Ductile and scalable 3 GPa steel via a hierarchical microstructural architecture Authors: Ruming Geng, Yue Liu, Peiwen Tang, Xiaoyuan Yuan, Leilei Li, Yu Liu, Hanqi Wang, Lijing Zheng, Simin Lei, Yandong Sun, Yuxian Cao, Xinyang Li, Shun Han, Yong Li, Robert O Ritchie, Shiteng Zhao, Huibin Xu, Chunxu Wang Published: 2026 ABSTRACT AbstractUltrahigh-strength steels with yield strengths exceeding 3 GPa and appreciable ductility represent a transformative frontier in materials science. Conventional strengthening approaches, such as cold-work hardening, however, inevitably trade strength for total elongation, and thus are unable to deliver ductile 3 GPa steel. Here, we present a hierarchical microstructural architecture strategy applied to a designed maraging steel (Fe-16Ni-7Mo-1.6Ti-15Co-0.1Al, wt.%) that synergizes tailored precipitations with high-temperature severe plastic deformation. The multi-misfit nano co-precipitates, a high but spatially uniform density of dislocations, and ultrafine equiaxed grains are orchestrated into a hierarchical microstructure that delocalizes stress concentrations that commonly embrittle ultrahigh-strength materials and converts brittle cleavage fracture to ductile dimpling. Remarkably, the steel demonstrates 3 GPa yield strength with 47.5% reduction in area, 7.0% tensile elongation, and fracture toughness (KIc) of 26.0 ± 0.2 MPa·m1/2, which far surpasses existing 3 GPa steels. Moreover, as it exhibits an excellent fatigue strength of 1.24 GPa at ambient temperature and a yield strength of 2 GPa at 500 °C, which is achieved in industrially viable billets, this strategy unlocks scalable damage-tolerant components for extreme environments.Subject terms: Metals and alloys, Mechanical properties Introduction Steel remains the cornerstone of modern civilization, laying foundations for critical infrastructure, transportation, and industry. Its performance is evolving, especially for extreme load-bearing components where plastic deformation demarcates failure, making yield strength, not ultimate tensile strength, the critical design metric. Following nearly a century of advancement, the yield strength of high-strength alloys, particularly for aerospace applications, has now reached approximately 2 GPa1–5. However, the imperative for lighter, safer, and more efficient systems necessitates the paramount 3 GPa objective6,7, which is fundamental to unlocking next-generation technologies such as transmission shafts for space exploration vehicles. Crucially, achieving this ultra-high strength cannot come at the expense of damage tolerance. Operational safety further mandates acceptable resistance to cyclic fatigue and impact, measured by total elongation, reduction in area, and fatigue life. The pursuit of -strength steels with 3 GPa yield strength confronts a fundamental barrier, namely, the catastrophic loss of damage tolerance. For example, in the strength benchmarks such as AerMet 100 and AerMet 360, the yield strength rises from 1724 to 2245 MPa, while the total elongation drops sharply from 14 to 6.6% due mainly to profuse semi-coherent carbides causing stress concentration for fracture (https://www.carpentertechnology.com/alloy-finder). Similar results are observed in maraging steels where the yield strength of 18Ni(400) can be as high as 2.6 GPa with a very high volume fraction of intermetallic phase, whereas its ductility is constrained to 6%, as these intermetallic strengtheners are quite brittle8,9. Work hardening can further improve the strength of maraging steels to 3 GPa, yet its total elongation further deteriorates10,11. Cold-drawn pearlitic steel wire is another example of high-strength yet highly brittle materials. Its nano-scale layered pearlite structure drastically enhances the strength, as a large number of immobile dislocations are introduced and clustered around interfaces between ferrite and cementite12,13. However, when the ultimate tensile strength reaches 2.5 GPa, the sub ~10 nm interlamellar spacing loses its dislocation storage capability, and the corresponding total elongation drops to 2.5%14,15. More recent efforts have been focused on developing thermal-mechanical processing, such as severe rolling followed by tempering, to enhance the mechanical properties of steels. A series of ~2–2.4 GPa steels have been developed16–21. However, these methods usually require that the austenitic phase (as in TWIP and TRIP steels) undergo dislocation slip before martensite forms. This inherent characteristic poses a fundamental challenge in attaining the 3 GPa ultra-high yield strength. Moreover, visible lamellar microstructures formed along the rolling direction exhibit strong anisotropy, which is generally undesirable for industrial applications. Thus, the core challenge persists: as yield strength approaches 3 GPa, existing steels suffer a drastic loss in total elongation (below 3%), accompanied by a pronounced transition to brittle fracture. Overcoming this strength-ductility trade-off to achieve both 3 GPa yield strength and engineering-relevant ductility (e.g., elongation exceeding 5% and a reduction in area above 40%) remains an unmet goal. Nevertheless, it has been established that a hierarchical microstructural architecture involving the accumulation of dislocations, precipitates, and interfaces is essential for 3 GPa strength22–24. Yet conventional single-mechanism strengthening (e.g., nano-precipitates in maraging steels or dislocation walls in cold-worked wires) concentrates stress, promoting premature crack initiation. We tackle this challenge by synergistically regulating microstructures synergistically across different scales in a martensitic steel: (i) high density of homogeneously distributed nano co-precipitates to strengthen the material; (ii) a spatially uniform dislocation network to facilitate ductility without localized pile-ups and enable strain hardening, even at ultra-high stresses; (iii) equiaxed sub-micrometer scale grains with a high density of randomly-orientated high-angle grain boundaries to harden the steel through a strong Hall-Petch effect. Furthermore, these microstructural targets must be realized within a compositional design framework that ensures the absence of harmful solute segregation or brittle phase formation at grain boundaries, and full phase uniformity to prevent premature failure at soft or hard heterogeneities. This hierarchical microstructural architecture strategy minimizes dislocation mean free paths, approaching the theoretical limit, while ensuring exceptional microstructural homogeneity, thereby preventing local stress concentrations. The resulting steel achieves unprecedented mechanical properties of 3 GPa yield strength with 7.0% total elongation and 47.5% reduction in area, surpassing all existing 3 GPa materials. Moreover, this material exhibits a high fatigue strength (the fatigue limit at 107 cycles) of 1240 MPa and excellent high-temperature strength exceeding 2 GPa at 500 °C. Critically, we demonstrate scalability using industrially produced 30-mm-diameter bars (≥1 m in length), thereby validating real-world viability for safety-critical applications. Results and discussion Synergistic precipitation design To achieve ultra-high yield strength exceeding 3 GPa, a high density of nano-scale precipitates in a martensitic matrix is essential. As Al, Ti, and Mo are the most effective precipitate formers, the influence of different precipitate types on the strength and plasticity of our maraging steel (Supplementary Fig. 1 and Supplementary Table 1) was systematically investigated. In Ti-strengthened steel, the η-Ni3Ti phase forms semi-coherent interfaces with the martensitic matrix. However, titanium (Ti) exhibits a strong tendency to segregate at grain boundaries, leading to substantial η-Ni3Ti precipitation at these sites (Supplementary Fig. 2a); this grain-boundary precipitation adversely affects the plasticity of Ti-strengthened steel. In contrast, Al-strengthened steel demonstrates superior plasticity (Supplementary Fig. 1). Mo-strengthened steel offers a favorable strength-ductility balance. In a coupled strengthening approach, adding Al to Ti-strengthened steel did not suppress η-Ni3Ti precipitation at grain boundaries; instead, it increased precipitate size (Supplementary Fig. 2f). Overall, among these six steels, the Mo-Ti alloy exhibited a marked strength enhancement (~1.65 GPa) with an acceptable loss in total elongation (~7.5%) after aging (Supplementary Fig. 1c, d). Moreover, to ensure a fully martensitic microstructure after a cryogenic treatment at −73 °C, the martensite-start temperature (Ms) should exceed 140 °C, a temperature that is readily achievable with a dry ice and alcohol mixture, which is well suited for industrial implementation. Furthermore, by adjusting the weight fraction of Mo, Ni, Co, and Ti to 7%, 16%, 15%, and 1.6%, a balance between precipitation strength, full-melting temperature (Supplementary Fig. 3a), and Ms temperature (Supplementary Fig. 3b) is achieved. The microstructure in the pre-aging state also significantly influences the precipitation process. Based on the current Mo-Ti strengthened composition, we designed a high-temperature severe plastic deformation process to refine the microstructure. A holding temperature of 950 °C and a final deformation temperature above 800 °C, with a cumulative true strain of 10 or above, ensured a high dislocation density and ultrafine grains. The full thermal-mechanical processing routes are provided in the Supplementary Materials. We term the high-temperature server-plastic-deformation (SPD) processed materials as SD (Severely-Deformed) grade. For comparison, a Cold-Rolled (CR) grade was also produced. A large number of dislocations (Fig. 1a) and precipitations (Fig. 1a, c) are distributed within the SDA grade matrix. High-resolution transmission electron microscopy (HRTEM) imaging and energy-dispersive X-ray spectroscopy (EDS) analysis indicate that the precipitates comprise spherical ω phases and needle-like η-Ni3Ti (Fig. 1d, e). The ω phases formed by displacement fluctuations exhibit an ordered atomic lattice at the interface with the matrix (Fig. 1f), which will develop a sharp interface with further aging25. The η-Ni3Ti precipitates display a specific orientation relationship26 (Fig. 1f): [0001]η-Ni3Ti//[0–11]α, (11–20)η-Ni3Ti//(111)α. Three-dimensional atom probe tomography (3D-APT) (Fig. 1g–k) further indicates that the ω phases are primarily composed of Ni and Mo elements, whereas η-Ni3Ti consists of Ni and Ti with a trace amount of Mo. These two precipitates, with high number densities (4.14 × 1023/m3 for η-Ni3Ti and 3.36 × 1023/m3 for ω phases) and ultrafine radii (2.09 nm for η-Ni3Ti and 1.80 nm for ω phases), greatly contribute to the enhancement of strength. Compared with single-type precipitation, this co-precipitation strengthening strategy, involving precipitates with varying lattice misfits, not only effectively impedes dislocation slip under high-stress conditions but also serves as a broader source of individual dislocation initiation, thereby delaying local deformation and plastic instability (necking)27,28. Furthermore, the composition and morphology of the precipitates in the CRA sample were identical to those in the SDA sample (Supplementary Fig. 4). The high-temperature SPD process significantly refined the precipitate size and increased their number density, resulting in age-hardening with a yield strength enhancement of up to 2.1 GPa (Supplementary Table 2). Characterization of the multi-misfit nano-precipitates in the 3 GPa steel. a Low-magnification bright-field scanning transmission electron microscopy (STEM) image. b Local bright-field STEM image in (a). c Local dark-field STEM image in (a). d HRTEM image. e EDS maps of (d) for different elements (Ni, Ti, Mo, Fe) distributed in the HRTEM image. f The ω phases and η-Ni3Ti nano-precipitates viewed in the HRTEM image in (d). g Atomic distribution maps of Ni, Ti, Mo, Co, Fe, and η-Ni3Ti nano-precipitates and ω phases defined as the iso-concentration surfaces encompassing zones containing more than 40 at.% (Ni + Ti) and 15 at.% Mo, respectively. h Proximity histograms of η-Ni3Ti nano-precipitates. i Proximity histograms of ω phases. j Equivalent spherical radius distribution diagram of η-Ni3Ti nano-precipitates, the equivalent spherical radius, number density, and volume fraction of η-Ni3Ti nano-precipitates are 2.09 nm, 4.14 × 1023 m−3, and 1.4%, respectively. k Equivalent spherical radius distribution diagram of ω phases, the equivalent spherical radius, number density, and volume fraction of ω phases are 1.8 nm, 3.36 × 1023 m−3, and 0.8%, respectively. Ultrafine and homogeneous microstructure Alloy design involving precipitation hardening can effectively strengthen the steel, yet it alone cannot meet the 3 GPa target. Grain refinement to sub-micrometer scales is resorted to further enhance the strength of metallic materials29,30. EBSD analysis highlights a key microstructural distinction: cold-rolled (CR) grade exhibits pronounced grain elongation along the rolling direction (RD)31 (Supplementary Fig. 5c), whereas the severely deformed (SD) grade shows a remarkable isotropic microstructure between the normal direction (ND) and RD (Fig. 2a). Furthermore, the matrix of the SPD-processed and aged (SDA) grade consists almost entirely of martensite, with minimal retained austenite (Fig. 2b). This severe plastic deformation process successfully produces equiaxial ultrafine grains (Fig. 2c), and concurrently generates a high density of high-angle grain boundaries (Fig. 2d). Critically, the nature of the grain boundaries differs significantly. Cold-rolled (CR) grade contains abundant low-angle grain boundaries within the grains, formed by dislocation substructures (Supplementary Fig. 5c, d). Compared to the cold-rolled and aged (CRA) grade, SDA grade displays much finer grains (average effective size: 0.53 μm) and a higher density of larger high-angle grain boundaries (80.4%) (Supplementary Fig. 5e, f). The microstructures of the 3 GPa steel subjected to severe plastic deformation. a Reconstructed three-dimensional inverse pole figure map and the corresponding RD, ND, and TD directions, the white areas on the specimen are unresolved regions caused by excessive stress concentration. The EBSD measurement was performed with a step size of 0.03 μm. b Reconstructed three-dimensional phase map (the red and white regions represent martensite and austenite, respectively). c Inverse pole figure map parallel to the rolling direction. Inset in (c) is the grain size distribution map; the average grain size of the SDA grade is shown to be 0.53 μm. d Kernel average misorientation (KAM) and grain-boundary maps parallel to the rolling direction. Inset in (d) is the KAM values distribution map; the average KAM value of the SDA grade is shown to be 0.43°. XRD analysis (Supplementary Fig. 6) indicated that the dislocation density introduced through high-temperature severe plastic deformation is as high as that introduced through cold rolling, measuring 1.05 × 1016/m² and 1.14 × 1016/m², respectively. A Kernel average misorientation (KAM) diagram indicates that cold rolling introduces high-density dislocations and causes stress concentration at the low-angle grain boundaries formed along the rolling direction (Supplementary Fig. 7a–f). In contrast, the SPD-processed (SD) grade exhibits significantly lower stress concentrations (Fig. 2d; KAMave is 0.43° for SDA vs. 0.74° for CDA), especially at the interface, thereby reducing the susceptibility to interface cracking. This also indicates that high-temperature SPD introduces a more homogeneous spatial distribution of dislocations, corresponding to a much higher age-hardening (2.08 GPa for SDA vs. 1.66 GPa for CRA) despite a nearly identical dislocation density. which explains the enhanced age-hardening response in the SDA specimen despite a nearly identical dislocation density. Exceptional mechanical properties The precise precipitation design and exceptional microstructural control translate directly to superior mechanical properties. Figure 3a shows the engineering stress-strain curves of the two grades after cryogenic treatment and aging. The SPD-processed and aged (SDA) grade achieves ultra-high strength (>3000 MPa) while retaining exceptional ductility. Specifically, the SDA grade exhibits an ultimate tensile strength of 3009.5 ± 12.5 MPa, a yield strength of 2977 ± 1 MPa, a total elongation of 7.0 ± 0.25%, and a reduction in area of 47.5 ± 1.5%. The cold-rolled and aged (CRA) grade has a comparable ultimate tensile strength (2998 MPa) but significantly reduced ductility (a total elongation of 2.2 ± 1.2% and a reduction in area of 23.5 ± 1.5%), similar to commercial 3000 MPa-grade cold-drawn pearlitic steel wires and other maraging steels. Furthermore, a counterpart alloy without SPD processing (hereafter denoted as the coarse-grained (CG) grade) was fabricated. Figure 1 presents its mechanical properties and microstructure, demonstrating that the SPD process enhances strength by approximately 500 MPa (3009.5 MPa for SDA vs. 2455 MPa for CGA), accompanied by a minor reduction in total elongation (7.0% for SDA vs. 7.8% for CGA). Mechanical properties of the 3 GPa steel. a Engineering stress-strain curves of SDA and CRA grades; insert: tensile sample dimension. b Staircase diagram of the fracture toughness of the SDA grade compared with those of other reported steels (including low alloy UHSSs (https://www.carpentertechnology.com/alloy-finder), maraging steels (https://www.carpentertechnology.com/alloy-finder), secondary hardening steels (https://www.carpentertechnology.com/alloy-finder)28, and ultra-high strength stainless steels (https://www.carpentertechnology.com/alloy-finder)). c Staircase diagram of the fatigue limit strength of the SDA at a stress ratio of 0.1 and a frequency of 100 Hz. d Room temperature tensile properties of our steel compared with those of other reported steels (including Medium Mn steels16–18,23, low alloy UHSSs (https://www.carpentertechnology.com/alloy-finder)44, secondary hardening steels (https://www.carpentertechnology.com/alloy-finder)45–49, nano bainitic steels50, maraging steels (https://www.carpentertechnology.com/alloy-finder)4,10,51–53, ultra-high strength stainless steels (https://www.carpentertechnology.com/alloy-finder)11,54,55, cold-draw pearlitic steels14, and Q&P steels56). e Industrial scale product image of SD grade. The key to achieving such an impressive ductility in our 3 GPa steel lies in the ultra-high density of co-nano-precipitates, ultrafine microstructures, and mobile dislocations. The latter two features are primarily governed by the deformation process, with the aging process exerting negligible influence. Consequently, the microstructural and mechanical characteristics prior to aging were additionally characterized. Supplementary Fig. 8a further presents the engineering stress-strain curves of the SDC and CRC grades before aging. As their tensile strengths are similar, the SDC grade exhibits a lower yield strength-to-ultimate tensile strength ratio (0.70 for SDC vs. 0.90 for CRC) and better work-hardening ability (Supplementary Fig. 8b). Regarding uniform elongation, the SDC grade achieves 2.2%, representing a 69% increase over the CRC grade. A significant layered microstructure and dislocation walls were formed after cold rolling (Supplementary Fig. 9c), consistent with the absence of dislocation entanglement in the SDC grade (Supplementary Fig. 9a), indicating that a higher proportion of mobile dislocations was introduced by the high-temperature SPD. While subsequent aging triggers massive precipitation, which sacrifices uniform elongation for the 3 GPa yield strength (Fig. 3a), the ability to accommodate non-uniform deformation is the key to enhancing the ductility of 3 GPa steel. Figure 3d further compares the yield strength and total elongation of the 3 GPa steel with those of other ultra-high-strength steels. Steels with yield strengths exceeding 2.7 GPa typically exhibit very limited total elongation (<3%) and are susceptible to brittle fracture. In contrast, our 3 GPa steel achieves a favorable balance of yield strength and total elongation. Moreover, a high non-uniform elongation implies strong damage-tolerance capability. The plane-strain fracture toughness (KIC) is measured at 26.0 ± 0.2 MPa m1/2, offering significant advantages over existing steels (Fig. 3b). The fatigue performance was also evaluated in uniaxial tension with a stress ratio (ratio of minimum to maximum stress) of R = 0.1, which generated a measured fatigue strength (the fatigue limit at 107 cycles) of 1240 MPa (Fig. 3c). Compared to other reported maraging steels, this novel developed 3 GPa steel exhibited the highest fatigue performance (Supplementary Fig. 8d). Moreover, this material also shows excellent high-temperature performance, with its ultimate tensile strength exceeding 2 GPa under 500 °C with good ductility (Supplementary Fig. 8c, e, f). Crucially, all of these mechanical properties were obtained from a representative bulk sample of the 3 GPa steel, measuring 30 mm in diameter and 1 m in length (Fig. 3e), which underscores the great potential for industrial applications. Plastic deformation mechanism Following elastic deformation, the material undergoes plastic deformation in two stages: (i) uniform plastic deformation (from yield to peak stress), and (ii) non-uniform plastic deformation (from peak stress to fracture). Numerous nano-precipitates formed in aging markedly restrict dislocation movement. Both the SDA and CRA grades reach their fracture strength shortly after yielding, corresponding to low uniform elongations (Fig. 3a), during which dislocation slip governs plasticity. During the high-temperature SPD process, it is postulated that mobile dislocations are generated and stored in the materials, which can serve as carriers of plasticity. As a result, the SPD-processed and cryogenic (SDC) grade exhibits a low yield strength-to-ultimate tensile strength ratio, indicating high work-hardening capacity (Supplementary Fig. 1). By comparison, traditional cold rolling produces more dislocations and decreases their mobility, thus leading to reduced work hardening ability14. The engineering stress-strain curves of both the SDA and CRA grades (Fig. 3a) reveal that non-uniform deformation governs plastic deformation. However, the total elongation of the SDA steel is almost twice that of the CRA (and other reported 3 GPa steels, Fig. 3d), demonstrating a sound damage tolerance. Fractography of the SDA grade exhibits pronounced necking with extensive fibrous zones (Fig. 4a) and uniform equiaxed dimples (Fig. 4b, c), confirming ductile failure via microvoid coalescence. In contrast, the cold-rolled and aged (CRA) grade exhibits mixed-mode fracture with shear dimples and cleavage planes (Fig. 4g–i), accompanied by computed-tomography (CT)-observed laminar cracks that coalesce into through-thickness fractures (Fig. 4j–l). Fracture surface morphology of the experimental steel. a Computed tomography (CT) image of the SDA tensile sample after fracture (ductile). b Low magnification SEM image corresponding to the fracture surface in (a). c Local SEM image corresponding to (b). d Sampling position of transmission Kikuchi diffraction (TKD) tests from SDA tensile samples. e Inverse pole figure (IPF) map corresponding to the region in (d) (the white areas on the specimen are unresolved regions caused by excessive stress concentration). f High-angle (HAGB) and low-angle (LAGB) grain boundary map corresponding to the region in (d); insert: bright-field STEM image in (e). g CT image of the CRA tensile sample after fracture (brittle). h Low magnification SEM image corresponding to the fracture surface in (g). i Local SEM image corresponding to (h). j CT image corresponding to the fracture surface in (g). k CT image corresponding to cross-sectional view 1.5 mm from the fracture surface in (g). l CT image corresponding to the lateral view in (g). m Inverse pole figure (IPF) map corresponding to the yellow region in (i) (the white areas on the specimen are unresolved regions caused by excessive stress concentration). n High-angle and low-angle grain boundary map corresponding to the yellow region in (i); insert: bright-field STEM image in (k). Microvoid-mediated ductile fracture initiates at stress concentrations such as dislocation pile-ups at interfaces32–34. KAM mapping demonstrates pronounced deformation localization adjacent to low-angle boundaries in the cold-rolled and aged (CRA) grade, observed both parallel and vertical to rolling direction (Supplementary Fig. 7a–f), where interfacial stress concentration is expected due to the high density of geometrically necessary dislocations. Under tensile conditions, this stress localization accounts for CRA’s relatively brittle morphology dominated by cleavage. In sharp contrast, high-temperature SPD produced a low-stress, equiaxed-grained microstructure in the SDA grade (Fig. 2d), which prevents stress concentrations and thus avoids cleavage cracking along the interface under tensile loading. Transmission Kikuchi diffraction (TKD) analysis of material near the fracture surface of SDA grade (Fig. 4d) shows elongated grains (Fig. 4e) where a high-density of high-angle grain boundaries deflect cracks, creating tortuous propagation paths (Fig. 4f). This contrasts sharply with the linear crack propagation in the CRA specimens (Fig. 4m, n). In conclusion, our work demonstrates the successful realization of a ductile and scalable 3 GPa steel through hierarchical microstructural architecture. The exceptional ductility originates from a fundamental transformation in fracture micromechanics: the homogeneously distributed nano co-precipitates and the high-density spatially uniform dislocations within our steel facilitate vital stress relaxation through plastic deformation at crack tips35,36, preventing the localized stress accumulation and cleavage failure observed in conventional counterparts37. Critically, the profuse equiaxed sub-micrometer scale grains with a high density of high-angle boundaries act synergistically, forcing crack paths to zigzag and substantially amplifying energy dissipation during microvoid growth38. This deliberate orchestration of hierarchical microstructural architecture, simultaneously achieving refined grains, high densities of mobile dislocations, and stress-relieving interfaces, overcomes the intrinsic brittleness barrier at the 3 GPa strength threshold. Furthermore, the tailored microstructure should be grounded in a compositional design framework that ensures phase uniformity and suppressed harmful grain-boundary segregation or brittle phases, thereby preventing premature failure at heterogeneities. This hierarchical microstructural architecture strategy provides a universal and scalable pathway, leveraging industrially viable high-temperature severe plastic deformation, to push structural materials towards their theoretical performance limits. Methods Synergistic precipitation design experiments In order to investigate the influence of precipitate types on strength and ductility, six different grades of experimental steels were designed, with their chemical compositions detailed in Supplementary Table 1. The steels were melted in a 25 kg vacuum induction furnace and cast into ingots. These ingots were homogenized at 1220 °C, then forged into 35 mm square billets. After air cooling, the billets were held at 1080 °C and forged into 15 mm diameter rods. Tensile specimens were taken from these rods. Each specimen was subjected to a heat treatment consisting of solution treatment at 840 °C for 1 h, followed by oil quenching, deep cryogenic treatment at −73 °C for 2 h, followed by air cooling, and aging at 500 °C for 5 h, followed by air cooling. Samples preparation Alloy ingot with a nominal composition of Fe-16Ni-7Mo-1.6Ti-15Co-0.1Al (wt. %) was prepared by a three-ton vacuum-induction furnace and a vacuum-consumption furnace, prior to being cast into ingots with a diameter of 410 mm. All raw materials employed for smelting were high-purity grade, with Fe, Mo, Ni, Co, and Ti introduced as pure metals (each with a main component purity exceeding 99.9%). The as-cast ingot was first subjected to homogenization treatment at 1220 °C and then forged into a 100-mm diameter bar using upsetting and drawing repeated three times, with a finished forging temperature above 900 °C. After cooling to room temperature, the bar was subsequently annealed at 650 °C. To achieve an strength of 3 GPa, a server-plastic-deformation (SPD) process was designed. Following air cooling to room temperature, the bar was held at 950 °C over 1 h to ensure thorough and uniform heating and forged into a round rod with a diameter of 30 mm, as the final forging temperature was over 800 °C to avoid the intermetallic compounds forming. Over 3 times, piling and pulling were carried and the forging ratio exceeded 10 in the final pulling step. Multiple sequences were completed after heating one time and then immediately water cooling to obtain ultrafine grains and high-density dislocations. As cold deformation is currently the main deformation process used and reported to achieve 3 GPa strength. We also prepared cold-rolled samples with a total deformation of 70%. Due to the high strength of the material and the capacity limits of the available rolling equipment, the initial thickness of the samples was set at 20 mm, which was reduced to a final thickness of 6 mm. This reduction was achieved through 28 rolling passes, with an approximate reduction of 0.5 mm per pass. Specimens cut from the bars were subjected to a cryogenic treatment at −73 °C for 2 h and aged at 500 °C for 5 h. Samples after cryogenic treatment and after aging treatment were all analyzed; these are separately named as “SPD-processed and cryogenic” (SDC) and “SPD-processed and aged” (SDA) grade samples, and “cold-rolled and cryogenic” (CRC) and “cold-rolled and aged” (CRA) grade samples. Samples without the SPD process were named as coarse-grained (CG) grade. Room temperature tensile tests were performed on a universal testing machine (GNT50, NCS, China) at a constant strain rate of 2.5 × 10−4/s, using standard cylindrical specimens of stretch section with 3 mm diameter and 15 mm gauge length. Uniaxial tensile tests were carried out at room temperature following the specifications of the Chinese National Standards (GB/T 228.1). To ensure statistical reliability, each test condition was repeated twice. Fracture toughness tests were conducted using a multifunctional testing machine (MTS 370.10, 100 kN, MTS, USA) on specimens with dimensions of 10 × 20 × 100 mm. High-temperature tensile tests were conducted using a multifunctional testing machine (MTS 370.10, 100 KN, MTS, USA) equipped with a high-temperature environmental furnace. The high-temperature tensile tests were performed at temperatures ranging from 100 to 600 °C, with strain control applied until the maximum load was reached. At that point, the sample was subjected to displacement control applied until the sample fractured. The strain rate and displacement rate were 2.5 × 10−4/s and 6.25 × 10−3 mm/s, respectively. The strain variation after exceeding the maximum load was calculated by fitting a linear relationship between displacement and strain prior to reaching the maximum load. These specimen dimensions were also cylindrical, 65 mm long and 3 mm in diameter, with a gauge length of 15 mm. The mathematical description of maximum elongation at fracture (FE) and reduction in area (RA) can be expressed by the following equations: where LO represents the original length of the gauge section, LF represents the length of the gauge section after fracture, SO represents the original area of the gauge section, and SF represents the area of the gauge section after fracture. Room-temperature high-cycle fatigue tests were conducted using an electromagnetic resonance high-frequency fatigue testing machine (QBG 100, CCQB, China), at a frequency of ~100 Hz and a corresponding stress ratio (the ratio of minimum to maximum applied stresses) of 0.1, employing cylindrical specimens 5 mm in diameter with a length of 100 mm. The fatigue strength is referred to as the fatigue limit at 107 cycles. Microstructure characterization Microstructural analysis was performed using a field emission scanning electron microscope (SEM5000, CIQTEK, China, 30 kV) equipped with an electron backscatter diffraction (EBSD) detector (Oxford Symmetry2, Oxford, England). For EBSD analysis, the samples were electropolished for 10 s in a 10 vol.% chromic acid solution. The step size was set to 0.03 μm for the SD and CR grades, and 0.15 μm for the CG grade. The acquired EBSD data were analyzed and processed using AZtecCrystal software. Precipitates were analyzed using transmission electron microscopy (TEM; Talos F200X, Thermo Fisher, USA) equipped with EDS (Spuer X, Thermo Fisher, USA) and transmission electron microscopy (TEM; Thermo Spectra Ultra, Thermo Fisher, USA) equipped with EDS (Ultra X, Thermo Fisher, USA). TEM samples for the Talos F200X were mechanically ground to approximately 50 μm and electropolished for about 40 s at −15 °C using a 6 vol.% perchloric acid alcohol solution. TEM samples for Thermo Spectra Ultra were prepared by the SEM microscope equipped with a Ga focused-ion-beam (FIB; 30 kV, Thermo Scientific Scios 2, Thermo Fisher, USA) and an omniprobe manipulator. The pieces were initially pre-cut from the bulk samples by using a current of 7 nA; ion beam currents from 0.5, 0.3, 0.1 nA to 10 pA were then used in sequence to further mill the piece into electron-transparent slices with a thickness of 70 nm. Dislocation density measurements were performed using X-ray diffraction (XRD; D8 Advance, Bruker, Germany) with a Co target (λ = 1.79 Å, 35 kV, 40 mA). The XRD scan range was from 45° to 115°, with a scanning speed of 2°/min. The samples were electropolished for 10 s in a 10 vol.% chromic acid solution. The dislocation density was calculated using the modified Williamson–Hall (MWH) equation39–43. where the diffraction vector is K = 2sinθ/λ and ΔK = 2cosθ(Δ2θ)/λ. Here, Δ2θ represents the full-width at half-maximum (FWHM) of diffraction peaks at θ, θ is the diffraction angle, and λ is the wavelength of the X-rays. D and ρ represent average grain size and dislocation density, respectively; b = 0.248 nm is the magnitude of the Burgers vector. A is a constant that depends on both the effective outer cut-off radius of dislocations and the dislocation density. O is the higher-order term of KC¯1/2. Ch00 and q are associated with the elastic constants of the material, with detailed information available in refs. 39–42. The terms h, k. l refer to the Miller indices of the crystal plane. The distribution of precipitate elements was analyzed using 3D-APT (LEAP 4000HR, CAMECA, USA) at a reference temperature of 50 K, with high-voltage pulses (4.5 × 10−11 torr, 20% pulse fraction, 200 kHz repetition frequency). APT samples were prepared by electropolishing the aged samples, with data reconstruction carried out using IVAS 3.6.2 software. TKD sample tests were conducted using a dual-beam electron microscope (Thermo Fisher Scios 2, Thermo Fisher, USA), equipped with a coaxial TKD (Bruker OPTIMUS, USA). The parameters selected were: voltage of 30 kV, beam current of 2 nA, working distance of 5 mm, and a data collection step size of 0.12 nm. Fracture surfaces were characterized using a scanning electron microscope (SEM; EVO 25, Zeiss, Germany). The three-dimensional morphology of the tensile fracture was characterized using a high-precision micro-focus X-ray three-dimensional detection system (CT, 190KV, YXLON FF35CT, JEOL, Japan). In-situ SEM experiments were conducted to observe fracture behavior. The tensile tests were carried out on an in-situ tensile test bench (PicoFemto-SEP SEM5000N, Zeptools Technology Co., Ltd., China) at a constant strain rate of 2 μm/s, using a plate-shaped specimen with 0.5 mm thickness and 1.5 mm width. The surface of the specimen, polished and corroded before tensile testing, was observed using an equipped scanning electron microscope (SEM5000, CIQTEK, China). Supplementary information Supplementary Information Transparent Peer Review file Source data Source data Acknowledgements The authors thank Changsha Kaiple Technology Co., Ltd. for providing technical support about EBSD and TEM experiments; NCS Testing Technology Co., Ltd. for mechanical properties testing; Q. Liu (State Key Laboratory of Advanced Metallurgy, University of Science & Technology Beijing, China) for the CT scanning experiment. Author contributions C.X.W. and Y.L. (Yong Li) proposed the idea. C.X.W., Y.L. (Yong Li), and R.M.G. finished experimental steel design. C.X.W., S.T.Z., R.M.G. and Y.L. (Yue Liu) designed the main experiments. Y.L. (Yue Liu) carried out the main experiments. R.M.G. and Y.L. (Yue Liu) analyzed the data. R.M.G., Y.L. (Yue Liu), and S.T.Z. drafted the manuscript of the paper. R.O.R., H.B.X. and C.X.W. revised the manuscript. P.W.T. and H.Q.W. edited the figures in the paper. X.Y.Y. prepared the 3D-APT experiments. L.L.L. assisted with the XRD experiments. Y.L. (Yu Liu) and S.M.L. assisted the thermodynamic calculations. Y.D.S. and Y.X.C. assisted with the EBSD analysis. L.J.Z, X.Y.L. and S.H. proposed the suggestions for the experiments. All authors contributed to the discussion. Peer review Peer review information Nature Communications thanks Hongbiao Dong, Lichu Zhou, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding C.X.W. and Y.L. (Yong Li) disclose support for the research of this work from the National Key R&D Program of China (Grant No. 2022YFB3705200). Data availability All data supporting the findings in this study are available within the main text, Supplementary Information, and the Source data file. The raw data have been deposited on Figshare: 10.6084/m9.figshare.32233842. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Ruming Geng, Yue Liu. Contributor Information Shiteng Zhao, Email: szhao@buaa.edu.cn. Chunxu Wang, Email: wangchunxu@nercast.com. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-74216-4. References Associated Data Supplementary Materials Supplementary Information Transparent Peer Review file Source data Data Availability Statement All data supporting the findings in this study are available within the main text, Supplementary Information, and the Source data file. The raw data have been deposited on Figshare: 10.6084/m9.figshare.32233842. Source data are provided with this paper.