Universal synthesis of high-entropy alloy nanostructures with femtosecond lasers Authors: Zikang Su, Lan Jiang, Yiheng Dai, Shilong Yuan, Xianze Zhang, Haozhe Gai, Xingdong Wang, Jihan Zhou, Xueqiang Zhang Published: 2026 ABSTRACT High-entropy alloy nanostructures (HEA-NSs) hold promise in advanced catalysis and materials science, but lack a robust and universal synthetic strategy designed at an atomic level. Here, femtosecond lasers are used to prepare HEA-NSs, from single atoms to 100 nm nanoparticles, either free-standing or supported on a variety of substrates. Ultrafast excitation enables precise control over electron dynamics and instantaneously formed solvated electrons induce rapid, nonselective ion reduction ∼100 picoseconds post-irradiation, followed by stochastic atom nucleation and formation of atomically dispersed HEA-NSs through diffusion-controlled dynamics beyond nanoseconds. These processes are governed by the spatiotemporal confinement effect (STCE), with atomic diffusion restricted within nanoseconds, thereby enabling kinetic trapping of metastable nanoclusters with strict atomic-level dispersity. As a demonstration, FeCoNiRuPt achieved a peak power density of 2.1 W cm−2 for oxygen reduction reaction (ORR). Femtosecond lasers are robust and effective tools for the universal preparation of difficult-to-synthesize metastable nanoparticles by controlling electron dynamics. INTRODUCTION High-entropy alloy nanostructures (HEA-NSs), defined as alloys comprising five or more principal elements each at 5–35 at.% with configurational entropy ≥1.5R, have garnered increasing interest owing to their tunable compositions, tailored electronic structures, and exceptional stability under extreme conditions. These attributes make them promising for diverse applications, including catalysis, energy storage, and structural materials (1–3). For example, by integrating transition metal atoms with various radii, the activity and product selectivity of HEA-NSs can be fine-tuned (4, 5). However, chemical and physical variability across different elements introduces challenges in the synthesis of HEA-NSs, leading to phase separation and elemental segregation at the atomic level, which adversely affect their catalytic performance (6, 7). Over the past decade, several methods for synthesizing HEA-NSs have been developed, including carbothermal shock (CTS) (8), electrosynthesis (9), and other approaches (10–13). These methods typically rely on strong driving forces to bypass thermodynamic equilibrium and trap the system in a metastable alloying state, thereby enabling unique properties such as increased catalytic activity. However, they often have limitations related to compositional and substrate flexibility (text S1, tables S1 and S2, and fig. S1). Notably, these approaches have achieved considerable success in enabling the synthesis of high-entropy phases through nonequilibrium pathways. Laser-assisted methods have also been explored (14), where nanosecond laser synthesis has demonstrated effective multicomponent alloy formation. Femtosecond laser approaches have likewise shown promising results (15), offering access to nonthermal pathways. Existing strategies have successfully synthesized compositionally homogeneous high-entropy materials through a variety of thermal, electrochemical, and nonequilibrium processes. These approaches typically involve reduction, nucleation, and growth processes occurring over nanosecond-to-second timescales. In contrast, accessing reaction environments on substantially shorter timescales remains relatively unexplored. Such ultrafast conditions provide opportunities to investigate distinct reaction pathways and transient states during the formation of multicomponent nanostructures. The ultrafast nucleation dynamics confines atom motion in the earliest growth stages and intrinsically suppress phase segregation, ensuring true atomic-level mixing among elements in HEA nanostructures that would otherwise be immiscible under thermal-driven conditions. Femtosecond lasers provide a unique platform for accessing such ultrafast reaction environments, particularly in liquid-phase systems. Femtosecond laser irradiation, which involves the delivery of ultrashort pulses with high peak intensity and leads to nonthermal effects such as multiphoton ionization, offers a fundamentally distinct regime of reaction control (16–20). The ultrashort pulse duration and high instantaneous field strength give rise to nonthermal synthetic pathways, in which photoinduced ionization replaces traditional heat-driven mechanisms as the dominant driving force (21, 22). In this regime, a dense population of high-energy electrons is confined in both space and time and exhibits strong reducing power. This enables near-concurrent reduction and nucleation of diverse metal ions, largely independent of their intrinsic redox properties, allowing arbitrary mixing of elements. Importantly, starting from a homogeneously mixed solution eliminates the need for long-range atomic diffusion, while the process occurs within a sub-100 ps window where diffusion is strongly restricted. As a result, selective reduction and secondary redox interactions are suppressed, promoting highly homogeneous multielement distributions at an atomic scale and suppressing detectable phase segregation. This capability provides a unique approach for kinetic control over reduction, nucleation, and growth processes, which is essential for producing metastable nanostructures such as compositionally homogeneous HEA-NSs. Moreover, owing to these attributes, femtosecond laser irradiation is not only an efficient energy delivery method but also a versatile platform for precise material formation in solution environments, offering advantages such as scalability, ease of precursor handling, and broad substrate compatibility. The persistent limitations of existing methods, including restricted compositional tunability, substrate incompatibility, and lack of atomic-level precision, underscore the importance of developing a generally applicable femtosecond laser-based approach for diverse material systems. Here, we develop a generally applicable strategy for synthesizing HEA-NSs across diverse compositions and supports, including systems containing up to ten elements and structures ranging from isolated single atoms and subnanometer clusters to high-entropy nanoparticles approaching 100 nm in size (fig. S2). Transmission electron microscopy (TEM), transient absorption spectroscopy (TAS), and molecular dynamics (MD) simulations were used to characterize the structures and mechanisms of the products. By regulating the electron dynamics at an ultrafast timescale and leveraging kinetic (e.g., atomic rearrangement and diffusion) and thermodynamic (e.g., surface energy and redox potential) factors, the composition, size, and morphology of the HEA-NSs were precisely tailored across a wide range of elements and substrates. FeCoNiRuPt/Graphene was synthesized using the proposed strategy and exhibited a notable peak power density of 2.1 W cm−2 with extended stability during the oxygen reduction reaction (ORR), surpassing commercial Pt/C in terms of stability. Electron dynamics-controlled synthesis of HEA-NSs offers an avenue for the general preparation of advanced nanocatalysts, with potential applications in energy conversion and environmental remediation. RESULTS General strategy for synthesizing HEA-NSs via femtosecond lasers To determine if the general strategy developed in this work involving femtosecond laser-assisted reduction of metal ions could be used to synthesize HEA-NSs, we synthesized different materials on various supports and examined their morphological properties via TEM and scanning electron microscopy (SEM). An ink mixture containing metal salts and substrates, including combinations such as Fe, Co, Ni, Ru, Pt with graphene, carbon black, zirconia, zeolite, and metal-organic frameworks (MOFs), was continuously pumped into a customized flow cell, and HEA-NSs were obtained in one step using spatially shaped femtosecond lasers (Fig. 1A, fig. S3 and Materials and Methods). The precursor concentration is selected to balance optical transparency and reduction efficiency. High concentrations may lead to optical shielding by formed nanoparticles, while overly dilute solutions reduce production yield. Because ultrashort laser pulses deliver energy faster than thermal equilibration can occur, bulk heating is minimized, allowing the synthesis to proceed entirely under ambient conditions. Consequently, the process is compatible with various supports, including oxides such as zirconia and alumina, carbon-based materials such as carbon black and graphene, and microporous materials such as zeolites and metal organic frameworks (MOFs) (Fig. 1, B to H and figs. S4 to S14). This approach exhibits broad generality and compatibility with various substrates. In the case of three-dimensional (3D) frameworks such as zeolites and MOFs, HEA-NSs were observed within the confined nanopores or subnanopores (Fig. 1, E to F). At higher energy densities of 0.5–1.0 J/cm2, the laser can also induce in situ substrate modification, leading to structural defects or transformation of the support structure (fig. S15). At such fluences, additional nonlinear optical effects such as optical breakdown, plasma formation, and self-focusing may occur, which increase the complexity of the reaction environment. In the present study, we focused on this energy density window to maintain a relatively controllable and modellable system, facilitating mechanistic investigation and precise observation. This choice does not imply that higher fluences are inherently disadvantageous; rather, they may offer additional opportunities for simultaneous tailoring of both HEA-NSs and their local environment in future applied studies. This ability to simultaneously tailor both HEA-NSs and their local environment expands the design flexibility of functional nanomaterials. Universal synthesis of HEA-NSs on diverse substrates. (A) Schematic illustration of the flow-cell approach used for synthesizing HEA-NSs via femtosecond laser irradiation. This setup enables continuous precursor delivery and in situ control of laser exposure. (B to F) TEM images showing HEA-NSs synthesized on different substrates: (B) graphene, (C) XC-72 carbon black, (D) ZrO2, (E) ZSM-5 zeolite molecular sieve, and (F) MIL-101 MOF. (G and H) HAADF-STEM images of HEA-NSs on graphene and ZSM-5, respectively. (I) HAADF-STEM image and EDS elemental maps of Fe, Co, Ni, Cu, Pd, In, Pt, and Au in the HEA-NSs on graphene, demonstrating uniform elemental distribution. (J) XANES spectra at the Fe, Co, Ni, and Ru K-edges and Pt L3-edge for the FeCoNiRuPt HEA-NSs on graphene, compared with reference samples, confirming the metallic state and coordination environment of each element. To further evaluate the robustness of the femtosecond laser-assisted synthesis strategy, the reproducibility of HEA-NSs formation was systematically examined through multiple independent experiments. Under identical synthesis conditions, three independent batches were prepared for representative systems, and consistent morphological and structural characteristics were observed. Statistical analysis of TEM images (fig. S16) shows that the nanoparticle size distributions remain highly consistent across different runs, with only minor variations within experimental uncertainty, confirming the stability of the nucleation and growth process. To elucidate the composition of the samples and verify that they were HEA-NSs, elemental mapping, x-ray photoelectron spectroscopy (XPS), and inductively coupled plasma-mass spectrometry (ICP-MS) were performed. Energy-dispersive x-ray spectroscopy (EDS) mapping of a representative sample, FeCoNiCuInPdPtAu (Fig. 1I), confirmed a homogeneous distribution of all the constituent elements. X-ray photoelectron spectroscopy (fig. S17) and ICP-MS confirmed the elemental composition, and the calculated mixing entropy values (1.53 R to 2.01 R; table S3) met the criteria for HEA-NSs formation (section S2). X-ray absorption near-edge structure (XANES) spectroscopy was employed to investigate the electronic structures and atomic arrangements of the FeCoNiRuPt HEA-NSs (Fig. 1J). For Fe, Co, and Ni, the edge positions were between those of the metallic and oxidized states, indicating partial oxidation, whereas Ru and Pt maintained edge positions and “white line” intensities closer to those of their metallic foils, confirming their predominantly metallic states (23). As shown in the valence state fittings (fig. S18), Fe, Co, and Ni exhibited partial oxidation, Ru demonstrated a slightly positive valence, and Pt displayed a slightly negative valence, aligning well with the general trend in elemental electronegativity (Pt > Ru > Fe, Co, and Ni). In the k3-weighted spectra of the Fourier transform extended x-ray absorption fine structure (FT-EXAFS, fig. S19), the Fe-metal (M) shell distance in the FeCoNiRuPt HEA-NSs (2.24 Å) was slightly larger than that in Fe foil (2.21 Å). Similarly, the Co-M (2.39 Å) and Ni-M (2.27 Å) shell distances were slightly greater than those in standard Co (2.18 Å) and Ni (2.18 Å) foils, respectively, which is attributed to the larger atomic radii of Ru and Pt coordinated with Fe, Co and Ni. These larger shell distances indicate structural expansion due to alloying, contributing to understanding key structural features of HEA-NSs. In contrast, the Ru-M (2.30 Å) and Pt-M (2.45 Å) shell distances were shorter than those of the respective foils. Wavelet transform maps confirmed the altered coordination environments characteristic of chemically disordered multimetallic structures (figs. S20 to S24). Structural parameters derived from fitting Fe, Co, Ni, and Ru K-edge and Pt L3-edge EXAFS spectra (table S4 and figs. S25 to S29) indicated generally higher coordination numbers (CNs) for M-M interactions, reflecting densely packed structures, whereas the lower CN values for M-O suggest more dispersed oxygen coordination. These densely packed structures and dispersed oxygen coordination correspond to desirable HEA-NSs properties for increased stability and enhanced catalysis. Mechanistic study of HEA-NSs formed via femtosecond laser liquid-phase synthesis To elucidate the temporal sequence of physical and chemical events that underpin femtosecond laser synthesis, in situ TAS was employed to track electron and ion dynamics from the femtosecond to nanosecond timescales (section S3 and figs. S30 and S31). The synthesis of Au nanoparticles by irradiating an HAuCl4 solution with a femtosecond laser was selected as a model reaction because of its relatively well-characterized photophysicochemical signatures reported in the literature (16). By controlling the flow rate of the precursor ink, the proportions of different components in the flow cell, i.e., metal ions, nanoclusters and nanoparticles of different sizes, can be precisely regulated, thereby decoupling the TAS signals of different components (figs. S32 to S34). By comparing the in situ TAS responses of HAuCl4, irradiated HAuCl4, and Au nanoparticles, a spectroscopic baseline for tracking electron and ion dynamics was established (Fig. 2, A to C; figs. S35 and S36). The early-time TAS signatures reveal a broad positive band near 600 ∼ 720 nm, which can be attributed to solvated electrons (24–26). Additionally, characteristic photoinduced absorption (PIA) and ground-state bleaching (GSB) features of Au nanoparticles were observed at ∼500 nm and ∼550 nm (27). Notably, the solvated-electron signals of HAuCl4 decay faster than those in pure water within the 1–100 ps window, indicating rapid electron scavenging by Au3+ ions (Fig. 2D). Moreover, an unexpected upward shift near 550 nm indicated the onset of ultrafast nucleation (Fig. 2E). These observations confirm that femtosecond-laser synthesis is driven not by thermal pathways but by solvated electron-mediated reduction occurring on sub-100-ps timescales. TAS in multicomponent systems are affected by spectral overlap and cannot fully resolve individual species. Additional measurements on single- and binary-precursor systems show that the rise and decay time windows remain similar after mixing (fig. S37). This suggests that the overall temporal evolution is mainly governed by ultrafast electron dynamics rather than the specific combination of metal ions. This mechanistic distinction highlights the nonthermal nature of the process and provides a basis for understanding the subsequent structural evolution. Mechanistic study of HEA-NSs formation via femtosecond laser liquid-phase synthesis. (A to C) In situ TAS of 0.01 M HAuCl4 solutions under different flow rates in the flow cell: (A) 5 mL/min, with the intermediate species irradiated by a 1 femtosecond laser pulse, approximating a standard HAuCl4 solution; (B) 0.5 mL/min, with the intermediate species irradiated by 12 pulses; (C) 0.005 mL/min, with the solution exposed to thousands of pulses, approximating dispersed Au nanoparticles. These spectra track the evolution of precursor, intermediate, and nanoparticle species under ultrafast laser irradiation. ΔT, change in absorbance. (D) TAS dynamics at 700 nm, showing the evolution of solvated electrons in H2O and HAuCl4 under femtosecond laser irradiation, revealing early-stage reduction kinetics. Stage I involves femtosecond ionization and the generation of solvated electrons. Stage II corresponds to the confined reduction and nucleation that occur within the sub-100-ps window. Stage III is characterized by diffusion-mediated ripening on the nanosecond scale. (E) TAS signals from panels (A to C) at 520 nm, representing a combination of GSB from Au nanoparticles and PIA from HAuCl4. (F) Kinetic traces showing the relative contribution of precursor HAuCl4 and Au nanoparticles signals to the total TAS signal over time. Red: fraction of HAuCl4 obtained from in situ TAS spectral fitting. Black: nucleation kinetics of Au nanoparticles obtained from MD simulations. Rapid reduction occurs at ∼70 ps, followed by nucleation and growth. (G) Schematic illustration of femtosecond laser-induced liquid-phase nanoparticle formation. The top axis represents temporal evolution, while the bottom axis shows the diffusion distance at corresponding times. Five stages, including ionization, reduction, nucleation, growth, and ripening, occur at defined timescales and diffusion lengths. After the ultrafast burst of solvated electron reduction, the femtosecond laser-irradiated system rapidly enters a regime in which both temporal and spatial constraints limit the mobility of reactive species. Global kinetic analysis and molecular dynamics simulations indicate that nucleation finishes within several tens of picoseconds, whereas subsequent growth requires nanoseconds to proceed (Fig. 2F and figs. S38 to S43). Although transient local heating can occur during this process, it relaxes rapidly and does not persist long enough to drive notable atomic diffusion. As a result, diffusion remains strongly limited within the sub-100 ps window. According to the Stokes-Einstein equationD=kBT6πηr(1)where D is the diffusion coefficient, kB is the Boltzmann constant, T is the absolute temperature, η is the viscosity of the solvent, and r is the hydrodynamic radius of the diffusing species. For solvated electrons and metal ions in an aqueous solution, the estimated diffusion distance within ∼100 ps is approximately 3.3 Å, which is comparable to the interatomic spacing. Reasonable variations in temperature, viscosity, and species size may change this value. However, even under extreme but physically relevant conditions, the diffusion length remains within the angstrom scale (fig. S44). This distance is comparable to the spacing between atoms and therefore prevents substantial diffusion in the early synthesis stage. We refer to this phenomenon as the spatiotemporal confinement effect (STCE). Even considering transient thermalization of hot electrons, the available time is too short to enable meaningful atomic migration. Under this effect, atoms remain effectively localized before diffusion can occur. Such confinement suppresses premature aggregation and stabilizes mixed local configurations of different metal species. As a result, compositions that would normally segregate under thermal conditions remain homogeneously distributed, which provides the fundamental physical basis for achieving atomic-level dispersion in HEA-NSs. We integrated TAS, kinetic reconstruction, and MD simulations to establish a time-stratified reaction framework consisting of three consecutive stages. The first stage involves femtosecond ionization and the generation of solvated electrons. The second stage corresponds to the confined reduction and nucleation that occur within the sub-100-ps window under the influence of the STCE. The third stage is characterized by diffusion-mediated ripening on the nanosecond scale (Fig. 2G and fig. S45). Each stage evolves on a distinct timescale and is governed by different transport constraints, rendering femtosecond-laser synthesis inherently modular. In contrast to thermally driven processes, in which reduction, nucleation, and growth often overlap and proceed with notable atomic diffusion, these steps are decoupled under femtosecond excitation. The reduction occurs on an ultrafast timescale, followed by confined nucleation and slower growth. This separation limits atomic diffusion and helps suppress phase segregation. By adjusting the pulse energy, repetition rate, dispersion, or photon flux, components of individual stages can be selectively tuned, allowing control over reduction kinetics, diffusion behavior, and structural evolution. To our knowledge, this mechanistic study reveals for the first time that the formation of HEA-NSs during femtosecond laser liquid-phase synthesis is governed by ultrafast electron dynamics rather than thermal processes. The identification of the STCE provides a principle-level explanation of how femtosecond excitation suppresses detectable phase separation and promotes intrinsically homogeneous multielement distributions. These insights elevate femtosecond laser synthesis from an empirical approach to a mechanism-guided strategy and establish ultrafast electronic control as a powerful means for constructing metastable, compositionally complex nanomaterials. Electron-dynamics-driven control of composition, structure, and growth in HEA-NSs To establish the generality of the femtosecond-laser-driven synthesis approach, we systematically examined how electron dynamics mediate the formation pathways of HEA-NSs under varying precursor and laser conditions. The generation and consumption of solvated electrons, together with the restricted diffusion of metal atoms within the irradiated region, define a controllable kinetic framework that enables precise regulation of nucleation and early-stage growth. By tuning the precursor composition, laser fluence and repetition rate, the distribution of the reduced atoms and their mobility can be modulated, thereby determining the elemental composition, size and morphology of the final HEA-NSs. At single-pulse energies of 0.1–0.5 J cm−2, rapid solvated electron accumulation drives nearly simultaneous multimetal reduction, enabling uniform nucleation within the sub-100-ps STCE window and producing compositions closely matching those of the precursor (Fig. 3A). Lower fluences reduce the electron density and slow the reaction, resulting in selective reduction that favors noble metals and leads to compositional deviations. These observations show that the pulse energy governs the synchronization of reduction processes, providing an effective means to achieve compositional homogeneity in complex HEA-NSs systems. Laser-controlled tuning of composition, phase, and size of HEA-NSs. (A) Elemental distributions of FeCoNiCuRuPdPtAu HEA-NSs synthesized under varying laser parameters, demonstrating that the pulse energy can modulate the composition across multiple metals. (B) HAADF-STEM image and corresponding EDS maps of Au, Pt, Zn, Cu and Ni within an HEA-NS. Left: CuZnNiPtAu HEA-NSs, right: CuPtAu core-shell nanoparticles. Heat-induced transition from metastable high-entropy structures to stable core-shell configurations: (C) HAADF-STEM images of FeCoNiRuPt HEA-NSs. Inset: statistical particle size distribution. Regulation of the nanoparticle size, expanding the particle diameter from a single atom to 100 nm: (D) HAADF-STEM images showing the increase in the particle diameter from isolated single atoms to ∼100 nm nanoparticles under sequential laser treatments, demonstrating tunable size control across a wide range. (E) Three-dimensional atomic model of an HEA-NP with a fivefold crystal structure, reconstructed from AET. Local atomic coordination environments are colored according to the legend. To enhance structural clarity, atoms within 3.5 Å of the particle surface have been removed from the images in the second panel onward. Scale bar: 2 nm. The overall electronic energy input can also be regulated to control the structural phase accessible during synthesis. High fluence above 1 J cm−2 and repetition rates of 1–10 kHz enhance nonthermal energy transfer and support nucleation within ultrafast, confined reaction environments, stabilizing metastable configurations such as single-atom states and HEA-NSs. Reducing the nonthermal contribution at a lower fluence or repetition rate yields metallic-glass-like phases (fig. S46), whereas extended irradiation introduces thermal accumulation that drives the system toward thermodynamic relaxation, producing phase segregation or core-shell structures (Fig. 3B). Smaller particles, which are quenched more rapidly, tend to retain high-entropy features, whereas larger particles more readily undergo post-growth segregation (28). Spatial nucleation behavior also arises from electron-driven regulation: higher pulse energies promote uniform nucleation, while lower energies lead to heterogeneous nucleation at defects (fig. S47). Adjusting the repetition rate enables independent control of the nucleation density without changing the average particle size (figs. S48 and S49), which modulates the balance between nucleation and atomic supply during growth. Consequently, particle sizes ranging from isolated single atoms and subnanometer clusters to crystalline nanoparticles near 100 nm can be produced (Fig. 3, C and D). Notably, the formation of isolated single atoms under such ultrafast laser conditions is also governed by kinetic constraints (29–32). TEM reveals a growth sequence from discrete nuclei to amorphous aggregates and ultimately to faceted crystallites (fig. S50). These results demonstrate that femtosecond-laser-controlled electron dynamics provide a unified strategy for tuning phase and size distributions across a broad material space. To further reveal the structural implications of this kinetic framework, atomic-resolution electron tomography (AET) was used to probe the three-dimensional topology of the synthesized HEA-NSs. The 3D reconstructions reveal particles with elongated pentagonal bipyramid (EPB) shapes and multiple face-centered cubic (FCC) domains separated by distinct grain boundaries (Fig. 3E and fig. S51). Pair distribution function (PDF) and bond-orientational order (BOO) analyses confirm high crystallinity (fig. S52), and polyhedral template matching reveals the presence of coherent strain between domains. These results imply desirable lattice strains in HEA-NSs, resulting in optimized catalytic binding energies. Full 3D strain tensor maps highlight localized lattice distortion (figs. S53 and S54), and similar structural motifs are observed in a different HEA-NSs featuring stacking faults (figs. S55 to S57), indicating the generality of the observed deformation modes across different particles. The observed defects and lattice distortions reflect the nonequilibrium conditions of the femtosecond laser synthesis. While the AET measurements were primarily performed on relatively large HEA-NSs (>5 nm), which are more influenced by thermodynamic factors, smaller nanoparticles are expected to form under stronger kinetic control. Direct tomography of these smaller particles is limited by current resolution and signal constraints. However, high-resolution imaging reveals lattice distortions, defects, and low crystallinity in this size regime, consistent with kinetically trapped structures. Taken together, these observations suggest a size-dependent structural evolution. Smaller particles reflect kinetically dominated formation with suppressed diffusion, whereas larger particles show partially relaxed structures. Despite this difference, no obvious phase segregation is observed, supporting a consistent picture of homogeneous multielement distribution across different size regimes. Electrocatalytic performance of an HEA-NSs As a proof-of-concept, we examined the electrocatalytic performance of an FeCoNiRuPt HEA-NSs synthesized using the developed laser-based approach for the ORR (fig. S58). When the FeCoNiRuPt HEA-NSs was integrated into a membrane electrode assembly (MEA), remarkable performance under both H2-O2 and H2-air fuel cell conditions was achieved, with peak power densities reaching 2109.3 mW cm−2 and 1394.5 mW cm−2, respectively (Fig. 4A), which are comparable to or exceed those of benchmark Pt-based catalysts (Fig. 4C, fig. S59, tables S5 and S6). Mass activity (MA) analysis confirmed the efficacy of the HEA-NSs catalyst (Fig. 4B). At 0.5 V, the MA reached 85.26 A mgPt−1 (H2-O2) and 59.62 A mgPt−1 (H2-air), surpassing those of many Pt- and PtRu-based systems in the literature. This high mass activity reflects the notable intrinsic activity of the FeCoNiRuPt HEA-NSs and the improved mass transport within the catalyst layer (33–35). Under a high current density, where oxygen transport becomes limited, the HEA-NSs maintained superior performance, suggesting that the catalyst structure facilitates efficient oxygen accessibility (36). ORR performance of HEA-NSs catalysts in fuel cells. (A) Polarization curves and power density plots of MEAs using FeCoNiRuPt HEA-NSs catalysts under H2-O2 fuel cell conditions and H2-air fuel cell conditions, before and after a stability test involving 20,000 ADT cycles. The anode catalyst in all MEAs was Pt/C with a loading of 0.1 mgPt−1 cm−2, and the cathode Pt loading was 0.04456 mgPt−1 cm−2. Operating temperature: 353.15 K; relative humidity: 100%; back pressure: 0.2 MPa. H2 and O2 flow rates: 1000 and 1500 ccm, respectively; H2 and air flow rates: 1000 and 1500 ccm, respectively. (B) Comparison of the peak power density and cathode Pt loading of FeCoNiRuPt-based MEAs under low and high current densities, demonstrating the increased utilization of Pt in HEA catalysts relative to commercial Pt/C. (C) Comparison of the peak power density and cathode Pt loading of PtRuNiCoFe and commercial Pt/C-based fuel cells under low and high current densities. (D to F) In situ XANES spectra and EXAFS of FeCoNiRuPt HEA-NSs catalysts at OCP, −0.6 V, and −1.0 V: (D) Ru K-edge, (E) Pt L3-edge, and (F) Fe K-edge. The insets show zoomed-in views of the dashed boxes to highlight detailed spectral changes under different potentials. The stability was evaluated under both rotating disk electrode (RDE) and MEA configurations. After 5000 accelerated durability test (ADT) cycles in the RDE configuration, only a 0.3 mV shift in the half-wave potential was observed (fig. S60). Similarly, under MEA conditions, after 20 k potential cycles, the peak power density decreased from 1394.5 to 1310.0 mW cm−2, corresponding to only a 6.0% decrease (Fig. 4A, table S7). These results indicate that HEA-NSs synthesized by this method can sustain prolonged operation without notable performance degradation. In contrast, the electrocatalytic performance decreased in control experiments with PtFeCoNi, PtRuCoNi, PtRuNi and PtCoNi (figs. S61 and S62), highlighting the strong synergistic effect across the constituent elements in the FeCoNiRuPt HEA-NSs. To further explore the electrocatalytic mechanism, in situ XAFS spectroscopy was conducted during the ORR (fig. S63). XANES analysis reveals potential-dependent changes in the valence states of Pt, Ru, Fe, Co, and Ni (Fig. 4, D to F for Pt, Ru, and Fe; figs. S64 to S66 for Co and Ni, with Pt, Ru, Fe shown in the main text due to their central catalytic roles). Pt underwent reversible redox shifts, becoming metallic under a cathodic bias (37–41). Ru largely retained its metallic character, whereas Fe, Co, and Ni showed partial reduction, suggesting interelement charge transfer and dynamic redox coupling. EXAFS and wavelet transform analyses (tables S8 to S12, figs. S67 to S71) reveal coordination rearrangements linked to intermediate oxygen adsorption. The Pt-O bond signal appears at intermediate potentials and vanished under strongly cathodic conditions, suggesting active intermediate turnover at the Pt sites. The Ru-M signal intensifies during potential cycling, indicating ligand exchange processes, whereas Fe, Co, and Ni maintain relatively stable coordination environments. This in situ analysis suggests a structurally robust core in the HEA-NSs supporting a catalytically dynamic surface. The FeCoNiRuPt HEA-NSs demonstrates superior ORR performance and stability, validating the utility of the proposed synthesis method. DISCUSSION The synthesis of HEA-NSs has long been constrained by the difficulty of achieving compositional homogeneity, tunability and compatibility with diverse substrates. In this study, these limitations are addressed through a femtosecond laser strategy that operates through ultrafast electron-driven pathways rather than thermal processes. Mechanistic analyses, including the generation of solvated electrons, suppression of atomic motion at early times, and the emergence of the STCE, show that femtosecond excitation creates a reaction environment distinct from conventional methods. In this regime, atomic diffusion is strongly restricted, enabling multiple elements to be trapped in a highly mixed state and suppressing phase segregation. This ultrafast mechanism allows control over composition, structure, and phase, from isolated single atoms to nanoparticles up to 100 nm. The approach is applicable to a wide range of elements and substrates and produces nanostructures with tunable properties. Although demonstrated here through ORR electrocatalysis as a representative high-entropy catalyst, the underlying principle may be applicable to a broader range of catalytic reactions and nonequilibrium material systems. Further validation across additional compositions and reaction environments will be the subject of future studies. At the same time, several limitations remain. The synthesis involves a complex and coupled parameter space, including laser conditions, precursor chemistry, and transient processes. Their relationships to final structures are not yet fully predictable, and optimization still relies on empirical exploration. In addition, ultrafast electron dynamics, plasma effects, and fluid processes occur simultaneously, which makes it difficult to establish a unified quantitative model. From a practical perspective, the current production yield is on the order of milligrams per hour, and the energy efficiency is not yet optimized for large-scale use. Future work should focus on improving throughput and energy efficiency, for example by using higher repetition rate lasers, beam parallelization, and optimized flow-cell designs. Further studies on parameter mapping and modeling will also help improve controllability and expand the accessible materials space. MATERIALS AND METHODS Materials Titanium (IV) (triethanolaminato) isopropoxide solution (C9H19NO4Ti, 80 wt %, MREDA), Vanadium (III) Acetylacetonate (C15H21O6V (98%, MREDA), Manganese (II) chloride (MnCl2, 99.9%, MREDA), Ferric (III) chloride (FeCl3, 99.9%, MREDA), Cobalt (II) chloride (CoCl2, 99.9%, MREDA), Nickel (II) chloride (NiCl3, 99.9%, MREDA), Copper (II) chloride (CuCl2, 99.9%, MREDA), Zinc (II) chloride (ZnCl2, 99.9%, MREDA), Gallium (III) chloride (GaCl3, 99.9%, MREDA), Zirconium (IV) chloride (ZrCl4, 99.9%, MREDA), Niobium (IV) chloride (NbCl4, 99.9%, MREDA), Sodium molybdate (Na2MoO4, 99.9%, MREDA), Ruthenium (III) chloride (RuCl3, 99.9%, MREDA), Rhodium (III) chloride (RhCl3, 99.9%, MREDA), Palladium (II) chloride (PdCl2, 99.9%, MREDA), Silver (I) nitrate (AgNO3, 99.9%, MREDA), Chromium (III) chloride (CdCl3, 99.9%, MREDA), Indium (II) chloride (InCl2, 99.9%, MREDA), Stannous (II) chloride (SnCl2, 99.9%, MREDA), Hafnium (IV) chloride (HfCl4, 99.9%, MREDA), Graphene and Graphene Oxide (Single layer graphene/multilayer graphene, XFNANO-Nanjing, China), Potassium hydroxide (99.99%, Meryer), Methanol (analytical grade, Meryer), Absolute ethyl alcohol (analytical grade, Meryer), Pt/C powders (20 wt %, Shanghai Hesen Electric Co., Ltd.), Nafion D-521 dispersion (5 wt %, Sinero), MIL 101 MOF (Aladdin) and ZSM-5 zeolite were synthesized according to the method reported in Ref. (42). Synthesis of HEA-NSs by femtosecond laser flow-cell Metal nitrates or chlorides were selected based on solubility and chemical compatibility, and dissolved in deionized water. For poorly soluble precursors or substrates, ethanol or other non-reactive solvents were used. The precursor solution was ultrasonically mixed with substrates for 1 h to form a uniform ink. The ink was delivered into a flow cell using a peristaltic pump under continuous stirring. A Spectra Physics laser (800 nm, 35 fs, 1 kHz) was used. The laser power was controlled by an attenuator, and a cylindrical lens was used to shape the beam for uniform irradiation. The focal position was adjusted to the center of the flow channel. After irradiation, the mixture was collected, centrifuged, washed with ethanol to remove residual salts, and dried under vacuum. The synthesis can be influenced by precursor concentration, substrate properties, and laser parameters (e.g., power and pulse conditions). These parameters were adjusted depending on the system. In a typical experiment, a solution containing 0.05 M Fe, Co, and Ni chlorides, 0.01 M H2PtCl6, 0.01 M RuCl3, and 1 mg mL−1 graphene was processed at a flow rate of 1 mL min−1 under 500 mW laser irradiation. Characterizations The morphology of the samples was characterized by TEM (FEI Tecnai G2 20) with an accelerating voltage of 200 kV. HAADF-STEM images were obtained with an FEI Themis Z microscope at 200 kV. The crystal phases of each sample were identified via powder XRD using a Y-2000 x-ray diffractometer with Cu Kα radiation (40 kV, 40 mA). XPS measurements were conducted using a PHI VersaProbe III XPS spectrometer with nonmonochromatic Al Kα radiation. The metal content was determined by ICP-OES (Agilent 5110). Experimental procedure for MEA testing As shown in fig. S61, the MEA was prepared using a catalyst-coating membrane method. The cathode catalyst loading was calculated based on 20 wt % metal contents. For the MEAs tested, the anode and cathode loadings were 0.1 mgPt cm−2 and 0.4 mgM cm−2, respectively. The anode employed commercial 40 wt % Pt/C (Johnson Matthey), while the cathode utilized the target catalyst. The actual Pt loading was determined by ICP-OES to be 11.14 wt %, corresponding to a Pt loading of 0.04456 mgPt cm−2. Metal activity (MA) was calculated as the measured current divided by the Pt loading (0.04456 mgPt cm−2). The MEA consisted of a Gore membrane (thickness: 8 μm), SGL-29 BC carbon paper as gas diffusion layers (GDLs), Nafion D520 (Dupont, 5 wt %) ionomers, and the catalysts. The catalysts were ultrasonically mixed with deionized water, isopropanol and D520 Nafion into a homogeneous slurry with an ionomer to carbon ratio of 0.65. The ink mixture was sonicated in a 0°C water bath for 2 hours with continuous stirring to ensure proper dispersion. Catalyst-coated membranes (CCMs) were fabricated by spray-coating the ink onto both sides of the membrane at 50°C. The assembled MEAs (active area: 5 cm2) were hot-pressed at 120°C under a pressure of 5 kg cm−2 for 90 seconds. Single-cell performance tests were conducted on a fuel cell test system (Scribner 850) with the operating temperature and relative humidity set to 80°C and 100%, respectively. Backpressure and gas flow rates were held at 200 kPa and 1000/1500 sccm. For H2-O2 and H2-Air polarization testings, the cell was first stabilized at testing conditions under N2 purge. After reaching thermal equilibrium and desired pressure, the gas supply was switched to H2 (anode) and O2 (cathode). A linear current scan mode was employed, starting from 0 A. The scan ended at the current corresponding to a cell voltage lower than 0.3 V. XAFS measurements and in situ XAFS measurements XAFS data were collected at BL14W1 station in Shanghai Synchrotron Radiation Facility (SSRF, operated at 3.5 GeV with a maximum current of 250 mA), respectively. The data were collected at room temperature (Fe, Co, Ni, Ru K-edge and Pt L3-edge in fluorescence excitation mode). All samples were pelletized as disks of 13 mm diameter with 1 mm thickness using graphite powder as binders. In situ XAFS measurements were conducted at the BL14W1 beamline of the SSRF, operated by the Institute of High Energy Physics, Chinese Academy of Sciences. A custom-designed electrochemical cell was used for these experiments. The electrochemical measurements were performed using a computer-controlled electrochemical workstation in a standard three-electrode configuration. The working electrode consisted of carbon paper coated with a PtRuFeCoNi catalyst, the counter electrode was a platinum plate, and an Ag/AgCl (saturated KCl) electrode served as the reference. All electrode potentials were converted to the reversible hydrogen electrode (RHE) scale for reporting. The electrochemical cell was fabricated from polytetrafluoroethylene (PTFE), offering excellent chemical stability and x-ray transparency. It featured a flat wall with a circular aperture of 15 mm in diameter to allow unobstructed x-ray transmission for in situ measurements. The catalyst-coated carbon paper was electrically connected via a copper slip, with the catalyst layer facing inward toward the x-ray beam path. The cell chamber was filled with 0.1 M HClO4 electrolyte solution, and copper tape ensured reliable electrical contact between the electrodes and the electrochemical station. A PTFE top cover was used to securely position the Ag/AgCl reference electrode and to maintain a fixed distance between the reference and working electrodes throughout the measurement. In situ XAFS data were collected in fluorescence mode using an argon-filled Lytle detector. Measurements were performed at three different potentials: open circuit potential (OCP), −0.6 V, and −1.0 V versus the RHE, to investigate the electronic and structural evolution of the catalyst under electrochemical working conditions. Data acquisition and reconstruction of AET The HEA-NSs were deposited on thin Si3N4 films supported on silicon chips. HAADF-STEM tilt series were acquired on an aberration-corrected ThermoFisher Titan Themis Z microscope operated at 300 kV with the convergence semi-angle of 30.0 mrad. The beam current was set to 10 pÅ and the camera length of 145 mm was used. The pixel size of the tilt series is 0.352 Å. Three sequential images were captured with a dwell time of 2 μs to reduce sample drift. The electron dose used for a full tilt series was ∼105 e−/Å2. The details of data processing are presented in the supporting materials.