# Fileset

[Adv Funct Materials - 2025 - Hirai - Non-Equilibrium Magnon Engineering Enabling Significant Thermal Transport Modulation.pdf](https://mdr.nims.go.jp/filesets/57a8a1ca-dd66-498b-98c7-2a12a6359195/download)

## Creator

[Takamasa Hirai](https://orcid.org/0000-0002-5577-8018), Toshiaki Morita, Subrata Biswas, [Jun Uzuhashi](https://orcid.org/0000-0003-2023-8158), [Takashi Yagi](https://orcid.org/0000-0002-2008-872X), [Yuichiro Yamashita](https://orcid.org/0000-0003-1376-0263), Varun Kumar Kushwaha, Fuya Makino, Rajkumar Modak, [Yuya Sakuraba](https://orcid.org/0000-0003-4618-9550), [Tadakatsu Ohkubo](https://orcid.org/0000-0003-3548-1951), Rulei Guo, Bin Xu, [Junichiro Shiomi](https://orcid.org/0000-0002-3552-4555), [Daichi Chiba](https://orcid.org/0000-0002-6631-5131), [Ken-ichi Uchida](https://orcid.org/0000-0001-7680-3051)

## Rights

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

## Other metadata

[Non-Equilibrium Magnon Engineering Enabling Significant Thermal Transport Modulation](https://mdr.nims.go.jp/datasets/30a33da6-4746-4b8b-b5dc-ca37e3c0da85)

## Fulltext

Non‐Equilibrium Magnon Engineering Enabling Significant Thermal Transport ModulationRESEARCH ARTICLEwww.afm-journal.deNon-Equilibrium Magnon Engineering Enabling SignificantThermal Transport ModulationTakamasa Hirai,* Toshiaki Morita, Subrata Biswas, Jun Uzuhashi, Takashi Yagi,Yuichiro Yamashita, Varun Kumar Kushwaha, Fuya Makino, Rajkumar Modak,Yuya Sakuraba, Tadakatsu Ohkubo, Rulei Guo, Bin Xu, Junichiro Shiomi, Daichi Chiba,and Ken-ichi Uchida*Thermal conductivity, a fundamental parameter characterizing thermaltransport in solids, is typically determined by electron and phonontransport. Although electrical transport properties are material-specific,recent advance in understanding phonon transport has led to new insightson controlling the thermal conductivity via phonon engineering techniques.Here, an unconventional way of artificially modulating the heat conductionin solids is demonstrated via engineering of magnon transport. Time-domainthermoreflectance measurements in ferromagnetic metal/insulator junctionsystems reveal that the thermal conductivity of the ferromagnetic metals andinterfacial thermal conductance vary substantially depending on the spatialdistribution of non-equilibrium spin currents. Systematic measurements of thethermal transport properties with changing the boundary conditions for spincurrents unveil that magnons significantly modulate the heat conduction by≈10 Wm−1 K−1 even in ferromagnetic metals at room temperature, upsettingthe conventional wisdom that the magnon thermal conductivity is verysmall in metals except at low temperatures. This magnon-engineered thermaltransport offers a new principle and manner for active thermal management.T. Hirai, T. Morita, S. Biswas, J. Uzuhashi, V. K. Kushwaha, F. Makino,R. Modak, Y. Sakuraba, T. Ohkubo, K. UchidaNational Institute for Materials Science (NIMS)Tsukuba, Ibaraki 305-0047, JapanE-mail: HIRAI.Takamasa@nims.go.jp;UCHIDA.Kenichi@nims.go.jpT.Morita,D. ChibaSANKENOsakaUniversityOsaka, Ibaraki 567-0047, JapanS. BiswasDepartment of PhysicsIndian Institute of TechnologyGuwahatiGuwahati 781039, IndiaThe ORCID identification number(s) for the author(s) of this articlecan be found under https://doi.org/10.1002/adfm.202506554© 2025 The Author(s). Advanced Functional Materials published byWiley-VCH GmbH. This is an open access article under the terms of theCreative Commons Attribution License, which permits use, distributionand reproduction in any medium, provided the original work is properlycited.DOI: 10.1002/adfm.2025065541. IntroductionDevelopments in understanding and con-trolling thermal transport are critical forthermal management in densely packed,high performance electronic devices.[1,2] Amaterial showing high thermal conductiv-ity, 𝜅, is desirable for heat dissipation andexchange, while a material showing low𝜅 for thermal insulation and thermoelec-tric conversion. However, as device dimen-sions shrinks, materials’ 𝜅 alters, deviatingfrommacroscale behavior, due to more pro-nounced scattering processes of multipleheat carriers, e.g., electrons and phonons,at micro/nanoscale. The presence of multi-ple heat carriers imposes another thermaltransport problem: interfacial thermal re-sistance, R, that leads to large bottlenecksto heat flow.[3] In metal/insulator junctions,commonly installed in many electronic de-vices, since the heat conduction in metalsT. Yagi, Y. YamashitaNational Institute of Advanced Industrial Science and Technology (AIST)Tsukuba, Ibaraki 305-8563, JapanF. Makino, Y. Sakuraba, K. UchidaGraduate School of Science and TechnologyUniversity of TsukubaTsukuba, Ibaraki 305-8577, JapanR. Modak, K. UchidaDepartment of Advanced Materials ScienceGraduate School of Frontier SciencesThe University of TokyoKashiwa, Chiba 277-8561, JapanR. Guo, B. Xu, J. ShiomiDepartment of Mechanical Engineering, Graduate School of EngineeringThe University of TokyoBunkyo, Tokyo 113-8656, JapanD. ChibaCenter for Spintronics Research NetworkOsaka UniversityToyonaka, Osaka 560-8531, JapanAdv. Funct. Mater. 2025, 35, 2506554 2506554 (1 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbHhttp://www.afm-journal.demailto:HIRAI.Takamasa@nims.go.jpmailto:UCHIDA.Kenichi@nims.go.jphttps://doi.org/10.1002/adfm.202506554http://creativecommons.org/licenses/by/4.0/http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadfm.202506554&domain=pdf&date_stamp=2025-10-01www.advancedsciencenews.com www.afm-journal.de(insulators) is typically dominated by electrons (phonons), thelargeR inhibits the heat flow across the heterojunction, accompa-nying with a finite temperature drop at the interface via electron-phonon coupling, even when using metals showing high 𝜅.To fully realize the multiscale thermal management depend-ing on intended applications, technologies tailoring both 𝜅and interfacial thermal conductance G (= 1/R) by consider-ing mean free paths (MFPs) of different heat carriers need tobe developed. While electron contribution to 𝜅 is connected toelectrical conductivity as per the Wiedemann-Franz law, withtheir MFP limited to a few nanometers, significant modula-tion of electron 𝜅 is challenging without resorting to phasetransitions or carrier doping.[4,5] In contrast, since MFP ofphonons is much longer than that of electrons and variousphonon modes with different MFPs contribute to heat con-duction, the micro/nanoscale thermal transport properties canbe dramatically modulated without changing electrical conduc-tivity by selectively controlling phonon scattering or phonon-phonon and electron-phonon interactions.[6,7] Various strategiesfor phonon-engineered thermal transport have been proposedso far, e.g., reducing 𝜅 and/or G by tuning nano-structuralproperties[8–11] and improving 𝜅 and/or G through electricallymodulating chemical bonding,[12] lattice strain,[13] and electricalpolarization.[14]Recently, advancements in the field of spin caloritronics haveprovided novel thermal management principles and function-alities by introducing the spin degree of freedom into ther-mal transport and thermoelectric conversion.[15–17] While muchfocus has been on magneto-thermoelectric and thermo-spinconversion in spin caloritronics,[17–20] several studies have re-ported the manipulation of 𝜅 in magnetic metal films ormultilayers through magnetic field application.[21–25] Consid-ering heat conduction dominated by electrons in metals, themagnetic-field-induced 𝜅 change is attributed to the thermalanalog of magnetoresistance effects, i.e., the change in elec-trical conductivity depending on a magnetization configura-tion. Since phonon thermal conductivity is magnetic-field- orspin-independent, the change ratio of 𝜅 should be compara-ble to or less than that of the electrical conductivity as per theWiedemann-Franz law and spin-dependent electron transporttheories.[22,26] Nevertheless, a few reports show a change in 𝜅enough to breakdown the Wiedemann-Franz law at and aboveroom temperature,[24,25] hinting at additional magnetic-field- orspin-dependent heat conduction mechanisms beyond conven-tional electron contribution. A potential mechanism is the con-tribution of magnons, quanta of collective motion of magneticmoments, to the heat conduction; magnons are known to carrynot only spin angular momentum but also heat energy,[27] andD. ChibaDivision of Spintronics Research NetworkInstitute for Open and Transdisciplinary Research InitiativesOsaka UniversitySuita, Osaka 565-0871, JapanD. ChibaInternational Center for Synchrotron Radiation Innovation SmartTohoku UniversitySendai, Miyagi 980-8577, Japanmagnon-driven 𝜅 change in a magnetic insulator (MI) was the-oretically predicted.[28] Although the recent theoretical simula-tion predicts that magnons contribute to heat conduction evenin a ferromagnetic metal (FM) at room temperature,[29] the di-rect experimental observation of magnon thermal conductiv-ity in FM has been proved only by freezing out the magnonexcitation with high magnetic fields at very low temperature(< 5 K).[30,31]In this study, we demonstrate that heat conduction in FM thinfilms and at FM/insulator interfaces can be significantly modu-lated by tuning the spatial distribution of non-equilibrium spincurrents even at room temperature (Figure 1a,b). Inspired bythe spin transport theory on magnetic metal/insulator junctions,we design the experimental system to elucidate spin-current-induced thermal transport engineering. Systematic investiga-tions of thermal transport properties for various materials us-ing an ultrafast optical pump-probe technique obtain the sur-prising evidence that magnons carry substantial heat even atroom temperature in FM, enabling the engineering of not only𝜅 of FM, 𝜅FM, but also G at the FM/insulator interface. Our ex-periment reports a groundbreaking approach in understandingheat conduction by magnon at various temperatures, even underlowmagnetic fields. The functionality of themagnon-engineeredthermal transport paves the way for thermal engineering re-search and promotes spintronic thermal management[32] forapplications.2. Results and Discussion2.1. Phenomenological Prediction of Thermal TransportEngineering by Non-Equilibrium Spin CurrentTo discuss the modulation of heat conduction by spin currents,we examine a transport of thermally excited spin currents intwo types of FM/insulator junctions: FM/non-magnetic insula-tor (NI) and FM/MI junctions. When a temperature gradient isapplied across the FM layer and the junction interface, a non-equilibrium spin current is excited in the FM layer. In FM/NIjunctions, the spin current flowing in the thickness directionmust dissipate at the interface because the spin current is notinjected into NI. In contrast, in FM/MI junctions, the spin cur-rent can transmit to the MI interface via the conversion of thespin current in FM into the magnon spin current in MI despitethe absence of conduction electrons in MI.[33] The difference inthe boundary conditions for spin currents can induce differentspin-current and spin-accumulation distributions following thespin diffusion equation.[34–36] Figure 2 presents schematics ofthe spatial profiles of the normalized spin current density jsnand corresponding spin chemical potential µsn in FM/NI andFM/MI junction systems. This assumes a 1D system along thethickness direction with continuous spin currents at the FM/MIinterface. Such a spin transport is governed by a spin diffu-sion length, a characteristic length in which spin currents andspin accumulation persist. As shown in Figure 2a,b, the totalamount of spin currents in the FM layer adjacent to MI be-comes greater than that on NI, attributed to the disappearanceof spin currents at the FM/NI interface. If these spin currentsconvey heat, 𝜅FM on MI should be larger than that on NI. How-ever, when the spin diffusion length of FM 𝜆FM is much smallerAdv. Funct. Mater. 2025, 35, 2506554 2506554 (2 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 1. Thermal transport engineering by non-equilibrium spin currents. a,b) Schematic of magnon-engineered thermal transport in ferromag-netic metal (FM)/insulator junction structures. Thermal conductivity 𝜅 of FM in FM/non-magnetic insulator (NI) system gets smaller than that inFM/magnetic insulator (MI) system due to disappearance (a) or transmission (b) of the spin current Js at the FM/insulator interface.than the thickness of FM tFM, this effect should be minimal.In contrast, when 𝜆FM is larger than tFM, a noticeable differencein 𝜅FM between FM/MI and FM/NI systems emerges, owing tothe stark reduction in the effective population of spins in the FMlayer attached to NI (Figure 2c,d). Following this scenario, themagnitude ofG at the FM/insulator interface can also be affectedby the type of attached insulator as it alters the transmission ofspin currents and the resultant heat transfer across the interface.Although the above-mentioned spin currents in FM may com-prise both conduction-electron and magnon spin currents, their𝜆FM values are quite different; typically, 𝜆FM for the conduction-electron (magnon) spin current is a few nm (>100 nm).[37–41]Consequently, the change in 𝜅FM of 100-nm-order thick FMfilms depending on the spin-current distribution is expectedto be minor for conduction-electrons’ spins but significant formagnons, reflecting their dominant contribution to that thermaltransport.2.2. Experimental Details of Time-Domain ThermoreflectanceFor the proof-of-concept demonstration of the spin-current-induced thermal transport engineering, we measured the cross-plane 𝜅FM and G at the FM/NI and FM/MI interfaces by meansof the time-domain thermoreflectance (TDTR) method, an opti-cal pump-probe technique, in a front-heating and front-detectionFigure 2. Distributions of spin current and spin chemical potential. a–d)Schematics of spatial profiles of the normalized spin current density jsnand corresponding spin chemical potential µsn following spin diffusionequation in FM/NI and FM/MI junction systems, where the spin diffusionlength of the FM layer 𝜆FM is smaller (larger) than the thickness of the FMlayer tFM in (a,b) and (c,d). Orange arrows represent heat currents.configuration (Figure 3a). The irradiation of ultrafast pump laserpulses heats up the surface of a metallic transducer deposited ona target thin film and that of probe laser pulses with varying atime delay detects the transient response of surface temperaturevia thermoreflectance, i.e., temperature dependence of reflectiv-ity, enabling the determination of thermal transport propertiesof thin films and their interfaces.[42–45] Since the TDTRmeasure-ments do not require any microfabrication processes and electri-cal contacts, TDTR facilitates high-throughput and reliable ther-mal transport measurements on a single wafer with a wedgestructure by merely shifting the position of laser spots.[46,47] Forsystematic and quantitative investigations, we mainly measuredthermal transport properties in a double-wedge structure com-prising a thickness-wedged garnet substrate and composition-spread FM thin film. Here, paramagnetic Gd3Ga5O12 (GGG) andferrimagnetic Y3Fe5O12 (YIG) were selected asNI andMI, respec-tively, because YIG is a prominent MI for spin current physics inspintronics[48,49] and spin caloritronics[50,51] and GGG serves asa standard reference insulator for YIG owing to the same crystalstructure as and the close lattice constant to YIG. By preparinga thickness-wedged-YIG/GGG substrate, we compared TDTRsignals of the FM/NI and FM/MI systems using a single sample(Figure 3a,b). As one of the FM layers, we used a CoFe thin filmwith a gradient in the Co atomic content ratio q along the film-plane direction (y direction) perpendicular to the YIG-thickness-gradient direction (x direction) to investigate the FM composi-tion dependence of 𝜅FM and G (Figure 3b). The top surface ofthe sample was covered by an Al transducer, chosen for its well-known thermoreflectance coefficient.[52] We found that the lay-ered structure of Al/CoFe/YIG was free from atomic diffusionand that the crystallinity and in-plane electrical conductivity didnot depend on the attached insulators, confirming similar filmquality of CoFe on both GGG and YIG (see Experimental Sectionand Figures S1 and S2, Supporting Information, for the detailsof structural and electrical transport measurements). All TDTRmeasurements were performed at room temperature and in anair atmosphere while applying a magnetic field (= 50 mT alongthe film plane unless otherwise specified; also see ExperimentalSection) to assist the alignment of the magnetization direction.Here, the magnetization of CoFe and YIG was parallelly alignedvia interfacial exchange interaction (Note S1 and Figure S3,Supporting Information). In our TDTR experiments, we mea-sured the transient response of heat diffusion from the samplesurface to the substrate since the heat diffusion length in FM in-duced by laser heating ismuch larger than tFM (Experimental Sec-tion and Figure S4, Supporting Information).Adv. Funct. Mater. 2025, 35, 2506554 2506554 (3 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 3. TDTR signals and thermal transport properties in CoFe/insulator junction structure. a) Schematic of the time-domain thermoreflectance(TDTR) measurement for double-wedge FM/insulator structure. Al, CoFe, Gd3Ga5O12 (GGG), and Y3Fe5O12 (YIG) were used as transducer, FM, NI,and MI, respectively, where the YIG thickness tYIG was linearly wedged along the x direction and content of Co q in CoFe was gradually changed from0 to 1 along the y direction. By changing the x (y) directional position of laser spots, TDTR signals of CoFe on different insulators (with different CoFecompositions) can be obtained. During TDTR measurements, a magnetic field H with a magnitude of 50 mT was applied along the x direction. b)Schematic of the sample from a top view and profiles of measured tYIG values along the x direction and of Co and Fe contents along the y direction(see also Experimental Section for details). Black arrows in the tYIG profile indicate positions at which TDTR measurements were performed for GGG (x= 8.5 mm) and YIG (x = 1.5 mm). c) Temporal response of 𝜑TR for CoFe with tFM = 100 nm and q = 0.58 (Co58Fe42) on GGG and YIG. d,e) Thermalconductivity of Co58Fe42 𝜅FM with tFM = 100 nm (d) and interfacial thermal conductance G at Co58Fe42/GGG and Co58Fe42/YIG interfaces (e).2.3. TDTR Measurements in Double-Wedge CoFe/GarnetSubstrateFigure 3c presents the thermoreflectance signal 𝜑TR for CoqFe1-qwith q = 0.58 and tFM = 100 nm on GGG (gray symbols) and YIGwith a thickness tYIG of 46 μm (green symbols) as a function of thedelay time between pump and probe laser pulses. The 𝜑TR datafor CoFe/YIG lay below that for CoFe/GGG, suggesting fasterheat diffusion in the CoFe/YIG junction than in the CoFe/GGGjunction. In contrast, in the absence of the CoFe layer, the TDTRsignals were nearly identical for the GGG and YIG regions, in-dicating minimal difference in thermal transport properties ofGGG and YIG (Figure S5, Supporting Information). This resultconfirms that the observed change in thermal transport is signifi-cantly influenced by the CoFe layer and its interface. Note that thesame TDTR signals were observed when applying out-of-planemagnetic field or in the absence of magnetic field (Figure S6,Supporting Information). The results at different tYIG values areshown in Note S2 and Figure S7 (Supporting Information). Theanalysis of the TDTR signals based on a heat diffusion modelallows estimation of the 𝜅FM values on GGG and YIG (𝜅FMGand 𝜅FMY, respectively), as well as the G values at CoFe/GGGand CoFe/YIG (GFM/GGG and GFM/YIG, respectively) (see the fit-ting curves shown as solid curves in Figure 3c and Experimen-tal Section for details). Figure 3d shows the 𝜅FMG and 𝜅FMY val-ues for Co58Fe42, revealing that 𝜅FMY (= 76 ± 5 Wm−1 K−1) waslarger than 𝜅FMG (= 63 ± 4 Wm−1 K−1). This behavior cannot beexplained only by electrons’ contribution based on Wiedemann-Franz law (Figure S2, Supporting Information) and is qualita-tively consistent with our expectation (Figure 2). Surprisingly,the difference in 𝜅FM between the YIG and GGG regions, Δ𝜅FM= 𝜅FMY − 𝜅FMG, was comparable to changes due to the giantmagneto-thermal resistance effect in spintronic multilayers[23]and the change ratio Δ𝜅FM/𝜅FMG was estimated to be ≈22%.Additionally, GFM/YIG was found to be several times larger thanGFM/GGG (Figure 3e), supporting the existence of heat conduc-tion concomitant with spin currents in the FM/MI junction. Thisresult further corroborates the validity of our phenomenologicalprediction. Note that the GFM/GGG value (≈0.2 × 109 Wm−2 K−1)was comparable to G values at conventional metal/insulatorinterfaces[53] and that the 𝜅FMG value was independent of tFM(Figure S8, Supporting Information), which suggests the sizeeffect of phonon heat conduction is negligible in the presentsample.We next focus on the TDTR measurements along the ydirection to investigate the CoFe composition dependence.Figure 4a shows the q dependence of 𝜅FMG and 𝜅FMY (seeFigure S9, Supporting Information, for the temporal responseof 𝜑TR for various q values). The 𝜅FM value of CoFe exhibitscomposition-dependent behavior; it tends to be lower than thatAdv. Funct. Mater. 2025, 35, 2506554 2506554 (4 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 4. a–d) q dependence of 𝜅FM (a), change in 𝜅FM, Δ𝜅FM = 𝜅FMY − 𝜅FMG, and change ratio, Δ𝜅FM/𝜅FMG, with 𝜅FMG(Y) being the magnitudeof 𝜅FM on GGG (YIG) (b), G at FM/GGG and YIG interfaces (c), and change in G, ΔG = GFM/YIG − GFM/GGG, and change ratio, ΔG/GFM/GGG, withGFM/GGG(YIG) being the magnitude of G at the FM/GGG (FM/YIG) interface (d).of pure Fe (q = 0) and Co (q = 1) when the Co content isq < 0.3 or 0.7 < q, but higher in the intermediate compositionrange [note that a similar tendency has been observed for theelectrical conductivity (Figure S2, Supporting Information) andfor 𝜅FM in a bulk Co50Fe50 alloy.[24]] About the difference be-tween 𝜅FMG and 𝜅FMY, similar trends to the results for Co58Fe42,i.e., revealing 𝜅FMY > 𝜅FMG, were observed in whole q range,though the differences for q ≈ 0 and 1 were less pronounced.The dome-shaped q dependence of Δ𝜅FM and Δ𝜅FM/𝜅FMG is de-picted in Figure 4b, peaking at ≈10 Wm−1 K−1 and ≈20% for0.2 < q < 0.7 and diminishing with further increasing or de-creasing q. We show the q dependence ofGFM/GGG andGFM/YIG inFigure 4c and the corresponding changeΔG=GFM/YIG −GFM/GGGand change ratio ΔG/GFM/GGG in Figure 4d. While the GFM/GGGvalues of 0.1–0.3 × 109 Wm−2 K−1 were similarly observed for allcompositions, the GFM/YIG values were enhanced by 200–400%from those of GFM/GGG. These results robustly sustain the exper-imental demonstration of substantial modulation of nanoscaleand interfacial thermal transport in FM/insulator structures atroom temperature by engineering the boundary condition forspin currents.2.4. Conduction-Electron Spin Current versus Magnon SpinCurrentThe remaining question is which conduction-electrons ormagnons dominantly contribute to spin-current-induced ther-mal transport engineering in the present FM/insulator systems.The difference in 𝜆FM of conduction-electron and magnon spincurrents (hereinafter referred to as 𝜆FMe and 𝜆FMm, respectively)provides a useful hint for clarifying the origin. First, the 𝜆FMevalues of Co, Fe, and CoFe are at most a few or several tensof nanometers (Table S1, Supporting Information), which is toosmall to explain the large change in 𝜅FM in 100-nm-order thickFMs (Figure 2a,b). In addition, the values of 𝜆FMe are inconsis-tent with the q dependence of Δ𝜅FM/𝜅FMG; the change in 𝜅FMshould be smaller for samples with smaller 𝜆FMe, but the exper-imental results sometimes show the opposite tendency (e.g., forFe and Co60Fe40). The small 𝜆FMe values also exclude the pos-sibility of additional contributions coming from the injectionof conduction-electron spin currents into FM from YIG due tothe spin Seebeck effect (SSE).[50] Subsequently, we focus on themagnon contribution. To discuss 𝜆FMm, it is necessary to deter-mine the energy scale of themagnons of interests. In typical MIs,such as YIG, a temperature gradient induces the flow of incoher-ent non-equilibrium high-energy magnons with a high magnonfrequency (∼THz).[54,55] MFPs of such high-energy magnons areshorter than those of coherent low-energy magnons excited bymicrowaveswithGHz-order frequencies. Thus, the value of 𝜆FMmof high-energy magnons contributing to thermal transport inCoFe should be smaller than the magnon diffusion length es-timated by microwave experiments (≈5–20 μm);[40,41] however,this fact alone still does not show whether 𝜆FMm of high-energymagnons is larger than 𝜆FMe. Then, let us consider the case ofSSE and its reciprocal called the spin Peltier effect (SPE),[51] inwhich high-energy sub-thermal magnons primarily contribute tothe conversion between heat and spin currents. In SSE/SPE, thelength scale of magnon transport reaches the order of several mi-crometers for MIs[56] and several tens of nanometers for FMs,[57]which is much larger than 𝜆FMe. Therefore, although 𝜆FMm ofhigh-energy magnons for heat conduction has not been reporteddirectly, it is reasonable that it becomes larger than 𝜆FMe in a sim-ilar manner to SSE/SPE. Considering the fact that 𝜆FMe is toosmall to explain our results, we can conclude that the magnonsdominantly contribute that spin-current-induced thermal trans-port modulation.To further verify the magnon contribution, we performed theTDTRmeasurements in a Ni80Fe20/wedged-YIG system as a con-trol experiment. Given the notably small 𝜆FMe of NiFe alloys (≈3–5 nm; Table S1, Supporting Information), the contribution ofconduction-electron spin currents in Ni80Fe20/YIG is expected tobe less than in CoFe/YIG. As shown in Figure 5a,b, the distinctchange in𝜑TR similar to the case of CoFewas obtained inNi80Fe20with tFM = 100 nm, resulting in the Δ𝜅FM/𝜅FMG of ≈25% andΔGFM/sub./GFM/GGG of ≈500% (see also Figure S10, SupportingInformation, for the details of structural characterization). Im-portantly, by inserting a 10-nm-thick or 5-nm-thick Cu layer be-tween the Ni80Fe20 and insulator layers, the difference in the tem-poral response of𝜑TR was visibly decreased, and the difference in𝜅FM and G disappeared (Figure 5c,d; see Figure S11, SupportingInformation, for the results of the film with the 5-nm-thick CuAdv. Funct. Mater. 2025, 35, 2506554 2506554 (5 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.deFigure 5. a,c) Temporal response of 𝜑TR for Ni80Fe20 (a) and Ni80Fe20/Cu(c) films with a Ni80Fe20 (Cu) thickness of 100 nm (10 nm) on GGG andYIG. b,d) 𝜅FM (upper panel) and G (lower panel) for the Ni80Fe20 (b) andNi80Fe20/Cu (d) films on GGG and YIG.insertion layer). Because of the absence of magnons in the Culayer, the disappearance of the substrate dependence of 𝜅FM andG supports our interpretation that the spin-current-induced ther-mal transport engineering is of magnon origin.Finally, we discuss the magnitude ofΔ𝜅FM. Assuming that theΔ𝜅FM values (≈10 Wm−1 K−1) observed here originate from themodulation of the magnon thermal conductivity of FM, one cansee that this change is quite large because it surpasses 𝜅 of YIG,a proficient magnon conductor. This leads to additional possibili-ties ofmagnon-related 𝜅modulationmechanisms, e.g., magnon-electron and magnon-phonon interactions, akin to the magnon-drag effect in thermoelectric conversion.[58,59] For further clari-fication of microscopic mechanisms of thermal transport engi-neering, including the obvious composition dependence ofΔ𝜅FMand ΔG, the low-temperature TDTR measurements, which caneasily suppress thermal magnon excitation with a strong mag-netic field, would be insightful; such experiments should revealthe energy and length scales of magnons instrumental in ther-mal transport. Nevertheless, our measurement method uniquelyallows for the assessment of magnon-driven thermal transportproperties at room temperature, making it pertinent for investi-gating the applicability of magnon engineering in thermal man-agement technologies.3. ConclusionIn summary, we experimentally demonstrated the engineeringof nanoscale thermal transport properties in FM/insulator junc-tion systems at room temperature by controlling the boundarycondition for spins. Using the TDTR technique, we found thatthe interfacial transmission of thermally excited spin currentsnot only enhanced 𝜅FM by several tens of percent but also sig-nificantly improved G compared to conventional non-magneticmetal/insulator interfaces. The observed 𝜅FM and G engineeringshows the significant contribution of magnons in the heat con-duction even at room temperature. The concept demonstrated inthis study has broader application beyond a single interface; itcould be extended to multilayers and nanocomposites, enablingthe thermal transport controlling through magnon engineeringeven inmacroscale materials. Our findings will thus provide newengineering opportunities for active thermal management basedon spintronics.4. Experimental SectionPreparation of Thickness-Wedged-YIG/GGG Substrate: Single-crystalline YIG (111) with a thickness of 75 μm was grown on asingle-crystalline GGG (111) substrate with a thickness of 0.5 mm usinga liquid epitaxy method, where a portion of Y in YIG was substituted withBi to improve lattice matching between GGG and YIG; the actual com-position of YIG was Bi0.1Y2.9Fe5O12. The thickness-wedged-YIG/GGGsubstrates were prepared by obliquely polishing uniform YIG/GGGsubstrates with SiC sandpapers and alumina slurry. The range of the YIGthickness after polishing was estimated to be 0–50 μm (Figure 3b).Thin Film Deposition: The thin films were deposited on the wedged-YIG/GGG substrate using a magnetron sputtering system (CMS-A6250 × 2, Comet, Inc.) at an Ar gas pressure of 0.6 Pa and room tem-perature. Before the deposition, the substrates were flushed at 500 °Cfor 30 min to clean the surface. The CoqFe1-q composition-spread filmswere prepared by means of a layer-by-layer wedge-shaped depositiontechnique,[47,60] consisting the repetition of the following three steps:1) deposition of a wedge-shaped Co layer over a length of 7.0 mm or-thogonal to the YIG wedge using a linear moving shutter, 2) rotation ofthe substrate by 180° in the in-plane direction, and 3) deposition of awedge-shape Fe layer on the same area of (1), where the thickness ofthe CoqFe1-q films after completing one (1)–(3) cycle was designed tobe 0.5 nm. this sequence was repeated 200 times to fabricate the 100-nm-thick CoqFe1-q composition-spread film. 80, 150, and 200-nm-thickCoqFe1-q composition-spread films on GGG were also prepared to investi-gate the size effect on 𝜅FM of CoFe. Although the films were fabricated bylayer-by-layer deposition, the CoqFe1-q films were formed with uniformlymixing Co and Fe atoms (Figure S1, Supporting Information). In additionto the CoFe composition-spread films, uniform Ni80Fe20 (100 nm) andNi80Fe20 (100 nm)/Cu (10 and 5 nm) films were fabricated on the wedged-YIG/GGG substrate. The samples for the TDTR (electrical conductivity)measurements were covered with 48-nm-thick (5-nm-thick) Al layer with-out breaking the vacuum. An Al (48 nm)/wedged-YIG/GGG sample with-out the FM layer was also prepared specifically for TDTR measurements.SEM and STEM-EDS Measurements: The thickness gradient of theYIG layer was measured by cross sectional scanning electron microscopy(SEM) observation using a focused-ion-beam (FIB)-SEM dual beam sys-tem (CarlZeiss, CrossBeam550). The structural and chemical characteriza-tion were carried out using high-angle annular dark field (HAADF) scan-ning transmission electron microscopy (STEM) with energy-dispersive X-ray spectroscopy (EDS) (Thermo Fisher Scientific, Titan G2 80–200 andSpectra Ultra) using the CoFe film with tFM = 200 nm and NiFe film withtFM = 100 nm. A different FIB-SEM system (Thermo Fisher Scientific,Helios G4 UX DualBeam) was used to prepare the STEM specimens byAdv. Funct. Mater. 2025, 35, 2506554 2506554 (6 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.destandard lift-out techniques. The STEM specimens were thinned down to25 nm based on backscattered electrons intensity.[61] The results are dis-played in Figures S1 and S10 (Supporting Information).XRDMeasurement: The crystallographic investigation was conductedby the X-ray diffraction (XRD) with Cu-K𝛼 radiation (Rigaku, SmartLab).The X-ray was incident on the CoFe sample with tFM = 100 nm through alength limit slit of 0.5 mm in the direction perpendicular to the YIG thick-ness gradient. The irradiation center of X-ray was positioned at (x, y)= (8.5mm, 5.0 mm) for CoFe on GGG and (1.5 mm, 5.0 mm) for CoFe on YIG,as indicated in Figure 3b. The result of the XRD measurement is shown inFigure S1 (Supporting Information).Electrical Conductivity Measurement: To measure the q dependence ofthe in-plane electrical conductivity of the CoFe film 𝜎FMip, the sampleswere cut out into a rectangular strip with a width of 1 mm along the direc-tion parallel to the CoFe composition gradient from both sides of the GGGand YIG edges using a wire saw. The magnitude of 𝜎FMip was quantifiedusing a standard four-probe method, where the charge current of 1 mAwas applied to the film through indium contacts bonded on the ends ofthe strip and the voltage was measured using a probe unit with a distanceof 100 μm between probes. The 𝜎FMip value for each q was ascertained byadjusting the probe position along the longitudinal direction of the stripusing a micrometer.TDTR Measurement and Analysis: The schematics of the TDTR sys-tem is shown in Figure S4 (Supporting Information). The system utilizes amode-lock Er-doped fiber laser as both the pump and probe laser sources,which generates a train of pulses at a repetition rate of ≈20 MHz with apulse width of ≈0.5 ps and a central wavelength at 1550 nm. The pumplaser beam was modulated by a lithium niobate modulator chopping thepulses with a rectangular wave modulation at fmod = 200 kHz. Meanwhile,the probe laser beam was passed through a second-harmonic generator,altering its central wavelength to 775 nm. The laser power and 1/e2 spotdiameter were set at 20 mW (< 1 mW) and 90 μm (30 μm) for the pump(probe) laser beam, respectively. Given that the diameters of the pump andprobe laser spots were significantly larger than the heat diffusion length√Dsub∕𝜋fmod ≈2 μm, whereDsub is the thermal diffusivity of GGG or YIG,1D heat diffusion could be assumed in this experiment. The steady-statetemperature rise at the surface of samples[42] was <1 K. The TDTR systemwas equipped with an electrical delay control system in which the oscilla-tions of two lasers were synchronized to the electrical signal from a func-tion generator, allowing for a delay between pump and probe laser pulses𝜏d with 0–50 ns without requiring a mechanical delay stage.[43,45,47,62] Thepenetration depth of thermal diffusion in FM was roughly estimated as√DFM𝜏d, with DFM being the thermal diffusivity of FM. Considering theDFM values of CoFe (NiFe) is ≈2 × 10−5 m2 s−1 (≈1 × 10−5 m2 s−1), themagnitude of√DFM𝜏d reaches ≈1.0 μm (0.7 μm) for CoFe (NiFe) at 𝜏d =50 ns, which was substantially greater than the FM thickness used in thisstudy. To measure thermoreflectance signals, the modulated pump laserbeam was irradiated onto the sample surface, i.e., the Al surface, and theprobe laser beam was focused on the same spot with a specific time delay.By detecting the reflected probe laser beam using a Si adjustable balancedphotoreceiver connected to a lock-in amplifier, the lock-in amplitude andphase components of the thermoreflectance signals synchronized with themodulation frequency at the time delay were detected. Through the useof both the lock-in detection technique and a resonant filter between thephotodiode and the lock-in amplifier, responses to higher harmonics ofthe rectangular-wave-modulated pump laser pulses and dc offset were ef-fectively removed. Note that the variation of q within the laser spot sizewas estimated to be ≈0.4 at.% by taking into account the magnitude ofthe composition gradient in the CoqFe1-q film, which was negligibly small.During the TDTRmeasurements,Hwas applied along the in-plane (out-of-plane) direction using an electromagnet (by placing a permanent magnetbehind the sample).The values of 𝜅FM and G were determined through numerical simula-tions based on 1D heat diffusion model.[63] For the analysis of TDTR data,the lock-in phase 𝜑TR was used, which corresponds to the ratio betweenthe in-phase and out-of-phase components of lock-in signals.[42,64] The𝜑TR data was fitted at the delay time of >100 ps, where the electrons andphononswere thermalized in the Al layer. For TDTR analyses, some param-eters were usually set as the fixed parameters on the heat diffusion model.In this study, the bulk values of the volumetric heat capacity 𝜌C with 𝜌 andC respectively being the density and specific heat were used for Al, Co,Fe, and Cu and the 𝜌C value of CoqFe1-q (Ni80Fe20) was calculated to bethe weighted average of the bulk values of Co (Ni) and Fe by consideringthe compositionmeasured by STEM-EDS. The thicknesses estimated fromthe STEM measurement were used for each layer in the thermal model.The thermal effusivity e (=√𝜅𝜌C = 𝜌C√Dsub) of GGG and YIG was esti-mated to be 4.0 × 103 and 3.8 × 103 Jm−2 s−0.5 K−1, respectively, based onthe TDTR experiment using the Al/wedged-YIG/GGG sample. To double-check the accuracy, the magnitude of e of GGGwas also verified by individ-ually measuringDsub using the laser flash method, C using the differentialscanning calorimetry, and 𝜌 using the Archimedesmethod for a plain GGG(111) substrate. The estimated value of 3.96 ± 0.08 × 103 Jm−2 s−0.5 K−1showed good consistency for GGG. The error bar for 𝜅FM (G) was es-timated by calculating 𝜑TR with deviation of DFM and C (R) of FM ac-cording to rectangular distribution in statistics Figure S12, (SupportingInformation). It was confirmed that the model has proper sensitivityto evaluate DFM by the sensitivity calculations Figure S13, (SupportingInformation).Supporting InformationSupporting Information is available from the Wiley Online Library or fromthe author.AcknowledgementsT.H. and T.M. contributed equally to this work. The authors thank G. E. W.Bauer, R. Iguchi, T. Koyama, A. Srinivansan, and P. Alagarsamy for valuablediscussions and T. Hiroto, K. Suzuki, R. Toyama, V. Barwel, M. Isomura,A. Takahagi, K. Takamori, and R. Nagasawa for technical support. Thiswork was partially supported by CREST “Creation of Innovative Core Tech-nologies for Nano-enabled Thermal Management” (No. JPMJCR17I1) andERATO “Magnetic Thermal Management Materials Project” (No. JPM-JER2201) from JST, Japan; Grant-in-Aid for Research Activity Start-up (No.22K20495) and Grant-in-Aid for Scientific Research (S) (No. 22H04965)from JSPS KAKENHI, Japan; and NEC Corporation. A part of this workwas supported by the Electron Microscopy Unit, National Institute forMaterials Science (NIMS). T.H. acknowledged support from the Thermaland Electric Energy Technology Foundation. T.M. was supported by Pro-gram for Leading Graduate Schools “Interactive Materials Science CadetProgram”.Conflict of InterestThe authors declare no conflict of interestData Availability StatementThe data that support the findings of this study are available from thecorresponding author upon reasonable request.Keywordsinterfacial thermal conductance, magnon, spin current, thermal conduc-tivity, time-domain thermoreflectanceReceived: March 13, 2025Revised: May 23, 2025Published online: October 1, 2025Adv. Funct. Mater. 2025, 35, 2506554 2506554 (7 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.dewww.advancedsciencenews.com www.afm-journal.de[1] P. Ball, Nature 2012, 492, 174.[2] D. G. Cahill, P. V. Braun, G. Chen, D. R. Clarke, S. Fan, K. E. Goodson,P. Keblinski, W. P. King, G. D. Mahan, A. Majumdar, H. J. Maris, S. R.Phillpot, E. Pop, L. Shi, Appl. Phys. Rev. 2014, 1, 011305.[3] J. Chen, X. Xu, J. Zhou, B. Li, Rev. Mod. Phys. 2022, 94, 025002.[4] D. W. Oh, C. Ko, S. Ramanathan, D. G. Cahill, Appl. Phys. Lett. 2010,96, 151906.[5] Q. Yang, H. J. Cho, Z. Bian, M. Yoshimura, J. Lee, H. Jeen, J. Lin,J. Wei, B. Feng, Y. Ikuhara, H. Ohta, Adv. Funct. Mater. 2023, 33,2214939.[6] M. Nomura, J. Shiomi, T. Shiga, R. Anufriev, Jpn. J. Appl. Phys. 2018,57, 080101.[7] X. Qian, J. Zhou, G. Chen, Nat. Mater. 2021, 20, 1188.[8] D. Li, Y. Wu, P. Kim, L. Shi, P. Yang, A. Majumdar, Appl. Phys. Lett.2003, 83, 2934.[9] P. E. Hopkins, J. C. Duda, S. P. Clark, C. P. Hains, T. J. Rotter, L. M.Phinney, G. Balakrishnan, Appl. Phys. Lett. 2011, 98, 161913.[10] M. N. Luckyanova, J. Garg, K. Esfarjani, A. Jandl, M. T. Bulsara, A.J. Schmidt, A. J. Minnich, S. Chen, M. S. Dresselhaus, Z. Ren, E. aFitzgerald, G. Chen, Science 2012, 338, 936.[11] M. D. Losego, M. E. Grady, N. R. Sottos, D. G. Cahill, P. V. Braun,Nat.Mater. 2012, 11, 502.[12] J. Cho, M. D. Losego, H. G. Zhang, H. Kim, J. Zuo, I. Petrov, D. G.Cahill, P. V. Braun, Nat. Commun. 2014, 5, 4035.[13] S. Ning, S. C. Huberman, Z. Ding, H. H. Nahm, Y. H. Kim, H. S. Kim,G. Chen, C. A. Ross, Adv. Mater. 2019, 31, 1903738.[14] Y. Zang, C. Di, Z. Geng, X. Yan, D. Ji, N. Zheng, X. Jiang,H. Fu, J.Wang,W. Guo, H. Sun, L. Han, Y. Zhou, Z. Gu, D. Kong, H. Aramberri, C.Cazorla, J. Íñiguez, R. Rurali, L. Chen, J. Zhou, D. Wu, M. Lu, Y. Nie,Y. Chen, X. Pan, Adv. Mater. 2022, 34, 2105778.[15] G. E. W. Bauer, E. Saitoh, B. J. van Wees, Nat. Mater. 2012, 11, 391.[16] S. R. Boona, R. C. Myers, J. P. Heremans, Energy Environ. Sci. 2014, 7,885.[17] K. Uchida, Proc. Jpn. Acad., Ser. B 2021, 97, 69.[18] K. Uchida, J. P. Heremans, Joule 2022, 6, 2240.[19] K. Uchida, T. Hirai, F. Ando, H. Sepehri-Amin, Adv. Energy Mater.2024, 14, 2302375.[20] T. Hirai, F. Ando, H. Sepehri-Amin, K. Uchida, Nat. Commun. 2024,15, 9643.[21] T. Jeong, M. T. Moneck, J. J. Zhu, IEEE Trans. Magn. 2012, 48, 3031.[22] F. K. Dejene, J. Flipse, G. E. W. Bauer, B. J. Van Wees,Nat. Phys. 2013,9, 636.[23] J. Kimling, R. B. Wilson, K. Rott, J. Kimling, G. Reiss, D. G. Cahill,Phys. Rev. B 2015, 91, 144405.[24] H. Nakayama, B. Xu, S. Iwamoto, K. Yamamoto, R. Iguchi, A. Miura,T. Hirai, Y. Miura, Y. Sakuraba, J. Shiomi, K. Uchida, Appl. Phys. Lett.2021, 118, 042409.[25] M. R. Natale, D. J. Wesenberg, E. R. J. Edwards, H. T. Nembach, J. M.Shaw, B. L. Zink, Phys. Rev. Mater. 2021, 5, L111401.[26] T. T. Heikkilä, M. Hatami, G. E. W. Bauer, Phys. Rev. B 2010, 81,100408(R).[27] T. An, V. I. Vasyuchka, K. Uchida, A. V. Chumak, K. Yamaguchi, K.Harii,J. Ohe, M. B. Jungfleisch, Y. Kajiwara, H. Adachi, B. Hillebrands, S.Maekawa, E. Saitoh, Nat. Mater. 2013, 12, 549.[28] P. Tang, K. Uchida, G. E. W. Bauer, Phys. Rev. B 2025, 111, L180407.[29] X. Wu, Z. Liu, T. Luo, J. Appl. Phys. 2018, 123, 085109.[30] W. B. Yelon, L. Berger, Phys. Rev. B 1972, 6, 1974.[31] Y. Hsu, L. Berger, Phys. Rev. B 1976, 14, 4059.[32] K. Uchida, R. Iguchi, J. Phys. Soc. Jpn. 2021, 90, 122001.[33] J. Xiao, G. E. W. Bauer, K. Uchida, E. Saitoh, S. Maekawa, Phys. Rev. B2010, 81, 214418.[34] R. Ramos, T. Kikkawa, M. H. Aguirre, I. Lucas, A. Anadon, T. Oyake,K. Uchida, H. Adachi, J. Shiomi, P. A. Algarabel, L. Morellon, S.Maekawa, E. Saitoh, M. R. Ibarra, Phys. Rev. B 2015, 92, 220407(R).[35] L. J. Cornelissen, K. J. H. Peters, G. E. W. Bauer, R. A. Duine, B. J. vanWees, Phys. Rev. B 2016, 94, 014412.[36] K. Uchida, R. Iguchi, S. Daimon, R. Ramos, A. Anadón, I. Lucas, P. A.Algarabel, L. Morellón, Phys. Rev. B 2017, 95, 184437.[37] J. Bass, W. P. Pratt, J. Phys. Condens. Matter 2007, 19, 183201.[38] H. Yu, R. Huber, T. Schwarze, F. Brandl, T. Rapp, P. Berberich, G.Duerr, D. Grundler, Appl. Phys. Lett. 2012, 100, 262412.[39] M. Madami, E. Iacocca, S. Sani, G. Gubbiotti, S. Tacchi, R. K. Dumas,J. Åkerman, G. Carlotti, Phys. Rev. B 2015, 92, 024403.[40] H. S. Körner, M. A. W. Schoen, T. Mayer, M. M. Decker, J. Stigloher, T.Weindler, T. N. G. Meier, M. Kronseder, C. H. Back, Appl. Phys. Lett.2017, 111, 132406.[41] L. Flacke, L. Liensberger, M. Althammer, H. Huebl, S. Geprägs, K.Schultheiss, A. Buzdakov, T. Hula, H. Schultheiss, E. R. J. Edwards,H. T. Nembach, J.M. Shaw, R. Gross,M.Weiler,Appl. Phys. Lett. 2019,115, 122402.[42] D. G. Cahill, Rev. Sci. Instrum. 2004, 75, 5119.[43] N. Taketoshi, T. Baba, A. Ono, Rev. Sci. Instrum. 2005, 76, 094903.[44] P. Jiang, X. Qian, R. Yang, J. Appl. Phys. 2018, 124, 161103.[45] T. Yamazaki, T. Hirai, T. Yagi, Y. Yamashita, K. Uchida, T. Seki, K.Takanashi, Phys. Rev. Appl. 2024, 21, 024039.[46] Y. Song, D. Shoemaker, J. H. Leach, C. McGray, H. L. Huang, A.Bhattacharyya, Y. Zhang, C. U. Gonzalez-Valle, T. Hess, S. Zhukovsky,K. Ferri, R. M. Lavelle, C. Perez, D. W. Snyder, J. P. Maria, B. Ramos-Alvarado, X. Wang, S. Krishnamoorthy, J. Hwang, B.M. Foley, S. Choi,ACS Appl. Mater. Interfaces 2021, 13, 40817.[47] R. Modak, Y. Sakuraba, T. Hirai, T. Yagi, H. Sepehri-Amin, W. Zhou,H. Masuda, T. Seki, K. Takanashi, T. Ohkubo, K. Uchida, Sci. Technol.Adv. Mater. 2022, 23, 767.[48] Y. Kajiwara, K. Harii, S. Takahashi, J. Ohe, K. Uchida, M. Mizuguchi,H. Umezawa, H. Kawai, K. Ando, K. Takanashi, S.Maekawa, E. Saitoh,Nature 2010, 464, 262.[49] B. Heinrich, C. Burrowes, E. Montoya, B. Kardasz, E. Girt, Y. Y. Song,Y. Sun, M. Wu, Phys. Rev. Lett. 2011, 107, 066604.[50] K. Uchida, J. Xiao, H. Adachi, J. Ohe, S. Takahashi, J. Ieda, T. Ota,Y. Kajiwara, H. Umezawa, H. Kawai, G. E. W. Bauer, S. Maekawa, E.Saitoh, Nat. Mater. 2010, 9, 894.[51] J. Flipse, F. K. Dejene, D. Wagenaar, G. E. W. Bauer, J. Ben Youssef, B.J. van Wees, Phys. Rev. Lett. 2014, 113, 027601.[52] Y. Wang, J. Y. Park, Y. K. Koh, D. G. Cahill, J. Appl. Phys. 2010, 108,043507.[53] A. Giri, P. E. Hopkins, Adv. Funct. Mater. 2020, 30, 1903857.[54] S. R. Boona, J. P. Heremans, Phys. Rev. B 2014, 90, 064421.[55] B. L. Giles, Z. Yang, J. S. Jamison, R. C. Myers, Phys. Rev. B 2015, 92,224415.[56] L. J. Cornelissen, B. J. van Wees, Phys. Rev. B 2016, 93, 020403.[57] R. Iguchi, A. Yagmur, Y. C. Lau, S. Daimon, E. Saitoh, M. Hayashi, K.Uchida, Phys. Rev. B 2018, 98, 014402.[58] M. E. Lucassen, C. H. Wong, R. A. Duine, Y. Tserkovnyak, Appl. Phys.Lett. 2011, 99, 262506.[59] M. V. Costache, G. Bridoux, I. Neumann, S. O. Valenzuela,Nat.Mater.2012, 11, 199.[60] H. Masuda, R. Modak, T. Seki, K. Uchida, T.-C. Lau, Y. Sakuraba, R.Iguchi, K. Takanashi, Commun. Mater. 2020, 1, 75.[61] J. Uzuhashi, Y. Yao, T. Ohkubo, T. Sekiguchi,Microscopy 2025, 74, 279.[62] Y. Yamashita, K. Honda, T. Yagi, J. Jia, N. Taketoshi, Y. Shigesato, J.Appl. Phys. 2019, 125, 035101.[63] K. Kobayashi, T. Baba, Jpn. J. Appl. Phys. 2009, 48, 05EB05.[64] A. J. Schmidt, X. Chen, G. Chen, Rev. Sci. Instrum. 2008, 79, 114902.Adv. Funct. Mater. 2025, 35, 2506554 2506554 (8 of 8) © 2025 The Author(s). Advanced Functional Materials published by Wiley-VCH GmbH 16163028, 2025, 40, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/adfm.202506554, Wiley Online Library on [17/10/2025]. 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 Licensehttp://www.advancedsciencenews.comhttp://www.afm-journal.de Non-Equilibrium Magnon Engineering Enabling Significant Thermal Transport Modulation 1. Introduction 2. Results and Discussion 2.1. Phenomenological Prediction of Thermal Transport Engineering by Non-Equilibrium Spin Current 2.2. Experimental Details of Time-Domain Thermoreflectance 2.3. TDTR Measurements in Double-Wedge CoFe/Garnet Substrate 2.4. Conduction-Electron Spin Current versus Magnon Spin Current 3. Conclusion 4. Experimental Section Supporting Information Acknowledgements Conflict of Interest Data Availability Statement Keywords