Improved oxidation behavior of Hf0.11Al0.20B0.69 in comparison to Hf0.28B0.72 magnetron sputtered thin films

The oxidation resistance of Hf0.28B0.72 and Hf0.11Al0.20B0.69 thin films was investigated comparatively at 700 °C for up to 8 h. Single-phase solid solution thin films were co-sputtered from HfB2 and AlB2 compound targets. After oxidation at 700 °C for 8 h an oxide scale thickness of 31 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document}± 2 nm was formed on Hf0.11Al0.20B0.69 which corresponds to 14% of the scale thickness measured on Hf0.28B0.72. The improved oxidation resistance can be rationalized based on the chemical composition and the morphology of the formed oxide scales. On Hf0.28B0.72 the formation of a porous, O, Hf, and B-containing scale and the formation of crystalline HfO2 is observed. Whereas on Hf0.11Al0.20B0.69 a dense, primarily amorphous scale containing O, Al, B as well as approximately 3 at% of Hf forms, which reduces the oxidation kinetics significantly by passivation. Benchmarking Hf0.11Al0.20B0.69 with Ti–Al-based boride and nitride thin films with similar Al concentrations reveals superior oxidation behavior of the Hf-Al-based thin film. The incorporation of few at% of Hf in the oxide scale decelerates oxidation kinetics at 700 °C and leads to a reduction in oxide scale thickness of 21% and 47% compared to Ti0.12Al0.21B0.67 and Ti0.27Al0.21N0.52, respectively. Contrary to Ti–Al-based diborides, Hf0.11Al0.20B0.69 shows excellent oxidation behavior despite B-richness.

HfB 2 coatings have first been synthesized by chemical vapor deposition and have been studied in terms of wear resistance for tribological applications [18][19][20][21][22] .However, recently chemical composition quantification 23 and oxidation behavior 24 of direct current magnetron sputtered (DCMS) HfB 2 coatings have been reported.Glechner et al. 24 performed oxidation experiments on Hf 0.30 B 0.70 in synthetic air with significantly lower water vapor pressure than in ambient air.An increase in mass during dynamic oxidation experiments was observed up to a temperature of 1200 °C, followed by a decrease in mass.This behavior was explained by the evaporation of B 2 O 3 above 1200 °C.After 120 min of oxidation at 700 °C, a crystalline oxide containing O, Hf, and B with a scale thickness of 171 ± 21 nm was observed.After oxidation at 800 °C and 900 °C, an additional porous B-rich layer formed on top of a Hf-rich oxide containing O, Hf, and B 24 .However, when oxidation was carried out in ambient air within the same report, no boron oxide was observed which was attributed to the hygroscopic character of B 2 O 3 forming volatile compounds 5,24 .Based on secondary ion mass spectrometry (SIMS) data the authors concluded that Hf 0.30 B 0.70 films show very limited oxygen inward diffusion (through the scale) and that these films are hence, efficient oxygen diffusion barriers 24 .
For bulk samples, it was shown that the oxidation resistance of ZrB 2 and HfB 2 can be improved by the addition of SiC [25][26][27] .For coatings, it has been demonstrated that the addition of Si into TMB 2 (TM = Ti, Cr, Hf, Ta, W) 28,29 or Al 30 into TiB 2 leads to enhanced oxidation resistance: For Ti 0.29 Al 0.13 B 0.58 a protective Al-containing scale forms during oxidation at 700 °C for up to 8 h which reduced the oxide scale thickness from ~ 1350 nm for the Ti 0.29 B 0.71 coating to ~ 460 nm for the Ti 0.29 Al 0.13 B 0.58 coating 30 .
Furthermore, the chemical composition and especially the boron-to-metal ratio play a very important role in the oxidation resistance of TMB 2 .Under-stoichiometric Ti 0.41 B 0.59 (boron-to-metal ratio = 1.4) films have been reported to exhibit superior oxidation resistance compared to over-stoichiometric Ti 0.32 B 0.68 and Ti 0.27 B 0.73 (boron-to-metal ratio of 2.2 and 2.7 respectively) films due to the absence of the B-tissue phase, resulting in the formation of a denser TiO x oxide scale with reduced B 2 O 3 evaporation, thereby limiting oxide scale growth 31 .
One method to deposit stoichiometric Ti-Al-B coatings was shown by Stüber et al. 32 to co-sputter elemental Al to TiB 2 forming a Ti 0.84 Al 0.16 B 2.0 solid solution.Another deposition approach was chosen by Navidi Kashani et al. 33 who studied the effect of the B concentration on the mechanical properties and the oxidation resistance at 700 °C for up to 8 h for Ti-Al-B coatings.Using elemental B targets together with a TiAl target, Ti-Al-B coatings with an Al concentration range of 22 ± 2 at% and varying B concentration from 63 via 67 to 71 at% and hence, above and below stoichiometry, were synthesized.The most favorable combination of the oxidation resistance and mechanical behavior was obtained for the stoichiometric Ti 0.12 Al 0.21 B 0.67 coating with an oxide scale thickness of 39 ± 7 nm.In contrast, the over-stoichiometric Ti 0.10 Al 0.19 B 0.71 coating had a factor 5.2 thicker oxide scale of 204 ± 16 nm after 8 h 33 .Recently, oxidation experiments in ambient air at 700 °C, 800 °C, and 900 °C revealed a strong Al concentration dependence of the oxidation resistance.Stoichiometric films containing ≥ 21 at% of Al formed a dense and passivating oxide scale, while Al concentrations ≤ 15 at% resulted in the formation of a non-passivating scale 34 .
Here, we seek to explore, if the oxidation resistance of Hf 0.28 B 0.72 thin films can be further improved by Al incorporation.An oxidation temperature of 700 °C was chosen to study the onset of oxidation and compare it with literature data for potential coating materials.To this end, the oxidation resistance and oxide scale formation of Hf 0.11 Al 0.20 B 0.69 is compared to Hf 0.28 B 0.72 as well as to the available literature data for Hf-B, Ti-Al-B, and Ti-Al-N films exhibiting similar Al concentrations.

Results and discussion
Sputtering only the HfB 2 target leads to a film with 27.5 ± 3.5 at% Hf, 72.2 ± 3.6 at% B, and 0.3 ± 0.1 at% O according to elastic recoil detection analysis (ToF-ERDA) and will be referred to as Hf 0.28 B 0.72 in the following discussion.The B-to-Hf ratio of 2.6 indicates that the film is B over-stoichiometric compared to stoichiometric HfB 2 .
This B-richness is in good agreement with other literature reports for DCMS-deposited HfB x thin films from stoichiometric HfB 2 compound targets 24,35 .The deviation of the film composition from the target composition for constituents with large differences in mass has been proposed to be caused by preferential resputtering of one of the sputtered constituents [36][37][38] and/or preferential sputtering of the lighter constituent along the target normal 39 .For TiB x thin films, Neidhardt et al. 40 have demonstrated experimentally and with simulations that there is a complex interaction between the angular distribution of the sputtered flux as well as gas scattering which influences the film composition.The preferential emission of B along the target normal is observed while the angular distribution of Ti is much broader and similar to the ideal spherical cosine distribution.These emission characteristics are modified at higher pressures by the effect of gas scattering.Because of the larger atomic radius, Ti has a more efficient energy transfer with Ar and a shorter thermal mean free path (MFP) than B. This results in more effective stopping and thus less scattering of Ti than B. The complex interdependence between emission characteristics and scattering processescan explain the linear dependence of the Ti/B ratio with pressure and the substrate-to-target distance.As a result, at low pressure, no scattering collisions are expected for B leading to a Ti-deficient film.At high pressures, however, B is scattered preferentially, resulting in a B-deficient film 40 .Mráz et al. 41 studied the angle-resolved thin film composition evolution during sputtering of a Cr 2 AlC compound target by varying the angle α between the target normal and the substrate surface between 0° and 67.5°.As the mass differences in this system are larger than for TiB 2 , significant differences between film and target composition were reported: for α ≤ 22.5°heavy element deficient thin films compared to the target composition are observed where the deficiency decreases with increasing pressure.However, this heavy element deficiency is decreasing with increasing angle between the target normal and the substrate surface 41 .
In Hf 0.28 B 0.72 the mass difference between the target constituents is larger and α = 0°.Hence, the angular dis- tribution of Hf and B differs more severely than in TiB 2 40 and the formation of Hf-deficient thin films is expected based on the reasoning in 40 and 41 .
To estimate the effect of gas phase scattering the MFP was calculated using Eq.(1) which includes the deposition temperature (T = 673 K), deposition pressure (p = 0.6 Pa), covalent radii of Ar (r Ar = 71 pm 42 ), Hf (r Hf = 208 pm 42 ) or B (r B = 87 pm 42 ).
The MFP of B is 14.0 cm leading at a substrate-to-target distance of 10 cm to an average number of 0.7 collisions.An average of < 1 collision implies that B atoms will likely not undergo collisions on route to the target and will therefore likely be incorporated in the growing thin film.In contrast, the MFP of Hf is 4.5 cm resulting in 2.2 collisions on average, causing less efficient transport of Hf to the surface of the growing film which is in turn leading to a Hf-deficient and hence B-rich thin film 40 .Therefore, the here observed Hf deficiency (compared to the target composition) can be attributed to preferential emission of B along the target normal as well as efficient scattering of Hf.
When AlB 2 is co-sputtered with HfB 2 at the same angle α between the target normal and the substrate surface, the B over-stoichiometry decreases to a B/(Hf + Al) ratio of 2.3 which corresponds to 69.2 ± 3.5 at% B and Al and Hf concentrations of 19.8 ± 2.2 at% and 10.5 ± 1.2 at%, respectively.Henceforth, the film will be referred to as Hf 0.11 Al 0.20 B 0.69 .The oxygen concentration was 0.5 ± 0.1 at%.The reduced B over-stoichiometry compared to Hf 0.28 B 0.72 can be explained by the smaller mass (m Al = 26.98 u) and atomic radius (r Al = 118 pm 42 ) of Al compared to Hf with m Hf = 178.49u and r Hf = 208 pm 42 , which leads to an MFP of 9.8 cm and therefore to an average number of 1.0 collisions and thus to more efficient Al transport towards the surface of the growing film.Hence, as more metal is incorporated into the film the B-richness is reduced.
Figure 1 depicts X-ray diffraction (XRD) patterns of the as-deposited films grown on α-Al 2 O 3 (0001) to investigate the phase formation.The positions of HfB 2 and AlB 2 reference lines according to the JCPDS cards 00-038-1398 and 00-008-0216 are highlighted with circle and triangle markers, respectively.The diffractogram of Hf 0.28 B 0.72 reveals the formation of a single-phase hexagonal HfB 2 structure.The peak positions of the Hf 0.11 Al 0.20 B 0.69 films are located between those of the binary HfB 2 and AlB 2 phases, indicating the formation of a solid solution.Based on our DFT data this solid solution is metastable as witnessed by the positive mixing enthalpy of 0.08 eV/atom for stoichiometric Hf 0.12 Al 0.21 B 0.67 .
(1)  43 , the c parameter exhibits a non-linear behavior.The deviation of the experimental to the density functional theory (DFT) calculated lattice parameter of Hf 0.28 B 0.72 is − 0.3% for a and − 0.8% for c and of Hf 0.11 Al 0.20 B 0.69 it is + 0.9% for a and − 0.6% for c, respectively.According to Paier et al., deviations of up to 2.1% between DFT-calculated and experimental lattice parameters have to be expected based on the here-applied exchange-correlation functionals 44 .Hence, very good agreement between prediction and experiment is obtained even though the B over-stoichiometry and the actual stress state of the as-deposited thin films were not considered.
To study the morphology, cross-sectional and plan view scanning transmission electron microscopy (STEM) images in bright-field (BF) mode were taken which are shown in Fig. 3. Hf 0.28 B 0.72 exhibits a small-grained, dense, and nanocolumnar microstructure.While this observation is consistent with literature reports from Mayrhofer et al. 3 where the excess boron is reported to surround the columns and between smaller sub-columns, the formation of a B-rich tissue phase cannot be inferred from the images presented here.
The ternary solid solution film also shows a dense, columnar morphology; however, the column size is increased as depicted in Fig. 3b,d compared to Fig. 3a,c.The hardness and elastic modulus determined for the Hf 0.28 B 0.72 film are 38 ± 2 GPa and 541 ± 19 GPa, respectively, which is reduced to 30 ± 1 GPa and 397 ± 9 GPa after Al addition forming Hf 0.11 Al 0.20 B 0.69.As depicted in Fig. 2, the elastic modulus of Hf 0.28 B 0.72 matches the values predicted by DFT within the error bar.The experimentally determined elastic modulus of Hf 0.11 Al 0.20 B 0.69 is 18.5% higher than the DFT calculated value of 334 GPa, which is according to Paier et al. 44 within the expected range of bulk modulus deviations between different exchange-correlation functionals and experiments.As the Al concentration is increased from 0 at% via 20 at% to 33 at% in Hf 0.33−x Al x B 0.67 , the cohesive energy increases linearly from − 7.87 eV/atom via − 6.44 eV/atom to − 5.68 eV/atom, respectively.The increase in cohesive energy indicates that the reduction in elastic modulus can be explained by bond weakening caused by Al addition as observed in Ti 0.33−x Al x B 0.67 34 .This bond weakening leads to an elastic modulus reduction of ~ 7 GPa per at% of Al in Hf 0.33−x Al x B 0.67.Al concentration induced bond weakening has also been reported for Ti 0.33−x Al x B 0.67 , where also a ~ 7 GPa reduction was observed for each at% Al added 34 .
The time-dependent oxidation resistance at 700 °C of the ternary solid solution Hf 0.11 Al 0.20 B 0.69 thin film was compared to the binary Hf 0.28 B 0.72 thin film.In Fig. 4 the diffractograms of the as-deposited and oxidized films after 4 and 8 h are presented.The positions of HfO 2 reference lines according to the JCPDS card 01-070-2832 is shown by the square marker in addition to the positions of AlB 2 and HfB 2 from Fig. 1.It is evident that for Hf 0.28 B 0.72 crystalline HfO 2 is formed between 4 and 8 h of oxidation.In contrast, for Hf 0.11 Al 0.20 B 0.69 no additional diffraction signals are visible, even after 8 h of oxidation, indicating the absence of X-ray crystalline oxide scales.
Figure 5 shows a high angle annular dark field (HAADF) high-resolution STEM (HRSTEM) micrograph along with corresponding energy dispersive X-ray spectroscopy (EDX) maps and line scans capturing the oxide scale including the interfacial region as well as the underlying unaffected thin film for oxidized Hf 0.28 B 0.72 and Hf 0.11 Al 0.20 B 0.69 after 8 h at 700 °C.It should be noted that due to the difference in oxide scale thickness of Hf 0.28 B 0.72 and Hf 0.09 Al 0.21 B 0.70 , the scale bars in the images and the corresponding EDX maps differ.
The HAADF image of Hf 0.28 B 0.72 , see Fig. 5a, shows a bi-layered oxide scale morphology, where the ~ 100 nm thick top layer is more porous than the ~ 160 nm thick layer below.The corresponding EDX line scan and maps indicate the formation of an O, Hf, and B-containing oxide.The B concentration gradually decreases while the Hf concentration increases from the oxide surface towards the film.The fact that Pt, utilized during lamellae preparation, was able to penetrate the oxide scale could amplify the decrease in the Hf concentration measured in the line scan.Furthermore, literature shows that B quantification in TEM-EDX spectra is strongly affected by the lamellae thickness due to preferential B-K α X-ray absorption with increasing film thickness 45 .Hence, we 0.0 0.1 0.2 0.3 3.00 assume that the increasing lamellae thickness towards the substrate is causing the above-discussed composition variations.
Similar to the present study, Glechner et al. 24 reported a bi-layered oxide scale formation for Hf 0.30 B 0.70 thin films oxidized in synthetic air, with a B-rich surface-near oxide layer and a Hf-rich oxide layer underneath based on SIMS data.However, the porous boron oxide layer was based on electron microscopy cross-sections only described for Hf 0.30 B 0.70 thin films oxidized above 800 °C and the boron oxide on the surface only for synthetic air experiments.When the experiment was repeated in ambient air, no boron oxide was found on the surface which was explained by the formation of volatile H m BO n in the presence of water vapor which evaporates earlier.However, there is no detailed description of the composition and morphology of the oxide scale at 700 °C for the oxidation experiment conducted in ambient air 24 .
In this study, after oxidation at 700 °C for 8 h, the formation of crystalline HfO 2 is detected by XRD concurrent with EDX which indicates that the oxide scale consists of O, Hf, and B, while the HRSTEM image shows a porous morphology.Based on the oxidation model for bulk HfB 2 from Parthasarathy et al. 46 , for oxidation temperatures < 1000 °C a glassy B 2 O 3 film is expected to form on top of a porous, crystalline HfO 2 scale, where the pores are filled with B 2 O 3 .At temperatures > 1000 °C B 2 O 3 at the surface starts to evaporate 46 .
However, experiments and thermodynamic calculations with water-saturated synthetic air, with a water vapor pressure of 0.031 atm at 25 °C, showed that in the presence of water vapor B 2 O 3 reacts and forms hydrated B-O compounds like HBO 2 and H 3 BO 3 which evaporate below 1000 °C5,47 .
There are opposing reports in the literature about the rate of B 2 O 3, or more specifically H m BO n evaporation, at ~ 700 °C in ambient air.Some authors report an almost complete absence of the B 2 O 3 phase for Ti-B films oxidized in synthetic air with 40% humidity 47 and Hf-B films oxidized in ambient air 24 .However, there are also reports of the presence of B 2 O 3 in the oxide scale forming on TiB 2 bulk samples oxidized in ambient air at this temperature 48 or even at 1200 °C for 1 h of oxidation of ZrB 2 bulk samples oxidized in water-saturated synthetic air 5 .This suggests that a Hf and B containing oxide scalesimilar to the oxidation model for bulk HfB 2 has formed on the here investigated Hf 0.28 B 0.72 where some B 2 O 3 has started to form H m BO n and thus has evaporated leaving pores in the oxide scale in the vicinity of the surface.However, also here the discussed limitation in B quantification for films with decreasing thickness applies 45 .Therefore, we conclude that based on the EDX and TEM results, a Hf-B containing oxide scale has formed where some B has formed volatile H m BO n and evaporated leaving pores.Still, significant amounts of B are still present in the scale after oxidation at 700 °C for 8 h.The oxide scale forming on Hf 0.11 Al 0.20 B 0.69 is dense and contains besides O also equal amounts of Al and B as well as approximately 3 at% of Hf.The EDX maps show that the elements are homogeneously distributed.Based on ERDA measurements, the O concentration in the remaining film is below 1 at% and thus significantly lower than the ~ 18 at% O concentration measured by EDX.This observation can be explained by the sample preparation procedure.For the HRSTEM EDX analysis, a ∼ 60 nm thin lamella is prepared in the focused ion beam (FIB) and subsequently transferred through ambient air.Thus, it is reasonable to assume that oxidation by atmosphere exposure causes the formation of native oxides on each side of the lamella, which will affect the measured amount of O in EDX.Therefore, the Hf 0.28 B 0.72 film exhibits ~ 10 at% of O.The 8 at% higher O content in Hf 0.11 Al 0.20 B 0.69 might be related to the higher Al concentration in the film.Moreover, EDX is not well positioned to quantify the boron content, since light elements have low characteristic X-ray energies and a low yield resulting in difficulties in accurate resolution of their signals against background noise and limitations in detector sensitivity 49 .As a consequence, the accuracy of boron concentration measurements obtained by EDX is limited, and previous studies have reported systematic overestimation of boron concentration in Hf-B thin films using this technique when compared to ERDA 23 .Nevertheless, from the here obtained HRSTEM and EDX data it is evident that Hf 0.11 Al 0.20 B 0.69 forms a dense oxide scale containing around 30 at% of Al and B as well as 3 at% of Hf, as determined within the known limitations of EDX.
Based on XRD the oxide forming on Hf 0.11 Al 0.20 B 0.69 is amorphous.Transmission electron microscopy (TEM) dark field (DF) images of the interface between the oxide scale and film, which are depicted in Fig. 6, also show a mostly amorphous oxide scale.However, nanocrystalline regions, marked with red arrows, are visible preferentially in the vicinity of the interface which may have formed during oxidation or electron beam exposure.The unoxidized film is crystalline and the interface between oxide and film is well defined.
To study the oxide growth kinetics, Fig. 7 shows the resulting oxide scale thicknesses of the film determined by STEM imaging on FIB-prepared lamellae as a function of oxidation time at 700 °C.Also, the relevant literature data, namely magnetron sputtered Hf 0.30 B 0.70 24  To study the oxidation behavior and the effect of the Al addition on the oxidation kinetics, a power law fit according to 50 was utilized.compared to the available literature data from Glechner et al. 24 and Navidi Kashani et al. 33,34 .The dashed line for Hf 0.28 B 0.72 is included as a guide for the eye.
and cubic oxidation kinetics, respectively.The pre-exponential factor K ′ and the time exponent n determined for the films compared in Fig. 7 are listed in Supplementary Table 1.Initially, the Hf 0.28 B 0.72 thin film shows a parabolic oxidation behavior with n = 0.49.During that initial state, no oxide peaks occur in XRD diffractograms, indicating the formation of an X-ray amorphous Hf and B containing oxide scale.However, beyond 4 h of oxidation at 700 °C, the oxide scale crystallizes and pore formation in the oxide scale increases leading to the acceleration of oxidation kinetics as indicated by the dashed line to guide the eye and thus a non-passivating behavior.The observation that Hf 0.29 B 0.71 shows initially slow oxidation kinetics at 700 °C which then increases after a certain time has been reported previously 24 .
In contrast, Hf 0.11 Al 0.20 B 0.69 shows with n = 0.16 a much slower, sub-cubic oxidation rate.The substantially improved oxidation resistance of Hf 0.11 Al 0.20 B 0.69 compared to Hf 0.29 B 0.71 can be rationalized based on the XRD, EDX, HRSTEM and TEM data: firstly, the morphologies of the formed oxides vary significantly: While on Hf 0.29 B 0.71 the formation of a porous, crystalline oxide scale is observed, oxidation of Hf 0.11 Al 0.20 B 0.69 results in a dense, amorphous and passivating oxide scale.Secondly, the chemical composition of the oxide varies significantly: while Hf 0.29 B 0.71 forms a Hf-rich oxide scale containing significant amounts of B, Hf 0.11 Al 0.20 B 0.69 forms an oxide scale with only small amounts of Hf but high Al and B concentration.
Next, the influence of the transition metal on the oxidation resistance is discussed by comparing Hf 0.09 Al 0.21 B 0.70 with over-stoichiometric Ti 0.10 Al 0.19 B 0.71 and stoichiometric Ti 0.12 Al 0.21 B 0.67 33 .It is evident from Fig. 7 that Ti 0.12 Al 0.21 B 0.67 and Hf 0.11 Al 0.20 B 0.69 show a similar oxidation behavior, however, the oxide thickness of Ti 0.12 Al 0.21 B 0.67 is 26% thicker than for Hf 0.11 Al 0.20 B 0.69 after 8 h of oxidation.It is noteworthy that the here investigated Hf 0.11 Al 0.20 B 0.69 is both, over-stoichiometric with a boron-to-metal ratio of 2.2 and superior in oxidation behavior compared to stoichiometric Ti 0.12 Al 0.21 B 0.67 .As pointed out in the introduction the boron-to-metal ratio plays a major role in the oxidation resistance of Ti-Al-B thin films as a comparison with the over-stoichiometric Ti 0.10 Al 0.19 B 0.71 33 illustrates.The over-stoichiometric Ti 0.10 Al 0.19 B 0.71 thin film shows an increase in oxide scale thickness by factor 5.2 resulting in 204 ± 16 nm compared to stoichiometric Ti 0.12 Al 0.21 B 0.67 after oxidation at 700 °C for 8 h.The reason is the formation of a porous oxide scale, which is enriched in Ti and O as well as depleted in Al and B and is therefore less oxidation resistant.In contrast, the stoichiometric Ti 0.12 Al 0.21 B 0.67 film forms a dense, Al-rich oxide scale which is highly oxidation resistant 33 as a passivating scale is formed 34 .The Hf 0.11 Al 0.20 B 0.69 film shows excellent oxidation resistance even though a significant amount of B is in the Al-rich oxide scale.
Another advantage of Hf 0.11 Al 0.20 B 0.69 compared to Ti-Al-B films with a similar Al concentration 33 is the enhancement of mechanical properties.The highly oxidation-resistant Ti 0.12 Al 0.21 B 0.67 exhibits hardness and elastic modulus values of 19 ± 1 GPa and 395 ± 12 GPa, respectively.The elastic modulus is similar to the Hf 0.11 Al 0.20 B 0.69 film, however, the hardness is significantly reduced compared to 30 ± 1 GPa of Hf 0.11 Al 0.20 B 0.69 .Higher hardness is observed for the over-stoichiometric Ti 0.10 Al 0.19 B 0.71 with 24 ± 1 GPa, but still significantly lower than for Hf 0.11 Al 0.20 B 0.69 .Moreover, the elastic modulus of over-stoichiometric Ti 0.10 Al 0.19 B 0.71 is only 330 ± 9 GPa, and the oxidation resistance is much inferior with an oxide scale thickness that is 558% larger than the one of Hf 0.11 Al 0.20 B 0.69 33 .Finally, Hf 0.11 Al 0.20 B 0.69 is compared to the industrial benchmark coating Ti 0.27 Al 0.21 N 0.52 34 .After oxidation at 700 °C for 8 h, the nitride film exhibits an oxide scale thickness of 57 ± 5 nm corresponding to an 84% increased thickness compared to Hf 0.11 Al 0.20 B 0.69 .This data shows that Hf 0.11 Al 0.20 B 0.69 exhibits superior combined mechanical and oxidation behavior at 700 °C compared to Ti-Al-based diborides and nitrides, even when the boron-tometal ratio is over-stoichiometric.The fact that the oxidation resistance of Hf-Al-based diborides, in contrast to Ti-Al-based diborides, is not adversely affected by over-stoichiometric B concentrations results in less stringent requirements concerning the composition distribution obtained within an industrial deposition system.
Vol:.( 1234567890 This study highlights the untapped application potential of Hf-Al-based diboride thin films for oxidation protection.Very significantly, in comparison to Ti-Al-based diborides, Hf-Al-based diborides have the advantage of forming passivating oxide scales for B-rich thin film compositions, thereby improving processability while exhibiting a high hardness of 30 ± 1 GPa.

Experimental methods
The Hf 0.28 B 0.72 and Hf 0.11 Al 0.20 B 0.69 thin films were synthesized by direct current magnetron sputtering using an in-house built high-vacuum growth system with a base pressure of less than 6 × 10 -4 Pa at the deposition temperature of 400 °C.Stoichiometric 2-inch HfB 2 (99.8% purity) and AlB 2 (99.8% purity) compound targets (Plansee Composite Materials GmbH) were mounted at a 45° angle with respect to the substrate normal as depicted in Supplementary Fig. 1.
For the Hf 0.11 Al 0.20 B 0.69 deposition, the HfB 2 and AlB 2 targets were co-sputtered at a constant power density of 3.9 W/cm 2 and 7.4 W/cm 2 , respectively.To deposit the reference Hf 0.28 B 0.72 thin film only the HfB 2 target was operated at a constant power density of 7.4 W/cm 2 .All films were grown in pure Ar atmosphere (99.9999% purity) at 0.6 Pa and floating substrate bias potential onto 10 × 10 mm 2 α-Al 2 O 3 (0001) substrates with a targetto-substrate distance of 10 cm.Substrate rotation was used to ensure a homogenous composition.The deposition time was chosen so both film thicknesses were 1 µm with deposition rates of 13.3 nm/min for Hf 0.11 Al 0.20 B 0.69 and 11.1 nm/min for Hf 0.28 B 0.72 .
The samples were oxidized in ambient air at 700 °C for exposure times of 1, 4, and 8 h using a GERO tube furnace.The as-deposited samples were placed into an Al 2 O 3 crucible with an attached Ni/Ni-Cr thermocouple and transferred into the pre-heated furnace at 700 °C and immediately removed from the furnace after the exposure time.
For the structural analysis, a Bruker AXS D8 Discover General Area Detector Diffraction System was used where the Cu K α (λ = 1.5406Å) X-ray source was set to 40 kV at a current of 40 mA and the incident angle was fixed at 15°.The peaks were fitted with a pseudo-Voigt II function using the TOPAS software V.3.Based on the fitted peak positions, lattice parameters were calculated using the CellCalc software V. 2.10.
For determining the chemical composition, time-of-flight elastic recoil detection analysis (ToF-ERDA) was carried out at the Tandem Accelerator Laboratory of Uppsala University [https:// doi.org/https:// doi.org/ 10. 1088/ 1748-0221/ 17/ 04/ P04011].Recoils were generated using a 36 MeV 127 I 8+ primary ion beam and timeenergy coincidence spectra were recorded with a detector telescope, equipped with thin carbon foils for ToF measurement and a gas detector for energy discrimination 51 .Further information on the detector telescope and the ToF setup is available in reference 52 .Uncertainties regarding the stopping power values as well as the detection efficiency dominate the measurement uncertainty 53 and beam straggling and multiple scattering diminish the depth resolution for heavy transition metals like Hf 54 .For Ti-B films it has been reported that ERDA and Rutherford backscattering spectrometry deviate less than 3% and thus coincide very well 23 .Since ERDA is employed as a stand-alone technique in the present work, a maximum total uncertainty of 5% of the deduced values is assumed for the light element B and aliquot fractions thereof for the metals (Hf,Al).The composition was quantified from the surface near region of the depth profile to prevent an influence of multiple scattering on the chemical composition analysis 23,54,55 .
The mechanical properties of as-deposited films were determined by quasistatic nanoindentation in a Hysitron TI-900 TriboIndenter.Indentation modulus as well as hardness were derived according to the method from Oliver and Pharr 56 .A diamond indenter (Poisson's ratio of 0.07, elastic modulus of 1140 GPa) with Berkovich geometry was used.Load-controlled measurements with 6 mN were applied, resulting in maximum contact depths of 60 nm for as-deposited samples, corresponding to < 6% of the film thickness.50 indents were performed for each sample and the indenter area function was derived from measurements on fused silica.Measured indentation modulus values were converted to the elastic modulus with a Poisson's ratio of 0.154 for HfB 2 , obtained by averaging published data from density functional theory (DFT) calculations [57][58][59][60] .For Hf 0.11 Al 0.20 B 0.69 a Poisson's ratio of 0.131 was used, based on the interpolation between HfB 2 and the calculated value of 0.118 for AlB 2 34 .Scanning transmission electron microscopy (STEM) was performed to study the microstructure of as-deposited and oxidized samples in a dual-beam microscope FEI Helios NanoLab 600 with a Ga + focused ion beam (FIB) and equipped with an EDAX Octane Elect energy dispersive X-ray spectroscopy (EDX) detector.First, the lamellae were prepared with a standard lift-out procedure, and then STEM imaging was carried out at an acceleration voltage of 30 kV and a current of 50 pA.
Transmission electron microscopy (TEM) was utilized to study the microstructure and chemical composition using a Thermo Fisher Scientific Titan Themis 200 G3 equipped with a SuperX detector for EDX mapping.The TEM lamellae were first prepared in the FEI Helios NanoLab 600 as well and were subsequently thinned and polished at 5 kV using a Tescan Lyra FIB/SEM system.

Computational methods
The lattice constants of relaxed Hf 0.33-x Al x B 0.67 with x = 0, 0.04, 0.09, 0.17, 0.24, and 0.33 were calculated based on DFT 61,62 as implemented in the Vienna ab initio simulation package (VASP) 63,64 .The projector augmentedplane-wave method 65,66 was used with the generalized gradient approximation with PBE 67 functional described the exchange-correlation effects.The plane-wave cutoff energy was set to 500 eV, and a 7 × 7 x 7 k-point mesh was used.For the calculations, 3 × 3 x 3 supercells of the AlB 2 -type (P6/mmm) were created.The metal sublattice was populated with Hf and Al using the Special Quasi-random Structures method as implemented in 68 .The X-ray diffraction (XRD) patterns from the calculated structures were determined with the VESTA software 69 .The elastic constants were calculated in the form of the full stiffness tensor (C ij ), from which the Young's modulus was derived with the Voigt-Reuss-Hill scheme 70 .The stiffness tensor was obtained with the strain-stress method 71 .

67 Fig. 2 .
Fig. 2. Change in lattice parameters (a) a and (b) c as well as (c) the elastic modulus as a function of the Al concentration x in Hf 0.33−x Al x B 0.67 calculated by DFT and compared with experimental data.

Fig. 3 .
Fig. 3. STEM images of as-deposited Hf 0.28 B 0.72 (a and c) and Hf 0.11 Al 0.20 B 0.69 (b and d) thin films in BF-mode from cross-sectional (a and b) and plan view (c and d).

Fig. 6 .
Fig. 6.Typical cross-sectional DF TEM image of the oxide scale of Hf 0.11 Al 0.20 B 0.69 .The red arrows mark the nanocrystalline areas.
Figure 2 shows the calculated and experimentally obtained lattice constants a and c as a function of Al concentration x in Hf 0.33−x Al x B 0.67.While the calculated lattice constant a decreases almost linearly with the Al concentration following Vegard´s law Fig. 1.Diffraction patterns of as-deposited Hf 0.28 B 0.72 and Hf 0.11 Al 0.20 B 0.69 thin films.Vol:.(1234567890)Scientific Reports | (2024) 14:21653 | https://doi.org/10.1038/s41598-024-72134-3www.nature.com/scientificreports/ 33l 2 O 3 and B 2 O 3 containing oxide scale is slowed down by the incorporation of approximately 3 at% of Hf even if the B concentration in the Hf 0.11 Al 0.20 B 0.69 oxide scale seems to be larger than for Ti 0.12 Al 0.21 B 0.67 .Hf 0.11 Al 0.20 B 0.69 outperforms well-established Ti 0.27 Al 0.21 N 0.5233regarding oxidation behavior since the oxide scale thickness of the nitride is 84% thicker than on Hf 0.11 Al 0.20 B 0.69 .