Microstructure and Materials Properties of Understoichiometric TiBx Thin Films Grown by HiPIMS

of Understoichiometric TiBx Thin Films Grown In the present research article we report synthesis of TiB x , 1.43n-situ mass- and energy-spectroscopy is used to explain the obtained compositional range. Excess B in overstoichiometric TiB x thin films from DCMS results in a hardness up to 37.7±0.8 GPa, attributed to the formation of an amorphous B-rich tissue phase separating stoichiometric TiB 2 columnar structures. With a particular focus on characterization of the understoichiometric samples, we show that understoichiometric TiB 1.43 thin films synthesized by HiPIMS exhibit a superior hardness of 43.9±0.9 GPa, where the deficiency of B is found to be accommodated by Ti planar defects. The apparent fracture toughness, electrical resistivity and thermal conductivity of the same sample is 4.2±0.1 MPa√m, 367±7 μΩ·cm and 5.1 W/(m.K), respectively, as compared to corresponding values for overstoichiometric TiB 2.20 Abstract TiB x thin films with a B content of 1.43 ≤ x ≤ 2.70 were synthesized using high-power impulse magnetron sputtering (HiPIMS) in comparison to direct-current magnetron sputtering (DCMS). HiPIMS allows compositions ranging from understoichiometric to overstoichiometric dense TiB x thin films with a B/Ti ratio between 1.43 and 2.06, while DCMS yields overstoichiometric TiB x films with a B/Ti ratio ranging from 2.20 to 2.70. Excess B in overstoichiometric TiB x thin films from DCMS results in a hardness up to 37.7±0.8 GPa, attributed to the formation of an amorphous B-rich tissue phase separating stoichiometric TiB 2 columnar structures. We furthermore show that understoichiometric TiB 1.43 thin films synthesized by HiPIMS, where the deficiency of B is hardness of 43.9±0.9 GPa. The apparent fracture toughness and thermal conductivity of understoichiometric TiB 1.43 HiPIMS films are 4.2±0.1 MPa√m and 5.1 W/(m.K), respectively, as compared to corresponding values for overstoichiometric TiB 2.20 DCMS film samples of 3.2±0.1 MPa√m and 3.0 W/(m.K). This work increases the fundamental understanding of understoichiometric TiB x thin films and their materials properties, and show that understoichiometric films have properties matching or going beyond those with excess B. Abstract TiB x thin films with a B content of 1.43 ≤ x ≤ 2.70 were synthesized using high-power impulse magnetron sputtering (HiPIMS) in comparison to direct-current magnetron sputtering (DCMS). HiPIMS allows compositions ranging from understoichiometric to overstoichiometric dense TiB x thin films with a B/Ti ratio between 1.43 and 2.06, while DCMS yields overstoichiometric TiB x films with a B/Ti ratio ranging from 2.20 to 2.70. Excess B in overstoichiometric TiB x thin films from DCMS results in a hardness up to 37.7±0.8 GPa, attributed to the formation of an amorphous B-rich tissue phase separating stoichiometric TiB 2 columnar structures. We furthermore show that understoichiometric TiB 1.43 thin films synthesized by HiPIMS exhibit a superior hardness of 43.9±0.9 GPa, where the deficiency of B is found to be accommodated by Ti-enriched planar defects. The apparent fracture toughness and thermal conductivity of HiPIMS films with a B/Ti ratio of 1.43, are 4.2±0.1 MPa√m and 5.1 W/(m.K), respectively, as compared to corresponding values for overstoichiometric TiB 2.20 DCMS thin film samples of 3.2±0.1 MPa√m and 3.0 W/(m.K). This work increases the fundamental understanding of understoichiometric TiB x thin films and their materials properties, and show that understoichiometric films have properties matching or going beyond those with excess B.

In the present research article we report synthesis of TiB x , 1.43n-situ mass-and energy-spectroscopy is used

Introduction
Titanium diboride, TiB 2 , is one of the more thoroughly investigated transition-metal diborides to date. This hard ceramic has attractive properties motivating this interest, [1][2][3] including high thermal and electrical conductivity, [4] good thermal and chemical stability [5] as well as good oxidation and mechanical erosion resistance. [6][7][8] While the most common method to grow TiB 2 is direct-current magnetron sputtering from compound targets, several reports show other approaches in choice of synthesis, including electroplating, arc-PVD, pulsed DCMS, RF-sputtering, and HiPIMS. [9][10][11][12][13][14][15][16] DCMS typically produce overstoichiometric material, with a B/Ti ratio in the range of 2.4 to 3.5. Previous work has shown that the excess B in these TiB 2 films move towards grain boundaries and form nanostructures, a mechanism that is well described. [17,18] Essentially, excess B forms an amorphous B-rich tissue-phase in between the nanocolumnar morphology of TiB 2 . The structure, in turn, prevents propagation of dislocations across columns from occurring, which increases the hardness of the TiB x thin film. Few reports present understoichiometric TiB x thin films. [16,19,20] An emerging method for TiB 2 growth is HiPIMS. [16,[19][20][21] In contrast to TiB x thin films grown with DCMS, reports indicate that TiB x films grown with HiPIMS realize understoichiometric compositions. The mechanism for this behavior has not yet been addressed, with limited investigation of resulting mechanical properties. M.N. Polyakov et. al. [20] demonstrated hardness of HiPIMS-grown TiB x thin films with B/Ti ratio of 1.62 to 1.97 in the range of 39-45 GPa, and attributed this to film densification. Also, explanation is lacking in literature on how the B deficiency is accommodated in the lattice of understoichiometric TiB x films. To ameliorate this, we here investigate the effect of gas pressure on HiPIMS and DCMS ion flux, atomic structure, microstructure, and expand on the properties of over-and understoichiometric TiB x films grown with HiPIMS and compare with corresponding DCMS grown films.

Experimental Details
The TiB x thin films with 1.43 ≤ x ≤ 2.70 are grown using HiPIMS and DCMS with a 3.0'' x 0.125'' (99.5% purity) TiB 2 target (Kurt J. Lesker Co.) in an ultrahigh vacuum (UHV) system with a base pressure of 7x10 -8 mTorr on sapphire (001) substrates with a TiN buffer layer. The substrates were ultrasonically cleaned in acetone and isopropanol for 10 min respectively then blow dried in N 2 . 45 and 90 nm TiN buffer films, used for increased adhesion between the substrate and the TiB x films, were deposited using DCMS with a pure Ti target (99.995% purity) with Ar and N gas, both 99.9999% pure, introduced through mass flow controllers to partial pressures of 2.75 and 0.25 mTorr, respectively. Substrate temperature was kept at 670 °C with a floating bias and a target power fixed at 185 W during buffer layer deposition. TiB 2 films, approximately 1 μm thick, where grown by HiPIMS and DCMS in pure Ar at 5 mTorr and 20 mTorr at substrate temperature of 500 o C.
In-situ mass-and energy-spectroscopy analyses of ion fluxes generated in HiPIMS and DCMS were performed using a Hiden Analytical EQP1000 instrument placed at 15 cm from the target surface. In both cases, the diagnostic was done in the time-averaged mode of the analyzer, [22] i.e. the data acquisition time window of the analyzer at each measurement point (10 ms) was set the same for both HiPIMS and DCMS. [23] For each of the studied here operational conditions, the generated plasma was characterized through mass-scans at fixed ion energy and energy-scans at fixed mass-to-charge ratio for all detected ions. The energy scans were recorded in steps of 0.5 eV/charge up to 50 eV. Each scan was recorded at least three times to ensure consistency of the data. To determine the plasma composition, total intensities of the elements were calculated as sums of intensities of corresponding ions on all steps of the measurement. All values presented are repeatable with an error of 5 %. The natural isotope distribution of the elements was taken into account in line with Ref. [24]. X-ray diffraction (XRD) θ-2θ scans on the thin film samples were performed using a PANalytical X'Pert powder diffractometer, with Cu source (λ = 1.54 Å). The optics utilized for these measurements were a graded Bragg-Brentano HD with ½° divergent and ½° antiscatter slits for the incident beam side, and a 5-mm antiscatter slit together with a Soller slit for the diffracted beam side. A 5-120° continuous scan was performed on the sample using a step size of 0.016° with 10 s time per step.
Film compositional and structural analysis was done with a scanning electron microscope (SEM) LEO 1550 Gemini equipped with an Oxford Instruments energy-dispersive X-ray (EDX) detector operating with an acceleration voltage between 5-15 keV, time-of-flight elastic recoil detection analysis (ToF-ERDA), with a 36 MeV 127 I +8 beam incident at 67.5° to the sample normal with a recoil angle of 45°, and Rutherford backscattering spectrometry (RBS) with a 2 MeV 4 He + beam incident at 5° with a scattering angle of 170°. [13][14][15] The two latter analyses were carried out at the Tandem Laboratory at Uppsala University.
High-resolution scanning transmission electron microscopy (HRSTEM) imaging and selective area electron diffraction (SAED) was used to determine the atomic structure, and to further verify the composition and crystallographic relations of the film.
Microscopy was performed using the Linköping double corrected FEI Titan 3 60-300 operated at 300 kV. STEM high-angle annular dark-field (STEM-HAADF) imaging was performed by using a 21.5-mrad convergence semi-angle, which provided sub-Ångstrom resolution probes with ~ 60-pA beam current and using an angular detection range of 46-200 mrad. TEM specimens were prepared by focused ion beam technique employing a Carl Zeiss Cross-Beam 1540 EsB system.
Mechanical properties of the films were studied with the aid of a Hysitron TI 950 nanoindenter system and an ultra-micro indentation system (UMIS). To investigate the indentation hardness, H, and combined elastic modulus of the contacting bodies, E*, every sample was subjected to 90 indentations with a Berkovich diamond tip within the load range of 0.1-13 mN. The H and E* values were obtained by evaluating the loaddisplacement curves by means of the Oliver and Pharr method. [25] Subsequently, the elastic modulus of the films, E, was calculated assuming a Poisson's ratio, ν, of 0.15. [26,27] Given the impact of the stoichiometry on the elastic constants of TiB x , the upper and lower limits of E were calculated for the Poisson's ratios of 0.1 and 0.3, respectively. [28,29] After plotting H and E values as a function of penetration depth, only H values from a fully developed plastic zone and yet unaffected by the substrate were considered film-only hardness, while a smooth curve was fitted to the E values and subsequently extrapolated back to zero depth to obtain film-only elastic modulus. [30] A qualitative analysis of the fracture behavior of the system film/substrate was carried out by means of nanoindentation utilising UMIS equipped with a cube-corner indenter tip. The applied load, P, was varied within the range 5-300 mN, incremented in steps 10 mN within the load range 10-50 mN and in steps 50 mN within the load range 50-300 mN. By examining the imprints with the aid of an SEM and measuring the average crack length, c, -for an applied load, P, -the apparent fracture toughness of the system film/substrate, K C , can be evaluated as follows: where α is an empirical calibration constant taken for 0.036. [31] To achieve a statistical distribution of K C , five indents were made at every load. After plotting K C values as a function of penetration depth, a linear fit to the data points was extrapolated back to 1/10 of the film thickness to minimise any substrate effect. [32] The electrical resistivities of the TiB x samples, with the TiN buffer layer, were obtained by measuring the sheet resistance with a four-point probe (Jandel RM3000). The effective resistivity of the TiB x layer was obtained using the following equation [33]: Where ρ is the electrical resistivity of the material and t is the film thickness. The electrical resistivity of a TiN film of 45 nm deposited on Al 2 O 3 was also measured for reference.
Thermal characterization of the TiB 2 films was performed using a modulated photothermal radiometry (MPTR) setup, used and previously reported in Refs. [34,35].
The laser beam was modulated in frequency in the 0.7 to 5 kHz range and was focused at the surface of the films by an appropriate optical path. The films are capped with a platinum layer of approximately 100 nm which acts as the optical-to-thermal transducer.
The infrared (IR) radiation from the heated surface is collected by two parabolic mirrors and focused at the sensitive element of an IR detector. The phase between the reference excitation and the measured signal from the detector is measured using a lock-in amplifier. The equivalent thermal conductivity accounting for the TiB x film and the film-substrate interface is then identified by minimizing the quadratic gap between the measured phase and that calculated from the model that describes the heat diffusion within the sample in the experimental configuration. The heat diffusion model is based on the heat diffusion within the layer and the substrate, assuming isotropic properties for both materials. The model requires knowing the heat capacity per unit volume of the layer and the thermal properties of the substrate. The minimization is achieved by using the Levenberg-Marquardt algorithm.

Results and Discussion
The elemental composition of the films determined by ToF-ERDA is given in Tab. 1; the composition is henceforth used for sample notation. The residual gas contamination for the DCMS and HiPIMS thin films were also measured displaying similar impurity The two thin films grown by HiPIMS, TiB 1. 43 and TiB 2.06 , were investigated by planview STEM imaging, and the results are shown in Fig. 3

a)-c) and d)-e), respectively.
From the overview images, it is clear that both thin films exhibit a similar structure with column widths of the order of ~50 nm. The local increased contrast originates from grains, which are oriented with their (001) zone axis along the electron beam. Under such condition the electron probe channels very strongly along the atomic columns, which together with thermal diffused scattering results in increased intensity at the annular detector, overwhelming the atomic number contrast imaging. [37][38][39] The grains deviating from the zone axis appear dark. At higher magnifications apparent differences can be found between two samples. The most striking difference is the presence of a tissue phase between the dendritic/fractal-like columns in the TiB 2.06 thin film, see Fig. 3 d)-f), while the structure is dense and free from any tissue phase for the TiB 1.43 thin film which at high magnification appear to be a high density of stacking faults, see Fig. 3

c).
These striations were further investigated by HRSTEM methods on a TiB 1.44 thin film deposited using the same experimental setup though with an elevated temperature, HiPIMS at 900 o C at 20 mTorr, owing to a higher crystalline quality with fewer stacking faults causing overlap and projection effects. The results are shown in Fig. 4 and with a more detailed investigation of the stacking faults presented in Ref. [40]. The HRSTEM image in Fig. 4 presents the core of a column, viewed on-axis, e.g., parallel to the <001> direction. The structure clearly exhibits a hexagonal appearance, however, stacking faults on the prismatic planes are visible throughout the grain. The stacking faults make up an intricate pattern, but remains fixed to the { 11 00 } prismatic planes of the TiB 2 crystal structure. Further, it was found that stacking faults are deficient in B and rich in Ti, as judged from electron energy loss spectroscopy analysis Ref. [40]. The  The elastic modulus, E, of respective film can be seen in Fig. 5 a).  films, yet higher than bulk and single-crystal films at ~20 μΩ·cm. [41] The TiB x thin film values obtained, 3677 and 3094 μΩ·cm for TiB 1. 43 and TiB 2.70 respectively, are also higher than those encountered for single crystal bulk samples (~15 μΩ·cm), [42] however they are in the same range as the values reported for e-beam evaporation and RF sputtered thin films (267 and 230 -330 μΩ·cm, respectively). [43,44] The stoichiometric TiB 2.06 film has the lowest resistivity which can be attributed to the lack of amorphous B-rich tissue phase, which is also in accordance with Shutou et al. [44] The resistivity is more than 2.3 times larger for both the over-stoichiometric and understoichiometric films. The densities of each films were deduced through RBS and crosssectional SEM and were multiplied by the specific heat of stoichiometric TiB 2 (617 J/g.K), taken from the literature. [4] The phase of the thermal IR response obtained for the TiB x films are given as a function of frequency for the different B/Ti ratios and are shown in Fig. 6

Conflicts of interest
There are no conflicts to declare.
Previous work has shown that the excess B in these TiB2 films move towards grain boundaries and form nanostructures, a mechanism that is well described. [17,18] Essentially, excess B forms an amorphous B-rich tissue-phase in between the nanocolumnar morphology of TiB2. The structure, in turn, prevents propagation of dislocations across columns from occurring, which increases the hardness of the TiBx thin film. Few reports present understoichiometric TiBx thin films. [16,19,20] An emerging method for TiB2 growth is HiPIMS. [16,[19][20][21] In contrast to TiBx thin films grown with DCMS, reports indicate that TiBx films grown with HiPIMS realize understoichiometric compositions. The mechanism for this behavior has not yet been

Experimental Details
The TiBx thin films with 1. 44  In-situ mass-and energy-spectroscopy analyses of ion fluxes generated in HiPIMS and DCMS were performed using a Hiden Analytical EQP1000 instrument placed at 15 cm from the target surface. In both cases, the diagnostic was done in the time-averaged mode of the analyzer, [22] i.e. the data acquisition time window of the analyzer at each measurement point (10 ms) was set the same for both HiPIMS and DCMS. [23] For each of the studied here operational conditions, the generated plasma was characterized through mass-scans at fixed ion energy and energy-scans at fixed mass-to-charge ratio for all detected ions. The energy scans were recorded in steps of 0.5 eV/charge up to 50 eV.
Each scan was recorded at least three times to ensure consistency of the data. To determine the plasma composition, total intensities of the elements were calculated as sums of intensities of corresponding ions on all steps of the measurement. All values presented are repeatable with an error of 5 %. The natural isotope distribution of the elements was taken into account in line with Ref. [24] X-ray diffraction (XRD) θ-2θ scans on the thin film samples were performed using a PANalytical X'Pert powder diffractometer, with Cu source (λ = 1.54 Å). The optics utilized for these measurements were a graded Bragg-Brentano HD with ½° divergent and ½° antiscatter slits for the incident beam side, and a 5 mm antiscatter slit together with a Soller slit for the diffracted beam side. A 5-120° continuous scan was performed on the sample using a step size of 0.016° with 10 s time per step. to the sample normal with a recoil angle of 45°, and Rutherford backscattering spectrometry (RBS) with a 2 MeV 4 He + beam incident at 5° with a scattering angle of 170°. [13][14][15] The two latter analyses were carried out at the Tandem Laboratory at Uppsala University.
High-resolution scanning transmission electron microscopy (HRSTEM) imaging and selective area electron diffraction (SAED) was used to determine the atomic structure, and to further verify the composition and crystallographic relations of the film.
Microscopy was performed using the Linköping double corrected FEI Titan 3 60-300 equipped operated at 300 kV. STEM high angle annular dark field (STEM-HAADF) imaging was performed by using a 21.5 mrad convergence semi-angle, which provided sub-Ångstrom resolution probes with ~ 60 pA beam current and using an angular detection range of 46-200 mrad.
Mechanical properties of the films were studied with the aid of a Hysitron TI 950 nanoindenter system and an ultra-micro indentation system (UMIS). To investigate the indentation hardness, H, and combined elastic modulus of the contacting bodies, E*, every sample was subjected to 90 indentations with a Berkovich diamond tip within the load range of 0.1-13 mN. The H and E* values were obtained by evaluating the loaddisplacement curves by means of the Oliver and Pharr method. [25] Subsequently, the elastic modulus of the films, E, was calculated assuming a Poisson's ratio, ν, of 0.15. [26,27] Given the impact of the stoichiometry on the elastic constants of TiBx, the upper and lower limits of E were calculated for a Poisson's ratios of 0.1 and 0.3, respectively. [28,29] After plotting H and E values as a function of penetration depth, only H values from a fully developed plastic zone and yet unaffected by the substrate were considered film-only hardness, while a smooth curve was fitted to the E values and subsequently extrapolated back to zero depth to obtain film-only elastic modulus. [30] A qualitative analysis of the fracture behavior of the system film/substrate was carried out by means of nanoindentation utilising UMIS equipped with a cube-corner indenter tip. The applied load, P, was varied within the range 5-300 mN, incremented in steps 10 mN within the load range 10-50 mN and in steps 50 mN within the load range 50-300 mN. By examining the imprints with the aid of an SEM and measuring the average crack length, c,for an applied loadthe apparent fracture toughness of the system film/substrate, KC, can be evaluated as follows: where α is an empirical calibration constant taken for 0.036. [31] To achieve a statistical distribution of KC, five indents were made at every load. After plotting KC values as a function of penetration depth, a linear fit to the data points was extrapolated back to 1/10 of the film thickness to minimise any substrate effect. [32] The electrical resistivities of the TiBx samples, with the TiN buffer layer, were obtained by measuring the sheet resistance with a four-point probe (Jandel RM3000). The effective resistivity of the TiBx layer was obtained using the following equation [33]: The electrical resistivity of a TiN film of 45 nm deposited on Al2O3 was also measured for reference.
Thermal characterization of the TiB2 films was performed using a modulated photothermal radiometry (MPTR) setup, used and previously reported in Refs. [34,35].
The laser beam was modulated in frequency in the 0.7 to 5 kHz range and was focused at the surface of the films by an appropriate optical path. The films are capped with a platinum layer of approximately 100 nm which acts as the optical-to-thermal transducer.
The infrared (IR) radiation from the heated surface is collected by two parabolic mirrors and focused at the sensitive element of an IR detector. The phase between the reference excitation and the measured signal from the detector is measured using a lock-in amplifier. The equivalent thermal conductivity accounting for the TiBx film and the filmsubstrate interface is then identified by minimizing the quadratic gap between the measured phase and that calculated from the model that describes the heat diffusion within the sample in the experimental configuration. The heat diffusion model is based on the heat diffusion within the layer and the substrate, assuming isotropic properties for both materials. The model requires knowing the heat capacity per unit volume of the layer and the thermal properties of the substrate. The minimization is achieved by using the Levenberg-Marquardt algorithm.

Results and Discussion
The elemental composition of the films determined by ToF-ERDA is given in Tab. 1; the composition is henceforth used for sample notation. The residual gas contamination for the DCMS and HiPIMS thin films were also measured displaying similar impurity of roughly 0.5±0.1% for N, O and C. The composition of the thin films deposited with DCMS was TiB2.70 and TiB2.20 for a deposition pressure of 5 mTorr, and 20 mTorr, respectively. This overstoichiometry is consistent with previous reports on TiBx thin films from DCMS. [17,18] Furthermore, the reduced B content with increasing pressure is in line with previous work, [36] suggesting B transported preferentially along the target normal and Ti having a wider distribution angle at lower pressure, and with an increased B scattering with increased pressure reducing the overstoichiometry. [36] The films deposited with HiPIMS have composition TiB1.43 and TiB2.06 for a pressure of 5 mTorr and 20 mTorr, respectively. The reasons for this drastic reduction in B content compared to DCMS are discussed in Ref. [19]; The two thin films grown by HiPIMS, TiB1.43 and TiB2.06, were investigated by plan-view STEM, and the results are shown in Fig. 3 a) Fig. 4 and with a more detailed investigation of the stacking faults presented in Ref. [40]. The STEM image in Fig. 4 presents the core of a column, viewed on-axis, e.g., parallel to the <001> direction. The structure clearly exhibits a hexagonal appearance, however, stacking faults on the prismatic planes are visible throughout the grain. The stacking faults make up an intricate pattern, but remains fixed to the (1-100) prismatic planes of the TiB2 crystal structure. Further, it was found that stacking faults are deficient in B and rich in Ti, as judged from the elemental contrast in Fig. 4. The presence of such stacking faults thus serves to accommodate B deficiency in the bulk of the crystals. The inset in Fig. 3 a) and Fig. 4 shows an SAED and Fast Fourier Transform (FFT) patterns, respectively, from the TiB1.43 and TiB1.44 (900 o C) thin films, which indicates the predominantly polycrystalline with <001> texture, in agreement with XRD, Fig. 1 The elastic modulus, E, of respective film can be seen in Fig. 5 a). The phase of the thermal IR response obtained for the TiBx films are given as a function of frequency for the different B/Ti ratios and are shown in Fig. 6. The equivalent thermal conductivity of each film is identified from the measured phase and are reported in Tab

Conclusions
High-power impulse magnetron sputtering (HiPIMS) paves the way for a control of the stoichiometry of TiBx thin films. TiBx thin films with a B content of 1. 43

Conflicts of interest
There are no conflicts to declare.