nuclear applications. The CrFeNiMn alloy showed good phase stability during tensile testing by not undergoing phase transformation contrary to metastable austenitic stainless steel where martensite transformations take place.


Introduction
High entropy alloys are a novel single phase metal alloys, which consist of three or more components typically in equiatomic proportions.High entropy alloys were first proposed by Yeh et al. [1] over 17 years ago and CoCrFeNiMn, which was discovered by Cantor et al. [2] is perhaps the most well-known.The alloy was discovered when studying mixtures of 20 elements.It has an equiatomic CoCrFeNiMn composition, but still a single-phase FCC microstructure.Since its discovery CoCrFeNiMn has been investigated extensively [2][3][4][5] including various processing methods.Ji et al. [4] used mechanical alloying (MA) and spark plasma sintering (SPS; also referred to PECS; Pulsed Electric Current Sintering) to produce the CoCrFeNiMn alloy.SPS is typically used for its rapid sintering process while obtaining fully dense bulk material [6].This processing route resulted in a compressive strength of 1987 MPa. Liu et al. [5] showed that mechanical milling of the gas atomized powder can also be used to increase the tensile strength of the equiatomic CoCrFeNiMn produced by SPS.After 10 h of milling the tensile strength increased from 480 MPa to over 1000 MPa.
The most important strengthening factors in metal alloys are solid solution strengthening, precipitation hardening, dislocation strengthening, twinning, phase transformation hardening and grain size hardening.In single phase HEAs the effective factors are typically dislocation strengthening, twinning, and grain size hardening Grain size of the sintered compact can be refined by ball milling of the powder [7].Decrease in grain size increases the strength based on grain boundary strengthening (Hall-Petch effect) [8,9].Hardness and strength can also increase with the presence of precipitates (precipitation hardening) [10].During deformation metallic materials undergo strain hardening [11] enabled by a variety of mechanisms such as dislocation induced cell-formation, deformation twinning and phase transformations [12,13].Twinning is also often observed during annealing and recrystallization in FCC metals [14].Sun et al. [15] studied grain refinement via deformation twinning in equiatomic CoCrFeMnNi alloy and observed that deformation twinning only takes place at a grain size larger than 3 μm, below which the critical twinning stress became too high and inhibited the twinning process.Fu et al. [16] reported that the three major contributors to strengthening of an FCC Co 25 Ni 25 Fe 25 Al 7.5 Cu 17.5 HEA were twin boundary strengthening 4%, dislocation strengthening 33% and grain boundary strengthening 62%.Dislocation cell formation has been reported in HEAs during deformation by Otto et al. [17].CoCrFeNiMn alloy, which was deformed by 20.6% in tensile loading showed dislocation cells ranging from 200 to 300 nm.The cell-forming deformation process is commonly observed for many polycrystalline materials as a result of the grain boundaries inhibiting the motion of dislocations [18,19].During the cell-forming process the accumulated lattice dislocations re-arrange to minimize their total energy by forming dense dislocation walls (DDWs) and dislocation tangles (DTs) inside the grains [20].With increasing strain the dislocations accumulate at the DDWs, transforming them into sub-grain boundaries (SGBs) with a larger misorientation angle [20].At sufficiently high strains the SGBs transform into new grain boundaries and potentially form an extremely fine-grained microstructure, achieved for example by a severe plastic deformation process [21].
One potential area of applications of high strength multielement alloys is in nuclear field, as it has been discovered that the multielement single phase solid solutions possess also an improved resistance to irradiation damage [22].However, alloys consisting cobalt are not attractive because the activation of cobalt to Co 60 under radiation.Therefore, interest is focused on cobalt free alloys [23][24][25].As a possible Co-free alternative the CrFeNiMn-alloy system has been studied to some extent.Wu et al. [23] found that the FeNiMnCr 18 alloy has a single-phase FCC structure and its tensile strength increases as the test temperature decreases, a similar trend has been observed in CrMnFeCoNi [26].Twinning is observed to take place during deformation at temperatures below 77 K. FeNiMnCr 18 has also been studied under exposure to ion irradiation by Kumar et al. [24] The alloy showed good resistance to point defect mobility and resistance to swellings in comparison with conventional austenitic Fe-Ni-Cr-based alloys.Li et al. [25] studied the neutron irradiation response of FeNiMnCr 18 and observed more sluggish diffusion compared to traditional FCC alloys.Elbakhshwan et al. [27] studied the corrosion properties and thermal stability of FeNiMnCr 18 , and they predicted that at an elevated temperature (700 • C) a Cr rich BCC phase is formed.Stepanov et al. [28] have studied Cr-Fe-Ni-Mn in non-equiatomic composition by varying the Cr content.Fe 40 Mn 28- Ni 32-x Cr x (where x = 4, 12,18,24).They obtained a yield strength of 210 MPa for the lowest Cr content and 310 MPa for the highest Cr content with elongations of 71% and 54%, respectively.
Equiatomic CrFeNiMn has been studied very little, because of it being classified as a multiphase alloy by previous investigations, see e.g.Refs.[29,30].Bloomfield et al. [30] homogenized an equiatomic CrMnFeNi at several temperatures ranging between 500 and 1170 • C and none of them yielded a single-phase structure.Recently, gas-atomized powder of the equiatomic CrFeNiMn alloy was made by Lehtonen et al. [31], with a near single phase FCC structure.An equiatomic CrFeNiMn coating having FCC structure was produced from the powder via cold spray process by Lehtonen et al. [32] using nitrogen as the process gas.These results indicate that a single-phase FCC microstructure may be achieved for bulk CrFeNiMn, if the production process is optimized.
In the current work an equiatomic CrFeNiMn high entropy alloy was chosen to be studied, with the aim to optimize the sintering process to achieve a single-phase structure with a high density and good mechanical properties.The influence of process parameters on microstructural characteristics (e.g. grain size) and obtained mechanical properties are determined.Gas atomized powder with a 45-90 μm powder fraction is chosen for this study and it is consolidated using PECS at varying temperatures.The effect of powder milling on the densification and microstructure of CrFeNiMn is also studied.The deformation mechanisms are investigated using TEM, SEM and EBSD, utilizing the recently developed domain misorientation approach [33] for resolving dislocation sub-structures.

Powder preparation
The equiatomic CrFeNiMn powder was made using raw materials with commercial purity (Cr, Fe, Ni > 99.6 wt-%, and Mn 98.9 wt-%).Atomization was carried out at Bremen University, Department of Production Engineering.An Al 2 O 3 crucible was used to melt the alloy under argon.Argon gas atomization was performed at a temperature of 1723 • C with a pressure of 12 bar using a 2.5 mm diameter close-coupled atomizer, a tangential feed pressure of 210 mbar and a temperature of 200 • C. A more detailed description of the process is given in Ref. [30].A powder fraction of 45-90 μm was sieved from the atomized powder and chosen to be used in this study.
Milling of the powder prior to PECS was studied to determine its effect on sintering and mechanical properties of the sintered material.Milling was performed in a planetary ball mill (Fritsch Pulverisette 6) at 250 rpm using a tungsten carbide milling pot and hardened steel balls (Ø6 mm).The powder was milled for 2, 10 or 25 h.One mass-% of stearic acid was used as a process control agent (PCA) to prevent cold welding during milling.

Pulsed electric current sintering (PECS) of the samples
The sieved powder was sintered using a FCT HP D 25-2 unit (FCT, Germany).The powders were compacted in a cylindrical graphite mould having inner diameter of 20.8 mm.Graphite foils with a thickness of 0.4 mm were placed in between the powder and the graphite surfaces.Graphite felt was placed around the mould for insulation and avoiding the thermal gradients.Five 20 mm diameter samples were sintered at 950 • C, 1000 • C, 1050 • C, 1100 • C and 1130 • C to study the effect of temperature on the phase structure and mechanical properties.Three samples were also made by milling the powder for 2, 10 and 25 h prior to sintering at a temperature of 1100 • C.During PECS compaction a heating rate of 100 • C/min was used with a pulse timing of 10/5 ms (pulse/pause).A constant 35 MPa pressure was applied during sintering.All PECS experiments were performed under vacuum (<9 mbar).
Consolidated bulk samples had a thickness of approximately 3 mm and a diameter of 20 mm.They were ground down to P2500 grit SiC paper.Cross sections of the PECS compacts were cut and mounted in Polyfast (Struers), and polished using 5 μm, 1 μm and 0.3 μm aluminium oxides, the final polishing was done with 40 nm colloidal silica using vibratory polishing using Buehler Vibromet 2 to finish the surfaces to facilitate the microstructural investigations.

Tensile sample preparation
In order to prepare tensile samples, compacts having thicknesses of 5 mm and 9 mm were made using a 40 mm diameter mould.The first specimen (PECS 40) was made directly of the sieved powder and was sintered at 1100 • C for 5 min.For the second sample (PECS 40 M) the powder was milled for 25 h and sintered at 1050 • C for 1 min.Flat tensile samples with a total length of 40 mm, a gauge length of 17.5 mm (PECS40 M) or 20 mm (PECS40) was used with thickness of 3.5 mm and width of 5 mm, were then cut and machined out of the disks.A minimum of 1 mm was removed from all surfaces to eliminate the BCC phase.Tensile samples were cut lengthwise perpendicular to compression direction of the PECS process.Reference tensile test samples were made from a 3 mm thick plate of the commercial AISI 316 stainless steel.Samples of similar size were made in order to compare tensile properties to a known material with non-standard tensile samples.The hardness of the commercial stainless steel tensile samples was 199 HV1 with a standard deviation of 9.2HV1.A summary of the sintered samples is given in Table 1.

Characterization of the samples
Density of the compacted samples was measured using Archimedes method, i.e., by weighing the sample in air and water.Vicker's hardness was measured according to the standard SFS-EN-ISO 6507-1:2018 using Innovatest Nexus 4303 Vickers indenter under a load of 1 kg.X-ray diffraction (XRD) was performed to identify the phase structures applying Panalytical Xpert Pro Powder XRD equipment and Cu-K α -radiation.Scanning electron microscopy (SEM) was performed using Tescan Mira 3 with an energy-dispersive X-ray spectrometer (EDS) made by Thermo Fisher Scientific.Tensile tests were performed under strain rate of 0.002 s −1 using an Instron 4204 Universal tester with a 50 kN load cell.Sample extension was measured by an optical extensometer or from jaw displacement.A Fischer FMP30 Feritescope was used to check the tensile samples for ferromagnetic phases after tensile testing.Transmission electron microscopy (TEM) was performed using JEOL JEM-2800.Disks of 3 mm in diameter were cut for the preparation of TEM samples.Final thinning of the samples by electropolishing from a thickness of 100 μm was carried using a Struers Tenupol-5 with a voltage of 9 V and a temperature of −17 • C in a solution of ethanol and nitric acid (vol%-70/30).
Electron backscatter diffraction (EBSD) analyses were performed using an Oxford Instruments, Channel 5 attached to a Zeiss Ultra 55 FEG-SEM on samples vibratory polished for a minimum duration of 16 h.Grain size was calculated using EBSD images, and grain boundaries were analyzed using fast multiscale clustering (FMC) [34] and MTEX version 5.3 [35,36] in MATLAB.The grain boundary segmentation parameter C Maha was set at 1.0 for grain size analyses.Calculation was carried out with MTEX from EBSD maps using the line and point-sampled intercept length methods as described by Lehto et al. [37,38].The methodology is available at Refs. [39,40].For each sample an area having at least 2000 individual grains in minimum was included.Fraction of twin boundaries was analyzed based on grain boundary misorientation angles that corresponded to twin boundaries using MTEX in MATLAB.Twin boundaries were defined as having a misorientation angle of greater than 57 • and at most 10 • of separation from the (111) axis.
Deformation induced dislocation sub-structures were analyzed from EBSD data using the domain misorientation approach developed by Lehto [33], available as open source at Refs. [41,42].The methodology is tailored for capturing the deformation induced grain sub-structures for the cell-forming deformation process.The measurement principle utilizes measurement domains that are grown radially until a specified misorientation value has been reached.This enables stochastic analysis of local misorientation to be carried out within individual grain sub-structures.To characterize the plastic deformation state of the materials, the stages of grain sub-structural evolution must be measured.
Using the domain misorientation approach this can be carried out by measuring boundaries with different misorientation criteria.The misorientation across the DDWs is typically in the range of 0.5-0.8• [20].With continued deformation the DDWs transform into sub-grain boundaries (SGBs) with a larger misorientation in the range of 2-5 • [20].In the current study the dense dislocation walls (DDW) were measured using misorientation values of Δθ DDW = 0.5 • and Δθ DDW = 1.0 • and the domain misorientation approach.For sub-grain boundaries (SGB) the misorientation value Δθ SGB = 2.0 • was used.To analyze the degree of grain size refinement, the grain boundaries were defined using a misorientation angle threshold because the fast multiscale clustering (FMC) is prone to over-segmentation when deformation induced grain sub-structures are present.The grain boundaries were defined by a misorientation of at least 2 • between adjacent datapoints, yielding similar average grain size values for the undeformed specimens as FMC.
The initial kernel size varied between 60 and 140 nearest neighbors according to the average grain size and step size of the EBSD maps.The domain misorientation is also used to estimate the size distribution of the sub-structures [33].The analysis was carried out using MTEX version 5.3 [43].The EBSD data was post-processed to reduce the spatially independent measurement noise of the Hough-based EBSD using the Half-Quadratic filtering available in MTEX; see Table 2.The deformation analysis was carried out for the non-milled and milled tensile samples in three different conditions: 1) As-sintered before deformation (reference), 2) location near the fracture surface after the tensile test (ultimate strain, <2 mm from fracture surface)), and 3) location far away from the fracture surface (high strain, 5 mm from fracture surface) to map the evolution of dislocation sub-structures.Both strained samples are located at the midline of the specimen, with ultimate strain sample located in the necked region approximately 1 mm away from the fracture surface.The highly strained location is approximately 5 mm from the fracture surface, located near the edge of the tensile specimen's constant width gauge length.

Density
Porosity is detrimental to the properties of the HEA's, and thus the first step of process parameter optimization is to achieve a high relative density for the PECS manufacturing process.Fig. 1 shows the evolution of the density as the sintering temperature is increased while sintering time is kept constant at 5 min.It is clear that an increase of sintering temperature increases the final density of the bulk material.Above 1050 • C the density increase has a knee-point and a plateau beyond it.At a temperature of 1130 • C partial melting of the samples occurred during sintering and a small amount of material squeezed out of the mould.The density of the 40 mm compacts used for making the tensile specimens is also shown in Fig. 1.The density of the two samples made at 1100 • C (PECS1100 and PECS40) show that the larger 40 mm mould does not result in a lower density in the sample.The density of the PECS40 M (blue circle) has a slightly higher density than the PECS1050 sample which is non-milled.The inset in Fig. 1 shows the effect of milling time on density at a sintering temperature of 1100 • C for 5 min.It is noted that milling for more than 10 h resulted only in a minor increase in densification.The improvement in densification is caused by the increased surface area of the powder, which in turn increases the surface

General microstructure and chemical composition
As the density of the material depends on the sintering parameters such as heat cycle and pressure as well as possible powder activation by milling, the microstructure of the sintered materials is influenced likewise.Fig. 2a shows the microstructure in the cross section of bulk CrFeNiMn sintered at 1100 • C. The thin darker layer within about 100 μm from the sample surface (top of the image) shows the presence of a second phase, which was identified to have a BCC structure by XRD.The BCC phase is likely formed when the carbon from graphite used in the PECS molds or in the graphite paper, reacts with the outer layer of the powder.This results in the formation of a Cr-rich BCC phase.The composition of this phase as measured by the EDS is shown in Table 3.The chromium weight fraction in the BCC phase is significantly higher compared to the other elements shown in Table 3.According to the XRD the lattice parameter of the phase is approximately 2.88 Å suggesting    that the phase is Cr-rich ferrite.
The microstructure near the surface of the compacts is very fine with a duplex microstructure.Below this surface layer the FCC grains are noticeably larger down to a depth of approximately 200 μm, after which becoming significantly finer again.In the bulk FCC region the presence of annealing twins is frequent, see Fig. 2b where several twin boundaries are visible inside a single grain.
To determine the phase structure achieved with different processing parameters, an XRD analysis was carried out.XRD of the flat surface of the PECS-compacts show a mixture of FCC and BCC near the surface in an approximately 200 μm thick layer, followed by a fully FCC structure below it.This was observed in all sintered samples.The consequent analyses are focused on the bulk structure, ignoring the surface layer which is caused by reaction with the graphite mould.Cross section XRD spectra are presented in Fig. 3 showing the peaks of FCC and BCC phases as a function of sintering temperature.Table 4 shows the calculated phase volume fractions through Rietveld refinement.There is a small amount of BCC present in all samples, except the sample sintered at 1130 • C, which has no detectable BCC phase.There is however uncertainty in this measurement as the samples are thin (~2 mm thick) and the area reacted to form the BCC phase represents a significant portion of the sample cross area.Overall the amount of BCC phase nevertheless is low and shows a diminishing trend as the temperature is increased.

Grain size
The sintering temperature and time, and especially milling of the powder, are expected to greatly influence the grain size of the material.The calculated grain size parameters from EBSD maps are shown in Fig. 4a.The average grain size increases with sintering temperature increases, as expected.Grain size increases from approximately 3 μm-9 μm when the temperature increases from 950 • C to 1130 • C. At the same time, grain size becomes more uniform as indicated by the reduction in the relative grain size dispersion parameter (Δd/d).The inset in Fig. 4a shows the effect of milling on the grain size, decreasing it to 2.2 μm and 1.9 μm when milled for 10 or 25 h, respectively and sintered at 1100 • C.
The PECS 40 M sample, which was sintered at a lower temperature of 1050 • C after milling for 25 h has a significantly lower average grain size of approximately 0.5 μm.
Fig. 4b shows the fraction of twin boundaries of the total grain boundary determined from EBSD maps for each sample.Increasing the sintering temperature from 950 • C to 1100 • C increases twin boundary fraction from 0.1 to 0.2.Sintering at 1100 • C after milling further increases the fraction of twin boundaries, the corresponding fractions being 0.32, 0.41 and 0.37 for 2 h, 10 h and 25 h milling times.The PECS40 M sample has a slightly lower fraction of twins, i.e., 0.27.A summary of the grain size and twin boundary measurements is shown in Table 5.The twin boundaries present in the samples is also given per unit surface area.The amount of twin boundaries of the surface area increases drastically especially after milling and is the highest in the PECS40 M sample despite the twin boundaries representing only 27% of the total grain boundary.When comparing the PECS samples with the PECS40 M sample a 15-fold increase can be observed for amount of twin boundary present of the total surface area.The increase in twin boundary area of the total volume is due to the decrease in grain size and the increase in total grain boundary area.

Hardness and its correlation to grain size
Hardness was measured in the middle of the cross sections of the samples to obtain the bulk mechanical properties without the influence of the thin BCC/FCC dual phase layer on the surface.It is noted that generally hardness is noticeably higher when measured on the surface.Fig. 5 shows the HV1 hardness values measured from cross sections of the PECS compacts sintered at different temperatures.Hardness increases with the sintering temperature and the highest hardness is obtained at 1050 • C.This trend is consistent with the observed increase in relative density.Above 1050 • C hardness begins to decrease again, indicating a slight reduction in strength of the non-milled samples.Effect of milling on the hardness of samples sintered at 1100 • C is shown in the inset of Fig. 5, showing an increase in hardness as milling time increases up to 25 h.The PECS 40 M sample, sintered at 1050 • C after 25 h of milling, deviates from other measurement points considerably, showing a high hardness of over 300 HV1.This indicates that this combination of processing parameters results in excellent mechanical properties.
The correlation between grain size and hardness for all tested samples is shown in Fig. 6.In general, all PECS samples have a lower hardness than the samples sintered from milled powder.A linear trend can be established between the volume-weighted average grain size and hardness of all samples.A decrease in grain size leads to a considerable increase in hardness.The 950 • C PECS sample deviates from this trend due to its low relative density.

Tensile properties
A summary of tensile properties is shown in Table 6.The sample sintered using milled powder has a considerably high tensile strength.Fig. 7a shows the engineering stress-displacement curves for the tensile specimens sintered of gas atomized and milled gas atomized powder as well as the curve for reference AISI 316 specimen.The displacement is given instead of strain because of difficulties in measuring the exact strain in some specimen.The yield strength, R e , is indicated by the markers in Fig. 7a.The specimen sintered of gas atomized powder yielded a relatively low yield strength of 250 MPa, while the specimen sintered of milled gas atomized powder showed a considerably higher yield strength of ≈800 MPa.The ultimate tensile strength (R m ) of the specimen sintered of gas atomized powder was 692 MPa, while the specimen sintered from milled powder showed a value of approximately 1 GPa.The former specimen had a 35% strain at failure, while the latter specimen failed at a strain of 26%.The reduction in elongation at break is typical when the tensile strength increases, however, it remains at a good level also for the specimen sintered of milled powder.Overall, milling the powder prior to sintering generated a considerable increase in yield and tensile strength while maintaining good ductility.The elastic modulus measured utilizing an optical extensometer for the specimen sintered from gas atomized powder obtains a value of 182 GPa.Comparison of strain hardening behavior of specimen sintered from gas atomized powder and reference AISI 316 specimen is shown in Fig. 7d.Strain hardening shows a very smooth trend from the initially Fig. 3. XRD of Cross sections.high values at yield to stabilized values further in the plastic region for both samples.Ferritescope measurements were performed on all three tensile specimens to reveal the possible presence of a ferromagnetic phase in the tensile test specimens.It was confirmed that CrFeNiMn samples showed no ferromagnetic phase before or after tensile testing while more than 20% of strain induced α ′ -martensite was observed in the fractured AISI 316 tensile test specimen after tensile testing.Strain hardening exponents (n) were calculated for the 316 reference samples as well as for the non-milled tensile sample and a value of 0.24 was obtained for both samples.Strength coefficients (k) were also calculated for both the samples and values of 1150 and 1160 MPa were obtained for the 316 reference sample and the non-milled CrFeNiMn samples, respectively.
To verify that the milled sample maintained a primarily ductile failure mode, the fracture surfaces were imaged using scanning electron microscopy, and are shown in Figs. 8 and 9 for the two types of samples.Both show a highly ductile appearance, with small dimples observed for both.
Dimple sizes observed on the tensile fracture surfaces range from 0.2 to 5 μm for the specimen sintered from gas atomized powder and from 0.05 to 1.4 μm for the specimen sintered from milled gas atomized powder, with average values of 1.26 μm and 0.33 μm, respectively.
While the fracture surfaces of the specimen sintered of gas atomized powder were quite planar, strong faceting was observed for the specimen sintered using milled gas atomized powder.The size of the features is influenced by the geometric characteristics of the microstructure.The inset shows the influence of milling time on hardness.

Evolution of microstructure
In order to better understand the differences observed in tensile properties and the associated fracture surfaces, the microstructures of the tensile specimen sintered from gas atomized powder and milled gas atomized powder must be compared.Here the former specimen refers to the specimen sintered at 1100 • C for 5 min directly from gas-atomized powder, while the latter tensile specimen refers to the specimen that had the powder been milled for 25 h, followed by sintering at 1050 • C for 1 min.Particular attention is given to the deformation mechanisms, i. e. the cell-forming dislocation mediated plastic deformation process as well as twinning.Before deformation the specimen sintered from gas atomized powder has a relatively large but uniform grain size, with many annealing twins visible; see Fig. 10a.As shown in Fig. 10b, tensile deformation has altered the grain morphology, induced severe orientation gradients within the grains, as well as formed deformation twins.In comparison the specimen sintered from the milled powder has a significantly smaller grain size prior to straining, as shown in Fig. 11a (note the scale difference).The main difference to the specimen sintered of gas atomized powder is the presence of a bi-modal grain size distribution, showing ultrafine grains with a size of only 200-300 nm, and fine grains in the 1-3 μm regime.The fine grains have annealing twins, while the ultrafine grains do not seem to have grain sub-structures.The ultrafine grains form a necklace structure, showing up as continuous veins throughout the microstructure.After the tensile test the amount of ultrafine grained regions has increased, which are difficult to index due to the high dislocation densities; Fig. 11b.At the same time, the finegrained areas show some orientation gradients and significant grain sub-structure.No deformation twins were observed for the specimen sintered of milled powder.
To investigate the deformation mechanisms and explain the increased tensile strength for the milled sample, a detailed analysis was carried out to analyze the differences in grain sub-structures.Before deformation the specimen sintered of gas atomized powder has relatively large grains, with very few dislocation sub-structures present, i.e.SGBs and DDWs (Fig. 12).In the uniformly strained area of the fractured specimen (high strain area) the IPF shows clear division into sub-grains and presence of significant orientation gradients.The domain misorientation reveals small sub-grains and dislocation cells.Refinement is even higher in the ultimate strain region (in the neck area of the fractured tensile specimen) with both the grain size and dislocation substructures being considerably smaller.New grains have been formed during straining, the size of which corresponds to the size of the smallest dislocation cells shown by Δθ DDW = 1 • .
The microstructural evolution during tensile testing of the specimen sintered of milled powder is shown in Fig. 13.The undeformed state shows much smaller grains compared to the non-milled sample; the volume-weighted average grain size is approximately 1 μm compared to 10 μm for the non-milled sample.The grain size is bi-modal, with the ultrafine grains being relatively free of dislocation sub-structure, while the fine grains show some dislocation cells.The tensile strained samples show no significant changes in the grain size, exhibiting a similar bimodal distribution of ultrafine grains and fine grains.Straining has caused the formation of sub-grains and dislocation cells in the fine grains.For the ultimate strain location the size of the smallest dislocation cells is 100-300 nm, being similar to the ultrafine grains observed in the undeformed state.This indicates that the material is achieving its maximum degree of grain refinement, and that the plastic strain is accommodated by the evolution of the sub-structural boundaries from dense dislocation walls to sub-grains and finally into new grain boundaries.
To quantify the deformation process, the average size of grains and dislocation sub-structures is shown in Fig. 14.The analysis is carried out only for the undeformed and ultimate strain locations, as they contain a sufficient amount of grains for stochastic analysis.The specimen sintered of gas atomized powder shows a decrease in all size parameters after the tensile test.While the volume-weighted average grain size has decreased approximately from 10 μm to 5 μm, the dislocation substructures have become considerably smaller.For example, the subgrains (Δθ SGB = 2.0 • ) have decreased in average size from 9 μm to 1.5 μm.Before deformation the dislocation cells have an average size of 4-7 μm.After tensile deformation the cell size has decreased to 0.5 μm and 0.8 μm, defined at misorientation of Δθ DDW = 0.5 • and Δθ DDW = 1 • , respectively.The most significant influence of milling the powder prior to sintering is the reduction of initial grain size down to the ultrafine range at a volume-weighted average of 0.92 μm.Almost no sub-structure is observed as the size of the SGBs and DDWs is similar to the grain size.In the strained specimen the volume-weighted average grain size has reduced slightly to 0.78 μm and the sub-grain size to 0.47 μm.At the same time, the average dislocation cell sizes have refined to 0.33 μm (Δθ DDW = 1.0 • ) and 0.25 μm (Δθ DDW = 0.5 • ), respectively.When viewed in terms of the d −0.5 of the Hall-Petch scaling law this reduction is considerable compared to the corresponding values of 0.75 μm and 0.51 μm of the non-deformed reference sample.Twins are observed before and after deformation in all tensile specimens.For both samples the length of grain boundaries increases more than the length of twin boundaries, and thus the relative amount of twins shown in Fig. 13 is lower after straining.Fig. 15a shows the deformation twinning present in the ultimate strain region of the tensile specimen and Fig. 15b shows in red the twin boundaries detected using the criteria of >57 • misorientation angle and <10 • degrees from a 111 axis.The TEM micrograph in Fig. 15c shows the dislocation structures present in the milled tensile specimen after 25% of uniform deformation.Dislocation networks can be seen gathering near annealing twin boundaries and grain boundaries.Dislocation  cell sizes calculated from TEM micrographs ranged from 60 to 250 nm in the tensile specimen sintered from milled gas atomized powder and from 100 to 500 nm in the specimen sintered from the gas atomized powder.

Discussion
Bulk single-phase FCC structured CrFeNiMn-alloy can be made by pulsed electric current sintering of corresponding prealloyed gas atomized powder.Although the parent powder had a minor amount of BCC phase prior to sintering, no BCC phase could be detected by XRD when  the temperature of the PECS process exceeded 1100 • C. The BCC phase observed at the edge regions of all sintered disks seems to result from some kind of an interaction between the alloy and the graphite molds used in the PECS process.This was also observed by Bloomfield et al. [30] whom reported a Cr-rich BCC precipitation phase in the CrFeNiMn-alloy system.As a result, a duplex structure containing chromium rich ferrite (chromium is a ferrite stabilizer [44]) and chromium rich BCC forms near the edges of the PECS samples.When this layer is removed the structure consists of single-phase FCC.
Processability of the alloy using PECS is excellent and a good >98% relative density is obtained above 1050 • C as shown in Fig. 1.Maximum hardness of the alloy is also obtained at 1050 • C (Fig. 5) as hardness   decreases due to grain growth at higher temperatures.Milling improves the processability of the powder as well as the properties of the consolidated bulk material, illustrated in the inset of Fig. 1.Milling reduces the particle and crystallite size of the powder [45].This increases the active surface area of the powder, which in turn aids the sinterability of the powder by lowering the T s [46].This can be observed when comparing the T s and T f for the 2, 10 and 25 h milled samples.Milling for 2 h had a very small impact on T s and T f temperatures, as well as on the density after PECS.However, grain size did decrease noticeably, from 6 to 3.6 μm, increasing hardness by more than 10%.Milling the powder for 10 h increased the density of the sintered sample significantly to ≈99% while the grain size further decreased down to 2.2 μm and the hardness improved in accordance with the Hall-Petch relationship.Milling for 25 h did not yield a significant increase in density when comparing to the 10 h, but the grain size was further refined to 1.9 μm and hardness again increased in accordance with the Hall-Petch relationship.The main strengthening mechanism of milling is thus the grain size strengthening (Hall-Petch effect), and the main deformation process is the cell-forming plastic deformation mechanism [47] that has been observed for a variety of polycrystalline materials [12,13].This is in contrast to many Fe-Cr-Ni-stainless steel alloys, where planar deformation structures prevail due to the low stacking fault energy (SFE) as well as to Cantor alloys, where the main deformation mechanism is mechanical slipping at lower strains (10-30%) and mechanical twinning at higher (30-50%) [48].It thus seems that the equiatomic FeCrNiMn-alloy has a relatively high SFE.Galindo-Nava and Rivera-Díaz-del-Castillo [49] have analyzed the data for austenitic steels to correlate the transition in mechanism with stacking fault energy.Twinning is observed in alloys for which SFE <30 mJ m −2 , whereas martensitic transformation is achieved when SFE <18 mJ m −2 .Lower SFE in HEAs enables the propensity of twinning and transformations.Deformation twinning has been observed in FeNiMnCr 18 [5] at a temperature of 77 K, however none were detected at room temperature as has been observed in this work.This is most likely due to the higher Cr content present in this work.In the CrFeNiMn alloy system Cr, Fe and Mn all lower the stacking fault energy (SFE) of pure Ni, however Cr is the most significant SFE reducer [50].Deformation twinning is very unlikely to occur in the tensile sample made from the milled powder, as the grain size is very fine and the stresses to induce twinning is extremely high (>1.2GPa) [18].XXXXXXXXXXXXXXXXX.
Bozzolo and Bernacki [51] observed a difference between twins formed during recrystallization and annealing.There is evidence of recrystallization in all samples made using milled powder.This can be seen from the increased fraction of twin boundaries present in each milled sample as well as the increased fraction of incoherent twins.
The T f temperature which indicates completion of densification was achieved for the 10 h and 25 h milled samples at 1075 • C and 1045 • C, respectively.As sintering of milled samples was performed at 1100 • C this explains the lack of noticeable improvement in density when comparing the 10 and 25 h milled samples.The particle size of the powder reaches a critical level after 10 h of milling and does not get reduced significantly.This is evident as the surface energy is a large driver of densification during PECS.However, the grain and microstructure continue to evolve as milling is performed for 25 h, as can be seen from the hardness and grain size of the bulk materials.There is a slightly larger amount of BCC phase detectable by XRD on the tensile sample sintered from the milled powder, when compared to the sample sintered directly from GA powder (1050 • C).
The compacts, from which the milled tensile samples were cut, were sintered at a lower temperature of 1050 • C. The temperature was chosen Fig. 12. Evolution of the dislocation sub-structure for tensile specimen sintered of gas atomized powder, showing the inverse pole figure, sub-grain boundaries and dense dislocation walls.Analysis is shown only for a small representative sub-section of the entire maps, indicated in Fig. 10.
for the maximum hardness and density.Grain growth was also significant at temperatures above 1050 • C, as shown in Fig. 4 and Table 4. Milling the powder prior to the PECS process improved the density, and hardness as well as yielded a smaller grain size.
The tensile samples made directly from the non-milled powder had properties similar to the PECS 1100 • C sample.Tensile samples sintered using both milled and nonmilled powder had a density of ~99% confirming good processability of powder with PECS.The grain size decreased from 5.7 μm to 500 nm after milling and PECS and hardness increased from 190HV1 to 310HV1, i.e. by 60%, with the decreasing grain size.Ultimate tensile strength increased by 45% from 700 MPa to 1000 MPa when milling the powder prior to PECS, while the elongation at break decreased from 35% to 26%.The measured dimple sizes (0.33 μm and 1.26 μm) correspond very closely to the dislocation cell sizes Fig. 13.Evolution of the dislocation sub-structure in the tensile specimen sintered of milled powder as showed by the inverse pole figure, sub-grain boundaries and dense dislocation walls.Note that the magnification is four times higher compared to Fig. 12. Analysis is shown only for a small representative sub-section of the entire maps, indicated in Fig. 11.Fig. 16 compares the tensile properties obtained in this work with those of stainless steels (AISI 304 and AISI 316) and other similar high entropy alloys, including TRIP and TWIP steels.
Twin boundary fraction of total grain boundaries increases as the sintering temperature increases.During grain growth twin boundaries multiply [47].This is also observed in this work as the fraction of twin boundaries increase linearly with the sintering temperature as shown in Fig. 4b.The further increase in twin boundary caused by milling is due to the reduction in starting particle size of the powder and the deformation [14] the powder is subjected prior to sintering.
CrFeNiMn exhibits a strong correlation between hardness (HV) and tensile strength (UTS).The tensile specimen sintered of milled powder shows hardness of about 300HV1, i.e., about 2940 MPa, while the tensile specimen sintered of gas atomized powder shows hardness of about 200HV1, i.e., 1960 MPa.Thus the hardness is approximately 3 times the ultimate tensile strength for both specimen having different grain structure.Literature values for the ratio of hardness to UTS range from 2.5 to 4.4 [53], thus the obtained values of 3.7 and 3.3, for the non-milled and milled tensile samples, respectively, are well within the literature values.
The tensile specimen sintered of the gas atomized powder shows a tensile to yield strength (TS:YS) ratio of 2.1 and the tensile specimen sintered using milled powder shows a TS:YS ratio of 1.2.As the literature values reported for TS:YS ratios range from 2.1 to 1.0 [53] the former specimen shows a TS:YS ratio in the higher end of the range and the latter specimen shows a value close to the minimum value reported.A high TS:YS ratio (>1.56) [53] indicates good strain-hardenability.When comparing the strain hardening behavior of the AISI 316 reference sample to the tensile specimen made using gas atomized powder it can be observed that the CrFeNiMn-alloy sample hardens at a faster rate when compared to AISI 316 stainless steel.Ferritescope measurements confirmed that no phase transformations occur in CrFeNiMn during the tensile testing making it a more stable phase alloy in high strain applications.Both samples have similar yield points and ultimate tensile strengths but AISI 316 work hardening behavior is partly related to formation of strain induced martensite while that of the CrFeNiMn relies on dislocation cell formation and at very high strain twinning.When comparing the strain hardening behavior of CrFeNiMn to that of Cantor alloys, a noticeable difference can be observed.CrFeNiMn strain hardening rate levels off around the 1000 MPa, while Cantor alloy levels off noticeably lower at 800 MPa [54].Transformation induced plasticity (TRIP) Dual Phase (DP-HEAs) consisting of the CoCrFeMnNi-alloy system [54,55] shows similar strain hardening behavior as CrFeNiMn, however this is through the aid of an FCC to HCP transformation occurring during deformation.Equiatomic CrFeNiMn made using PECS shows excellent potential to be used in cryogenic applications, this is supported by previous research conducted on the Cantor alloy [26] identical observations have also been made in the case of FeNiMnCr 18 alloy [23].

Conclusions
In this work an equiatomic CrFeNiMn was consolidated from gas atomized powder using PECS.The effect of milling of the powder on the PECS process was also investigated as well as the microstructure and properties of the alloy.Tensile samples were made of the compacts sintered directly from the gas atomized powder as well as milling the powder with optimized PECS parameters to obtain an ultra fine grained microstructure.Milling the powder increased the hardness and strength of the alloy by 60% and 45%, respectively, while elongation at break decreased by only 25%.
Deformation mechanisms were analyzed from a milled and nonmilled tensile sample with grain sizes of 0.54 μm and 5.69 μm respectively.The non-milled sample shows dislocation cell formation as the main deformation mechanism along with the formation of new finer grains and a small amount of deformation twins were observed at the highest strain location.The milled sample due to its ultra fine grained microstructure showed dislocation cell formation as its main deformation mechanism.
CrFeNiMn also shows good strain hardenability as well as very good phase stability under deformation in comparison to its parent alloy (CoCrFeNiMn) as well as AISI 316.This is especially beneficial to it not having Co in its composition showing great promise in nuclear applications, and its phase stability in comparison to AISI 316, which formed transformation induced martensite after deformation.
energy and allows sintering to occur at a lower temperature.While milling for 2 h did not yield a densification finish temperature (T f, which was measured from the PECS process curves, refers to a temperature above which no noticeable densification occurs), the milling times of 10 and 25 h yielded T f temperatures of 1075 • C and 1025 • C, respectively.In contrast, none of the non-milled PECS samples reached a densification finish temperature even at 1130 • C. Milling the powder also lowers the sintering start temperature (T s, which was measured from the PECS process curves, refers to a temperature at above which densification of the powder begins).Milling the powder for 2 h initially increases the T s from 850 • C to 930 • C, which is most likely due to the powder accumulating stress without a noticeable decrease in T f .Further milling the powder for 10 and 25 h decreases the T s to 850 • C and 725 • C, respectively.These findings indicate that high relative density can be achieved with the 45-90 μm powder both in the as-received and milled conditions.In summary, 98% relative density could be obtained by PECS of the gas atomized powder at 1100 • C in 5 min and the density increases to 99% if the powder is milled for 10 or 25 h.

Fig. 1 .
Fig. 1.Relative density of all samples as a function of sintering temperature for a constant sintering time of 5 min, with the inset showing the effect of milling time when sintered at 1100 C for 5 min.

Fig. 4 .
Fig. 4. (a) Calculated average grain sizes (d) from EBSD images and (b) twin boundary surface area of total material volume Insets shows the effect of milling for 2 h, 10 h and 25 h.

Fig. 5 .
Fig. 5. Vickers hardness (HV1) measured for cross sections of PECS compacts.The inset shows the influence of milling time on hardness.

Fig. 7 .
Fig. 7. Tensile curves for the two tensile samples.(a) Tensile stress vs. displacement measured from the gross-head movement, (b) stress vs. strain (obtained based on the optical extensometer), (c) true stress vs true strain and (d) strain hardening behavior.

Fig. 8 .
Fig. 8. Fracture surface of the tensile sample sintered from the gas atomized powder at 1100 • C for 5 min.

Fig. 9 .
Fig. 9. Fracture surface of the tensile sample sintered at 1050 • C for 1 min from the gas atomized powder milled for 25 h prior to sintering.

Fig. 10 .
Fig. 10.Non-milled tensile specimen showing the orientation map (IPF-Z) overlaid on the band contrast map to highlight locations of grain boundaries: a) Reference state before deformation, and b) location of ultimate strain after tensile test.Tensile axis is horizontal, and the dashed regions show the location of microstructural evolution analysis in Fig. 12.

Fig. 11 .
Fig. 11.Tensile specimen sintered from milled powder showing the orientation map (IPF-Z) overlaid on the band contrast map to highlight locations of grain boundaries: a) Reference state before deformation, and b) location of ultimate strain after tensile test.Note the scale difference compared to Fig. 12. Tensile axis is horizontal, and the dashed regions show the location of microstructural analysis in Fig. 13.

Fig. 14 .
Fig. 14.The volume-weighted average grain size and the average size of dislocation sub-structures in the tensile specimens sintered of milled gas atomized gas atomized powder.Analysis is carried out for the full EBSD maps shown in Figs.11 and 12.

Fig. 15 .Fig. 16 .
Fig. 15.Deformation structures in the PECS specimen.a) Orientation map (IPF-Z) overlaid on band-contrast map showing deformation twins in the ultimate strain region of the specimen made of gas atomized powder, b) Twin boundaries (red color) analyzed based on the criteria >57 • in misorientation and <10 • from 111 axis in the same area, and c) TEM micrograph showing dislocation sub-structures in the high strain region of the specimen sintered of gas atomized and milled powder.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Table 1
Summary of the processing parameters of the prepared PECS samples.

Table 2
Parameters used for EBSD data de-noising with the half-quadratic filtering.
• Threshold for sub-grain boundaries (point-to-point)

Table 3 EDS
results for the CrFeNiMn sample sintered at 1100 • C. Compositions are given in at-%.

Table 4
Phase composition through XRD/Rietveld analysis of PECS discs cross sections.

Table 5 -
Measured grain sizes (Bolded samples are sintered using same PECS process parameters).