A Proposal for a Composite with Temperature-Independent Thermophysical Properties: HfV2–HfV2O7

The HfV2–HfV2O7 composite is proposed as a material with potentially temperature-independent thermophysical properties due to the combination of anomalously increasing thermoelastic constants of HfV2 with the negative thermal expansion of HfV2O7. Based on literature data, the coexistence of both a near-zero temperature coefficient of elasticity and a coefficient of thermal expansion is suggested for a composite with a phase fraction of approximately 30 vol.% HfV2 and 70 vol.% HfV2O7. To produce HfV2–HfV2O7 composites, two synthesis pathways were investigated: (1) annealing of sputtered HfV2 films in air to form HfV2O7 oxide on the surface and (2) sputtering of HfV2O7/HfV2 bilayers. The high oxygen mobility in HfV2 is suggested to inhibit the formation of crystalline HfV2–HfV2O7 composites by annealing HfV2 in air due to oxygen-incorporation-induced amorphization of HfV2. Reducing the formation temperature of crystalline HfV2O7 from 550 °C, as obtained upon annealing, to 300 °C using reactive sputtering enables the synthesis of crystalline bilayered HfV2–HfV2O7.


Introduction
Volume expansion upon heating is probably the most prominent example of the influence of temperature on the physical properties of materials. However, for electronic, optical, and other high-precision devices, whose performances are critically affected by slight variations in volume, near-zero thermal expansion materials are desired [1][2][3][4][5]. Moreover, for mechanical components of precision instruments, temperature-independent volumes and elastic moduli are required [6][7][8]. A combination of both properties has only been obtained in gum metals [9,10] and Fe-Ni alloys [11] after intense deformation, thereby promoting the development of materials with intrinsic near-zero expansion and temperature-invariant elastic behavior irrespective of the plastic deformation route.
To compensate for thermally-induced volume changes, implementation of materials with negative thermal expansion (NTE) in a composite is a widely propagated approach [4,5,12]. Here, we propose to join a material with NTE properties with another exhibiting anomalously increasing thermoelastic constants in an attempt to obtain a combination of temperature-invariant elastic behavior and volume. This is a pioneering approach for the design of materials with temperature-independent thermophysical properties. substrate remained at floating potential. For the synthesis of stoichiometric HfV 2 , the employed target power densities were 2.0 and 10.2 W cm −2 for Hf and V, respectively. Ar (99.9999%) was used as sputtering gas to achieve a working pressure of 0.4 Pa. The base pressure (at deposition temperature) was below 1 × 10 −6 Pa. HfV 2 films were deposited without intentional heating and at substrate temperatures of 500 and 700 • C. To ensure high purity of deposited HfV 2 , the following measures were taken: The backsides of the single-crystalline sapphire (0001) substrates were deposited beforehand with approximately 250 nm of Nb to overcome their partial transmissivity to radiation during heating. This consequently reduced the heat impact on the chamber walls, and thus, a considerable decrease in base pressure at elevated deposition temperatures was achieved. Second, prior to each deposition, all targets were sputtered for 5 min with closed shutters (positioned approximately 2 cm opposite of the target) to make the surfaces free from condensed impurities and to getter residual gases. Third, elemental Zr was additionally sputtered at 20 W against a shutter during all depositions to exploit its pronounced affinity for oxygen [46,47] as a getter pump to further reduce residual gas incorporation [48] into the growing HfV 2 thin film. The Zr concentration in the as-deposited films was below 0.7 at.% based on energy-dispersive X-ray spectroscopy (EDX). Subsequently, selected films were capped after cooling to room temperature with an approximately 10 nm thick Nb layer to prevent impurity incorporation into the as-deposited thin film during air exposure.
HfV 2 O 7 thin films were deposited at a substrate temperature of 450 • C in a reactive Ar/O 2 (99.999%) atmosphere at a constant working pressure of 0.86 Pa while varying the O 2 partial pressure between 0.05 and 0.13 Pa. Further information can be found elsewhere [49]. For the HfV 2 -HfV 2 O 7 bilayer deposition, the synthesis procedure was the following: First, HfV 2 was deposited at 700 • C for 75 min. Afterwards, the system was cooled down in vacuum to the synthesis substrate temperatures for HfV 2 O 7 , which were 250, 300, and 350 • C. Pure Ar was used for plasma ignition before O 2 (p O2 = 0.09 Pa) was introduced to reactively sputter HfV 2 O 7 for 100 min.
Annealing experiments were performed in a GERO F 40-200/13 air furnace (Carbolite Gero, Neuhausen, Germany). The Hf-V ratio in the synthesized films was measured by EDX carried out in a JEOL JFM-6480 SEM (JEOL Ltd., Tokyo, Japan) equipped with an EDAX Genesis 2000 device (EDAX Inc., Mahwah, NJ, USA) at an acceleration voltage of 20 kV. Chemical composition depth profiling of HfV 2 was done by time-of-flight elastic recoil detection analysis (ToF-ERDA) at the Tandem Accelerator Laboratory of Uppsala University. 36 MeV 127 | 8+ primary ions and a time-of-flight telescope in combination with a Si solid-state detector for energy discrimination were used. Further details on the detector telescope can be found elsewhere [50]. The time and energy coincidence spectra were evaluated using the CONTES software package [51]. O and H depth profiles were characterized by ToF-ERDA, while the Hf-V ratio was obtained from EDX spectra. It should be noticed that Nb-capped und uncapped HfV 2 were measured by ToF-ERDA under identical conditions, hence, systematic uncertainties do not affect this comparison. The structure of the synthesized films was studied using X-ray diffraction (XRD) carried out with a Bruker AXS D8 Discover General Area Detector Diffraction System (Bruker Corporation, Billerica, MA, USA). A Cu K α source (current 40 mA, voltage of 40 kV) was used with a 0.5 mm pinhole collimator. Scans were performed at a fixed incidence angle of 15 • . Selected samples were measured between room temperature and 475 • C using a DHS 1100 Hot Stage (Anton Paar, Ostfildern-Scharnhausen, Germany) equipped with a NI-NiCr thermocouple to measure the surface temperature. Peak fitting was conducted using TOPAS software (version 3) with a pseudo-Voigt II function. Lattice parameters of the cubic structures were consequently calculated employing Bragg's law [52]. The lattice parameters of HfV 2 were determined by averaging the data from (220), (311), (222), (331), (422), (333), and (044) reflections. Based on the obtained changes in lattice parameters with temperature, the linear coefficient of thermal expansion for cubic HfV 2 O 7 was calculated by averaging the data from (200), (210), (211), (211), (220), (311), and (222) reflections. Standard deviations are added to evaluate the fitting quality. Morphology of bilayered HfV 2 -HfV 2 O 7 was studied using scanning transmission electron microscopy (STEM) carried out in an FEI Helios Nanolab 660 dual-beam microscope (Thermo Fisher Scientific, Waltham, MA, USA). Cross-sectional sample preparation was conducted by focused ion beam techniques with a Ga + source following a standard lift-out procedure [53].

Composite Assessment
First, the suitability of HfV 2 -HfV 2 O 7 as a composite with temperature-independent physical properties is evaluated, which relies on the coexistence of a near-zero coefficient of thermal expansion and a near-zero temperature coefficient of elasticity (TCE) for a certain phase fraction ratio. The TCE is defined by and is estimated for HfV 2 -HfV 2 O 7 composite using a rule-of-mixture approach (weighted average based on the volume fractions). While the temperature dependence of the elastic modulus of HfV 2 is taken from the literature [35], no thermoelasticity data for HfV 2 [40], respectively, and share common structural features, both forming an openly-packed network structure of octahedral (Zr/Hf)O 6 and polyhedral (W/V)O 4 units connected by corner-sharing oxygen atoms [63]. The TCE of the composite was averaged over a temperature range from 120 (onset of NTE behavior in HfV 2 O 7 [40,41,64]) to 300 • C. On the other hand, the coefficient of thermal expansion of a composite usually does not follow a simple rule-of-mixture behavior [65]. NTE materials are generally less stiff (lower E) than expected based on the bond strengths [66], and thus normally constitute the more compliant component in the composite. As a result, the elastic deformation during expansion and contraction is predominantly concentrated on the NTE component reducing its impact on the overall expansion coefficient. Consequently, theoretical models to describe the expansion coefficient of a composite typically also take the elastic properties of the individual components into account [66]. In the Turner model the overall coefficient of thermal expansion of the composite α c is described by where B i , α i , and φ i denote the bulk modulus, coefficient of thermal expansion, and volume fraction, respectively, of component i [65]. The bulk modulus and the coefficient of thermal expansion of HfV 2 (HfV 2 O 7 ) of 117 GPa [32] (56 GPa [49]) and 9.9 × 10 −6 K −1 [36] (−7.2 × 10 −6 K −1 [40]), respectively, were used for the calculation of α c . The resulting volume fraction dependent expansion coefficient α c and averaged TCE of HfV 2 -HfV 2 O 7 composite are plotted in Figure 1. Materials 2020, 13, x FOR PEER REVIEW 5 of 17 The coefficient of thermal expansion of the composite was calculated to reach temperature invariance (α = 0) at a volume fraction of HfV2 of around 0.25 (ϕ = 0.75). A TCE of zero was achieved for ϕ of around 0.35 (ϕ = 0.65) with a corresponding α of 1.9 × 10 −6 K −1 , which complies with the class of very low thermal expansion materials [1]. These results not only suggest that α and TCE of this composite can individually be adjusted to thermal invariance but also that a near-zero α and TCE may be achieved simultaneously. Based on the applied data and the assumptions outlined above, we predict that the corresponding volume phase fractions of HfV2 (HfV2O7) are between 20 and 40  vol.% depending on the optimization criterion. After assessing the capability of composite HfV2-HfV2O7 to exhibit temperature-independent properties, its synthesis is studied in the following.

Composite Formation by Oxidation of HfV2
A potential synthesis pathway for a composite has been demonstrated by oxidizing TiN thin films, where the time-and temperature-dependent formation of a TiO2 scale on top of TiN was observed with varying thickness, and hence phase fraction ratios [67,68]. Consequently, the first synthesis strategy to form HfV2-HfV2O7 composites comprises synthesis of HfV2 and subsequent heat treatment in air to partly oxidize HfV2 in an attempt to form a HfV2O7 oxide scale on top.

Phase Formation of HfV2
First, the phase formation for co-sputtered Hf-V thin films using magnetron sputtering is described. No previous reports on the synthesis of HfV2 thin films by magnetron sputtering are currently available. However, the synthesis of isoelectronic and isostructural ZrV2 thin films is discussed in the literature [69][70][71]. EDX analysis of the as-deposited thin films shows deviations from the desired 1Hf:2V stoichiometry below ±1 at.%. The results of the structural analysis to study the influence of the deposition temperature on the phase formation are depicted in Figure 2. The coefficient of thermal expansion of the composite was calculated to reach temperature invariance (α c = 0) at a volume fraction of HfV 2 of around 0.25 (φ HfV2O7 = 0.75). A TCE of zero was achieved for φ H f V2 of around 0.35 (φ H f V2O7 = 0.65) with a corresponding α c of 1.9 × 10 −6 K −1 , which complies with the class of very low thermal expansion materials [1]. These results not only suggest that α c and TCE of this composite can individually be adjusted to thermal invariance but also that a near-zero α c and TCE may be achieved simultaneously. Based on the applied data and the assumptions outlined above, we predict that the corresponding volume phase fractions of HfV 2 (HfV 2 O 7 ) are between 20 and 40  vol.% depending on the optimization criterion. After assessing the capability of composite HfV 2 -HfV 2 O 7 to exhibit temperature-independent properties, its synthesis is studied in the following.

Composite Formation by Oxidation of HfV 2
A potential synthesis pathway for a composite has been demonstrated by oxidizing TiN thin films, where the time-and temperature-dependent formation of a TiO 2 scale on top of TiN was observed with varying thickness, and hence phase fraction ratios [67,68]. Consequently, the first synthesis strategy to form HfV 2 -HfV 2 O 7 composites comprises synthesis of HfV 2 and subsequent heat treatment in air to partly oxidize HfV 2 in an attempt to form a HfV 2 O 7 oxide scale on top.

Phase Formation of HfV 2
First, the phase formation for co-sputtered Hf-V thin films using magnetron sputtering is described. No previous reports on the synthesis of HfV 2 thin films by magnetron sputtering are currently available. However, the synthesis of isoelectronic and isostructural ZrV 2 thin films is discussed in the literature [69][70][71]. EDX analysis of the as-deposited thin films shows deviations from the desired 1Hf:2V stoichiometry below ±1 at.%. The results of the structural analysis to study the influence of the deposition temperature on the phase formation are depicted in Figure 2. Materials 2020, 13, x FOR PEER REVIEW 6 of 17 A broad hump around 40° was obtained for HfV2 deposited without intentional heating (RT), indicating an amorphous structure, as reported for magnetron sputtered Zr-V thin films deposited without heating [71] and at 400 °C [70]. In this temperature regime, in which the adatom mobility is low, the formation of amorphous solid solution in Zr-V can be understood based on its higher stability compared to a random bcc solid solution, as demonstrated by DFT calculations [71]. It is reasonable to assume that this may also apply to Hf-V.
For Zr-V, a phase formation sequence from amorphous (400 °C) to a phase mixture of bcc V and hcp Zr (500 °C) to intermetallic ZrV2 (600 °C) was obtained with increasing deposition temperatures [70]. The same sequence was obtained by annealing amorphous films [70]. The intermediate formation of bcc V and hcp Zr is contradictive to the accepted equilibrium phase diagram and is mostly discussed with respect to difficult nucleation kinetics of the Laves phase structure [69,70]. However, it may also be explained by the thermodynamic instability of cubic ZrV2, as predicted by DFT in the ground state [71][72][73], exhibiting an energy of formation of 150 meV atom −1 [71], which persists at elevated temperatures. In comparison, several theoretical studies also predict cubic HfV2 to be energetically unstable in the ground state, reporting energies of formation between 20 and 35 meV per atom [29,74,75]. Furthermore, experiments show a transformation of cubic HfV2 upon cooling into an orthorhombic structure at around −160 °C [76,77] possibly due to kinetically limited decomposition into elemental V and Hf, since orthorhombic HfV2 also exhibits positive energy of formation [29,75]. However, for sputtered HfV2 at 500 °C (see Figure 2), the change in shape of the main peak at around 38° suggests the first formation of nanocrystals, and the emerging hump around 20° points towards intermetallic HfV2 nanocrystals, since their presence cannot be explained by hcp Hf or bcc V.
A further increase in deposition temperature to 700 °C resulted in the formation of sharp diffraction peaks which all can be assigned to cubic HfV2 [78]. Thus, no phase mixture of hcp Hf and bcc V, unlike for Zr-V, was observed, but a sequence from amorphous to crystalline HfV2 with increasing deposition temperature was obtained. Consequently, the high temperature of 700 °C required to form crystalline HfV2 is attributed to the kinetically limited formation of the Laves phase structure and not to an energetic instability of the cubic structure up to these temperatures. This A broad hump around 40 • was obtained for HfV 2 deposited without intentional heating (RT), indicating an amorphous structure, as reported for magnetron sputtered Zr-V thin films deposited without heating [71] and at 400 • C [70]. In this temperature regime, in which the adatom mobility is low, the formation of amorphous solid solution in Zr-V can be understood based on its higher stability compared to a random bcc solid solution, as demonstrated by DFT calculations [71]. It is reasonable to assume that this may also apply to Hf-V.
For Zr-V, a phase formation sequence from amorphous (400 • C) to a phase mixture of bcc V and hcp Zr (500 • C) to intermetallic ZrV 2 (600 • C) was obtained with increasing deposition temperatures [70]. The same sequence was obtained by annealing amorphous films [70]. The intermediate formation of bcc V and hcp Zr is contradictive to the accepted equilibrium phase diagram and is mostly discussed with respect to difficult nucleation kinetics of the Laves phase structure [69,70]. However, it may also be explained by the thermodynamic instability of cubic ZrV 2 , as predicted by DFT in the ground state [71][72][73], exhibiting an energy of formation of 150 meV atom −1 [71], which persists at elevated temperatures. In comparison, several theoretical studies also predict cubic HfV 2 to be energetically unstable in the ground state, reporting energies of formation between 20 and 35 meV per atom [29,74,75]. Furthermore, experiments show a transformation of cubic HfV 2 upon cooling into an orthorhombic structure at around −160 • C [76,77] possibly due to kinetically limited decomposition into elemental V and Hf, since orthorhombic HfV 2 also exhibits positive energy of formation [29,75]. However, for sputtered HfV 2 at 500 • C (see Figure 2), the change in shape of the main peak at around 38 • suggests the first formation of nanocrystals, and the emerging hump around 20 • points towards intermetallic HfV 2 nanocrystals, since their presence cannot be explained by hcp Hf or bcc V.
A further increase in deposition temperature to 700 • C resulted in the formation of sharp diffraction peaks which all can be assigned to cubic HfV 2 [78]. Thus, no phase mixture of hcp Hf and bcc V, unlike for Zr-V, was observed, but a sequence from amorphous to crystalline HfV 2 with increasing deposition temperature was obtained. Consequently, the high temperature of 700 • C required to form crystalline HfV 2 is attributed to the kinetically limited formation of the Laves phase structure and not to an energetic instability of the cubic structure up to these temperatures. This notion is supported by theoretical predictions suggesting the energetic stabilization of cubic HfV 2 at temperatures as low as −120 • C due to lattice vibrations [29]. For ZrV 2 , due to the considerably higher energy of formation in the ground state [71], its energetic stabilization is expected at higher temperatures, potentially explaining the discussed phase formation differences between Hf-V and Zr-V thin films. For comparison, bulk synthesis of HfV 2 includes heat treatments at temperatures above 1200 • C [34,77,79]. The reduction in synthesis temperature to 700 • C is enabled by surface diffusion of adatoms during sputtering [80].
No indications for impurity phases based on the presented XRD results were obtained in these samples, which is ascribed to the measures outlined above. However, thin films synthesized at higher base pressures or without additional co-sputtering of Zr contained traces of Hf 3 V 3 O and HfO 2 (not shown). This is in agreement with the observation of ZrV 3 O 3 as an impurity phase in sputtered Zr-V thin films [70].

Stability of HfV 2
As a next step, the stability of the synthesized HfV 2 films upon air exposure is examined. For this purpose, the structure of uncapped HfV 2 was studied as a function of the cumulated air-exposure time after removing it from the high-vacuum deposition system. The results are summarized in Figure 3.
Materials 2020, 13, x FOR PEER REVIEW 7 of 17 notion is supported by theoretical predictions suggesting the energetic stabilization of cubic HfV2 at temperatures as low as −120 °C due to lattice vibrations [29]. For ZrV2, due to the considerably higher energy of formation in the ground state [71], its energetic stabilization is expected at higher temperatures, potentially explaining the discussed phase formation differences between Hf-V and Zr-V thin films. For comparison, bulk synthesis of HfV2 includes heat treatments at temperatures above 1200 °C [34,77,79]. The reduction in synthesis temperature to 700 °C is enabled by surface diffusion of adatoms during sputtering [80].
No indications for impurity phases based on the presented XRD results were obtained in these samples, which is ascribed to the measures outlined above. However, thin films synthesized at higher base pressures or without additional co-sputtering of Zr contained traces of Hf3V3O and HfO2 (not shown). This is in agreement with the observation of ZrV3O3 as an impurity phase in sputtered Zr-V thin films [70].

Stability of HfV2
As a next step, the stability of the synthesized HfV2 films upon air exposure is examined. For this purpose, the structure of uncapped HfV2 was studied as a function of the cumulated air-exposure time after removing it from the high-vacuum deposition system. The results are summarized in Figure 3. In the as-deposited state, referring to a minimized air-exposure time of approximately 15 min, the peak positions coincide well with literature data for cubic HfV2 [78]. With increasing air-exposure time, a continuous peak shift to lower 2θ values is obtained, indicating an increase in the lattice parameter of the cubic structure from 7.38 to 7.51 Å (+1.8%) after four and to 7.54 Å (+2.2%) after ten weeks in air (standard deviations ≤ 0.01 Å). The obtained increase in lattice parameter may suggest continuous interstitial incorporation of impurities into the HfV2 lattice, which consequently also explains reported property changes in (Hf,Zr)V2 bulk samples after a one-year storage period [81]. HfV2 has been reported to act as a strong (weak) getter for hydrogen (oxygen) by dissociation of water [82]. In the as-deposited state, referring to a minimized air-exposure time of approximately 15 min, the peak positions coincide well with literature data for cubic HfV 2 [78]. With increasing air-exposure time, a continuous peak shift to lower 2θ values is obtained, indicating an increase in the lattice parameter of the cubic structure from 7.38 to 7.51 Å (+1.8%) after four and to 7.54 Å (+2.2%) after ten weeks in air (standard deviations ≤ 0.01 Å). The obtained increase in lattice parameter may suggest continuous interstitial incorporation of impurities into the HfV 2 lattice, which consequently also explains reported property changes in (Hf,Zr)V 2 bulk samples after a one-year storage period [81]. HfV 2 has been reported to act as a strong (weak) getter for hydrogen (oxygen) by dissociation of water [82]. Thin metal capping layers were shown to serve as effective oxidation barriers during air exposure at room temperature [83]. Hence, as-deposited HfV 2 was capped with approximately 10 nm Nb due to its passivating properties [84]. The functionality of this capping layer was investigated systematically by comparing the stabilities of capped and uncapped HfV 2 using XRD (see Figure 3). Other than a small additional peak around 38 • , measured for the Nb-capped film, which is attributed to bcc Nb, no difference in phase composition between both HfV 2 films in the as-deposited state was observed. However, in contrast to the uncapped HfV 2 film, no peak shift with increasing air-exposure time was obtained, demonstrating that the Nb capping ensures protection of synthesized HfV 2 against the incorporation of impurities from the ambient at room temperature.
To identify the incorporated impurities, the chemical compositions of both samples, uncapped and Nb-capped, were analyzed four weeks after deposition using ToF-ERDA. The measured oxygen concentration as a function of the film thicknesses is presented in Figure 4. The depth scale was calculated with the atomic masses of Hf and V under the assumption of stoichiometric HfV 2 with a density of 9.3 g cm −3 [32]. Thin metal capping layers were shown to serve as effective oxidation barriers during air exposure at room temperature [83]. Hence, as-deposited HfV2 was capped with approximately 10 nm Nb due to its passivating properties [84]. The functionality of this capping layer was investigated systematically by comparing the stabilities of capped and uncapped HfV2 using XRD (see Figure 3). Other than a small additional peak around 38°, measured for the Nb-capped film, which is attributed to bcc Nb, no difference in phase composition between both HfV2 films in the as-deposited state was observed. However, in contrast to the uncapped HfV2 film, no peak shift with increasing air-exposure time was obtained, demonstrating that the Nb capping ensures protection of synthesized HfV2 against the incorporation of impurities from the ambient at room temperature.
To identify the incorporated impurities, the chemical compositions of both samples, uncapped and Nb-capped, were analyzed four weeks after deposition using ToF-ERDA. The measured oxygen concentration as a function of the film thicknesses is presented in Figure 4. The depth scale was calculated with the atomic masses of Hf and V under the assumption of stoichiometric HfV2 with a density of 9.3 g cm −3 [32]. Uncapped HfV2 exhibited an averaged oxygen concentration of 4.3 ± 0.7 at.% in the bulk part of the film (neglecting the top surface oxidation) and a minor amount of hydrogen (<0.3 at.%, not shown) while the average oxygen concentration in the Nb-capped film was 0.5 ± 0.3 at.% and hydrogen was below the detection limit. Thus, the increasing lattice parameter for uncapped HfV2 is primarily attributed to a continuous interstitial uptake of oxygen into the cubic structure upon exposure to air. It has been shown theoretically that interstitially incorporated oxygen contributes to the energetic and mechanical stabilization of the cubic structure [29]. The oxygen concentration was measured to be constant throughout the analyzed film thickness of approximately 400 nm, suggesting high mobility of oxygen in cubic HfV2 already at room temperature. This may also have implications for the oxidation behavior of these films at elevated temperatures, which is discussed in the following.

Oxidation of HfV2
One sample of uncapped HfV2 was cyclically annealed in air for approximately 30 min at temperatures between 150 and 650 °C in 100 °C intervals. After each annealing step, the structure of the sample was studied by XRD. The resulting diffractograms are shown in Figure 5. Uncapped HfV 2 exhibited an averaged oxygen concentration of 4.3 ± 0.7 at.% in the bulk part of the film (neglecting the top surface oxidation) and a minor amount of hydrogen (<0.3 at.%, not shown) while the average oxygen concentration in the Nb-capped film was 0.5 ± 0.3 at.% and hydrogen was below the detection limit. Thus, the increasing lattice parameter for uncapped HfV 2 is primarily attributed to a continuous interstitial uptake of oxygen into the cubic structure upon exposure to air. It has been shown theoretically that interstitially incorporated oxygen contributes to the energetic and mechanical stabilization of the cubic structure [29]. The oxygen concentration was measured to be constant throughout the analyzed film thickness of approximately 400 nm, suggesting high mobility of oxygen in cubic HfV 2 already at room temperature. This may also have implications for the oxidation behavior of these films at elevated temperatures, which is discussed in the following.

Oxidation of HfV 2
One sample of uncapped HfV 2 was cyclically annealed in air for approximately 30 min at temperatures between 150 and 650 • C in 100 • C intervals. After each annealing step, the structure of the sample was studied by XRD. The resulting diffractograms are shown in Figure 5. Materials 2020, 13, x FOR PEER REVIEW 9 of 17 In the initial state, the sample was already exposed to air for 9 weeks. As a consequence, peak positions were already shifted towards smaller 2θ values compared to the as-deposited (see Figure  3) and Nb-capped films (see Figure 5). For all annealing steps up to 350 °C, the structural development is characterized by a continuous peak shift and broadening suggesting the successive incorporation of oxygen into the cubic structure finally yielding amorphization. as observed at 450 °C. Complementary annealing experiments with additional samples at fixed temperatures of 300 and 400 °C indicate that this process is determined by kinetics and may occur already at lower temperatures (not shown).
After annealing at 550 °C, small peaks emerged which may already indicate the formation of HfV2O7 nanocrystals [41], while the two main peaks at 32° and 37° cannot be attributed to any reported equilibrium phase in the Hf-V-O materials system. The peak positions of the unknown phase fit reasonably well to a cubic (fcc-based) structure exhibiting a lattice parameter of 4.9 Å. The reported cubic structures in the Hf-V-O systems, VO (Fm3 m) [85] and HfO2 (Fm3 m) [86], can be excluded due to the difference in the lattice parameters. Furthermore, the formation of a hypothetic cubic Hf1−xVxO monoxide appears unlikely based on conducted DFT studies, predicting a lattice parameter between 4.19 and 4.58 Å for x = 0 and x = 1, respectively. Nevertheless, the experimentally obtained lattice parameter lies within the predicted lattice parameter range of a previously unreported Hf1-xVxO2 dioxide (Fm 3 m) which is between 5.08 and 4.74 Å for x = 0 and x = 1, respectively. Hence, the formation of a metastable ternary (Hf,V)O2 phase is assumed henceforth. A high V solubility in cubic HfO2 has also been reported elsewhere [87]. At 650 °C, the formation of XRD phase pure HfV2O7 is evident [41].
The observation that the first crystallites of HfV2O7 may form at temperatures of 550 °C while oxygen is highly mobile in HfV2 already at room temperature suggests that the formation of HfV2- In the initial state, the sample was already exposed to air for 9 weeks. As a consequence, peak positions were already shifted towards smaller 2θ values compared to the as-deposited (see Figure 3) and Nb-capped films (see Figure 5). For all annealing steps up to 350 • C, the structural development is characterized by a continuous peak shift and broadening suggesting the successive incorporation of oxygen into the cubic structure finally yielding amorphization. as observed at 450 • C. Complementary annealing experiments with additional samples at fixed temperatures of 300 and 400 • C indicate that this process is determined by kinetics and may occur already at lower temperatures (not shown).
After annealing at 550 • C, small peaks emerged which may already indicate the formation of HfV 2 O 7 nanocrystals [41], while the two main peaks at 32 • and 37 • cannot be attributed to any reported equilibrium phase in the Hf-V-O materials system. The peak positions of the unknown phase fit reasonably well to a cubic (fcc-based) structure exhibiting a lattice parameter of 4.9 Å. The reported cubic structures in the Hf-V-O systems, VO (Fm3m) [85] and HfO 2 (Fm3m) [86], can be excluded due to the difference in the lattice parameters. Furthermore, the formation of a hypothetic cubic Hf 1−x V x O monoxide appears unlikely based on conducted DFT studies, predicting a lattice parameter between 4.19 and 4.58 Å for x = 0 and x = 1, respectively. Nevertheless, the experimentally obtained lattice parameter lies within the predicted lattice parameter range of a previously unreported Hf 1-x V x O 2 dioxide (Fm3m) which is between 5.08 and 4.74 Å for x = 0 and x = 1, respectively. Hence, the formation of a metastable ternary (Hf,V)O 2 phase is assumed henceforth. A high V solubility in cubic HfO 2 has also been reported elsewhere [87]. At 650 • C, the formation of XRD phase pure HfV 2 O 7 is evident [41].
The observation that the first crystallites of HfV 2 O 7 may form at temperatures of 550 • C while oxygen is highly mobile in HfV 2 already at room temperature suggests that the formation of HfV 2 -HfV 2 O 7 composites is not feasible using conventional annealing experiments in air. The oxygen mobility in HfV 2 at the formation temperature of crystalline HfV 2 O 7 appears to be critical for the endeavor of HfV 2 -HfV 2 O 7 composite formation. However, lowering of the synthesis temperature of crystalline HfV 2 O 7 by using non-equilibrium-based synthesis approaches appears promising and is discussed in the following. The formation of crystalline HfV 2 O 7 at a substrate temperature of 350 • C using reactive magnetron sputtering has been demonstrated previously [49], but the effect of the O 2 partial pressure has not been investigated yet. However, for the synthesis of bilayered HfV 2 -HfV 2 O 7 by magnetron sputtering, the O 2 partial pressure during the synthesis of HfV 2 O 7 is expected to be decisive for the purity of the underlying HfV 2 film, due to its high oxygen affinity. Hence, the influence of the O 2 partial pressure on the phase formation of HfV 2 O 7 was studied systematically for a synthesis temperature of 450 • C. The results of the structural analysis of synthesized thin films, exhibiting a V-Hf ratio of 2.0 based on EDX, are shown in Figure 6. HfV2O7 composites is not feasible using conventional annealing experiments in air. The oxygen mobility in HfV2 at the formation temperature of crystalline HfV2O7 appears to be critical for the endeavor of HfV2-HfV2O7 composite formation. However, lowering of the synthesis temperature of crystalline HfV2O7 by using non-equilibrium-based synthesis approaches appears promising and is discussed in the following.

Phase Formation of HfV2O7
The formation of crystalline HfV2O7 at a substrate temperature of 350 °C using reactive magnetron sputtering has been demonstrated previously [49], but the effect of the O2 partial pressure has not been investigated yet. However, for the synthesis of bilayered HfV2-HfV2O7 by magnetron sputtering, the O2 partial pressure during the synthesis of HfV2O7 is expected to be decisive for the purity of the underlying HfV2 film, due to its high oxygen affinity. Hence, the influence of the O2 partial pressure on the phase formation of HfV2O7 was studied systematically for a synthesis temperature of 450 °C. The results of the structural analysis of synthesized thin films, exhibiting a V-Hf ratio of 2.0 based on EDX, are shown in Figure 6. The thin film sputtered at an O2 partial pressure of 0.05 Pa exhibits broad peaks at around 32°, 37°, 53°, and 63°. The peak positions coincide well with the ones of the cubic (Hf,V)O2 structure observed during annealing of HfV2 at 550 °C (see Figure 5). Cubic HfO2 usually constitutes the hightemperature polymorph, which is stable above 2600 °C at ambient pressure [86] but can be stabilized down to room temperature by V doping in specific atmospheres (oxidation state ≤ V 4+ ) [88]. Furthermore, cubic HfO2 thin films form by ion beam assisted deposition under oxygen deficiency and substrate cooling [89]. Thus, it is assumed that the formation of cubic (Hf,V)O2 obtained for the lowest O2 partial pressure is triggered by oxygen deficiency. This notion is supported by the fact that the increase in the O2 partial pressure from 0.05 to 0.09 Pa resulted in a sixfold decrease in deposition rate (with respect to the mass gain), which indicates a more pronounced poisoning of the target  Figure 5). Cubic HfO 2 usually constitutes the high-temperature polymorph, which is stable above 2600 • C at ambient pressure [86] but can be stabilized down to room temperature by V doping in specific atmospheres (oxidation state ≤ V 4+ ) [88]. Furthermore, cubic HfO 2 thin films form by ion beam assisted deposition under oxygen deficiency and substrate cooling [89]. Thus, it is assumed that the formation of cubic (Hf,V)O 2 obtained for the lowest O 2 partial pressure is triggered by oxygen deficiency. This notion is supported by the fact that the increase in the O 2 partial pressure from 0.05 to 0.09 Pa resulted in a sixfold decrease in deposition rate (with respect to the mass gain), which indicates a more pronounced poisoning of the target racetrack. For O 2 partial pressures of 0.09 and 0.13 Pa, all obtained peaks can be assigned to the HfV 2 O 7 structure [41], indicating the formation of phase pure HfV 2 O 7 based on XRD.

Thermal Expansion of Sputtered HfV 2 O 7
While numerous NTE materials have intensively been studied in bulk, studies on thin films are scarce [42][43][44][45]. To verify the NTE behavior of the synthesized HfV 2 O 7 thin films, temperature-dependent in situ XRD measurements were performed to measure the change in the lattice parameter upon heating. The results are shown in Figure 7.
Materials 2020, 13, x FOR PEER REVIEW 11 of 17 racetrack. For O2 partial pressures of 0.09 and 0.13 Pa, all obtained peaks can be assigned to the HfV2O7 structure [41], indicating the formation of phase pure HfV2O7 based on XRD.

Thermal Expansion of Sputtered HfV2O7
While numerous NTE materials have intensively been studied in bulk, studies on thin films are scarce [42][43][44][45]. To verify the NTE behavior of the synthesized HfV2O7 thin films, temperaturedependent in situ XRD measurements were performed to measure the change in the lattice parameter upon heating. The results are shown in Figure 7. A continuous increase in the lattice parameter between 20 and 120 °C was measured, whereas between 155 and 475 °C the lattice parameter decreased, indicating NTE. The obtained transition from PTE to NTE is in agreement with reported transition temperatures of HfV2O7 being around 120-130 °C produced by solid-state reaction synthesis approaches [40,41,64]. Additional wafer curvature measurements confirm this transition, as indicated by a change from increasing compressive to increasing tensile stress upon heating (not shown). The NTE properties of HfV2O7 originate from its openly-packed network structure consisting of octahedral HfO6 and tetrahedral VO4 units, constituting quasi-rigid building blocks that are interconnected by corner-sharing oxygen atoms [4,90,91]. Three of four oxygen atoms of the VO4 tetrahedron are shared with neighboring HfO6 octahedra, while one is shared with another VO4 tetrahedron, thereby forming a V2O7 group [92]. The lattice symmetry (space group: Pa3 ) restricts these O3V-O-VO3 bonds to retain an angle of 180°, allowing for transverse vibrations that may give rise to NTE [4,93]. The loss of the NTE properties at lower temperatures is related to structural transitions. HfV2O7 transforms upon cooling via an intermediate incommensurate structure to a 3 × 3 × 3 superstructure [41,64]. It has been shown for isostructural ZrV2O7 that the majority of the O3V-O-VO3 linkages bend away from 180° in the 3 × 3 × 3 superstructure [94][95][96]. However, the additional weak reflections of the HfV2O7 superstructure [97] could not be resolved by XRD in this work. Based on the presented data, the linear coefficients of thermal expansion were determined to be (2.8 ± 0.5) × 10 -5 K −1 (20 °C ≤ T ≤ 120 °C) and (−9.9 ± 0.9) × 10 −6 K −1 (155 °C ≤ T ≤ 475 °C), respectively, agreeing reasonably well with the reported linear coefficients of thermal expansion of 2.5 × 10 −5 K −1 [64] and −7.2 × 10 −6 K −1 [40]. After establishing a synthesis recipe for NTE material HfV2O7 that is characterized by a low required O2 partial pressure, the way is paved for the synthesis of bilayered HfV2-HfV2O7 composites using a two-staged sputtering process. A continuous increase in the lattice parameter between 20 and 120 • C was measured, whereas between 155 and 475 • C the lattice parameter decreased, indicating NTE. The obtained transition from PTE to NTE is in agreement with reported transition temperatures of HfV 2 O 7 being around 120-130 • C produced by solid-state reaction synthesis approaches [40,41,64]. Additional wafer curvature measurements confirm this transition, as indicated by a change from increasing compressive to increasing tensile stress upon heating (not shown). The NTE properties of HfV 2 O 7 originate from its openly-packed network structure consisting of octahedral HfO 6 and tetrahedral VO 4 units, constituting quasi-rigid building blocks that are interconnected by corner-sharing oxygen atoms [4,90,91]. Three of four oxygen atoms of the VO 4 tetrahedron are shared with neighboring HfO 6 octahedra, while one is shared with another VO 4 tetrahedron, thereby forming a V 2 O 7 group [92]. The lattice symmetry (space group: Pa3) restricts these O 3 V-O-VO 3 bonds to retain an angle of 180 • , allowing for transverse vibrations that may give rise to NTE [4,93]. The loss of the NTE properties at lower temperatures is related to structural transitions. HfV 2 O 7 transforms upon cooling via an intermediate incommensurate structure to a 3 × 3 × 3 superstructure [41,64]. It has been shown for isostructural ZrV 2 O 7 that the majority of the O 3 V-O-VO 3 linkages bend away from 180 • in the 3 × 3 × 3 superstructure [94][95][96]. However, the additional weak reflections of the HfV 2 O 7 superstructure [97] could not be resolved by XRD in this work. Based on the presented data, the linear coefficients of thermal expansion were determined to be (2.8 ± 0.5) × 10 −5 K −1 (20 • C ≤ T ≤ 120 • C) and (−9.9 ± 0.9) × 10 −6 K −1 (155 • C ≤ T ≤ 475 • C), respectively, agreeing reasonably well with the reported linear coefficients of thermal expansion of 2.5 × 10 −5 K −1 [64] and −7.2 × 10 −6 K −1 [40]. After establishing a synthesis recipe for NTE material HfV 2 O 7 that is characterized by a low required O 2 partial pressure, the way is paved for the synthesis of bilayered HfV 2 -HfV 2 O 7 composites using a two-staged sputtering process.

Phase Formation of HfV 2 -HfV 2 O 7 Bilayers
Besides a low O 2 partial pressure during reactive sputtering of HfV 2 O 7 , the synthesis temperature for HfV 2 O 7 is decisive for avoiding substantial oxygen incorporation into HfV 2 . Hence, the influence of synthesis temperature of HfV 2 O 7 , which was varied between 250 and 350 • C, on the phase formation of HfV 2 -HfV 2 O 7 composite was investigated by maintaining a low O 2 partial pressure of 0.09 Pa (see Figure 6). While a Nb passivation layer ensures protection of HfV 2 in air at room temperature (see Figure 3), additional bilayer depositions revealed that the deposition of a Nb interlayer, separating the two layers in the composite, does not prevent oxygen incorporation in HfV 2 during reactive sputtering of HfV 2 O 7 (not shown) and is therefore not discussed further. The results of the structural analysis are shown in Figure 8.

Phase Formation of HfV2-HfV2O7 Bilayers
Besides a low O2 partial pressure during reactive sputtering of HfV2O7, the synthesis temperature for HfV2O7 is decisive for avoiding substantial oxygen incorporation into HfV2. Hence, the influence of synthesis temperature of HfV2O7, which was varied between 250 and 350 °C, on the phase formation of HfV2-HfV2O7 composite was investigated by maintaining a low O2 partial pressure of 0.09 Pa (see Figure 6). While a Nb passivation layer ensures protection of HfV2 in air at room temperature (see Figure 3), additional bilayer depositions revealed that the deposition of a Nb interlayer, separating the two layers in the composite, does not prevent oxygen incorporation in HfV2 during reactive sputtering of HfV2O7 (not shown) and is therefore not discussed further. The results of the structural analysis are shown in Figure 8. For a synthesis temperature of 250 °C, peaks that are attributed to HfV2 are barely shifted [78] indicating a low concentration of interstitially incorporated oxygen into the structure during the reactive sputtering process of HfV2O7. However, no distinct HfV2O7 peaks were obtained indicating an amorphous or nanocrystalline structure. It is expected that amorphous HfV2O7 does not exhibit NTE. While crystalline ZrW2O8 exhibits NTE in its entire stability range [38,39], PTE was obtained for amorphous ZrW2O8 films synthesized by reactive sputtering [42].
An increase in synthesis temperature to 300 °C results in more pronounced oxygen incorporation in HfV2, as indicated by the increasing peak shift, which corresponds to the shift obtained for a 10weeks exposure time of uncapped HfV2 to air (see Figure 3). Besides, peaks of crystalline HfV2O7 emerge, demonstrating the first synthesis of a crystalline HfV2-HfV2O7 composite. The effect of minute concentrations of interstitially solved oxygen on the thermoelastic anomaly of HfV2 has not been studied yet and is hence proposed for future investigations. Figure 8b shows the corresponding film cross-section studied by STEM. The image suggests the formation of a bilayered structure with a defined interface separating the HfV2 bottom layer (1.7 µm) from HfV2O7 (0.5 µm) top layer. Based on the atomic number (z) contrast using high-angle annular dark-field imaging, the distinct formation of HfV2 (high z) and HfV2O7 (low z) is supported. The deposition rates for HfV2 and HfV2O7 were determined to be approximately 22 and 5 nm min −1 , respectively. Thus, tuning the volume phase fractions to be in line with the proposed ones (see Figure 1) can simply be achieved by adjusting the deposition times for the individual layer. However, further investigations are required to evaluate For a synthesis temperature of 250 • C, peaks that are attributed to HfV 2 are barely shifted [78] indicating a low concentration of interstitially incorporated oxygen into the structure during the reactive sputtering process of HfV 2 O 7 . However, no distinct HfV 2 O 7 peaks were obtained indicating an amorphous or nanocrystalline structure. It is expected that amorphous HfV 2 O 7 does not exhibit NTE. While crystalline ZrW 2 O 8 exhibits NTE in its entire stability range [38,39], PTE was obtained for amorphous ZrW 2 O 8 films synthesized by reactive sputtering [42].
An increase in synthesis temperature to 300 • C results in more pronounced oxygen incorporation in HfV 2 , as indicated by the increasing peak shift, which corresponds to the shift obtained for a 10-weeks exposure time of uncapped HfV 2 to air (see Figure 3). Besides, peaks of crystalline HfV 2 O 7 emerge, demonstrating the first synthesis of a crystalline HfV 2 -HfV 2 O 7 composite. The effect of minute concentrations of interstitially solved oxygen on the thermoelastic anomaly of HfV 2 has not been studied yet and is hence proposed for future investigations. Figure 8b shows the corresponding film cross-section studied by STEM. The image suggests the formation of a bilayered structure with a defined interface separating the HfV 2 bottom layer (1.7 µm) from HfV 2 O 7 (0.5 µm) top layer. Based on the atomic number (z) contrast using high-angle annular dark-field imaging, the distinct formation of HfV 2 (high z) and HfV 2 O 7 (low z) is supported. The deposition rates for HfV 2 and HfV 2 O 7 were determined to be approximately 22 and 5 nm min −1 , respectively. Thus, tuning the volume phase fractions to be in line with the proposed ones (see Figure 1) can simply be achieved by adjusting the deposition times for the individual layer. However, further investigations are required to evaluate whether these bilayered composite films allow for accurate determination of their integral temperature coefficient of elasticity and coefficient of thermal expansion. A further increase in the synthesis temperature of HfV 2 O 7 to 350 • C reveals the onset of amorphization of HfV 2 due to oxygen incorporation, as indicated by the pronounced peak shifting and broadening.

Conclusions
The HfV 2 -HfV 2 O 7 composite has been proposed as a material with temperature-independent thermophysical properties due to the combination of the anomalously increasing thermoelastic constants of HfV 2 and the negative thermal expansion of HfV 2 O 7 . Both the temperature coefficient of elasticity and the coefficient of thermal expansion were predicted to be near zero for a phase fraction of approximately 30 vol.% HfV 2 and 70 vol.% HfV 2 O 7 .
Two synthesis pathways for HfV 2 -HfV 2 O 7 composites were studied: (1) annealing of magnetron sputtered HfV 2 thin films in air to form a HfV 2 O 7 oxide scale on the thin film surface and (2) magnetron sputtering of HfV 2 O 7 /HfV 2 bilayers. The onset of the oxidation behavior of HfV 2 thin films is characterized by continuous interstitial incorporation of oxygen, occurring already at room temperature, finally yielding amorphization. Crystalline HfV 2 O 7 forms at 550 • C. The high oxygen mobility in HfV 2 is suggested to inhibit the formation of crystalline HfV 2 -HfV 2 O 7 composites by annealing HfV 2 in air. Reducing the formation temperature of crystalline HfV 2 O 7 down to 300 • C using reactive magnetron sputtering enables the synthesis of a crystalline bilayered HfV 2 -HfV 2 O 7 composite. The NTE properties of HfV 2 O 7 were verified for monolithic magnetron sputtered HfV 2 O 7 thin films using temperature-dependent in situ X-ray diffraction. Funding: Simulations were performed with computing resources granted by JARA-HPC from RWTH Aachen University under project JARA0131. Support for the operation of the accelerator laboratory in Uppsala by VR-RFI (contract 2017-00646-9) and the Swedish Foundation for Strategic Research (SSF, contract RIF14-0053) is gratefully acknowledged. The authors also kindly acknowledge funding through the Max Planck Fellow Program.

Conflicts of Interest:
The authors declare no conflict of interest.