# Fileset [MDR-Cost_effective_NiAu.pdf](https://mdr.nims.go.jp/filesets/9486d548-05f6-4668-9454-892c648142e1/download) ## Creator A. Riss, F. Garmroudi, M. Parzer, C. Eisenmenger-Sittner, A. Pustogow, [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 [Material-efficient preparation and thermoelectric properties of metallic

{Ni}_{x} {Au}_{1 - x}

films with large power factor](https://mdr.nims.go.jp/datasets/2a3a3097-c104-4ba6-a093-34d7be490c01) ## Fulltext Material-efficient preparation and thermoelectric properties of metallic NixAu1-x films with large power factorA. Riss,1, ∗ F. Garmroudi,1 M. Parzer,1 C. Eisenmenger-Sittner,1 A. Pustogow,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, JapanThermoelectric materials can utilize the Seebeck effect to convert electrical into thermal energyand vice versa. State-of-the-art thermoelectric materials are semiconductors as the band gap in theirelectronic density of states results in large values of the Seebeck coefficient. However, recent resultssuggest that certain metallic alloys such as NixAu1–x can also feature promising thermoelectricproperties, relevant for technological applications (Fabians Paper ausgeschrieben zitieren). In thiswork, we study the thermoelectric potential of NiAu films on different substrates, with additionalfocus on a cost-efficient synthesis. Ultimately, a large power factor of 11.2 mW/mK2 is achieved at614 K in Ni0.27Au0.73. Furthermore, by neglecting the phonon contribution to thermal conduction,which is negligibly small in these metallic systems well-obeying the Wiedemann-Franz law, a figureof merit of zT = 0.15 is obtained at 763 K.INTRODUCTIONIn times of a rapidly changing climate and increasingglobal temperature, the necessity of using energy more ef-ficiently and lowering the amount of waste heat is clearlyapparent. As of 2012, the vast majority of the global pri-mary energy (≈ 72,%) is lost at conversion, with ther-mal losses having the largest share, as estimated by For-man et al. [1]. Thermoelectric materials have demon-strated to be able to partially convert thermal into elec-tric energy by making use of the Seebeck effect. The ef-ficiency of such a material depends on the figure of meritzT = S2σT/λ, where S is the Seebeck coefficient, σ theelectrical conductivity and λ = λph + λel the thermalconductivity, composed of contributions from phononsand electrons, respectively. All but the phonon ther-mal conductivity strongly depend on the charge carrierconcentration [2], thus preventing the simultaneous opti-mization of all electronic properties by means of simplealiovalent doping. Instead, more sophisticated strategies,such as band convergence, modulation doping or mobilityenhancement are necessary [3]. The electrical conductiv-ity and electron thermal conductivity are further coupledtrough the Wiedemann-Franz law λelσ = LT with theLorenz number L. The phonon thermal conductivity isthe only parameter than can be partially manipulatedwithout effecting electronic transport to a significant ex-tent.In good metals, λel � λph and the figure of meritreduces tozT =S2L+O(λphλel). (1)In the metallic limit, the performance thus only dependson the Seebeck coefficient of the material.Since A. Ioffe’s proposal of the superiority of semicon-ductors over metals [4], thermoelectric research has al-most entirely focused on finding, designing and improv-ing systems composed of semiconductors. This is basedon the assumption that a finite band gap is necessary toobtain a sufficient asymmetry of the electronic densityof states N(E) with respect to the Fermi level. How-ever, the Seebeck coefficient additionally depends on thescattering time τ [5], which can also strongly vary withenergy, causing an asymmetry. In a first approximation,the Seebeck coefficient can be determined from the well-known Mott formulaS = −π2k2B3eT(∂ lnN∂E− ∂ ln τ−1∂E)EF. (2)Garmroudi et al. combined experimental data and the-oretical calculations of NixAu1–x alloys to demonstratethat the Seebeck coefficient is significantly increased intransition metals when the Fermi energy closely alignswith the d states [6].The presence of flat d bandes in Ni results in a consid-erable ∂ lnN/∂E, but their lower conductivity diminishestheir overall contribution to the total Seebeck coefficient,as described by the formulaS =Ssσs + Sdσdσs + σd, (3)where s and d denote the s and d states, respectively.Consequently, the dominant contribution to S primarilyoriginates from the highly mobile and more conductive sstates.In the context of the Mott formula, the first term inEquation 2 is neglectable in s and p states due to theirbroad bandwidth. Conversely, in the case of Au s satesscattering into Ni d states, there is a significant increasein the scattering rate τ−1 within the energy range of Ni dstates. This results in a substantial enhancement of thesecond term in Equation 2, ∂ ln τ−1/∂E. By partiallysubstituting Ni with Au, effectively shifting the Fermienergy closer to the edge of the d band, Garmroudi et al.2achieved a remarkable increase in the power factor, reach-ing 34 mW/mK2 in Ni0.1Au0.9 at 560 K [6].In this study, we attempt to replicate the crystal struc-ture and thermoelectric properties of the bulk in filmsdeposited on various substrates. We place particular em-phasis on developing a cost-effective synthesize method,addressing the challenge associated with the costly prepa-ration of Au-rich bulk materials.EXPERIMENTAL DETAILSThe sputter target was prepared from a Ni disk witha diameter of 25.4 mm, onto which either two or threeflattened Au pieces were affixed through spot-welding.Once in place, the target was compressed with a pres-sure of 200 bar. This not only led to great electrical andthermal contacts but also ensured a decent mechanicalconnection, although strong vibrations could potentiallydislodge the Au components. Each Au piece weightedapproximately 0.5 g and a single target could produce be-tween 20 to 30 films (depending on the desired thickness),which highlights the cost-efficiency of the approach.It’s worth noting that the overall target thickness is re-stricted to < 1.5 mm due to the ferromagnetic behavior ofNi. Otherwise, the target could disrupt the magnetic fieldwithin the sputter chamber by causing a short-circuitingof the field. Consequently, the thickness has a significantimpact on the required gas pressure for initiating the gasdischarge.The sputtering process was conducted at a tempera-ture of 500 ◦C utilizing direct-current (DC) magnetronsputtering, with a power of either 5 W or 10 W, a target-substrate distance of 30 mm and a working gas pressureof 0.05 mbar. Prior to the deposition, the substrates werecleaned with acetone and ethanol. The substrates encom-passed unpolished Si and yttria-stabilized zirconia (YSZ)as well as polished quartz glass.The films’ composition was characterized using energy-dispersive X-ray (EDX) spectroscopy, conducted with theFEI Quanta 250 FEG. The crystal structure was deter-mined using the X’Pert MPDII diffractometer from Pan-alytical in the Bragg-Brentano geometry. To minimizereflection peaks from the single-crystalline substrates, anoffset of ω = 3 ◦ was applied. For electrical resistivity andSeebeck coefficient measurements, the ZEM-3 equipmentfrom ULVAC was used.RESULTSComposition and crystal structureThe composition of all samples is depicted in Figure 1a.The x-axis shows the sample number in chronological se-Ni + AuPhase separationNixAu1-xSolid solution937 K1068 K a)b)Figure 1: a) Composition of all films versus the samplenumber. The purple and green points belong to filmssputtered from the target with two and three Au piecesshown in the inset, respectively. Half-filled symbolsindicate films made at a power of 5 W, while filledsymbols represent films made at 10 W. The gray dashedlines highlight the mean amount of Ni in the films. b)Phase diagram of binary Ni-Au [7]. The blue areamarks the region of solubility. The purple and greenareas highlight the necessary annealing temperature toobtain a single phase. For x < 0.3, perfect solubilitywas found without quenching, as illustrated by theblack dashed line [6].quence, effectively representing the progression of the tar-get ablation. It is evident that the amount of Ni increasesas more material is removed from the target. This phe-nomenon is likely attributed to the finite surface area andvolume of the Au pieces in contrast to the quasi-infiniteNi layer beneath. Films originating from the target withless Au exhibited an average Ni content of x = 0.56,roughly twice the amount found in films deposited fromthe Au-rich target (x = 0.24 on average). It is essen-3tial to highlight that precise control over the compositionthrough the strategic placement of Au is challenging dueto several uncertainties, e.g. the precise positioning rel-ative to the magnetron-sputtering ablation ring and thesize, shape and even thickness of the Au pieces, whichcan manipulate the ion current distribution on the tar-get. Instead, achieving the desired composition necessi-tates adjustments to parameters like working gas pres-sure, target-substrate distance and offset and power, allof which are known to impact the film’s composition [8–11]. For example, examining the purple data points inFigure 1a reveals that the amount of Ni is slightly higherat 10 W compared to 5 W. It is important to note thatthe conventional theory, which posits that the ratio ofejected atoms equals the target’s composition due to sur-face adjustments, does not apply in this context. Thisis because the two elements are spatially separated andnot intermixed, making the individual energy-dependentsputtering yields a critical factor to consider [8].The phase diagram of binary Ni–Au alloys is presentedin Figure 1b [7]. In the blue-shaded region, the elementsexhibit complete solubility, forming a solid solution. Atlower temperatures, they tend to segregate, leading tothe formation of Ni-rich and Au-rich phases. To obtainsingle-phase films, it is necessary to sputter or annealat specific temperatures, with upper limits of 937 K and1068 K for films produced from targets with 3 and 2 Aucomponents, respectively. Garmroudi et al. achieved sta-bilization of the solid solution in bulk materials at roomtemperature by rapidly quenching the samples in wa-ter, effectively preserving the metastable crystal struc-ture [6]. However, it’s worth noting that for x < 0.3 theauthors observed that a single phase could also be ob-tained through gradual furnace cooling, as indicated bythe black dashed line in Figure 1b.Figure 2a-c shows the diffraction pattern of films onquartz glass, Si and YSZ before and after annealing at600 ◦C for three days. Prior to annealing, the films de-posited on glass, with a composition of Ni0.58Au0.42, ex-hibit a phase resembling Ni0.6Au0.4 with minor traces ofpure Au (see Figure 2a). Upon annealing, the structureundergoes a transformation, shifting towards an Au-richNi0.12Au0.88 phase along with pure Ni. This transforma-tion aligns with findings from bulk alloy measurements[6].On Si substrates (Figure 2b), the same Ni0.6Au0.4phase is formed within the Ni0.52Au0.48 sample, while theAu-rich film predominantly comprised Ni0.2Au0.8 despitea subtle presence of a Ni-rich peak. This agrees with thealloy’s stability predicted for x < 0.3. Subsequent to theannealing process, an almost pure Au phase (Ni0.03Au0.97forms irrespective of the initial composition. Any re-maining Ni disperses from the primary phase, manifest-ing as NiSi2. Notably, the crystal structure exhibits onlyslight variations between the samples with x = 0.22 andx = 0.52, primarily reflecting differences in the NiSi2 con- ZrO2:Yc) ZrO2Ni0.6Au0.4Ni0.12Au0.88Au Sib)Ni0.03Au0.97Ni0.2Au0.8NiSi2SiO2CuAuUnknownphaseNi0.6Au0.4 Glassa)Ni Ni0.12Au0.88AuSiO2Ni0.6Au0.4 Nix = 0.58x = 0.55x = 0.55x = 0.19x = 0.19x = 0.52x = 0.52x = 0.22x = 0.22x = 0.58NixAu1-xFigure 2: X-ray diffraction pattern of NiAu films on a)polished quartz glass, b) unpolished Si and c)unpolished YSZ. In each case, the blue and red linesrepresent samples before and after annealing at 600 ◦Cfor three days, respectively. The Ni content, denoted asx, is displayed on the left side of each line for reference.tent. The Au-rich sample additionally comprised tinyamounts of a CuAu phase, which could originate fromthe copper plate of the sputtering chamber.The films deposited on YSZ with a composition ofNi0.55Au0.45 show a tendency similar to that observedin the samples on quartz glass, as depiced in Figure 2c.Without annealing, these films adopt a Ni0.6Au0.4 phasetogether with pure Au. Upon annealing, this phase dis-perses into Ni0.12Au0.88 along with the emergence of pureNi. Much like the films on Si, the same crystal structureis observed on YSZ after annealing, irrespective of the4composition. However, in the case of Au-rich films, thereduced Ni content leads to a decrease in the presence ofpure Ni.In summary, no significant difference in the crystalstructure is observed among the three substrates con-cerning the NixAu1–x phase. On Si, however, NiSi2forms, while in the other two substrates, pure Ni per-sists. Prior to annealing, most samples contain minortraces of pure Au, hinting at a potential lack of thermalenergy to stabilize into the favourable crystal structure.This is likely attributed to the sputtering setup wherethe heater is positioned within a copper plate beneath thesubstrate. The weak thermal contact between the copperand the substrate causes a decrease of the temperaturewhich is exacerbated by the low thermal conductivity ofglass (λ ≈ 1 W/mK) and YSZ (λ ≈ 2 W/mK) [12, 13].Conversely, on Si, the temperature is expected to be sev-eral degrees higher due to the significantly larger thermalconductivity (λ ≈ 150 W/mK) [14] and the thinner sub-strate (280 µm compared to 560 µm of YSZ and 1 mm ofglass). These circumstance facilitates the formation ofthe desired crystal structure.Thermoelectric propertiesConsidering that pure Ni and Au exhibit Seebeck coef-ficients of only ≈ −10µV/K and ≈ 3µV/K above roomtemperature, respectively, and given that multiple, non-interacting phases tend to have a detrimental effect onthe overall material properties [composite paper zitieren],it is reasonabel to expect that films exhibiting multiplephases would yield a lower performance compared to thebulk material with the same NiAu phase.The measured thermoelectric properties, along withthe estimated figure of merit zT derived from Equation 1,are illustrated in Figure 3. For comparative purposes,the data of bulk Ni0.2Au0.8 and Ni0.6Au0.4 alloys are in-cluded, representing compositions that closely resemblethose of the films.Due to the exceedingly low electrical resistivity of NiAualloys, a contribution of any of the substrates can be ruledout at room temperature. However, at elevated temper-atures, the substantial decrease in electrical resistivityin Si results in a modification of the measured Seebeckcoefficient and electrical resistivity of the film-substratesystem, leading to non-physical values above 550 K [15].The Seebeck coefficient does not exhibit a significantdependence on the substrate, as corroborated by the X-ray diffraction pattern. An exception to this trend isobserved in annealed films on the Si substrate, where aconsiderable reduction in thermopower is evident. Thisdecline is presumably a consequence of Si diffusion andthe substantial formation of a NiSi2 phase.Films deposited from the target with less Au uniformlyexhibit the Ni0.6Au0.4 phase prior to annealing (indicatedby the light blue lines) and closely align with the respec-tive bulk properties. Although the resistivity is higherdue to limited ordering, the Seebeck coefficient, and con-sequently zT , show a slight improvement. This enhance-ment is attributed to the decomposition of the bulk forx ≥ 0.3, a phenomenon observed only in the films afterannealing. In the case of Au-rich films on Si, a Ni0.2Au0.8phase forms upon sputtering, in contrast to films on YSZ,where a Ni0.12Au0.88 phase coexists with pure Au. Con-sequently, the film on Si exhibits a Seebeck coefficientcloser to the respective bulk value, albeit substantiallyreduced. Ultimately, on Si, the power factor and figureof merit reach PF = 6.7 mW/mK2 in Ni0.22Au0.78 at418 K and zT = 0.1 Ni0.52Au0.48 at 344 K.Annealed films deposited on Si manifest a Au-dominant Ni0.03Au0.97 phase alongside NiSi2, causing adetrimental effect on the Seebeck coefficient and, conse-quently, overall thermoelectric performance. Thus, theinteraction of Ni with Si renders Si substrates unsuitablefor the deposition of NiAu alloys.The comparable diffraction pattern observed in filmsdeposited on glass (Ni0.58Au0.42) and YSZ (Ni0.55Au0.45),both forming Ni0.12Au0.88 upon annealing, is also pro-nounced by closely aligned thermoelectric properties.Nonetheless, the performance is below that of bulkNi0.2Au0.8 and Ni0.1Au0.9 (PF ≈ 20 − 30 mW/mK2around room temperature [6]). The Seebeck coefficientis notably increased in Au-rich films on YSZ despite thesimilar crystal structure. However, the reduced Ni con-tent mitigates the adverse effect of multi-phase contri-butions. Ultimately, PF = 11.2 mW/mK2 at 614 K andzT = 0.15 at 763 K in Ni0.19Au0.81 deposited on YSZ.Adopting the method suggested by Garmroudi et al.[6], an attempt was made to achieve a single phase by an-nealing samples in an Ar-filled quartz tube at 937 K and1068 K, followed by rapid quenching in water. Unfortu-nately, this approach proved challenging as the thermalshock caused the substrates to fracture. Further exper-iments involving rapid cooling in air also proved unsuc-cessful and had a detrimental effect on the thermoelectricproperties of the materials. Thus, future work should fo-cus on using different substrates with better mechanicalproperties to withstand the thermal shock during waterquenching.CONCLUSIONNixAu1–x films have been successfully prepared viaDC magnetron sputtering, employing a single Ni targetto with Au pieces were affixed. The Au components hada combined weight ranging from 1 to 2 g, allowing for theproduction of 20 to 30 films per target, demonstratingthe cost-effectiveness of this single-target approach.Achieving precise control over the composition provedto be challenging due to various uncertainties, including5 � � � � � � � � � � � � � �0S [µV/K]N i 0 . 2 A u 0 . 8N i 0 . 6 A u 0 . 402 55 07 51 0 01 2 51 5 01 7 5� [µΩcm]S i N i 0 . 2 2 A u 0 . 7 8 N i 0 . 5 2 A u 0 . 4 8 N i 0 . 2 2 A u 0 . 7 8 a n n e a l e d N i 0 . 5 2 A u 0 . 4 8 a n n e a l e dZ r O 2 : Y N i 0 . 1 9 A u 0 . 8 1 N i 0 . 5 5 A u 0 . 4 5 N i 0 . 1 9 A u 0 . 8 1 a n n e a l e d N i 0 . 5 5 A u 0 . 4 5 a n n e a l e dG l a s s N i 0 . 5 8 A u 0 . 4 2 N i 0 . 5 8 A u 0 . 4 2 a n n e a l e d051 01 52 02 53 03 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 7 0 0 7 5 0 8 0 0024681 01 2PF [mW/mK²]T [ K ]3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 7 0 0 7 5 0 8 0 00 . 0 00 . 0 50 . 1 00 . 1 5zTT [ K ]d )a ) b )c )Figure 3: a) Seebeck coefficient, b) electrical resistivity, c) power factor and d) figure of merit of films deposited onSi, YSZ and glass substrates. Blue and red data points correspond to samples before and after annealing,respectively. The gray lines show bulk properties of Ni0.1Au0.9 and Ni0.6Au0.4 for comparison, taken from Ref. [6].the positioning of Au pieces relative to the magnetron-sputtering ablation ring and the shapes of the pieces.Additionally, the thickness of the Ni layer influenced themagnetic field and, consequently, the ion current ontothe target surface. To manipulate the composition effec-tively, other sputtering-related parameters, such as work-ing gas pressure, power and substrate-target distance andoffset, have to be tuned.After deposition, the films exhibited phases close tostoichiometry, though the limited degree of orderingand traces of secondary phases hampered overall perfor-mance. Upon annealing, the phases transformed, shift-ing towards Au-rich phases and pure Ni. The thermo-electric properties depend on the amount of pure Ni orrelated phases, such as NiSi2. Ultimately, a power factorof 11.2 mW/mK2 and figure of merit of 0.15 was achievedin Ni0.19Au0.81 deposited on YSZ.To optimize the performance of NiAu films, rapidquenching in water appears necessary. However, the chal-lenge lies in finding a suitable setup or substrate to pre-vent film failure during this process.ACKNOWLEDGEMENTSFinancial 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-6edged for providing the scanning electron microscope tostudy the composition of NiAu films.∗ alexander.riss@tuwien.ac.at[1] C. Forman, I. K. Muritala, R. Pardemann, and B. Meyer,Renewable and Sustainable Energy Reviews 57, 1568(2016).[2] G. J. Snyder and E. S. Toberer, Nature materials 7, 105(2008).[3] G. Tan, L.-D. Zhao, and M. G. Kanatzidis, Chemicalreviews 116, 12123 (2016).[4] A. Ioffe, The socialist reconstruction and science 1, 23(1932).[5] A. Zevalkink, D. M. Smiadak, J. L. Blackburn, A. J. Fer-guson, M. L. Chabinyc, O. Delaire, J. Wang, K. Kovnir,J. Martin, L. T. Schelhas, et al., Applied Physics Reviews5 (2018).[6] F. Garmroudi, M. Parzer, A. Riss, C. Bourgès,S. Khmelevskyi, T. Mori, E. Bauer, and A. Pustogow,arXiv preprint arXiv:2303.03062 (2023).[7] J. Wang, X.-G. Lu, B. Sundman, and X. Su, Calphad 29,263 (2005).[8] N. Matsunami, Y. Yamamura, Y. Itikawa, N. Itoh,Y. Kazumata, S. Miyagawa, K. Morita, R. Shimizu, andH. Tawara, Atomic data and nuclear data tables 31, 1(1984).[9] A. Rogov and Y. V. Kapustin, Instruments and Experi-mental Techniques 63, 776 (2020).[10] 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).[11] 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).[12] C. Kittel, Physical Review 75, 972 (1949).[13] J.-F. Bisson, D. Fournier, M. Poulain, O. Lavigne, andR. Mévrel, Journal of the American Ceramic Society 83,1993 (2000).[14] H. Shanks, P. Maycock, P. Sidles, and G. Danielson,Physical Review 130, 1743 (1963).[15] A. Riss, M. Stöger, M. Parzer, F. Garmroudi,N. Reumann, B. Hinterleitner, T. Mori, and E. Bauer,Physical Review Applied 19, 054024 (2023).mailto:alexander.riss@tuwien.ac.at Cost-effective preparation and thermoelectric properties of metallic NiAu films Abstract Introduction Experimental details Results Composition and crystal structure Thermoelectric properties Conclusion Acknowledgements References