Many applications of boron carbide (B4C) films entail deposition on non-planar substrates, necessitating a better understanding of oblique angle deposition phenomena. Here, we systematically study the effect of substrate tilt on properties of B4C films with thicknesses up to 10 μm deposited by direct current magnetron sputtering. Results show that all films are amorphous and columnar with an average column width of 100 nm, independent of substrate tilt. Column tilt angles are limited to 20° even for substrate tilt of 80°. Film density, residual stress, and the refractive index weakly (within 20%) depend on substrate tilt. Oxygen impurities bond preferentially with carbon atoms in inter-columnar regions. Substrate tilt has a major effect on mechanical properties that decrease by 50%, suggesting weak interconnection between nano-columns. Implications of these observations for the deposition onto non-planar substrates are discussed.

Boron carbide, with typical stoichiometry of B4C,1 possesses a unique combination of properties of interest to several applications such as light-weight armor, nuclear reactor components, and a diverse range of coatings for x-ray optics,2–4 neutron detectors,5,6 cutting and abrasive tools,7,8 bearings,9 shaving razor blades,10 chemically resistant components in semiconductor processing tools,11 the first wall of tokamaks,12 and hydrogen fuel ablator capsules for inertial confinement fusion (ICF).13–17 Most of these applications require the coating of non-planar substrates in regimes when the substrate normal is tilted away from the preferential direction of the depositing species flux. Such deposition on tilted substrates, commonly referred to as oblique angle deposition (OAD),18–20 is influenced by complex film growth phenomena related to atomic shadowing and diffusional processes of both surface adatoms and point defects in the near surface layer. For the development of robust coating processes onto non-planar substrates, it is desirable to better understand OAD-related phenomena and to identify deposition regimes agnostic to such phenomena.

The OAD regime of physical vapor-deposition has been studied for many decades, albeit mostly for metallic materials.18–27 It has been observed that, with increasing substrate tilt angle (α) above some critical value (typically 30°), most materials exhibit a transition from a regime when film properties are weakly dependent on α to a regime extremely sensitive to α. This transition is often so rapid that a scaling theory with large exponent values of 7 has been invoked to quantify some observations.20,27 For substrate tilts above the critical angle, film porosity and surface roughness rapidly increase, which are often undesirable. Hence, deposition onto non-planar substrates can sometimes benefit from the addition of apertures between the source and the substrate to collimate depositing species flux in order to minimize the deposition with effective substrate tilt angles above the critical one. In addition to the increased cost and complexity of the deposition apparatus, a penalty associated with such collimation is a reduction in the deposition rate.

We are not aware of any previous OAD studies for sputter-deposited B4C that measure critical substrate tilt angles, which are necessary for designing appropriate flux collimation schemes. Here, we systematically study the OAD of B4C films deposited by direct current magnetron sputtering (DCMS) under a set of conditions suggested by our recent study17 which reported a close-to-zero residual stress state. We use a faceted substrate holder designed as a mono-block to ensure thermal uniformity at elevated substrate temperatures required for the deposition of low-stress B4C films. Deposition onto such a faceted substrate holder is used as an experimental platform to identify conditions for coating non-planar substrates. Contrary to expectations based on previous OAD studies of other material systems,18–27 we find that residual stress, density, and the refractive index of B4C films are relatively weakly dependent on substrate tilt in the entire range studied. However, mechanical properties exhibit a significant (50%) decrease with increasing tilt, attributed to the development of a peculiar columnar microstructure in the OAD regime.

Films were made by DCMS in a cylindrical high-vacuum chamber, 44 cm in diameter and 36 cm in height. Prior to turning on the substrate heater, the base pressure of the chamber was 5×107 Torr with the substrate and chamber walls kept at 30°C. The base pressure increased to 5×106 Torr when the substrate holder was heated to 450 °C prior to deposition. The chamber was equipped with a 76-mm-diameter planar magnetron gun (MAK model from MeiVac Inc.). The B4C target (with a density of 2.51gcm3 and electrical resistivity of 2×104Ω cm) was 64 mm in diameter with an initial thickness of 9 mm. The target was bonded to a 76-mm-diameter, 3-mm-thick Cu backing plate.

A customized faceted substrate holder, machined from a solid Mo block, was used (Fig. 1). The holder featured a central facet positioned parallel to the target surface, corresponding to the substrate tilt angle (α) of 0°. The other four side facets were tilted at 20°, 40°, 60°, and 80°. Substrates were mechanically clamped to each facet, as also shown in Fig. 1. Two types of substrates were used: (i) 10×10mm2 Si (100) chips with an 200-nm-thick Ta metal layer sputter deposited on top in a separate DCMS run and (ii) 12×3mm2, 262-μm-thick Si (100) cantilevers. All substrates were cleaned with ethanol and air plasma prior to deposition. The Ta layer on Si chips was used as a marker in the areal density measurements by high-energy ion scattering as described below. Both chips and cantilevers were exposed to air before film deposition. Therefore, they are covered with their native oxides.

FIG. 1.

Photograph of the faceted substrate holder (machined from a Mo block), with 10×10mm2 Si chips and 12×3mm2 Si cantilevers clamped, positioned on the heater assembly on the floor of the coating chamber.

FIG. 1.

Photograph of the faceted substrate holder (machined from a Mo block), with 10×10mm2 Si chips and 12×3mm2 Si cantilevers clamped, positioned on the heater assembly on the floor of the coating chamber.

Close modal

The faceted substrate holder was placed on an electrically grounded Mo-body resistive heater with a diameter of 76 mm (HeatWave Labs Inc.). Temperatures of the heater, the faceted substrate holder, and substrates were monitored separately. Temperatures of the heater body and the faceted substrate holder were measured by thermocouples, while the substrate temperature was measured by imaging pyrometry (Optris GmbH, model PI 640). When the substrate holder was held at 450 °C during deposition, the substrate surface temperature was 330 °C. This temperature value is consistent with estimates based on radiative energy transfer between the holder and Si substrates. We note that, in our recent work,17 the same substrate temperature was used as in the present study, and the quoted temperature of 450 °C in Ref. 17 referred to the substrate holder temperature (measured with a thermocouple in both studies) rather than the surface temperature of substrates (measured by imaging pyrometry in the present study).

Sputter-deposition conditions for B4C films are summarized in Table I. These are based on the deposition conditions identified in our previous study resulting in close-to-zero residual stress.17 However, to ensure better deposition uniformity over the faceted holder, we have doubled the target-to-substrate distance (i.e., the throw) and halved the Ar pressure to maintain the pressure–distance product constant. Two separate deposition runs were made, referred to below as Series A and B, with the same deposition parameters, including the deposition rate, but with different deposition times, resulting in films with a thickness of 2μm (Series A) and 10μm (Series B) for the α=0° facet.

TABLE I.

Conditions used to deposit B4C films in the present study, including the chamber base pressure before deposition with the substrate holder heated to 450 °C (Pbase), Ar working gas (99.998% purity) pressure (PAr), target-to-substrate distance (L) defined as the distance between the target surface and the top facet of the substrate holder, target voltage (Vt), target power (Wt), substrate surface temperature (Ts), deposition rate (R), and the thickness of films for zero-tilt facets of the holder (h). The flow rate of Ar was 25 standard cubic centimeters per minute, and a 76-mm-diameter magnetron source was operated in a DCMS constant-power mode.

PbasePArLVtWtTsRh
(Torr)(mTorr)(cm)(V)(W)(°C)(μm/h)(μm)
6 × 10−6 10 400 300 330 0.25 1.94, 9.90 
PbasePArLVtWtTsRh
(Torr)(mTorr)(cm)(V)(W)(°C)(μm/h)(μm)
6 × 10−6 10 400 300 330 0.25 1.94, 9.90 

The Ta marker layer on square Si chips was deposited by DCMS in the sputter-down configuration with a 50-mm-diameter planar magnetron source (MAK model from MeiVac Inc.) powered at 100 W, an Ar pressure of 12 mTorr, and a target-to-substrate distance of 180 mm, resulting in a deposition rate of 0.43 μm/h. Substrates were placed on a holder held at 30°C. Such Ta-coated square Si chips were used only for density measurements. All the other physical properties were measured for films deposited onto cantilevers. The presence of this 200-nm-thick Ta layer on Si is expected to have a negligible effect on the properties of deposited films since films, as described below, are amorphous, which rules out epitaxy on the underlying substrate. Our scanning electron microscopy (SEM) examination of Si chips before and after Ta layer deposition (and prior to B4C deposition) has revealed minimal changes to surface morphology. We have also found that the presence of the Ta maker layer has no observable effect on the morphology of films as studied by SEM for films on chips and cantilevers from the same substrate holder facet.

The B/C stoichiometric ratio and oxygen content in all films were measured by Rutherford backscattering spectrometry (RBS) with 2 MeV 4He+ ions incident normal to the sample surface and backscattered into a detector located at 165° from the incident beam direction. The areal density was measured with 2 MeV 1H+ ions in the same scattering geometry. The energy shift of the signal from the Ta marker layer was used to measure the areal density. The analysis of RBS spectra was done with the RUMP code.28 

The physical thickness of films was first measured by conventional stylus profilometry (KLA-Tencor D-100). More accurate thickness measurements, required for mass density analysis, were done by cross-sectional SEM in a Thermofisher Apreo instrument operating at 2 kV. For all films, cross sections were prepared by mechanical fracture at room temperature. Mass density was calculated by dividing the areal density measured with RBS by the physical thickness measured by SEM. For selected films from Series B, cross sections were also prepared by focused Ga ion beam milling.

Crystallography and elemental mapping were investigated with transmission electron microscopy (TEM) in an FEI Titan 80-300 microscope (Thermo Fisher Scientific Inc.) operating at 300 kV and equipped with a OneView detector (Gatan Inc.) for TEM imaging and diffraction and a Gatan 965 GIF Quantum ER (Gatan Inc.) for electron energy loss spectroscopy (EELS). The analysis of EELS data including thickness and elemental maps was done with DigitalMicrograph v3.11 (Gatan Inc.). Cross sections for TEM were prepared by conventional focused Ga ion beam milling. A Keyence VK-X1000 laser confocal microscope was used for root mean square surface roughness (Rq) measurements over an area of 20×20μm2.

Mechanical properties were evaluated by nanoindentation in the load-controlled mode with an MTS XP nanoindenter with a Berkovich diamond tip. Meyer’s hardness (HM) was defined as average contact pressure, and Young’s modulus (EY) was calculated based on the Oliver–Pharr method.29 In Oliver–Pharr calculations, we assumed Poisson’s ratios of diamond and B4C films of 0.07 and 0.17, respectively, and a Young’s modulus of diamond of 1141 GPa.1,30 Measurements were performed over the indenter penetration depth range of 1020% of film thickness.

Residual stress in films was calculated with the Stoney equation based on the change in cantilever curvature measured by profilometry before and after deposition.31 The thermal stress component (σTE) originating from the difference in coefficients of thermal expansion between the film and the (Si) substrate was calculated as follows:

σTE=EY(1νf)(αfαs)ΔT,
(1)

where ΔT is the difference between film growth and stress measurement temperatures; EY and νf are the Young’s modulus and Poisson’s ratio of the film, respectively; and αf and αs are linear thermal expansion coefficients of the film and substrate, respectively. For ΔT=310 K of the present study, νf=0.17,1,αs=3.6×106K1,32 and αf=4.6×106K1,33 σTE is tensile and relatively small, in the range of 60130 MPa (tensile), decreasing with increasing α due to the α dependence of EY also measured in this study as described above. Hence, an average value of σTE of 100 MPa was used.

To resolve both in-plane and randomly distributed nanoscale morphology within the films, grazing incidence small-angle x-ray scattering (GISAXS) was employed. All films were studied with Cu–Kα x rays in a Xeuss 3.0 instrument (Xenocs Inc.), and a high-flux synchrotron x-ray beam was used for films with a low scattering signal. In both cases, samples were secured to a holder such that an 10 mm long film in-plane direction was parallel to the x-ray beam propagation direction. Films were illuminated with a 0.02(V) × 0.12(H) mm2, 13.3 keV monochromatic x-ray beam at tilt angles of 0.000.12° at beamline 12ID-B located within the Advanced Photon Source (APS). Samples were also measured with a lab-based Xuess 3.0 instrument under vacuum with a 2.0×0.5mm2 beam at tilt angles of 0.0°0.2°, which corresponds to roughly equivalent penetration depths as the measurements at beamline 12ID-B. A Pilatus 2M detector was used to collect the scattered x rays at 12ID-B, while a Pilatus3 300k detector was used in “line-erasure” mode whereby two images are obtained to remove the dead zones on the detector. The 2D images were analyzed with FitGISAXS module for Igor Pro34 and shown in reciprocal space units (qx, qy, qz); qx and qy are parallel to the substrate and qz is perpendicular to it. To a first approximation, the magnitude vector (|q|) is proportional to the dimensions of scattering centers (d) by the relationship: |q|2πd. In our study, GISAXS was analyzed in two different q-ranges to resolve large (d50 nm) and small (d5 nm) scattering features within the films. More details of the GISAXS experimental geometry and data reduction analysis can be found in the supplementary material.

The refractive index of the films was measured by spectroscopic ellipsometry (J.A. Woollam Co.) at several different locations on each sample with an incident beam diameter and angle of 125 μm and 70°, respectively. Optical constants were obtained by fitting the amplitude (Ψ) and phase difference (Δ) components to a mathematical spline model under Kramers–Kroning conditions to ensure physical consistency.

Monte Carlo modeling of gas-phase transport of sputtered B and C atoms was performed with the SiMTra code35 for our specific sputtering chamber and faceted substrate holder geometry described above. Trajectories of 107 atoms were tracked at room temperature separately for B and C, and values of landing energy (E) and incident angle (θ) were recorded for each atom. Results for B and C atoms were stoichiometrically combined for each facet for an area of 2 × 2 cm2. The initial energy and angular distributions of sputtered atoms, required for SiMTra, were calculated with the TRIM code36 in the monolayer collision step mode. The surface binding energy, lattice binding energy, and bulk displacement energy, for both B and C, were assumed to be 6.06, 3, and 20 eV, respectively.37 Predictive capabilities of Monte Carlo simulations, and the TRIM code,36 in particular, for sputtering related phenomena have well known limitations.38 Hence, the simulation predictions reported here should be treated as best available estimates when experimental data on sputtering yields and particle emission characteristics are not available.

We start by presenting the results of Monte Carlo simulations as they are instrumental in the subsequent discussion of experimental data. Distributions of atom landing energies (E) and incident angles (θ) for each substrate facet are shown in Fig. 2, while α dependencies of the major moments for such distributions are plotted in Fig. 3. It is seen from Fig. 2 that both E and θ distributions non-trivially change with α. Although all E distributions [Fig. 2(a)] extend to large values of 100 eV and have positive skewness [Fig. 3(c)] for all the facets, average landing energies [E¯, Fig. 3(a)] are only 7 eV and weakly dependent on α. The slight decrease in E¯ and ΔE with increasing α observed in Figs. 3(a) and 3(b) is related to a corresponding increase in the thermalized fraction of atoms, expressed here as counts in the energy bin of 1 eV and a decrease in the intensity of the high energy tail of the E distribution [Fig. 2(a)]. Such weak E¯(α) dependence is due to the fact that energy loss of light B or C atoms by collision with heavier Ar gas atoms is small, and the initial energy distribution of B and C atoms leaving the target surface, rather than energy loss in gas-phase collisions, determines their landing energy distribution. This was discussed in more detail in our previous study.17 

FIG. 2.

Distributions of (a) kinetic landing energy, with the cumulative fraction shown in the inset, (b) incident atom angles for all atoms, and (c) incident atom angles for atoms with energies above 10 eV. The legend in (a), relating symbols to different substrate tilt angles, applies to all panels and the inset. Results of SiMTra/TRIM code simulations for B and C atoms for the sputtering geometry and deposition conditions of this study.

FIG. 2.

Distributions of (a) kinetic landing energy, with the cumulative fraction shown in the inset, (b) incident atom angles for all atoms, and (c) incident atom angles for atoms with energies above 10 eV. The legend in (a), relating symbols to different substrate tilt angles, applies to all panels and the inset. Results of SiMTra/TRIM code simulations for B and C atoms for the sputtering geometry and deposition conditions of this study.

Close modal
FIG. 3.

Substrate tilt dependencies of (a) average kinetic landing energy (E¯) and average incident angle (θ¯) of depositing atoms, (b) the standard deviation of distributions of the landing energy (ΔE) and incident angle (Δθ), and (c) the skewness of E and θ distributions. Results of SiMTra/TRIM code simulations for B and C atoms. Results for all the atoms and atoms with energies above a threshold of 10 eV are shown as closed and open symbols, respectively. In all panels, the left axis applies to the E-related points are shown by circles, while the right axis refers to θ-related points depicted by squares.

FIG. 3.

Substrate tilt dependencies of (a) average kinetic landing energy (E¯) and average incident angle (θ¯) of depositing atoms, (b) the standard deviation of distributions of the landing energy (ΔE) and incident angle (Δθ), and (c) the skewness of E and θ distributions. Results of SiMTra/TRIM code simulations for B and C atoms. Results for all the atoms and atoms with energies above a threshold of 10 eV are shown as closed and open symbols, respectively. In all panels, the left axis applies to the E-related points are shown by circles, while the right axis refers to θ-related points depicted by squares.

Close modal

Figure 2(b) shows θ distributions for different α values. It reveals that all the θ distributions are bell-shaped with positive skewness for α in the range of 0°40° and negative skewness for larger α angles [Fig. 3(c)]. Note that skewness represents symmetry of the distribution curve where positive and negative values imply the distribution is skewed to the right and left, respectively. All these θ distributions, except for the α=0° facet, have a peculiar shape with a double peak structure, which deserves a further discussion. The θ distribution during magnetron sputtering is determined by a combination of the geometry of the magnetron racetrack, the initial angular distribution of sputtered atoms, and gas-phase scattering of the sputtered atoms on their journey from the target to the substrate. The deviation of θ distributions in Fig. 2(b) from the expected bell-shaped curve is due to gas-phase transport of the sputtered atoms specific to the current experimental conditions. This conclusion is supported by Fig. 2(c), showing energy-filtered θ distributions made only from high-energy atoms with energies >10 eV. A comparison of the total θ distributions [Fig. 2(b)] with such energy-filtered distributions [Fig. 2(c)] clearly reveals that the sharp peaks in the total distributions originate from high energy atoms, while thermalized atoms contribute to the broad tail causing the change in the skewness sign with increasing α [Fig. 3(c)].

Simulations further predict that there are no Ar atoms backscattered from the target for all the facets. The absence of backscattering/reflection of heavy 40Ar ions from the low atomic number B4C target is expected from ballistics.

These findings of Monte Carlo simulations will be used to interpret experimental data below.

Figure 4 summarizes substrate tilt dependencies of the major physical properties of B4C films with two different thickness ranges made in Series A and B runs. The areal density, directly measured by RBS [Fig. 4(a)], is the number of atoms deposited per unit area. It reflects the efficiency of sputtering and gas-phase atomic transport and is independent of film mass density (porosity). Hence, the α dependence of the areal density measured by RBS can be directly compared with the areal density predicted by SiMTra simulations, which is also shown in Fig. 4(a), normalized to the experimental data for the case of α=0°. It is seen from Fig. 4(a) that, for both Series A and B, the areal density of films monotonically decreases with increasing α since the same atomic flux is being deposited onto a larger substrate area with increasing substrate tilt.

FIG. 4.

Substrate tilt dependencies of (a) the areal density measured by RBS and the relative areal density calculated with SiMTra (cross symbols); (b) physical thickness; (c) mass density (ρ, where ρo=2.52g/cm3 is the theoretical maximum density of crystalline B4C); (d) residual stress (σ, where σTE is the stress contribution due to the thermal expansion mismatch between the substrate and the film); (e) surface roughness (Rq) measured over an area of 20×20μm2; (f) column tilt angle (β), with predictions of empirical cosine and tangent rules shown by solid and dashed lines, respectively; and (g) the O content in B4C films. Closed and open symbols show data for nominally 2-μm-thick (Series A) and 10-μm-thick (Series B) films, respectively. The legend in (a) relating the symbol type to nominal film thickness applies to all panels.

FIG. 4.

Substrate tilt dependencies of (a) the areal density measured by RBS and the relative areal density calculated with SiMTra (cross symbols); (b) physical thickness; (c) mass density (ρ, where ρo=2.52g/cm3 is the theoretical maximum density of crystalline B4C); (d) residual stress (σ, where σTE is the stress contribution due to the thermal expansion mismatch between the substrate and the film); (e) surface roughness (Rq) measured over an area of 20×20μm2; (f) column tilt angle (β), with predictions of empirical cosine and tangent rules shown by solid and dashed lines, respectively; and (g) the O content in B4C films. Closed and open symbols show data for nominally 2-μm-thick (Series A) and 10-μm-thick (Series B) films, respectively. The legend in (a) relating the symbol type to nominal film thickness applies to all panels.

Close modal

The shape of plots in Fig. 4(a) is determined by contributions from both ballistic and diffusive components of the deposition flux, the extreme cases of which are a cosine dependence and a constant, respectively. Indeed, for a case, when gas phase collisions of B and C atoms (with Ar atoms of the working gas) during their journey from the target to the substrate can be completely neglected, deposition flux is purely ballistic, and sputtered atoms do not change their directions or energies during their flight. In this purely ballistic case with collimated flux, the number of arriving B and C atoms onto the substrate (and, hence, the deposition rate) would follow a cosine function of the substrate tilt angle. In contrast, in the opposite extreme case when there are so many collision events of the depositing B and C atoms (with Ar gas atoms) that their direction is completely randomized, deposition flux is purely diffusive, and the deposition rate is independent of a substrate tilt angle.

All three curves in Fig. 4(a) are far from such extreme cases and are close to a linear dependence with a similar slope. The two experimental curves for Series A and B films essentially overlap, while SiMTra-derived areal densities are consistently slightly smaller than experimental ones with the normalization procedure used. This suggests limitations of the present SiMTra/TRIM code simulations, including those mentioned in Sec. II C, or experimental errors and warrants further studies.

The α dependence of physical thickness of films is plotted in Fig. 4(b). As expected, thickness decreases with increasing α for both Series A and B. The difference between α dependencies of the areal density [Fig. 4(a)] and physical thickness [Fig. 4(b)] reflects changes in mass density, which is plotted in Fig. 4(c). Density remains unchanged for the α range of 0°40°, followed by a monotonic decrease for α of 60° and 80°. To better understand this density dependence, we examine the film microstructure.

The microstructure of representative films from Series A and B (for α=0°, 40°, and 80°) is illustrated in SEM micrographs in Figs. 5 and 6, respectively. In these figures, left columns are plan-view and center and right columns are corresponding cross-sectional micrographs. These cross sections were prepared by cleaving or ion-milling the film and substrate such that the cross sectional plane is parallel to the plane defined by the line between the target and facet centers and facet normal direction. In this case, cross sections could be used to measure the inclination angle of columns (β) from the film normal direction. Conventional cross sections prepared by mechanical fracture of the film/substrate sandwich are typically used for visualizing grains (in polycrystalline films) and columns in columnar films (both crystalline and amorphous). Contrast is caused by the preferential propagation of cracks through the mechanically weak regions between grains and/or columns. However, with the fracture cross sections typically dominated by the columnar/grain structure, voids cannot be readily seen. In contrast, FIB-prepared cross sections have smooth surfaces, often decorated only by FIB-related artifacts (typically visible as high-symmetry streaks) of surface ripples formed during high-dose, large incident angle sputtering used for FIB milling. The intersection of such smooth FIB-prepared cross-sectional surfaces with voids can often be identified in SEM micrographs.

FIG. 5.

(Left column) plan-view and (right column) fracture cross-sectional SEM micrographs of Series A films deposited at three representative substrate tilt angles given in the legends in the left column. All scale bars are 1 μm.

FIG. 5.

(Left column) plan-view and (right column) fracture cross-sectional SEM micrographs of Series A films deposited at three representative substrate tilt angles given in the legends in the left column. All scale bars are 1 μm.

Close modal
FIG. 6.

(Left column) plan-view, (center column) fracture-cross-sectional, and (right column) FIB-cross-sectional SEM micrographs of Series B films deposited at three representative substrate tilt angles given in the legends in the left column. All scale bars are 3 μm. The vertical length scale of the FIB-cross-sectional SEM micrographs are off the scale due to the tilted stage (52°) during imaging. Representative voids are marked by arrows in FIB cross sections.

FIG. 6.

(Left column) plan-view, (center column) fracture-cross-sectional, and (right column) FIB-cross-sectional SEM micrographs of Series B films deposited at three representative substrate tilt angles given in the legends in the left column. All scale bars are 3 μm. The vertical length scale of the FIB-cross-sectional SEM micrographs are off the scale due to the tilted stage (52°) during imaging. Representative voids are marked by arrows in FIB cross sections.

Close modal

Both plan-view and fracture-prepared cross-sectional micrographs in Figs. 5 and 6 reveal a columnar structure for all the films in both Series A and B. An average column width of 100 nm is independent of substrate tilt and film thickness. The right column of Fig. 6, showing FIB-cross-sectional micrographs of thicker (Series B) films clearly reveals small (100 nm wide) inter-columnar voids that nucleate in films thicker than 1 and 2μm for cases of α of 60° and 80°, respectively. Representative voids are marked with arrows in Fig. 6. Interestingly, Fig. 6 does not reveal any evidence of an increase in porosity with further increasing film thickness after void nucleation begins. This is in agreement with the observation of similar densities for Series A and B films [Fig. 4(c)].

The dependence of β on α is plotted in Fig. 4(f). Columns have the same width of 100 nm for all substrate tilt conditions, and the columnar structure and tilt do not change with increasing film thickness up to the maximum value studied here (10μm). The β(α) dependence such as plotted in Fig. 4(f) has attracted much attention in the coating community and has been studied extensively for other materials (not B4C) over many decades.18–20 Two most frequently used correlations are the empirical cosine rule45 and the tangent rule,46 which are also plotted in Fig. 4(f) by solid and dashed lines, respectively. It is evident that the actual β(α) dependence for B4C is sublinear and much weaker than these empirical predictions. In fact, β saturates at 20° despite the presence of a well-defined ballistic component of depositing flux revealed by Figs. 2 and 3, described above. The presence of a ballistic component is critical for the formation of a tilted columnar structure as it is related to atomic shadowing. For isotropic flux of depositing species, the average angle of incoming atomic flux is independent of the substrate tilt angle, and β=0°.

Figure 7 plots β as a function of the average atom landing angle (θ¯) predicted by SiMTra/TRIM code simulations from Fig. 3(a). A comparison of Figs. 4(f) and 7 shows that, in contrast to the sublinear β(α) dependence, the β(θ¯) dependence is close to linear, suggesting that column tilt is determined by the incident angle of landing atoms. However, β is much lower than θ¯ and the β(θ¯) is much weaker than that predicted by the empirical cosine and tangent rules. Additional studies are needed to better understand how the very small column tilt and the existence of a tilt angle saturation regime are related to specific B4C film growth processes.

FIG. 7.

Column tilt angle (β), measured by SEM, plotted as a function of the average incident angle (θ¯) of depositing B and C atoms estimated by SiMTra/TRIM simulations. Corresponding values of substrate tilt angles (α) are given in the top axis. The solid line is a linear fit to all data points. The data points shown are the average of 35 measurements for each film, with error bars representing the standard deviation.

FIG. 7.

Column tilt angle (β), measured by SEM, plotted as a function of the average incident angle (θ¯) of depositing B and C atoms estimated by SiMTra/TRIM simulations. Corresponding values of substrate tilt angles (α) are given in the top axis. The solid line is a linear fit to all data points. The data points shown are the average of 35 measurements for each film, with error bars representing the standard deviation.

Close modal

Figure 4(e) shows the α dependence of surface roughness (Rq), revealing much larger Rq values for thicker films from Series B than for Series A films and a qualitatively different behavior for Series A and B for large tilt angles of 60° and 80°. Indeed, for thinner Series A films, Rq increases with increasing α, in agreement with the evolution of nanocolumns clearly visible in plan-view SEM micrographs of Fig. 5. This is consistent with the lack of driving force for surface faceting for an amorphous phase and suggests a relatively high mobility of adatoms, preventing the growth of energetically unfavorable surface hillocks and pores. Without adatom mobility, film growth in the OAD regime is expected to be unstable, resulting in highly porous films.18–20,25,39,40 Adatom mobility is required to counteract growth instabilities and atomic shadowing effects in the OAD regime.

In contrast, for thicker Series B films, Rq decreases with increasing α despite a similar size and morphology of nanocolumns in both Series A and B films, as evident from a comparison of SEM micrographs as shown in Figs. 5 and 6. Both observations of much larger Rq values (100 vs 3 nm) and a decrease in Rq with increasing α for Series B films are related to the formation of the so-called “nodules” (surface hillocks) for films with thicknesses 5μm, as illustrated by both plan-view and cross-sectional SEM micrographs of Fig. 6(a). The presence of such nodules greatly increases Rq, which is measured over an area of 20×20μm2. The density of nodules decreases with increasing α, which explains the decreasing Rq(α) dependence for films from Series B.

Similar nodule formation has been previously observed for sputter deposited B4C (Refs. 14, 17, and 41) and for many other materials systems,42–44,47 although there is still a limited understanding of their formation and growth mechanisms. Nodules are commonly associated with particulates deposited onto the surface during or before film growth. Our systematic study aimed at a better understanding of the mechanisms of nodule nucleation and growth in B4C films is currently under way and will be reported separately.

Two most common impurities in sputter deposited films are implanted working gas atoms (Ar in the present study) and O from the target impurities and the background water in the chamber. For all the films of this study, the Ar content was below the detection limit of RBS measurements (i.e., <0.2 at. %), consistent with our recent study conducted at similar conditions in the same deposition chamber,17 with expectations based on ballistics and with predictions of Monte Carlo simulations described above.

On the other hand, the O content shows a monotonic increase with α for both Series A and B films, as plotted in Fig. 4(g). This can be attributed to a corresponding reduction in the deposition rate with increasing α and an α-independent impingement rate of oxygen containing molecules from the residual gas in the chamber, particularly, given a relatively high chamber base pressure when the heater is turned on (Table I). The RBS data shown in Fig. 4(g) was acquired 5 days and 4 months after film deposition for Series B (thick) and Series A (thin) films, respectively. Hence, we cannot rule out a possibility that O uptake has occurred shortly after exposure of the films to air after deposition. The difference in the O content in Series A and B films is not related to film storage in air since a repeated RBS measurement has revealed that the O content has not changed for both Series B and A films stored in laboratory air for 7 and 11 months, respectively, since their deposition. This is in agreement with our earlier (unpublished) observations of negligible O uptake for other sputter deposited B4C films stored over a year in the laboratory.

The film microstructure and the distribution of B, C, and O in Series A films deposited at α of 0° and 60° are illustrated in bright-field (BF) TEM micrographs and EELS-derived elemental and thickness maps in Fig. 8. The film deposited at α=0° [Fig. 8(a)] exhibits an almost featureless BF TEM micrograph and little correlation of the elemental distribution, thickness, and BF STEM contrast. In contrast, the film deposited at α=60° [Fig. 8(b)] shows strong BF TEM contrast, reflecting the columnar structure and clear correlations in STEM-EELS maps. The nanocolumns, visible as wider darker contrast regions in the BF STEM image, are thicker and B-rich. The inter-columnar regions are thinner, exhibit brighter BF STEM contrast, and are O rich. The fluctuation of the C content is less prominent, but C is slightly more abundant in O-rich inter-columnar regions. Such periodic column boundaries can also be clearly observed in the BF TEM image. Despite these significant differences in the microstructure, both films are amorphous, as revealed by diffuse rings in the diffraction patterns shown as insets in Fig. 8. Such stoichiometric fluctuations with B-rich columns and O- and C-rich inter-columnar regions in obliquely deposited films suggest a difference in mobilities of B and C adatoms and preferential adsorption of O at intercolumnar regions. These stoichiometric fluctuations are local as our RBS measurements, with the scattering signal averaged over macroscopically large areas (with beam spot diameters 100μm), have revealed the same film stoichiometry as that of the starting sputtering target (B4C). This is in agreement with previous studies of sputter deposited compounds, reporting stoichiometric imbalance only for materials with a large mass difference of the target constituents.48–50 

FIG. 8.

Cross-sectional bright-field TEM micrographs of Series A films deposited at substrate tilt angles of (a) 0° and (b) 60°. Bright-field STEM micrograph and corresponding thickness and elemental maps acquired by STEM-EELS are shown in upper-corner insets. Lower corner insets depict selected-area diffraction patterns from these films. Scale bars in both (a) and (b) are the same.

FIG. 8.

Cross-sectional bright-field TEM micrographs of Series A films deposited at substrate tilt angles of (a) 0° and (b) 60°. Bright-field STEM micrograph and corresponding thickness and elemental maps acquired by STEM-EELS are shown in upper-corner insets. Lower corner insets depict selected-area diffraction patterns from these films. Scale bars in both (a) and (b) are the same.

Close modal

Mechanical properties derived from nanoindentation measurements are presented in Fig. 9. All the load-displacement curves have no discontinuities with a clear trend of a monotonic reduction of both EY and HM with increasing α. A corresponding decrease in mass density [Fig. 4(c)], associated with the development of porosity (Fig. 6), contributes to this reduction in mechanical properties for α60°. For α40°, where the density remains constant, however, a reduction in EY and HM could be attributed to the peculiar columnar microstructure revealed by Fig. 8. Boron-rich nanocolumns are separated by O-rich regions. A monotonic increase in the O content with increasing α [Fig. 4(g)] suggests an increase in the fraction of O-rich inter-columnar regions, weakening mechanical properties of films.

FIG. 9.

(a) Representative nanoindentation load-displacement curves for Series A films deposited at different substrate tilt angles, indicated in the legend. Only every 10th data points are depicted for clarity. (b) Substrate tilt dependence of (EY, left axis) Young’s modulus and (HM, right axis) Meyer’s hardness measured by nanoindentation for B4C films from both Series A and B.

FIG. 9.

(a) Representative nanoindentation load-displacement curves for Series A films deposited at different substrate tilt angles, indicated in the legend. Only every 10th data points are depicted for clarity. (b) Substrate tilt dependence of (EY, left axis) Young’s modulus and (HM, right axis) Meyer’s hardness measured by nanoindentation for B4C films from both Series A and B.

Close modal

Figure 4(d) shows the substrate tilt dependence of the residual stress in both Series A and B films, revealing close-to-zero intrinsic stress for all the cases. It is in agreement with our previous report17 for films deposited on substrates at α=0° under conditions characterized by similar landing atom ballistics.

Figure 10 shows GISAXS results for two representative films from Series A deposited at α of 0° and 60°. The modeling of the isotropic scattering component in GISAXS data at high-q (i.e., the diffuse halo in the left column of Fig. 10) suggests the presence of scattering centers with a mean size of 3 nm, which could be attributed to voids. The increase of scattering intensity for the α=60° film (compared to the α=0° case) suggests an increase in void density at larger tilt angles, which is consistent with a decrease in mass density [Fig. 4(c)]. In addition, the scattering data at low-q (the right column of Fig. 10) indicate the presence of anisotropic heterogeneities with dimensions of 50 nm parallel to the surface. The intensity scale of such anisotropic heterogeneities is larger for the α=60° film, which indicates these features are better defined in terms of electron density. These observations are consistent with a columnar structure with periodic compositional fluctuations as revealed in Fig. 8. More details of the GISAXS modeling and data reduction analysis can be found in the supplementary material.

FIG. 10.

GISAXS results for Series A films deposited at substrate tilt angles of (a) 0° and (b) 60°. Data shown for high-q (left column) and low-q (right column) range are obtained from beamline 12ID-B within APS and from the lab-based Xuess 3.0 instrument, respectively. Here, qx and qz indicate film in-plane and out-of-plane directions, respectively.

FIG. 10.

GISAXS results for Series A films deposited at substrate tilt angles of (a) 0° and (b) 60°. Data shown for high-q (left column) and low-q (right column) range are obtained from beamline 12ID-B within APS and from the lab-based Xuess 3.0 instrument, respectively. Here, qx and qz indicate film in-plane and out-of-plane directions, respectively.

Close modal

Finally, Fig. 11 shows the refractive index obtained by fitting spectroscopic ellipsometry data for Series A films. During fitting, we observed a tilting of interference oscillations on both Ψ and Δ as a function of wavelength, which is consistent with the existence of anisotropy in the films observed by SEM, TEM, and GISAXS data and discussed above. The best fit to the data was achieved with a uniaxial anisotropic model where in-plane (ordinary, nx=ny) optical properties are different from the ones across the film thickness (extraordinary, nz). Figure 11 shows that nx and ny reach a maximum (3.1) for α=20°40°, followed by a decrease for larger α. The initial increase at α=20°40° could be attributed to the development of the columnar structure with periodic density fluctuations, while a decrease at α=40°80° is consistent with a corresponding decrease in mass density [Fig. 4(c)]. The nz(α) dependence follows a similar trend but with values that are much higher than for nx and ny. This could be understood by the fact that the columnar structure of films with relatively small column tilt angles of β20° corresponds to an effectively higher electronic density and, hence, a higher refractive index along the out-of-plane (z) direction than for the in-plane (x and y) directions.

FIG. 11.

Substrate tilt dependence of in-plane (nx, ny) and out-of-plane (nz) refractive indices for Series A films measured by spectroscopic ellipsometry.

FIG. 11.

Substrate tilt dependence of in-plane (nx, ny) and out-of-plane (nz) refractive indices for Series A films measured by spectroscopic ellipsometry.

Close modal

We have systematically studied the effect of substrate tilt on the major properties of B4C films with thicknesses of 10μm deposited by direct current magnetron sputtering in a regime with close-to-zero residual stress characterized by low landing atom energetics. The main results of this work can be summarized as follows:

  • For the entire range of tilt angles studied (0°80°), films are stoichiometric B4C and amorphous with an anisotropic columnar structure with column tilt angles limited to 20° and column widths of 100 nm.

  • The columns are B-rich, while inter-columnar regions are O- and C-rich, with weaker bonding, contributing to a reduction in film density and mechanical properties with increasing substrate tilt and suggesting different mobilities of B and C adatoms.

  • Small-angle x-ray scattering suggests the presence of 3 nm voids in all the films, and 100 nm inter-columnar voids are observed by SEM for 1-μm-thick films deposited at oblique tilt angles of 60°.

  • The overall weak dependence of film properties of B4C on substrate tilt, compared to many other materials,19,20 is promising for applications requiring coating onto non-planar substrates.

See the supplementary material for more details of the GISAXS experimental geometry and data reduction analysis.

This work was performed under the auspices of the U.S. DOE by LLNL under Contract No. DE-AC52-07NA27344 and by GA under Contract No. 89233119CNA000063 and was supported by the LLNL-LDRD program under Project No. 20-ERD-029. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility, operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.

The data that support the findings of this study are available within the article and its supplementary material.

1.
F.
Thevenot
, “
Boron carbide—A comprehensive review
,”
J. Eur. Ceram. Soc.
6
,
205
(
1990
).
2.
D. L.
Windt
, “
Reduction of stress and roughness by reactive sputtering in W/B4C x-ray multilayer films
,”
Proc. SPIE
6688
,
R6880
(
2007
).
3.
R.
Soufli
,
A. L.
Aquila
,
F.
Salmassi
,
M.
Fernández-Perea
, and
E. M.
Gullikson
, “
Optical constants of magnetron-sputtered boron carbide thin films from photoabsorption data in the range 30 to 700 eV
,”
Appl. Opt.
47
,
4633
(
2008
).
4.
J. L.
Wu
,
R. Z.
Qi
,
Q. S.
Huang
,
Y. F.
Feng
,
Z. S.
Wang
, and
Z. H.
Xin
, “
Stress, roughness, and reflectivity properties of sputter-deposited B4C coatings for x-ray mirrors
,”
Chin. Phys. Lett.
36
,
120701
(
2019
).
5.
C.
Hòglund
,
J.
Birch
,
K.
Andersen
,
T.
Bigault
,
J. C.
Buffet
,
J.
Correa
,
P.
van Esch
,
B.
Guerard
,
R.
Hall-Wilton
,
J.
Jensen
,
A.
Khaplanov
,
F.
Piscitelli
,
C.
Vettier
,
W.
Vollenberg
, and
L.
Hultman
, “
B4C thin films for neutron detection
,”
J. Appl. Phys.
111
,
104908
(
2012
).
6.
Z.
Zhang
,
Y.
Liang
,
W.
Li
,
Z.
Wang
, and
H.
Chen
, “
Stress evolution in B4C and Cr mono-layer and B4C/Cr multilayer with variable layer thickness for neutron detectors
,”
Thin Solid Films
531
,
302
(
2013
).
7.
E.
Eichen
and
J. D.
Flasck
, “Disordered coating,” U.S. patent 4,716,083 (December 29, 1987).
8.
O.
Knotek
,
E.
Lugscheider
, and
C. W.
Siry
, “
Superhard PVD coatings in the B-C-N triangle
,”
Int. J. Refract. Met. Hard Mater.
17
,
157
(
1999
).
9.
G. L.
Doll
, “Bearing with amorphous boron carbide coating,” U.S. patent 6,517,249 B1 (February 11, 2003).
10.
R. F.
Boland
,
C. A.
Hultman
,
W. E.
Vreeland
, and
P. S.
Williams
, “Shaving razors,” U.S. patent 5,088,202 (February 18, 1992).
11.
H.
Shih
,
N.
Han
,
S. S. Y.
Mak
, and
G. Z.
Yin
, “Boron carbide parts and coatings in a plasma reactor,” U.S. patent 6,120,640 (September 19, 2000).
12.
O. I.
Buzhinskij
and
Y. M.
Semenets
, “
Thick boron carbide coatings for protection of tokamak first wall and divertor
,”
Fusion Eng. Des.
45
,
343
(
1999
).
13.
T. R.
Dittrich
,
S. W.
Haan
,
S.
Pollaine
,
A. K.
Burnham
, and
G. L.
Strobel
, “
NIF capsule design update
,”
Fusion Sci. Technol.
31
,
402
(
1997
).
14.
A. K.
Burnham
,
C. S.
Alford
,
D. M.
Makowiecki
,
T. R.
Dittrich
,
R. J.
Wallace
,
E. C.
Honea
, and
C. M.
King
, “
Evaluation of B4C as an ablator material for NIF capsules
,”
Fusion Sci. Technol.
31
,
456
(
1997
).
15.
A. S.
Moore
,
S.
Prisbrey
,
K. L.
Baker
,
P. M.
Celliers
,
J.
Fry
,
T. R.
Dittrich
,
K. J. J.
Wu
,
M. L.
Kervin
,
M. E.
Schoff
,
M.
Farrell
,
A.
Nikroo
, and
O. A.
Hurricane
, “
A simulation-based and analytic analysis of the off-Hugoniot response of alternative inertial confinement fusion ablator materials
,”
High Energy Density Phys.
20
,
23
(
2016
).
16.
R.
Chen
,
J.
Qi
,
Q.
Shi
,
X.
Guo
,
D.
Wu
,
T.
Lu
, and
Z.
Liao
, “
Rapid preparation and uniformity control of B4C ceramic double-curvature shells: Aim to advance its applications as ICF capsules
,”
J. Alloys Compd.
762
,
67
(
2018
).
17.
A. M.
Engwall
,
L. B.
Bayu Aji
,
S. J.
Shin
,
P. B.
Mirkarimi
,
J. H.
Bae
, and
S. O.
Kucheyev
, “
Sputter-deposited low-stress boron carbide films
,”
J. Appl. Phys.
128
,
175301
(
2020
).
18.
A. G.
Dirks
and
H. J.
Leamy
, “
Columnar microstructure in vapor-deposited thin films
,”
Thin Solid Films
47
,
219
(
1977
).
19.
L.
Abelmann
and
C.
Lodder
, “
Oblique evaporation and surface diffusion
,”
Thin Solid Films
305
,
1
(
1997
).
20.
A.
Barranco
,
A.
Borras
,
A. R.
Gonzalez-Elipe
, and
A.
Palmero
, “
Perspectives on oblique angle deposition of thin films: From fundamentals to devices
,”
Prog. Mater. Sci.
76
,
590153
(
2016
).
21.
A. M.
Engwall
,
L. B.
Bayu Aji
,
A. A.
Baker
,
S. J.
Shin
,
J. H.
Bae
,
S. K.
McCall
,
J. D.
Moody
, and
S. O.
Kucheyev
, “
Effect of substrate tilt on sputter-deposited AuTa films
,”
Appl. Surf. Sci.
547
,
149010
(
2021
).
22.
J. A.
Thornton
, “
Influence of apparatus geometry and deposition conditions on the structure and topography of thick sputtered coatings
,”
J. Vac. Sci. Technol.
11
,
666
(
1974
).
23.
J. A.
Thornton
and
D. W.
Hoffman
, “
Stress-related effects in thin films
,”
Thin Solid films
171
,
5
(
1989
).
24.
S.
Sun
,
B.
Yu
,
T.
Guo
,
S.
Yao
,
Y.
Liu
,
W.
Deng
,
C.
Li
, and
C.
Jin
, “
Mo/Si multilayers sputtered onto inclined substrates: Experiments and simulations
,”
Opt. Express
28
,
13516
(
2020
).
25.
J.
Dalla Torre
,
G. H.
Gilmer
,
D. L.
Windt
,
R.
Kalyanaraman
,
F. H.
Baumann
,
P. L.
O’Sullivan
,
J.
Sapjeta
,
T.
Dıiaz de la Rubia
, and
M.
Djafari Rouhani
, “
Microstructure of thin tantalum films sputtered onto inclined substrates: Experiments and atomistic simulations
,”
J. Appl. Phys.
94
,
263
(
2003
).
26.
A.
Siad
,
A.
Besnard
,
C.
Nouveau
, and
P.
Jacquet
, “
Critical angles in DC magnetron glad thin films
,”
Vacuum
131
,
305
(
2016
).
27.
M.
Backholm
,
M.
Foss
, and
K.
Nordlund
, “
Roughness of glancing angle deposited titanium thin films: An experimental and computational study
,”
Nanotechnology
23
,
385708
(
2012
).
28.
L. S.
Doolittle
, “
Algorithms for the rapid simulation of Rutherford backscattering spectra
,”
Nucl. Instrum. Methods Phys. Res. B
9
,
344
(
1985
).
29.
W. C.
Oliver
and
G. M.
Pharr
, “
An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments
,”
J. Mater. Res.
7
,
1564
(
1992
).
30.
S. O.
Kucheyev
,
A. V.
Hamza
,
J. H.
Satcher
, and
M. A.
Worsley
, “
Depth-sensing indentation of low-density brittle nanoporous solids
,”
Acta Mater.
57
,
3472
(
2009
).
31.
A. J.
Perry
,
J. A.
Sue
, and
P. J.
Martin
, “
Practical measurement of the residual stress in coatings
,”
Surf. Coat. Technol.
81
,
17
(
1996
).
32.
H.
Watanabe
,
N.
Yamada
, and
M.
Okaji
, “
Linear thermal expansion coefficient of silicon from 293 to 1000 K
,”
Int. J. Thermophys.
25
,
221
(
2004
).
33.
L. B.
Bayu Aji
,
S. J.
Shin
,
J. H.
Bae
,
A. M.
Engwall
,
J. A.
Hammons
,
X.
Lepro Chavez
,
N.
Catarineu
,
P. B.
Mirkarimi
, and
S. O.
Kucheyev
, “Effect of substrate temperature on sputter-deposited boron carbide films” (unpublished).
34.
D.
Babonneau
, “
FitGISAXS software package for modelling and analysis of GISAXS data using IGOR Pro
,”
J. Appl. Cryst.
43
,
929
(
2010
).
35.
K.
Van Aeken
,
S.
Mahieu
, and
D.
Delpa
, “
The metal flux from a rotating cylindrical mangetron: A Monte Carlo simulation
,”
J. Phys. D: Appl. Phys.
41
,
205307
(
2008
).
36.
J. F.
Ziegler
,
M. D.
Ziegler
, and
J. P.
Biersack
, “SRIM-2013 - The stopping power and range of ions in matter,”
Nucl. Instrum. Methods Phys. Res. Sect. B
268
(
11
),
1818
1823
(
2010
).
37.
E.
Hechtl
,
A.
Mazanec
,
W.
Eckstein
,
J.
Roth
, and
C.
Garcia-Rosales
, “
Sputtering behavior of boron and boron carbide
,”
J. Nucl. Mater.
196
,
713
(
1992
).
38.
K.
Wittmaack
, “
Reliability of a popular simulation code for predicting sputtering yields of solids and ranges of low-energy ions
,”
J. Appl. Phys.
96
(
5
),
2632
(
2004
).
39.
R. P. U.
Karunasiri
,
R.
Bruinsma
, and
J.
Rudnick
, “
Thin-film growth and the shadow instability
,”
Phys. Rev. Lett.
62
,
788
(
1989
).
40.
G. S.
Bales
and
A.
Zangwill
, “
Macroscopic model for columnar growth of amorphous films by sputter deposition
,”
J. Vac. Sci. Technol. A
9
,
145
(
1991
).
41.
C.
Mattsson
,
G.
Randall
,
H.
Xu
,
J.
Knipping
,
D.
Wall
,
B.
Blue
, and
A.
Nikroo
, “Boron carbide materials for inertial confinement fusion,” in Poster Presentation at 55th Annual Meeting of the APS Division of Plasma Physics, Denver, Colorado, November 11–15 (2013).
42.
T.
Spalvins
and
W. A.
Brainard
, “
Nodular growth in thick-sputtered metallic coatings
,”
J. Vac. Sci. Technol.
11
,
1186
(
1974
).
43.
R. J.
Tench
,
R.
Chow
, and
M. R.
Kozlowski
, “
Characterization of defect geometries in multilayer optical coatings
,”
J. Vac. Sci. Technol. A
12
,
2808
(
1994
).
44.
R. N.
Tait
,
T.
Smy
,
S. K.
Dew
, and
M. J.
Brett
, “
Nodular defect growth and structure in vapor deposited films
,”
J. Electron. Mater.
24
,
935
(
1995
).
45.
J. M.
Nieuwenhuizen
and
H. B.
Haanstra
, “
Microfractography of thin films
,”
Philips Techol. Rev.
27
,
87
(
1966
).
46.
R. N.
Tait
,
T.
Smy
, and
M.
Brett
, “
Modeling and characterization of columnar growth in evaporated films
,”
Thin Solid Films
226
,
196
(
1993
).
47.
P.
Panjan
,
A.
Drnovsek
,
P.
Gselman
,
M.
Cekada
, and
M.
Panjan
, “
Review of growth defects in thin films prepared by PVD techniques
,”
Coatings
10
,
447
(
2020
).
48.
C.
Mitterer
, “
Borides in thin film technology
,”
J. Solid State Chem.
133
,
279
(
1997
).
49.
J.
Neidhardt
,
S.
Mraz
,
J. M.
Schneider
,
E.
Strub
,
W.
Bohne
,
B.
Liedke
,
W.
Moller
, and
C.
Mitter
, “
Experiment and simulation of the compositional evolution of Ti-B thin films deposited by sputtering of a compound target
,”
J. Appl. Phys.
104
,
063304
(
2008
).
50.
B.
Bakhit
,
I.
Petrov
,
J. E.
Greene
,
L.
Hultman
,
J.
Rosen
, and
G.
Greczynski
, “
Controlling the B/Ti ratio of TiBx thin films grown by high-power impulse magnetron sputtering
,”
J. Vac. Sci. Technol., A
36
,
030604
(
2018
).

Supplementary Material