# Fileset [MDR-Iterative-Manuscript-Alex.pdf](https://mdr.nims.go.jp/filesets/21cd3e93-73bc-45f7-aa35-e12607bcddd8/download) ## Creator [A. Riss](https://orcid.org/0000-0002-9707-8394), E. Lasisch, S. Podbelsek, [K. Schäfer](https://orcid.org/0009-0006-8641-6665), [M. Parzer](https://orcid.org/0000-0003-3509-7474), [F. Garmroudi](https://orcid.org/0000-0002-0088-1755), [C. Eisenmenger-Sittner](https://orcid.org/0000-0002-7096-6092), [T. Mori](https://orcid.org/0000-0003-2682-1846), E. Bauer ## Rights ©2024 American Physical Society[In Copyright](http://rightsstatements.org/vocab/InC/1.0/) ## Other metadata [Iterative composition optimization in

{Fe}_{2} VA 1

–based thin-film thermoelectrics using single-target sputtering](https://mdr.nims.go.jp/datasets/f1579143-14e5-487a-8460-76a8f02bf6ed) ## Fulltext Iterative composition optimization in Fe2VAl-based thin-film thermoelectrics usingsingle-target sputteringA. Riss,1, ∗ E. Lasisch,1 S. Podbelsek,1 K. Schäfer,1 M. Parzer,1 F.Garmroudi,1 C. Eisenmenger-Sittner,1 T. Mori,2, 3 and E. Bauer11Institute of Solid State Physics, Technische Universität Wien, 1040 Vienna, Austria2International Center for Materials Nanoarchitectonics (WPI-MANA),National Institute for Materials Science, Tsukuba 305-0044, Japan3University of Tsukuba, Tsukuba 305-8577, JapanMagnetron sputtering inherently exhibits the advantage of dislodging particles from the target ina ratio equivalent to the target stoichiometry. Nevertheless, film compositions often deviate due toelement-dependent scattering with the working gas, necessitating the adjustment of the sputteringprocess. In this work, we explore an unconventional approach of addressing this issue, involvingthe employment of an off-stoichiometric target. The required composition is obtained throughan iterative process, which is demonstrated by Fe2VAl and Fe2V0.9Ti0.1Al films as case studies.Ultimately, the correct stoichiometry is obtained from Fe1.86V1.15Al0.99 and Fe1.88V1.02Ti0.13Al0.97targets, respectively. Despite the thermoelectric properties falling below expectations, mainly due toimperfect film crystallization, the strategy successfully achieved the desired stoichiometry, enablingaccurate film synthesis without the need of advanced sputtering setups.I. INTRODUCTIONThermoelectric materials exhibit the ability to convertheat into electric energy and vice versa. The perfor-mance of a material is expressed by the figure of meritzT = S2σT/λ, comprising the Seebeck coefficient S, theelectrical conductivity σ and the thermal conductivity λ.Since the discovery of the Seebeck effect over 200 yearsago, thermoelectric research has mainly focused on bulkmaterials, with current state-of-the-art materials reach-ing values of zT > 1 [1–3]. Nevertheless, setups utilizingthermoelectric films deposited on various substrates havealso demonstrated excellent properties and have provensuitable for thermoelectric applications [4–7].Over time, various different coating techniques havebeen developed, encompassing solid, liquid and vapor de-position processes [8]. Among the techniques within va-por deposition, sputtering is a notable example of phys-ical vapor deposition, nowadays commonly employed asmagnetron sputtering. The sputter yield, i.e. the amountof material removed from the sputter target per incidention, differs severely between elements [9]. More specif-ically, factors such as the sublimation energy, the scat-tering cross section, the energy and incident angle of theimpinging ion and the crystal structure of the target in-fluence the sputter yield [10]. Notably, however, a sig-nificant advantage of sputtering, in contrast to variousother vapor-deposition processes like thermal evapora-tion, is that after an initial depletion phase, the particleratio of the ejected atoms aligns with the compositionof the target [11, 12]. Nevertheless, it is important toemphasize that obtaining the desired composition in theresulting film is anything but guaranteed, which presentsa serious issue in obtaining reproducible thermoelectric∗ alexander.riss@tuwien.ac.atproperties. This is attributed to two factors: i) the an-gular distribution of the ejected atoms is not the same[13–16] and ii) the ejected atoms experience scatteringwith gas atoms while moving to the substrate, causingdeviations from their intended paths [14, 17, 18].This constitutes a significant challenge, especially interms of designing high-performance thermoelectric ma-terials, where tiny variations in the composition can havea severe impact on their properties. It that can be ad-dressed through various methods. The simplest approachinvolves multi-target sputtering [5, 19–21], employingmultiple targets with distinct discharge power to con-trol the composition. These targets may consist of singleelements, alloys or compounds. However, this approachrequires a suitable sputter chamber equipped with multi-ple target holders and power supplies. Another strategyis the use of chips placed on the target to compensatefor deficiencies in one or more elements [22–24]. Thesechips can be composed of one or multiple elements andmay vary in size. While this allows for a simple adap-tation of the composition, the accuracy is not very highcompared to other methods. Furthermore, adjustmentsof the composition can be achieved by modifying sputterparameters such as the sputter angle [23, 25, 26], power[27] or gas pressure [14, 28, 29], among others. This ap-proach works well for two-element alloys, but may yieldinsufficient results in case of more constituents.All of these strategies provide the ability to tailor thecomposition of sputtered films to meet the required sto-ichiometry, each with its own set of advantages and dis-advantages. Several studies on full-Heusler Fe2VAl-basedfilms utilized one of these approaches to obtain better sto-ichiometry and thermoelectric properties [22, 23, 30–33].In this work, we present an unconventional approach,which involves the adaptation of the target. By varyingthe composition of the target, films with the desired ratioof constituents are obtained. While it requires the abil-mailto:alexander.riss@tuwien.ac.at2StartCompositioncorrect?YesNoDetermination of the initial target compositionSynthesis of a target with the selectedcompositionDetermination of the new targetcompositionSynthesis of a film from the targetCalculation of the error between target and substrateMeasurement of the composition of target and filmEndFigure 1: Flow chart illustrating the iterativeoptimization process for refining the film compositionthrough adjustments to the stoichiometry of the target.The iterative process includes the synthesis of both atarget and a respective film as well as a measurement ofthe compositions. The loop repeats until the film’sstoichiometry matches the desired composition.ity to synthesize adjusted targets, it avoids the necessityof a multi-target sputter chamber and does not hold thesame challenges regarding accuracy and reproducibilityas placing a chip on a target. This approach is success-fully demonstrated for Fe2VAl-based Heusler compounds,yielding films with the nominal stoichiometry and crystalstructure.II. METHODOLOGYThe stoichiometry of the target is changed accordingto the deviation of the film’s composition. This processis conducted iteratively, starting with the synthesis of atarget and the respective film, followed by a compositionmeasurement. If the composition deviates from the de-sired stoichiometry, an error calculation is performed, fol-lowed by the synthesis of another target with an adaptedcomposition. The procedural steps are visually presentedin the flow chart shown in Fig. 1. The initial targetmay comprise either the stoichiometric composition, or,when information about deficiencies in one or more con-stituents is pre-known, an educated guess regarding theoff-stoichiometry.In this study, the precise composition of the targets hasbeen determined through X-ray fluorescence (XRF) spec-troscopy, utilizing the Zetium XRF spectrometer fromPanalytical. For thin films, the application of XRF spec-troscopy is more complex due to the interdependenceof intensity, measured composition and thickness [34].Consequently, the films’ composition has been deter-mined using the secondary-electron microscope (SEM)FEI Quanta 250 FEG with an energy-dispersive X-ray(EDX) detector. To minimize measurement errors, therespective target sample has been simultaneously mea-sured as reference. Details of the calculation process areprovided in Appendix A.Despite the impression that the correct film composi-tion is obtained after the first adoption of the target, asubsequent alteration in the target’s composition inducesa change in the sputter behavior. This arises e.g. from in-homogeneities within the target, resulting from finite sol-ubility of its constituents and the emergence of impurityphases [35]. These impurities may also possess magneticmoments, influencing the magnetic field in the vicinityof the target. Nevertheless, through an iterative process,the deviation of the film’s composition from the nominalvalue can be gradually reduced, as shown below. Eachadjustment in the target composition refines the sputterdynamics, ultimately yielding the desired stoichiometry.The synthesis of the targets target was conductedby weighing in high-purity (> 99.9 %) elements, whichwere melted together in a homemade high-frequencyinduction-melting setup. From the obtained ingot, discswith a diameter of 25.4 mm and a thickness of 2 mmwere cut out. The films were deposited with a direct-current magnetron sputtering setup, employing a dis-charge power of 10 W and a working gas pressure of2 Pa. Furthermore, the target-substrate distance was setto 50 mm, ensuring a uniform thickness distribution.The crystal structure of all samples was obtained fromX-ray diffraction (XRD) measurements, using a Panalyt-ical X’Pert MPDII diffractometer. To minimize reflectionpeaks from the single-crystalline substrates, an offset of4◦ was applied. The thermoelectric properties of boththe bulk and film samples were measured utilizing an UL-VAC ZEM-3. Furthermore, the thermal conductivity ofthe bulk material was calculated by determining the den-sity from the lattice parameter obtained from the XRDpattern as well as measuring the thermal diffusivity andheat capacity, employing the Lineis LFA500 Light Flash.III. RESULTSTo elaborate on the practicality and applicability ofthis approach, Fe2VAl and Fe2V0.9Ti0.1Al films with thedesired composition were synthesized and measured withrespect to their structural and thermoelectric proper-ties. In addition, we aimed to synthesize stoichiomet-ric Fe2TaAl thin films, which were recently predicted toshow superior thermoelectric performance compared toFe2VAl-based systems [36, 37]. However, significant chal-lenges related to the stability of the target material were3Fe 2VAlFe 1.83V1.17AlFe 1.9V 1.15Al 0.95Fe 1.86V1.15Al 0.990.80.91.02.02.12.22.3nominal compositionatomsperformulaunitFe 2V0.9Ti 0.1AlFe 1.92V1.05Ti 0.12Al 0.92Fe 1.86V1.07Ti 0.13Al 0.94Fe 1.88V1.02Ti 0.13Al 0.970.080.10.80.91.02.02.12.2atomsperformulaunita)b)TiFe2VAlFe2V0.9Ti0.1AlVAlFeVAlFeFigure 2: Atoms per formula unit of films made fromdifferent a) Fe2VAl-based and b) Fe2V0.9Ti0.1Al-basedtargets. Nominal stoichiometry is denoted by dashedlines, while solid lines represent the target’scomposition. The films’ composition resulting from thefinal targets is Fe2.02V0.99Al and Fe2.02V0.9Ti0.1Al0.99.faced, as elaborated in Appendix B.All films had a thickness of 2µm and were sputteredon unpolished yttria-stabilized zirconia (YSZ) substrates,ensuring a negligible contribution to the measured See-beck coefficient and electrical conductivity due to thehigh electrical resistance [38].A. Composition and crystal structureFig. 2 presents the composition of Fe2VAl- andFe2V0.9Ti0.1Al-based films for each synthesized tar-get, refined according to the above-presented approach.Films sputtered from the stoichiometric targets ex-hibited severe off-stoichiometry, i.e. Fe2.21V0.85Al0.94and Fe2.10V0.79Ti0.08Al1.04 instead of Fe2VAl andFe2V0.9Ti0.1Al were obtained. This underscores the im-portance for investigating the composition rather thanrelying on the frequently made assumption that it alignswith the target’s stoichiometry. Notably, an excess of Feis evident in both systems, while deficiencies in V andTi are observed, similar to previously reported resultson Fe2VAl-based films [22, 39]. However, it is impor-tant to emphasize that the off-stoichiometry can not bea single-element offset but rather represents variations inthe concentration of multiple elements.A total of four iterations (see Fig. 2) were necessary toachieve a satisfactory stoichiometry in the films for bothsystems. Ultimately, a composition of Fe2.02V0.99Al andFe2.02V0.9Ti0.1Al0.99 was obtained from Fe1.86V1.15Al0.99and Fe1.88V1.02Ti0.13Al0.97 targets, respectively, closelyresembling the desired stoichiometry.To elucidate the structural characteristics, the X-raydiffraction powder patterns of stoichiometric Fe2VAl andFe2V0.9Ti0.1Al bulk specimens, along with diffractionpatterns of the final stoichiometric films, are presented inFig. 3. The analysis of the targets reveals the presenceof all dominant peaks characteristic of the full-Heuslerstructure. Of particular significance are the (111) and(200) lines, appearing at ≈ 27 ◦ and ≈ 31 ◦, respectively.These lines serve as indicators of the degree of disorderwithin the material. In the event of B2 disorder, wherea complete disordering of the V and Al sites occurs, the(111) peak vanishes, while the fully disordered A2 struc-ture lacks both lines [40]. Although not clearly visible inFig. 3, both peaks are present in the targets, suggestingnearly complete ordering.On the contrary, the films do not exhibit a fully or-dered structure. Aside from peaks originating from theYSZ substrate and disorder-independent peaks from thefull-Heusler structure, films annealed at 873 K reveal onlythe (200) peak due to the lack of thermal energy neces-sary for L21 ordering. Furthermore, upon annealing at1073 K, both the (111) and (200) lines are absent. Simul-taneously, additional impurity peaks, presumable oxides,form during the annealing process.B. Thermoelectric propertiesThe thermoelectric transport properties of the sto-ichiometric targets were measured, alongside those offilms with the closest composition. The results forFe2VAl are presented in Fig. 4. The electronic struc-ture of Fe2VAl features a close-to-zero band gap, resem-bling a small-gap semiconductor or semimetal [42–44].Consequently, disorder or changes in the charge carrierconcentration from off-stoichiometry result in significantvariations of all thermoelectric quantities [45]. Consis-tent with this, an opposite sign of the Seebeck coefficientwas previously reported in B2-disordered Fe2VAl [41],aligning with the measured Seebeck coefficient of the B2-4a)Fe2VAlFe2VAlYSZsecondary phasesecondary phase20 30 40 50 60 70 80 90 100sample annealingtarget 873 Kfilm 873 Kfilm 1073 KIntensity[arb.u.]20 30 40 50 60 70 80 90 100Intensity[arb.u.]2 [deg.]b)Fe2V0.9Ti0.1AlFigure 3: X-ray diffraction pattern of a) Fe2VAl and b)Fe2V0.9Ti0.1Al, featuring the stoichiometric targets(gray line) alongside the films with the closestcomposition, annealed at 873 K and 1073 K for 3 days.Peaks from the full-Heusler structure andyttria-stabilized zirconia (YSZ) substrate arehighlighted with green and orange marks, respectively.Different impurity phases present in the films aredepicted with pink and blue symbols.disordered film annealed at 873 K shown in Fig. 4a. Therespective electrical resistivity exhibits an increased valueand weak temperature dependence, suggesting imperfectformation of the Heusler structure within the film due tolimited thermal energy during sputtering and annealing.Furthermore, upon annealing at 1073 K, resulting in theA2-disordered metallic ground state [46], the Seebeck co-efficient is deteriorated to < 10µV/K. In addition, theelectrical resistivity decreases due to an increase of thecarrier concentration and increasingly metallic groundstate, further corroborating the findings from the XRDmeasurement.The measured Fe2VAl target exhibits a positive See-beck coefficient of ≈ 60µV/K at room temperature andan electrical resistivity resembling semiconductorlike be-havior, consistent with previous reports in literature [47–50]. Given the coinciding composition of target and film,the differences in the temperature-dependent propertiesand the reduced thermoelectric performance of the filmare attributed to finite crystallization and limited degreeof order in the structure.Ultimately, a maximum power factor of5.6 · 10−2 mW/(mK2) at 450 K was achieved. As-suming a constant value for the thermal conductivity of3 W/(mK) [30], a figure of merit of 8.2 · 10−3 at 550 Kwas obtained.The thermoelectric properties of both bulk and filmFe2V0.9Ti0.1Al are depicted in Fig. 5. Similar to theFe2VAl films discussed above, films with a compositionresembling Fe2V0.9Ti0.1Al exhibit a disorder-inducednegative Seebeck coefficient, characteristic for the effectof disorder in Fe2VAl-based materials [41]. After an-nealing at 873 K, the Seebeck coefficient reaches approxi-mately −40µV/K. Concurrently, the resistivity exhibitsa small negative temperature dependence, indicative ofdisorder-dominated transport. Eventually, a broad max-imum power factor of 0.12 mW/(mK2) is achieved be-tween 330 and 490 K (highlighted in the inset of Fig. 5c).Recently, thermoelectric properties of films sputteredfrom a stoichiometric Fe2V0.9Ti0.1Al target without con-trolling the composition were reported [39]. Comparingthese values (depicted as gray data points in Fig. 5) withour measured results shows the superiority of stoichio-metric films and highlights the importance of meticulouscomposition tuning for performance optimization. Themeasured, fully ordered, target displays a Seebeck coeffi-cient of 65µV/K alongside a small, linearly increasing re-sistivity, consistent with previously reported results [51].Moreover, a power factor exceeding 2 mW/(mK2) and afigure of merit of 0.05 are achieved. The distinctive dif-ferences, compared to the measured films, can be under-stood by the formation of impurity phases in the latterupon annealing at 1073 K and a reduction in the degreeof ordering, thereby diminishing the thermoelectric per-formance. Notably, this further causes the performanceto fall below previously reported values of fully orderedFe2V0.9Ti0.1Al films [52].IV. CONCLUSIONIn this study, we have proposed and demonstrated apromising approach for tailoring the composition of ther-moelectric thin films deposited via sputtering techniques.Traditional methods, such as tuning of deposition param-eters and the use of compensatory chips, pose challengesin achieving precise stoichiometry due to various factors5a) b)c) d)300 350 400 450 500 550 600 650 700 750020406080S[µV/K]film = 3 W/(mK)Ref. [41]sample annealingtarget 873 Kfilm 873 Kfilm 1073 K300 350 400 450 500 550 600 650 700 7500250500750100012501500[µΩcm]300 350 400 450 500 550 600 650 700 7500.00.10.20.30.40.50.60.7PF[mW/(mK2 )]T [K]300 350 400 450 500 550 600 650 700 7500.0000.0020.0040.0060.0080.010zTT [K]300 400 500 600 7000.000.020.040.06Fe2VAlFigure 4: a) Seebeck coefficient S, b) electrical resistivity ρ, c) power factor PF and d) figure of merit zT of theFe2VAl target (black symbols) and the films with the closest stoichiometry after annealing at 873 K and 1073 K(dark red symbols) as a function of temperature. The zT of the films was calculated assuming atemperature-independent thermal conductivity of 3 W/(mK), consistent with prior findings [30]. For comparison,the Seebeck coefficient of B2-disordered Fe2VAl is included in (a) as gray points, taken from Ref. [41].influencing sputter dynamics. In contrast, our methodinvolves iteratively adapting the composition of the sput-ter target to obtain the desired film composition, thus cir-cumventing the need for complex sputter chamber setupsand providing greater accuracy compared to chip-basedadjustments.Films with desired compositions of Fe2VAl andFe2V0.9Ti0.1Al were successfully synthesized, demon-strating the applicability of the method and allowing forthe gradual refinement of film compositions to closelymatch the desired stoichiometry, which is crucial for de-signing high-performance thermoelectric materials. Al-though the thermoelectric properties of the films were be-low expectations, structural measurements revealed dis-order within the sample, presumably caused by a lackof thermal energy upon annealing, as well as additionalsecondary impurity phases. Thus, carefully refining theannealing process in future studies will likely yield per-formances coinciding with the respective bulk material.Nevertheless, compared to films sputtered from the sto-ichiometric target, which exhibit substantially reducedperformances, our results show the potential of the herepresented method to control the stoichiometry and im-prove the properties of films deposited by single-targetsputtering.ACKNOWLEDGMENTSFinancial support for the research in this paper wasgranted by the Japan Science and Technology Agency(JST) programs MIRAI, JPMJMI19A1. We acknowledgethe X-ray Center at TU Wien for providing their equip-ment. Furthermore, USTEM at TU Wien is acknowl-6a) b)c) d)300 350 400 450 500 550 600 650 700 750020406080S[µV/K]Ref. [39]300 350 400 450 500 550 600 650 700 750025050075010001250[µΩcm]sample annealingtarget 873 Kfilm 873 Kfilm 1073 K300 350 400 450 500 550 600 650 700 7500.00.51.01.52.0PF[mW/(mK2 )]T [K]300 350 400 450 500 550 600 650 700 7500.000.010.020.030.040.05zTT [K]300 400 500 600 7000.000.050.100.15film = 3 W/(mK)Fe2V0.9Ti0.1AlFigure 5: a) Seebeck coefficient S, b) electrical resistivity ρ, c) power factor PF and d) figure of merit zT of theFe2V0.9Ti0.1Al target (black symbols) and the films with the closest stoichiometry after annealing at 873 K and1073 K (dark red symbols) as a function of temperature. In addition, the thermoelectric properties of the finalFe1.88V1.02Ti0.13Al0.97 target are included (blue symbols). The zT of the films was calculated assuming atemperature-independent thermal conductivity of 3 W/(mK), consistent with prior findings [30]. For comparison,the Seebeck coefficient, electrical resistivity and power factor of off-stoichiometric Fe2V0.9Ti0.1Al films are includedas gray points, taken from Ref. [39].edged for providing the scanning electron microscope tostudy the composition of the synthesized bulk and filmsamples.Appendix A: Calculation process of the targetcompositionFig. 6 depicts the process of calculating the film com-position as well as the deviation between the target andfilm stoichiometry. From the figure it becomes evidentthat the EDX measurement yields slightly inaccurate re-sults. Assuming that the XRF results depict the correctstoichiometry, the EDX error ε can be computed for eachelement. These values are subsequently used to derive theactual composition of the film from the measured values.Furthermore, a comparison of the EDX results from tar-get and films enables the determination of the deviationδ between the stoichiometries due to different behaviorof the constituents during sputtering. Ultimately, δ isemployed to evaluate the required target composition forachieving the correct film stoichiometry.Appendix B: Synthesis of Fe2TaAl filmsA stoichiometric Fe2TaAl target was prepared, similarto the method described in the main article, by weigh-ing in and melting high-purity Fe, Ta and Al. Thefilms deposited from the target had a composition of7XRF (Target):Fe: 51.11 %V: 25.01 %Al: 23.88 %EDX (Target):Fe: 45.12 %V: 23.96 %Al: 30.92 %EDX (Film):Fe: 48.93 %V: 20.36 %Al: 30.71 %Actual film composition:Deviation of film to target δ:δFe: +8.44 %δV: -15.03 %δAl: -0.68 %EDX error ε:εFe: -11.72 %εV: -4.20 %εAl +29.48 %Nominal composition:Fe: 50.00 %V: 25.00 %Al: 25.00 %FeEDX - FeXRFFeXRFεFe=Fefilm - FetargetFetargetδFe=FeEDX1 + εFeFeact=Feprev1 + δFeFenext=Fe2VAlNext target composition:Fe: 45.79 %V: 29.22 %Al: 24.99 %Fe1.83V1.17AlFe: 55.21 %V: 21.17 %Al: 23.62 %Fe2.21V0.85Al0.94Figure 6: Visualization of the sequential process ofdetermining both the film and subsequent targetcomposition. The blue area shows the initiallyweighed-in, nominal composition of Fe2VAl.Measurement data are depicted in orange, while greenareas represent calculated values. Equations presentnext to the arrows highlight the utilized formalism,using Fe as an illustrative example.Fe2.27Ta0.58Al1.15. The X-ray diffraction patterns of boththe target and the film, following annealing at 873 K,are depicted in Fig. 7. The abundance of peaks ob-served in both samples distinctly indicates the presenceof multiple phases. Guided by the deviation in the film’scomposition, an optimized target with a composition ofFe1.81Ta0.87Al1.31 was synthesized. However, the severeoff-stoichiometry caused a complete breaking of the tar-get during the cooling phase subsequent to inductionmelting due to thermal stress.Consequently, small adjustments were madeto the composition, leading to the synthesis ofFe1.87Ta0.91Al1.21. Despite these refinements, thesample’s stability remained compromised, ultimatelycausing the abandonment of the refinement process forthis particular composition.[1] G. Rogl, A. Grytsiv, K. Yubuta, S. Puchegger, E. Bauer,C. Raju, R. Mallik, and P. Rogl, Acta Materialia 95, 201(2015).[2] L.-D. Zhao, S.-H. Lo, Y. Zhang, H. Sun, G. Tan, C. Uher,C. Wolverton, V. P. Dravid, and M. G. Kanatzidis, na-ture 508, 373 (2014).[3] Y. Pei, A. LaLonde, S. Iwanaga, and G. J. Snyder, En-ergy & Environmental Science 4, 2085 (2011).[4] R. Venkatasubramanian, E. Siivola, T. Colpitts, andB. O’quinn, Nature 413, 597 (2001).[5] P. Fan, Z.-h. Zheng, Y.-z. Li, Q.-y. Lin, J.-t. Luo, G.-x. Liang, X.-m. Cai, D.-p. Zhang, and F. Ye, AppliedPhysics Letters 106 (2015).[6] B. Hinterleitner, I. Knapp, M. Poneder, Y. Shi,H. Müller, G. Eguchi, C. Eisenmenger-Sittner, M. Stöger-Pollach, Y. Kakefuda, N. Kawamoto, et al., Nature 576,85 (2019).[7] Z.-H. Zheng, X.-L. Shi, D.-W. Ao, W.-D. Liu, M. Li, L.-Z. Kou, Y.-X. Chen, F. Li, M. Wei, G.-X. Liang, et al.,Nature Sustainability 6, 180 (2023).[8] A. Makhlouf, in Nanocoatings and ultra-thin films (Else-vier, 2011) pp. 3–23.[9] N. Matsunami, Y. Yamamura, Y. Itikawa, N. Itoh,Y. Kazumata, S. Miyagawa, K. Morita, R. Shimizu, andH. Tawara, Unknown (1983).[10] K. Wasa, Handbook of Sputtering Technology , 41(2012).[11] M. Ohring, Materials Science of Thin Films: Depositonand Structure (Elsevier, 2001) pp. 113–115.[12] R. A. Levy, Microelectronic materials and processes, Vol.164 (Springer Science & Business Media, 2012) pp. 164–173.[13] S. Mahieu, G. Buyle, D. Depla, S. Heirwegh, P. Ghekiere,and R. De Gryse, Nuclear Instruments and Methods inPhysics Research Section B: Beam Interactions with Ma-terials and Atoms 243, 313 (2006).[14] J. Neidhardt, S. Mráz, J. M. Schneider, E. Strub,W. Bohne, B. Liedke, W. Möller, and C. Mitterer, Jour-nal of applied physics 104 (2008).[15] Y. V. Martynenko, A. Rogov, and V. Shul’Ga, Technicalphysics 57, 439 (2012).[16] A. Rogov and Y. V. Kapustin, Instruments and Experi-mental Techniques 63, 776 (2020).[17] J. Shi, C. Chen, H. Yu, and S. Zhang, Vacuum 83, 249(2008).[18] E. Särhammar, E. Strandberg, J. Sundberg, H. Nyberg,T. Kubart, S. Jacobson, U. Jansson, and T. Nyberg, Sur-face and Coatings Technology 252, 186 (2014).[19] A. Boyd, C. O’Kane, and B. Meenan, Surface and Coat-ings Technology 233, 131 (2013).[20] Z.-k. Cai, P. Fan, Z.-h. Zheng, P.-j. Liu, T.-b. Chen, X.-m. Cai, J.-t. Luo, G.-x. Liang, and D.-p. Zhang, Applied8Fe2TaAlFe2.27Ta0.58Al1.1520 30 40 50 60 70 80 90 100Intensity[arb.u.]2 [deg.]sample annealingtarget 873 Kfilm 873 KFigure 7: X-ray diffraction pattern of the Fe2TaAltarget (gray line) alongside the respective film with thecomposition of Fe2.27Ta0.58Al1.15 (orange line), annealedat 873 K for 3 days.surface science 280, 225 (2013).[21] R. Lan, S. L. Otoo, P. Yuan, P. Wang, Y. Yuan, andX. Jiang, Applied Surface Science 507, 145025 (2020).[22] S. Hiroi, M. Mikami, and T. Takeuchi, Materials trans-actions 57, 1628 (2016).[23] N. Fukatani, Y. Kurosaki, S. Yabuuchi, A. Nishide, andJ. Hayakawa, Applied Physics Letters 112 (2018).[24] F. Machda, S. Singh, K. Kuga, A. K. Ang, M. Mat-sunami, and T. Takeuchi, Japanese Journal of AppliedPhysics 62, 055502 (2023).[25] C. Wu, C. Tseng, Y. Fanchiang, C. Cheng, K. Lin, S. Yeh,S. Yang, C. Wu, T. Liu, M. Wu, et al., Scientific Reports8, 11087 (2018).[26] S. Hou, J. Kwo, R. Watts, J.-Y. Cheng, and D. Fork,Applied physics letters 67, 1387 (1995).[27] S. Kim, Y.-S. Lee, and N.-H. Kim, Materials 16, 6087(2023).[28] Y. Hakuraku, S. Miyata, T. Inoue, and S. Nozoe, Super-conductor Science and Technology 11, 179 (1998).[29] T. Selinder, G. Larsson, U. Helmersson, and S. Rudner,Journal of applied physics 69, 390 (1991).[30] Y. Furuta, K. Kato, T. Miyawaki, H. Asano, andT. Takeuchi, Journal of electronic materials 43, 2157(2014).[31] S. Hiroi, M. Mikami, K. Kitahara, and T. Takeuchi, In-ternational Journal of Nanotechnology 13, 881 (2016).[32] Y. Kurosaki, S. Yabuuchi, A. Nishide, N. Fukatani, andJ. Hayakawa, AIP Advances 10 (2020).[33] Y. Kurosaki, S. Yabuuchi, A. Nishide, N. Fukatani, andJ. Hayakawa, Applied Physics Express 15, 085502 (2022).[34] T. Huang, X-Ray Spectrometry 20, 29 (1991).[35] H. Miyazaki, S. Tanaka, N. Ide, K. Soda, and Y. Nishino,Materials Research Express 1, 015901 (2013).[36] D. I. Bilc, G. Hautier, D. Waroquiers, G.-M. Rignanese,and P. Ghosez, Physical review letters 114, 136601(2015).[37] S. A. Khandy, I. Islam, D. C. Gupta, M. A. Bhat, S. Ah-mad, T. A. Dar, S. Rubab, S. Dhiman, and A. Laref,RSC advances 8, 40996 (2018).[38] A. Riss, M. Stöger, M. Parzer, F. Garmroudi,N. Reumann, B. Hinterleitner, T. Mori, and E. Bauer,Physical Review Applied 19, 054024 (2023).[39] W. Gao, Z. Liu, T. Baba, Q. Guo, D.-M. Tang,N. Kawamoto, E. Bauer, N. Tsujii, and T. Mori, ActaMaterialia 200, 848 (2020).[40] S. Maier, S. Denis, S. Adam, J.-C. Crivello, J.-M. Jou-bert, and E. Alleno, Acta Materialia 121, 126 (2016).[41] F. Garmroudi, M. Parzer, A. Riss, A. V. Ruban,S. Khmelevskyi, M. Reticcioli, M. Knopf, H. Michor,A. Pustogow, T. Mori, et al., Nature Communications13, 3599 (2022).[42] D. Singh and I. Mazin, Physical Review B 57, 14352(1998).[43] A. Berche, M. T. Noutack, M.-L. Doublet, and P. Jund,Materials Today Physics 13, 100203 (2020).[44] B. Hinterleitner, F. Garmroudi, N. Reumann, T. Mori,E. Bauer, and R. Podloucky, Journal of Materials Chem-istry C 9, 2073 (2021).[45] E. Alleno, A. Diack-Rasselio, M. T. Noutack, andP. Jund, Physical Review Materials 7, 075403 (2023).[46] E. Alleno, A. Berche, J.-C. Crivello, A. Diack-Rasselio,and P. Jund, Physical Chemistry Chemical Physics 22,22549 (2020).[47] Y. Nishino, M. Kato, S. Asano, K. Soda, M. Hayasaki,and U. Mizutani, Physical review letters 79, 1909 (1997).[48] B. Hinterleitner, P. Fuchs, J. Rehak, F. Garmroudi,M. Parzer, M. Waas, R. Svagera, S. Steiner, M. Kishi-moto, R. Moser, et al., Physical Review B 102, 075117(2020).[49] F. Garmroudi, A. Riss, M. Parzer, N. Reumann,H. Müller, E. Bauer, S. Khmelevskyi, R. Podloucky,T. Mori, K. Tobita, et al., Physical Review B 103, 085202(2021).[50] M. Parzer, F. Garmroudi, A. Riss, S. Khmelevskyi,T. Mori, and E. Bauer, Applied Physics Letters 120,071901 (2022).[51] M. Mikami, K. Ozaki, H. Takazawa, A. Yamamoto,Y. Terazawa, and T. Takeuchi, Journal of electronic ma-terials 42, 1801 (2013).[52] M. Mikami, T. Kamiya, and K. Kobayashi, Thin SolidFilms 518, 2796 (2010). Iterative composition optimization in Fe2VAl-based thin-film thermoelectrics using single-target sputtering Abstract Introduction Methodology Results Composition and crystal structure Thermoelectric properties Conclusion Acknowledgments Calculation process of the target composition Synthesis of Fe2TaAl films References