Doubled strength and ductility via maraging effect and dynamic precipitate transformation in ultrastrong medium-entropy alloy

Demands for ultrahigh strength in structural materials have been steadily increasing in response to environmental issues. Maraging alloys offer a high tensile strength and fracture toughness through a reduction of lattice defects and formation of intermetallic precipitates. The semi-coherent precipitates are crucial for exhibiting ultrahigh strength; however, they still result in limited work hardening and uniform ductility. Here, we demonstrate a strategy involving deformable semi-coherent precipitates and their dynamic phase transformation based on a narrow stability gap between two kinds of ordered phases. In a model medium-entropy alloy, the matrix precipitate acts as a dislocation barrier and also dislocation glide media; the grain-boundary precipitate further contributes to a significant work-hardening via dynamic precipitate transformation into the type of matrix precipitate. This combination results in a twofold enhancement of strength and uniform ductility, thus suggesting a promising alloy design concept for enhanced mechanical properties in developing various ultrastrong metallic materials.


Dear Reviewers,
We would like to submit the following revised manuscript for publication in Nature Communications, on behalf of all co-authors. This paper reports ultrahigh strength and good ductility in a tempered FeCo-based MEA owing to both metastability nature of the introduced hard precipitates at grain boundaries and the staking-fault-induced plasticity of relatively stable grain interior precipitates. TRIP effect has been frequently reported in ultrahigh strength steels (e.g., Aermet100) in which the thin metastable foil austenite has no downside to the strength. Nevertheless, strengthening by deformable hard precipitates was not reported before and may have certain importance for further development of ultrahigh strength materials. Overall, the methodology needs to improve substantially and some issues raised below must also be addressed.

Question #1-1>
The strength increment after the applied ageing is very high, exceeding 1000 MPa. The authors have estimated strengthening from different contributors. However, the models applied were not clearly elaborated. Since the paper reported that the precipitates can be deformed by SFs or by TRIP effect, which should be described by neither cutting nor Orowan looping. Is the strengthening response related to the critical stress and/or strain to initiate SF or TWIP? The authors showed that their energy is nearly identical, which means that one of them is highly unstable?
Reply #1-1> We appreciate the reviewer's helpful comment, allowing us to improve the explanation of strengthening mechanisms more clearly. We had regarded the hP24 precipitates as non-shearable ones due to their relatively large size (average length and diameter: ~160 nm and ~25 nm) for the 24H alloy. According to the reviewer's suggestion, we have performed additional TEM analyses to directly unravel the dislocation behaviours as shown in the revised As the reviewer pointed out, the formations of SF and metastability of L12 are crucial mechanisms in this work. However, they are more likely associated with work-hardening mechanisms and consequent plasticity, rather than directly determining the increment of yield strength. We observed that the interior hP24 precipitates show the glide of SFs within them 4 during deformation as well as the formation of Orowan loops with matrix dislocations.
Although the precipitates after ageing already contain dense SFs (see Fig. 2k and Supplementary Fig. 5), the FFT image of deformed precipitates shown in Fig. 4d reveals more evident pattern of the hP24 structure. This result indicates that partial dislocations in the precipitates can glide by applied stress, leading to the L12 to hP24 transformation. Therefore, it is concluded that the interior precipitates accompany dislocation glides, which contributes to plasticity as well as the precipitate strengthening via the Orowan mechanism.
As well as the formation of SFs in interior hP24 precipitates, the grain-boundary L12 precipitates also significantly contribute to work hardening by TRIP effect. It is expected that the contribution of L12 to the yield strength increment is relatively negligible because dislocations hardly interact with the grain-boundary precipitates before yielding in contrast to the grain interior precipitates. They can pile up at the L12 interfaces (i.e., the location near grain boundaries) in the later stage of deformation. Thus, partial dislocations motion and SF formation within the L12 precipitates result in the dynamic phase transformation into hP24, leading to a considerable work-hardening and large uniform ductility.
Moreover, the grain-boundary L12 precipitate is thought to be metastable rather than highly unstable. If the L12 precipitates are highly unstable, they are prone to phase transformation under stress levels before yielding (known as stress-induced transformation), which can hardly contribute to plasticity. It was confirmed that TRIP occurs after yielding (Fig. 4e,f), and the high critical stress for TRIP can be estimated by an increasing high work hardening rate after yielding in contrast to the SA sample (Fig. 3b). Fig. 4e,f clearly shows the gradual progress of deformation and resulting phase transformation during deformation: partial dislocations were emitted from the grain boundary, resulting in the extended SFs and precipitate transformation into hP24. Therefore, it can be concluded that the mechanical stability of the grain-boundary 5 L12 precipitate is enough to contribute to the high work hardening and prolonged plastic deformation. Therefore, the following sentences and relevant references have been added to support the explanation of strengthening mechanisms in the discussion part.

Discussion
-… shearing or Orowan bowing mechanism. The dark-field image in Fig. 4a shows the top view of the rod-shaped PPTs, where the direct observation of dislocationprecipitate interactions indicates that gliding dislocations bypass by the Orowan bowing mechanism. The strength increment from the precipitation strengthening of interior hP24 is estimated to be ~1030 MPa (see Methods and Supplementary Fig. 8 for detailed information). On the other hand, the strengthening contribution… -… it is observed that the interior hP24 PPTs show the glide of SFs within them as well as the formation of Orowan loops with matrix dislocations. Although the interior PPTs after ageing already contain dense SFs (see Fig. 2k and Supplementary Fig. 5), the FFT image of deformed PPTs shown in Fig. 4d reveals a more evident pattern of the hP24 structure. This result indicates that partial dislocations in the PPTs can glide by applied stress, leading to the well-defined highly faulted structure. Therefore, it is concluded that the interior PPTs accompany dislocation glides, which contributes to plasticity as well as the precipitate strengthening via the Orowan mechanism.
-… Plastic deformation introduces partial dislocations motion and SFs formation within the L12 that result in the dynamic phase transformation into hP24, leading to considerable work-hardening and large uniform ductility. It is confirmed that the TRIP occurs after yielding (Fig. 4e,f), and the high critical stress for TRIP can be estimated by an increasing high work hardening rate after yielding in contrast to the SA sample ( Fig. 3b) [39][40][41] . Question #1-2> It is well known that the stability of microstructure is also critical for cracking resistance of ultrahigh strength alloys. The authors characterized evolution of the nanoprecipitates after deformation in detail, but the information regarding deformed substructure of the matrix and that adjacent to the interfaces between the nanoprecipitates and matrix were not provided. While these substructures are definitely important for the understanding of both the strengthening and ductilization mechanism, which may also separate the contributions from interior precipitation and from grain boundary precipitates. The discussion in the current version is too simple, and both strengthening and ductilization mechanisms should be discussed more systematically.

Reply #1-2>
We appreciate the reviewer for providing constructive suggestions to unravel the deformation mechanism concerning the matrix. According to the reviewer's suggestions, we have conducted more TEM analyses as shown in the revised Fig. 4b,c (please see the figure set in Reply #1-1). Fig. 4b,c shows the deformed substructure of the matrix and that adjacent to the interfaces between the interior hP24 precipitates and matrix. High-density dislocations form homogeneous deformation substructures as indicated by a blue arrow, and the high fraction and small interspacing of hP24 precipitates lead to massive dislocation interactions in the matrix. The hP24 effectively acts as obstacles to dislocation motions. To further understand the strengthening and ductilisation mechanism, details are discussed as follows.
The significant strengthening and ductilisation of the present alloy are attributed to the following three dominant mechanisms: 1) dislocation behaviours in the matrix; 2) SF formation in interior hP24; 3) TRIP effect in grain-boundary L12. First, aforementioned, high-density dislocations generate homogeneous deformation substructures, leading to massive interfacial interactions with hP24 obstacles. Before ageing, the as-quenched SA alloy possesses high dislocation density initially due to inherent characteristics of martensitic transformations. This initial high-density tangled dislocation has limited capability of work hardening as represented in Fig. 3a,b. However, the ageing treatment (24H alloy) enables thermal recovery (dislocation annihilation and rearrangement) and increases the mean-free path of dislocations (MFP). In addition to the recovery, dislocations are consumed by formations of semi-coherent interfaces of hP24 and L12 as measured from 1.43×10 15 m -2 to 8.40×10 14 m -2 via XRD. This increased MFP allows uniform dislocation glides at a certain regular spacing (Fig. 4b,c) and consequent high ductility. However, it cannot be concluded that this mechanism is solely dominant due to 8 massive interior precipitates with the average interparticle spacing of ~58 nm, which limits the substantial increase of MFP. Nevertheless, the uniform dislocation glides and homogeneous deformation substructures prevent premature cracking in ultrahigh strength alloys.
In this regard, the deformation mechanisms of hP24 and L12, described in Reply #1-1, contribute to the enhanced work hardening and delayed plastic instability, resulting in both strengthening and ductilisation. It is reasonable that deformable hP24 and transformable L12 precipitates play an essential role in enhancing ductilisation, resulting in the doubled uniform elongation to 4% with the tensile strength of ~2100 MPa. However, the quantitative contributions of described three mechanisms are challenging to estimate due to their complex and interdependent effects. The following sentences have been added to discuss the strengthening and ductilisation mechanisms in the discussion part.
-Discussion. The increased strength and ductility of the present alloy are attributed to the following three dominant mechanisms: 1) dislocation behaviours in the matrix; 2) SF formation in interior PPT (hP24); and 3) TRIP effect in grain-boundary PPT (L12).
First, Fig. 4b,c shows the deformed substructure of the matrix and that adjacent to the interfaces between the interior PPTs and matrix. High-density dislocations form homogeneous deformation substructures as indicated by a blue arrow, and the high fraction and small interspacing of PPTs lead to massive dislocation interactions in the matrix. Before ageing, the as-quenched SA alloy initially possesses high dislocation density due to inherent characteristics of martensitic transformations. This initial highdensity tangled dislocation has limited capability of work hardening as represented in Fig. 3a,b. However, the reduction in dislocation density due the ageing treatment (24H alloy) increases the mean free path of dislocations (MFP). This increased MFP allows uniform dislocation glides at a certain regular spacing (see Fig. 4b,c) and consequent 9 high ductility. However, it cannot be concluded that this mechanism is solely dominant in strengthening and ductilisation due to massive interior PPTs with the average interparticle spacing of ~58 nm, which limits the substantial increase of MFP.
Nevertheless, the uniform dislocation glides and homogeneous deformation substructures contribute to preventing premature cracking in ultrahigh-strength alloys.
Question #1-3> In page 6, the authors defined the driving force for the formation of hP24, i.e., the free energy difference between the parent phase (bcc) and hP24 structure. The composition used for the calculation is the stoichiometric composition of the involved precipitates phase, rather than the alloy composition. As such, the DFT calculation might be able to explain origins of the TRIP effect of the metastable phase, but cannot reveal any direct relationship with the designed alloy compositions. That is, from alloy design point of view, these calculations are insufficient to provide any hints.

Reply #1-3>
We appreciate that the comment allows us to explain specific relationships of alloy compositions with precipitate compositions calculated by DFT. Our primary objective was to design a metastable ordered-structure phase as a precipitate. We targeted M3V-type (A3B-type with M: Fe, Co, Ni) structures reported from Liu et al. [Ref. 46], as the structures vary from ordered fcc to hcp depending on the electron concentration (e/a, valence electrons per atom) adjusted by M composition. Ni having the highest number of valence electrons was excluded from the candidate as it can result in high hexagonality and the absence of cubic structure. Therefore, we chose Fe and Co as candidates for the M site to target the metastable intermetallic phase between Co3V and Fe3V. Stoichiometric compositions from Co3V to Fe3V were considered in the DFT calculation, resulting in the lowest energy gap between L12 and hP24 structure at Co2Fe1V1. However, the calculated composition represents that of precipitate, which cannot reveal any direct relationship with the designed alloy compositions as the reviewer commented. We entirely agree with it. In fact, we had designed the alloy composition via CALPHAD approaches via thermodynamic calculations to form the aimed Co2Fe1V1 precipitates. At the first submission, we omitted details of thermodynamic calculations to construct a concise manuscript and also highlight the narrowest energy gap. The calculation details are as follows and have been added to the supplementary materials (Supplementary Fig. 2 and 3).
As aforementioned, three elements of Fe, Co, and V are the candidates for our strategy. The phase diagrams under fixed V content at 10 at% and 20 at% are shown in Supplementary Fig.   2. For the FexCo80-xV20 phase diagram, the sigma () phase is present along the Fe-rich regions, which is likely to cause embrittlement. In addition, sufficient Fe is necessary in order to obtain martensitic microstructure. Thus, the V content of 10 at% was considered to avoid brittle  phase and also to obtain aimed M3V phase. At 10% V, the next step was determining the proportion of Fe and Co. The calculations in Supplementary Fig. 3 demonstrate that more abundant Co leads to a massive fraction of M3V. Interestingly, the M3V phase has the composition aimed Co2Fe1V1 in the Fe50Co40V10 alloy composition. With increasing the nominal Co content in the alloy, the M3V phase possesses more Co content (>50 at%), which deviates from the targeted Co2Fe1V1. Thus, the Fe50Co40V10 was chosen as the alloy composition in order to obtain the aimed Co2Fe1V1 precipitate for M3V precipitate and also to form fully martensitic structures from a single fcc phase at a high-temperature range (>900 °C).
Based on the calculation results, the initial heat treatments were conducted at 1000 °C for 1 h and subsequent ageing at 550 °C for 24 h. We additionally modified the heat-treatment conditions to optimise the mechanical properties; the alloys were annealed at 900 °C for 10 min in order to obtain a smaller prior austenite (fcc) grain size and fully martensitic structure (bcc) after quenching. Thus, the annealing and ageing temperatures slightly deviate from the equilibrium temperature (920 °C and 546 °C, respectively).
In addition, we conducted additional DFT calculations with the actual composition of the precipitate to ensure the trend of the phase stability between the different crystal structures.
The stability of L12 and hP24 was confirmed to intersect between 550-600 K (exactly at 578 K), where L12 structure is more stable at ageing temperature and hP24 at room temperature, respectively. We have included the additional results in Supplementary Fig. 1 and Table 4. Therefore, the following sentences, figures, table, and reference have been added to explain the alloy design approaches.

Results
-Alloy design. …in varying compositions of (Fe,Co)3V1 (see details of computational methodology and additional data in Methods and Supplementary Fig. 1). … -…at typical temperatures for the ageing treatment. Then, we performed thermodynamic calculations to form the desired precipitates in the Fe-Co-V ternary system (see details in Methods and Supplementary Fig. 2 and 3). Thus, FeCo0.8V0.2 was selected in order to obtain fully martensitic microstructure after quenching as the matrix and form precipitates (PPTs) of the desired M3V phase (M: Fe, Co) after ageing without other phases causing embrittlement. To fabricate…

Methods
-Alloy design & fabrications. As aforementioned, three elements of Fe, Co, and V are the candidates for our strategy. To determine the alloy composition to embody the desired precipitates, we performed thermodynamic calculations based on CALPHAD approaches using Thermo-Calc software with a TCFE2000 database and its upgraded version 58-61 . The phase diagrams under fixed V content at 10 at% and 20 at% are shown in Supplementary Fig. 2. For the FexCo80-xV20 phase diagram, the sigma () phase is present along the Fe-rich regions, which is likely to cause embrittlement. As mentioned in the manuscript, however, sufficient Fe is necessary in order to obtain martensitic microstructure. Thus, the V content of 10 at% was considered to avoid brittle  phase, and also to obtain aimed M3V phase in the martensitic matrix. At 10% V, in determining the proportion of Fe and Co, calculation results in Supplementary Fig. 3 demonstrate that more abundant Co leads to a massive fraction of M3V. Interestingly, the M3V phase has the composition aimed Co2Fe1V1 in the Fe50Co40V10 alloy composition. Further increase of nominal Co content leads to more Co content (>50 at%) in the M3V phase, which deviates from the targeted Co2Fe1V1. Thus, considering the calculation results and the possibility of the formation of brittle sigma phase in the abundant V composition, we selected Fe50Co40V10 as a bulk alloy composition with relatively higher Fe and Co content compared to that of V. This composition is expected to obtain the aimed Co2Fe1V1 precipitate for M3V precipitate and also to form fully martensitic structures from single fcc phase at a high-temperature range (>900 °C). FexCo80-xV20.   Table 1).

Question #1-4>
The authors pointed out that boundary precipitates is L12 rather than hp24 because of the difference in both lattice misfit and variant amount. First, the close-packed plane for L12 and hp24 should be identical and they are both the parallel plane in K-S and Burgers OR. Therefore, I guess the misfit of the L12-matrix and hp24-matrix is very similar. Also, the authors should provide the basic data such as lattice parameter, used for the misfit calculation.
Second, Figure 2 shows that boundary precipitates mainly grow into one side of grain boundary with obvious orientation relationship. According to the description that they maintain semicoherent interfaces with both side matrix, the orientation relationship with the other side of grain should be also provided. Therefore, ℎ 24 is considered as lattice constant along a axis of hP24 and as √3fold of the lattice constant of bcc martensite.

Reply #1-4>
Regarding the orientation relationship (OR) between L12 and matrix at both side grains, our statement was misleading and should have been carefully examined. It has been wellestablished that a precipitate at grain boundary has a rational OR, e.g., K-S, with an adjacent grain, while the interface is incoherent with the other adjacent grain and is highly dependent on the grain boundary characteristics. However, it was also shown that precipitates at grain boundary become partially coherent by the formation of ledges and misfit compensating defects, as they try to reduce the increment of interfacial energy at most. Therefore, both the intragranular precipitates with a highly faulted structure and the intergranular precipitates showing K-S OR with an adjacent grain are expected to minimise the interfacial energy.
In this regard, we have conducted further TEM analyses to investigate the interface structure between grain-boundary precipitate and the other adjacent grain. As expected, the irrational OR was observed where BDbcc = <111> and BDL12 ~ <114>, respectively; however, the HRTEM image ( Supplementary Fig. 7) shows an interesting contrast from the phase interface with partially coherent bonding and its arrangement changes with the curved interface. A similar interface structure was observed in three different precipitates. This might be related to the formation of ledges and misfit compensating defects. It is also noted that the activation energy increased with increasing the tilt angle between the low-energy interface boundary and the original matrix grain boundary, when the tilt angle is below the critical value. The closest {111}fcc to the matrix grain boundary (~{011}bcc) was selected as a low-energy interface. The observed flat interfaces in the present study ( Fig. 2j,m) are parallel to the {011}bcc, and thus, it seems to be the low-energy interface, while the crystal structure of precipitate and matrix is different from the reference [Ref. 32]. Although further systematic studies are definitely required to fully understand the precipitation behaviour and the interface structure depending on grain boundary characteristics, which is beyond the scope of this study, we have revised the discussion on the precipitation behaviour and added relevant references as follows.

Results
-Microstructure and precipitation behaviour. …experimental findings. Notably, PPTs possess two different crystal structures depending on the nucleation sites. The interior hP24 PPTs develop Burgers OR with the matrix, while the grain-boundary L12 PPTs develop K-S OR with the matrix. Based on the ORs of each PPT, the measured lattice misfits exhibited a value of 0.46% for hP24 and 1.84% for L12. PPTs having low-energy interfaces, e.g., Burgers and K-S ORs, exhibit flat interfaces leading to rod-shaped morphology for hP24 and polygonal shape for L12 and both interfaces are parallel to the {011}bcc. It has been well-established that a PPT at grain boundary has a rational OR, e.g., K-S, with an adjacent grain, while the interface is incoherent with the other adjacent grain and highly dependent on the grain boundary characteristics 29 .
However, it was also shown that PPTs at grain boundary become partially coherent by formations of ledges and misfits compensating defects 30 , as they try to reduce the increment of interfacial energy at most. It was also noted that the activation energy increases with increasing the tilt angle between the low-energy interface and the original matrix grain boundary, when the tilt angle is below the critical value 31 . In other words, the closest {111}fcc to the matrix grain boundary (~{011}bcc) is selected as a low-energy interface 32 . Therefore, both the interior PPTs with a highly faulted structure and the grain-bounday PPTs showing K-S OR with an adjacent grain are expected to minimise the interfacial energy. While it is confirmed that L12 develops K-S OR with an adjacent grain (Fig. 2o), the interface structure with the other adjacent grain does not show exact K-S OR. As expected, the irrational OR was observed where the beam directions (BDs) were BDbcc = <111> and BDL12 ~ <114>, respectively; however, the HRTEM image ( Supplementary Fig. 7) shows an interesting contrast from the phase interface with partially coherent bonding and its arrangement changes 19 with the curved interface. This might be related to the formation of ledges and the misfit compensating defects to minimise the interfacial energy 30 .
To further explain the origin of the precipitation behaviour of the two phases, a selection of structure based on electron concentrations (e/a) of PPTs and their heterogeneous nucleation and growth were further considered. According to the theory based on e/a from Liu et al. 33 , the M3V phase begins to show hexagonality mixed with a cubic crystal structure over e/a of 7.89. The e/a value exhibits ~7.81 for our actual composition of the present PPTs, which is lower than the critical concentration forming hexagonality. Therefore, it is likely for the PPTs to form cubic ordered structure, i.e., L12. However, in the process of heterogeneous nucleation and growth of PPTs, those nucleating on dislocations are dominated by the strain-field effect where the lattice misfit becomes a critical factor 34 . The interior PPTs would accommodate numerous stacking faults and transform in the direction of the hP24 structure during growth to minimise the misfit (L12-bcc: 1.84% vs. hP24-bcc: 0.46%). The narrow energy stability gap between two phases also seems to allow the formation of SFs and local hP24 structure. On the other hand, the PPTs at grain boundaries consume the prior fcc grain boundaries and forms the semi-coherent interface with an adjacent grain to lower the interface energy. This well corresponds to a conventional heterogeneous precipitation mechanism at grain boundaries. When L12 forms a semi-coherent interface with an adjacent grain, it seems that the formation of SFs does not further reduce the energy, but the locally flat interfaces form with the other adjacent grain, resulting in the polygonal-shaped PPT aforementioned. Question #1-5> This paper mainly focused on semi-coherent precipitates. Given the large size, is the interface still semi-coherent?
Reply #1-5> We thank the reviewer for this question, allowing us to elaborate on the semicoherent interface of the precipitate. As mentioned in Reply #1-4, it is hard to determine that all precipitates maintain semi-coherent interfaces under prolonged ageing conditions. Most other maraging alloys are subjected to ageing less than 10 h; thus, precipitates in the present alloy are relatively larger compared to others. In other words, nanoprecipitates in pre-existing maraging alloys exhibit few nanometre sizes, while the interior hP24 precipitates have an average length and diameter of ~160 nm and ~25 nm, respectively. In order to unravel the uncertainty of interfaces, we have performed additional TEM analyses, as shown in HRTEM images in Supplementary Fig. 5, showing two precipitates intersecting one another with interfaces of the matrix/precipitates. The provided image shows an unclear and very diffused interface where it is hard to identify the exact boundary owing to the highly faulted structure of the hP24 precipitate. Although facing a hurdle in direct observation, the FFT images support and provide information that the interface is semi-coherent on <1 ̅ 11>bcc//<112 ̅ 0>hcp direction indicated by a yellow arrow. The following sentences and HRTEM image have been revised and added to support the semi-coherent interfaces of hP24.

Results
-Microstructure and precipitation behaviour. …because of the high density of SFs The overall pattern indicates Burgers orientation relationship in <1 ̅ 11>bcc//<112 ̅ 0>hcp direction.
As well as the additional TEM works, we have conducted further ageing treatment for 1 week to investigate whether the semi-coherent interfaces are still valid for very larger precipitates ( Supplementary Fig. 6). The average length and diameter of ~160 nm and ~25 nm of 24H alloy increase to ~300 nm and ~57 nm. Interestingly, although both length and diameter are expected to increase sufficiently, their increments are not considerable, resulting from the low-energy characteristics of semi-coherent interfaces. Besides, the flat facets are still maintained along the longitudinal side of hP24 and also at grain-boundary L12. Thus, this result 23 supports the retention of semi-coherent interfaces even for large precipitates to a certain degree.
It is worthwhile that the further aged alloy shows decreased strength but increased ductility, while their amounts are not significant despite the increase of ageing time from 1 day to 7 days.
The following sentence has been added to support the semi-coherent interfaces of hP24.

Results
-Microstructure and precipitation behaviour. …they have a diffused phase interface because of the SFs. These semi-coherent interfaces enable the PPT to maintain the nanometre size after further ageing up to 1 week ( Supplementary Fig. 6). Unlike the grain interior hP24, the PPTs along… - Supplementary Fig. 6 Question #1-7> Figure 3c, data for recently developed maraging steels should be included. It would be nice to discuss possible difference of underlying mechanisms involved.

Reply #1-7>
We appreciate the constructive comments to compare and highlight our approaches with existing ultrastrong alloys. The data for recently developed and also conventional maraging steels have been added in Figure 3c. The region of ultrastrong alloys, i.e., maraging steels group, has been additionally magnified in order to clarify the difference.
The tensile properties comparable to our work are reported in maraging steels by Jiang et al.
[ Ref. 2], which exhibits uniform elongation of 4%. The two alloys (Jiang's and this work) possess similar tensile properties; however, they show very different mechanisms. The primary work hardening mechanism in Jiang's work is slip-band refinements induced by extremely fine and nanosized shearable precipitates having coherent interfaces with minimal lattice misfit. In contrast, the present precipitates have relatively larger size and semi-coherent interfaces, which occurs the Orowan bowing mechanisms with matrix dislocations. Besides, deformable hP24 and transformable L12 precipitates play an essential role in both strengthening and ductilisation.
Therefore, the following sentences and figure have been revised and added relevant references as follows.

Results
MEAs, and maraging steels. The inset in Fig. 3c  27

Reviewer #2 (Remarks to the Author):
This article presents work assessing the maraging effect in a FeCo0.8V0.2 alloy. It comprises some interesting analysis and is a good demonstration of a joint modelling & experiment approach. Unfortunately, I'm not convinced of the significance of the phase transformation mechanisms proposed, nor that the alloy developed has particularly special mechanical properties. I have the following specific comments (in order of appearance in the text): Question #2-1> Introduction text: "preliminary" should be primarily?
Reply #2-1> We appreciate the reviewer for revising the expression more appropriately. The text has been revised as follows: -Introduction. … The technological importance of martensite primarily comes from its high strength based on hierarchy substructures… Question #2-2> To say that the technological importance of martensite originates from maraging is an odd statement. Most martensitic microstructures in use today (i.e., those in steels) are not age hardened, but rather just tempered (softened). It is tempering that allows martensite to be useful in most cases. The statement seems to contradict the statement later in the paragraph, which says that current maraging alloys aren't applied much (indicating that maraging isn't very technologically important).

Reply #2-2>
We agree with the reviewer's comment, and thank the reviewer very much for pointing out the odd statement. The maraging alloys are still technologically important, while those do not result from the inherent characteristics of martensite, as the reviewer commented, but they are associated with a special class of very low-carbon steels. Maraging alloys are not hardened by carbon or carbides but by precipitations of intermetallic compounds, which allows 28 for achieving combinations of high strength and toughness while maintaining relatively high ductility. Therefore, we have revised the following sentences and added reference. Please kindly refer to Supplementary Fig. 2 and 3, and Reply #1-3 for details of CALPHAD approaches. At the first submission, we omitted details of thermodynamic calculations to construct a concise manuscript and also highlight the narrowest energy gap. Based on the phase 30 diagram under fixed V content at 10 at% and 20 at%, FexCo90-xV10 was selected to obtain aimed M3V phase and also to avoid sigma () phase which is likely to cause embrittlement. In FexCo90-xV10 compositions ( Supplementary Fig. 2), sufficient Fe is necessary in order to obtain martensitic microstructure, while more abundant Co leads to a massive fraction of M3V.
Interestingly, the M3V phase has the composition aimed Co2Fe1V1 in the Fe50Co40V10 alloy composition. The M3V phase possesses more Co content (>50 at%) with increasing the nominal Co content, which deviates from the targeted Co2Fe1V1. Thus, the Fe50Co40V10 was chosen as the alloy composition in order to obtain the aimed Co2Fe1V1 precipitate for M3V precipitate and also to form fully martensitic structures from single fcc phase at a high-temperature range (>900 °C). Therefore, the following sentences to explain the design of alloy composition, precipitate composition, and overall microstructures have been revised as follows.

Results
-Alloy design. …in varying compositions of (Fe,Co)3V1 (see details of computational methodology and additional data in Methods and Supplementary Fig. 1).
-…at typical temperatures for the ageing treatment. Then, we performed thermodynamic calculations to form the desired precipitates in the Fe-Co-V ternary system (see details in Methods and Supplementary Fig. 2 and 3) Supplementary Fig. 2. For the FexCo80-xV20 phase diagram, the sigma () phase is present along the Fe-rich regions, which is likely to cause embrittlement. As mentioned in the manuscript, however, sufficient Fe is necessary in order to obtain martensitic microstructure. Thus, the V content of 10 at% was considered to avoid brittle  phase, and also to obtain aimed M3V phase in the martensitic matrix. At 10% V, in determining the proportion of Fe and Co, calculation results in Supplementary Fig. 3 demonstrate that more abundant Co leads to a massive fraction of M3V. interaction, which at the same time contributes to plasticity with partial dislocation gliding within them. Additionally, the L12 precipitates along the grain boundaries go through dynamic phase transformation, providing high work hardening and uniform ductility. Consequently, the complex and interdependent mechanisms result in an impressive combination of strength (~2 GPa) and uniform ductility (~4%). This kind of unique behaviour has never been reported before and therefore gives a novelty to this study. Therefore, the following sentences have been revised as follows.

Results
-Mechanical properties. …reported for ultrastrong precipitation-strengthened HEAs, MEAs, and maraging steels. The inset in Fig. 3c is the magnified region of ultrahigh strength maraging group to clearly distinguish the properties. When it comes to 34 ultrahigh strength metallic materials with strength reaching near 2 GPa, most of them show uniform ductility of less than 2%, while the present 24H alloy reaches 4%.

Discussion
-In summary, we demonstrate a design strategy resulting in an ultrahigh strength of ~2 especially aerospace products such as aircraft landing gear, missile casings, high-performance shafts in jet engines, coil springs, and bolts. Those applications require primarily high yield strength and ductility due to the margin of safety in service as described in Reply #2-5.
Accordingly, FeCo0.8V0.2 alloy was suggested as a model alloy system to implement the requirements with unprecedented metallurgical mechanisms. Details of element selections to obtain martensitic matrix and metastable precipitates are described in Reply #2-3.
As well as mechanical properties and associated strengthening-ductilisation mechanisms, the absence of carbon in the present alloy and consequent relatively ductile martensite before ageing exhibit good formability, enabling the homogenised cast ingot to gain a cold-rolling reduction of approximately 80%. In addition, no retained austenite is attainable when quenching to room temperature, which might be attributed to high Ms temperature due to high Co content. In commercial maraging alloys using TRIP effect, such as Aermet 100 and Aermet 35 the excessive retained austenite significantly decreases the yield strength. In this regard, the current alloy only needs simple heat treatment to obtain superior properties, i.e., solid-solution treatment and post ageing treatment, without any cryogenic treatment or deformation prior to ageing. Therefore, the following sentences have been added to the discussion part.
-Discussion. …TRIP effect in grain-boundary PPTs. These complex metallurgical mechanisms with simple heat treatment are suggested to implement high-performance and load-bearing application requirements. We expect these… Therefore, combining two ideas of (i) the initial crystal structure can be controlled through compositional configurations and (ii) the stacking faults induced by deformation are able to transform L12 into hexagonal ordered structures is the original concept. In other words, the transformation from L12 to hP24 structures was targeted by comparing the phase stability depending on the composition; Co3V, Co2Fe1V1, Co1.5Fe1.5V1, Co1Fe2V1, and Fe3V. The L12 phase possesses a crystal structure of ordered fcc with a stacking sequence of 'abcabcabc…' and the hP24 has an ordered-hcp structure with a stacking sequence of 'abcacbabc…' which exhibits a twin-like formation. Thus, shear displacements and stacking faults by deformation can induce dynamic precipitate transformation as the stacking sequence of the two phases are akin to one another, as illustrated in Supplementary Fig. 9.
Regarding the second comment, the dynamic precipitate transformation in the present study has a certain novelty because the strengthening through deformable hard precipitate has not been reported before. Similar to the concern in Question #2-12 regarding the work hardening of TRIP effects, the existence of a soft austenite phase in TRIP-assisted steels generally allows a dramatic increase in work-hardening rate and ductility. Although a high fraction of metastable austenite contributes to considerable plasticity, it also accompanies a decrease in yield strength, and thus this kind of microstructure cannot implement a class of ultrahigh strength steel. In 37 order to maintain the ultrahigh strength level and gain additional ductility, a fraction of the soft phase should keep a minimum, or its morphology should be tuned to possess high strength level and high mechanical stability. As for the represented case, Aermet100 contains 1-6% metastable austenite, which presents as thin foils with no downside to the strength. Although it is hard to exhibit a significant increment of work-hardening rate or ductility in a large scope such as general high-strength TRIP steels, it can contribute to preventing premature failure and improving toughness effectively. Therefore, it is worth mentioning that the TRIP effect in ultrahigh strength alloys exhibits different performances from the conventional high-strength TRIP steels.
In this respect, the present work exploits inherently hard intermetallic phases, which also have no downside to the strength, and their deformable and transformable characteristics provide a two-fold enhancement in strength and ductility via ageing. With the newly demonstrated mechanism, the authors envisage the dynamic precipitate transformation concept to be more developed for designing future structural materials. Please kindly refer to Reply #2-12 regarding the detailed plastic accommodation mechanisms of the present alloy. Therefore, the following sentences and relevant references have been added as follows.
-  during the quenching process. The substructure division also depends on the prior austenite grain size. The larger prior austenite grain size accommodates more strain change than the 40 small grain size and therefore requires more division of substructures. The example of this case is observable in Supplementary Fig. 4. Therefore, the following sentence has been revised to indicate the substructures.

Results
-Microstructure and precipitation behaviour. …consequent coarse-grained prior fcc phase ensure the characteristic martensitic structure, including packet, block, and lath substructures ( Supplementary Fig. 4). … Question #2-9> It is stated that "the proximity histogram across the PPT and bcc matrix ( Fig.   2h) implies that precipitation involving atomic diffusion was not yet completed." What aspect implies this is the case? The smooth transition in composition? And what is the reasoning?
Reply #2-9> We appreciate the comment regarding the ambiguous expression. We had stated that the atomic diffusion would not complete forming the aimed precipitates yet based on comparing precipitate compositions between the experimental and calculated results. For the reviewer's concern, we had intended the composition of precipitate, not the smooth transition of composition at the interface. In order to improve readability, the relevant sentence has been removed, and the following sentences have been revised in results part.

-Microstructure and precipitation behaviour. … An APT reconstruction of the 1H
alloy clearly reveals the very small size (width of ~4 nm) of the PPTs (Fig. 2h), whereas those in 24H alloy are determined to exhibit average length and diameter of ~160 nm and ~25 nm, respectively (Fig. 2p). The local lattice… Question #2-10> It is suggested that the tendency of one precipitate to nucleate at grain boundaries compared to the other can be related to the number of variants in the OR. This is an unusual suggestion -can the authors provide evidence of where this has been demonstrated before? It still seems unlikely that good correspondence would be found on both sides of the boundary very often, even when there are lots of possible variants.

Reply #2-10>
We are sincerely thankful for the question as it allows us to explain the mechanisms of multiphase precipitation more elaborately. Our suggestion was based on the idea that when a secondary phase nucleates at the boundaries, the interfacial energy has a critical impact on the driving force for nucleation along with the chemical driving force.
Precipitates tend to nucleate with the lowest interfacial energy with the surrounding matrix as possible constructing orientation relationship. As in the case of intergranular or interphase precipitates, they possess partially coherent interfaces as much as possible with adjacent matrix the growth process in order to minimise the misfit (L12-bcc: 1.84%, hP24-bcc: 0.46%). The narrow energy stability gap between the two-phase also seems to allow the formation of SFs and local hP24 structure. On the other hand, the precipitate at the grain boundaries consumes the prior austenite grain boundaries and forms the semi-coherent interface with an adjacent grain to lower the interface energy. This is a conventional heterogeneous precipitation mechanism at grain boundaries. When L12 forms a semi-coherent interface with an adjacent grain, it seems that the formation of SFs does not further reduce the energy, but the locally flat interfaces with the other adjacent grain were observed, resulting in the polygonal-shaped precipitate (Fig. 2g). Although the further systematic investigation is required for the fundamental understanding of precipitation behaviour, which is beyond the scope of this study, we have revised our manuscript as follows. However, it was also shown that PPTs at grain boundary become partially coherent by formations of ledges and misfits compensating defects 30 , as they try to reduce the increment of interfacial energy at most. It was also noted that the activation energy increases with increasing the tilt angle between the low-energy interface and the original matrix grain boundary, when the tilt angle is below the critical value 31 . In other words, the closest {111}fcc to the matrix grain boundary (~{011}bcc) is selected as a low-energy interface 32 . Therefore, both the interior PPTs with a highly faulted structure and the grain-bounday PPTs showing K-S OR with an adjacent grain are expected to minimise the interfacial energy. While it is confirmed that L12 develops K-S OR with an adjacent grain (Fig. 2o), the interface structure with the other adjacent grain does not show exact K-S OR. As expected, the irrational OR was observed where the beam directions (BDs) were BDbcc = <111> and BDL12 ~ <114>, respectively; however, the HRTEM image ( Supplementary Fig. 7) shows an interesting contrast from the phase interface with partially coherent bonding and its arrangement changes with the curved interface. This might be related to the formation of ledges and the misfit compensating defects to minimise the interfacial energy 30 .

Results
To further explain the origin of the precipitation behaviour of the two phases, a selection of structure based on electron concentrations (e/a) of PPTs and their heterogeneous nucleation and growth were further considered. According to the theory based on e/a from Liu et al. 33 , the M3V phase begins to show hexagonality mixed with a cubic crystal structure over e/a of 7.89. The e/a value exhibits ~7.81 for our actual composition of the present PPTs, which is lower than the critical concentration forming 44 hexagonality. Therefore, it is likely for the PPTs to form cubic ordered structure, i.e., L12. However, in the process of heterogeneous nucleation and growth of PPTs, those nucleating on dislocations are dominated by the strain-field effect where the lattice misfit becomes a critical factor 34 . The interior PPTs would accommodate numerous stacking faults and transform in the direction of the hP24 structure during growth to minimise the misfit (L12-bcc: 1.84% vs. hP24-bcc: 0.46%). The narrow energy stability gap between two phases also seems to allow the formation of SFs and local hP24 structure. On the other hand, the PPTs at grain boundaries consume the prior fcc grain boundaries and forms the semi-coherent interface with an adjacent grain to lower the interface energy. This well corresponds to a conventional heterogeneous precipitation mechanism at grain boundaries. When L12 forms a semi-coherent interface with an adjacent grain, it seems that the formation of SFs does not further reduce the energy, but the locally flat interfaces form with the other adjacent grain, resulting in the polygonal-shaped PPT aforementioned.
at least meet the properties found in this work (e.g., Aermet 340).

Reply #2-11>
We appreciate the helpful suggestion. We have included some conventional and recently developed maraging alloys as revised in Fig. 3c. Please kindly refer to Reply #1-7 for the details of explanations.
- Question #2-12> More discussion is needed on the mechanism proposed for the work hardening -presumably, the precipitate phase transformation leads to a harder precipitate than originally? How does this mean that more plastic deformation can be accommodated after the transformation (which is the next sentence)?
Reply #2-12> We appreciate the reviewer for this constructive comment, allowing us to discuss the work hardening mechanism in more detail. The conventional TRIP steels that accommodate phase transformation from fcc to bcc (or bct, ´ martensite) exhibit the following mechanisms contributing to the increased work hardening. The deformation accumulates strain energy in the soft fcc phase, which is later consumed for phase transformation. Then, the diffusionless transformation of fcc to bcc eventually leads to a harder phase, and the local hard phase prevents initiation of plastic instability. Furthermore, the volume expansion during the transformation creates geometrically necessary dislocations (GND) along the interfaces, which grants additional work hardening. Another example of TRIP effects is associated with fcc to hcp martensite transformation. Similar to twinning-induced plasticity (TWIP), partial dislocation glides on slip planes are known to be required for deformation-induced fcc to hcp transformation. Therefore, the fcc to hcp martensite transformation introduces phase boundaries including stacking faults, and thus, effectively reduces the mean free path of dislocation and even hinders the activation of secondary slip system due to the limited number of slip systems in the hcp structure. This mechanism, therefore, significantly contributes to work hardening through the dynamic Hall-Petch effect. For the case of our alloy, the characteristics of transformation consuming the accumulated strain energy and introducing phase interfaces are similar to the TRIP associated with fcc to hcp transformation. Furthermore, the cubic to hexagonal transformation in the A3B-type structure significantly increases frictional stress due to the interplanar-locking effects. The precipitates themselves deform by adopting partial dislocations and stacking faults, corresponding to the accommodating mechanism of plastic deformations. Therefore, unlike the conventional ordered precipitates or carbides that contribute to only strengthening, precipitates in the present alloy are able to implement both strengthening and ductilitisation mechanisms. Thus, the following sentences have been revised in the discussion part and relevant references have been added.

47
-Discussion. In this respect, the glide of partial dislocations enables L12 to hP24 precipitate transformation during deformation, introducing additional phase boundaries and SFs. The introduced interfaces effectively reduce the mean free path of dislocations and even hinder the activation of secondary slip systems due to the limited number of slip systems in the hcp structure 13,48 . This mechanism, therefore, significantly contributes to work hardening through the dynamic Hall-Petch effect.
The phase transformation also relieves the strain energy accumulated during tensile deformation, enabling further plastic deformation to be accommodated. Furthermore, the cubic to hexagonal transformation in the A3B-type structure is known to significantly increase frictional stress due to the interplanar-locking effects 33,49 .
Therefore, unlike the conventional ordered PPTs or carbides that contribute to only strengthening, PPTs in the present alloy are able to implement both strengthening and ductilitisation mechanisms. Notably, this… Reply #2-13> We sincerely appreciate the comment pointing out critical points, allowing us to discuss mechanisms more thoroughly. The grain interiors are important for work hardening, and the grain interior precipitations are preferred for effective precipitation hardening because the interior sites where most dislocations glide and interact with obstacles. However, the interior precipitations influence on reducing mean-free paths of dislocations, which in turn leads to the limited work hardening effect. Thus, usual precipitation affects the increment of strength rather than that of work hardening. Regarding the interactions between dislocations and interior hP24 precipitates, please kindly refer to Reply #1-2, explaining the massive formation of Orowan bowing loops around the precipitates. It has been demonstrated that the metastable austenite at the grain boundaries significantly contributes to work hardening indeed through the TRIP effect in those conventional steels.
Moreover, a few percent of metastable austenite (1-6%) can still contribute effectively, as observed in several studies of Aermet100, PH13-8 Mo, and Mn steels. Thus, the following sentences have been revised in the discussion part with relevant references added.
plastic deformation by phase transformations would prevent premature exhaustion of dislocation sources at the boundaries through the repeated generation of dislocation.
The current dynamic precipitate transformation at boundaries has similar effects to TRIP effects in Mn steels or quenching and partitioning (Q&P) steels that consist of metastable austenite at the boundaries 41,42 . It is well known that the dynamic phase transformation of the metastable phase in TRIP steels and alloys postpones plastic instability and thus enhances ductility and work hardening in a large scope 43,44 . The high fraction of metastable austenite contributes to considerable plasticity; however, it also accompanies a decrease in yield strength, and thus this kind of microstructure cannot implement a class of ultrahigh strength steel. In order to maintain the ultrahigh strength level and gain additional ductility, a fraction of the soft phase should keep a minimum, or its morphology should be tuned to possess high strength level and high mechanical stability. As for the represented case, Aermet100 contains 1-6% metastable austenite, which presents as thin foils with no downside to the strength.
Although it is hard to exhibit a significant increment of work-hardening rate or ductility in a large scope such as general high-strength TRIP steels, it can contribute to preventing premature failure and improving toughness effectively, as observed in several studies of Aermet100, PH13-8 Mo, and Mn steels 40,45,46 . Therefore, it is worth mentioning that the TRIP effect in ultrahigh strength alloys exhibits different performances from the conventional high-strength TRIP steels. In this respect, the present work exploits inherently hard intermetallic phases, which also have no downside to the strength, and their deformable and transformable characteristics provide a two-fold enhancement in strength and ductility via ageing.
- Reply #2-14> Thank the reviewer very much for the interesting question. The precipitates in conventional high-strength maraging alloys mostly contribute to only strengthening as efficient dislocation glide inhibitors through the shearing mechanism or Orowan looping mechanism.
Conventional high-strength maraging alloys possess precipitates, namely, NiAl (B2), Ni3Ti (D024), and M2C carbides in the case of secondary hardening steels. The strengthening mechanism depends on interdependent factors; precipitate size, morphology, and crystal structure. For example, the NiAl usually precipitates in a very small size having coherency with the bcc matrix. Owing to the small size (few nanometres) and coherent interfaces, the precipitates are shearable and therefore contribute to strengthening as a shearing mechanism.
On the other hand, the Ni3Ti has a different structure from the bcc matrix, which generally results in a larger size than NiAl and thus forms a semi-coherent interface with the matrix. The  Fig. 2 and 3).
Question #3-3> How the authors obtained the dislocation densities for 3 conditions heattreated alloys?
Reply #3-3> We appreciate the helpful comment. The dislocation densities had been obtained using a modified Williamson-Hall method based on the measured data from XRD analysis.  (310), were extracted to estimate dislocation densities of the matrix. We plotted K and KC 1/2 from the five representative bcc peaks data and fitted linear slope, quantitatively analysing the line broadening. The equation used for fitting the slope is as follows: where  is the diffraction angle,  is the wavelength of the X-rays, K is 2sin/, K is (2(2)cos/), D characterises the crystallite size, b is the Burgers vector, C is the average contrast factor of dislocations, and  is a constant depending on the effective outer cut-off radius of the dislocations. O stands for the higher-order terms where O The XRD data should move to the main text.
Reply #3-4> We sincerely thank the reviewer for the helpful suggestion. The XRD data have been moved to Fig. 2a as a part of microstructural characterisations and the following sentences have been added in results section.

Results
-Microstructure and precipitation behaviour. Question #3-5> The current status of the paper should be re-organized. For instance, some parts of the strengthening and plasticity mechanisms section should be the discussion part and 57 there is no scientific discussion in the current discussion part. The current discussion part looks like a summary part to me.

Reply #3-5>
We sincerely appreciate the suggestion that requires a re-organisation of discussions. We have thoroughly revised the discussion parts including strengthening, ductilisation, and plasticity mechanisms originating from matrix, hP24, and L12 precipitates.
The following discussions have been revised in the manuscript, with subheadings to help the reviewer's understanding (not provided in the manuscript).

Strengthening mechanism in yield strength increment
-Discussion. To elucidate the strengthening and ductilisation mechanisms by maraging effects, the deformed microstructures (tensile strained by 1%) of 24H alloy were investigated through TEM analyses as shown in Fig. 4. It is likely that fine PPTs significantly enhance the yield strength; either by shearing or Orowan bowing mechanism. The dark-field image in Fig. 4a  MPa as the alloy undergoes ageing for 24 h. The reduction in dislocation density is attributed to the combination of thermal recovery (dislocation annihilation and rearrangement) and consumption by forming PPTs with semi-coherent interfaces. As the ageing proceeds for 1 h and 24 h, the dislocation density of the alloy gradually decreases from 1.43×10 15 m -2 in SA to 9.49×10 14 m -2 in 1H and 8.40×10 14 m -2 in 24H.
These two different contributions have a counter effect; however, the large increase in 58 strength due to precipitation renders the decrease in strength due to the reduction in dislocations relatively negligible. However, this precipitation strengthening, on its own, cannot be a unique mechanism for the notable performance of the alloys examined in this study. The aged 24H alloy exhibits greater ductility, specifically uniform elongation, despite its ultrahigh strength level. First, Fig. 4b,c shows the deformed substructure of the matrix and that adjacent to the interfaces between the interior PPTs and matrix. High-density dislocations form homogeneous deformation substructures as indicated by a blue arrow, and the high fraction and small interspacing of PPTs lead to massive dislocation interactions in the matrix. Before ageing, the as-quenched SA alloy initially possesses high dislocation density due to inherent characteristics of martensitic transformations. This initial highdensity tangled dislocation has limited capability of work hardening as represented in Fig. 3a,b. However, the reduction in dislocation density due the ageing treatment (24H alloy) increases the mean free path of dislocations (MFP). This increased MFP allows uniform dislocation glides at a certain regular spacing (see Fig. 4b,c) and consequent high ductility. However, it cannot be concluded that this mechanism is solely dominant in strengthening and ductilisation due to massive interior PPTs with the average interparticle spacing of ~58 nm, which limits the substantial increase of MFP.

Mechanisms
Nevertheless, the uniform dislocation glides and homogeneous deformation substructures contribute to preventing premature cracking in ultrahigh-strength alloys.  Fig. 2k and Supplementary Fig. 5), the FFT image of deformed PPTs shown in Fig. 4d reveals a more evident pattern of the hP24 structure. This result indicates that partial dislocations in the PPTs can glide by applied stress, leading to the well-defined highly faulted structure. Therefore, it is concluded that the interior PPTs accompany dislocation glides, which contributes to plasticity as well as the precipitate strengthening via the Orowan mechanism.

Mechanisms for the enhanced work hardening and ductilisation III: dynamic phase
transformation of the grain-boundary L12 precipitate -Discussion. Thirdly, whereas the interior PPTs (hP24) possess dense SFs prior to deformation, the grain-boundary PPTs (L12) remain as defect-free states (Fig. 2n).
Plastic deformation introduces partial dislocations motion and SFs formation within the L12 that result in the dynamic phase transformation into hP24, leading to considerable work-hardening and large uniform ductility. It is confirmed that the TRIP occurs after yielding (Fig. 4e,f), and the high critical stress for TRIP can be estimated by an increasing high work hardening rate after yielding in contrast to the SA sample ( Fig. 3b) [39][40][41] . Figure 4e,f clearly shows the gradual progress of deformation and resulting phase transformation during deformation. Partial dislocations were emitted 60 from the grain boundary, resulting in the extended SFs and dynamic precipitate transformation into hP24. The presence of grain-boundary phases allowing to accommodate plastic deformation by phase transformations would prevent premature exhaustion of dislocation sources at the boundaries through the repeated generation of dislocation.
The current dynamic precipitate transformation at boundaries has similar effects to TRIP effects in Mn steels or quenching and partitioning (Q&P) steels that consist of metastable austenite at the boundaries 41,42 . It is well known that the dynamic phase transformation of the metastable phase in TRIP steels and alloys postpones plastic instability and thus enhances ductility and work hardening in a large scope 43,44 . The high fraction of metastable austenite contributes to considerable plasticity; however, it also accompanies a decrease in yield strength, and thus this kind of microstructure cannot implement a class of ultrahigh strength steel. In order to maintain the ultrahigh strength level and gain additional ductility, a fraction of the soft phase should keep a minimum, or its morphology should be tuned to possess high strength level and high mechanical stability. As for the represented case, Aermet100 contains 1-6% metastable austenite, which presents as thin foils with no downside to the strength.
Although it is hard to exhibit a significant increment of work-hardening rate or ductility in a large scope such as general high-strength TRIP steels, it can contribute to preventing premature failure and improving toughness effectively, as observed in several studies of Aermet100, PH13-8 Mo, and Mn steels 40,45,46 . Therefore, it is worth mentioning that the TRIP effect in ultrahigh strength alloys exhibits different performances from the conventional high-strength TRIP steels. In this respect, the present work exploits inherently hard intermetallic phases, which also have no 61 downside to the strength, and their deformable and transformable characteristics provide a two-fold enhancement in strength and ductility via ageing.
The underlying mechanism of dynamic structural changes from L12 to hP24 can be understood based on the stacking faults pair. As observed in Ni3(Al,Ti) precipitationhardened nickel-based alloys, L12 shearing by matrix dislocations can result in different structures of SnNi3-type D019, TiNi3-type D024, and VCo3-type hP24 47 . The stacking sequence of L12 is ABCABCA…, whereas that of hP24 is ABCACBA…, which exhibits twin-like formation. The stacking sequence of L12 can be changed to hP24 through the shear displacements of the type {111}1/3<112> with superlattice intrinsic (S-ISF) and extrinsic (S-ESF) stacking faults pair 47 . To further illustrate the sequence changes, a schematic drawing of the layers is provided ( Supplementary Fig. 9), where the modification of sequences via SFs is shown. Through <112>-type shear displacement of partial dislocations, S-ESF adds a C´ layer between A2 and B2, whereas S-ISF removes C2 layer from the sequence. As a result, the extrinsic/intrinsic stacking faults pair leads to a sequence from A1B1C1A2B2C2 to A1B1C1A2C´B2_A3 as ABCACBA…, which is that of hP24, i.e., the VCo3 type.
In this respect, the glide of partial dislocations enables L12 to hP24 precipitate transformation during deformation, introducing additional phase boundaries and SFs.
The introduced interfaces effectively reduce the mean free path of dislocations and even hinder the activation of secondary slip systems due to the limited number of slip systems in the hcp structure 13,48 . This mechanism, therefore, significantly contributes to work hardening through the dynamic Hall-Petch effect. The phase transformation also relieves the strain energy accumulated during tensile deformation, enabling further plastic deformation to be accommodated. Furthermore, the cubic to hexagonal transformation in the A3B-type structure is known to significantly increase frictional 62 stress due to the interplanar-locking effects 33,49 . Therefore, unlike the conventional ordered PPTs or carbides that contribute to only strengthening, PPTs in the present alloy are able to implement both strengthening and ductilitisation mechanisms.
Notably, this dynamic phase transformation and consequent ductilisation of ultrastrong alloys are attributed to the narrow stability gap of the desired multiple PPTs. The deformation mechanisms of the PPTs were sketched in Fig. 5 with microstructural evolutions during ageing.
Question #3-6> The target applications for developed M/HEAs are mostly high-temperature.
Have authors tried to investigate the mechanical behaviour at elevated temperature for the present alloy?
Reply #3-6> We appreciate very interesting suggestions. Thanks to the suggestion, we have additionally investigated high-temperature mechanical behaviours for the present 24H alloy, specifically at 500 °C at a strain rate of 10 -3 s -1 . The obtained stress-strain curves are as follows. the absence of an extensometer in the atmosphere chamber when measuring tensile strains.
Although the work-hardening ability seems to deteriorate compared to the room-temperature property, the alloy still shows a very high strength level of approximately 1600 MPa even at 500 °C. We would like to thank the reviewer for the valuable comments on making this interesting approach possible, and we will proceed with further and more diverse follow-up studies based on these results.

Dear Reviewers,
We would like to submit the following revised manuscript for publication in Nature Communications, on behalf of all co-authors. We cordially thank the reviewers for considering our work and providing helpful and valuable comments. We carefully revised the manuscript accordingly, and a detailed point-by-point reply to each item is enclosed below. Revised or added sentences have been highlighted in blue in the revised manuscript. Please kindly consider the attached reply and revised manuscript for further consideration.
Thank you for your consideration.
Sincerely yours, Seok Su Sohn Associate Professor, Korea University E-mail. sssohn@korea.ac.kr deformation. Besides, the KAM maps reveal that the dislocation density increases homogeneously throughout the entire grains, where massive dislocation-precipitate interactions occur in nanometre scale as presented in Fig. 4b.
Moreover, evolutions of basic deformation patterns were observed through SEM images in Supplementary Fig. 9c-f. Deformation occurs by forming parallel micro shear bands that pass through the grain interior without crack initiation at the precipitates, indicating the precipitate and matrix well accommodates plastic strain ( Supplementary Fig. 9c). At more strained region, more micro shear bands are distributed densely and homogeneously through entire grains ( Supplementary Fig. 9d). Then, the homogeneous deformation bands are severely deepened and eventually leads to crack initiation mostly at the prior fcc boundary ( Supplementary Fig.   9e). Finally, macro-scale observation on the fractured region exhibit that the matrix sufficiently accommodates deformation bands and thus homogeneous deformation substructures contribute in preventing premature cracking. In other words, the present precipitates can effectively accommodate plastic strains without initiating brittle cracks, which enables a highly sustained strain hardening rate. Thus, the following texts, figures, figure captions, and a reference have been added as follows.
Discussion: …Nevertheless, the uniform dislocation glides and homogeneous deformation substructures contribute to preventing premature cracking in ultrahigh-strength alloys as observed in Supplementary Fig. 9. Orientation gradients and geometrically necessary dislocations in ultrafine grained dual-phase steels studied by 2D and 3D EBSD. Mater. Sci. Eng. A 527, 2738-2746 (2010).