Mechanically derived short-range order and its impact on the multi-principal-element alloys

Chemical short-range order in disordered solid solutions often emerges with specific heat treatments. Unlike thermally activated ordering, mechanically derived short-range order (MSRO) in a multi-principal-element Fe40Mn40Cr10Co10 (at%) alloy originates from tensile deformation at 77 K, and its degree/extent can be tailored by adjusting the loading rates under quasistatic conditions. The mechanical response and multi-length-scale characterisation pointed to the minor contribution of MSRO formation to yield strength, mechanical twinning, and deformation-induced displacive transformation. Scanning and high-resolution transmission electron microscopy and the anlaysis of electron diffraction patterns revealed the microstructural features responsible for MSRO and the dependence of the ordering degree/extent on the applied strain rates. Here, we show that underpinned by molecular dynamics, MSRO in the alloys with low stacking-fault energies forms when loaded at 77 K, and these systems that offer different perspectives on the process of strain-induced ordering transition are driven by crystalline lattice defects (dislocations and stacking faults).

. The arrangement of the sub-figures is to be improved. In addition, the measurement on the Moiré-fringe was wrong, which should be deleted though, as it shows no practically useful information.
Page 33, line 847. It is necessary to explain why using in-situ heating test to introduce oxidation to the TEM lamellar when one can simply remove the sample from TEM and wait for it. Is it for the purpose of accelerating the oxidation process? If so, the piece of information should be included in the methods.
Lastly, I would like to appreciate the vast amount of work the authors have done and included in the extended data.
2 order in Fe50Mn30Co10Cr10 high-entropy alloy. Mater. Today Nano 16, 100139 (2021)] providing the experimental results of chemical SRO (CSRO) in non-equiatomic quaternary Fe50Mn30Co10Cr10 (at%) HEA. They reported that even though Cr, Mn, Fe, and Co are close in atomic size, unusual or particular CSRO does exist in the Fe50Mn30Co10Cr10 HEA, unlike equiatomic ternary VCoNi MEA. In particular, their previous TEM results demonstrated a general tendency toward 'preference for unlike pairs and avoidance of like pairs', i.e., Fe-Mn, Fe-Co, Fe-Cr and Mn-Co all exhibited a relatively strong tendency to be neighbours, elevating the variability of atomic packing in the non-equiatomic Fe50Mn30Co10Cr10 HEA. A similar trend was also shown in our Fe40Mn40Cr10Co10 (at%) HEA. In addition to this similarity, our MC simulations showed a 'preference for Cr-Cr like pairs' in the Fe40Mn40Cr10Co10 system at 750 K. This intriguing feature of Cr-Cr like pairs was even demonstrated in a ternary equiatomic NiCrCo MEA, suggested in the earlier MC simulations and seen experimentally.
This indicates that even though concentrated solid solutions including Cr had different chemical compositions, 'preference for Cr-Cr like pairs' would be favouable. Reason for this is presumably due to the complex thermodynamic driving force for generating the CSRO. In fact, thermodynamic driving force for forming the CSRO in a few M/HEAs can stem from diverse paths, for example, fluctuation in the local strain, bonding state, electronic and magnetic interaction. A further in-depth understanding of this issue (determination of CSRO type) demands careful energy computations. However, this additional work to reveal why there is a tendency of 'preference for Cr-Cr like pairs' in the nonequiatomic quaternary is beyond the scope of this study, because we primarily focus on mechanically stimulated SRO.
In addition to the type of CSRO, we fully agree that reviewer's suggestion about plotting the statistical size distribution of CSRO domains via atomistic MC simulations would be beneficial for strengthening our work. Unfortunately, quantitative measurements on the exact size distribution of CSROs, based on the MC simulations, is still an open issue on account of the limitation of the simulation methodology. In other words, it would be impossible to define clear boundaries between the CSROs and the matrix region in the captured snapshot-slices of atomic species configurations with a thickness of 2 or 3 nm, given by the MC simulations. Instead, we could provide the qualitative trend in CSROs with respect to the temperatures of the current HEA system. Furthermore, based on our TEM diffraction patterns, the intensity of CSRO-stimulated diffuse scattering signals for boron-free Fe40Mn40Cr10Co10 HEA (before tensile deformation) was limited to quantitatively determine the degree/extent of the extremely small CSROs. In turn, TEM foils taken from real bulk samples of the Fe40Mn40Cr10Co10 HEA before straining would have possessed the limited degree/extent of chemical SRO domains. All authors, therefore, conclude that even though a comparison of the SRO size distribution between atomistic MC simulations and TEM results is important but challenging, it seems to be beyond the scope of this study. 3 With consideration of the length of this manuscript, we addressed the statistical size distribution of mechanically derived SRO, determined from dark-filed TEM imaging.

The following changes were made:
Please see the Introduction section on lines between 74 and 106, pages #4~5, "Extensive efforts to monitor the ordering transition in MPEAs have provided insights into the thermally induced CSROs via simulations 14,15,26 and X-ray absorption 30 . However, X-ray-based measurements provide rather low analytical capabilities, restricting the experimental observation of CSROs in the individual grains of the materials, as they are averaged over a comparatively large volume of material 16 . Later, transmission electron microscopy (TEM) and atom probe tomography went through a revolution to record the extremely small CSROs within individual fcc grains for metallic alloys [16][17][18][19][20][21] . For instance, TEM dark-field images, formed with CSRO-induced additional diffuse scattering in reciprocal-space selected-area electron diffraction patterns (EDPs) with specific zone axes, enable quantitative measurements of the CSRO degree/extent in fcc-structured NiCrCo and VNiCo MEAs [16][17][18] . This approach is even applicable for binary Ti-6Al concentrated solution 33 and high-Mn (25 wt%) steel sample 34 . Overall, CSRO sizes ranged from ~0.7 to ~2.0 nm for a given fcc-structured alloy composition 34,35 .
In contrast to the superlattice reflections caused by long-range-ordered nanoprecipitates, elastic diffuse scattering between normal fcc Bragg reflections in TEM-EDPs is ascribed to structural deviations (e.g., vacancies, dislocations, stacking faults, and SROs) from a periodic crystalline lattice 36,37  Other TEM-based studies suggested that SFEs of MPEAs do not influence the degree of chemical ordering 24,25 . Therefore, the following fundamental questions still remain: (i) whether SFs are phenomenologically intertwined with the clustered ordering; (ii) whether ordering can affect mechanical twinning and deformationinduced martensitic transformation 12 ; and (iii) the mechanism and reason for the 'deformation-derived disorder-to-SRO transition at low temperatures', suggested based on experimental observations 35 and theoretical calculations 38 ." "Hereafter, such strain-induced ordering is denoted as mechanically derived SRO (MSRO) for convenience." Please see Results on lines between 130 and 142, page #6, "…..in the Methods section. Although the near-random spatial distribution of each individual element is observed at 1373 K (Fig. 1a), CSRO formation in the singlecrystal system is potentially favourable at 873 and 750 K ( Fig. 1b and c, respectively).
The associated radial distribution function profiles revealed three features. First, by lowering the temperature, Fe-Co, Mn-Cr, Mn-Co, Fe-Mn, and Fe-Cr pairs, all showed a relatively stronger tendency to be neighbours, promoting the variability of atomic packing sequence in the first neighbouring shell (Fig. 1d-f). This holds for a general CSRO definition, i.e., 'preference for unlike pairs and avoidance of like pairs' 39 . Our MC simulations for the non-equiatomic Fe40Mn40Cr10Co10 HEA were consistent with the type of ordering, reported by a recent TEM work 39 highlighting Fe-Mn/Co/Cr preference in the first neighbouring shell in an fcc Fe50Mn30Cr10Co10 (at%) system. Second, of all unlike pairs, Fe-Co exhibited the strongest tendency to be neighbors. Third, a primarily strong tendency toward 'preference for Cr-Cr like pairs' was generated by lowering the temperatures. Besides these results" Please see lines 268 to 272 on page #12, "Concurrent with previous TEM-based studies in which CSROs and chemical medium-range-order domains form an interchangeable preferential ordering in a 5 NiCrCo MEA 17,18,21 , our observations suggest that a period of {311} inter-planar spacing in MSRO doubles that of the fcc lattice for the current alloy. This can explain the detection of the diffuse scattering introduced by MSRO at the ½{311} locations in all [112] TEM-EDP and HRTEM-or STEM-FFT results." Please see lines 297 to 311 on page #13, "The B-free Fe40Mn40Cr10Co10 reference alloy before straining was chemically ordered at 750 K, predicted by MC simulations (especially, Cr-Cr pairs in Fig. 1).
However, for the system containing CSROs, our MD-EDPs ([ 1 12] zone axis) revealed weak or faint diffuse scattering at the ½{311} locations ( Supplementary Fig.   9). This indicates that the degree/extent of CSROs in the MC simulations for the reference structure may not yield clear diffuse scattering in the MD-EDPs.
Compared to the B-free system, B-doped Fe40Mn40Cr10Co10 HEA prior to straining was expected to have more and larger CSROs owing to a relatively stronger driving force for CSRO formation in the DTA thermal profiles (Supplementary Fig.   1) and to a strong negative solution enthalpy by B ingress. However, we experimentally detected weak diffuse scattering at the ½{311} locations in the [112] TEM-EDP for the HEA (Fig. 5a). Because our TEM samples include the extremely small CSROs in the real microstructure, a large inelastic background attributed to thermally induced diffuse scattering and plasmon scattering would lower the TEM spatial resolution 37 . This might be the primary cause for the weak diffuse scattering signals from the B-doped sample before straining." Please also see Discussion on lines between 404 and 418, pages #17~18 "Short-range diffusion-mediated CSRO in a few MPEAs can be described in terms of 'preference for unlike species and avoidance of like pairs' 17,18,21,39

Comment # 2
Both NiCrCo MEA and Ag are deformed under low, moderate and fast strain rate (as shown in Figure   5d and 6), while Fe40Mn40Cr10Co10 HEA is only deformed under low and moderate strain rate (as plotted in Figure 5b). Is there any reason why the results of Fe40Mn40Cr10Co10 HEA under fast strain rate are not included? Besides, this paper claims that the deformation is highly dependent on the strain rate.
However, the simulation strain rate (10 7 ~ 10 9 s -1 ) is several order of magnitude higher than the 7 experimental one (10 -5 ~ 10 -3 s -1 ). In that case, how to prove that the simulation predictions are able to support the experimental observations?
Author reply to comment #2 We fully agree with these issues related to the strain rates. First, we considered here the visualisation of Fe40Mn40Cr10Co10 HEA results for only low (2 × 10 7 s -1 ) and moderate (2 × 10 8 s -1 ) strain rates owing to the limitation of interatomic potential used for the non-equiatomic quaternary HEA system. Smaller strain (0.08) was used for the HEA structure, as amorphisation was observed at a fast strain rate (2 × 10 9 s -1 ). This amorphisation (grey colour in the figure below) becomes more severe for deformation with a fast strain rate, restricting the observation of the effect of strain rate under equivalent conditions. Therefore, we regrettably should present results for only low and moderate strain rates because the presence of result for the fast rate and discussion of the amorphisation would go beyond the scope of the manuscript and, from a more technical perspective, would make the presentation of the results far more complex and non-transparent. Instead, we briefly explained the reason for this in Figure 6b caption in the newly revised manuscript.  Materialia 197, 54-68 (2020)]. For example, a discrepancy in the dislocation density between the present MD simulations (10 16 ~ 10 17 /m 2 ) for Fe40Mn40Cr10Co10 HEA and the present TEM experiments (~10 10 /m 2 ) for boron-doped Fe40Mn40Cr10Co10 HEA may raise a question how the simulation predictions are able to support the experimental observations. Therefore, we could hardly say that there is a quantitative agreement between experiments and simulations. Instead, we could provide a possible insight into stronger slip planarity due to a higher loading rate, proved by experiments and simulations qualitatively. In addition, the use of a much smaller sample and narrower selected area could compensate the higher dislocation density for the analysis of diffraction. In this regard, an important point to compare the experimental and simulation results is not on the absolute size and loading rates but on whether the simulation can reproduce the slip mechanism shown in the experiments phenomenologically. As shown in the figure, the present MD simulation can reproduce well the slip mechanism via partial dislocations of usual fcc metals depending on specific values of SFE. To clarify this issue, we added the following sentence in the revised manuscript.

The following changes were made:
Please see lines between 118 and 123 on pages #5~6, "To phenomenologically underpin the TEM results obtained from respective fcc single-grains, we further conducted atomistic Monte Carlo (MC) simulations, molec ular dynamics (MD) simulations, and MD-based virtual diffraction analyses in an fcc Fe40Mn40Cr10Co10 single-crystal system. In this context, the experimental results and computational predictions provide insights into the microstructural features attributable to MSRO in the non-equiatomic FeMnCrCo HEA and other fcc nonequiatomic NiCrCo MEAs." Please see lines between 224 and 229, page #10 "Assuming that there was no atomic diffusion of principal species in the HEA during quasistatic deformation at 77 K, it is pertinent to attribute the additional diffuse scattering to the emergence of strain-induced SRO rather than to the initial CSRO prior to the deformation. Again, MSROs are distributed along the slip band. In contrast with MSRO, the distribution of initial CSROs is highly regular throughout 9 the individual grain, suggested computationally 14,15,22,26 and imaged experimentally [16][17][18][19][20][21]33,34 ." Please see lines between 371 and 375, page #16 "Deformation at a higher έ gives rise to more dislocations (or stronger slip planarity) and more SFs (Fig. 6c). This demonstrates that the increased intensity of MSROintroduced scattering in reciprocal-space MD-EDPs due to higher έ drives the Bdoped system towards stronger slip planarity and more SFs. Moreover, there were no LRO precipitates in all of the cell structures. This confirms that spot-like ½{311} scattering in the MD-EDPs is directly correlated with SFs and dislocations, but not with LRO structure." Please see Discussion on lines between 439 and 447, page #19 "These results were consistently supported by MD simulations and relevant MD-EDPs. However, three concerns can be raised: (i) a vast difference in the applied έ values between the bulk samples (in range of 10 -5 -10 -3 s -1 ) and the simulations (10 7 -10 9 s -1 ); (ii) the simulated Fe40Mn40Cr10Co10 system had a single-crystalline structure, while experimental B-doped Fe40Mn40Cr10Co10 sample was polycrystalline; (iii) the accuracy of the potential used in the present simulation. The first issue is

Comment # 3
The microstructural discrepancies between the experimental sample and the simulation model. The experimental Fe40Mn40Cr10Co10 sample is polycrystalline with a grain size from 3.5 to 5.1 μm, while the simulated Fe40Mn40Cr10Co10 sample is single crystalline. A bi-/poly-crystal model with grain boundaries is highly recommended to add for comparison.

Author reply to comment #3
Many thanks for this helpful comment, we fully agree. In principle, all TEM results in this work were taken from each grain in the B-doped Fe40Mn40Cr10Co10 polycrystalline samples. As the reviewer and general readers recognized, selected-area diffraction patterns and associated images via TEM are typically obtained from one single respective grain in most materials. Similarly, we here determined the degree and spatial extent of SROs by measuring those that exist within each fcc grain or within an individual grain of the HEA. Specifically speaking, dark-field TEM images formed with the extra scattering appearance in specific beam direction EDPs enable quantitative visualisation of the extremely small CSROs in the respective grain for NiCrCo and VNiCo MEAs 16-18 as well as binary Ti-6Al concentrated solution 33 . If selected-area TEM analysis includes two or three grains with different orientations at the same time, then the associated diffraction pattern cannot be interpreted, and thereby, complex spots are displayed in the diffraction patterns. Hence, selected-area diffraction patterns of TEM for the SRO-stimulated diffuse scattering must be taken from each fcc grain or individual grains of the M/HEA materials. This rationalizes that similar to the TEM diffraction patterns obtained from each grain in the polycrystalline samples, MD simulation-based virtual diffractions were acquired from a single crystalline structure in our original manuscript.
Nevertheless, the reviewer's point would be on different conditions of the dislocation nucleation between experiments and simulations, i.e., grain boundary (GB) can act as a heterogeneous site for the dislocation nucleation. With this consideration, we further performed the MD simulations and its virtual diffraction analysis for the deformed Fe40Mn40Cr10Co10 system with bi-crystal model following the reviewer's recommendation. The poly-crystal cannot be applied, as our MD simulation-based virtual diffraction cannot distinguish diffraction in a specific direction from the averaged diffraction from multiple grains, and it was difficult to find and apply the diffraction simulation on the specific crystal direction. In other words, a diffraction pattern can be obtained only in the crystal orientation initially specified for the simulation cell. Even for the bi-crystal, there are not many choices of initial crystal orientation appropriate for the analysis of the [112] zone axis. This is even true for TEM analysis, as reviewer recognized. In our additional simulations, we selected a tilt grain boundary with a boundary

Comment #4
The Discussion section is not sufficient. Please add some discussions on the effect of temperature on TSRO (such as 140K/300K vs. 77K), comparisons between mechanically stimulated TSRO and thermally activated CSRO, etc.

Author reply to comment #4
We thank the reviewer for the suggestion. We revised the corresponding paragraph according to the reviewer's comment. As well as such amendments, particular attention has been paid to the thorough overhaul throughout the manuscript to improve the readability of this manuscript. Furthermore, the arrangement of the main text was newly improved.

The following changes were made:
Please see Discussion section on lines between 419 and 438 on pages #18~19, "To reiterate our TEM findings from B-doped Fe40Mn40Cr10Co10 HEA after tensile deformation with different έ at 77 K, diffuse scattering at the ½{311} locations in the [112]-indexed EDPs came into focus at specific regions (strain-induced slip bands) in the deformation fcc structure. More SROs in the 77 K-deformed HEA were formed than those in the undeformed sample with the same alloy composition, and they were distributed along the slip bands. Particularly, applying higher έ under quasistatic conditions increased the intensity of MSRO-derived ½{311}fcc diffuse scattering from faint signals to spot-like ones, which in turn yielded more and larger SROs in the slip bands. We again highlight that the spot-like scattering monitored in the TEM-EDPs, HRTEM-or STEM-FFTs, and MD-EDPs is explicitly derived by a high degree of MSRO, but not by fcc-based LRO precipitates. The precursor of MSROderived diffuse discs and spot-like scattering for the strained MPEAs was directly correlated with strain-induced crystal imperfections (dislocations and mechanical SFs), which were originally driven by a strain-induced atomic packing mismatch.
Furthermore, from the viewpoint of mechanical response, the non-equiatomic quaternary alloy system exhibited no σYS dependence on the έ values. On account of these unique features, we herein regarded the localised SROs as mechanically derived 14 SROs (MSROs) rather than diffusion-mediated CSROs that would be destroyed by planar dislocations during deformation.
Our STEM images further revealed the microstructural features responsible for such ordering in the non-equiatomic quaternary HEA, i.e., MSROs in fcc grains were attributed to strain-induced structural deviations (specifically, dislocations and SFs) from a periodic lattice. ….." Also, see lines 448 to 463, on pages #19~20 "We showed that for the 77 K-strained Fe40Mn40Cr10Co10 HEA, the detection of the According to our findings and those previously reported, it is highly plausible that an increase in the diffuse-disc intensity due to a higher έ reflects a higher degree of MSROs, more SFs, and more dislocations in the deformation structures for the nonequiatomic MPEAs. To sum up, displacive MSRO belongs to isostructural disorderorder transition that occurs in slip bands during 77 K-tensile deformation (specifically, in the plastic strain regime). We assume that MSRO might be defined to be the degree of strain-induced local deviation from the average local-scale ordering in terms of either chemical or structural occupation. Unfortunately, a kinetics investigation of MSRO evolution has still not been performed. Further investigation to figure out how MSROs are mechanically stabilised with a higher έ, for example, at different low temperatures and levels of applied plastic strain, would be interesting." On lines between 467~487, pages #20~21 "If the observed slip bands are completely developed in the large strain regime, the associated MSROs will cause glide plane hardening 44,45 , based on our observations of non-shearable MSROs (Figs. 4 and 5d, and Supplementary Fig. 5a). However, indepth characterisation of non-shearable MSROs is still an open issue. A higher έinduced hardening after the onset of yielding arises from high degree/extent of MSRO, elevating severe lattice distortion effect 28,34,35,55 . Although twinning-induced 15 plasticity (TWIP) or transformation-induced plasticity (TRIP) effects can be active in the large-strain regime and arrest the strain localisation 2,40,56 , the current multilength-scale characterisation revealed that both were independent of the applied έ values. The TWIP or TRIP effects are less affected by MSRO formation, as the ordering emerges with the slip bands.
As the evolution of individual slip bands is suppressed by dislocation-source exhaustion in fcc-based metals 57,58 , the denser bands at higher έ are associated with the high density of MSRO. More and larger MSROs are likely to generate large amount of localised heat inside the slip bands, as the SRO phenomenon (either CSRO or MSRO) is a heat generation process. This MSRO-induced heat will stimulate the slip-band refinement, i.e., new slip bands can be generated near pre-existing ones, thereby lowering severe strain localisation in a single slip band 45 . Accordingly, new slip bands can be refined with further straining at a given έ. This is phenomenologically analogous to the so-called dynamic slip-band refinement 44 . We anticipate that this slip-band refinement effect is caused by a higher έ and governs the εtotal dependence on the έ. Thus, the current study can provide important insights into the mechanical properties of many fcc-structured alloys that are designed to take advantage of the disorder-order transition."

Comment #5
It seems that there are several typos: i) label in Figure 1c "at 750K"; ii) label of red curve in Figure 4c "HEA-F". Please double check.

Author reply to comment #5
We appreciate the hint and corrected the typos. The terminology of "HEA-F" is revised as "HEA-H" (for high strain rate).

The following changes in figures were made:
Please see Figure 1c, on page #32, Please also see Figure 5c (formerly, Figure 4c),

Comment #1
The manuscript talks about the mechanically driven SRO in HEA/MEA. If I understand correctly, the concept originates from the 1986 PRB paper by Prof. Littlewood. At that time, the Ge-Si alloy, which was considered as a model random alloy, showed some SRO when epitaxially grown on Si substrate AND annealed at 450 C. The PRB paper then rationalized the mechanism beneath the observation. After reading this manuscript, I am overall confused. First of all, what is the definition of the TSRO here?
The authors claimed that TSRO appears even in pure Ag metal, indicating a completely different definition than the one in the PRB paper. The authors should give their definition of TSRO both linguistically and mathematically.
Then how does the TSRO appear? The kinetics? In the PRB paper, the mechanically driven SRO appears by annealing, similar to the chemical SRO. How about the TSRO here? Intuitionally, the slip would destroy the ordering and make the system approach ideal random.

Author reply to comment #1
We would like to express our sincere appreciation to the reviewer for these immensely helpful and detailed comments. We would like to apologise for the overall confusing. We fully agree that the former version of the manuscript did cause a misunderstanding, particularly at "TSRO appears even in pure Ag metal". We fully agree that this statement in the former version of our manuscript was confusing. After authors' careful discussions on the systematic results, we concluded that Ag perfect crystal should be disordered irrespective of loading. With this consideration, we deleted the corresponding paragraph about Ag simulations to avoid the confusion. Furthermore, please see the authors' response on the comment #4 raised by referee #1 for the revised Discussion section.

The following changes were made:
Please see updated Abstract, lines between 28 and 40 on page #2, "Chemical short-range order in disordered solid solutions often emerges with specific heat treatments. Here, we demonstrate that unlike the thermally activated ordering, mechanically derived short-range order (MSRO) in a multi-principalelement Fe40Mn40Cr10Co10 (at%) alloy originates from tensile deformation at liquid-N2 temperature, and its degree/extent can be tailored by tuning loading rates under quasistatic conditions. The mechanical response and multi-length-scale characterisation consistently pointed to the minor contribution of MSRO formation to yield strength, mechanical twinning, and deformation-induced displacive transformation. Scanning and high-resolution transmission electron microscopy together with diffraction patterns revealed both the microstructural features responsible for such ordering and the dependence of the ordering degree/extent on the applied strain rates. Underpinned by the molecular dynamics simulations, we show that the MSRO is driven by strain-induced structural deviations (dislocations and stacking faults) from a periodic lattice, offering new perspectives on the ordering transition and mechanistic understanding of multi-principal-element alloys at low temperatures." Please see Introduction section, lines between 59 and 64 on page #3, "When compositionally homogeneous structures have specific low-coordinationnumber clusters, the preferential local ordering of principal elements generally dominates over the spatial order of a few nearest-neighbour spacings, i.e., short-range 18 order (SRO), often called chemical SRO (CSRO) 22,23 .
Formation of diffusion-mediated CSRO domains in fully disordered alloys belongs to thermally activated isostructural disorder-to-order transition at short ranges." "It has been widely conjectured that localised planar slip and leading dislocations would destroy the pre-existing CSRO domains in a face-centred-cubic (fcc) phase upon loading, which corresponds to the so-called glide plane softening 31,32 . This effect drives the non-random system towards ideal random, referred as the As shown in Supplementary Fig. 1, prominent exothermic peaks in the heating process of both samples were observed. We found that there was a shift in the exothermic peak to lower temperature due to B doping: 797 K for the B-doped HEA, while 806 K for the B-free case. Moreover, the peak height increases with B ingress for a given continuous heating rate. These results can imply a strong thermodynamic driving force for forming CSRO owing to the B ingress. This rationalizes our aforementioned anticipation that high degree or large extent of Cr rich CSROs can be introduced by both B doping and ageing plus furnace cooling employed herein.
2. In this study, we conducted tensile tests at different strain rates and at 77 K.
Reasons for this are as follows. First, our previous TEM result suggested that lowering of tensile testing temperatures from 298 to 77 K leads to both deformationinduced order transition and resultant planar dislocation glide in boron-doped Fe40Mn40Cr10Co10 (at%) HEA 3 . However, it failed to explore the origin of straindriven SRO in the alloy. Second, reducing the tensile testing temperatures from 298 to 77 K can affect the SFE values of the alloy substantially, as SFE is a fun ction of temperature. In other words, varying the loading temperatures cannot decipher whether planar dislocation glide in fcc metallic alloys is either due to SRO or due to SFE. In fact, consensus has not been reached on the origin of planar dislocation glide, mainly due to the fact that both SRO and SFE have the strong influence on the glide mode of dislocations and associated hardening in glide softening-or hardening-dominated fcc solid solutions. Lastly, chemical SRO is a thermally activated or diffusion-mediated process. This implies that the formation of chemical SRO in M/HEAs is sensitive to loading rates and loading temperatures owing to deformation-introduced heating. Hence, we intentionally selected the tensile testing parameters, to minimise the loading temperatures on the SFE value and to minimise the deformation-induced heating during tensile tests.
Please see the lines between 224 and 229 on page #10, "Assuming that there was no atomic diffusion of principal species in the HEA during quasistatic deformation at 77 K, it is pertinent to attribute the additional diffuse scattering to the emergence of strain-induced SRO rather than to the initial CSRO 21 prior to the deformation. Again, MSROs are distributed along the slip band. In contrast with MSRO, the distribution of initial CSROs is highly regular throughout the individual grain, suggested computationally 14,15,22,26 and imaged experimentally [16][17][18][19][20][21]33,34 ." Also, see the Discussion section, lines between 404 and 463 on pages #17~20, "Short-range diffusion-mediated CSRO in a few MPEAs can be described in terms of 'preference for unlike species and avoidance of like pairs' 17,18,21,39  An earlier TEM study demonstrated that highly intensified streaking along the {111} directions in [110] EDPs was directly correlated with high-degree CSROs in a NiCrCo MEA 16 . According to our findings and those previously reported, it is highly plausible that an increase in the diffuse-disc intensity due to a higher έ reflects a higher degree of MSROs, more SFs, and more dislocations in the deformation structures for the non-equiatomic MPEAs. To sum up, displacive MSRO belongs to isostructural disorder-order transition that occurs in slip bands during 77 K-tensile deformation (specifically, in the plastic strain regime). We assume that MSRO might be defined to be the degree of strain-induced local deviation from the average local-23 scale ordering in terms of either chemical or structural occupation. Unfortunately, a kinetics investigation of MSRO evolution has still not been performed. Further investigation to figure out how MSROs are mechanically stabilised with a higher έ, for example, at different low temperatures and levels of applied plastic strain, would be interesting.."

Comment #2
Technical side, a lot of MD and MC simulations are performed. How accurate is the potential? The prediction of the phase diagram is a non-trivial task. Does the ultrahigh loading strain rate in MD affect the conclusions? I am curious about the accuracy of these simulations. More validations and discussions are needed.

Author reply to comment #2
We thank the reviewer for the suggestion, we fully concur. According to the reviewer's comment, critical concerns can be raised: (i) the accuracy of interatomic potential is questionable, and (ii) there was a vast difference in the applied έ values between the bulk samples (in range of 10 -5 -10 -3 s -1 ) and the simulations (10 7 ~ 10 9 s -1 ).
Regarding the first potential accuracy issue, the accuracy and transferability of such interatomic potential (2NN MEAM) was detailed at the Methods section in the original paper, focusing on fundamental physical properties and properties related to the deformation (e.g., compositional dependence of the yielding tendency and the sluggish diffusion) in comparison with experiments. We must admit that the interatomic potential for HEAs used in the present study is not perfect and exhibits some weaknesses. We already confirmed the problem of unexpected amorphization at too high strain level (here we kindly refer to comment #2 of the first referee). However, we would like to clarify that the key outcome of the present MD simulations is independent of the selection of interatomic potential.  (2020)]. Therefore, we can hardly say that there is a quantitative agreement between experiment and simulation, but we can only provide a possible insight which can be derived by the qualitative agreement. For example, a discrepancy in the density of dislocation observed by the present MD simulations (10 16 ~ 10 17 /m 2 ) and the present experiment (~10 10 /m 2 ) may raise a question why there is a good agreement between observed TEM-EDPs and MD-EDPs. However, the use of very smaller sample and narrower selected area can compensate the very higher dislocation density for the analysis of diffraction. In this regard, an important point to compare the experimental and simulation results is not on the absolute size and rate of loading but on whether the simulation can reproduce the slip mechanism of the experiment. As shown in the figure, the present MD simulation can well reproduce the slip mechanism via partial dislocations of usual fcc metals depending on specific values of SFE. This critical issue is additionally described in Supplementary Discussion 3. Lastly, in the revised manuscript we newly suggest that future work to accurately decipher how the MSROs are mechanically stable, and whether the MSRO is either non-shearable or shearable would be necessary.

The following changes in supplementary information were made:
Please see newly added Supplementary Fig. 14  loading rates. The diffuse scattering at the ½{311} locations is shown, while the relative intensity of the diffuse scattering increases with a higher strain rate. These two features obtained from the Lennard-Jones (LJ) potential model are well consistent with those acquired from different potential model (2NN MEAM) (Fig. 6a). b, c, Corresponding cell structure strained at low and moderate loading rates, respectively, showing the distribution of deformation-induced SFs (middle panel) and dislocations with different Burgers vectors (right panel). Applying a higher loading rate elevated both stronger slip planarity and more SFs, which was well matched with those acquired from different potential model (2NN MEAM) (Fig. 6b).

Reviewer 3:
Chemical short-range ordering recently received significant interest, especially within multi-component alloys due to its important role in tailoring the properties of materials. Seol, JB et al. have presented a detailed and systematic manuscript reporting a mechanically driven "topological short-range order" (TSRO) and its effects in a B-doped Fe40Mn40Cr10Co10 high entropy alloy (HEA), which is novel and interesting.
The authors performed careful characterisation experiments, especially TEM analysis, on HEA samples deformed with strain rates of ~10-3, ~10-4 and ~10-5 at liquid nitrogen temperature. Results indicate that the thermal-induced chemical short-range ordering already pre-existed in the undeformed samples but with a low density. Tensile tests show strain-to-failure increases with strain rate while yield strength remains strain rate-independent. Supported by experimental observations and molecular dynamics simulation results, the authors claimed that the improving ductility with a high strain rate is due to the abundant and homogeneous slip band activities, correlated with the TSRO zones observed within the slip bands. Regardless of the intensive partial dislocation activities, the authors claimed they are minor contributors when compared to slip banding. The TSRO was argued to introduce a "glide plane hardening" effect, i.e., to obstruct dislocations gliding in the slip bands, consequently contributing to improved ductility at high strain rate levels. The authors provided a massive amount of high quality experimental and numerical simulation data; while some of the conclusions are not fully supported by the evidence, they should be addressed before considering for publication. My major comments are given below for reference,

Major comment # 1
The authors concretely attributed the super-lattice diffraction spots in the SAED patterns to diffusive scattering, thus evaluating short-range ordering. Yet no sign of diffusive scattering was seen in the diffraction pattern, such as streaks (except SFs), blurry backgrounds. I strongly suggest the authors explain why using the sign of long-range order, i.e. superlattice spots, to characterise short-range ordering.
Author reply to comment #1 We thank the reviewer for this critical comment. We fully concur the suggestion. We revise the text
These LRO-caused spots are also found in another fcc-based FeNiCrCoCuAl0 Fig. 8b. With this comparison, we suggest that locations of MSROderived diffuse scattering in the TEM-EDPs are well predicted by the MD-EDPs."

Major comment # 2
As presented by the authors, topological short-range ordering, that usually observed in amorphous materials, can be induced by deformation in the selected high entropy alloy, even in pure fcc Ag. This is somehow confusing as the evidence is not sufficient to show the topological nature of the ordering.
The cited literature by Littlewood (page 24, line 108) reported a potential strain-induced chemical shortrange order without mentioning any topological characteristic. I strongly suggest the authors be extremely careful using the term "topological short-range ordering" unless experimental evidence on the short-range ordering in deformed Ag is provided. Obviously, there is no chemical short-range order in pure Ag, but maybe it is not the case for TSRO.

Author reply to comment #2
Many thanks for this helpful comment, we fully agree. First of all, we would like to apologise for the overall confusing. We fully agree that our original manuscript did cause a misunderstanding, particularly at "TSRO appears even in pure Ag metal". We fully agree that this statement is confusing.
After authors' careful discussions on the systematic results, we concluded that Ag perfect crystal should be disordered irrespective of loading, and deleted the corresponding paragraph. Hence, please kindly look at the Author reply to comment #1 raised by the Reviewer #2.

Major comment # 3
A few things need to be clarified in the introduction to improve the readability of the manuscript, though some of them are given later in the main text.
Author reply to comment #3 We thank the reviewer for the comment. Particular attention has been paid to the thorough overhaul throughout the manuscript to improve the readability of this manuscript. Furthermore, the arrangement of the main text was newly improved.

The following changes were made:
Please see the Introduction section on lines between 59~64 on page #3, "When compositionally homogeneous structures have specific low-coordinationnumber clusters, the preferential local ordering of principal elements generally dominates over the spatial order of a few nearest-neighbour spacings, i.e., short-range order (SRO), often called chemical SRO (CSRO) 22,23 .
Formation of diffusion-mediated CSRO domains in fully disordered alloys belongs to thermally activated isostructural disorder-to-order transition at short ranges." " It has been widely conjectured that localised planar slip and leading dislocations would destroy the pre-existing CSRO domains in a face-centred-cubic (fcc) phase upon loading, which corresponds to the so-called glide plane softening 31,32 . This effect drives the non-random system towards ideal random, referred as the

Major comment # 4
The authors did not clarify the reason for conducting tensile tests at different strain rates and at a low temperature of 77 K.
Author reply to comment #4 We thank the reviewer for the suggestion, we fully concur. We added the reasons for conducting tensile tests at different strain rates and at a low temperature of 77 K in the revised manuscript.

The following changes were made:
Please see the Results section on lines 162~163, page #7 "The έ values and loading temperature were intentionally chosen to minimise deformation-induced heating during tensile tests (see the Supplementary Notes 2 for details)." Also, see the newly added Supplementary Notes 2 in Supplementary Information 2. In this study, we conducted tensile tests at different strain rates and at 77 K.
Reasons for this are as follows. First, our previous TEM result suggested that lowering of tensile testing temperatures from 298 to 77 K leads to both deformationinduced order transition and resultant planar dislocation glide in boron-doped Fe40Mn40Cr10Co10 (at%) HEA 35 . However, it failed to explore the origin of straindriven SRO in the alloy. Second, reducing the tensile testing temperatures from 298 to 77 K can affect the SFE values of the alloy substantially, as SFE is a fun ction of temperature. In other words, varying the loading temperatures cannot decipher whether planar dislocation glide in fcc metallic alloys is either due to SRO or due to SFE. In fact, consensus has not been reached on the origin of planar dislocation glide, mainly due to the fact that both SRO and SFE have the strong influence on the glide mode of dislocations and associated hardening in glide softening-or hardening-dominated fcc solid solutions. Lastly, chemical SRO is usually a thermally activated or diffusion-mediated process. This implies that the formation of chemical SRO in M/HEAs is sensitive to loading rates and loading temperatures owing to deformation-introduced heating. Hence, we intentionally selected the tensile testing parameters, to minimise the loading temperatures on the low SFE value and to minimise the deformation-induced heating during tensile tests."

Minor comment #5
The authors did not clarify the concept of topological SRO.
Author reply to comment #6 We fully agree. We used the term "mechanically derived SRO (MSRO)" instead of "TSRO" in the revised manuscript. We clarified this term by using a better expression in the revised manuscript. Please kindly see the Author reply to comment #1 raised by Reviewer #2 for details.

The following changes were made:
Please see Discussion section on lines between 428 and 435, pages #18~19 "The precursor of MSRO-derived diffuse discs and spot-like scattering for the strained MPEAs was directly correlated with strain-induced crystal imperfections (dislocations and mechanical SFs), which were originally driven by a strain-induced atomic packing mismatch. Furthermore, from the viewpoint of mechanical response, the non-equiatomic quaternary alloy system exhibited no σYS dependence on the έ values. On account of these unique features, we herein regarded the localised SROs as mechanically derived SROs (MSROs) rather than diffusion-mediated CSROs that would be destroyed by planar dislocations during deformation." Please see lines between 456 and 463, pages #19~20 "To sum up, displacive MSRO belongs to isostructural disorder-order transition that occurs in slip bands during 77 K-tensile deformation (specifically, in the plastic strain regime). We assume that MSRO might be defined to be the degree of strain-induced local deviation from the average local-scale ordering in terms of either chemical or structural occupation. Unfortunately, a kinetics investigation of MSRO evolution has still not been performed. Further investigation to figure out how MSROs are mechanically stabilised with a higher έ, for example, at different low temperatures and levels of applied plastic strain, would be interesting."
Author reply to comment #6 We thank the reviewer for the correction.

Minor comment #7
Page 6, line 146. As mentioned by the authors, SRO already exited before straining.
Author reply to comment #7 We appreciate the hint and deleted the typos in the revised manuscript.

Minor comment #8
Page 7, line 160. "Final strain to sample failure" is a strain not an "elongation", as provided in the brackets.
Author reply to comment #8 We appreciate the hint and revised the typos.

The following changes were made:
Please see lines 166~167 on page #7, "increased the maximum tensile strength (σUTS) and total elongation (εtotal)."

Minor comment #9
Page 8, line 180. Due to the limited resolution of EBSD, the high density of nanotwins can sometimes be mis-indexed as HCP phase, as presented in Figure 2 and Extended Data Figure 3. This confusion directly undermines the argument that phase transformation was observed in the material. Highresolution TEM on the HCP stacking sequence along <110> zone axis is essential.
Author reply to comment #9 We thank the reviewer for the suggestion, and we fully concur that the high density of nanotwins can sometimes be mis-indexed. In fact, for the density measurements on nano-twins, we used HRTEM and STEM images not EBSD. For the thicknesses of mechanical twins with > ~40 nm, we used EBSD images. We revised the text accordingly.

The following changes were made:
Please see lines 194~200 on page #9,

Minor comment #11
Page 8, line 196. If the authors were to identify the slip system by using the Burgers vector of the dislocation slipping in that system, then it should be a/2<011>{111} instead of "a<011>{111}" Author reply to comment #11 Many thanks for this comment and we revised the typos.

The following changes were made:
Please see lines between 213 and 214 on page #9 "Using an appropriate zone axis, we found that active slip system was a/2<011>{111} (a is the lattice parameter) rather than a/2<100>{001} (Fig. 3d)."

Minor comment #12
Figure 2. The arrangement of the sub-figures is to be improved. In addition, the measurement on the Moiré-fringe was wrong, which should be deleted though, as it shows no practically useful information.
Author reply to comment #12 We appreciate the hint and deleted the typos.

The following changes were made:
Please see lines between 234 and 235 on page #10 The Moiré-fringes were thus shown along the <112>-type twin in the high-angle annular dark-field (HAADF) and bright-field HRTEM images ( Supplementary Fig.   5).

Minor comment #13
Page 33, line 847. It is necessary to explain why using in-situ heating test to introduce oxidation to the TEM lamellar when one can simply remove the sample from TEM and wait for it. Is it for the purpose of accelerating the oxidation process? If so, the piece of information should be included in the methods.
Author reply to comment #13 Many thanks for this helpful comment. As the reviewer recognized, diffracted spots are also introduced when the surface of thin TEM foils oxidises 48 . Hence, we should distingush the MSRO-derived spotlike scattering from the surface oxidation-caused spots. With an attempt to distingush the MSROderived spot-like scattering from the surface oxidation-caused spots, we further performed in-situ TEM heating experiments of the current HEA samples for accelerating the oxidation process.

Minor comment #14
Lastly, I would like to appreciate the vast amount of work the authors have done and included in the extended data.
Is this MSRO a unique phenomenon in Fe-Mn-Cr-Co? I assume no. It is well-known that dislocations and stacking faults massively appear when any ductile metal is plastically deformed. So is MSRO an alternative terminology for this well-known phenomenon? Lattice distortions are well-known in HEAs even for single crystals. Is this lattice distortion MSRO according to your definition?
Author reply to comment #1 We thank the reviewer for the kind support and valuable comments. We fully concur that dislocations and stacking faults massively appear when any ductile metal is plastically deformed. We, therefore, deleted "unique" term in the 2 nd revised manuscript. Instead, in order to avoid any confusing, we outline that this MSRO can form in multi-principal-element alloys with low stacking-fault energies (SFEs) upon load at 77 K. In other words, in this revision, we addressed that MSRO in the multi-principal-element alloys with low SFEs tends to form upon load at 77 K, which is driven by crystalline lattice defects (dislocations and stacking faults).
This is related to the following two facts: i) Dislocations and stacking faults in alloys with lower SFEs are more active compared to that in alloys with higher SFEs; ii) The lower temperature (i.e., 77 K) further reduces the SFEs of FCC-based alloys and hence intensify the activities of dislocations and stacking faults.
As suggested by the reviewer, lattice distortions are also well-known in many HEAs and MEAs.

Comment #3
It has been proposed that CSRO can be used to tailor the alloy properties in the fabrication process. On the contrary, it seems that MSRO is caused during the plastic loading. Then how to make use of the MSRO?
Author reply to comment #3 We thank the reviewer for the kind support and valuable comments. In the revised version, we addressed the importance of MSRO.

The following changes were made:
In abstract on page #2, "Underpinned by molecular dynamics, we show that MSRO in the alloys with low stacking-fault energies tends to form upon load at 77 K, which is driven by crystalline lattice defects (dislocations and stacking faults), offering new perspectives on the strain-induced ordering transition." Please see pages #21~22, "Thus, the current study can provide important insights into the fundamental, practical, and mechanistic understanding of many concentrated solid solutions that are designed to take advantage of the disorder-order transition during deformation at low temperatures."

Comment #4
A detail about the CSRO. In Fig. 1 (iii) The microdomain model. In this case, well-ordered regions (the microdomains) embed in a random matrix with the same composition.

7
(iv) The lattice defect model. Lattice defects such as dislocations and vacancies cause the variation of the degree of order and inhomogeneity of composition.
Considering the diverse CSRO definitions, we deleted the plotted purple circles for indicating the CSRO in our MC simulations (Figure 1). Figure 1 were made: I would like to thank the authors for their great effort in improving the draft. The refs they provided are also valuable. However, I remain confused about their definition of the MSRO.

The following changes in
I propose a way of clarifying. In Fig. 1, you compared the configurations with and without CSROs. This shows very clear to me what is CSRO. And it is consistent with my understanding based on the WC-SRO definition.
For the MSRO, could you please also show similar comparisons, i.e., the real-space atomistic structures with and without MSRO?
In the ref you mentioned, https://www.sciencedirect.com/science/article/pii/S135964622100703X they also talked about the mechanically driven SRO. But I can understand their results -the L1_0 ordering forms when the alloy is plastically deformed. But for your definition of MSRO, I am sorry I remain confused.

Manuscript NCOMMS-22-06614B
Dear Reviewers Here, we reply to the reviewers' comments on our manuscript (NCOMMS-22-06614B) entitled Mechanically derived short-range order and its impact on the multi-principalelement alloys. All the amendments are outlined below in more detail and highlighted in the revised version of our paper with blue colour highlighting.
With best regards on behalf of the author team,

Comment #1
For the MSRO, could you please also show similar comparisons, i.e., the real-space atomistic structures with and without MSRO?
Author reply to comment #1 We appreciate very much for the valuable comment. Following the reviewer' suggestion, we compared the configurations with and without the mechanical loading (obtained by Molecular Dynamics simulations). In fact, the result exhibits only marginal differences in terms of the 2 chemical distribution of principal elements before and after straining. This is not surprising, since the MD simulation is only performed for a very short period of time during which atomic diffusion cannot occur (the chemical ordering can be dealt by the prior Monte Carlo (MC) simulations overcoming the time-scale limitation). As already explained in the previous manuscript version, the strain-induced MSRO was observed during the deformation at very low temperature (77 K) and comparably higher strain rates which is not sufficient for redistribution of principal elements. This situation corresponds well to the usual condition of MD simulation. Figure 15 were made:

The following changes in Supplementary
much similarly to the previously reported TEM results of CSRO structure in VCoNi and CrCoNi MEAs. Based on this similar location of the ½{311} diffuse scattering by MSRO, it is highly plausible that real-space atomistic structure with the MSRO in the FeMnCrCo HEA structure is identical to that of CSRO in VCoNi and CrCoNi MEAs. We note that the CSRO is constructed by the L11-type structure motif in the fcc MEA structures [5]. In turn, the MSRO unit cell is likely to have L11-type structure motif, and the atomic occupation in MSRO during mechanical loading at 77 K is re-arranged as a result of diffusionless process, i.e., stacking faults and edge dislocations inside slip band. Hence, according to the concept of CSRO structure motif, we herein provide a schematic of the MSRO atomistic structure motif ( Supplementary Fig. 16). Based on the TEM-and MD-EDPs along different zone axes, we sketch the diffraction patterns caused by normal fcc spots and extra MSRO scattering ( Supplementary Fig. 16a). These EDPs described here agree well with those by fcc plus CSRO shown in the literature. Next, we constructed atomic projections of MSRO along different zone axes. Based on the TEM observations (Fig. 4e), where the measured interplanar spacing (