# Fileset

[Small - 2026 - Jansen - Thermal Instability Pathways of Aluminum‐Based Nanocomposites  The Role of Alloying‐Element‐Driven.pdf](https://mdr.nims.go.jp/filesets/414316c2-5c56-42f8-9a6f-bd03dc383fef/download)

## Creator

[Hendrik C. Jansen](https://orcid.org/0009-0004-3513-4348), Amit Sharma, [Marcus Hans](https://orcid.org/0000-0002-1814-3101), [Fedor F. Klimashin](https://orcid.org/0000-0002-2152-7165), [Léo Lapeyre](https://orcid.org/0000-0002-6982-3131), Dominik Gutnik, Jochen M. Schneider, [Thomas E.J. Edwards](https://orcid.org/0000-0002-3089-0062), [Jakob Schwiedrzik](https://orcid.org/0000-0003-1239-1961), [Barbara Putz](https://orcid.org/0000-0001-9687-6346), Johann Michler

## Rights

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

## Other metadata

[Thermal Instability Pathways of Aluminum‐Based Nanocomposites: The Role of Alloying‐Element‐Driven Precipitation and Amorphous Interlayers](https://mdr.nims.go.jp/datasets/fb557a7b-a62f-4e0c-9d9e-5cafa03b8f18)

## Fulltext

Smallwww.small-journal.comRESEARCH ARTICLEThermal Instability Pathways of Aluminum-BasedNanocomposites: The Role of Alloying-Element-DrivenPrecipitation and Amorphous InterlayersHendrik C. Jansen1,2 Amit Sharma1 Marcus Hans3 Fedor F. Klimashin1 Léo Lapeyre1Dominik Gutnik4 Jochen M. Schneider3 Thomas E.J. Edwards1,5,6 Jakob Schwiedrzik1,7 Barbara Putz1,4Johann Michler1,21Laboratory For Mechanics of Materials and Nanostructures, Empa Swiss Federal Laboratories for Materials Science and Technology, Thun, Switzerland2École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland 3Materials Chemistry, RWTH Aachen University, Aachen, Germany4Department of Materials Science, University of Leoben, Leoben, Austria 5Research Center for Structural Materials, National Institute for Materials Science,Tsukuba, Japan 6Department of Materials Science, Institute of Pure and Applied Sciences, University of Tsukuba, Tsukuba, Ibaraki, Japan 7Laboratory ForHigh Performance Ceramics, Empa Swiss Federal Laboratories for Material Science and Technology, Dübendorf, SwitzerlandCorrespondence:Hendrik C. Jansen (hendrik.jansen@empa.ch)Received: 23 April 2026 Revised: 11 August 2026 Accepted: 13 August 2026Keywords: hybrid PVD / ALD | nanocrystalline Al | nanoindentation | phase transformation | thermal stabilityABSTRACTHybrid physical vapour deposition (PVD) combined with atomic layer deposition (ALD) enables the fabrication of metal/ceramicAlNi/AlO(H) nanolaminates with high room-temperature strength, controlled architecture, and promising functional perfor-mance, yet their thermal stability remains unresolved. Here, we establish design principles for thermally robust Al-basednanocomposites by systematically assessing their thermal instability through correlated synchrotron X-ray diffraction, high-resolution microscopy, atom probe tomography, and nanoindentation following vacuum annealing between 160◦C and 600◦C.Angle-resolved X-ray diffraction (XRD) reveals Al crystallite evolution while high-resolution microscopy unravels furtherinstability mechanisms: At homologous temperature Thom ≈ 0.52 (∼210◦C), Ni segregation to Al grain boundaries competes withAl3Ni formation. Hydroxylated ALD-AlOxHy interlayers become mobile, driving swelling and oxidation of adjacent Al. At Thom ≈0.77 (∼450◦C), κ-Al2O3 crystallises, followed by interlayer rupture, while at Thom ≈ 0.94 (∼600◦C) the nanolaminate architecturecollapses into a coarsened Al(Ni) matrix with dispersed Al2O3 inclusions. These coupled transformations progressively reducehardness, transitioning from confined layer slip to matrix–inclusion-dominated deformation. Beyond elucidating degradationmechanisms, this work shows how hydrogen-mediated interface dynamics and the dual role of alloyed-Ni limit the thermaland mechanical stability, precisely identifying the onset temperatures, and providing a pathway to robust Al-based PVD/ALDnanocomposites.1NbcTo©ShIntroductionanocrystalline Al offers exceptional strength and low density,ut suffers from limited thermal stability (∼200◦C) as rapid grainoarsening at elevated temperatures leads to a loss of strengthhis is an open access article under the terms of the Creative Commons Attribution Licenriginal work is properly cited.2026 The Author(s). Small published by Wiley-VCH GmbHmall, 2026; 0:e75393ttps://doi.org/10.1002/smll.75393[1]. The driving force for grain growth in Al is particularly highdue to its low melting point and high diffusivity [2], whichrenders its nanostructure thermodynamically unstable even atmoderate homologous temperatures (Thom = T/Tm; where T andTm are the operating and melting temperature and both in unitse, which permits use, distribution and reproduction in any medium, provided the1 of 20http://www.small-journal.comhttps://doi.org/10.1002/smll.75393https://orcid.org/0009-0004-3513-4348https://orcid.org/0000-0002-1814-3101https://orcid.org/0000-0002-2152-7165https://orcid.org/0000-0002-6982-3131https://orcid.org/0000-0002-3089-0062https://orcid.org/0000-0003-1239-1961https://orcid.org/0000-0001-9687-6346mailto:hendrik.jansen@empa.chhttp://creativecommons.org/licenses/by/4.0/https://doi.org/10.1002/smll.75393http://crossmark.crossref.org/dialog/?doi=10.1002%2Fsmll.75393&domain=pdf&date_stamp=2026-08-22odtpaarA=miswvAg1tastswfo0tfbseait1tat0AnaAarAabamwasonItct[T2 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativf Kelvin). In recent years, considerable effort has therefore beenevoted to enhancing the thermal stability of nanocrystalline Alhrough segregation engineering, either by forming nanoscalerecipitates that provide Zener pinning, or introducing solutetoms that segregate to grain boundaries to reduce their energynd mobility [3]. Kinetic stabilisation in nanocrystalline Al waseported by Huang, Kalidindi, and Schuh [4] for electrodepositedl-6.5 at.% Mn, maintaining sub-100 nm grain size after Thom0.61 (∼300◦C) annealing in inert atmosphere for up to 1onth through the formation of Al6Mn nanoprecipitates. Sim-larly, nanoscale Al3Mg2 precipitates were found to kineticallytabilise magnetron-sputtered Al-5 and Al-10 at.% Mg films,hichmaintained sub-10 nmgrain size afterThom = 0.61 (∼300◦C)acuum-annealing for 3 h [5]. In a ball milled Al90Zr10 alloy, L12-l3Zr nanoprecipitates were found to kinetically stabilise Al grainrowth to only 58 nm after annealing at Thom = 0.88 (∼550◦C) forh [6]. Gianola et al. [7] reported that grain boundary stabilisationoward higher yield strength in the sub-100 nm range canlready be achieved through∼10 at.% O impurities inmagnetron-puttered nanocrystalline Al upon base pressure variations (10−7o 10−5 mbar). In another study, intentional co-alloying in theputtering process gave rise to a binary 4-nm columnar Al94.5Fe5.5ith thermal stability up to Thom = 0.51 (∼200◦C) with 1.3 GPalow stress [8]. In summary, kinetic pinning by precipitates arisednly at low to moderate homologous temperatures of Thom =.61 (∼300◦C) and was limited by rapid coarsening at higheremperatures, leaving a narrow and fragile window of bene-it. Enhanced thermodynamic stabilisation of grain boundariesy complexions has however been reported through chemicalegregation of solutes that effectively lower the grain boundarynergy and reduce its driving force for grain coalescence uponnnealing [9, 10]. Yet, elements that refine Al grains may stillnduce deleterious segregation in the matrix and fail to enhancehermal resilience [11, 12]. Gianola and colleagues [11, 13, 14–7] pioneered studies on the thermodynamic stabilisation ofernary nanocrystalline Al-Mg-Y, Al-Fe-Y, Al-Ni-Y, and Al-Ni-Celloys. Recently, a magnetron-sputtered Al94.9Ni3.8Ce1.3 was ableo maintain sub-10 nm Al grain size upon annealing at Thom =.64 (∼325◦C) for 1 h, while only above Thom = 0.70 (∼380◦C)l grain coarsening did occur [16]. Notably, the stabilisation ofanocrystalline Al was correlated to the formation of NiCe-richmorphous intergranular films, while intermetallic Al3Ni andl11Ce3 phases precipitated above Thom = 0.70 (∼380◦C). Thedditions of Ni and Ce were therefore able to prevent previouslyeported softening at Thom = 0.42 (∼120◦C) for nanocrystallinel95.5Ni4.5 [18] and recrystallisation at Thom = 0.53 (∼220◦C) formorphous Al-Ni [19]. Notably, magnetron sputtering imposeseneficial ultrahigh cooling rates up to 1015 K s−1 [20], whichllow a high degree of disorder similar to the design of thin filmetallic glasses [21, 22]. However, grain boundary stabilisationas also observed for ball-milled Al95.7Mg2.1Y2.2, Al95.7Fe2.2Y2.1,nd Al95.7Ni2.2Y2.1 that were able to maintain sub-50 nm Al grainize up to Thom = 0.87 (∼540◦C) for 1 h through a combinationf amorphous Ni-Y complexion and amorphous Ni-Y core-shellanorod formation at grain boundaries [11].n contrast to such chemical modification, structural leveragehrough systematic control over layer thickness and architecturean enable improvements in both strength [23] and eventually,hermal stability [24]. While nanolaminating Al with e.g., AlN25], SiC [26], crystalline Al2O3 [27], and amorphous Al2O3of 20[28–30] has been widely studied to enhance mechanical proper-ties, the literature lacks studies on similar effects for enhancedthermal stability. Notably, enhanced thermal stability throughnanolaminating was not yet reported for Al and mainly investi-gated in crystalline / crystalline Cu / Nb, where the combinationof face centered cubic (FCC) Cu and body centered cubic (BCC)Nb enhanced the poor thermal stability of the FCC Cu through3D interfaces up to 300◦C for 1 h [31]. The combination ofcrystalline metallic and amorphous ceramic layers provides fur-ther opportunities, since e.g., amorphous Al2O3 exhibits higherthermal stability than FCC Al and does not crystallise below1000◦C [32]. Accordingly, the out-of-plane (OP) grain growth ofthe nanocrystalline layer is impeded, up to the stability limit ofthe amorphous interlayer.We have recently applied the interface-engineered design strategy of hybrid physical vapour deposition(PVD) and atomic layer deposition (ALD) to Al100-zNiz / AlOxHy(25 / 1 nm) [33], exploiting OP control by AlOxHy interlayers andin-plane (IP) control by chemical complexity through Al100-zNiz.In situ transmission electron microscopy (TEM) has alreadyunravelled the high thermal stability of thin 4-nm ALD-Al2O3layers up to 900◦C [34], which is significantly higher thanthe eutectoid temperature of Al-Ni around 640◦C [35]. Here,ultimately, we combine both chemical and structural leveragetoward Al100-zNiz / AlOxHy (25 / 1 nm) thin films in a novel wayand approach their thermomechanical stability.2 Results and Discussion2.1 Elucidating Thermal StabilityCrystallite size evolution of (111) FCC Al after vacuum anneal-ing serves as an initial indicator of thermal instability in thenanolaminates, providing a basis for broader analyses that willlater include phase separation and decomposition effects. Theresults from synchrotron XRD are displayed in Figure 1. WhileFigure 1a shows the derived OP-Al crystallite size (ψ = 0◦)as a function of Thom, Figure 1b displays IP-Al crystallite size(ψ = 70◦) approximations. In comparison to well-establishedcrystallite size determination through dark field (DF)-TEM, IP-Al crystallite size derived from synchrotron XRD at ψ = 70◦and DF-TEM crystallite size correlate with R2 = 0.89 – justifyingthe fast screening technique especially for Thom < 0.94 here.Representative DF-TEM images and grain size correlation (DF-EM vs. Scherrer-XRD) can be found in Section S1. Error bars inFigure 1 are derived from the quadrature rule by Pearson VII fitquality, instrument broadening corrected by a LaB6 standard atevery ψ, and three different stage rotations ϕ.The OP-Al crystallite size in Al – AlOxHy thin films stays constantaround 25 ± 2 nm until Thom = 0.77. Ni-doped films show almostconstant OP-Al crystallite size only until Thom = 0.52, with 19 ±2 and 20 ± 2 nm for 2 and 5 at.% doping, respectively. At Thom =0.77, the OP-Al crystallite size of Ni-doped films exceeds the purereference OP crystallite size with 30 ± 8 and 34 ± 8 nm for 2 and5 at.% Ni, respectively. These values represent a relative OP graingrowth of∼20%, 76%, and 84% for 0, 2, and 5 at.% doped films fromas-deposited toThom = 0.77. AtThom = 0.94, anOP-Al grain growthuntil 61 ± 13 and 48 ± 12 nm can be detected for 0 and 2 at.% Nicontaining films. Yet, Al95Ni5 – AlOxHy OP-Al crystallite size atThom = 0.94 is only slightly increased to 34 ± 9 nm.Small, 2026e Commons LicenseFIGURE 1 Evolution of Al microstructure over ex situ vacuum annealing for 1 h in Al100-zNiz–AlOxHy thin films: (a) Out-of-plane crystallite and(b) in-plane Al crystallite size as a function of homologous temperature. Note that the ordinate scales differ.Ttsss0rFsNt2sc∼tbtYnasmbOcsldlaivatcS 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creatihe IP-Al crystallite size in Figure 1b evolves differently overemperature and chemistry. UntilThom = 0.46, the IP-Al crystalliteize for all three samples stays constant with an IP-Al crystalliteize refinement upon Ni addition. Notably, the IP-Al crystalliteizes until Thom = 0.46 are 31 ± 2, 14.5 ± 1, and 10 ± 1 nm for, 2, and 5 at.% Ni, respectively. This agrees with our previouslyeported IP-Al crystallite size refinement by Ni addition [33].urther, until Thom = 0.77, the IP-Al crystallite size increasestepwise to 50 ± 8, 57 ± 15, and 50 ± 13 nm for 0, 2, and 5 at.%i. Similarly to the jump in OP-Al crystallite size at Thom = 0.94,he IP-Al crystallite size reaches 123 ± 26, 144 ± 36, and 83 ±2 nm for 0, 2, and 5 at.% Ni, respectively. Here, the IP crystalliteize significantly exceeds the OP crystallite size for all threehemistries and OP/IP crystallite size ratios of ∼2.0, ∼3.0, and2.4 are determined for 0, 2, and 5 at.% Ni. This is a first indicatorhat the AlOxHy interlayers serve as dominating coalescencearriers rather than any structural or chemical modification dueo Ni addition in the crystalline Al layers.et, several aspects require critical evaluation to assess the robust-ess of the data. In viewof the commonly reported∼100nmupperpplicability threshold of the Scherrer approach, the crystalliteizes obtained in this work fall securely within the reliableeasurement range of the method for all OP crystallite sizes,ut only until Thom = 0.77 (∼450◦C) for the IP determination.nly the undoped Al – AlOxHy shows almost constant OP-Alrystallite size in the lower Thom range, whereas OP-Al crystalliteizes in Ni-doped films at lower Thom are below 25 nm metalayer thickness. In the Scherrer formula, a lower-bound coherentomain size underestimation originates from the contribution ofattice distortions to the full width at half maximum (FWHM)nd/or peak position. However, as the Scherrer formula does notnclude microstrain, it is commonly seen as a lower boundaryalue. There is no peak shift of the (111) Al visible for samplesnnealed until Thom = 0.46, but only peak broadening. Due tohe constant layer thickness, the observed IP peak broadeningannot stem from grain refinement. Thus, the underestimationmall, 2026of OP crystallite size at lower Thom might be due to the presenceof lattice strain from the Al-Ni solid solution and formation of Ni-rich nanoclusters (discussed in Section 2.3). XRD was performedin the vicinity of both (111) and (200) FCC Al peaks, but thestrong (111) texture of the architecture here does not allow acomparable analysis of the (200) FCC Al diffraction. Hence, theWilliamson and Hall [36] method is not applicable to determinethe actual microstrain values. Similar peak broadening wasreported for ball-milledMg-3 at.% Zn alloys due the formation of asolid solution [37] and γ’ nanoprecipitates broadening diffractionpeaks of the γ matrix in Ni-based superalloys [38], which is inagreement with the formation of Ni-rich nanoclusters in the Ni-containing as-deposited films at Thom < 0.52 here. In summary,and although the classical Scherrer approach here neglectsmicrostrain broadening, the good quantitative agreement (R2 =0.89) obtained in the present fits with DF-TEM demonstratesthat a reliable size analysis can still be achieved even withoutexplicit strain determination. Activation energy analysis of graingrowth in Al100−zNiz / AlOxHy was performed and correlated withliterature reports onnanocrystallineAl; the findings are discussedin Section S2.Huang, Kalidindi, and Schuh [4] reported enhanced thermalstability of electrodeposited nanocrystalline Al93.5Mn6.5 with as-fabricated Al grain size of 66 ± 16 nm and Al grain growth toward100 ± 40 nm for annealing at both Thom = 0.51 (∼200◦C) and Thom= 0.61 (∼300◦C) for 1 h. In comparison, Al95Ni5 / AlOxHy retainsa more refined IP-Al crystallite size of 29 ± 1 and 63 ± 8 nmfor vacuum annealing at Thom = 0.52 (∼210◦C) and Thom = 0.61(∼300◦C), respectively. However, magnetron-sputtered AlNiCewith Al crystallite size below 10 nm was found to not coarsenduring 1 h annealing up to Thom ∼0.64 (∼325◦C) for 1 h due tothe formation ofNi-Ce-rich amorphous interphases atAl-Al grainboundaries [16]. The addition of low amounts of Ce (2 at.%) tolow Ni amounts (2 at.%) resulted in an amorphous intermixing ofNi and Ce at the grain boundaries, which is thermodynamicallymore favourable than matrix segregation.3 of 20ve Commons LicenseFIGURE 2 Diffraction data from lab-scale: Evolution of phaseformation in Al100-zNiz / AlOxHy thin films probed by lab-scale XRD inBragg-Brentano geometry of (a) Al / AlOxHy, (b) Al98Ni2 / AlOxHy, and(c) Al95Ni5 / AlOxHy.IadIao2t0lNtiaoiκaopu[rN4 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativn addition to the Al crystallite size, microstructure evolutionnd phase formation was probed by lab-scale XRD. Respectiveiffractograms (2θ range from 20 to 50◦) are displayed in Figure 2.n the reference Al – AlOxHy thin film, the Al layers have a (111)nd (200) texture, whereas at higher Thom, there is an indicationf crystallisation of orthorhombic κ-Al2O3 with only one peak atθ = 34.7◦ (Figure 2a). With increasing annealing temperature,he FCC Al peaks slightly shift to higher 2θ angles until Thom =.77, whereas annealing at Thom = 0.94 induces a peak shift toower 2θ angles again. In XRD and upon thermal annealing, thei-doped films show three major differences: a strong (111) Alexture and a slight shift of Al reflections to smaller 2θ anglesn the as-deposited state; the formation of orthorhombic Al3Ni;s well as an earlier and more pronounced crystallisation ofrthorhombic κ-Al2O3. Films containing 2 at.% Ni, Figure 2c,ndicatemid-range order (“amorphous hump”) formation toward-Al2O3 and Al3Ni after 1 h of annealing at Thom = 0.77, whereasfter annealing at Thom = 0.94, respective crystalline phase peaksccur. The addition of 2 and 5 at.% Ni to the Al layers leads to ahase separation from FCC Al→ FCC Al + orthorhombic Al3Nipon annealing, which are the thermodynamically stable phases35]. The Al3Ni crystallisation onset temperature (from XRD) iseduced with increasing Ni content, from Thom ≤ 0.94 with 2 at.%i to Thom ≤ 0.61 with 5 at.% Ni.of 20The crystallisation of amorphous AlOxHy to metastable crys-talline κ-Al2O3 is seen as a destabilisation mechanism in thecrystalline / amorphous microstructure. Here, three factors seemto influence the crystallisation: ultra-thin 1 nm thickness, H-enrichment through room temperature ALD, and Ni-doping ofthe crystalline Al layers. Vogl et al. [34] observed the amorphous-crystalline phase transition of ALD-deposited amorphous 4 nmAl2O3 onCu by in situ TEMat∼1000◦C. TheALDdeposition tem-perature of 120◦C in their study versus room-temperature ALDhere is expected to reduce the H concentration from > 20 at.%(thusAlOxHy) to< 10 at.% and approachhigher density from< 2.4to > 3 g/cm3 [39, 40]. Commonly, the crystallisation temperatureof amorphous Al2O3 by ALD depends on the substrate anddecreases with increasing annealing time and film thickness,while no crystallisation is usually observed at T < 700◦C [32].This hints toward H-enrichment, rather than the 1 nm thicknessof the AlOxHy interlayer, as a limiting factor for the integrityof the nanolaminated crystalline / amorphous architectureshere, which will be discussed in chapter 2.4. Literature lacksstudies on the influence of Ni on the crystallisation sequence ofamorphous AlOxHy with or without H. Tkalec et al. [41] applieddifferential scanning calorimetry (DSC), XRD, and (scanning)transmission electron microscopy (S/TEM) to investigate the roleof Ni (0–3 at.%) replacing Al in amorphous diphasic Al2O3 –SiO2 precursors for sol-gel manufacturing. DSC data showed thataddition of 3 at.% Ni facilitated the crystallisation of γ-Al2O3 at874◦C, which was not visible for samples with 0, 1, and 2 at.%Ni. Their findings therefore clearly agree with an earlier onsettemperature for crystallisation of Ni-containingAl2O3. In anotherstudy, crystallisation of ∼3 Å amorphous Al2O3 thin films grownon (100) NiAl substrates was reported to be temperature andtime sensitive in the annealing temperature range of 450◦C to500◦C [42], significantly lower than common ALD-Al2O3 crys-tallisation temperatures [32]. A well-established phenomenonthat can markedly reduce the thermal stability of crystalline/ amorphous laminates is metal-induced crystallisation (MIC),where facilitated crystallisation of an adjacent amorphous layercan generally be linked to a low-energy interface with a highdensity of defects and high diffusivity in the crystalline part [43].As the Al100-zNiz / AlOxHy nanolaminates offer a large amount ofinterfaces, Ni-MIC needs to be considered. The low bilayer periodand crystalline Al-Ni adjacent to an ionic-covalently bondedamorphous AlOxHy layer may indeed lead to bond weakeningand higher crystallisation propensity, when the PVD layers areNi-doped.Thus far it appears that the thermal instability of Al100-zNiz /AlOxHy nanolaminates is controlled by the combined effects ofroom-temperature ALD (with its pronounced H incorporation)and the presence of Ni in the (crystalline) layers. The detrimentaleffect of Ni on the thermal stability is of particular interest and indirect contrast to beneficial Ni-induced strengthening effects inthe as-deposited state [33]. This complex behaviour is examinedand elaborated in greater detail in the following chapters.2.2 Phase-Interface Competition in Al-NiThe first limiting factor for thermal stability inAl100-zNiz / AlOxHythin films is the phase separation of FCC Al100-zNiz → FCC Al+ orthorhombic Al3Ni, competing with an interface segregationSmall, 2026e Commons LicenseFIGURE 3 S/TEM results on Al95Ni5 / AlOxHy after annealing at Thom = 0.52: (a) HAADF-STEM image of several bilayer periods with SAEDpattern, automated crystal orientation mapping (b) inverse pole figure out-of-plane map for FCC Al and (c) phase map of Al and Al3Ni; HR-STEMimages and selected area FFT patterns of phase separation to (d) FCC Al and (e) orthorhombic Al3Ni, (f) HR-STEM image highlighting grain boundarysegregation with (g) linescan; as well as results from STEM-EDS including (h) HAADF-yield, (i) Al-O-yield and out-of-plane linescan, (j) Ni-yield, and(k) in-plane line scan. Al appears in blue, Ni in green, and O in orange.ogcbmsappcpeFhtAfoeFwrTiS 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativf Ni solutes in Al. It emphasises the difference between trueeometric boundaries (interfaces) and local microstructure andhemistry variations (interphases). The two phenomena are visi-le in the S/TEM analysis of the evolution of the Al95Ni5 / AlOxHyicrostructure after vacuum annealing at Thom = 0.52 (∼210◦C),ummarised in Figure 3. Figure 3a shows a representative highngle annular dark field (HAADF)-STEM image of 24 bilayereriods as well as an selected area electron diffraction (SAED)attern covering approximately 20 bilayer periods. Automatedrystal orientation mapping was performed to resolve the inverseole figure out-of-plane map for FCC Al and a phase map consid-ring FCC Al and orthorhombic Al3Ni, visualised in Figure 3b,c.igure 3d and e further demonstrate the phase separation throughigh resolution (HR)-STEM images and selected area fast Fourierransformation (FFT) patterns of an FCC Al and orthorhombicl3Ni grain, respectively. Figure 3f shows an HR-STEM imageocusing on grain boundary segregation including a linescanver the interface, Figure 3g. For chemical mapping, STEM-nergy dispsersive X-ray spectroscopy (EDS) was performed andigure 3h–k shows respective HAADF-, AlO-, Ni-yield mapsith out-of-plane linescan, as well as an in-plane linescan,espectively.he HAADF-STEM overview image in Figure 3a confirms themprint of an intact 25 / 1 nm architecture of crystalline Al-Nimall, 2026layers and continuous AlOxHy interlayers where both separationand grain boundary segregation are resolvable. The crystallinelayers show a pronounced Z contrast, which can be attributedto intermetallic phase formation of Al3Ni (Figure 3e). The SAEDpattern in Figure 3a confirms the strong (111) FCC Al OPtexture, which is additionally confirmed by the automated crystalorientation mapping inverse pole figure out-of-plane map inFigure 3b.Moreover, single diffraction events from small volumescan also be derived for the Al3Ni phase as well as κ-Al2O3. IP-Alcrystallite size is in good agreement with synchrotron XRD anda representative DF-TEM image with objective aperture on the(111) and (200) FCC Al diffractions can be found in Section S3.Compared to the as-deposited condition, the IP FCC Al grainsgrow around 290%up to 29± 1 nm,with an average IP-Al3Ni grainsize of 18± 13 nm. Despite Al3Ni formation, there is an additionalhint toward interface segregation at Al–Al grain boundaries aswell as the crystalline / amorphous interfaces (Figure 3c,f). Asthe S/TEM specimens were exclusively prepared by Xe- plasmafocused ion beam (pFIB), these artefacts in Figure 3e shall beattributed to a higher Z contrast that does not originate fromGa-FIB preparation [44]. The Al-O map in Figure 3i confirmscontinuous AlOxHy interlayers with a more pronounced surfaceroughness, while chemical analysis by STEM-EDS of 15 inter-layers in the very same map reveals a chemical compositionof Al78.0±1.2Ni4.5±1.3O17.5±0.4. Thus, there is still Ni detected in the5 of 20e Commons LicensevNTciseAVtg=SisgiwossmAsist3AtpHalYfTgF=A0ficrAHT0aTaeewltf6 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativicinity of the interlayers. The Ni map in Figure 3j indicatesi-diffusion and agglomeration in the crystalline Al-Ni layers.hree different crystalline phases depending on the local Niontent are formed, which is confirmed by the in-plane linescann Figure 3k: almost Ni-deficient FCC grains, FCC Al-Ni solidolution, as well as orthorhombic Al3Ni grains. Their chemistries,xcluding remaining O, can be quantified as Al96.4±0.9Ni0.5±0.3,l90.7±2.1Ni5.5±1.2, and Al82.0±1.8Ni14.3±0.7, respectively.acuum annealing at Thom = 0.52 (∼210◦C) was chosen dueo previously reported Mg-enrichment of nanocrystalline Al–Alrain boundaries in Al100-zMgz (z≤ 7 at.%) when annealed at Thom0.51 (∼200◦C) for 1 h and subsequent strength increase [45].imilarly, Balbus et al. [16] reported amorphous Ni-Ce formationn a nanocrystalline AlNiCe alloy promoting enhanced thermaltability and mechanical properties. The authors proposed intra-ranular amorphous phase formation sufficiently hindering thentergranular formation of the intermetallic Al11Ce3 and Al3Ni,hile Al3Ni was not observed below 300◦C. In another studyn Al93.5Mn6.5, Mn segregation at Al–Al grain boundaries wasuggested as a reason for stability at Thom = 0.51 (∼200◦C), whiletability at Thom = 0.61 (∼300◦C) occurred despite dual-phaseicrostructure and Al6Mn precipitation [4]. Here, Ni diffusion inl95Ni5 / AlOxHy nanolaminates seems to indeed trigger interfaceegregation to both Al grain boundaries and the Al / AlOxHynterfaces (Figure 3d). Such interface complexions in dopedystems can appear in various ways: clean, monolayer, bilayer,rilayer, nanolayer, or wetted layer; ordered and disordered [10,0]. A potentially ordered single atomic layer Ni segregation onl–Al grain boundaries like the one in Figure 3d would needo follow epitaxial growth on the Al plane [46], which makeslane spacing measurements from the STEM image redundant.owever, the plot profile of the grain interior to grain boundaryrea in Figure 3g shows slightly increased intensity for the atomicayer adjacent to the Al grain boundary.et, interface segregation effects here are mostly prevented by theormation of the thermodynamically more stable Al3Ni phase.he formation of intermetallic Al3Ni might decelerate grainrowth and thereby affect the evolution of IP-Al crystallite size.igure 1b shows almost constant IP-Al crystallite size until Thom0.46 (∼160◦C) for Al95Ni5 / AlOxHy, yet a strong increase of IP-l crystallite size, when Ni-doped films were annealed at Thom =.52 (∼210◦C). Here, this strong relative change likely originatesrom the formation and / or growth of Al3Ni upon anneal-ng originating from Ni-rich nanoclusters in the as-depositedondition. Huang, Kalidindi, and Schuh [4] reported grain sizeetention below 100 nm partially due to phase formation ofl6Mn in an electrodeposited nanocrystalline Al93.5Mn6.5 alloy.ere, the Al3Ni formation can only be confirmed by SAED athom = 0.52 (∼210◦C) (Figure 3a), but not by XRD at Thom =.77 (∼450◦C) (Figure 2). Yet, the Al3Ni phase is anticipated asn Al grain stabiliser in a Zener pinning mechanism only untilhom = 0.46 (∼160◦C) for 1 h; however upon Al3Ni formationnd growth, the relative IP-Al grain growth in Ni-doped filmsxceeds the undoped reference film (Figure 1b). Notably, thenthalpy of Al3Ni formation is approximately −60 kJ mol−1 [35],hile the enthalpy of Ni segregation at Al–Al grain boundariesies in the range from 0 to −25 kJ mol−1 [47]. Therefore, underhese annealing profiles close to thermodynamic equilibrium, theormation of Al3Ni is expected to be more favourable than Niof 20segregation at Al–Al grain boundaries. However, it should beemphasised that Al grain size in our study at similar annealingtemperature and duration (∼200◦C for 1 h) lies below 30 nm,which is significantly lower than 100 ± 50 nm for the Al93.5Mn6.5alloy [4]. Recently, Cunningham et al. [48] postulated the effectof flash DSC with varying cyclic cooling rates up to 10.000◦C s−1successfully enabling amorphous Al-Ni phases at Al–Al grainboundaries in a sputter-deposited Al98Ni2 alloy. Therefore, theeffect of such metastable heating profiles on the phase formationin the Al100-zNiz / AlOxHy here remains an open question.2.3 Nanoscale ChemistryWe have previously reported the enrichment of S/TEM(-EDS)datasets through APT on the as-deposited films that elucidatednanoscale composition data not-resolvable by S/TEM [33]. Thescreening by XRD (Figure 2) has already marked the crystalli-sation of AlOxHy. Here, APT offers the unique ability to trackthe evolution of matrix and alloying elements, including O andeven H in 3D for correlation to the stability of the interlayers.Therefore, a cross-correlation of S/TEM and APT on the role ofAl, Ni, O, and H is conducted in the following. The Al95Ni5 /AlOxHy at Thom = 0.32, 0.52, and 0.77 (as-deposited with ∼23◦Cas well as annealed at ∼210◦C and ∼450◦C) were chosen as filmsof particular interest. Figure 4 shows reconstructions of Al, Ni, O,and H atomic positions of two bilayer periods in each conditionand respective OP composition profiles from cylinders over twoAl95Ni5 / AlOxHy interfaces. Isoconcentration surfaces of 10, 5,and 10 at.% for Ni, O, andH, respectively, were chosen to visualiselocal enrichment of the respective elements. It should be notedthat the H content can be understood as an upper boundary valuedue to the likely contribution of residual gas from the analysischamber. The reconstructions correspond to slices of 10 nm andenable illustration on the order of one single grain depth. Similarphenomena were observed for Al98Ni2 / AlOxHy at Thom = 0.52(∼210◦C) and corresponding data can be found in Section S4.Mass-spectra for Al95Ni5 / AlOxHy are displayed in Section S5.In the as-deposited condition, the crystalline Al-Ni layers consistof a Ni-deficient solid solution and randomly distributed Ni-richnanoclusters. The adjacent AlOxHy interlayers show coincidingO and H signals that build up quasi-continuous interlayers. Forfurther details readers are referred to our previous publication[33]. At Thom = 0.52 (∼210◦C), the crystalline layers show Ni-richvolumes, which are consistent with individual grains and seemto impinge and bow the AlOxHy interlayers. Contiguous AlOxHyinterlayers exhibit reduced continuity, a certain broadening aswell as differences in O- and H-enrichment. Additionally, thereis a certain decoupling observable between composition peaksof O and H, when the AlOxHy interlayers collide with the Niagglomeration zones, indicating reduced AlOxHy stability in thevicinity of Ni-enriched volumes. Notably, the 10 at.%Ni isosurfacefor Thom = 0.52 in Figure 4c shows one Ni-enriched interface(annotated with a black arrow), which is in good agreementwith the more pronounced Z contrast at a vertical Al–Al grainboundary visualised through HR-STEM in Figure 3d. The mea-surement cylinder reveals a quasi-similar chemical compositionof thatNi-enriched interface and theAl3Ni phases. Therefore, andin contrast to the analysis of as-deposited films, STEM-EDS andAPT show good agreement on the competing mechanism of NiSmall, 2026e Commons LicenseFIGURE 4 APT reconstructions of atomic positions (10 nm slices) including Al, Ni, O, and H as well as composition profiles for Al95Ni5 / AlOxHyat three selected annealing conditions: (a) Thom = 0.32 (as-deposited, adapted from [33]), (b) Thom = 0.52 (∼210◦C), and (d) Thom = 0.77 (∼450◦C); as wellas respective out-of-plane composition profiles from cylinders. Al appears in blue, Ni in green, O in orange, and H in gray. The black arrow marks mostlikely a Ni-rich grain boundary.Small, 2026 7 of 20 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons LicensesmgiNbritbngbUAcOTAsOstAaaThgsbmfAatsOecciTAaTdqasuNftw[md8 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creatiegregation and the phase separation to Al and Al3Ni upon ther-al annealing. Lei et al. [11] reported the formation of amorphousrain boundary complexions and amorphous core-shell nanorodsn ball-milled nanocrystalline Al-Mg-Y, Al-Fe-Y, and Al-Ni-Y.otably, the core–shell nanorods were found to nucleate at grainoundaries. Likewise, we are assuming Ni diffusion from Ni-ich nanoclusters toward Al-Al grain boundaries, which resultsn the observed enrichment. Figure 3f however, demonstrates thathe atomic arrangement of the Ni-segregation at the Al–Al grainoundary is not confirmed to have an amorphous nature andeither several nanometres of thickness. Then, with sufficient Niathered, theAl3Ni phase can precipitate and grow from the grainoundary into the adjacent matrix.pon further annealing at Thom = 0.77 (∼450◦C), both crystallinel100-zNiz and amorphous AlOxHy layers alter significantly. In therystalline layer, the Ni-deficient Al matrix is growing in bothP and IP directions, and a larger Ni-rich volume is visible.he Ni-enrichment appears to be still confined by neighbouringlOxHy, although it distorts the Al grain underneath, but no Ni-egregation at vertical Al-Al grain boundaries is visible anymore.ne important aspect of PVD / ALD nanolaminating is thetabilisation of PVD-derived microstructure toward enhancedhermal stability. Multilayered Al / Ni systems without e.g.,lOxHy interlayers are commonly subject to chemical mixingt the interfaces and formation of Al-Ni compounds (uponnnealing) such as the primary formation ofAl3Ni at∼270◦C [49].he authors emphasised the role of triple junctions serving aseterogeneous nucleation sites, where upon annealing, Al3Ni canrow and coalesce as a continuous layer to reduce the thermaltability of Al / Ni films. Such coalescent Al3Ni formation cane prevented here by ALD interlayers, which is a sign towardore effective chemical control by thermally exposed PVD / ALDilms.t Thom = 0.77 the AlOxHy interlayer broadening continues,nd interlayer breaking is visible with several tens of nanome-res spacing between AlOxHy segments. The remaining AlOxHyegments additionally show a more pronounced decoupling ofand H signal in a sort of core-shell structure. Despite anffective spatial resolution limit of 1 nm by APT [50], the relativehanges of AlOxHy annealing indicate a volume expansion inorrelationwith instability, whichwill be subject of the discussionn Section 2.4.he chemistries of Al grains can be derived to approximatelyl87.9±2.6Ni6.8±1.5O1.2±0.6H4.1±1.1, Al97.2±0.6Ni0.8±0.3O0.4±0.1H1.6±0.3,nd Al100 for Thom = 0.32, 0.52, and 0.77, respectively.herefore, Al grains continuously become Ni, O, and H-eficient upon annealing. The Ni-rich volumes can beuantified to Al77.1±1.3Ni20.9±1.2O0.2±0.2H1.8±0.5 at Thom = 0.52nd Al70.0±1.8Ni25.0±2.1O1.0±0.4H4.0±1.3 at Thom = 0.77. Thus, thetoichiometry of Al3Ni with O and H impurities is reachedpon annealing. However, upon annealing, the H content in thei-rich grains is continuously increasing, while it is decreasingor the Al grains. The cross-correlation of STEM and APT linkhe Ni-rich grains to intermetallic Al3Ni, which is in agreementith recent studies of intermetallic alloys as hydrogen getters51], although Al3Ni was not-yet-reported as a hydrogen-getteraterial and no hydride formation is observed here throughiffraction.of 20vIn contrast, the AlOxHy concentrations evolve in thefollowing way from Thom = 0.32 to 0.52 and 0.77:Al75.0±0.7Ni8.5±1.1O5.9±0.8H10.6±1.6, Al70.2±2.9Ni4.0±1.2O10.8±1.2H15.0±2.1,and Al62.1±3.9Ni4.1±1.2O21.3±2.6H12.5±1.7. Upon annealing, theinterlayers are continuously detected to be O-deficient,but progression toward the ideal Al2O3 chemistry withH-contamination can be observed.Beyond the enhanced structural integrity, Al / α-Al2O3 coatingswere recently proposed as promising candidates for H barriercoatings [52]. Therefore, the role of H for the stability of theAl100-zNiz / AlOxHy architectures is of particular interest. Here,the bulk compositions, considering the entire APT specimens(Figure 4), reveal H outdiffusion from the Al95Ni5 / AlOxHyspecimens: average (upper boundary) H concentrations decreasefrom 4.2 at.% to 2.7 at.% to 1.2 at.% for Thom = 0.32, 0.52, and0.77, respectively. Figure 4b,d, and f give an overview of theAl, Ni, O and H compositions over two interfaces in Al95Ni5 /AlOxHy derived from the APT reconstructions in Figure 4 andmeasurement cylinders with 10 nm diameter. The compositionprofiles reveal strong composition variations of both Al100-zNiz(Al + Al3Ni) and AlOxHy upon annealing. Here, focus is shedon the analysis of light elements O and H, which are decreasingin Al100-zNiz upon annealing. Upon partial crystallisation andbreaking atThom = 0.77 (∼450◦C), theAl100-zNiz /AlOxHy interfaceadditionally becomes sharper. Al / α-Al2O3 coatings by (reactive)magnetron sputtering were proposed as H permeation barriersat ambient temperature due to enhanced damage resistance andlow H diffusivity [52]. Complementary density-functional theory(DFT) and Grand Canonical Monte Carlo simulations unravelledthe root cause in superabundant (several orders of magnitudeincreased [53]) metal Al vacancy concentrations at crystalline /crystalline Al / α-Al2O3 interfaces that allow H trapping, whileH barrier performance scales with increasing amount of Al /Al2O3 interfaces. The roughly 114 Al100-zNiz / AlOxHy interfacesthrough the ∼ 3 µm 25 / 1 nm bilayer period might indeedinduce similar H permeation barrier capabilities, but the AlOxHyinterlayers are already H-enriched and DFT simulations predictslightly accelerated diffusivity of H in amorphous AlOxHy [54].H diffusivity in Al at room-temperature can be estimated to1.5×10−14 m2 s−1 [55], which is several magnitudes higher thanH diffusion in any AlOxHy [54] and explains the apparent Hdiffusion visualised through APT in Figure 4b for Al95Ni5 /AlOxHy annealed at Thom = 0.52 (∼210◦C). We are expecting herefurther accelerated room temperature diffusivity in the AlOxHylayers deposited by room-temperature ALD, due to a significantlylower density of < 2.4 g cm−3 compared to 3.2 g cm−3 forclose-to stoichiometric Al2O3 [40], which creates more diffusionpaths. Thus, amorphous ALD-AlOxHy is seen as less promisingcandidate for multi-layered H barrier coatings. The questionremains on the H barrier performance of post-annealed Al100-zNiz/ AlOxHy, which are subject to AlOxHy decomposition and will bemore discussed in Section 2.4.On the cross-correlation of XRD, S/TEM, and APT at Thom =0.52, there are several points to be remarked. Generally, bothS/TEMandAPT agree on the formation of Ni-rich volumes. Here,S/TEM and XRD confirmed Al3Ni crystallinity and APT clarifiedthe impinging character on the adjacent AlOxHy interlayers,which is especially visible at Thom = 0.52 in Figure 4b. S/TEMin Figure 3 indicates slightly enhanced roughness of the AlOxHySmall, 2026e Commons LicenseFIGURE 5 S/TEM results on Al95Ni5 / AlOxHy after annealing at Thom = 0.77 (∼450◦C): S/TEM overview (a) HAADF-yield with SAED pattern, (b)Al-O-Ni-yield, and (c) corresponding out-of-plane EDS line scan, as well as overview of interlayer destabilisation mechanisms (d) expansion, (e) partialcrystallisation toward κ-Al2O3, and (f) breaking points.iwtaweasA1uisHfwHSAasS 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Cnterlayers after annealing compared to the as-deposited state,here APT hints toward increased surface roughness. Notably,his is not seen as an indicator for breaking of the interlayerst Thom = 0.52 due to continuous O- and H enrichment, thoughith more pronounced chemical heterogeneity (less continuousxcess as revealed by isoconcentration surfaces) compared to thes-deposited condition. The results from XRD in Figure 2 havehown the formation of superlattice fringes around the (111) FCCl peak at 2θ of 38.6◦ after annealing at Thom ≥ 0.61 (300◦C) forh only in 5 at.%Ni doped films. Although superlattice fringes aresually seen as sign for excellent structural coherency and smoothnterfaces in multi-layered or epitaxial thin films [56, 57], theuperlattice fringes here only appear upon AlOxHy roughening.ence, the superlattice fringes are assumed to stem from theormation of a thin adjacent FCC Ni layer from Ni outdiffusion,hich is in good agreement with the higher Z contrast layers inAADF-STEM images in Figures 3 and 5 (presented in the nextection 2.4).fter elucidating the role of intermetallic Al3Ni phase formationnd roughening of the AlOxHy interlayers on the microstructuretability and IP-Al crystallite size, further assessment is needed tomall, 2026reatiunravel the effect and mechanism of AlOxHy interlayer breakingand its implications for OP-grain growth.2.4 Interlayer Breaking and DewettingSignificant AlOxHy interlayer breaking tendencies in Al95Ni5 /AlOxHy at Thom = 0.77 were resolved by APT in Figure 4c. Theprevious discussion has unraveled increased destabilisation ten-dencies through H-enrichment in the AlOxHy and Ni diffusion.Therefore, cross-sectional S/TEM analysis was performed forboth Al98Ni2 / AlOxHy and Al95Ni5 / AlOxHy after annealing atThom = 0.77 for 1 h. Figure 5 illustrates an overview of STEM(-EDS) results for Al95Ni5 / AlOxHy (Figure 5a–c) and respectiveinterlayer breaking mechanisms (Figure 5d–f). RepresentativeSTEM-EDS images for Al98Ni2 / AlOxHy after annealing at Thom= 0.77 for 1 h are displayed in Section S6 and emphasise morecontinuous and therefore stable AlOxHy layers.The HAADF-STEM image of several bilayer periodsAl95Ni5 / AlOxHy after vacuum annealing Thom = 0.77 (∼450◦C)for 1 h in Figure 5a confirms the ongoing phase separation9 of 20ve Commons LicensebacFrTAOcstiaNbpptsroddwpsFAAtctκIaA(ooac7sXsTaTHeIoaHibc(i1 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativetween crystalline FCC Al and orthorhombic Al3Ni as wells the in-plane grain growth of both phases. Contrary to threerystalline domains in the Al-Ni layers at Thom = 0.52 (∼210◦C,igure 3k), the OP EDS linescan in Figure 5c confirms twoemaining domains: almost pure Al grains and Al3Ni grains.heir chemistries are quantified as Al90.4±2.3Ni0.2±0.2O9.4±1.8 andl71.6±3.4Ni18.1±1.8O10.3±2.4, similar to pure Al and Al3Ni withimpurities. Additionally, SAED in Figure 5a confirms therystalline nature of κ-Al2O3. However, this crystallisation iseen here as partial crystallisation, since there is no evidencehat all AlOxHy layers have crystallised. The Z contrast amongnterfaces is more pronounced, and the interlayers are broadernd less continuous than at lower annealing temperatures.otably, the Al-O-Ni maps in Figure 5b annotate interlayerreaking due to O-yield discontinuity. Additionally, the out-of-lane EDS linescan over a couple of Al95Ni5 / AlOxHy bilayereriods in Figure 5c confirms the lack of O-enrichment andhus, interlayer discontinuity with the formation of interlayeregments. These discontinuities are in good agreement with APTesults in Figure 4, which provided evidence for destabilisationf Al95Ni5 / AlOxHy at Thom = 0.77 (450◦C). Themore pronouncedestabilisation of amorphous AlOxHy interlayers for samplesoped with 5 at.% Ni compared to 0 and 2 at.% Ni is consistentith the results from (synchrotron) XRD that showed moreronounced Al2O3 crystallinity compared to the undopedample.ollowing the results in Figure 5d–f, we can determine threelOxHy interlayer degradation mechanisms on the example ofl95Ni5 / AlOxHy at Thom = 0.77: volume expansion, crystallisa-ion, and breaking. Bright field (BF)-TEM images in Figure 5d,eonfirm AlOxHy volume expansion (manifested as a layer-hickening) up to 2.3 ± 0.5 nm and partial crystallisation toward-Al2O3 confirmed by selected area FFT in the interlayer vicinity.t is noted that the BF-TEM image in Figure 5d illustratesn example of extraordinary layer swelling up to ∼4.7 nm.dditionally, Figure 5f shows discontinuous AlOxHy interlayersindicated by yellow arrows) that give rise to the formationf Al–Al grain boundaries and facilitated grain growth in theut-of-plane direction. The here observed degradation of 1 nmmorphous AlOxHy layers, fabricated by ALD, disagrees withommonly observed crystallisation of ALD-AlOxHy only above50◦C, depending on annealing time, film thickness and sub-trate, respectively [32], as alreadymentioned in the discussion ofRD results in Section 2.1. Local crystallisation of thin amorphouspecimens can be a consequence of electron irradiation duringEM at 200 kV [34]. Yet, the results from (synchrotron) XRDllow a determination on κ-Al2O3 crystallisation temperature athom > 0.61 (∼300◦C).ence, the remaining question needs to focus on the sequence ofxpansion – crystallisation – breaking of the AlOxHy interlayers.t is emphasised that volume expansion (Figure 5d) and formationf crystallisation sites (Figure 5d) in AlOxHy upon annealingre mechanisms originating from room temperature ALD andenrichment [39, 58, 59]. Such solid AlOxHy complexes canndeed contain up to 65%–75% of water [60], which is confirmedy the low density of room-temperature AlOxHy < 2.4 g cm−3ompared to stoichiometric Al2O3 (3.2 g cm−3) [39] and water1.0 g cm−3). It is remarked that hydrogen-containing speciesncorporated during room-temperature ALD—whether asmolec-0 of 20ular H2O, hydroxyl groups, or other H-related defects—aredifficult to distinguish experimentally, especially in 3D and ex situ[61]. Additionally, while the incorporated hydrogen originatespartially from the H2O precursor during room-temperature ALD[39], its exact chemical state after deposition is uncertain here.Consequently, the following discussion addresses the collectivethermal mobilisation of hydrogen-containing species duringannealing and contribution to the observed microstructural evo-lution solely as hydrogen-mediated mechanisms of hydroxylatedspecies. Therefore, quartz crystal microbalance (QCM) measure-ments were conducted for ALD-AlOxHy at room temperature bytrimethylaluminium (TMA) and deionised (DI) scanning -H2O.Mass gains for three ALD cycles are displayed in Section S7. Theresults indicate a mass gain of 9.3± 0.2 ng cm−2 upon the pulse ofDI-H2O at room temperature, an indicator for the physisorptionsubsequent hydroxylation. Henceforth, hydrogen-mediated vol-ume expansion (swelling) upon annealing at low temperatures issuggested by the combined QCM, APT, and S/TEM observationsthat indicate the initiation of AlOxHy decomposition. Sato [62]accordingly reported the thermal decomposition temperature ofamorphous AlOxHy precipitates from aqueous solutions, morespecifically amorphous and pseudo-boehmite AlO(OH), by dif-ferential thermal analysis around 250–300◦C, followed by thesubsequent formation of amorphous Al2O3. Decomposition ofAlOxHy then gives rise to various possible scenarios, two ofthe main ones include the reaction of hydroxylated species withadjacent phases or evaporation from the amorphous layers.Notably, hydrolysis can be assumed due to quasi-linear AlOxHythickness increase from 1.1 ± 0.2 to 1.4 ± 0.3, and 2.3 ± 0.5 nmfor Al95Ni5 / AlOxHy annealed at Thom = 0.32, 0.52, and 0.77,respectively. However, the formation of bubbles on the surface ofthe films after annealing is observed, which is an indicator forevaporation of hydroxylated species and in good agreement withdecreasing O and H contents in APT specimens upon annealing.Representative images from scanning electronmicroscopy (SEM)investigation are displayed in Section S8. Then, crystallisationof AlOxHy is a compression-induced phase decomposition atheterogeneous nucleation sites from a less-dense to more-densestructure with an imminent molar volume change [39, 59], andthus needs to result in volume contraction. Ultimately, thiscrystallisation is estimated to result in stress release and interlayerbreaking [63], giving rise to out-of-plane grain growth as grainboundaries replace the crystalline / amorphous interfaces.Here, ex situ analysis of residual stress in the metallic compo-nents of the nanolaminates (after cooling and therefore stressrelaxations) was performed using the sin2ψ method. The errorbars in residual stress are derived from the R-squared value ofthe fitted linear trendline of the ln(1/sin(θ)) over sin2ψ dataset foreach condition.The results in Figure 6 shows increasing tensile residual stress(σres) with increasing vacuum annealing temperature for all threeAl100-zNiz / AlOxHy thin film architectures. Only the as-depositedAl95Ni5 / AlOxHy has a slightly compressive residual stress of σres= -82 ± 60 MPa. For the Al / AlOxHy films, the residual stress islinearly increasing fromThom = 0.32 to 0.94 and σres = 125± 22MPato 540 ± 2 MPa, respectively. For Ni-doped films, the residualstress is almost linearly increasing only until Thom = 0.77 withresidual stress values for Al98Ni2 / AlOxHy and Al95Ni5 / AlOxHyranging from σres = 56 ± 42 to 563 ± 6 MPa and σres = -82 ± 60Small, 2026e Commons LicenseFIGURE 6 Ex situ residual stress of the Al100-zNiz / AlOxHy thinfilms as a function of homologous temperature / vacuum annealingtemperature for 1 h derived from sin2ψ XRD.taf7Dcstuwr(K(tAada≤8≤rrsHtAispacS 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativo 532 ± 5 MPa, respectively. In contrast to pure Al layers, furthernnealing at Thom = 0.94 for 1 h induces residual stress reductionor 2 and 5 at.% Ni doped films toward σres = 416 ± 36 and 395 ±MPa, respectively.ue to differences in thermal expansion between individualonstituents of the Al100-zNiz / AlOxHy thin films and the Siubstrate, thermal stress cannot be neglected. The evolution ofhermal stress within the nanolaminates can be quantified bysing the following simplified relationship:𝜎𝑡ℎ𝑒𝑟𝑚𝑎𝑙 =𝐸𝑓𝑖𝑙𝑚1 − 𝜈𝑓𝑖𝑙𝑚(𝛼𝑠𝑖 − 𝛼𝑓𝑖𝑙𝑚)(𝑇 − 𝑇𝑑) (1)ith thermal stress σthermal, Young’s modulus Efilm, Poisson’satio νfilm, Si substrate thermal expansion coefficient (CTE) αSi2.55×10−6–4.10×10−5 K−1), film CTE αfilm (2.65×10−5–3.55×10−5−1), and temperature difference from deposition temperature Td23◦C) and vacuum annealing temperature T (160◦C to 600◦C)o room temperature. Due to the marginal volume fraction oflOxHy, Efilm, νfilm, and αfilm are assumed as EAl, νAl, and αAl fromn Al film. Henceforth derived thermal stress from the coolingown after annealing to 160◦C, 210◦C, 300◦C, 450◦C, and 600◦Cre all of compressive nature and calculated to 328 MPa< σthermal430 MPa, 448 MPa< σthermal ≤ 587 MPa, 663 MPa< σthermal ≤70 MPa, 1023 MPa< σthermal ≤ 1341 MPa, and 1382 MPa< σthermal1812 MPa, respectively. Calculated thermal stresses exceed theesidual stresses for every annealing condition. Notably, stresselaxation and recovery upon cooling will result in lower residualtress obtained from ex situ sin2ψ than calculated thermal stress.owever, the order of stress magnitude for both residual andhermal stress is expected to give rise to plastic deformation in thel100-zNiz layers. In fact, the phase-separated Al and Al3Ni grainsn this particular grain size regime are expected to possess yieldtress in the range of several hundred MPa [64]. The assumedlastic deformation in the crystalline Al and Al3Ni phases uponnnealing at temperatures higher than 450◦C will thereforeontribute to the limited thermal stability of the architecturesmall, 2026probed here. It is reasonable to expect that the tensile stresswill additionally facilitate diffusion and increase grain migrationtendencies [65], which is supported by the abnormal grain growthat Thom > 0.77 (see Figure 1).Ultimately, the observed crystallisation of 1 nm AlOxHy at Thom> 0.61 (∼300◦C) disagrees with crystallisation kinetics expectedfrom literature for ultra-thin ALD films. Conjunction of intrinsicstress, Ni and H impurities and the interlayer roughness of PVD-deposited Al100-zNiz grains allows stress localisation and breakingpoints. Upon crystallisation from 1 nm amorphous AlOxHy toκ-Al2O3, mechanical stresses in the range of several hundredMPa are expected from the CTEs of Al and Al2O3 mentionedabove.To probe harsher conditions and microstructure evolutionbeyond on-set destabilisation in the range of 300 to 450◦C,the samples were also annealed at Thom = 0.94 (∼600◦C). Theevolution of Al grain size in Figure 1 gave rise to the strongestgrain growth from Thom = 0.77 to 0.94 both in the OP- and IP-orientation. Due to the destabilising effects of Ni already visibleat Thom = 0.77, the reference Al / AlOxHy film was chosen to beinvestigated more closely by S/TEM. Figure 7 gives an overviewof the structure (Figure 7a–c,f) and chemistry (Figure 7d,e) of Al/ AlOxHy after vacuum annealing at Thom = 0.94 for 1 h.The BF-TEM overview image in Figure 7a shows discontinuousAlOxHy interlayers and strong Al grain growth in both out-of-plane and in-plane orientation. The corresponding SAED patternconfirms both FCC Al and orthorhombic κ-Al2O3 crystallinity.Again, the crystallisation cannot be confirmed for all the Al2O3inclusions and is therefore seen as partial crystallisation only.The low amount of Al diffraction events compared to lowerannealing temperatures confirms the presence of larger Al crystalsizes of 62 ± 30 nm (derived from crystal orientation mappingof inverse pole figure out-of-plane map in Figure 7f). Increasingthe magnification, HAADF-STEM in Figure 7b further confirmsthe formation of Al-Al grain boundaries between adjacent sub-layers, partially replacing the original AlOxHy interlayers. Thebreaking of AlOxHy interlayers gives rise to the encapsulationof mainly spherical AlOxHy fragments with an average diameterof 18 ± 7 nm. Note that this size was measured parallel tothe long axis for a minimum of 75 AlOxHy fragments. TheO-rich fragments are encapsulated within the surrounding Algrains, while some are also located at Al-Al grain boundaries.Chemically, O-rich particles and Al-rich matrix are quantified bySTEM-EDS to Al71.5±4.2O28.5±6.0 and Al93.8±1.8O6.2±1.3, respectively.The automated crystal orientation mapping in Figure 7f confirmsthe broader grain size distribution and equiaxed Al matrix grainstructure.Literature proposes mainly three mechanisms for the thermalinstability of crystalline / amorphous nanocomposites, namelyMIC [43] as already discussed in Section 2.1, as well as grainboundary grooving [66] / Rayleigh instability [67] in combinationwith solid-state dewetting [68]. Thermal grooving commonlyrequires a polycrystalline nature [66] and as-deposited AlOxHyis amorphous. However, the decomposition and subsequentdensification and crystallisation upon vacuum annealing giverise to nanocrystalline domains (Figure 5e) that are expected toallow subsequent grooving. Through the process of dewetting,11 of 20e Commons LicenseFIGURE 7 S/TEM results of Al / AlOxHy after annealing at Thom = 0.94 (∼600◦C): (a) BF-TEM image over several original bilayer periods, (b)HAADF-STEM of several hundred nanometres with SAED inset, (c) higher magnification HAADF-STEM image; EDS investigation including (d) Al–Oyield, and (e) representative out-of-plane linescan over a couple of bilayer periods; and (f) automated crystal orientation mapping of inverse pole figureout-of-plane map.mnads[dAfwauascmtOmpAfqca1 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativetastable crystalline thin films frequently evolve into discreteanoparticles on the substrate surface, driven by void formationnd the minimisation of the total free energy [68]. Althoughewetting is not expected for amorphous AlOxHy due to itstill strong bonding character and low mobility, Freddi et al.69] recently reported on the crystalline nature of Ge islandsewetted from monolithic amorphous Ge films. In the Al /l2O3 system, Hieke, Dehm and Scheu [70] reported on theormation of faceted voids through dewetting of crystalline Alith a native amorphous oxide grown on α-Al2O3. At 600◦Cnnealing inAr atmosphere, the crystalline Al filmwas retractingnder the continuous surface oxide to give rise to void formationnd pinning of the retracting Al film induced irregular voidhapes. Simultaneously the surface oxide was transformed torystalline γ-Al2O3. This observation is similar to the observedicrostructure change in phase-transformed Al2O3 here, wherehe spherical shape of Al2O3 can be a result of competing Al andmass flux upon annealing. It is worth noting that such dewettedicrostructures could still be used to tailor functional materialroperties. Recently, advancesweremade by alternating dewettedu particles with ALD-Al2O3 toward disordered metasurfacesor enhanced optical properties [71]. Hence, it remains an openuestion whether dewetted AlOxHy in an architecture like therystalline / amorphous here could indeed be used for similarpplications.2 of 20One mechanism not yet considered for the nanolaminated filmsare phase transformations upon annealing, which release stress.Krautheim et al. [72] investigated mechanical stress in ALD-Al2O3 films on Si–SiO2/Si3N4 substrates through wafer curvaturemeasurements during thermal cycling until 870◦C and com-plementary X-ray techniques. Films with layer thicknesses inthe range of 25–60 nm showed a sharp tensile stress increaseat 780◦C–790◦C on the order of 1 GPa, which was linked tovolume reduction from crystallisation. Yet, 6 nm films showeda continuous reversible tensile stress rise upon annealing upto almost 6 GPa with no crystallisation detected. Taking theabove-mentioned points into consideration, the thermal insta-bility mechanisms in Al100-zNiz / AlOxHy can be summarisedas depicted in Figure 8. Here, Ni exhibits a dual role in thenanolaminates: it enhances as-deposited hardness but reducesthermal stability by promoting Al3Ni precipitation and AlOxHydestabilisation. Under the present processing conditions, Ni istherefore not an effective stabilising dopant unless these phase-separation and oxide-crystallisation pathways are suppressed.2.5 Consequences for Mechanical PropertiesUltimately, the evolution of chemistry and microstructurewith annealing temperature is put into perspective withSmall, 2026e Commons LicenseFIGURE 8 Schematic illustration of the derived thermal instability mechanisms for Al100-zNiz / AlOxHy nanolaminates, with and without Ni-addition in the PVD layers, depending on Thom referencing to Al melting point around of ∼660◦C.tfndFettABAiataArS 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativhe mechanical properties of the Al100-zNiz / AlOxHy thinilms to evaluate their thermomechanical robustness. Here,anoindentation hardness serves as a screening parameter and isisplayed as a function of homologous annealing temperature inigure 9. Due to its nanocomposite nature, a double abscissa ismployed for (1) the already established Thom (T/Tm) respectinghe Al metal character as well as (2) relative decompositionemperature for the AlOxHy interlayers (T/Tdecomp.) assuminglOxHy decomposition around 300◦C [62].ased on the data shown in Figure 9, vacuum annealing ofl100-zNiz / AlOxHy nanolaminates up to Thom = 0.94 for 1 hs gradually lowering the nanoindentation hardness of all threerchitectures. The evident strengthening in the Ni-doped archi-ectures in the as-deposited state is continuously reduced uponnnealing.Notably, the nanoindentationhardness of as-depositedl95Ni5 / AlOxHy film is 98% higher (5.3 ± 0.1 GPa) than theeference Al / AlOxHy (2.7 ± 0.1 GPa) and this difference ismall, 2026significantly reduced to only 12.5% after vacuum annealing atThom = 0.94 (1.8 ± 0.2 GPa versus 1.6 ± 0.1 GPa). An exep-tion to the linear hardness decay is the capability of all threearchitectures to quasi-maintain hardness after vacuum annealingat Thom = 0.46. Upon further annealing until Thom ≤ 0.61, twomechanisms are expected to influence the steep hardness decayconcurrently: diffusion and reaction within the Al-Ni layers,and the simultaneous decomposition of the AlOxHy layers at0.6<T / Tdecomp. < 1.0. Still, the hardness of Ni-doped films iscontinuously exceeding the reference Al / AlOxHy hardness upto Thom = 0.77. Correlation of residual stresses in Figure 6 andnanoindentation hardness in Figure 9 additionally reveals thedecay in nanoindentation hardness with a rise in tensile residualstress for all samples. However, the direct effect of residualstress on the measured hardness appears to be limited. All filmsretain nearly constant hardness after vacuum annealing at Thom= 0.46 (160◦C for 1 h; Figure 9), despite an increase in tensileresidual stress of approximately 200 MPa (Figure 6). Assuming13 of 20e Commons LicenseFIGURE 9 Ex situ nanoindentation hardness of the Al100-zNiz /AlOxHy thin films as a function of homologous temperature / vacuumannealing temperature for 1 h.ntaWNipdtmdmioihfIdbqAvcg0dpiadH7/FIGURE 10 Yield stress σYS from nanoindentation hardness H (YS∼H/3) of the Al100-zNiz / AlOxHy as a function of effective Al grain size(IP grain size x √3 [28]) in a Hall-Petch plot. Additionally visualised areconfined layer slip strengthening [73] (in turquoise) from the respectiveFCC Al grain size regimes and experimentally observed yield stress forultrafine-grained (UFG) Al including the inverse Hall-Petch regime [64,74, 75] (in brown). Relevant strengthening regimes stem from confinedlayer slip (stable AlOxHy) [73], Al + Al3Ni rule of mixture, and Orowanbowing [76] around spherical Al2O3 inclusions.1 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativegligible microstructural evolution during this annealing step,his indicates that moderate changes in residual stress have onlymarginal influence on nanoindentation hardness.e have previously shown that the strengthening in as-depositedi-doped Al100-zNiz / AlOxHy is a combined consequence of a risen (crystalline / amorphous) interfaces, dislocation density, andrecipitation strengthening in the crystalline matrix that impedeislocation motion through combined efforts [33]. However, ashe microstructures change significantly upon annealing, theechanisms need to be put into consideration. The followingiscussion is therefore used to predict validity ranges for theechanisms considering the change in architecture and chem-stry. As for nanolaminated materials, yield stress as a functionf effective layer thickness (IP grain size x √3 [28]) is displayedn Figure 10. It is explicitly emphasised that the yield stressere is derived from nanoindentation hardness with a Taboractor of 3 for a facilitated comparison of strength contributions.t is annotated that this relation and therefore the followingiscussion involves simplification, and the comparisons shoulde regarded as semi-quantitative estimates rather than preciseuantitative validation. The yield stress as a function of effectivel grain size will then allow a more direct comparison ofarying strength upon changing the architecture and chemistry atomparable Al grain size. Here, we propose that, as the effectiverain size approaches ∼0.1 (∼100 nm) after annealing at Thom >.61 (300◦C), concurrent with interlayer rupture, the dominanteformation mechanism transitions from confined layer slip torecipitation-/inclusion-mediated strengthening. This transitions a mechanistic interpretation due to significant strength dropnd the (partial) lack of reinforcing interlayers, rather than airect experimental verification.ere, Hall-Petch coefficient around 6.0 × 103 MPa nm−1/2,.5×103 MPa nm−1/2, and 5.9×103 MPa nm−1/2 can be derived for AlAlOxHy, Al98Ni2 / AlOxHy, and Al95Ni5 / AlOxHy, respectively.4 of 20Upon annealing, the Al grains are growing and the confinedlayer slip (CLS) strengthening [73] is therefore progressivelydecreasing. Simultaneously, but not annotated here, replacingthe crystalline / amorphous interfaces through crystalline /crystalline interfaces should reduce strength as reported for Cu /CuZr nanolaminates due to the replacement of strain-compatibleamorphous interlayers [77]. When comparing to nanocrystallineAl in the grain size regime < 0.2 nm−1/2 and ∼ 200–500 MPa [64,74, 75], the derived yield stress for Al / AlOxHy of 534–890 MPais in good agreement with strengthening from CLS, confirmingmainly architectural strengthening.In Ni-containing films, the softening upon annealing due toAl grain growth is partially compensated by the formation ofAl3Ni and grain growth. Zhang et al. [78] reported 15% to19% strengthening from the formation of 100–150 nm Al3Ninanoparticles in the matrix of friction-stirred Al7075. Similarly,the phase-separated PVD layers should gain from Al + Al3Nirule-of-mixtures strengthening σROM according to:𝜎𝑅𝑂𝑀 = 𝑉𝐴𝑙𝜎𝐴𝑙 + 𝑉𝐴𝑙3𝑁𝑖𝜎𝐴𝑙3𝑁𝑖 (2)with VAl volume fraction of Ni-deficient Al, σAl yield stress of theNi-deficient Al, and VAl3Ni volume fraction of Al3Ni, and σAl3Niyield stress of Al3Ni. At effective Al grain size 0.10 nm−1/2< d <0.14 nm−1/2, the Al95Ni5 / AlOxHy has a 54% to 62% higher derivedyield stress than the reference Al / AlOxHy (Figure 10).At Thom > 0.77 (∼450◦C), and effective grain size < 0.10 nm−1/2in Figure 10, the CLS model is not applicable anymore dueto breaking of the AlOxHy interlayers. This is confirmed bythe strongly reduced yield stress of Ni-containing films suchthat all Al / AlOxHy, Al98Ni2 / AlOxHy, and Al95Ni5 / AlOxHypossess a similar yield stress of 567 ± 28 MPa (average Al100-zNizSmall, 2026e Commons License/afsi1gia∼sr[bewPiA0i1bonOAtCHaNossid3A(cmm(fS 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable CreativAl2O3). Here, the growth of Al3Ni is not contributing tony strengthening, which is mainly expected from the thinilm architecture. Rather, the formation of partially crystalline,pherical κ-Al2O3 inclusions in the Al100-zNiz matrix at 600◦Cnduces strengthening. Strengthening in the range of 10% to70MPawas also observed in accumulative roll-bonded ultrafine-rained (UFG) Al with 0.5 vol.% Al2O3 particles that additionallynduced grain refinement [79]. Here, the Al100-zNiz / Al2O3 filmfter vacuum annealing at Thom = 0.94 (∼600◦C) has a yield stress567 ± 28 MPa (average Al100-zNiz / Al2O3) at effective Al grainize in the range of 150–200 nm. At that respective grain sizeange, pure Al is expected to have a yield stress of 280–400 MPa64, 74, 75]. Taking the spherical Al2O3 as obstacles for Orowanowing into account, strengthening σBow arises from the followingquation [76]:𝜎Bow = 0.4𝑀𝐺𝑏𝜋𝜒𝑙𝑛(2𝑟𝑠𝑏)(1 − 𝜈)−0.5 (3)ith Taylor factor M, shear modulus G, Burger’s vector b,oisson’s ratio ν, intersection radius of particles rs, and derivednter-particle spacing χ derived from volume fraction f as in [80]:𝜒 = 𝑟𝑠(√𝜋∕𝑓 − 2)(4)pplying an average diameter of 18± 7 nm and volume fraction of.68% for spherical Al2O3 inclusions, derived from STEM imagesn Figure 7, Orowan strengthening in the range of σBow 90–70 MPa has to be expected. Hence, the strengthening by Orowanowing is in good agreement with the yield stress differencebserved for Al100-zNiz / AlOxHy (567 ± 28 MPa) compared toanocrystalline Al in the same grain size regime (400 MPa [64]).verall, the discussion on softening upon annealing Al100-zNiz /lOxHy demonstrates that the mechanical evolution of the archi-ectures can be quantitatively tracked through themechanisms ofLS, rule-of-mixture strengthening, and Orowan strengthening.ere, the combined framework of synchrotron XRD, S/TEMnd APT enables tracking of the microstructural evolution. Ini-doped films, the formation and growth of Al3Ni grains notnly pose obstacles to Al grain growth at low Thom, but induceignificant strengthening too. However, the most significanttrengthening can be derived from crystalline / amorphousnterlayers due to the significant reduction in yield stress uponestabilisation and interlayer breaking.Conclusionssystematic approach was presented to unravel the thermalin)stability of Al100-zNiz / AlOxHy nanolaminates by cross-orrelating (synchrotron) XRD, high-resolution microscopy, andechanical testing. Following a 1 h vacuum-annealing treat-ent at five homologous temperatures from 0.46 to 0.94∼160◦C to ∼600◦C), the results can be summarised in theollowing manner:∙ Synchrotron XRD allowed a non-destructive, reliable, andefficient investigation of angle-dependent FCC Al crystallitesize (out-of-plane and in-plane).mall, 2026∙ At low homologous temperatures (Thom ≤ 0.52, ∼210◦C),two competing mechanisms dominate the microstructureevolution: (1) Ni-segregation at Al grain boundaries and (2)phase separation FCC Al100-zNiz → FCC Al and orthorhombicAl3Ni. This phase separation effectively limits the thermalstability, reducing the driving force for stabilising Al grainboundaries.∙ Hydroxylated 1 nm ALD-AlOxHy interlayers deposited atroom temperature decompose upon annealing; the combinedanalytical framework suggests both interlayer swelling andoxidation of the adjacent Al layers. Crystallisation to κ-Al2O3and subsequent interlayer breaking are observed with anonset below Thom = 0.77 (∼450◦C). Together, these findingssuggest that the hydrogen impurities in the ALD layerspromote out-of-plane Al grain growth occurring at lowertemperatures than expected.∙ Upon further annealing at Thom = 0.94 (∼600◦C), thenanolaminated architecture is fully dissolved and partiallycrystalline, spherical Al2O3 particles embedded in the coarsergrained Al(100-zNiz) matrix are found.∙ Combined, the presented degradation mechanisms causea continuous decay in nanoindentation hardness. Whileat (close-to) ambient conditions, chemistry variation fromAl100-zNiz significantly influences hardness, at higher temper-atures (Thom > 0.77), the dominant mechanisms change, andthe hardness is controlled primarily by the cooperative effectsbetween the alloy matrix and the oxide precipitates.Overall, the combined structural and mechanical analyses revealthat the thermal instability of Al100-zNiz / AlOxHy nanolami-nates is governed by a sequence of coupled phase separationand decomposition processes that progressively dissolve thenanoscale architecture. Future work toward more stable PVD-AlQ (Q dopant elements) / ALD-Al2O3 nanolaminates shouldtherefore focus on applying (1) Al-Q layers without matrixprecipitation and crystalline Al2O3 stabilisation as well as (2)altered ALD-Al2O3 processing to reduce the water content(hydroxylated species). It is emphasised that great potential isseen here for ALD-Al2O3 from O3 / O-plasma and TMA, whilehigh temperature ALD is limited by significant coarsening ofthe nanocrystalline Al grains during deposition. Finally, the herepresented findings establish a comprehensive framework to cor-relate atomic-scale transformations with macroscopic propertydegradation in complex metal/oxide nanocomposites, which is ofinterest for a broad range of laminated coatings such as emergingH barrier coatings.4 Experimental SectionThe thermal stability of nanolaminated Al100-zNiz / AlOxHywas investigated after 1 h vacuum annealing over a homol-ogous temperature range of Thom = 0.46–0.94 (≈160–600◦C)using a correlative multi-scale characterisation approach com-bining (synchrotron) XRD, HR-S/TEM, and atom probe tomog-raphy (APT). Orientation-resolved synchrotron XRD providedstatistically robust grain-size analysis over macroscopic sam-ple areas, complementing nanoscale structural and composi-tional characterization to establish processing–microstructure15 of 20e Commons License–ntmdnTaeNrdbcasa2dowwmdbpa1oTbAt5afWmbta=tnd0gsttfFtAtASc1 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creativproperty relationships. This integrated framework reveals howanolamination, solute segregation, and interfacial transforma-ions control microstructural evolution and the retention/loss ofechanical strengthening during thermal exposure, providingesign insights for improving the thermal stability of PVD/ALDanocomposites.he thin filmswere deposited bymeans of hybrid PVD (Al100-zNiz)nd ALD (AlOxHy) in a SwissCluster AG SC-1. Al100-zNiz lay-rs were deposited from two elemental Al targets and onei target at room temperature, while AlOxHy was grown byoom-temperature ALD from trimethylaluminium (TMA) andeionised (DI)-H2O. Themetallic Al100-zNiz layers were depositedy physical vapour deposition (PVD) under controlled vacuumonditions: the base and working pressure were always < 9×10−7nd∼5×10−3 mbar. Applying 2×280mA direct currentmagnetronputtering on 2“ Al (99.999%, HMW Hauner AG, Germany)nd 1×15-33 mA high power impulse magnetron sputtering on” Ni (99.999%, HMW Hauner AG, Germany), the resultingeposition rate was ∼0.08 nm s−1. All films were depositedn Si+80 nm Si3N4 (Microchemicals AG, Germany) substratesithout rotation. Prior to this, film thicknesses and chemistriesere calibrated by depositing monolithic Al100-zNiz on TEMembranes by varying deposition time and magnetron powerensities. The AlOxHy interlayers were subsequently introducedy room-temperature ALD using TMA and DI-H2O with definedulse/purge sequences: TMA pulse of 200 ms at 250 sccm Ar,nd purge of 20 000 ms at 500 sccm Ar; DI H2O pulse of50 ms at 250 sccm, and purge of 20000 ms at 500 sccm Ar. Anxide growth per cycle of around 0.14 nm cycle−1 was reached.he multilayer architecture consisted of 114 Al100-zNiz – AlOxHyilayers.n Advance Rico Mini Lamp Annealer MILA-5050 under con-inuous Ar flow of 1000 sccm and vacuum backpressure around×10−2 mbar at heat ramping of 10 K min−1 was utilised for thennealing experiments at 160◦C, 210◦C, 300◦C, 450◦C, and 600◦Cor 1 h (in the manuscript, referred to as vacuum annealing).hile the residual oxygen/water partial pressure was not directlyonitored but affect the interfacial chemistry, the reducedackpressure combined with continuous Ar flow is expectedo substantially limit residual oxidizing species compared tombient conditions. The lower homologous temperatures Thom0.46 (∼160◦C) and Thom = 0.52 (∼210◦C) were chosen torigger mainly Ni diffusion in the Al matrix, while keeping theanolaminated structure stable, similarly to previous studies onoped nanocrystalline Al [45]. The higher temperatures Thom =.61, 0.77, and 0.94 (∼300◦C, ∼450◦C, and ∼600◦C) allow investi-ation of microstructure evolution under harsher conditions andcreening of potential application temperatures. It is emphasisedhat all the samples were isothermally annealed without priorhermal treatment, and the homologous temperature was derivedrom Tm = 933 K (660◦C) to allow comparison to Al alloys.or analysis, we combine (synchrotron) XRD, S/TEM, and APTo complementary track the phase and chemistry evolution inl100-zNiz – AlOxHy. Synchrotron XRD is a powerful techniqueo track the angle-dependent evolution of coherent (111) FCCl crystallite size, residual stress, and crystallisation of AlOxHy./TEM adds information on the local phase evolution andhemistry, as well as interface segregation tendencies, while APT6 of 20provides information on the diffusion behaviour of the (light)elements.XRD patterns of Al – AlOxHy, Al98Ni2 – AlOxHy, and Al95Ni5 –AlOxHy thin films were acquired directly after the sampleswere taken out of the vacuum furnace in a Bruker D8 Dis-covery in 2θ-range from 20 to 100◦ at 40 kV and 40 mA.The instrument broadening was corrected by measuring a LaB6standard.Synchrotron XRDwas applied to approach both out-of-plane andin-plane (111) FCCAl coherent crystallite size (in the followingOPand IP crystallite size) analysis for each sample. Eight psi anglesfrom ψ = 0 up to 70◦ were chosen to approach OP and almostIP (111) FCC Al crystallite size. Three phi angles (Φ = 0, 45, and90◦) were chosen for comparability. Awavelength of 0.128 nmwasused, and the instrument broadeningwas corrected bymeasuringa LaB6 standard at every angle ψ. Ultimately, the ψ -dependent(111) FCC lower-bound coherent domain size of each sample wasanalysed by applying the Scherrer equation with a shape factorequal to 1, allowing the best fit to as-deposited layer thicknessof ∼25 nm under the assumption of equiaxed grains. Here, theScherrer lower-bound coherent domain size for the IP conditionis termed “crystallite size” as a short-hand in the following. Fromthe same dataset, sin2ψ analysis [81] was performed to determinethe residual stress and stress-free lattice parameter evolution overchemistry and temperature variation.TEM samples were prepared using a Thermo Fisher ScientificHelios 5 Hydra DualBeam pFIB – scanning electron microscope(SEM) equipped for Xe+ milling, and their crystallography andchemistry were subsequently analysed by S/TEM alike the pro-cedure described in [44]. TEM was conducted on a ThermoFisher aberration corrected (probe) Themis 200 G3 operated at200 kV. Analyses of S/TEM images, fast fourier transform (FFT)images, and selected area electron diffraction (SAED) patternswere conducted using the CrysTBox software [82]. STEM-EDSmaps were acquired with 20 eV dispersion and 5 µs dwell timeand absorption-corrected assuming ∼100 nm sample thicknessand 2.7 g cm−3 density.Atom probe specimens were also prepared in a Thermo FisherHelios 5 Hydra UX dual-beam microscope with a pFIB. Xe ionswere used at 12 kV for deposition of a protection layer, using aprecursor mixture of 99% Pt and 1% C, while milling operationswere conducted at 30 kV. Plan-view lift-outs were done (seesupplementary material of [83]) in order to align the Al100-zNiz –AlOxHy interfaces parallel to the specimen shank. Sharpeningwas carried out at 12 kV with the final step at 5 kV. APT was donein a CAMECA local electrode atom probe (LEAP) 4000X HR.Field evaporation was assisted by voltage pulsing with 20% pulsefraction, 100 kHz pulse frequency, 60 K base temperature and0.5% detection rate. Up to 50 million ions were acquired and datareconstruction was done with CAMECA AP Suite 6.1, employingthe shank angle protocol. Visualisation of the interfaces wasrealised with 10-nm-thin slices of the reconstructions. H wasincluded in the ranging of mass spectra and the content canbe understood as an upper boundary value due to the likelycontribution of residual gas from the analysis chamber.Moreover,C and N impurities with the sum < 1 at.% were detected and arenot discussed further.Small, 2026e Commons LicenseTtmCcSwpdfutbNtZBsicdtpt1fPfaAfiafAH&iIFIeAFCJRAH“(a(S 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creatio track the water precursor adsorption during room-emperature ALD, in situ quartz crystal microbalance (QCM)onitoring was performed in an SC-Qube ALD system (Swissluster) using a 6 MHz RC-cut quartz crystal with masshanges (∆m/A, in ng/cm2) calculated from frequency shifts viaauerbrey’s equation:Δ𝑓 = −2𝑓20𝐴√𝜌𝑞𝜇𝑞Δ𝑚 (5)here f0 is the initial resonance frequency (6 MHz), A is theiezoelectrically active crystal area (in cm2), ρq is the quartzensity (2.65 g cm−3) and μq is the shear modulus of quartzor RC-cut crystal (2.947⋅1011 g cm−1⋅s−2). A conservative relativencertainty of ±3% was applied to all QCM mass measurementso account for instrumental noise, thermal drift, and long-termaseline fluctuations.anoindentation was performed on a Zwick Roell nanoinden-er system (ZHN Nanoindenter with a 2 N measuring head,wickRoell GmbH & Co. KG, Ulm, Germany) equipped with aerkovich diamond tip. A total of 25 indents were collected perample condition. The indentation experiments were conductedn a quasi-continuous stiffness measurement performed in load-ontrolled mode with maximum indentation depth of 1 µm. Aetailed description of the measurement procedure, includinghe tip area function and instrument compliance calibration, isrovided in [84]. Indentations were conducted with a Berkovichip (Synton-MDP Ltd., Switzerland) up to a maximum load of00 mN. Depth-resolved hardness (H) profiles were extractedrom the load-displacement curves according to the Oliver andharr method [85]. The film-only hardness values were derivedrom a plateau region between 100–300 nm indentation depths proposed by Fischer-Cripps [86] to minimise substrate effects.fter discarding outliers, at least 20 valid indents were collectedor each condition, and the measurement error is derived fromndent-to-indent variation and indent-over-depth variation. It isnnotated that slight pile-up from plastic behaviour was observedor every sample.uthor Contributions.C. Jansen: Investigation, Writing – original draft, Writing – reviewediting, Conceptualisation, Methodology. B. Putz: Investigation, Writ-ng – review & editing, Validation, Funding Acquisition. A. Sharma:nvestigation, Writing – review & editing. L. Lapeyre: Investigation..F. Klimashin: Investigation, Writing – review & editing. D. Gutnik:nvestigation.M. Hans: Investigation, Visualisation, Writing – review &diting. J.M. Schneider: Writing – review & editing, Resources, Fundingcquisition. J.J. Schwiedrzik: Writing – review & editing, Supervision,unding Acquisition. T.E.J. Edwards: Writing – review & editing,onceptualisation, Methodology, Investigation, Validation, Supervision.. Michler: Writing – review & editing, Conceptualisation, Supervision,esources, Funding Acquisition.cknowledgementsCJ was supported by the Innosuisse Innovation project 109.352.1ORALCOAT”. TEJE gratefully acknowledges the support of Innosuisse42220.1 IP-ENG) and JSPS KAKENHI Start-Up grant (24K23036). MHnd JMS acknowledge funding from Deutsche ForschungsgemeinschaftDFG) within the Collaborative Research Centre SFB 1394, “Structuralmall, 2026vand Chemical Atomic Complexity – From Defect Phase Diagrams toMaterials Properties” (project ID 409476157). B.P. acknowledges fundingfrom the Swiss National Science Foundation under the Ambizionegrant agreement No. PZ00P2_202089. We acknowledge SOLEIL forprovision of synchrotron radiation facilities and thank Cristian Mocutaand Dominique Thiaudiere for assistance in using the beamline DiffAbsunder proposals 20230260, and 20230080. The authors emphasise support& assistance in this work by Dr. Daniele Casari for TEM support, aswell as Prof. Damien Faurie and Prof. Pierre-Olivier Renault for theirsupport in synchrotron analysis and crystallographic input. The authorsultimately acknowledge ScopeM for their support & assistance in thiswork, especially by Dr. Joakim Reuteler.Open access publishing facilitated by ETH-Bereich Forschungsanstalten,as part of theWiley - ETH-Bereich Forschungsanstalten agreement via theConsortium Of Swiss Academic Libraries.Conflicts of InterestThe authors declare no conflicts of interest.Data Availability StatementThe data that support the findings of this study are available from thecorresponding author upon reasonable request.References1. C. J.Hung, S. K.Nayak, Y. Sun, et al., “NovelAl-XAlloysWith ImprovedHardness,”Materials &Design 192 (2020): 108699, https://doi.org/10.1016/j.matdes.2020.108699.2. A. Jankowski, J. Ferreira, and J. Hayes, “Activation Energies of GrainGrowth Mechanisms in Aluminum Coatings,” Thin Solid Films 491(2005): 61–65.3. D. Raabe, M. Herbig, S. Sandlöbes, et al., “Grain Boundary SegregationEngineering in Metallic Alloys: A Pathway to the Design of Interfaces,”Current Opinion in Solid State and Materials Science 18 (2014): 253–261,https://doi.org/10.1016/j.cossms.2014.06.002.4. T. Y. Huang, A. R. Kalidindi, and C. A. Schuh, “Grain Growth andSecond-Phase Precipitation in Nanocrystalline Aluminum–ManganeseElectrodeposits,” Journal ofMaterials Science 53 (2018): 3709–3719, https://doi.org/10.1007/s10853-017-1764-4.5. A. Devaraj, W. Wang, R. Vemuri, et al., “Grain Boundary Segregationand Intermetallic Precipitation in Coarsening Resistant NanocrystallineAluminum Alloys,” Acta Materialia 165 (2019): 698–708, https://doi.org/10.1016/j.actamat.2018.09.038.6. V. M. S. Muthaiah and S. Mula, “Effect of Zirconium on ThermalStability of Nanocrystalline Aluminium Alloy Prepared by MechanicalAlloying,” Journal of Alloys and Compounds 688 (2016): 571–580, https://doi.org/10.1016/j.jallcom.2016.07.038.7. D. S. Gianola, B. G.Mendis, X.M. Cheng, andK. J. Hemker, “Grain-SizeStabilization by Impurities and Effect on Stress-Coupled Grain Growthin Nanocrystalline Al Thin Films,” Materials Science and Engineering: A483–484 (2008): 637–640, https://doi.org/10.1016/j.msea.2006.12.155.8. Q. Li, J. Cho, S. Xue, et al., “High Temperature Thermal andMechanicalStability of High-Strength Nanotwinned Al Alloys,” Acta Materialia 165(2019): 142–152, https://doi.org/10.1016/j.actamat.2018.11.011.9. T. J. Rupert, J. C. Trenkle, and C. A. Schuh, “Enhanced Solid SolutionEffects on the Strength of Nanocrystalline Alloys,” Acta Materialia 59(2011): 1619–1631, https://doi.org/10.1016/j.actamat.2010.11.026.10. P. R. Cantwell, M. Tang, S. J. Dillon, J. Luo, G. S. Rohrer, and M. P.Harmer, “Grain Boundary Complexions,”ActaMaterialia 62 (2014): 1–48,https://doi.org/10.1016/j.actamat.2013.07.037.11. T. Lei, J. Shin, D. S. Gianola, and T. J. Rupert, “Bulk Nanocrystalline AlAlloys With Hierarchical Reinforcement Structures via Grain Boundary17 of 20e Commons Licensehttps://doi.org/10.1016/j.matdes.2020.108699https://doi.org/10.1016/j.cossms.2014.06.002https://doi.org/10.1007/s10853-017-1764-4https://doi.org/10.1016/j.actamat.2018.09.038https://doi.org/10.1016/j.jallcom.2016.07.038https://doi.org/10.1016/j.msea.2006.12.155https://doi.org/10.1016/j.actamat.2018.11.011https://doi.org/10.1016/j.actamat.2010.11.026https://doi.org/10.1016/j.actamat.2013.07.037S11rJ1“TA21aAs1“EJ11Iah1LA01T11obh2oF2Bdh2A12Nh2a82“Wh2C(2Wt2T1 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creegregation and Complexion Formation,” Acta Materialia 221 (2021):17394, https://doi.org/10.1016/j.actamat.2021.117394.2. H. A. Murdoch and C. A. Schuh, “Estimation of Grain Boundary Seg-egation Enthalpy and its Role in Stable Nanocrystalline Alloy Design,”ournal of Materials Research 28 (2013): 2154–2163.3. F. Tang, D. S. Gianola, M. P. Moody, K. J. Hemker, and J. M. Cairney,Observations of Grain Boundary Impurities in Nanocrystalline Al andheir Influence onMicrostructural Stability andMechanical Behaviour,”cta Materialia 60 (2012): 1038–1047, https://doi.org/10.1016/j.actamat.011.10.061.4. T. Lei, M. Xu, J. Shin, D. S. Gianola, and T. J. Rupert, “Growthnd Structural Transitions of Core-Shell Nanorods in Nanocrystallinel-Ni-Y,” Scripta Materialia 211 (2022): 114502, https://doi.org/10.1016/j.criptamat.2022.114502.5. J. Shin, F. Wang, G. H. Balbus, T. Lei, T. J. Rupert, and D. S. Gianola,Optimizing Thermal Stability and Mechanical Behavior in Segregation-ngineered Nanocrystalline Al–Ni–Ce Alloys: A Combinatorial Study,”ournal of Materials Research 37 (2022): 3083–3098, https://doi.org/10.557/s43578-022-00715-x.6. G. H. Balbus, J. Kappacher, D. J. Sprouster, et al., “Disorderednterfaces Enable High Temperature Thermal Stability and Strength inNanocrystalline Aluminum Alloy,” Acta Materialia 215 (2021): 116973,ttps://doi.org/10.1016/j.actamat.2021.116973.7. G. H. Balbus, F. Wang, and D. S. Gianola, “Suppression of Shearocalization in Nanocrystalline Al–Ni–Ce via Segregation Engineering,”cta Materialia 188 (2020): 63–78, https://doi.org/10.1016/j.actamat.2020.1.041.8. Y. F. Zhang, R. Su, D. Y. Xie, et al., “Design of Super-Strong andhermally Stable Nanotwinned Al Alloys via Solute Synergy,” Nanoscale2 (2020): 20491–20505.9. K. Masui, S. Maruno, S. Sakakibara, and T. Kawaguchi, “Formationf Amorphous Al-Transition Metal (TM: Fe, Co, Ni) Binary Alloy Filmsy RF-Sputtering,” Journal of Non-Crystalline Solids 74 (1985): 271–284,ttps://doi.org/10.1016/0022-3093(85)90074-2.0. T. W. Barbee, W. H. Holmes, D. L. Keith, andM. K. Pyzyna, “Synthesisf Amorphous Niobium-Nickel Alloys by Vapor Quenching,” Thin Solidilms 45 (1977): 591–599, https://doi.org/10.1016/0040-6090(77)90251-6.1. Y. Kawamura, H. Mano, and A. Inoue, “Nanocrystalline Aluminumulk Alloys With a High Strength of 1420 MPa Produced by the Consoli-ation of Amorphous Powders,” Scripta Materialia 44 (2001): 1599–1604,ttps://doi.org/10.1016/S1359-6462(01)00781-3.2. A. Inoue, “Stabilization of Metallic Supercooled Liquid and Bulkmorphous Alloys,” Acta Materialia 48 (2000): 279–306, https://doi.org/0.1016/S1359-6454(99)00300-6.3. I. J. Beyerlein, Z. Li, and N. A. Mara, “Mechanical Properties of Metalanolaminates,” Annual Review of Materials Research 52 (2022): 281–304,ttps://doi.org/10.1146/annurev-matsci-081320-031236.4. S. Pathak, N. Velisavljevic, J. K. Baldwin, et al., “Strong, Ductile,nd Thermally Stable bcc-Mg Nanolaminates,” Scientific Reports 7 (2017):264.5. D. Bhattacharyya, N. A. Mara, R. G. Hoagland, and A. Misra,Nanoindentation and Microstructural Studies of Al/TiN Multilayersith Unequal Volume Fractions,” Scripta Materialia 58 (2008): 981–984,ttps://doi.org/10.1016/j.scriptamat.2008.01.054.6. J. Li, Q. Tang, N. Chawla, and M. Hassani, “Deformation of Metal-eramic Nanolaminates at Extreme Strain Rates,” Scripta Materialia 2732026): 117089, https://doi.org/10.1016/j.scriptamat.2025.117089.7. P. Baral, S. Jaddi, H. Wang, et al., “Al2O3/Al Hybrid Nanolaminatesith Superior Toughness, Strength and Ductility,” Nature Communica-ions 16 (2025): 1355, https://doi.org/10.1038/s41467-025-56512-7.8. T. E. J. Edwards, T. Xie, N. Maria Della Ventura, et al., “On thehinnest Al2O3 Interlayers in Al-Based Nanolaminates to Enhance8 of 20Strength, and the Role of Constraint,” Acta Materialia 240 (2022): 118345,https://doi.org/10.1016/j.actamat.2022.118345.29. B. Putz, T. E. J. Edwards, E. Huszar, et al., “In Situ Fragmentation ofAl/Al2O3 Multilayers onFlexible Substrates inBiaxial Tension,”Materials& Design 232 (2023): 112081, https://doi.org/10.1016/j.matdes.2023.112081.30. P. R. Cantwell, T. Frolov, T. J. Rupert, et al., “Grain BoundaryComplexion Transitions,” Annual Review of Materials Research 50 (2020):465–492, https://doi.org/10.1146/annurev-matsci-081619-114055.31. J. Y. Cheng, Z. Li, D. L. Poerschke, J. K. Baldwin, B. L. Bresnahan, andN. A. Mara, “Thermal Stability of 3D Interface Cu /Nb Nanolaminates,”Scripta Materialia 254 (2025): 116319, https://doi.org/10.1016/j.scriptamat.2024.116319.32. V. Miikkulainen, M. Leskelä, M. Ritala, and R. L. Puurunen, “Crys-tallinity of Inorganic FilmsGrown byAtomic Layer Deposition: OverviewandGeneral Trends,” Journal of Applied Physics 113 (2013): 021301, https://doi.org/10.1063/1.4757907.33. H. C. Jansen, A. Sharma,M.Hans, et al., “Nanolaminated Al100−zNiz/ AlOxHy Thin Films by Hybrid PVD / ALD: An Approach TowardsInterface-Engineered Thin Films by Dual-Route Tailoring,” Materials &Design 263 (2026): 115580, https://doi.org/10.1016/j.matdes.2026.115580.34. L. M. Vogl, P. Schweizer, L. Pethö, A. Sharma, J. Michler, and I.Utke, “From Metal Nanowires to Ultrathin Crystalline ALD Nanotubes:Process Development and Mechanism Revealed by In Situ TEM HeatingExperiments,” Nanoscale 15 (2023): 9477–9483, https://doi.org/10.1039/D3NR01185B.35. Y. Du and N. Clavaguera, “Thermodynamic Assessment of the Al-NiSystem,” Journal of Alloys and Compounds 237 (1996): 20–32, https://doi.org/10.1016/0925-8388(95)02085-3.36. G. K. Williamson and W. H. Hall, “X-Ray Line Broadening from FiledAluminium and Wolfram,” Acta Metallurgica 1 (1953): 22–31, https://doi.org/10.1016/0001-6160(53)90006-6.37. M. Yahyazameh, M. Kavanlouei, M. Shahbaz, and Y. Beygi-khosrowshahi, “Zr Addition and Milling Time Impact on theMicrostructure and Phase Analysis of New With High-Energy BallMilling and the Subsequent Sintering,” Journal of Materials Engineeringand Performance 66 (2024): 1678–1692.38. R. Gilles, D. Mukherji, M. Hoelzel, P. Strunz, D. M. Toebbens, and B.Barbier, “Neutron and X-Ray Diffraction Measurements on Micro- andNano-Sized Precipitates Embedded in a Ni-Based Superalloy and AfterTheir Extraction from the Alloy,” Acta Materialia 54 (2006): 1307–1316,https://doi.org/10.1016/j.actamat.2005.11.004.39. C. Guerra-Nuñez, M. Döbeli, J. Michler, and I. Utke, “Reaction andGrowth Mechanisms in Al2O3 Deposited via Atomic Layer Deposition:Elucidating theHydrogen Source,”Chemistry ofMaterials 29 (2017): 8690–8703, https://doi.org/10.1021/acs.chemmater.7b02759.40. C. Cancellieri, S. Gramatte, O. Politano, et al., “Effect of Hydrogenon the Chemical State, Stoichiometry and Density of Amorphous Al2O3FilmsGrownbyThermalAtomic LayerDeposition,”Surface and InterfaceAnalysis 56 (2024): 293–304, https://doi.org/10.1002/sia.7282.41. E. Tkalcec, S. Kurajica, and J. Schmauch, “Crystallization ofAmorphous Al2O3–SiO2 Precursors Doped With Nickel,” Journal ofNon-Crystalline Solids 353 (2007): 2837–2844, https://doi.org/10.1016/j.jnoncrysol.2007.06.011.42. H. Qin, P. Sutter, and G. Zhou, “The Crystallization of AmorphousAluminum Oxide Thin Films Grown on NiAl (100),” Journal of theAmerican Ceramic Society 97 (2014): 2762–2769, https://doi.org/10.1111/jace.13036.43. Z. M. Wang, J. Y. Wang, L. P. H. Jeurgens, and E. J. Mittemeijer,“Explosive″ Crystallisation of Amorphous Germanium in Ge/Al LayerSystems; Comparison With Si/Al Layer Systems,” Scripta materialia 55(2006): 987–990.44. H. C. Jansen, A. Sharma, K. Wieczerzak, et al., “On the Efficacy ofXe+-pFIB Preparation to Avoid Ga+-FIB Induced Phase TransformationsSmall, 2026ative Commons Licensehttps://doi.org/10.1016/j.actamat.2021.117394https://doi.org/10.1016/j.actamat.2011.10.061https://doi.org/10.1016/j.scriptamat.2022.114502https://doi.org/10.1557/s43578-022-00715-xhttps://doi.org/10.1016/j.actamat.2021.116973https://doi.org/10.1016/j.actamat.2020.01.041https://doi.org/10.1016/0022-3093(85)90074-2https://doi.org/10.1016/0040-6090(77)90251-6https://doi.org/10.1016/S1359-6462(01)00781-3https://doi.org/10.1016/S1359-6454(99)00300-6https://doi.org/10.1146/annurev-matsci-081320-031236https://doi.org/10.1016/j.scriptamat.2008.01.054https://doi.org/10.1016/j.scriptamat.2025.117089https://doi.org/10.1038/s41467-025-56512-7https://doi.org/10.1016/j.actamat.2022.118345https://doi.org/10.1016/j.matdes.2023.112081https://doi.org/10.1146/annurev-matsci-081619-114055https://doi.org/10.1016/j.scriptamat.2024.116319https://doi.org/10.1063/1.4757907https://doi.org/10.1016/j.matdes.2026.115580https://doi.org/10.1039/D3NR01185Bhttps://doi.org/10.1016/0925-8388(95)02085-3https://doi.org/10.1016/0001-6160(53)90006-6https://doi.org/10.1016/j.actamat.2005.11.004https://doi.org/10.1021/acs.chemmater.7b02759https://doi.org/10.1002/sia.7282https://doi.org/10.1016/j.jnoncrysol.2007.06.011https://doi.org/10.1111/jace.13036i14TSE044rJj4SUd4DFo5C45t15iJi5AP5“A5i5Po5OS5T(5MTd6oBi16L16m0S 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Library on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Crn Al-Ni Alloys,” Scripta Materialia 260 (2025): 116589, https://doi.org/10.016/j.scriptamat.2025.116589.5. S. C. Pun, W. Wang, A. Khalajhedayati, J. D. Schuler, J. R.relewicz, and T. J. Rupert, “Nanocrystalline Al-Mg With Extremetrength due to Grain Boundary Doping,” Materials Science andngineering: A 696 (2017): 400–406, https://doi.org/10.1016/j.msea.2017.4.095.6. J. W. Matthews, Epitaxial Growth (Academic, 1975).7. H. A. Murdoch and C. A. Schuh, “Estimation of Grain Boundary Seg-egation Enthalpy and its Role in Stable Nanocrystalline Alloy Design,”ournal ofMaterials Research 28 (2013): 2154–2163, https://doi.org/10.1557/mr.2013.211.8. W. S. Cunningham, T. Lei, H. C. Howard, T. J. Rupert, and D.. Gianola, “Kinetics of Amorphous Defect Phases Measured Throughltrafast Nanocalorimetry,” Acta Materialia 304 (2026): 121764, https://oi.org/10.1016/j.actamat.2025.121764.9. E. Ma, C. V. Thompson, and L. A. Clevenger, “Nucleation and Growthuring Reactions in Multilayer Al/Ni Films: The Early Stage of Al3Niormation,” Journal of Applied Physics 69 (1991): 2211–2218, https://doi.rg/10.1063/1.348722.0. F. De Geuser and B. Gault, “Metrology of Small Particles and Solutelusters by Atom Probe Tomography,” Acta Materialia 188 (2020): 406–15.1. F. Cuevas and M. Latroche, “Intermetallic Alloys as Hydrogen Get-ers,” Journal of Alloys andCompounds 905 (2022): 164173, https://doi.org/0.1016/j.jallcom.2022.164173.2. J. Kim, X. Yao, V. Somjit, et al., “Multilayer Alumina/AluminumCoat-ngs for Damage-Resistant Hydrogen Permeation Barrier,” Internationalournal of Hydrogen Energy 106 (2025): 226–230, https://doi.org/10.1016/j.jhydene.2025.01.300.3. Y. Fukai, “Superabundant Vacancies Formed in Metal – Hydrogenlloys Superabundant Vacancies Formed in Metal – Hydrogen Alloys,”hysica Scripta 11 (2003): 10–14.4. A. B. Belonoshko,A.Rosengren,Q.Dong,G.Hultquist, andC. Leygraf,First-Principles Study of Hydrogen Diffusion in α− Al2O3 and Liquidlumina,” Physical Review B 69 (2004): 024302.5. G. A. Y. Jr and J. R. Scully, “The Diffusion and Trapping of Hydrogenn High Purity Aluminum,” Acta Materialia 46 (1998): 6337–6349.6. L. Dai, Z. Deng, F. Auras, et al., “Slow Carrier Relaxation in Tin-BasederovskiteNanocrystals,”Nature Photonics 15 (2021): 696–702, https://doi.rg/10.1038/s41566-021-00847-2.7. M. Kneiß, P. Storm, A. Hassa, et al., “Growth, Structural andptical Properties of Coherent κ-(AlxGa1-x)2O3/κ-Ga2O3quantum Welluperlattice Heterostructures,” APL Materials 8 (2020): 051112.8. M. D. Groner, F. H. Fabreguette, J. W. Elam, and S. M. George, “Low-emperature Al2O3 Atomic Layer Deposition,” Chemistry of Materials 162004): 639–645.9. L. G. Gosset, J. F. Damlencourt, O. Renault, et al., “Interface andaterial Characterization of Thin Al2O3 Layers Deposited by ALD UsingMA/H2O,” Journal of Non-Crystalline Solids 303 (2002): 17–23, https://oi.org/10.1016/S0022-3093(02)00958-4.0. A. Wei, P. Tao, L. Dai, J. Hong, S. Li, and W. Zhao, “Enhanced DryingfWater-Containing PorousAluminumHydroxide Particles via a Cycloneased on Microchannel Oscillation,” Journal of Environmental Chem-cal Engineering 12 (2024): 113801, https://doi.org/10.1016/j.jece.2024.13801.1. P. Felfer, B. Ott, M. Monajem, et al., “An Atom Probe With Ultra-ow Hydrogen Background,” Microscopy and Microanalysis 28 (2022):255–1263, https://doi.org/10.1017/S1431927621013702.2. T. Sato, “Thermal Decomposition of Aluminium Hydroxides to Alu-inas,” Thermochimica Acta 88 (1985): 69–84, https://doi.org/10.1016/040-6031(85)85415-0.mall, 202663. G. Krautheim, T. Hecht, S. Jakschik, U. Schröder, and W. Zahn,“Mechanical Stress in ALD-Al2O3 Films,” Applied Surface Science 252(2005): 200–204.64. R.W. Hayes, D.Witkin, F. Zhou, and E. J. Lavernia, “Deformation andActivation Volumes of Cryomilled Ultrafine-Grained Aluminum,” ActaMaterialia 52 (2004): 4259–4271, https://doi.org/10.1016/j.actamat.2004.05.042.65. F. C. Larche and J. W. Cahn, “The Interactions of Composition andStress in Crystalline Solids,” Acta Metallurgica 89 (1984): 467–500.66. W. W. Mullins, “Theory of Thermal Grooving,” Journal of AppliedPhysics 28 (1957): 333–339, https://doi.org/10.1063/1.1722742.67. S. Zheng, J. S. Carpenter, J. Wang, N. A. Mara, and I. J. Beyerlein,“An Interface Facet Driven Rayleigh Instability in High-Aspect-RatioBimetallic Nanolayered Composites,” Applied Physics Letters 105 (2020):111901.68. C. V. Thompson, “Solid-State Dewetting of Thin Films,” AnnualReview of Materials Research 42 (2012): 399–434, https://doi.org/10.1146/annurev-matsci-070511-155048.69. S. Freddi, G. Sfuncia, M. Gherardi, et al., “Morphological Evolu-tion and Structural Study of Annealed Amorphous-Ge Films: InterplayBetween Crystallization and Dewetting,” Materials Science in Semicon-ductor Processing 174 (2024): 108228, https://doi.org/10.1016/j.mssp.2024.108228.70. S. W. Hieke, G. Dehm, and C. Scheu, “Annealing Induced VoidFormation in Epitaxial Al Thin Films on Sapphire (α-Al2O3),” ActaMaterialia 140 (2017): 355–365, https://doi.org/10.1016/j.actamat.2017.08.050.71. M. Chen, A. Sharma, J. Michler, X. Maeder, P. Lalanne, and A.Xomalis, “Assembling and Modeling Stacked Disordered Metasurfaces,”Nano Letters 25 (2025): 13221–13228.72. G. Krautheim, T. Hecht, S. Jakschik, U. Schröder, and W. Zahn,“Mechanical Stress in ALD-Al2O3 films,” Applied Surface Science 252(2005): 200–204.73. A. Misra, J. P. Hirth, and R. G. Hoagland, “Length-Scale-DependentDeformation Mechanisms in Incoherent Metallic Multilayered Com-posites,” Acta Materialia 53 (2005): 4817–4824, https://doi.org/10.1016/j.actamat.2005.06.025.74. G. Jeong, J. Park, S. Nam, et al., “The Effect of Grain Size OnThe Mechanical Properties Of Aluminum,” Archives of Metallurgy andMaterials 60 (2015): 1287–1291, https://doi.org/10.1515/amm-2015-0115.75. H. J. Choi, S. W. Lee, J. S. Park, and D. H. Bae, “Positive Deviationfrom a Hall-Petch Relation in Nanocrystalline Aluminum,” Materialstransactions 50 (2009): 640–643.76. E. Orowan, “Dislocations in Metals,” American Institute of Mimingand Metal lurgical Engineers 4 (1954): 69.77. X. Zhou and C. Chen, “Molecular Dynamic Simulationsof the Mechanical Properties of Crystalline/Crystalline andCrystalline/Amorphous Nanolayered Pillars,” Computational MaterialsScience 101 (2015): 194–200, https://doi.org/10.1016/j.commatsci.2015.01.033.78. S. Zhang, Y. Li, A. Frederick, et al., “In-Situ Formation of Al3Ni NanoParticles in Synthesis of Al 7075 Alloy by Friction Stir Processing With NiPowder Addition,” Journal of Materials Processing Technology 311 (2023):117803, https://doi.org/10.1016/j.jmatprotec.2022.117803.79. T. Hausöl, V. Maier, C. W. Schmidt, M. Winkler, H. W. Höppel, andM. Göken, “TailoringMaterials Properties by Accumulative Roll Bonding**,” Advanced Engineering Materials 12 (2010): 740–746.80. A. Kelly and R. Nicholson, “Dislocation-Particle Interactions,” inStrengthening methods in crystals (Elsevier Pub. Co., 1971), 627.81. D. Faurie, F. Zighem, P. Godard, et al., “In Situ X-Ray DiffractionAnalysis of 2D Crack Patterning in Thin Films,” Acta Materialia 165(2019): 177–182, https://doi.org/10.1016/j.actamat.2018.11.040.19 of 20eative Commons Licensehttps://doi.org/10.1016/j.scriptamat.2025.116589https://doi.org/10.1016/j.msea.2017.04.095https://doi.org/10.1557/jmr.2013.211https://doi.org/10.1016/j.actamat.2025.121764https://doi.org/10.1063/1.348722https://doi.org/10.1016/j.jallcom.2022.164173https://doi.org/10.1016/j.ijhydene.2025.01.300https://doi.org/10.1038/s41566-021-00847-2https://doi.org/10.1016/S0022-3093(02)00958-4https://doi.org/10.1016/j.jece.2024.113801https://doi.org/10.1017/S1431927621013702https://doi.org/10.1016/0040-6031(85)85415-0https://doi.org/10.1016/j.actamat.2004.05.042https://doi.org/10.1063/1.1722742https://doi.org/10.1146/annurev-matsci-070511-155048https://doi.org/10.1016/j.mssp.2024.108228https://doi.org/10.1016/j.actamat.2017.08.050https://doi.org/10.1016/j.actamat.2005.06.025https://doi.org/10.1515/amm-2015-0115https://doi.org/10.1016/j.commatsci.2015.01.033https://doi.org/10.1016/j.jmatprotec.2022.117803https://doi.org/10.1016/j.actamat.2018.11.0408oS8PMn8Ppt8mS18o(SAIS2 16136829, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/smll.75393 by National Institute For, Wiley Online Li2. M. Klinger, “More Features, More Tools, More CrysTBox,” Journalf Applied Crystallography 50 (2017): 1226–1234, https://doi.org/10.1107/1600576717006793.3. M. Hans, Z. Czigány, D. Neuß, et al., “Probing the Onset of Wurtzitehase Formation in (V,Al)N Thin Films by Transmission Electronicroscopy and Atom Probe Tomography,” Surface and Coatings Tech-ology 442 (2022): 128235, https://doi.org/10.1016/j.surfcoat.2022.128235.4. T. Chudoba, D. Schwenk, P. Reinsta, and M. Griepentrog, “High-recision Calibration of Indenter Area Function and Instrument Com-liance: Chudoba, Schwenk, Reinstädt, and Griepentrog,” Jom Journal ofhe Minerals Metals and Materials Society 74 (2022): 2179–2194.5. W. C. Oliver and G. M. Pharr, “An Improved Technique for Deter-ining Hardness and Elastic Modulus Using Load and Displacementensing Indentation Experiments,” Journal ofMaterials Research 7 (1992):564–1583, https://doi.org/10.1557/JMR.1992.1564.6. A. C. Fischer-Cripps, “Critical Review of Analysis and Interpretationf Nanoindentation Test Data,” Surface and Coatings Technology 2002006): 4153–4165, https://doi.org/10.1016/j.surfcoat.2005.03.018.upporting Informationdditional supporting information can be found online in the Supportingnformation section.upportingFile: smll75393-sup-0001-SuppMat.docx.0 of 20 Small, 2026brary on [01/09/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttps://doi.org/10.1107/S1600576717006793https://doi.org/10.1016/j.surfcoat.2022.128235https://doi.org/10.1557/JMR.1992.1564https://doi.org/10.1016/j.surfcoat.2005.03.018 Thermal Instability Pathways of Aluminum-Based Nanocomposites: The Role of Alloying-Element-Driven Precipitation and Amorphous Interlayers 1 | Introduction 2 | Results and Discussion 2.1 | Elucidating Thermal Stability 2.2 | Phase-Interface Competition in Al-Ni 2.3 | Nanoscale Chemistry 2.4 | Interlayer Breaking and Dewetting 2.5 | Consequences for Mechanical Properties 3 | Conclusions 4 | Experimental Section Author Contributions Acknowledgements Conflicts of Interest Data Availability Statement References Supporting Information