Age hardening in superhard ZrB 2 -rich Zr 1-x Ta x B y thin ﬁlms

We recently showed that sputter-deposited Zr 1-x Ta x B y thin ﬁlms have hexagonal AlB 2 -type columnar nanostructure in which column boundaries are B-rich for x < 0.2, while Ta-rich for x ≥ 0.2. As-deposited layers with x ≥ 0.2 exhibit higher hardness and, simultaneously, enhanced toughness. Here, we study the mechanical properties of ZrB 2.4 , Zr 0.8 Ta 0.2 B 1.8 , and Zr 0.7 Ta 0.3 B 1.5 ﬁlms annealed in Ar atmosphere as a function of annealing temperature T a up to 1200 ° C. In-situ and ex-situ nanoindentation analyses reveal that all ﬁlms undergo age hardening up to T a = 800 ° C, with the highest hardness achieved for Zr 0.8 Ta 0.2 B 1.8 (45.5 ± 1.0 GPa). The age hardening, which occurs without any phase separation or decomposition, can be explained by point-defect recovery that enhances chemical bond density. Although hardness decreases at T a > 800 ° C due mainly to recrystallization, column coarsening, and planar defect annihilation, all layers show hardness values above 34 GPa over the entire T a range.

Although TM diborides are categorized as hard materials [24] , this alone is not enough to prevent failure.Hardness is usually accompanied by brittleness causing crack formation and propagation at the presence of high stresses [25] .To avoid brittle cracking, diborides require to be both hard and relatively ductile.We recently showed that alloying ZrB 2 films with Ta causes an ~20% increase in hardness with a simultaneous increase of ~30% in toughness [26] .Zr 1-x Ta x B y alloys with x ≥ 0.2 exhibited a self-organized columnar core/shell nanostructure in which crystalline Zr-rich cores provide high hardness, while disordered Ta-rich, B-deficient, metallic-glasslike shells give enhanced ductility [26,27] .
In addition to increasing mechanical properties, the stability of nanostructure and mechanical properties at high temperatures are also essential.Here, thermal stability of ZrB 2.4 , Zr 0.8 Ta 0.2 B 1.8 , and Zr 0.7 Ta 0.3 B 1.5 thin films are studied at annealing temperatures T a ranging from 600 to 1200 °C in Ar atmosphere.We demonstrate that crystal structure does not change up to 1200 °C.Hardness increases for all films up to 800 °C due to retaining planar defects and chemical-bond recovery, followed by a decrease at T a > 800 °C during recrystallization.
The films are grown on Al 2 O 3 (0 0 01) in a CC800/9 CemeCon system equipped with rectangular (8.8 × 50 cm 2 ) ZrB 2 and Ta targets.The base pressure is 3.8 × 10 −6 Torr (0.5 mPa).The deposition temperature is 475 °C.ZrB 2.4 films are deposited by DCmagnetron sputtering (DCMS) at a 50 0 0-W power and a negative 100-V DC-substrate bias.The alloys are deposited in a hybrid scheme [28] in which the ZrB 2 target is continuously sputtered by 50 0 0-W DCMS, while the Ta magnetron is operated in highpower-impulse-magnetron-sputtering (HiPIMS) mode with 50-μs pulses.The average powers (and pulse frequencies) applied to the Ta target for growing Zr 0.8 Ta 0.2 B 1.8 and Zr 0.7 Ta 0.3 B 1.5 are 1200 W (200 Hz) and 1800 W (300 Hz), respectively.A negative 100-V bias is applied synchronously to substrates with a 30-μs offset and 100μs pulse width.Deposition times are adjusted such that all films are ∼1.6-μmthick.
θ -2 θ X-ray diffraction (XRD) scans are carried out using a PANalytical Empyrean X-ray diffractometer with a Cu K α source ( λ = 1.5406Å) to determine the crystal structure and orientations of the layers.Substrate curvatures are determined from XRD rocking-curve measurements to obtain films' residual stresses, based on modified Stoney equation [26,29,30] .To obtain film compositions, time-of-flight elastic recoil detection analyses carried out in a tandem accelerator with a 36 MeV 127 I 8 + probe beam.Chemical bonding in the layers is evaluated by X-ray photoelectron spectroscopy (XPS) using a Kratos Axis Ultra DLD instrument employing monochromatic Al K α radiation (h ν = 1486.6eV).For XPS experimental details see reference 26.
Plan-view transmission-electron-microscopy (TEM) analyses are carried out in a monochromated and double-corrected FEI Titan 3 60-300 electron microscope operated at 300 kV.Imaging is performed using scanning TEM (STEM) high-angle annular dark-field (HAADF-STEM), annular bright-field (ABF) imaging and conventional TEM modes.Energy-dispersive X-ray (EDX) and electron energy-loss spectroscopy (EELS) elemental maps are obtained using SuperX and GIF Quantum ERS spectrometers embedded in the FEI instrument.TEM specimens are prepared by focused ion beam method using Carl Zeiss Cross-Beam 1540 EsB system.
The films are annealed in a furnace with a continuous Ar flow.The annealing peak temperature T a is varied from 600 °C to 1200 °C in 100 °C increments.After reaching T a , with heating rate of 10 °C/min, temperature is held constant for 30 min.Then, samples are cooled down to room temperature, while the furnace is turned off.
Nanoindentation analyses are performed in an Ultra-Micro Indentation System at room temperature (ex-situ measurements) and Anton-Paar-TriTec UNHT 3 ultra-nanoindentation system at high temperature (in-situ measurements) with sharp Berkovich diamond tips calibrated using a fused-silica standard.The in-situ measurements are performed in Ar atmosphere up to 800 °C, in which the tip is heated with infrared heating.For hardness H and elastic modulus E measurements, the films are indented 35 times using a fixed load of 12 mN, while indention depths are maintained below 10% of film thickness.The results are analyzed using Oliver and Pharr method [31] .E are calculated from reduced elastic moduli using the diamond indenter's elastic modulus (1141 GPa) and Poisson's ratio ν = 0.07.The ν values of Zr 1-x Ta x B y required for obtaining E are unknown.To provide E values, ν are used from a linear interpolation between the Poisson ratio of ZrB 2 (0.13 [32] ) and that of TaB 2 (0.21 [33,34] ).
XRD patterns of as-deposited and annealed films reveal that all peaks detected in the 2 θ range from 20 °to 100 °originate from the hexagonal AlB 2 -type structure.All reflections are preserved throughout the entire T a range, and no new peaks are observed.Hence, only (0 0 01) reflections of as-deposited, 800-°C, and 1200-°C annealed films are shown in Fig. 1 .The (0 0 01) reflection also shifts toward higher 2 θ values from 25.1 °for x = 0, to 25.6 °for x = 0.2, to 25.7 °for x = 0.3 exhibiting a decrease in out-ofplane c-lattice parameters from 3.54 Å to 3.48 Å to 3.46 Å, respectively, due to replacing Zr by Ta atoms which have a smaller covalent radius [35] .The (0 0 01)-reflection intensities of all annealed films increase with increasing T a , in good agreement with results  [42] and TaB 2 [43] , respectively. of sputter-deposited Ti 0.71 Al 0.29 B 3.08 annealed at 10 0 0 °C [36] .Over the entire T a range, XRD does not provide evidence for phase separation in the annealed films such as secondary phase precipitation or spinodal decomposition, which are classical cases of age hardening in metastable TM nitride alloys [37][38][39][40][41] .
XPS is employed to probe changes in chemical bonding and composition of layers as a function of T a .As previously determined for the as-deposited films [26] , the B 1s peak shifts slightly toward higher binding energy from 188.1 eV for ZrB 2.4 to 188.4 eV for both Zr 0.8 Ta 0.2 B 1.8 and Zr 0.7 Ta 0.3 B 1.5 due to higher electronegativity of Ta compared to Zr, while no visible shift is observed for Zr 3d signals.
In the present experiment, no detectable change in B 1s, Zr 3d, and Ta 4f core-level spectra is observed as a function of T a , indicating that annealing has no effect on bonding states.This is in accordance with XRD results shown in Fig. 1 exhibiting no phase separation by annealing.In addition, there is no significant change in the films' compositions up to T a = 1200 °C, which reveals no B evaporation or oxidation.The as-deposited alloys, Fig. 4 (a), show higher H values than ZrB 2.4 (35.0 ±0.8 GPa for ZrB 2.4 , 42.3 ±0.9 GPa for Zr 0.8 Ta 0.2 B 1.8 , and 40.5 ±0.5 GPa for Zr 0.7 Ta 0.3 B 1.5 ).The higher H values of asdeposited alloys are primarily attributed to solid-solution hardening [44] and their narrow columns (Hall-Petch effect [45,46] ).H increases for all films as a function of T a up to 800 °C; 38.5 ±0.7 GPa for ZrB 2.4 , 45.5 ±1.0 GPa for Zr 0.8 Ta 0.2 B 1.8 , and 42.7 ±1.0 GPa for Zr 0.7 Ta 0.3 B 1.5 .These results are confirmed by the in-situ nanoindentation measurements, which reveal a similar increase in H for all layers up to 800 °C: from 33.0 ±2.1 to 37.1 ±2.3 GPa for ZrB 2.4 , from 42.The combination of XRD, XPS, STEM, EDX, and EELS results reveals that no secondary phase precipitation or spinodal decomposition takes place in the Zr 1-x Ta x B y films annealed up to 800 °C.This can be mainly attributed to their high melting points (3245 °C for ZrB 2 and 30 0 0 °C for TaB 2 [47] ).Bulk diffusion which results in recrystallization typically occurs at a homologous temperature T h (annealing temperature to melting-point temperature ratio) above ~0.5 [48] .Hence, diffusion is significantly limited in the Zr 1-x Ta x B y films up to 800 °C as their T h is ~0.2 at 800 °C.
Fig. 5 compares the plan-view high-resolution HAADF-STEM images of as-deposited, 80 0-°C and 120 0-°C annealed Zr 0.8 Ta 0.2 B 1.8 .The nanostructure of as-deposited films, Fig. 5 (a), consists of crystalline columns (cores) surrounded by narrow disordered boundaries (shells) [27] .Planar defects, most noteworthy stacking faults, can be observed inside the columns.In addition, there are some contrast changes in the columns, which result from local residual strain in the lattice caused by compositional inhomogeneities at a length scale shorter than that required for spinodal decomposition.Annealing the alloys at 800 °C does not significantly change the nanostructure; the columns retain the planar defects, indicated by arrows in Fig. 5 (a) and 5 (b).This confirms that no considerable atomic rearrangement (e.g.recrystallization) occurs in Zr 0.8 Ta 0.2 B 1.8 nanostructure by annealing up to 800 °C.However, the plan-view high-resolution HAADF-STEM image of the alloy annealed at 1200 °C in Fig. 5 (c) reveals a significant planar defect annihilation inside the columns, while the column boundaries still appear disordered.
Excluding the two classical mechanisms of age hardening (secondary phase precipitation and spinodal decomposition) and establishing no considerable residual-stress effect on hardening (compressive stresses decrease during annealing up to 800 °C), we instead consider the role of point defects.Sputter-deposited films typically have significant fractions of vacancies, interstitials, and anti-site substitutions of atoms [49] that require lower activation energies for annihilation at T h < ~0.5 than recrystallization [50] .Hence, the increase in the hardness of the Zr 1-x Ta x B y films by annealing up to 800 °C can be mainly due to point-defect recovery.Correspondingly, the elastic moduli of layers increase during the recovery of strong chemical bonds, Fig. 4 (b).In addition, stacking faults that serve as barriers against dislocation glide are preserved up to 800 °C and contribute to maintain hardness high.
However, the ex-situ nanoindentation hardness decreases to 36.5 ±0.8 GPa for ZrB 2.4 , 41.6 ±0.8 GPa for Zr 0.8 Ta 0.2 B 1.8 , and 40.2 ±0.5 GPa for Zr 0.7 Ta 0.3 B 1.5 at 1200 °C, Fig. 4 (a).This softening can be attributed to recrystallization and column coarsening.For these layers, T h is ~0.4 at 1200 °C that is sufficiently high for diffusion and activating recrystallization.This atomic rearrangement is also demonstrated in Fig. 5 (c) indicating stacking fault annihilation inside the columns, which contributes to a decrease in hardness.The films' elastic moduli continue increasing at T a > 800 °C, Fig. 4 (b), due to recrystallization, which reduces the volume of the disordered metallic-glass-like boundary regions [51] .In general, all layers have high hardness values (H > 34 GPa) over the entire T a range.
To conclude, we study thermal stability and mechanical properties of ZrB 2.4 , Zr 0.8 Ta 0.2 B 1.8 , and Zr 0.7 Ta 0.3 B 1.5 thin films as a function of annealing temperature.All films become harder during annealing up to 800 °C, with the highest hardness achieved for Zr 0.8 Ta 0.2 B 1.8 (45.5 ±1.0 GPa).This age hardening is attributed to point-defect recovery that results in enhanced chemical-bond density, in combination with retaining stacking faults.Eventually, hardness decreases at T a > 800 °C due to recrystallization, column coarsening, and stacking fault annihilation.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2
compares plan-view HAADF-STEM images and corresponding EDX maps of ZrB 2.4 and Zr 0.8 Ta 0.2 B 1.8 before and after annealing at T a = 800 °C and 1200 °C.Dark regions in the HAADF-STEM image of as-deposited ZrB 2.4 correspond to low-Z column boundaries, Fig. 2 (a).The Zr EDX map in Fig. 2 (d) indicates that these dark regions are Zr deficient, while the corresponding B EELS map in the inset of Fig. 2 (d) confirms that the column boundaries are B rich compared to the columns.Contrary to as-deposited ZrB 2.4 , the HAADF-STEM image of as-deposited Zr 0.8 Ta 0.2 B 1.8 film reveals column boundaries with lighter contrast, indicating an enrichment with heavier elements, Fig. 2 (g).Complementary Zr and Ta EDX map in Fig. 2 (j) shows that the column boundaries contain more Ta than the columns, in agreement with APT and XPS results published in reference 26.The B EELS map in the inset of Fig. 2 (j) exhibits that the Ta-rich column boundaries are B deficient compared to the Zr-rich columns.Annealing at T a = 800 °C does not have a significant effect on ZrB 2.4 nanostructure, Figs. 2 (b) and 2 (e).However, Fig. 2 (c) shows

Fig. 2 .
Fig. 2. Plan-view HAADF-STEM images with corresponding EDX elemental maps of as-deposited (a and d) ZrB 2.4 and (g and j) Zr 0.8 Ta 0.2 B 1.8 films, 800-°C annealed (b and e) ZrB 2.4 and (h and k) Zr 0.8 Ta 0.2 B 1.8 films, and 1200-°C annealed (c and f) ZrB 2.4 and (i and l) Zr 0.8 Ta 0.2 B 1.8 films.B EELS maps are shown as insets.

Fig. 3
Fig.3shows plan-view high-resolution HAADF-STEM, ABF-STEM, and TEM images acquired from a B-rich region of ZrB 2.4 annealed at 1200 °C.The HAADF-STEM image in Fig.3(a) exhibits that the dark B-rich area is surrounded by crystalline columns, while the ABF-STEM image reveals that this B-rich region is not crystalline, Fig.3(b).The high-resolution TEM image, which is acquired from another area, shows that the B-rich regions, indicated by dashed lines, have amorphous nanostructure, Fig. 3 (c).All as-deposited films have compressive residual stresses (-0.5 ±0.1 GPa for ZrB 2.4 , -1.5 ±0.3 GPa for Zr 0.8 Ta 0.2 B 1.8 , and -1.8 ±0.3 GPa for Zr 0.7 Ta 0.3 B 1.5 ).Annealing at 800 °C results in a decrease in stress to -0.3 ±0.2 GPa for ZrB 2.4 , -1.2 ±0.2 GPa for Zr 0.8 Ta 0.2 B 1.8 , and -1.4 ±0.3 GPa for Zr 0.7 Ta 0.3 B 1.5 .The decrease in the residual

Fig. 4 .
Fig. 4. Ex-situ nanoindentation (a) hardness H and (b) elastic modulus E of ZrB 2.4 , Zr 0.8 Ta 0.2 B 1.8 , and Zr 0.7 Ta 0.3 B 1.5 thin films as a function of annealing temperature T a .The vertical dashed line corresponds to the growth temperature, indicating H and E values of as-deposited films.

Fig. 5 .
Fig. 5. Plan-view HAADF-STEM images of (a) as-deposited, (b) 800-°C, and (c) 1200-°C annealed Zr 0.8 Ta 0.2 B 1.8 thin films.Arrows indicate stacking faults and microtwins.Semicoherent column boundaries can be also seen.The images are trimmed much to offer lattice resolution because the main purpose here is comparing planar defects at different temperatures.Yet, inspection of larger sets of images from wider areas of the samples confirm that the microstructure with defect density is the same between (a) and (b).