Unraveling the superlattice effect for hexagonal transition metal diboride coatings

Superlattice structures enable the simultaneous enhancement in hardness (H) and fracture toughness (K IC ) of ceramic-like coatings. While a deeper understanding of this effect has been gained for fcc-structured transition metal nitrides (TMN), hardly any knowledge is available for hexagonal diborides (TMB 2 ). Here we show that superlattices can — similarly to nitrides — increase the hardness and toughness of diboride films. For this purpose, we deposited TiB 2 /WB 2 and TiB 2 /ZrB 2 superlattices with different bilayer periods ( Λ ) by non-reactive sputtering. Nanoindentation and in-situ microcantilever bending tests yield a distinct H peak for the TiB

The fracture toughness of hard ceramic-like coatings can significantly be increased with a so-called superlattice (SL) structure, as recently discovered in the example of binary nitrides prepared by physical vapor deposition (PVD) [1][2][3][4].Their characteristic parameter is the bilayer period Λ, which is the combined thickness of the two constituting layers [5].The impact of a superlattice structure on hardness has been known for many years (especially for fcc nitrides), and one of the earlier works even shows twice the hardness for TiN/VN SLs with Λ = 5.2 nm compared to their monolithic counterparts [6].This SL effect on hardness is regularly observed for nitrides with Λ between 5 and 10 nm [7] and essentially results from competing dislocation activity mechanisms, which are well explained in [8].However, there is still a need to clarify the superlattice effect on fracture toughness since the measurement methodology excludes a significant contribution of dislocation movement [9].In the present work, the concept of superlattice structures was applied to hexagonal transition metal diboride coatings and thoroughly investigated concerning their fracture toughness and hardness.To unravel the primary effect of a different lattice parameter or different shear moduli individually-being the suggested key factors for hardness and fracture toughness increase-we chose TiB 2 /WB 2 with a high shear modulus difference combination (ΔG = 112 GPa and Δa = 0.01 Å [10]) and TiB 2 /ZrB 2 obtaining a significant difference in lattice parameters (ΔG = 27 GPa and Δa = 0.14 Å [10]).Therefore, we prepared TiB 2 /WB 2 and TiB 2 /ZrB 2 multilayer coatings using an AJA International Orion 5 unbalanced magnetron sputtering system equipped with two 2 ′′ (TiB 2 and WB 2 ) and one 3 ′′ (ZrB 2 ) cathode.The ultrasonically cleaned single crystalline substrates (Si (100) and Al 2 O 3 (1011)) were mounted on a rotating (1 s − 1 ) substrate holder and heated to the deposition temperature as well as Ar etched.The deposition was carried out at a total Ar pressure of 0.4 Pa using a computer-controlled shutter system to achieve the desired superlattice architecture.Details of the used deposition parameter and the final film thickness (determined using an FEI Quanta 250 FEGSEM) are presented in Table 1, the growth rates of our coatings were the following: 4.0 nm•min − 1 for TiB 2 , 29.4 nm•min − 1 for ZrB 2 , and 5.2 nm•min − 1 for WB 2 .
Fig. 1a shows a detailed view of the (001) peak for the TiB 2 /WB 2 system.The structural data were recorded on Si (100) using a Panalytical XPert Pro-MPD θ-θ diffractometer in Bragg-Brentano configuration equipped with a Cu K α radiation source.All our thin films are stabilized in the so-called α structure (SG 191); notably, we do not observe any signs of the ω-modification (SG 194), which is suggested to be detrimental to coherent growth.The (001) orientation is predominant for the monolithic coatings and bilayer periods of 2, 4, and 6 nm, respectively.Higher bilayer periods tend to show a mixed (001) and (101) orientation (the (101) peak positions at 2θ ~ 44 • shown in Fig. S1).We observe an increase in satellite peaks visible with increasing bilayer periods.However, as soon as the (101) orientation becomes apparent, the number of satellite peaks decreases.Fig. 1b shows the corresponding (001) peak region of the TiB 2 /ZrB 2 superlattices and monolithic coatings.Compared to the TiB 2 /WB 2 system, the number of satellite peaks increases, and the (001) orientation is maintained.The dominant structure is exclusively α.Furthermore, the cumulative peak for the Λ = 2 nm coating shifts to higher 2θ values, mainly due to a higher Zr content (see Table 1).To confirm the bilayer period derived from dividing the total film thickness by the number of bilayers, we also calculated the values, as described in [11], see Table 1 and supplementary materials.
To further prove and thoroughly describe the morphology of the superlattice structure, selected samples have been investigated in the TEM.All TEM images were recorded using a Tecnai F20 operated at 200 kV and equipped with a FEG and a Gatan Rio camera.Fig. 2a shows an HR-TEM image of the 6 nm TiB 2 /WB 2 sample, including an FFT of the entire image area, confirming the α structure.The bilayer period measured from this is 5.6 nm, fitting well with the SEM and XRD results.The dashed areas (marked with b, resp.c) show the sections used to generate an ABSF-filtered image [12], which was further investigated concerning the lattice spacing (Fig. 2d-f).We found that the lattice parameters of individual layers agreed with literature values: TiB 2 shows a lattice spacing d 100 of 2.67 Å (2.63 Å according to PDF 00-035-0741 [13]), WB 2 of 2.72 Å (2.62 Å according to PDF 04-003-6624 [13]).This crystal direction is also where coherence stresses should occur, but this could not be verified based on the analysis.By the same methodology, we determined the c-axis parameters and verified a good agreement with the literature for TiB 2 with 3.25 Å; the literature value for d 001 is 3.23 Å [13].For WB 2, we measured a c-axis parameter of 3.19 Å, well within the limits given in [14], considering the highly defective growth during PVD.
A different situation is observed for the TiB 2 /ZrB 2 sample with a 4 nm bilayer period.Here, we can also confirm the bilayer period measured by XRD and SEM in the TEM (3.76 nm) (Fig. 2g), but more considerable differences appear in the determined lattice parameters (Fig. 2h-l).While the d 100 lattice parameter for the ZrB 2 single layer with 2.81 Å is close to the literature value (2.76 Å according to PDF

Table 1
The used deposition parameter (the applied target current Ic, bias voltage Ub, and heater temperature Tc) for multilayer architectures and monolithic coatings, as well as the nominal bilayer period Λnom, the final experimental film thicknesses tf, and the experimental bilayer periods ΛSEM and ΛXRD.All three targets used had a purity of at least 99.5 at.% and have been manufactured by Plansee Composite Materials GmbH.For the growth of all films, a base pressure below 2•10− 6 mbar was maintained.An etching procedure was applied for 10 min by introducing 20 sccm Argon, regulating the pressure to 6 Pa at a bias voltage of − 750 V.The resulting bilayer period diverging from the nominal was calculated by dividing the film thickness by the number of bilayers.The TM/(TM+Ti) ratios were determined using an EDAX Octane Elite EDS detector using an FEI Quanta 200 SEM for Zr/(Zr+Ti) and measured by TOF-ERDA (see [24] and references therein) for W/(W+Ti), similar to the method described in [25].The ERDA results show a B/Ti ratio of 2.66 and a W/B ratio of 2.00 for the monolithic coatings.).In summary, for the analysis of the lattice parameters, it can be said that, as theoretically predicted, the TiB 2 /WB 2 system shows hardly any differences from the literature values, while the TiB 2 /ZrB 2 system shows significant deviations from literature values in the d 100 lattice parameter.Hence, coherency stresses should be present and influence the fracture properties of the latter system.Fig. 3 presents Young's moduli and hardness of both systems and the respective monolithic coatings.These mechanical parameters were measured using a UMIS nanoindentation system equipped with a diamond Berkovich tip.For the Young's moduli in Fig. 3a, we observe that the measured values of the superlattice coatings tend to show lower values than the average ones of their individual components, with a bilayer period depending relation, peaking at 4 and 6 nm for TiB 2 /ZrB 2 and TiB 2 /WB 2, respectively.Fig. 3b shows the hardness values of the TiB 2 /WB 2 system for both substrates, sapphire and silicon.We observe an apparent bilayer perioddependent behavior due to the difference in shear moduli (ΔG = 112 GPa) of TiB 2 and WB 2 .The hardness peaks around 6 nm and is slightly higher than the average of its constituents.It is also worth mentioning that the constituents do not possess any significant difference in lattice parameters perpendicular to the growth direction.Hence, their difference in shear modulus mainly accompanies the enhancement.Fig. 3c shows the hardness vs. bilayer period progression for the second system, TiB 2 /ZrB 2 , which does not significantly differ in their elastic constants (253 and 226 GPa for TiB 2 and ZrB 2 , respectively [10]).Importantly, we do not measure any bilayer period-dependent behavior in this case.The hardness values of the multilayers are at a level similar to the average of Fig. 2. High-resolution TEM images of the TiB 2 /WB 2 sample with Λ = 6 nm and the TiB 2 /ZrB 2 sample with Λ = 4 nm; the samples were prepared by applying a standard lift-out procedure using a ThermoFischer Scios 2 Dual Beam FIB.(a) shows an HR bright field image of a representative region of the TiB 2 /WB 2 sample, including an FFT image of the whole area as well as the areas for further detailed analysis (marked with b and c).Panels (b) and (c) show ABSF-filtered images, including the areas that were used to determine lattice spacings (marked with d, e, and f).The arrows thereby show the direction in which the intensity was summed and afterward plotted in (d), (e), and (f).The same procedure was applied to our TiB 2 /ZrB 2 sample: (g) shows the HR TEM image, including an FFT of the whole area, (h) and (i) a detailed view of the marked areas in (g), including marks for the intensity plots (j), (k), and (l).We measured 3 lattice distances for all intensity plots to obtain a more accurate average value for d 100 and d 001 .
their components.It is also interesting that coherency stresses alone (although large) do not guarantee any significant hardness increase.Hence, concluding this figure, we see a typical behavior for the hardness progression for diboride-based superlattices dominated by the shear modulus (112 GPa, TiB 2 /WB 2 ) difference, while the lattice parameters have a minor influence as described in [8].
Micro-cantilever bending tests have been conducted to prove the main influencing factors on fracture toughness.Therefore, the same SEM system used for the film thickness determination was equipped with a FemtoTools FT-NMT04 nanomechanical testing system for fracture toughness evaluation.FIB-milled cantilevers (ThermoFisher Scientific Scios2) were loaded in the growth direction with a wedge-shaped diamond tip until fracture (also in the growth direction).The cantilever dimensions were chosen to fulfill the criteria given by Brinckmann et al. [15,16].In this case, the typical dimension of the cantilever was ~ 14 × 2 × 2 µm 3 with an average initial crack length of ~ 400 nm.We performed at least four successful experiments for each material system and architecture to calculate the fracture toughness according to Matoy et al. [17].After testing and calculating the fracture toughness, we ensured that our values represented K IC values by estimating the plastic zone size and comparing it to the critical dimensions [9].Fig. 4a summarizes the results of our microcantilever bending tests.We observe a bilayer period-depending behavior for TiB 2 /ZrB 2 superlattices, with a peak in fracture toughness at K IC = 3.7 ± 0.26 MPa•m 1/2 for Λ = 6 nm, clearly above the monolithic systems.This increase in K IC has also been observed with other superlattice systems [2,3,18].In contrast, the TiB 2 /WB 2 system behaves differently: we do not observe a dependence of the fracture toughness on the bilayer period.In contrast, the measured K IC values tend to decrease with decreasing Λ, although not significantly.The highest value of 3.0 ± 0.24 MPa•m 1/2 was achieved at a bilayer period of 50 nm, comparable to the constituents.
In the case of macroscopically residual stress-free material and (c), respectively.To achieve specific statistics, we performed 31 measurements for each sample with varying maximum loads from 3 to 45 mN, thereby ruling out measurements with indentation depths larger than 1/10th of the coating thickness.The hardness was calculated according to Oliver and Pharr [26], whereas Young's Modulus was determined according to [27] for the two different substrate types.For the apparent fracture toughness K C , an indentation process serving crack initiation and propagation was repeated four times over four series of measurements, with the applied force gradually increasing from 100 mN for the first series to 400 mN for the last.
properties, these results highlight coherency stresses' crucial role in increasing fracture toughness.While the TiB 2 /ZrB 2 system (lattice mismatch of 0.14 Å) shows a pronounced peak in K IC , this behavior is absent in the TiB 2 /WB 2 system (difference in shear modulus 112 GPa).For comparison, the lattice parameters (in (100) direction) of the three material systems differ significantly for TiB 2 and ZrB 2 with 3.03 and 3.17 Å, respectively, whereas TiB 2 and WB 2 (3.020Å) show almost no differences.
Relying on the findings obtained by [4], the lattice parameter difference of Δa = 0.14 Å existing between TiB 2 and ZrB 2 leads to the formation of coherent stresses, compressive in the ZrB 2 layers and tensile in the TiB 2 layers.The distortion energy of the ZrB 2 lattice associated with the compressive stresses impedes the progressive growth of a previously initiated crack, with the effectiveness of the energetic resistance increasing with the layer thickness.The maximum K IC value was obtained for a bilayer period of Λ = 4 nm.With a progressive increase in layer thicknesses, the formation of interfacial misfit dislocations appears to be more energetically favorable for TiB 2 and ZrB 2 than their elastic lattice strain, resulting in the reduction of coherent stresses.Consequently, the distortion energy to be overcome by a crack in the ZrB 2 layer reduces, which decreases the intrinsic fracture toughness.
Fig. 4b shows the results of residual stress measurements (performed using a profilometer: Nanovea PS50) on Si substrates for TiB 2 /WB 2 and TiB 2 /ZrB 2 multilayer systems.Compressive residual stress states are predominant in all systems except the monolithic ZrB 2 layer.The residual stresses of the superlattice coatings show a slight dependence on the layer period for both variants.We observe an increase in the residual compressive stresses with decreasing layer period, in agreement with literature observations [19].
To further verify the observed findings, indentation fracture (IF) analysis took place in the same IBIS nanoindenter system described above, although a cube-corner diamond indenter tip was used to determine the fracture toughnesssee Fig. 4c.For each radial crack that appeared at the corner of a pyramidal plastic indentation, the length c fra was measured in the FEI Quanta 200 FEGSEM.In a further step, the associated K c value was calculated using [20].A penetration depth of the indenter tip smaller than ten percent of the layer thickness could not be observed; therefore, an influence of the substrate material on the fracture toughness had to be assumed.To reduce the effect of the violated condition on the K C value, the approach according to [21] was suitable.
Using the IF method, there is an initially dramatic increase in fracture toughness for the TiB 2 /ZrB 2 system from 1.62 ± 0.09 MPa√m for Λ = 2 nm to the maximum value of 3.91 ± 0.23 MPa√m for = 4 nm, followed by an equally significant decrease in the characteristic curve to K C = 1.66 ± 0.2 MPa√m, which subsequently takes a flat course leading to 1.62 ± 0.08 MPa√m for a bilayer period of 50 nm.Apart from the increase in intrinsic fracture toughness between Λ = 15 and 50 nm, the K C and K IC values of the TiB 2 /ZrB 2 system proceed in a congruent manner, as indicated by their high variance in a range from Λ = 2 to 15 nm and an explicit maximum for a layer period of 4 nm.Moreover, the characteristic values of those TiB 2 /ZrB 2 layers grown with Λ = 4 and 9 nm turn out to be higher than the K C and K IC values of the respective tougher monolithic diboride layer.
The TiB 2 /WB 2 system results in an initial decrease from K C = 3.92 ± 0.6 MPa√m for Λ = 2 nm to K C = 3.61 ± 0.26 MPa√m for Λ = 4 nm and a subsequent maximization to K C = 4.24 ± 0.25 MPa√m, with a final reduction to K C = 3.69 ± 0.12 MPa√m for a layer period of 50 nm.Although the residual stresses and the K C values of the TiB 2 /WB 2 layers do not show a consistent trend, their higher compressive residual stresses, considered globally, offer a comprehensible justification for exceeding the K C value exhibited by their constituents.
Because the K C and K IC evolution in the single-digit bilayer period hardly deviates, the main question, whether the shear moduli differences of two-layer materials dominate the superlattice effect on fracture toughness, can be negated for these hexagonal diborides.
Within this study, the enhancement of mechanical properties such as hardness and fracture toughness by superlattice structures could be successfully proven for physical vapor deposited hexagonal diborides.We could isolate the lattice parameter difference as the primary mechanism for increased toughness.At the layer interfaces, coherency stresses impede crack propagation due to the lattice mismatch (Δa ~ 4.6%), demonstrated for TiB 2 /ZrB 2 (i.e., K C improvement by 30%).In contrast, a difference in shear modulus results in a distinct hardness peak attributed to hindered dislocation motion across the interface and within a single layer, as shown for TiB 2 /WB 2 .Nevertheless, the anisotropic properties (predominant [001] texture [14,22]) of hexagonal diborides, combined with the shear modulus differences, need to be considered to interpret the hardness evolution in the case of the TiB 2 /WB 2 system.Additionally, for the TiB 2 /ZrB 2 system, the influence of a possible tissue phase at the absolute values of the fracture toughness-not their progression as the elemental composition is similar throughout the coating series [23].

Fig. 1 .
Fig. 1.X-ray diffraction patterns of (a) the (001) regions of TiB2/WB2 and (b) the (001) regions of TiB2/ZrB2.The bilayer period and the monolithic system of the XRDs are on the right side, and the triangles on the top mark the reference positions for the respective material system.The resulting cumulative peaksdue to overlapping contributions from the constituting layers, especially for small bilayer periodsare marked with "c," and the satellite peaks stemming from the layered structure are marked with "+n" and "-n" for positive and negative satellites, respectively.

Fig. 3 .
Fig. 3. (a) shows the Young's modulus of our coatings, hardness values for TiB 2 /WB 2 , and TiB 2 /ZrB 2 (on Si and Sapphire substrates) are presented in (b)and (c), respectively.To achieve specific statistics, we performed 31 measurements for each sample with varying maximum loads from 3 to 45 mN, thereby ruling out measurements with indentation depths larger than 1/10th of the coating thickness.The hardness was calculated according to Oliver and Pharr[26], whereas Young's Modulus was determined according to[27] for the two different substrate types.

Fig. 4 .
Fig. 4. (a) shows the intrinsic fracture toughness of our coatings.The residual stresses and indentation toughness on Si (100) are presented in (b) and (c), respectively.For the apparent fracture toughness K C , an indentation process serving crack initiation and propagation was repeated four times over four series of measurements, with the applied force gradually increasing from 100 mN for the first series to 400 mN for the last.