Archeology-inspired enhancement strategy for aluminum alloys

Summary

Metal casting naturally produces a fine-grained chill layer at the mold interface, although this surface layer is commonly removed during machining. Inspired by archeological observations of ancient Cu-Sn bronzes, we investigated the effects of preserving this naturally formed metallic “skin” in modern A356 aluminum alloys. In bronzes, retaining the chill layer reduced corrosion current densities in neutral, alkaline, and acidic environments. In A356 alloys, preserving the chill layer increased ultimate tensile strength by up to 10%, improved elongation from 4.1% to 5.8%, enhanced fatigue endurance by 14%, and increased surface hardness by 20%. These improvements are associated with a naturally formed hierarchical grain structure, in which a fine-grained surface and a coarser interior promote Hall-Petch strengthening and hetero-deformation-induced hardening. These results highlight the functional significance of the chill layer and suggest its preservation as a design consideration for improving cast aluminum alloys without alloy redesign or additional surface processing.

Graphical abstract

Highlights

• •Ancient bronzes inspire preservation of the naturally formed chill layer
• •Preserving the chill layer improves strength, fatigue, and corrosion resistance
• •A grain-refined surface promotes Hall-Petch and hetero-deformation strengthening

Interface science; Corrosion; Materials science; Alloys

Introduction

Metal casting naturally produces a fine-grained chill layer at the mold-metal interface during solidification. Driven by rapid heat extraction, this thin surface region exhibits a substantially finer microstructure than the bulk material, conferring reduced porosity and enhanced mechanical properties. In essence, the casting process intrinsically generates a surface-hardened metallic “skin,” which may provide a natural route for enhancing surface properties and reducing the reliance on additional surface-strengthening treatments, such as induction, laser, or electron-beam hardening.¹ Despite these potential advantages, the chill layer is often removed during subsequent machining to satisfy strict dimensional tolerances and surface-finish requirements. While this practice is essential for many engineering components, it may also result in the loss of a naturally formed surface microstructure with potentially beneficial functional properties. Evidence from ancient metallurgy provides an intriguing perspective on the value of such as-cast surface features. Cu-Sn bronze artifacts, for instance, exhibit remarkable long-term preservation, which is generally associated with the formation of stable corrosion products and passive oxide films (e.g., SnO₂ and Cu₂O).² Because these protective films develop directly on the as-cast microstructure, the fine-grained chill zone—with its high density of grain boundaries—may facilitate early-stage passivation and contribute to improved corrosion resistance. Thus, archeological bronzes provide inspiration for reconsidering the functional role of naturally formed chill layers and motivate their exploration as a potential enhancement strategy in modern aluminum castings.

Recent studies have demonstrated that the performance of aluminum-based materials can be enhanced through advanced structural design strategies, including additively manufactured lattice architectures, triply periodic minimal surface structures, and bio-inspired bimetallic metamaterials.^3,4,5 These approaches highlight the potential of deliberate structural engineering for tailoring the mechanical and functional properties of aluminum alloys. In contrast, the present work explores an intrinsic surface microstructure that forms naturally during conventional casting but has received limited attention as a functional design feature. A356 aluminum alloy, widely used in automotive and aerospace castings, naturally develops a fine-grained chill zone at the metal-mold interface.⁶ This refined surface region possesses distinct microstructural characteristics compared with the coarser interior. Previous studies have shown that the chill layer formed during casting can contribute positively to the mechanical properties of aluminum alloys. For example, Niu et al. demonstrated that a supersaturated chill layer formed during high-pressure die casting of Al-7Si-0.5 Mg alloys enhances the strength-ductility synergy through surface grain refinement and Mg₂Si precipitation.⁷ Nevertheless, these studies mainly focused on the formation mechanism of the chill layer and its influence on tensile behavior. In particular, it remains unclear whether preserving the chill layer, which is routinely removed during machining, can simultaneously enhance multiple performance metrics, including tensile properties, fatigue resistance, corrosion resistance, and surface hardness. Addressing this question is essential for evaluating the broader functional significance of the chill layer in cast aluminum alloys.

Motivated by archeological observations of ancient bronzes, we investigate whether the naturally formed chill layer can be exploited as a functional surface architecture for A356 aluminum alloys. Through electrochemical analysis, mechanical characterization, and atomistic simulations, we demonstrate that retaining the chill layer simultaneously improves corrosion resistance, tensile strength, fatigue endurance, and surface hardness without alloy redesign or additional surface processing. These findings broaden the current understanding of the chill layer from its formation to its multifunctional role in governing material performance, highlighting its preservation as a promising design consideration for cast aluminum alloys.

Results and discussion

Archeological inspiration from ancient bronzes

Ancient bronze artifacts provide archeological evidence that the naturally formed chill layer may serve as a durable protective surface. As illustrated in Figure 1A, archeological artifacts spanning different historical periods, including a 3,600- to 4,200-year-old large-bladed spear with a reversed hook and tubular socket (Qinghai Museum) as well as coins from the Warring States period (475–221 BCE) and Song dynasty (960–1279 CE), provide striking examples of the long-term preservation achievable in ancient bronze metallurgy. Microstructural analysis of these coins reveals a distinct heterogeneous structure, consisting of a fine-grained surface layer transitioning into a coarser interior. To explore whether this naturally formed surface refinement contributes to corrosion resistance, we evaluated laboratory-cast Cu-Sn specimens—with and without their chill layers—under neutral, alkaline, and acidic conditions (Figure 1B). The potentiodynamic polarization results show that specimens retaining the chill layer consistently exhibit lower corrosion current densities, indicating improved corrosion resistance. In neutral NaCl, the chill-layer sample displays a pseudo-passivation feature, characterized by a transient decrease in anodic current that delays active dissolution. In alkaline NaOH, while both samples undergo active-to-passive transitions, the chill-layer specimen exhibits a lower passivation current density, suggesting more favorable passive film formation. In highly corrosive acidic NaHSO₃, the chill-layer sample achieves a passive response, whereas the chill-free counterpart remains in an active dissolution state. Quantitative analysis (Figures 1C and 1D) further reveals that while the corrosion potentials (E_corr) are comparable, the corrosion kinetics (I_corr) are drastically reduced in the chill-layer samples, with the most pronounced improvement observed in alkaline media, where the corrosion rate decreases by ∼70% (from 0.670 to 0.198 mm year⁻¹). Post-polarization cross-sectional observations (Figure 1E) further demonstrate reduced localized corrosion and pitting susceptibility in the chill-layer specimens compared with the chill-free samples. Together, these results suggest that the naturally formed chill layer can facilitate early-stage corrosion protection, motivating further investigation of its functional potential in modern engineering alloys.

Figure 1.

Archeological observations and protective role of the chill layer in Cu-Sn bronzes

(A) Photographs of ancient bronze artifacts (a spear from the Qinghai Museum and coinage from the Warring States period and Song dynasty), accompanied by a representative metallographic microstructure from one of three independently prepared specimens (n = 3) and a schematic illustrating the naturally formed fine-grained chill layer. Scale bars, 100 μm.

(B) Representative potentiodynamic polarization curves of laboratory-cast Cu-Sn bronzes measured in neutral (NaCl), alkaline (NaOH), and acidic (NaHSO₃) environments.

(C) Comparison of the corrosion potential (E_corr) and corrosion current density (I_corr) for specimens with and without the chill layer.

(D) Calculated corrosion rates under different testing conditions.

(E) Representative cross-sectional corrosion morphologies and corresponding element-distribution maps after corrosion testing in NaOH solution, selected from one of three independently tested specimens (n = 3). Scale bars, 20 μm.

Electrochemical measurements were performed on three independent specimens (n = 3). Data in (C) and (D) are presented as mean ± SD. The polarization curves in (B) are representative of three independent measurements and, therefore, are shown without error bars.

The outstanding durability of archeological bronzes is generally attributed to the formation of stable SnO₂-rich passive films and other corrosion products. Our results suggest that the as-cast chill layer may contribute to the early stages of passivation behavior of Cu-Sn bronzes. Because passive films nucleate directly on the underlying microstructure, the dense grain-boundary network within the chill layer may provide additional diffusion pathways and interfacial sites for passive film development.⁸ Meanwhile, the reduced porosity and microsegregation associated with rapid solidification may help decrease localized galvanic interactions that promote pitting.⁹ To examine this early-stage passivation process, specimens were immersed in 0.6 M NaCl for 1 h and analyzed via X-ray photoelectron spectroscopy (XPS) (Figure 2). While both surfaces developed Sn⁴⁺-dominated passive films, depth profiling revealed a higher Sn⁴⁺ fraction throughout the passive layer in the chill-layer sample. This observation suggests enhanced formation or stabilization of Sn-containing protective species. Together with established mechanisms involving stable corrosion products and Sn enrichment, the chill layer may, therefore, represent one contributing factor to the long-term preservation of ancient bronzes, although the complex archeological corrosion history cannot be fully reproduced under laboratory conditions.

Figure 2.

Depth-resolved chemical composition of passive films formed on Cu-Sn bronzes

(A and B) Representative X-ray photoelectron spectroscopy (XPS) depth profiles showing the evolution of chemical states in specimens with (A) and without (B) the chill layer during sequential ion sputtering.

(C) Variation in the Sn⁴⁺ fraction as a function of sputtering depth.

The XPS depth profiles shown in (A)–(C) were obtained from one representative specimen for each condition and are presented without error bars.

Beyond serving as archeological inspiration, ancient casting practices also provide insights into controlling solidification-induced microstructures. Archeometallurgical records indicate that ancient artisans employed metal cores within bronze vessels to modify local cooling conditions.¹⁰ Acting as internal heat sinks due to their high thermal conductivity, these inserts could accelerate localized solidification and reduce casting defects. From a modern metallurgical perspective, such inserts functioned as internal chills, promoting microstructural refinement through directed heat extraction. This archeological insight motivates a modern enhancement strategy: preserving or deliberately engineering chill-induced grain refinement to utilize naturally formed surface microstructures for improving the performance of cast aluminum alloys.

Archeology-inspired enhancement strategy for A356 aluminum alloys

Chill layer microstructure

Inspired by the archeological observations described earlier, we next investigated whether preserving the naturally formed chill layer could be translated into an enhancement strategy for A356 aluminum alloys. Electron backscatter diffraction (EBSD) mapping (Figure 3A) reveals a continuous microstructural transition from a fine-grained surface to progressively coarser grains toward the interior. Owing to rapid heat extraction at the mold interface, the outer layer exhibits an average grain size of ∼20 μm (Figure 3B). This high cooling rate promotes nucleation while limiting grain growth, resulting in a dense, fine-grained surface layer.

Figure 3.

Crystallographic characterization of the naturally formed grain size hierarchy in A356 aluminum alloys

(A) A representative electron backscatter diffraction orientation map from one of three independently prepared specimens (n = 3), showing the naturally formed fine-grained chill layer adjacent to the casting surface. Scale bars, 100 μm.

(B) Grain size as a function of depth from the casting surface, illustrating the transition from the fine-grained chill layer to the coarser interior. Grain-size statistics were obtained from three independent specimens (n = 3). Data in (B) are presented as mean ± SD.

With increasing distance from the surface, the average grain size gradually increases to ∼35–55 μm. This evolution reflects the transition from nucleation-dominated to growth-dominated solidification under reduced undercooling within the melt. In this interior, the longer local solidification time facilitates grain-boundary migration and competitive grain growth along the thermal gradient.¹¹ Consequently, conventional casting naturally produces a hierarchical grain-size distribution governed by solidification kinetics. As demonstrated in the following text, this naturally formed microstructural hierarchy provides the structural basis for the enhanced mechanical and corrosion performance observed in A356 aluminum alloys.

Mechanical performance

Uniaxial tensile tests (Figure 4A) reveal that both the chill-layer and chill-free specimens exhibit nearly identical yield strengths (∼110 MPa), confirming comparable bulk defect populations and overall casting quality.¹² However, a marked divergence emerges during the strain-hardening stage. Retaining the chill layer increases the ultimate tensile strength from 215 ± 10 to 237 ± 12 MPa (∼10%) and the elongation from 4.1% ± 0.4% to 5.8% ± 0.5%. This simultaneous improvement in strength and ductility indicates that the naturally formed grain-size hierarchy can partially alleviate the conventional strength-ductility trade-off associated with homogeneous microstructures.^12,13 The dense dimple morphology in the chill-layer specimen (Figure 4A, inset) is consistent with more stable crack propagation and greater energy dissipation. The refined surface microstructure of the chill layer is expected to promote crack deflection and branching through its high grain-boundary density, thereby increasing resistance to crack propagation.¹⁴ Conversely, chill-free specimens exhibit smoother fracture surfaces indicative of accelerated crack propagation, where cracks initiate at surface defects and rapidly propagate along preferred weak interfaces, resulting in diminished fracture resistance.¹⁵ Fatigue performance exhibits a corresponding enhancement (Figure 4B), with the endurance limit at 10⁷ cycles increasing by ∼13.6% (from 49.5 to 56.2 MPa). Recent studies have demonstrated that fatigue performance is governed by microstructure-controlled crack initiation and propagation, making it essential to understand how surface microstructural refinement influences these damage processes.^16,17,18 Representative fractographic observations reveal distinct fatigue crack initiation locations in specimens with and without the chill layer. For the chill-free specimen, fatigue cracks initiate at the specimen surface, as shown in Figure S1A. In contrast, the crack initiation site is located beneath the surface in the chill-layer-retained specimen (Figure S1B). The difference in crack initiation behavior is likely associated with the hierarchical grain structure introduced by the chill layer, in which a fine-grained surface is seamlessly integrated with a coarser-grained interior. Such a microstructural hierarchy combines Hall-Petch strengthening with hetero-deformation-induced hardening, thereby reducing surface stress concentration and delaying surface fatigue crack initiation, which contributes to the improved fatigue resistance.^19,20 Nanoindentation mapping (Figures 4C and 4D) corroborates this mechanical heterogeneity. Hardness decreases gradually from the surface inward; the chill layer reaches ∼1.46 GPa, which is ∼21% harder than the bulk interior. The combination of a hard exterior and a compliant core establishes a mechanically graded architecture that distributes strain more uniformly under applied loads.^21,22

Figure 4.

Mechanical performance of A356 aluminum alloys with and without the chill layer

(A) Engineering stress-strain curves. The inset shows a representative fracture morphology from one of three independently tested specimens (n = 3). Scale bars, 50 μm.

(B) Stress-life (S-N) fatigue curves showing the enhanced fatigue endurance of the chill-layer specimens. Arrows indicate specimens that survived 10⁷ cycles (run-outs).

(C and D) Nanoindentation characterization of the near-surface mechanical heterogeneity, including the hardness profile as a function of depth (C) and representative load-depth curves (D).

Tensile tests were performed on three independent specimens (n = 3). The stress-strain curves in (A) are representative curves; corresponding quantitative mechanical properties are summarized in the main text. Fatigue tests were conducted using two specimens at each applied stress level, and the overlapping run-out data points correspond to two independent specimens surviving 10⁷ cycles. Nanoindentation hardness profiles were obtained from 15 independent indentations at each measurement depth and are presented as mean ± SD. Representative load-depth curves in (D) are shown without error bars.

Although the present strategy is demonstrated using A356 aluminum alloys, comparable trends observed in Cu-based (C90500) and Ti-based (Ti-6Al-4V) cast alloys (Figure S2) suggest that the underlying grain-refinement mechanism may extend to other cast metallic systems. This naturally formed hierarchical architecture resembles biological protective structures, combining a relatively hard surface with a tougher interior to improve resistance to damage while maintaining overall toughness.

Corrosion behavior

Potentiodynamic polarization tests in chloride (0.6 M NaCl) and sulfite (0.096 M) environments indicate improved electrochemical corrosion resistance for specimens retaining the chill layer (Figure S3). Under anodic polarization in NaCl solution, both samples are susceptible to pitting; however, the chill-layer specimen exhibits a higher pitting potential, indicating enhanced resistance to localized corrosion. Similarly, in the acidic NaHSO₃ environment, the chill-layer specimen shows a lower corrosion current density than the chill-free counterpart, suggesting enhanced corrosion resistance.²³ The improved corrosion behavior is consistent with the refined surface microstructure produced by rapid solidification. Reduced porosity and a finer grain structure may decrease the connectivity of corrosion-sensitive defects while promoting the formation of a more protective passive film. Similar trends were observed in the Cu-Sn bronze system, suggesting that preserving the naturally formed chill layer represents a promising strategy for improving the corrosion resistance of cast aluminum alloys and potentially other cast metallic materials.

Synergistic strengthening mechanism

Analogous to a biological protective layer, the naturally formed chill layer provides a mechanically robust surface for cast aluminum alloys. To elucidate this synergistic strengthening, we monitored the evolution of total dislocation density via X-ray diffraction (XRD) as a function of applied tensile strain (Figure 5A).²⁴ Consistent with surface-nanocrystallized architectures,²⁵ the chill-layer specimen possesses a substantially higher initial dislocation density than its chill-free counterpart. As deformation proceeds, the chill-layer sample exhibits sustained dislocation proliferation across all strain regimes. Kernel average misorientation (KAM) mapping further visualizes this localized strain partitioning. At 4% applied strain, the chill-layer sample (Figure 5C) exhibits dense dislocation tangles and extensive forest interactions. Conversely, the chill-free sample (Figure 5D) displays a sparse and diffuse dislocation distribution, lacking well-defined subgrain networks. This difference is consistent with the hierarchical grain structure introduced by the chill layer. The fine surface grains reduce the mean free path of dislocations, while the high density of grain boundaries promotes dislocation accumulation, lattice distortion, and strain hardening, which are expected to contribute to the improved fatigue resistance.²⁶

Figure 5.

Dislocation evolution and microstructural characteristics of A356 aluminum alloys with and without the chill layer

(A) Evolution of the total dislocation density as a function of applied tensile strain (0%–4%).

(B) Grain boundary misorientation distributions in the undeformed state, showing a higher fraction of high-angle grain boundaries in the chill-layer region.

(C and D) Representative kernel average misorientation (KAM) maps acquired at 4% tensile strain for specimens with (C) and without (D) the chill layer. Scale bars, 100 μm.

The data shown in (A)–(D) are representative results obtained from one specimen and are presented without error bars.

The grain boundary misorientation profiles (Figure 5B) provide further crystallographic context. The chill-layer region exhibits a suppressed fraction of low-angle grain boundaries (LAGBs, <15°) alongside a preponderance of high-angle grain boundaries (HAGBs). HAGBs provide effective barriers to the motion of geometrically necessary dislocations (GNDs), promoting dislocation accumulation and Hall-Petch strengthening.²⁷ Concurrently, the pronounced microstructural heterogeneity between the fine-grained surface and the coarser interior induces significant strain incompatibility during loading. This mechanical mismatch may contribute robust hetero-deformation-induced stresses,^26,28,29 thereby contributing to enhanced work hardening. Together, these observations provide mechanistic insight into how preserving the naturally formed chill layer enhances the mechanical performance of cast aluminum alloys.

To validate these atomistic mechanisms, we performed molecular dynamics (MD) simulations (Figure 6). The simulated stress-strain responses (Figure 6A) are consistent with the experimental observations, showing higher tensile strength for the chill-layer model. The simulated defect evolution (Figure 6B) further indicates enhanced dislocation activity in the chill-layer model. At 4% strain, the dislocation density in the chill-layer model reaches ∼11.8×10¹⁶ m⁻², nearly three times that of the chill-free model (∼4.0×10¹⁶ m⁻²). Atomistic configurations (Figures 6C–6F) reveal that partial dislocation activity near grain boundaries promotes GND accumulation. Together with the experimental observations, these simulations suggest that the naturally formed grain-size hierarchy promotes sustained dislocation accumulation and hetero-deformation hardening, providing a plausible mechanistic basis for the simultaneous improvements in strength, ductility, and fatigue resistance observed in the chill-layer specimens.³⁰

Figure 6.

Atomistic insights into deformation behavior and strengthening mechanisms in A356 aluminum

(A and B) Molecular dynamics simulated stress-strain responses (A) and corresponding defect evolution during deformation (B).

(C–F) Representative atomistic configurations and extracted dislocation structures at 4% tensile strain for models with (C and D) and without (E and F) the chill-layer architecture. Scale bars, 5 nm in (C) and (E).

The molecular dynamics simulations were performed using one representative model for each condition. The results shown in (A)–(F) are representative simulation results and are presented without error bars.

Design implications

Although the chill layer forms naturally during casting, its potential functional role has received relatively limited attention in the design of cast aluminum components. In industrial manufacturing, subsequent machining and heat treatment are often necessary to satisfy dimensional accuracy, surface quality, and microstructural consistency requirements. Consequently, the beneficial characteristics of the near-surface grain-refined region may not always be retained. Likewise, current quality-control practices primarily emphasize bulk characteristics, such as chemical composition, porosity, and defect density, whereas the microstructural features of the near-surface region are seldom considered as design parameters.

The present results suggest that the naturally formed chill layer represents a potentially valuable surface microstructure that merits consideration during casting design. Because its thickness and grain size are governed by solidification conditions—including local heat extraction, thermal gradients, and melt undercooling—these characteristics may be tuned through process parameters such as mold thermal conductivity, localized chilling, pouring temperature, and cooling rate. Although preserving the chill layer in industrial components remains challenging because of machining and dimensional requirements, manufacturing approaches such as near-net-shape casting, optimized machining allowances, and thermal management during solidification may provide practical routes for retaining this beneficial surface layer. Together, these findings highlight the potential of integrating solidification control with component design to utilize naturally formed surface microstructures for enhancing the mechanical performance and durability of cast aluminum alloys.

Traditionally regarded as a natural consequence of casting and often removed during machining, the chill layer is reconsidered here as a functional surface microstructure. Inspired by archeological observations of ancient Cu-Sn bronzes, this work shows that preserving the naturally formed chill layer provides an effective enhancement strategy for A356 aluminum alloys, simultaneously improving tensile strength, ductility, fatigue resistance, and corrosion resistance. Experimental characterization and atomistic simulations suggest that these improvements arise from the hierarchical grain structure, which combines grain-boundary strengthening with hetero-deformation-induced hardening. More broadly, this work highlights how archeological observations can inspire modern materials design, suggesting that preserving naturally formed casting microstructures offers a practical route for enhancing the performance of aluminum alloys without alloy redesign or additional surface processing.

Limitations of the study

This study was conducted under controlled laboratory conditions to evaluate the functional significance of preserving the naturally formed chill layer in cast aluminum alloys. Although the results consistently demonstrate improvements in mechanical and corrosion performance, the practicality of implementing chill-layer preservation in industrial manufacturing remains to be validated, particularly for components requiring stringent dimensional tolerances, surface-finish specifications, and high manufacturing efficiency. Consequently, the present work should be regarded as establishing the scientific basis for reconsidering the functional role of the chill layer rather than demonstrating a validated industrial enhancement strategy. In addition, the strengthening mechanism proposed here is supported by microstructural characterization, XRD, MD simulations, and previous studies on heterogeneous grain structures, but the respective contributions of Hall-Petch strengthening and hetero-deformation-induced hardening have not been directly quantified experimentally. Future investigations combining in situ mechanical testing with high-resolution strain mapping and three-dimensional dislocation characterization will be required to further validate the proposed mechanism. Finally, the chill-layer-retained and chill-layer-free specimens were prepared from parent castings of different thicknesses (3 and 6 mm, respectively) to preserve or eliminate the naturally formed chill layer while maintaining identical final specimen dimensions. Although identical machining procedures were applied after sampling, differences in parent casting thickness may introduce subtle variations in solidification history beyond the presence of the chill layer itself. Future studies employing identical parent castings with locally controlled chill-layer removal or other model systems would further isolate the specific contribution of the chill layer.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Zhen-Qiang Song (zqsong@ysu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •All data needed to evaluate the paper’s conclusions are presented in the article. The datasets generated and/or analyzed during the current study are available from the lead contact upon reasonable request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (grant nos. 52127808, 52125405, and 52471185) and Hebei Yanzhao Golden Platform Talent Gathering Plan (HY2024050013).

Author contributions

Conceptualization, S.F., Z.-Q.S., and R.-P.L.; methodology, B.W., Z.-Q.S., and Z.W.; investigation, B.Z., Y.L., S.F., Z.-Q.S., B.X., Z.Y., X.Z., L.-M.W., and R.-P. L.; writing – original draft, S.F., Z.-Q.S., and B. X.; writing – review and editing, S.F., Z.-Q.S., and R.-P. L.; funding acquisition, Z.-Q.S., S.F., X.Z., and R.-P. L.; resources, S.F. and R.-P. L.; supervision, S.F., Z.-Q.S., and R.-P. L.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
A356 aluminum alloy	Qinhuangdao Anyemetal Industry Co., Ltd.	http://www.anyemetal.com/
Ti-6Al-4V titanium alloy	Beijing Zhongnuo New Material Technology Co., Ltd.	http://www.znxc.cn/
Cu-Sn copper alloy	Laboratory preparation	N/A
Ancient Cu–Sn bronze artifacts	Bashu Xiaoyan Coin-Collectors Gallery	www.taobao.com
NaCl, 0.6 M (neutral electrolyte)	Laboratory preparation	N/A
NaOH, 0.6 M (alkaline electrolyte)	Laboratory preparation	N/A
NaHSO3, 0.096 M (acidic electrolyte)	Laboratory preparation	N/A
Keller’s reagent (2.5% HNO3, 1.5% HCl, 1% HF, balance H2O)	Laboratory preparation	N/A
SiC sandpaper (up to 5000 grit)	Keple	https://www.keple.cn/
Alumina polishing suspension (0.04 μm)	Keple	https://www.keple.cn/
Acetone	Inno-chem	CAS: 67-64-1
Ethanol	Inno-chem	CAS: 64-17-5
Software and algorithms
LAMMPS	Lammps.org	https://www.lammps.org/
Atomsk	Atomsk.org	https://atomsk.univ-lille.fr/
OVITO	OVITO.org	https://www.ovito.org/
Thermo Scientific Avantage	Thermo Scientific	https://www.thermofisher.cn/
EDAX OIM Analysis software	EDAX	https://www.edax.com.cn/
Origin Pro	OriginLab	https://www.originlab.com/
Other
Instron 5982 universal testing machine	Instron	https://www.instron.cn/
Instron 8872 fatigue machine	Instron	https://www.instron.cn/
KLA G200X nano indenter	KLA Instruments	https://www.kla-instruments.cn/
AxioLab5 Optical (OM)	Zeiss	https://www.zeiss.com.cn/
SU7000S (SEM with EDS)	Hitachi	https://www.hitachi-hightech.com/
Velocity Ultra (EBSD system)	EDAX	https://www.edax.com.cn/
JEM-2100 Plus (TEM)	JEOL	https://www.jeol.com/
SmartLab (XRD)	Rigaku	https://rigaku.com/
Thermo Scientific K-Alpha XPS	Thermo Fisher	https://www.thermofisher.cn/
3000A-DX electrochemical workstation	PARSTAT	https://www.par-solartron.com.cn/

Method details

Sample preparation

Laboratory-cast bronze specimens were prepared from an alloy containing 83 wt % Cu and 17 wt % Sn. An Al-7Si-0.3 Mg alloy (A356) was melted at 740 °C and cast into copper molds to produce rectangular plates. To evaluate the effect of the chill layer, two groups of specimens with identical dimensions were prepared. Specimens retaining the chill layer were machined directly from 3 mm-thick as-cast plates. Specimens without the chill layer were obtained from the central region of 6 mm-thick castings after removing approximately 1.5 mm from each surface, resulting in a final thickness of approximately 3 mm, as illustrated in Figures S4A and S4B. After sampling, both groups of specimens were machined using identical procedures prior to mechanical testing. The specimen geometries for tensile and fatigue testing are shown in Figures S4C and S4D. Cu–Sn bronze and Ti–6Al–4V (TC4) specimens were prepared using the same sampling procedure.

Mechanical and microstructural characterizations

Quasi-static tensile tests were conducted at a strain rate of 5 × 10⁻⁴ s⁻¹ using an Instron 5982 testing machine, following ASTM E8/E8M-16a. For each condition, three specimens were tested to ensure reproducibility. Tension-tension fatigue tests were performed on an Instron 8872 testing machine with a load ratio of R = 0.1 (σ_min/σ_max) and a frequency of 20 Hz under sinusoidal loading. Microstructural characterization was carried out using optical microscopy (OM), electron backscatter diffraction (EBSD; Hitachi SU7000 SEM), and transmission electron microscopy (TEM; JEM-2100 Plus). Prior to OM and EBSD observations, samples were etched using Keller’s reagent (2.5% HNO₃, 1.5% HCl, 1% HF, balance H₂O) to reveal grain boundaries. Hardness measurements were conducted using a KLA G200X nanoindenter equipped with a Berkovich tip, calibrated on fused quartz. Continuous stiffness measurement (CSM) was employed with a maximum load of 6 mN. X-ray photoelectron spectroscopy (XPS) was conducted using a Thermo Scientific K-Alpha spectrometer. Depth profiling was performed from the surface via sequential Ar⁺ sputtering at 10 nm intervals for a total of five cycles.

Electrochemical analysis

Electrochemical measurements were conducted at room temperature (298 K) using a PARSTAT 3000A-DX workstation. A conventional three-electrode cell was employed, consisting of a platinum mesh counter electrode, a saturated calomel electrode (SCE) as the reference electrode, and the specimen as the working electrode. Prior to testing, the samples were sequentially ground with SiC sandpaper up to 5000 grit and subsequently polished using a 0.04 μm alumina suspension to obtain a mirror-finished surface. The polished specimens were ultrasonically cleaned in acetone, ethanol, and deionized water, followed by air drying. Electrochemical corrosion tests were conducted in three representative electrolytes: 0.6 M NaCl, 0.6 M NaOH, and 0.096 M NaHSO₃. The exposed surface of each specimen was immersed in the electrolyte, and the open-circuit potential (OCP) was monitored until a stable value was reached. Potentiodynamic polarization measurement were then conducted from −1.5 V to 1 V versus SCE at a scan rate of 1 mV/s. Each condition was tested in triplicate to ensure reproducibility. The corrosion rate (V_corr) is calculated using Faraday’s law^31,32: V_corr = I_corrA_w/nρF, where I_corr is the corrosion current density, A_w is the average atomic weight of the dissolved alloy constituents, n is the number of electrons transferred, ρ is the density of the alloy, and F is the Faraday constant (96,500 C/mol).

Molecular dynamics simulation

Molecular dynamics (MD) simulations were performed using the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) to investigate the atomic-scale deformation behavior associated with the hierarchical grain structure.³³ Polycrystalline aluminum models were generated using Atomsk.³⁴ Two representative microstructures were constructed separately: a fine-grained chill layer (15 × 30 × 2.5 nm³) containing 40 randomly oriented equiaxed grains with an average grain size of ∼3.76 nm, and a coarse-grained layer (15 × 30 × 7.5 nm³) containing 20 grains with an average grain size of ∼6.86 nm. Based on these building blocks, two model structures were established: (1) a three-layer sandwich architecture consisting of fine-coarse-fine stacking (15 × 30 × 12.5 nm³, 346,537 atoms) and (2) a coarse-grained model (15 × 30 × 7.5 nm³, 207,857 atoms). Both models were energy minimized using the conjugate gradient algorithm and subsequently equilibrated at 300 K for 1 ns under the isothermal–isobaric (NPT) ensemble. Atomic interactions were described using an embedded atom method (EAM) potential³⁵ developed for aluminum. Uniaxial tensile loading was applied along the y-direction at a strain rate of 0.2/ns. Periodic boundary conditions were applied imposed along the x- and z-directions to minimize free-surface effects. Dislocation evolution was analyzed using the Dislocation Extraction Algorithm (DXA),³⁶ and the deformation process was visualized using OVITO.³⁷

Quantification and statistical analysis

Quantitative data are presented as mean ± standard deviation (SD) where applicable. Tensile tests were performed on three independent specimens (n = 3). Electrochemical measurements, including corrosion potential, corrosion current density, and corrosion rate, were obtained from three independent specimens (n = 3). Fatigue tests were conducted using two specimens at each stress level. Nanoindentation hardness values were obtained from 15 independent indentations at each measurement depth. Grain-size statistics were quantified from EBSD analyses of three independently prepared specimens (n = 3) using OIM software. Grain boundary misorientation distributions, KAM maps, and XRD-derived dislocation densities were obtained from a representative specimen. X-ray photoelectron spectroscopy (XPS) depth profiles were analyzed using Thermo Scientific Avantage software. Molecular dynamics simulation results, including stress–strain responses and dislocation evolution, were analyzed using the Dislocation Extraction Algorithm (DXA) and visualized with OVITO. All graphs and quantitative data analyses were performed using OriginPro software.