# Fileset

[Nat Comm-17-2026-4952-Xu.pdf](https://mdr.nims.go.jp/filesets/02af1cef-2460-42c9-a645-1d836d846927/download)

## Creator

[Ya Xu](https://orcid.org/0000-0001-9067-5244), [Keiji Oyoshi](https://orcid.org/0000-0002-0922-935X), Haruka Yoshikawa, [Hiroyuki Takeya](https://orcid.org/0000-0001-9445-4705), [Hiroshi Amekura](https://orcid.org/0000-0003-2148-8431), Takafumi D. Yamamoto, [Yoshitaka Matsushita](https://orcid.org/0000-0002-4968-8905), [Alexei A. Belik](https://orcid.org/0000-0001-9031-2355), [Miyoko Tanaka](https://orcid.org/0000-0003-2650-2845), [Akiko T. Saito](https://orcid.org/0000-0001-5920-5965), [Koji Kamiya](https://orcid.org/0000-0002-6765-4485), [Yoshihiko Takeda](https://orcid.org/0000-0003-4961-3687)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Enhancing hydrogen permeation barrier performance of ErCo2 magnetic refrigeration material via surface oxide layer formation](https://mdr.nims.go.jp/datasets/d1dbe0fa-0aa8-4d2c-98c0-3095b8b15791)

## Fulltext

Enhancing hydrogen permeation barrier performance of ErCo2 magnetic refrigeration material via surface oxide layer formationArticle https://doi.org/10.1038/s41467-026-71547-0Enhancing hydrogen permeation barrierperformance of ErCo2magnetic refrigerationmaterial via surface oxide layer formationYa Xu 1 , Keiji Oyoshi 1,6, Haruka Yoshikawa1,2,7, Hiroyuki Takeya1,Hiroshi Amekura 1, Takafumi D. Yamamoto3,8, Yoshitaka Matsushita 4,Alexei A. Belik 3, Miyoko Tanaka 4, Akiko T. Saito5, Koji Kamiya1 &Yoshihiko Takeda 1The ErCo2 intermetallic compound exhibits a significantmagnetocaloric effectat approximately 32 K and has potential applications as a magnetic refrigera-tion material for hydrogen liquefaction. However, exposure to a hydrogenatmosphere may lead to hydride formation, which weakens the magnetoca-loric effect. Thus, preventing hydrogen permeation into ErCo2 is crucial.Herein, we enhance the hydrogen permeation barrier (HPB) performance ofErCo2particles byusing electrolessCuplating followedbyoxidation treatmentto form a CuO layer with a thickness of a few micrometers. In experiments,ErCo2 particles, with a　1.5- to 5-µm-thick CuO surface layer, exhibited a largemagnetic entropy change of 24 J kg⁻¹ K⁻¹ even after exposure to a H2 atmo-sphere at 1.27MPa and 296K for 7 d. Experimental analyses and first-principlescalculations revealed the potential of CuO as an HPB material for magneticrefrigeration.Hydrogen liquefaction is crucial for its transportation and storage1–3.Magnetic refrigeration offers a more energy-efficient alternative toconventional gas compression/expansion refrigeration technology4–7,and the efficiency of such refrigeration systems can be furtherimproved using magnetic refrigeration materials with large magneto-caloric effects. Numerous magnetic refrigeration materials with largemagnetocaloric effects have been developed, such as DyNi28, ErCo29–11,ErNi212, Dy1-xErxNi213, HoB214, andHoAl215. However, the development ofmagnetic refrigerationmaterials with a largemagnetic entropy change(ΔSM) over a broad temperature range (77–20K) is challenging16–20. Alayered compositemagnetic refrigerant ofHoNi2, DyNi2, andTbNi2 hasbeen developed and obtained an average ΔSM value of 4.7 J kg–1 K–1under a 2 T field change in a broad temperature range of 7.5–53.4 K21.ErCo2 exhibits a ΔSM > 20 J kg–1 K–1 and an adiabatic temperaturechange (ΔTad) of ~10K under a 5 T field change at the Curie tempera-ture of ~32 K9–11, making it suitable for hydrogen liquefaction. Recentstrategies of alloying ErCo2 with Ni, Al, or Fe have expanded theavailable temperature range to 20–77 K22.The shape optimization of magnetic materials is important formaximizing the heat transfer between the magnetic material and heatexchange fluid within the active magnetic regenerative refrigeration(AMRR) system6,7. Magnetic refrigeration materials with a sphericalshape (diameter of 200–500 μm) are expected to increase the coolingefficiency of refrigeration systems because such particles can be den-sely packed in the system, while retaining a certain amount of gapsbetween them for the flow of heat exchange fluid23. Currently, AMRRReceived: 19 May 2025Accepted: 20 March 2026Check for updates1Research Center for Energy and Environmental Materials (GREEN), National Institute for Materials Science (NIMS), 3-13 Sakura, Tsukuba, Ibaraki, Japan.2Faculty of Advanced Engineering, Tokyo University of Science, 6-3-1 Niijyuku, Katsushika, Tokyo, Japan. 3Research Center for Materials Nanoarchitectonics(MANA), National Institute forMaterials Science (NIMS), Namiki 1-1, Tsukuba, Ibaraki, Japan. 4ResearchNetwork and Facility ServicesDivision, National Institutefor Materials Science (NIMS), Sengen 1-2-1, Tsukuba, Ibaraki, Japan. 5Research Center for Magnetic and Spintronic Materials, National Institute for MaterialsScience (NIMS), 3-13 Sakura, Tsukuba, Ibaraki, Japan. 6Present address: Research Network and Facility Services Division, National Institute for MaterialsScience (NIMS), Tsukuba, Japan. 7Present address: Mitsubishi Kakoki Kaisha, Ltd., Kawasaki, Kanagawa, Japan. 8Present address: Department of MaterialsScience and Technology, Tokyo University of Science, Tokyo, Japan. e-mail: xu.ya@nims.go.jpNature Communications |         (2026) 17:4952 11234567890():,;1234567890():,;http://orcid.org/0000-0001-9067-5244http://orcid.org/0000-0001-9067-5244http://orcid.org/0000-0001-9067-5244http://orcid.org/0000-0001-9067-5244http://orcid.org/0000-0001-9067-5244http://orcid.org/0000-0002-0922-935Xhttp://orcid.org/0000-0002-0922-935Xhttp://orcid.org/0000-0002-0922-935Xhttp://orcid.org/0000-0002-0922-935Xhttp://orcid.org/0000-0002-0922-935Xhttp://orcid.org/0000-0003-2148-8431http://orcid.org/0000-0003-2148-8431http://orcid.org/0000-0003-2148-8431http://orcid.org/0000-0003-2148-8431http://orcid.org/0000-0003-2148-8431http://orcid.org/0000-0002-4968-8905http://orcid.org/0000-0002-4968-8905http://orcid.org/0000-0002-4968-8905http://orcid.org/0000-0002-4968-8905http://orcid.org/0000-0002-4968-8905http://orcid.org/0000-0001-9031-2355http://orcid.org/0000-0001-9031-2355http://orcid.org/0000-0001-9031-2355http://orcid.org/0000-0001-9031-2355http://orcid.org/0000-0001-9031-2355http://orcid.org/0000-0003-2650-2845http://orcid.org/0000-0003-2650-2845http://orcid.org/0000-0003-2650-2845http://orcid.org/0000-0003-2650-2845http://orcid.org/0000-0003-2650-2845http://orcid.org/0000-0003-4961-3687http://orcid.org/0000-0003-4961-3687http://orcid.org/0000-0003-4961-3687http://orcid.org/0000-0003-4961-3687http://orcid.org/0000-0003-4961-3687http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-71547-0&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-71547-0&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-71547-0&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-71547-0&domain=pdfmailto:xu.ya@nims.go.jpwww.nature.com/naturecommunicationssystems use a helium gas atmosphere rather than a hydrogen gasatmosphere to liquefy hydrogen via helium gas flow6.The AMRR system could be simplified, and the energy efficiencycould be significantly improved if magnetic refrigeration materialscould be placed directly in a hydrogen atmosphere. However, ErCo2 isprone to hydride formation upon hydrogen exposure24–26, leading toparticle fragmentation and loss of themagnetocaloric effect. Since themagnetic refrigeration materials must be exposed to a hydrogenatmosphere at room temperature for a few days each year during theroutine maintenance period of magnetic refrigeration systems, it isimperative to develop ErCo2-based materials with superior hydrogenpermeation barrier (HPB) performance that are effective not only atcryogenic temperatures but also at room temperature.Previous studies have exploredHPB coatings for steelmaterials toprevent hydrogen embrittlement and leakage in nuclear powerplants27–29. Oxides such as Cr2O3-SiO2, Al2O3, Y2O3, Er2O3, La2O3, andSiO2 have been investigated owing to their lowhydrogen solubility andpermeability30–32. They are typically coated via a variety of techniques:hot-dipping, thermal spraying, plasma-spraying, physical vapordeposition (PVD), and chemical vapor deposition. We attempted touse hot-dipping, thermal spraying, and barrel PVD methods to form aCr2O3 or Al₂O₃ layer onto the ErCo₂ particles. However, particleaggregation is difficult to prevent via thermal spraying, and the coatinguniformity is hard to control on the small ErCo2 particles via the PVDmethod.Compared to these coatingmethods, the formation of aCuO layerby copper plating followed by oxidation treatment offers greatadvantages in terms of cost and efficiency. CuO has not previouslybeen recognized as an HPB material, and this study has demonstratedfor the first time that it possesses sufficient HPB properties near roomtemperature.Herein, we report our work to synthesize ErCo₂ magnetic refrig-eration materials with excellent HPB performance using well-established techniques, i.e., preparing ErCo2 particles with diametersof 212–355 µm using an electrode induction melting gas atomization(EIGA) process33–36, forming a Cu layer on the ErCo2 particles usingelectroless plating, and forming a CuO layer using the subsequentoxidation treatment. To compare the HPB performance of differentoxide layers, we formed a Co(Er) oxide layer through oxidation treat-ment alone and a CuO layer through a combination of electroless Cuplating and oxidation treatment on the particle surface. The HPBperformance of the particles with these layers was evaluated using aSieverts-type instrument (Fig. S1) at an initial H2 pressure of ~1.27MPaand 296K. The effects of the oxide layers on the magnetic propertiesof ErCo₂ particles were evaluated using a Quantum Design SQUIDmagnetometer. The ErCo2 particles with a Co(Er) oxide layer sig-nificantly improved the HPB performance, while particles with a CuOlayer further improved it, almost completely blocking hydrogen per-meation for 7 d under the abovementioned hydrogen exposure con-ditions. The CuO layer was analyzed via synchrotron X-ray diffraction(SXRD), scanning electron microscopy (SEM), and transmission elec-tron microscopy (TEM). Time-of-flight secondary ion mass spectro-metry (ToF-SIMS) provided experimental evidence that hydrogen isblocked at the surface. Density functional theory (DFT) calculationsrevealed that H2 is energetically favored for physisorption on the CuOsurface and that its dissociation to H atomsmust overcome an energybarrier of >20 kJ/mol, while diffusion of the H atom into CuO requiresovercoming an even larger energy barrier ( > 300 kJ/mol). The DFTcalculation results agreed with the experimental results, demonstrat-ing the potential of CuO as an HPB material.Results and discussionHPB performance and magnetic propertiesThe effects of different oxide layers on the HPB performance of ErCo2particles were investigated by comparing three types of samples: (1)particles after homogenization at 1123 K for 7 d in an Ar atmosphere(referred to as “as-homogenized”), (2) the homogenized particles afteroxidation at 773 K for 0.5 h in air (referred to as “oxidized”), and (3) thehomogenized particles after electroless Cu plating and subsequentoxidation at 773K for 0.5 h in air (referred to as “Cu-plated/oxidized”).Figure 1a shows the hydrogen pressure changes over time for thesesamples under an initial hydrogen pressure of ∼1.27MPa at 296 K, andFig. 1b shows the corresponding absorbed hydrogen amounts calcu-lated from the pressure changes.For the as-homogenized ErCo2 particles, the hydrogen pressurerapidly decreased from 1.27 to 1.08MPa within ∼1 h, corresponding toan absorbed hydrogen amount of ~1.311 mass% of the sample weight—comparable to that observed in typical hydrogen storage alloys37,38.Stereomicroscopy reveals that the ErCo2 particles pulverized afterhydrogen exposure (Fig. S2a, S2b), which could be due to the largeamount of absorbed hydrogen in ErCo2.In contrast, the oxidized ErCo2 particles underwent slowerhydrogen absorption, achieving a significantly reduced absorbedhydrogen amount. The initial hydrogen pressure of 1.271MPa startedto decrease only after 14 h and reached ~1.245MPa after 167.5 h, cor-responding to an absorbed hydrogen amount of 0.182 mass%. Ste-reomicroscopy images (Fig. S2c, S2d) showed that most of theparticles retained their shape, and only a few were pulverized. Theseresults demonstrate that the oxidation treatment substantiallyimproved the HPB performance of ErCo2 by restricting hydrogenpermeation.Further improvement was observed for the Cu-plated/oxidizedparticles. The hydrogen pressure (initially 1.272MPa) remained almostunchanged, decreased slightly after 60 h, and stabilized at 1.255MPaafter 166.75 h. The corresponding absorbed hydrogen amount was0.117 mass%, indicating even slower absorption than that upon oxi-dation alone. Stereomicroscopy confirmed that most particlesretained their shape after hydrogen exposure (Fig. S2e, S2f). The as-homogenized ErCo2 particles exhibited low HPB performance under1.27MPa hydrogen at room temperature, while oxidation significantlyimproved the HPB properties. Moreover, Cu plating followed by oxi-dation further improved the HPB performance, effectively preventinghydrogen permeation for over 100h.Pressure–composition–temperature (PCT) measurements wereperformed for the as-homogenized, oxidized, and Cu-plated/oxidizedErCo2 particles (Fig. S3). For the as-homogenized particles, no obvioushydrogen absorption was observed up to 4MPa at 303K, whereas theamount of hydrogen absorption reached approximately 1.3 mass% at333 K. The PCT results are consistent with the hydrogen exposure testresults, considering that the PCT measurement was completed in lessthan anhour (0.57 h) at 303 Kandhydrogen absorption commenced inthe as-homogenized particles during the hydrogen exposure test(Fig. 1a). In addition, the amount of hydrogen absorption obtainedfrom the PCTmeasurement at 333 K agrees well with the value of 1.311mass% obtained from the hydrogen exposure test. The absorbedhydrogen maintained approximately 1 mass% even after reducing thepressure to a low level ( < 0.005MPa), suggesting that hydrogenabsorbed by ErCo2 does not easily desorb.For the oxidized particles, no obvious hydrogen absorption wasobserved at 303 K, and only a small amount of hydrogen absorption( ~ 0.2 mass%) was observed at 333 K (Fig. S3b). For the Cu-plated/oxidized particles, no apparent hydrogen absorption was observed atboth 303 and 333K (Fig. S3c). These PCTmeasurements are consistentwith thoseof thehydrogen exposure test (Fig. 1a, b), indicating that theHPB performance was improved by the oxidation and furtherenhanced by Cu-plated/oxidation treatment.Furthermore, we performed long-term hydrogen exposure testsat room temperature up to 28 days for evaluating long-term hydrogenpermeation stability of oxidized only and Cu-plated/oxidized ErCo2particles. Particles after different Cu plating times (0.17 and 1.5 h) wereArticle https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 2www.nature.com/naturecommunicationsused for examining the influence of CuO layer thickness on HPB per-formance (Fig. S4). For all samples, the decrease in hydrogen pressurethroughout the test period was less than 2%, indicating an extremelyslow hydrogen absorption rate. This demonstrated that both oxidizedand Cu-plated/oxidized ErCo₂ exhibit excellent long-term HPB stabi-lity. Among these, the Cu-plated/oxidized particles showed slightlysuperior HPB characteristics compared to the oxidized particles. Fur-thermore, samples with shorter Cu plating time (0.17 h) exhibited HPBFig. 1 | Hydrogen permeation barrier performance andmagnetic properties ofthe ErCo2 particles. a Hydrogen pressure changes during the hydrogen exposuretest for the as-homogenized, oxidized only, andCu-plated/oxidizedErCo2 particles.The inset shows the hydrogen pressure change during the initial ten hours.b Corresponding absorbed hydrogen amount calculated from the pressure chan-ges. Thehydrogenexposure test startedwith an initial hydrogenpressure in the cellof 1.27MPa at approximately 296 K. c Temperature dependence of magnetizationat 0.01T for the as-homogenized ErCo2 particles. d Temperature dependence ofmagnetization at 0.01 T for the oxidized ErCo2 particles before and after thehydrogen exposure test. e Temperature dependence ofmagnetization at 0.01T forthe Cu-plated/oxidized ErCo2 particles before and after the hydrogen exposuretest. f Magnetic entropy change (ΔSM) for μ0 ΔH = 5 T of the as-homogenized andoxidized ErCo2 particles before and after the hydrogen exposure test. g Magneticentropy change (ΔSM) for μ0 ΔH = 5 T of the as-homogenized and Cu-plated/oxi-dized ErCo2 particles before and after the hydrogen exposure test (ΔSM is notshown for the as-homogenized particles after the hydrogen exposure test). Sourcedata are provided as a Source Data file.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 3www.nature.com/naturecommunicationscharacteristics no less than those with long Cu plating time (1.5 h)(TableS1). As the thicknessof theCuO layer formedon the samplewithshorter Cu plating time (0.17 h) was approximately 1.5μm(Fig. S5), thisresult suggests that a 1.5 μⅿ-thick CuO layer can provide a sufficientHPB effect.The effect of the Cu plating–oxidation treatment on themagneticproperties of ErCo2 was investigated by measuring the M–T curvesunder varyingfield up to 5 T for the as-homogenized, oxidized, andCu-plated/oxidized samples before and after H2 exposure test (Fig. S6) tocalculate the magnetic entropy change (4SM) using Eq. 3. Figure 1c, dshow theM–T curves of the as-homogenized, oxidized, and Cu-plated/oxidized particles, measured at 0.01 T before and after hydrogenexposure. In the as-homogenized particles, magnetization sharplyincreased at ~32 K (the Curie temperature) owing to the first-orderferrimagnetic transition9,10. After hydrogen exposure, this transitionwas absent, replaced by a rapid magnetization increase below 10K—suggesting ErCo2 conversion tohydride, which significantly altered themagnetic properties. Conversely, the oxidized and Cu-plated/oxidizedparticles retained a clear magnetization increase at ~32 K, both beforeand after hydrogen exposure, confirming that the Cuplating–oxidation treatment blocks hydrogen permeation and pre-serves the magnetic properties of ErCo2.Figure 1f, g present ΔSM for the as-homogenized, oxidized, andCu-plated/oxidized samples under μ0ΔH = 5 T. Themaximum ΔSM was~33 J kg⁻¹ K⁻¹ for the as-homogenized particles, ~26 J kg⁻¹ K⁻¹ after theoxidation treatment, and ~24 J kg⁻¹ K⁻¹ after the Cu plating−oxidationtreatment. The temperature range of large ΔSM (32–42 K) remainedunchanged. After hydrogen exposure, both ΔSM value and tempera-ture range remained stable for the oxidized and Cu-plated/oxidizedsamples. The reduction in ΔSM is due to the decrease in the weightfraction of ErCo2main phase in the ErCo2 particle after Cu plating and/or oxidation treatments. As the ΔSM data in the present study areshown in units of J per K per total mass of ErCo2 particles, their mag-nitude decreases proportionally to the weight fraction of the ErCo2phase. From the microstructural analysis results shown in Fig. 3, weobtained an average thickness of each phase by measuring 3–5 parti-cles and evaluated the weight fraction of each phase in the ErCo2particle: the innermost ErCo2 phase, the Er–Co–O layer, the Co–Olayer, and the outermost CuO layer (Fig. S7). The weight fraction of theErCo2 phase decreased by approximately 29% after Cu plating andoxidation,which roughly corresponds to the 30%decrease inmagneticentropy change. For the oxidized-only sample, we also confirmed thatthe decrease in the weight fraction of the ErCo2 phase corresponds tothe decrease inmagnetic entropy via the same process (Fig. S8). Theseresults indicate that while the magnetocaloric effect per unit weightafter oxidation or Cu plating–oxidation treatments is smaller, itmaintains a high intensity of this effect and effectively preventshydrogen-induced deterioration of the magnetic properties of ErCo2.Characterization of ErCo2 and surface oxide layerThe changes in the crystal structure of the ErCo2 particles during theCu plating–oxidation treatment were evaluated through SXRD. Fig-ure 2a presents the SXRD profiles of the as-homogenized, Cu-plated,and Cu-plated/oxidized ErCo2 particles before and after hydrogenexposure. For the as-homogenized particles, most peaks corre-sponded to the ErCo2 phase (cubic, with space group symmetryFd�3m), accompanied by weak peaks indexed to the ErCo3 (hexagonal,R�3m) and HT- & LT-Er2O3 (Ia�3and C2=m) phases. Qualitative analysisusing the reference intensity ratio revealed the mass percentages ofErCo2, ErCo3, and Er2O3 (Sum of HT- and LT-phase values) to be 94%,Fig. 2 | Structural characterization of the ErCo₂ particles via synchrotronX-raydiffraction. a XRD patterns of the as-homogenized (1), Cu-plated (2), and Cu-plated/oxidized ErCo2 particles (3) and the particles after the hydrogen exposuretest (4). bMagnified viewof the 2θ = 17°–22.5° range, indicating the presence of the111Cupeak afterCuplating, alongwith 111CuO and 111CoOpeaks afteroxidation. cUnitlattices of the identified phases: ErCo2, Cu, CuO, andCoO. Source data are providedas a Source Data file.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 4www.nature.com/naturecommunications5.5%, and 0.5%, respectively (Fig. S9). The trace amounts of the Er2O3phase likely resulted from the formation of a natural surface oxidelayer upon air exposure, as confirmed by cross-sectional analyses viaSEM and energy-dispersive X-ray spectroscopy (EDS; Fig. S10).After Cu plating, in addition to the aforementioned phases, traceamounts of the metallic Cu phase (cubic, Fm�3m) were detected. Fol-lowing the oxidation of the Cu-plated sample, ErCo2 remained as themain phase, while the metallic Cu phase disappeared and traceamounts of CuO (C2=c) and CoO (cubic, Fm�3m) were observed. Amagnified view of 2θ = 17°–22.5° clearly revealed the presence of a111Cu peak after Cu plating and 111CuO and 111CoO peaks after oxidation(Fig. 2b). After hydrogen exposure, the SXRDprofiles of the Cu-plated/oxidized particles remained unchanged, indicating that the oxidephases persist even after hydrogen exposure at 1.27MPa and ~296K.This reveals that CuO is not reduced to its metallic state under theseconditions, which is consistent with previous reports suggesting thatCuO is difficult to reduce at room temperature39,40. The unit latticeschematics of the identified phases are shown in Fig. 2c. These resultsindicate that small amounts of oxides (CuO, CoO, and LT-Er2O3) wereformed after Cu plating and the subsequent oxidation; however, theprimary phase remained as ErCo2.The surface and cross-sectional morphologies of the as-homogenized ErCo2 particles were examined through SEM. Figure 3ashows a secondary electron (SE) image revealing surface irregularities,and Fig. 3b shows a backscattered electron (BSE) image indicating alargely uniform composition, except for some small dark contrastareas. EDS identified these areas as Co-rich regions (Fig. S10), whichlikely corresponded to the ErCo3 phase, as supported by the SXRDresults. EDS line analysis along the particle diameter revealed a thin Er-and O-enriched surface layer, suggesting the presence of a natural Eroxide film (Fig. S10).ErCo2 particles with Er and Co oxide surface layers were preparedvia the oxidation treatment. Figure 3c shows the SE image and EDS lineanalysis results of an oxidized particle cross-section, revealing a two-layered structure. The outermost layer (1–2μm thick) exhibited a darkcontrast and consisted mainly of Co oxide. Beneath it, a 5- to 20-μm-thick intermediate layer appeared brighter than the outermost layerbut darker than the interior. EDS indicated the presence of Er and Cooxides in this layer, suggesting the partial oxidation of ErCo2. Crackswere observed within some areas of this oxide layer (Fig. 3d), indi-cating its brittleness and lack of densification.ErCo2 particles with a CuO surface layer were prepared via the Cuplating–oxidation treatment. To ensure uniform plating, a brief pre-treatment with dilute hydrochloric acid was conducted before Cudeposition, forming an oxide layer with a larger atomic ratio of Co/Ercompared with that of the subsurface region (Fig. S11). The dissolutionof Er was more pronounced than that of Co, leading to a higher Coconcentration on the surface. After Cu plating, a 10 to 15-μm-thick darkcontrast layer was formed (Fig. 3e, f). EDS identified Cu as the domi-nant element in the outermost region, with Co, O, and Er in theunderlying region within this layer (Fig. 3g). The Co/Er ratio in theunderlying region (1.9) was higher than that in the adjacent interior(1.2), likely because of the acid pretreatment.Subsequent oxidation of the Cu-plated particles led to furtherstructural changes. The subsurface layer formed beneath the surfacelayer with dark contrast exhibited an intermediate contrast (Fig. 3h, i).EDS analysis (Fig. 3j) confirmed that the outermost layer primarilyconsisted of CuO, whereas the underlying layer contained CoO withFig. 3 | Analysis of surface and cross-sectional morphologies of the ErCo2 par-ticles via SEM. a SE images of the ErCo2 particle produced by atomization afterhomogenization at 1173 K for 7 days in an Ar gas atmosphere. b BSE image of thecross-section of the as-homogenized particle. c Cross-sectional SE image of theoxidized ErCo2 particle and the results of EDS linear component analysis. d Cross-sectional SE image of the oxidized ErCo2 particle with an enlarged view showingsmall cracks in the surface layer. e Cross-sectional BSE image of the Cu-platedparticle. fCross-sectionalBSE imageof the surface layer ofCu-platedparticle.g EDSresults for the positions marked in (f). h Cross-sectional BSE image of the Cu-plated/oxidized ErCo2 particle. i Cross-sectional BSE image of the surface layerregion in h. j EDS results for the positions marked in (i).Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 5www.nature.com/naturecommunicationsminor Er oxides. The subsurface layer contained Co, Er, and O, with anO concentration (32 at%) lower than that in the surface layer (41 at%)but significantly higher than that in the interior (Fig. 3g), indicating thepresence of distinct Er and Co oxides.The surface-layer structures of the Cu-plated/oxidized particleswere further analyzed by TEM. Figure 4a presents a high-angleannular dark-field scanning TEM (HAADF-STEM) image with EDSelemental mapping, which reveals a continuous CuO outer layer ontop of an underlying Co-rich oxide layer withminor Cu and Er oxides.Figure 4b shows the TEM image of the selected region, with selected-area electron diffraction (SAED) patterns confirming the crystallinityof CuO and CoO. Figure 4c presents the high-resolution TEM(HRTEM) image and the fast Fourier transform (FFT) pattern of theCuO layer, indicating well-crystallized CuO(111). These results indi-cate that a well-crystallized CuO layer with large crystallites forms onthe outer surface, whereas the subsurface Co-, Er-, and Cu-oxidelayers exhibit some missing regions, which could be due to a weakinterface between the subsurface oxide layer and the outmost sur-face of the CuO layer.Mechanism of HPB performance improvementHydrogen permeation involves adsorption, dissociation, anddiffusion41–43. To determine the rate-limiting step for ErCo2 with theCuO surface layer, ToF-SIMSwas conducted to evaluate the deuteriumdistribution in the surface oxide layer after exposure to D2. Figure 5ashows the ToF-SIMS depth profiles of 167Er, 59Co, 16O, 2H, and 1H/100 inthe oxidized ErCo2 particles after D2 exposure. In the outermost ~3 µm,the peak intensity of 59Cowas higher, that of 167Er was lower, and that of16O was higher than that of the deeper layers ( ~ 40 µm), suggesting theformation of a Co-enriched oxide layer, consistent with SEM results(Fig. 3). Beneath this layer, up to ~20 µm, the peak intensities of 167Er,59Co, and 16O were relatively constant, whereas with further increase ofthe depth, the peak densities of 167Er and 16O decreased, while that of59Co remained constant, indicating the presence of an Er-rich oxidelayer—again consistent with the SEM results (Fig. 3). The calculated Dconcentration in the CoO layer significantly reduced, reaching nearlyzero at the interface of the next layer (Fig. 5b), suggesting that deu-terium was primarily blocked by the CoO layer.Figure 5c shows the ToF-SIMS depth profiles of 167Er, 59Co, 16O,63Cu, 2H, and 1H/100 in the Cu-plated/oxidized ErCo2 particles after D2exposure. The outermost surface exhibited strong signals for 63Cu, 2H,and 1H/100, along with a relatively high 16O intensity. Moreover, the59Co intensity was at similar levels, and the 167Er signal was weak com-pared to their intensities at deeper layers ( ~ 40 μm). With an increasein depth, the intensities of the 63Cu, 2H, and 1H/100 signals rapidlydecreased, while those of the 167Er, 59Co, and 16O signals increased,reaching a maximum at ~4 μm. Beyond this depth, the 59Co signalintensity gradually declined, while the 167Er and 16O signal intensitiesremained constant up to ~20 μm, after which all intensities declined.These findings align with the SXRD, SEM, and TEM results, confirmingthe formation of a CuO outer layer (several micrometers thick) fol-lowed by mixed Cu–Co–Er oxide layers. Figure 5d shows that the Dconcentration of the CuO layer significantly decreased and reachedzero in the underlying mixed oxide layer. This suggests that the CuOlayer prevents hydrogen permeation, with additional blocking fromthe mixed Cu–Co–Er oxides.These results indicate that both CoO and CuO have a significantinhibition effect on hydrogen permeation. However, as shown in Fig. 1,the oxidized particles with the Co(Er)O surface layer did not allowsufficient hydrogen permeation blocking. This was likely due to thepresence of fine cracks in the Co(Er)O oxide layer formed on someparticles (Fig. 3d), which resulted in hydrogen permeation into parti-cles. In contrast, the CuO layer was denser (Fig. 4) and exhibited astronger hydrogen blocking effect.DFT calculations were performed to further investigate thehydrogen adsorption, dissociation, and diffusion on the CuO(111)Fig. 4 | Microstructural characterization of the surface layer via TEM. aHAADF-STEM image of the surface layer area of the Cu-plated/oxidized ErCo2 particle andthe corresponding EDS elemental mapping images of Cu, O, Er, and Co. b TEMimage of the marked region in a and SAED patterns for the outer and subsurfacelayers. c HRTEM image of the CuO outer layer; the inset shows the correspondingFFT pattern.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 6www.nature.com/naturecommunicationssurface. The adsorption energies and optimized geometries were cal-culated for an H2 molecule at various sites on CuO(111), includingthree-coordinated O (hereafter denoted as O3) and Cu (as Cu3), four-coordinatedO (asO4) andCu (asCu4), 11 bridge sites, and8hollow sites(Fig. S12a). Five molecular orientations were considered: one perpen-dicular to the CuO(111) surface, two parallel, and two tilted approxi-mately 45° toward the z-axis from the two parallel orientations (Fig.S12c). Initially, H2 was positioned ~1.5 Å from the surface, with an H–Hbond length of 0.74 Å. After optimization, the H2 molecule movedaway from the surface (2.8–6Å) (Fig. 6a), with minimal bond lengthchange (0.74–0.75 Å) and low adsorption energies (−2.71–2.36 kJ/mol)(Table S2_1 and S2_2). For example, a H2 molecule initially at theCu4–Cu4 bridge site (shown as ⑩ Bridge-Cu4-Cu4 in Fig. S12a) andvertical to the bridge direction was set 1.7409Å from the nearestsurface atom (Fig. 6b) andmoved to 3.6999 Å from the nearest surfaceatom after optimization. In contrast, the H–H bond length changedonly from 0.7406 to 0.7497 Å (Fig. 6c), and the adsorption energy wasonly 0.68 kJ/mol (Table S2_1). The results indicate that H2 prefersphysisorption and requires an energy barrier for chemisorption anddissociation on the CuO(111) surface.The activation barriers for H2 dissociation at the optimized siteswere calculated for two cases: (i) bothH atoms adsorbedonto adjacentO3 atoms (forming twoH–Obonds) and (ii) one H atom adsorbed ontoO3 and the other onto Cu3 (forming H–O and H–Cu bonds). Theactivation barriers ranged from 27.7 to 152.4 kJ/mol for H–O bondformation and from 40.4 to 63.8 kJ/mol for H–O and H–Cu bond for-mation (Fig. S13). The lowest activation energy (27.7 kJ/mol) was foundat the Cu4–Cu4 bridge site for H–O bond formation, while 40.4 kJ/molwas the lowest for H–O and H–Cu bond formation. This suggests thatH2 dissociation favored H–O bond formation (Fig. 6d). In contrast, thediffusion barrier for H migration from H–Cu to H–O was much lower(13.5 kJ/mol) (Fig. 6e), suggesting that H atoms diffuse easily on theCuO surface. Moreover, the energy barrier for H diffusion from thesurface to subsurface was 337.4 kJ/mol (obtained from the differencebetween the total energy ofTS (240.8 kJ/mol) and thatof theH–O/H–Oabsorbed surface state (–96.6 kJ/mol, Fig. 6e), indicating that subsur-face diffusion is rate-limiting.The effects of H2 molecule surface coverage on H2 adsorptionstructure on CuO (111) surface were examined by setting various sur-face coverage rates of H2 molecules on CuO(111) surface. The resultsrevealed that H2molecules moved away from the surface at all surfacecoverage rates (0.0625–1ML, Fig. S14). The effect of H-atom surfacecoverage on H adsorption structure on CuO(111) was also examined bysetting various surface coverage rates of H atoms on the CuO(111)surface. The results revealed thatH atomsmost readily adsorb ontoO3atoms, followed by Cu3 atoms, with the increase of H coverage (Fig.S15). The adsorption energy of an H atom on the CuO (111) surfacedecreased with the increase of H coverage up to 0.5ML (Fig. S16).Fig. 5 | Analysis of hydrogen permeation behavior via time-of-flight secondaryion mass spectrometry (ToF-SIMS). a ToF-SIMS depth profiles of 167Er, 59Co, 16O,2H, and 1H/100 in the surface layer of the oxidized ErCo2 particles afterD2 exposure.b Normalized yield of D of the oxidized ErCo2 particles obtained by excluding thebackground H signal: D yield = detected 2H signal−detected 1H signal/100. c ToF-SIMS depth profiles of 167Er, 59Co, 16O, 63Cu, 2H, and 1H/100 in the surface layer of theCu-plated/oxidized ErCo2 particles after D2 exposure. d Normalized yield of D ofthe Cu-plated/oxidized ErCo2 particles. Note that the vertical axis in (a) and (c) isthe number of secondary ions counted and does not correspond to the elementalcomposition ratio. Source data are provided as a Source Data file.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 7www.nature.com/naturecommunicationsFurthermore, when the H surface coverage exceeds 0.5ML, additionalH atoms can no longer adsorb onto the surface. They detach from thesurface and form hydrogen molecules (Figs. S15 c.4 and d.4).The hydrogen atom solid solubility limit in CuO was evaluated bycalculating the solution energies when inserting H atoms sequentiallyup to 9 atoms into interstitial octahedral sites in a larger slab modelcomprising 256 atoms (Figs. S17 and S18). TheH solubility atΔG_sol = 0was determined as 0.01377 at% by extrapolation, indicating an extre-mely low H solubility in CuO at 298K and 1.27MPa (Fig. S19). TheseDFT results align with the ToF-SIMS findings (Fig. 5d), revealing thatthe CuO layer effectively blocks hydrogen permeation.We successfully synthesized an ErCo2 magnetic refrigerationmaterial with excellent HPB performance at 1.27MPa and ~296 Kthrough electroless Cu plating followed by oxidation. The CuO layereffectively blocked hydrogen permeation because of its high activa-tion energy for H diffusion ( ~ 337 kJ/mol). This low-cost and efficientapproach can be applied to other materials that require HPB proper-ties near room temperature.MethodsFabrication of ErCo2 spherical particlesA master alloy ingot of ErCo2 was prepared using high-frequencyvacuum melting equipment with pure raw materials of Er (99.9%,Nippon Yttrium Co., Ltd.) and Co (99.9%, SumitomoMetal Mining Co.,Ltd.). ErCo2 spherical particles were then fabricated from the ingotusing an EIGA process33–35. The ingot, serving as an electrode rod, wasinductivelymeltedwithout a crucible, enabling the production of high-purity spherical particles. After atomization, the particles were sieved,and those with diameters of 212–355μm were collected andhomogenized at 1123 K for seven days in an Ar gas atmosphere forsubsequent experiments.Oxidation treatments and electroless Cu platingOxidation treatments were conducted on both the as-homogenizedand Cu-plated particles using a box-type electric furnace in air. Theoxidation conditions were varied within the ranges of 473 to 773K and0.1 to 0.5 h. Oxidation at 773 K for 0.5 h wasmore effective for formingthe CuO layer. The samples were heated to 773 K at a rate of ~15 K/min,held at this temperature for 0.5 h, and then slowly cooled to roomtemperature in the furnace.Electroless Cu platingwas performed using a plating solutionwiththe composition listed in Table S3, at a pH of 12–12.5. The plating wasconducted at 248K for 1.5 hwithmagnetic stirring. Prior to plating, theErCo₂ particle surfaces were cleaned by immersion in 4 vol% HClaqueous solution at 298K for 20 s, followed by rinsing with purifiedwater. This pre-cleaning step was essential for ensuring uniform Cudeposition, as some particles failed to be plated without it (Fig. S20).SEM and EDS analyses revealed that a very thin Er–Co oxide layerexisted on the surface of as-homogenized ErCo2 particles, formedduring the homogenization heat treatment process or subsequentstorage at room temperature in air. This oxide layer is believed toaffect the copper plating process and needs to be removed by pre-cleaning. The short pre-cleaning time of 20 s was used to avoidunnecessary sample elution (Table S4).Evaluation test of HPB performanceThe HPB performance of ErCo₂ particles, both before and after Cuplating, was evaluated using a Sieverts-type instrument (SuzukiFig. 6 | Analysis of hydrogen adsorption, dissociation, and diffusion onCuO(111) surface via DFT calculation. aOptimizedpositions ofH2molecules at allinitial adsorption sites (shown in Fig. S4a), showing that all H2 molecules movedaway from the CuO(111) surface, to a distance exceeding 2.6 Å. bH2 set at the initialCu4–Cu4 bridge site⓾ shown in Fig. S12a, with an orientation parallel to the surfaceand vertical to the bridge direction. cOptimizedpositions of theH2molecule at theinitial adsorption site shown in (b).dActivation barriers for H2 dissociation to formtwo H–O bonds (H-O/H-O) or one H–Cu and one H–O bond (H-Cu/H-O) (TS: tran-sition state). e Activation barriers for H-atom diffusion from H–Cu/H–O to H–O/H–O on the surface and from surface H–O/H–O to subsurface H–O/H–O.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 8www.nature.com/naturecommunicationsSyokan Co., Ltd.). A schematic of the hydrogen exposure systemand detailed experimental setup are provided in the SupportingInformation (Fig. S1). For each measurement, 0.5 g of particleswas used. Hydrogen gas was introduced at an initial pressure of~1.27 MPa after vacuuming with a rotary pump for at least 15 min.The average relative error of the pressure gauge (PG-50KU-F,KYOWA Electronic Instruments Co., Ltd., Japan) is ±0.5%. Thecalibration sheet of the gauge shows its rate output is 1997 μV/V,and the calibration constant is 0.00245MPa/ 1μV/V. The pressurechange in the sealed cell was monitored for up to seven days at293–298 K to assess the HPB properties of the particles. We per-formed the D2 exposure experiment at the same temperature andpressure as that for the hydrogen exposure experiments, and theduration was also very close, i.e., 137.5 h for oxidized particles and139 h for Cu-plated/oxidized particles.The amount of hydrogen adsorbed by the ErCo₂ sample (Δn, mol)at each time point was calculated using Eq. (1), derived from the idealgas law:4n= ðP0VRT0� PiVRTiÞ ð1ÞwhereP0 andT0 are the initial hydrogenpressure (Pa) and temperature(K) in the cell, respectively, while Pi and Ti represent the pressure andtemperature at the i-th measurement. V is the total volume of the cell,reservoir, and gas pipeline (m³), and R is the molar gas constant(8.314462 J K⁻¹ mol⁻¹).The hydrogen absorption capacity of the ErCo₂ particles wascalculated as:Absorbed hydrogen ðmass%Þ=weight of absorbed hydrogen=weight of ErCo2 × 100ð2ÞCharacterizationThe morphology and composition of ErCo2 particles, before and afterCu plating and oxidation treatments, were analyzed using SEM (JEOL,JSM-7000F, JSM-6500F) and TEM (JEOL, JEM-ARM200F) coupled withX-ray energy-dispersive spectroscopy. Cross-sectional SEM sampleswere prepared by mechanically polishing the particles affixed to aconductive adhesive. TEM samples were prepared from SEM cross-sections using a focused ion beam technique, with details provided inthe Supporting Information (Fig. S21). The crystalline structures ofErCo₂ particles at various stages of treatment were analyzed via SXRD(SPring-8, BL02B2). The wavelength of the incident beam (λ) was0.775980Å.To assess the effects of Cu coating and oxidation on themagnetic properties of ErCo₂ particles, temperature-dependentmagnetization (M–T) measurements were conducted using aQuantum Design SQUID magnetometer. Measurements weretaken from 2 to 60 K under applied magnetic fields ranging from0.001 to 5 T. All the M(T) measurements were performed underfield-cooling processes. The temperature sweep rate was 0.6 Kper min; the mass of the samples was approximately 3–4mg.Demagnetization corrections were not performed in this studybecause it is difficult to evaluate the exact demagnetization fac-tors for a group of particles. The demagnetization effect does notsignificantly affect the evaluation of ΔSM for large magnetic fieldssuch as 5 T44. The magnetic entropy change (ΔSM) was calculatedusing the following equation45.�4SM = � μ0Z H0∂M∂T� �HdH ð3Þwhere μ0 is the permeability of vacuum, H is the external magneticfield, T is the temperature, and M is the magnetization. The measure-ment error of ΔSM is mainly from the weight error of the particlesamples used in the measurements, which is estimated to beapproximately 1–2%.ToF-SIMS analysis of the particles after the deuterium exposuretest was performed using a time-of-flight secondary ion mass spec-trometer (PHI, TRIFT V nanoTOF). A Ga emitter with 30 kV was used tosputter the particle surface. Deuterium (an isotope of hydrogen) wasselected because it exhibits diffusion behavior similar to that ofhydrogen46,47 and has a detection limit 1–2 orders of magnitude lowerthan that of hydrogen in ToF-SIMS48,49. D2 also eliminates interferencefrom trace hydrogen originally present in the sample and chamber.The D concentration was derived from the 2H signal by subtracting thebackground signal of 1H/100 intensity, which was confirmed to appearat the 2H position in all measurements.DFT calculationThe adsorption energies and the dissociation energy of molecularhydrogen on the CuO (111) surface were calculated using the PHASE0first-principles program package based on DFT (https://azuma.nims.go.jp/)50. The exchange-correlation energy was described using thegeneralized gradient approximation (GGA)51–53. A slab model com-prising 128 atoms in four layers (4 × 8 atoms per layer) was used with avacuum layer of 15 Å to prevent interactions between periodic images.The Brillouin zonewas sampled using a 2 × 2 × 1 k-pointmesh, with cut-off energies of 450eV for the wave function and 4050eV for thecharge density. The state determination was performed using a con-vergence criterionof 5 × 10–6 eV for total energy and a forcecriterionof0.01 eV/Å for force action.The zero-point energy (ZPE) correction was performed by calcu-lating vibration modes using the linear response function of PHASE0after optimizing the slabmodel and each adsorption state, and the ZPEenergy was calculated as follows:ZPE =12Xni = 1hγi� �ð4Þwhere γi is the frequency of each actual value vibration mode, n is thetotal numbers of actual value vibrationmode, and h is Planck constant(4.135667696 × 10-15eV·s). By this method, the ZPE value of the free H2molecule was calculated to be 0.273436 eV, which agrees well with theliterature value54.The adsorption energy of a single molecule on the surface (Eabs)was calculated as follows:Eads = ðEslab +ZPEslabÞ+ ðEH2+ ZPEH2Þ � ðEH2=slab+ZPEH2=slabÞ ð5Þwhere Eslab is the total energy of the slab, EH2is the total energy of anisolated hydrogen molecule, EH2=slabis the energy of the slab withadsorbed hydrogen, ZPEslab is the ZPE of the slab, ZPEH2is the ZPE of afree H2 molecule, and ZPEH2=slabis the ZPE of the slab with adsorbedhydrogen. A larger Eads indicates stronger hydrogen adsorption.The activation energies for H2 dissociation and atom diffusionon CuO(111) surface were determined using the climbing imagenudged elastic bandmethod55. The transition states were determinedby interpolating 6 images between the initial and final states. Thetransition state determination was performed using a convergencecriterion of 5 × 10–6 eV for total energy and a force criterion of0.01 eV/Å for force action. The k-point mesh was 2×2×1. The cut-offenergies were 450 eV for the wave function and 4050 eV for thecharge density.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 9https://azuma.nims.go.jp/https://azuma.nims.go.jp/www.nature.com/naturecommunicationsThe effects of H2 surface coverage on H2 adsorption structure onCuO (111) surface were examined by setting various surface coveragerates of H2 molecules on CuO(111) surface (0.0625–1ML) (Fig. S14). Inthe initial state, the H2 molecules were placed at positions approxi-mately 1.5Å above eachoutermostOorCu atom,with their orientationparallel to the x-axis. The effect of H surface coverage on H adsorptionstructure onCuO (111) surfacewas examined by setting various surfacecoverage rates of H atoms onCuO(111) surface (0.0625–1ML) (Fig. S15).In the initial state, the H atomswere placed at positions approximately1.5 Å above each outermost O or Cu atom. The slab model and calcu-lation conditions for these surface coverage effects were the same asthose for the single H2 molecule on the CuO(111) surface.The hydrogen atom solid solubility limit in CuO was evaluated bycalculating the solution energies as follows, when inserting H atomssequentially up to 9 atoms into interstitial octahedral sites in a larger slabmodel comprising 256 atoms (Fig. S17). Structural optimization of theslab model was first performed with all Cu and O atoms movable. Sub-sequently, the total energies of each state with solid solution H atomswere calculated by fixing the Cu and O atoms in the optimized model,and only the H atoms were movable. The Brillouin zone was sampledusing a 1×2×3 k-point mesh, with cut-off energies of 450eV for the wavefunction and 1800eV for the charge density. The state determinationwasperformed using a convergence criterion of 1 × 10–5 eV for total energyand a force criterionof 0.01 eV/Å for force action. The solution energy forsolid solution of H atoms into CuO (4Gsol) was calculated as follows:4Gsol = ðEðCuO+nHÞ +ZPEðCuO+nHÞÞ � ðECuO +ZPECuOÞ� n2ðEH2+ ZPEH2+μH2ðT ,pÞÞð6Þwhere n is the numbers of H atom (n = 1 ~ 9), EðCuO+nHÞ and ZPEðCuO+nHÞare the total energy andZPE ofCuOwith solubilizedH atoms, ECuO andZPECuO are the total energy andZPE of CuO, EH2andZPEH2are the totalenergy and ZPE of free H2 molecule, and μH2ðT ,pÞ is the chemicalpotential of gaseous H2 contributed by temperature (T) and pressure(p) as follows.μH2T ,pð Þ=μH2ðT ,p�Þ+ kBT lnðpp�Þ ð7ÞThe H concentration ðcHÞ was estimated as follows:cH � expð�4GsolkBTÞ ð8Þwhere kB is Boltzmann’s constant (1.380649×10–23 J/K), T is absolutetemperature (K), p° is the reference pressure, and p is the actualpressure. Moreover, μH2ðT ,p�Þ is the chemical potential of H2 at thereferencepressure (1 atm),whichwasobtained from theNIST standardreference database (https://doi.org/10.18434/T4D303).Data availabilityAll relevant data are included in the article and its supplementaryinformation files. Source data are provided with this paper. Furtherdata are available from the corresponding author uponrequest. Source data are provided with this paper.References1. Krasae-in, S., Stang, J. H. & Neksa, P. Development of large-scalehydrogen liquefaction processes from 1898 to 2009. Int. J. Hydrog.Energ. 35, 4524–4533 (2010).2. Al Ghafri, S. Z. S. et al. Hydrogen liquefaction: a review of the fun-damental physics, engineering practice and future opportunities.Energ. Environ. Sci. 15, 2690–2731 (2022).3. Aziz, M. Liquid hydrogen: a review on liquefaction, storage, trans-portation, and safety. Energies 14, ARTN 5917 https://doi.org/10.3390/en14185917 (2021).4. Franco, V. et al. Magnetocaloric effect: from materials research torefrigeration devices. Prog. Mater. Sci. 93, 112–232 (2018).5. Numazawa, T., Kamiya, K., Utaki, T. & Matsumoto, K. Magneticrefrigerator for hydrogen liquefaction. Cryogenics 62,185–192 (2014).6. Kamiya, K. et al. Active magnetic regenerative refrigeration usingsuperconducting solenoid for hydrogen liquefaction. Appl. Phys.Express 15, 053001 (2022).7. Zhang, L., Sherif, S. A., Veziroglu, T. N. & Sheffield, J. W. On exergylosses in AMR hydrogen liquefiers. Int J. Hydrog. Energ. 19,447–452 (1994).8. Cwik, J. et al. Magnetocaloric effect in Laves-phase rare-earthcompounds with the second-order magnetic phase transition:Estimation of the high-field properties. Acta Mater. 133,230–239 (2017).9. Wada, H., Tomekawa, S. & Shiga, M. Magnetocaloric properties of afirst-order magnetic transition system ErCo2. Cryogenics 39,915–919 (1999).10. Imai, H., Wada, H. & Shiga, M. Calorimetric Study on Magnetism ofErCo2. J. Magn. Magn. Mater. 140, 835–836 (1995).11. Wada, H., Tanabe, Y., Shiga, M., Sugawara, H. & Sato, H. Magneto-caloric effects of Laves phase Er(CoNi) compounds. J. Alloy Compd.316, 245–249 (2001).12. Llamazares, J. L. S., Ibarra-Gaytán, P., Sánchez-Valdés, C. F., Ríos-Jara, D. & Alvarez-Alonso, P. Magnetocaloric effect in ErNi melt-spun ribbons. J. Rare Earth 38, 612–616 (2020).13. Cwik, J. et al. Magnetic and magnetocaloric properties of DyErNisolid solutions and their promise for hydrogen liquefaction. J.Mater. Chem. C. 12, 14421–14432 (2024).14. de Castro, P. B. et al. Machine-learning-guided discovery of thegigantic magnetocaloric effect in HoB near the hydrogen lique-faction temperature. Npg Asia Mater. 12, 35 (2020).15. Campoy, J. C. P., Plaza, E. J. R., Coelho, A. A. & Gama, S. Magne-toresistivity as a probe to the field-induced change of magneticentropy in Al compounds (=Pr,Nd,Tb,Dy,Ho,Er). Phys. Rev. B 74,ARTN 134410 https://doi.org/10.1103/PhysrevB.74.134410(2006).16. Zhang, Y. et al. Excellent magnetocaloric properties in RE(2)Cu(2)Cd (RE = Dy and Tm) compounds and its composite materials. Sci.Rep. 6, 34192 (2016).17. Levinsky, J. J. B. et al. Giant magnetocaloric effect in a rare-earth-free layeredcoordinationpolymer at liquidhydrogen temperatures.Nat. Commun. 15, https://doi.org/10.1038/s41467-024-52837-x(2024).18. Terada, N. & Mamiya, H. High-efficiency magnetic refrigerationusing holmium. Nat. Commun. 12, 1212 (2021).19. Cwik, J. et al. Magnetocaloric performance of the three-componentHo(1-x)Er(x)Ni(2) (x = 0.25, 0.5, 0.75) Laves phases as compositerefrigerants. Sci. Rep. 12, 12332 (2022).20. Provenzano, V., Shapiro, A. J. & Shull, R. D. Reduction of hysteresislosses in the magnetic refrigerant Gd5Ge2Si2 by the addition ofiron. Nature 429, 853–857 (2004).21. Cwik, J. et al. Layered composite magnetic refrigerants for hydro-gen liquefaction. Int. J. Hydrog. Energ. 87, 485–494 (2024).22. Tang, X. et al. Magnetic refrigeration material operating at a fulltemperature range required for hydrogen liquefaction. Nat. Com-mun. 13, 1817 (2022).23. Kitazawa, H. & Takeya, H. Magnetic materials and processingtechniques: requirements for an efficient liquefaction system. inNIMS NOW International. Vol. 17, 11–13 (National Institute for Mate-rials Science (NIMS), 2019).Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 10https://doi.org/10.18434/T4D303https://doi.org/10.3390/en14185917https://doi.org/10.3390/en14185917https://doi.org/10.1103/PhysrevB.74.134410https://doi.org/10.1038/s41467-024-52837-xwww.nature.com/naturecommunications24. Kim, Y. G., Chung, U. I. & Lee, J. Y. Features of hydrogen-inducedamorphization of the Er(Co, Mn)2 pseudo-binary laves compound.Acta Met. Mater. 40, 1497–1503 (1992).25. Gualtieri, D. M. & Wallace, W. E. Hydrogen capacity and crystal-lography of Erfe2-based and Erco2-based ternary-systems. J. LessCommon Met 55, 53–59 (1977).26. Mushnikov, N. V., Gaviko, V. S. & Goto, T. Magnetic properties ofhydrogen-arnorphized RCoH (R = Gd, Tb, Dy, Ho, Er, Tm and Y)alloys. J. Alloy Compd. 398, 36–41 (2005).27. Nemanic, V. Hydrogen permeation barriers: basic requirements,materials selection, deposition methods, and quality evaluation.Nucl. Mater. Energy 19, 451–457 (2019).28. Yamada-Takamura, Y., Koch, F., Maier, H. & Bolt, H. Hydrogen per-meation barrier performance characterization of vapor depositedamorphous aluminum oxide films using coloration of tungstenoxide. Surf. Coat. Tech. 153, 114–118 (2002).29. Engels, J., Houben, A., Rasinski, M. & Linsmeier, C. Hydrogensaturation and permeation barrier performance of yttrium oxidecoatings. Fusion Eng. Des. 124, 1140–1143 (2017).30. Terai, T. Research and development on ceramic coatings for fusionreactor liquid blankets. J. Nucl. Mater. 248, 153–158 (1997).31. Smith, D. L. et al. Progress in coating development for fusion sys-tems. Fusion Eng. Des. 61-62, 629–641 (2002).32. Lee, Y. et al. Hydrogen barrier performance of sputtered La2O3films for InGaZnO thin-film transistor. J. Mater. Sci. 54,11145–11156 (2019).33. Antony, L. V. M. & Reddy, R. G. Processes for production of high-purity metal powders. J. Min. Met Mat. S 55, 14–18 (2003).34. Yamamoto, T. D. et al. Gas-atomized particles of giant magneto-caloric compound HoB2 for magnetic hydrogen liquefiers. Appl.Phys. A Mater. 127, 301 (2021).35. Yamamoto, T. D. et al. Magnetocaloric particles of the Laves phasecompound HoAl2 prepared by electrode induction melting gasatomization. J. Magn. Magn. Mater. 547, 168906 (2022).36. Yamamoto, T. D. et al. Tunable magnetic and magnetocaloricproperties by thermal annealing in ErCo2 atomizedparticles. J. AlloyCompd. 935, 168040 (2023).37. Liang, G., Huot, J. & Schulz, R. Hydrogen storage properties of themechanically alloyed LaNi5-based materials. J. Alloy Compd. 320,133–139 (2001).38. Schlapbach, L. & Züttel, A. Hydrogen-storage materials for mobileapplications. Nature 414, 353–358 (2001).39. Kim, J. Y., Rodriguez, J. A., Hanson, J. C., Frenkel, A. I. & Lee, P. L.Reduction of CuO and Cu2O with H-2: H embedding and kineticeffects in the formation of suboxides. J. Am. Chem. Soc. 125,10684–10692 (2003).40. Rodriguez, J. A., Kim, J. Y., Hanson, J. C., Perez, M. & Frenkel, A. I.Reduction of CuO in H-2: in situ time-resolved XRD studies. Catal.Lett. 85, 247–254 (2003).41. Hollenberg, G. W., Simonen, E. P., Kalinin, G. & Terlain, A. Tritiumhydrogen barrier development. Fusion Eng. Des. 28,190–208 (1995).42. Caravella, A., Barbieri, G. & Drioli, E. Modelling and simulation ofhydrogen permeation through supported Pd-alloymembraneswitha multicomponent approach. Chem. Eng. Sci. 63,2149–2160 (2008).43. Ward, T. L. & Dao, T. Model of hydrogen permeation behavior inpalladium membranes. J. Membr. Sci. 153, 211–231 (1999).44. Bez, H. N., Yibole, H., Pathak, A., Mudryk, Y. & Pecharsky, V. K. Bestpractices in evaluation of the magnetocaloric effect from bulkmagnetization measurements. J. Magn. Magn. Mater. 458,301–309 (2018).45. Yamamoto, T. D. et al. Adiabatic temperature change in ErAl2/metalPIT wires: A practical method for estimating the magnetocaloricresponse of magnetocaloric composites. J. Magn. Magn. Mater.513, 167207 (2020).46. Lozada-Hidalgo, M. et al. Sieving hydrogen isotopes through two-dimensional crystals. Science 351, 68–70 (2016).47. McMahon, G., Miller, B. D. & Burke, M. G. High resolution NanoSIMSimaging of deuterium distributions in 316 stainless steel specimensafter fatigue testing in high pressure deuterium environment. NpjMat. Degrad. 2, 2 (2018).48. Zhu, Z. H., Shutthanandan, V. & Engelhard, M. An investigation ofhydrogen depth profiling using ToF-SIMS. Surf. Interface Anal. 44,232–237 (2012).49. Paudel, B. et al. ToF-SIMS in material research: a view from nanos-cale hydrogen detection. Mater. Today 75, 149–165 (2024).50. Yamasaki, T. et al. Multi-axis decomposition of density functionalprogram for strong scaling up to 82,944 nodes on the K computer:compactly folded 3D-FFT communicators in the 6D torus network.Comput. Phys. Commun. 244, 264–276 (2019).51. Perdew, J. P. & Wang, Y. Accurate and simple analytic representa-tion of the electron-gas correlation-energy. Phys. Rev. B 45,13244–13249 (1992).52. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradientapproximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).53. Perdew, J. P. et al. Restoring the density-gradient expansion forexchange in solids and surfaces. Phys. Rev. Lett. 100,136406 (2008).54. Gross, A. & Scheffler, M. Role of zero-point effects in catalyticreactions involving hydrogen. J. Vac. Sci. Technol. A 15,1624–1629 (1997).55. Henkelman, G., Uberuaga, B. P. & Jónsson, H. A climbing imagenudged elastic band method for finding saddle points and mini-mum energy paths. J. Chem. Phys. 113, 9901–9904 (2000).AcknowledgementsThe authors thank Yuki Nishimiya and Yoshihiro Nemoto from NIMS forpreparing TEM samples and helping with TEM observations. SXRD wasperformed at BL02B2 of SPring-8 (Proposal no. 2022A1067). The calcu-lations in this study were performed on the Numerical Materials Simu-lator at NIMS. H.Y. expresses gratitude for support from the MaterialScience Human Resource Development Fellowship of Tokyo Universityof Science. This work is supported by the JST-Mirai Program, Japan(Grant No. JPMJMI18A3), the JST-CREST Program, Japan (Grant No.JPMJCR22O3), and the ARIM of MEXT, Japan (JPMXP1223NM5030 andJPMXP1224NM5154).Author contributionsY.X. plannedandwrote themanuscript. Y.X. andK.O. performed theHPBanalyses. Y.X. and H.Y. contributed to the DFT calculations. H.T. pre-pared the ErCo2 samples. T.D.Y. and A.T.S. performed the magneticproperty measurements. H.A. performed the ToF-SIMS measurements.Y.M. and A.B. conducted the SXRD measurements. M.T. and Y.X. con-tributed to the SEM and TEM analyses. A.T.S., K.K., and Y.T. supervisedthe project. All authors discussed the results and commented on themanuscript.Competing interestsThe authors declare no competing interests.Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 11www.nature.com/naturecommunicationsAdditional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-026-71547-0.Correspondence and requests for materials should be addressed toYa Xu.Peer review information Nature Communications thanks the anon-ymous reviewers for their contribution to the peer review of this work. Apeer review file is available.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jur-isdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, aslong as you give appropriate credit to the original author(s) and thesource, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in thisarticle are included in the article's Creative Commons licence, unlessindicated otherwise in a credit line to the material. If material is notincluded in the article's Creative Commons licence and your intendeduse is not permitted by statutory regulation or exceeds the permitteduse, you will need to obtain permission directly from the copyrightholder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2026Article https://doi.org/10.1038/s41467-026-71547-0Nature Communications |         (2026) 17:4952 12https://doi.org/10.1038/s41467-026-71547-0http://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/www.nature.com/naturecommunications Enhancing hydrogen permeation barrier performance of ErCo2 magnetic refrigeration material via surface oxide layer formation Results and discussion HPB performance and magnetic properties Characterization of ErCo2 and surface oxide layer Mechanism of HPB performance improvement Methods Fabrication of ErCo2 spherical particles Oxidation treatments and electroless Cu plating Evaluation test of HPB performance Characterization DFT calculation Data availability References Acknowledgements Author contributions Competing interests Additional information