The residual coating stress and its control is of key importance for the performance and reliability of silicon nitride (SiNx) coatings for biomedical applications. This study explores the most important deposition process parameters to tailor the residual coating stress and hence improve the adhesion of SiNx coatings deposited by reactive high power impulse magnetron sputtering (rHiPIMS). Reactive sputter deposition and plasma characterization were conducted in an industrial deposition chamber equipped with pure Si targets in N2/Ar ambient. Reactive HiPIMS processes using N2-to-Ar flow ratios of 0 and 0.28–0.3 were studied with time averaged positive ion mass spectrometry. The coatings were deposited to thicknesses of 2 μm on Si(001) and to 5 μm on polished CoCrMo disks. The residual stress of the X-ray amorphous coatings was determined from the curvature of the Si substrates as obtained by X-ray diffraction. The coatings were further characterized by X-ray photoelectron spectroscopy, scanning electron microscopy, and nanoindentation in order to study their elemental composition, morphology, and hardness, respectively. The adhesion of the 5 μm thick coatings deposited on CoCrMo disks was assessed using the Rockwell C test. The deposition of SiNx coatings by rHiPIMS using N2-to-Ar flow ratios of 0.28 yield dense and hard SiNx coatings with Si/N ratios <1. The compressive residual stress of up to 2.1 GPa can be reduced to 0.2 GPa using a comparatively high deposition pressure of 600 mPa, substrate temperatures below 200 °C, low pulse energies of <2.5 Ws, and moderate negative bias voltages of up to 100 V. These process parameters resulted in excellent coating adhesion (ISO 0, HF1) and a low surface roughness of 14 nm for coatings deposited on CoCrMo.
I. INTRODUCTION
Silicon nitride compounds are hard, fracture tough, wear resistant, and proved functional under extreme conditions due to their chemical as well as thermal stability.1 Silicon nitride was introduced as implant material due to its biocompatibility and solubility in aqueous solutions, where its bio-functional properties depend on the composition, synthesis method, and porosity.2–5 Silicon nitride coatings (SiNx) are currently investigated to be deposited on metal joint replacements.5–10 Such coatings may prevent the metal ion release from the CoCrMo bulk material and simultaneously improve the wear resistance of the implant surface while preserving the tough bulk material. Additionally, potential SiNx wear debris can dissolve in the aqueous solution4 with tuneable dissolution rates11 and may therefore reduce any undesirable biological response.
In an earlier study,12 we demonstrated the growth of amorphous SiNx coatings and presented the activation paths for the N2 molecule during reactive high power impulse magnetron sputtering (rHiPIMS) at different N2/Ar flow ( ratios. Here, the use of between 0.28 and 0.3 yield SiNx coatings with approximately 50 at. % N, where the Si-N bonds prevail, and the Si-Si bonds are absent. Such high N containing coatings showed generally favorable coating properties as coating densities of 2.98 g/cm3 and hardness values of up to ∼27.5 GPa as well as dissolution rates below 0.5 nm/day (Ref. 11) were measured. However, increasing N-contents resulted also in comparatively high residual coating stresses of up to −1.5 GPa, leading potentially to adhesive failure.
In literature, coating stresses of sputter-deposited films are often reported to be compressive. Next to the incorporation of impurities like oxygen and hydrogen as well as the inert sputter gas, compressive stresses were found to be generated by energetic particle bombardment also known as shot-peening or forward sputtering.13–17 As the deposition is carried out at comparatively low substrate-to-melting temperatures (Ts/Tm) < 0.2, forward sputtering induces a distortion of the growing film by a displacement of atoms to non-equilibrium sites. This implies that the intrinsic residual coating stresses of amorphous coatings depend on the deposition parameters that influence particle energies and ad-atom mobilities. These parameters, in turn, influence a number of coating properties among them are the density, morphology, and hardness. Moreover, thermal stress arises as an additional substrate heating is applied, and the substrate and the coating material possess different coefficients of thermal expansion (CTE).18,19
Compared to mid-frequency and direct current magnetron sputtering, HiPIMS was reported to yield thin films with comparatively low stress levels,20,21 while the film hardness and density were not sacrificed.22,23 The reduced stress in films deposited by HiPIMS is commonly attributed to the method's inherent increased flux of ionized target material together with a decreased flux of ionized working gas and implies a moderate total flux of ions.24–28 This is particularly the case for the metal-ion rich phase of the HiPIMS pulse.22,23 Here, the generation of the intrinsic coating stress induced by energetic particle bombardment can be substantially decreased if additionally the applied substrate bias voltage is synchronized to this metal-ion rich phase of the pulse to avoid bombardment by inert gas ions and doubly charged metal ions.22 However, this approach may only be of advantage for process setups using a specific target material - inert gas combinations providing a practical temporal discrimination of the plasma species and a setup that uses one cathode that is operated in HiPIMS mode as well as flat substrates, where the substrate rotation is not required.
As may be expected, bio-medical applications put highest requirements on the performance and reliability of SiNx coatings. Therefore, the understanding of the origin of residual coating stress and its control are of key importance, since the residual coating stress determines to a great extent the coating adhesion and thus, their performance. Although the generation of coating stresses is consistently studied, especially parameters of industrially relevant and scalable rHiPIMS processes for the control of residual stresses need material-specific investigation as thus far mainly crystalline films based on transition metals were explored.20,29,30 Furthermore, for the growth of SiNx on medical implants, the deposition process set-up needs to permit the growth of comparatively thick (4–8 μm) SiNx coatings exhibiting reproducible coating properties with excellent uniformity on complex shaped substrates. Our study focuses therefore on the stress defining sputter deposition parameters for amorphous SiNx coatings, and their impact on the coating growth rate, morphology, and hardness as well as adhesion.
II. EXPERIMENTAL SETUP
The SiNx film growth and the plasma characterization were carried out in an industrial coater CC800/9 ML (CemeCon AG, Germany). A base pressure of <2 mPa was achieved prior to sputtering. In this study, Si targets with a purity of 99.8% and N2-to-Ar flow ratios () of 0 and 0.28–0.3 were used to conduct the sputter processes. The (Eq. (1)) is defined as
where and are the N2 and Ar gas flows, respectively. Conventional P-doped Si(001) wafers served as substrates. A target – substrate distance of 60 mm, and a pulse width of 200 μs were used for all film depositions. The negative pulsed bias voltage (UB) was synchronized to the cathode pulses. The deposition pressure (pd), pulse frequency, average target power, and the substrate temperature (Ts) were varied between 200–600 mPa, 0.2–4 kHz, 600–4000 W, and 145–340 °C, respectively.
The composition, energy distribution, and temporal evolution of metallic and reactive discharges were investigated with time-averaged plasma mass spectrometry. Ion energy distribution functions (IEDFs) of most abundant, positively charged species were recorded using a Hiden EQP 1000. Time-averaged data of metallic processes were acquired with an orifice opening of 50 μm at an orifice to target distance of 60 mm, while the data of reactive processes were recorded with an orifice opening of 300 μm at an orifice to target distance of 320 mm. In the latter set-up, IEDFs of the isotopes 15N, 29Si, 30Si, and 36Ar were recorded due to the similarity of the masses and abundance of Si and N ions. All IEDFs were recorded between −0.4 eV and 30 eV using a step width of 0.5 eV. The dwell time of the spectrometer was set to average over at least 30 HiPIMS pulses.
Since the deposited SiNx coatings were amorphous, their residual stress was obtained from the substrate curvature of the Si(001) substrate. The curvature was assessed by X-ray diffraction (XRD, PANalytical Empyrean, Netherlands).31 The diffractometer, equipped with a Cu Kα1 source, was operated at 45 kV and 40 mA. As coating thicknesses of 2 μm and 5 μm are insignificant compared to the substrate thickness of 525 μm (Ref. 32), the Stoney formula for anisotropic single crystal Si(001) was used to extract the residual coating stress from the measured substrate curvature. Here, uniform plane stress in the coatings was assumed.33
The SiNx morphology, thickness, and hence its growth rate were studied by cross-sectional scanning electron microscopy (SEM, LEO 1550 Gemini, Zeiss, Germany). The microscope was operated at an acceleration voltage of 3 kV, and a working distance of ∼3 mm.
The SiNx hardness and the elastic modulus were measured with a CSIRO UMIS nanoindenter (Fischer-Cripps Laboratories, New South Wales, Australia) using a three-sided Berkovich tip. The coatings were tested in the load-controlled mode. At least 30 indents of increasing loads between 1 mN and 15 mN were performed. For evaluation of the hardness, the method suggested by Oliver and Pharr34 was applied.
A second set of SiNx coatings with a thickness of approximately 5 μm was prepared on CoCrMo substrates in order to probe their adhesion by Rockwell C indentation tests. The CoCrMo substrates were polished to an average roughness (Ra) of 11 ± 2 nm. Using a 1-fold substrate rotation, a Cr seed layer was deposited on the CoCrMo substrate prior to the SiNx growth. The indents were performed and classified according to ISO26443:2008,35 and additionally classified to the HF scale.36 The Rockwell indenter was equipped with a diamond tip with an angle of 120° and a load of 1471 N (150 kp) was applied. The indents were examined in an optical microscope using 10× magnification and a digital camera.
The average roughness (Ra) of SiNx coatings deposited on CoCrMo substrates was measured using surface profilometry (Dektak XT, BRUKER AXS SAS). A stylus with a tip radius of 2 μm and a force of 29.4 μN were used to perform at least three scans with a length of 5 mm on different sample positions.
The composition and chemical bond structure of the SiNx coatings was extracted from X-ray photoelectron spectroscopy measurements (Axis UltraDLD, Kratos Analytical, Manchester, UK) using monochromatic Al(Kα) X-ray radiation (hν = 1486.6 eV). XPS survey scans and core level spectra of the Si2p, Ar2p, N1s, C1s, and O1s regions were acquired from as-received coatings and after sputter cleaning for 120 s with a 2 keV Ar+ beam. The Ar+ beam was rastered over an area of 3 × 3 mm2 at an incidence angle of 20°. Automatic charge compensation was applied throughout the acquisition. The composition of sputter cleaned SiNx coatings was extracted from the core level spectra after subtraction of a Shirley-type background. Here, the elemental cross sections provided by Kratos Analytical were used. The Si2p core level spectra of as-received samples were studied with regard to their bond structure. All core level spectra were referenced to the C-C/C-CH bond at 285 eV (Ref. 37) after subtraction of the Shirley-type background. For the deconvolution of the spectra, Voigt peak shapes with a Lorentzian contribution of 30% were used. The full width at half maximum (FWHM) of all components was restricted to 1.8 eV.
III. RESULTS AND DISCUSSION
A. Global deposition parameter influencing the residual coating stress
Our previous study12 showed that increasing N-contents up to ∼51 at. % in the coating, result in increasing compressive stresses up to −1.5 GPa. The evolution of the residual stress was attributed to increased bond and coating densities induced by the reduced Si-N bond lengths, as well as increasing number of polyatomic species in the sputter plasma as the was raised up to ∼0.30.38 Moreover, the increase of the residual compressive stress was accompanied by an increasingly different CTE of the Si(001) substrate and the coating as the N-contents in the SiNx increased. In the following, the influence of substrate temperature and process pressure on the residual stress of SiNx coatings with a Si-to-N ratio of ≤1.1 deposited at of 0.28 are discussed.
Fig. 1(a) shows the residual stress of SiNx coatings over substrate temperatures ranging between 145 °C and 340 °C. The coatings were deposited at a process pressure of 200, 400, and 600 mPa using a of 0.28. Increasing the substrate temperature results in increasing compressive residual stresses from 0.87 GPa to 1.67 GPa and from 0.20 GPa to 1.02 GPa for coatings deposited at process pressures of 400 mPa and 600 mPa, respectively. The compressive residual stress increases for both sets of coatings with approximately the same rate of 4.1 MPa/K. This suggests that the increase has the same origin and may be governed by the differences in the CTE of the Si substrate and the SiNx coating.39,40 According to Ref. 39, the CTEs range between 2.8 × 10−6 and 3.8 × 10−6 K−1 for the Si substrate (S) and between 1.9 × 10−6 and 3.5 × 10−6 K−1 for the SiNx coating (C) within the range of applied substrate temperatures and are indicated in Fig. 1(a).
An opposite trend is observed for coatings deposited at a process pressure of 200 mPa. Here, the compressive stress is initially high and decreases from 2.09 GPa to 1.53 GPa with an increase of the substrate temperature. The effect induced by the difference of the CTEs appears to vanish, while reduced stress values are most likely induced by an increased ad-atom mobility and relaxation of the stress as the Ts/Tm approach 0.2 during deposition.17,41 A similar trend was reported for Si by Windischmann.42
Fig. 1(b) shows that the increase of Ts yields also a significant increase of the SiNx hardness when coatings are deposited at a process pressure of 400 and 600 mPa. This can be attributed to their increasing compressive stresses. Increasing hardness values are commonly observed as compressive stresses increase43 and are explained by a reduced amount of slip events or reduced viscous flow in crystalline materials or amorphous materials, respectively.44 In our previous study we found that the SiNx coatings were amorphous12 hence, the latter applies for the here investigated coatings.
The reduced hardness of coatings deposited at 200 mPa (cf. Fig. 1(b)) is likely caused by a slightly lower N-content of ∼47 at. % compared to coatings deposited at 400 mPa that exhibit N-contents between 51 and 52 at. % and those deposited at 600 mPa with N-contents of ∼49.5 at. %. At a low process pressure of 200 mPa, the hysteresis of the discharge changes slightly, and the transition regime can be observed for s between 0.36 and 0.40. Therefore, coatings deposited at 200 mPa using a of 0.28 comprise also the Si-Si bonds (cf. supplementary material, Fig. S1, XPS Si 2 p spectra) with a lower bond strength of ∼2.30 eV as compared to the strength of Si-N bonds of ∼3.68 eV. The presence of Si-Si bonds, in turn, promotes a lowered density and hardness of the coatings.12,44–46 For SiNx coatings with N-contents of ≥50 at. %, effects induced by an increased substrate temperature or forward sputtering on the tetrahedral bond coordination of Si, and thus also on the coating properties, could not be revealed but can also not be excluded, due to the amorphous nature of the coatings.
As expected and indicated in Fig. 1(a), the deposition pressure influences the residual stress of SiNx coatings significantly. Fig. 2 shows the residual stress of the above discussed SiNx coatings over the deposition pressure. At a substrate temperature of 145 °C, the residual stress drops by more than 1.8 GPa from −2.09 GPa to −0.2 GPa as the process pressure increases from 200 mPa to 600 mPa (cf. Fig. 1, blue squares). For sputter deposited coatings, the relief of compressive stresses by an increased process pressure is consistently reported.17,18,47–51 This is ascribed to lower energies and the amounts of ions in the plasma that impinge on the growing coating surface and induce a forward sputtering. The results are also corroborated by time-averaged ion mass spectrometry measurements shown in Figs. 3(a) and 3(b) as well as the collision mean free paths for plasma species at process pressures of 200 mPa, 400 mPa, and 600 mPa presented in Table I.52
. | Process pressure (mPa) . | ||
---|---|---|---|
. | 200 . | 400 . | 600 . |
Collisions mean free path (mm), Ar | 143 | 72 | 48 |
Collisions mean free path (mm), N2 | 135 | 68 | 45 |
Collisions mean free path (mm), Si | 130 | 70 | 40 |
. | Process pressure (mPa) . | ||
---|---|---|---|
. | 200 . | 400 . | 600 . |
Collisions mean free path (mm), Ar | 143 | 72 | 48 |
Collisions mean free path (mm), N2 | 135 | 68 | 45 |
Collisions mean free path (mm), Si | 130 | 70 | 40 |
The effect of the pressure on the energy and amount of different positively charged plasma species for a of ∼0.3 at room temperature is shown in Figs. 3(a) and 3(b). Here, the target-orifice distance was 320 mm using an orifice opening of 300 μm. Both processes were conducted using a pulse frequency of 300 Hz and an average target power of 1000 W, resulting in an energy per pulse (EpP) of 3.4 Ws. Comparing the IEDFs of Figs. 3(a) and 3(b), it is evident that the energies of abundant plasma species decrease as the process pressure is increased from 200 to 400 mPa. This is attributed to the decreased mean free path for collisions (cf. Table I)52 as well as a decreased average discharge voltage by 40 V and accounts for the reduced stress values as the process pressure is increased. From Table I and the substrate-to-target distance of 60 mm, it can be inferred that particles undergo at least one collision at a process pressure of 600 mPa, while the collision mean free path at 400 mPa and 200 mPa increases to approximately 70 mm and 140 mm, respectively.
Furthermore, Fig. 2 shows that the influence of the pressure on the residual stress decreases as the substrate temperature increases. This is mainly attributed to the evolution of the coating morphology as the deposition pressure and the substrate temperature are increased. Their influence on the coating morphology is presented by cross-sectional SEM images in Figs. 4(a)–4(d). The coating morphology changes from a featureless to columnar appearance as the growth is conducted at a substrate temperature of 145 °C, and the pressure is raised from 200 to 600 mPa (cf. Figs. 4(a) and 4(c)). An increasingly homogeneous appearance of coatings grown at 600 mPa is observed at an elevated substrate temperature of 340 °C (cf. Fig. 4(d)). Columnar morphologies, as observed for the coatings deposited at 600 mPa and 145 °C, may be linked to reduced coating stresses by Ref. 53. The presented SEM images corroborate also the above discussed results on the residual stress and mass spectrometry; The coating morphology at 200 mPa is strongly influenced by forward sputtering due to high amounts and energies of ions resulting in a featureless cross-section, while the growth at 145 °C and a process pressure of 600 mPa is governed by a low ad-atom mobility and the absence of forward sputtering resulting in a coarse columnar appearance.54
B. HiPIMS parameters influencing the residual coating stress
Fig. 5(a) shows the residual stress over the pulse energy, EpP, obtained from coatings deposited at pulse frequencies between 0.4 and 4 kHz. The corresponding coatings were grown using a substrate temperature of 230 °C, a deposition pressure of 400 mPa, and a of 0.28. The substrate temperature of 230 °C results in coatings with low levels of impurities and simultaneously provides a Ts/Tm < 0.2 to avoid additional effects on the coating stress by elevated Ts. Fig. 5(b) presents the residual stress over the corresponding pulse frequency normalized growth rate, RG. From Figs. 5(a) and 5(b), it is evident that the compressive residual stress increases with the EpP, the applied pulse frequency, and the related SiNx growth rate. Similar observations were also reported by Chason et al.55
Using the same pulse frequency, an increased EpP implies an increasing average target power and hence elevated sputter rates and ionization. Therefore, an elevated compressive stress at increasing pulse energies can be partly ascribed to forward sputtering induced by ion bombardment as discussed above. Ion energies of Si sputter processes in pure Ar are exemplified in Fig. 6(a)–6(d) for different power and frequency settings. Although, ion energies are generally lower in processes using of 0.28 due to the temporarily poisoned target and N2 dissociation, ionization as well as excitation,12 the results in Figs. 6(a)–6(d) suggest higher energies and amounts of ions as the average target power (cf. Fig. 6(c)) and/or the pulse frequency (cf. Fig. 6(d)).
In Figs. 6(a) and 6(b), IEDFs of discharges in pure Ar using pulse frequencies of 300 Hz and 1000 Hz at an average power of 1000 W, corresponding to pulse energies (EpP) of 3.4 Ws and 1 Ws, respectively, are shown. IEDFs presented in Figs. 6(c) and 6(d) were recorded for discharges in pure Ar carried out at 300 Hz and 1000 Hz with an average target power of 3000 W resulting in pulse energies of 10 Ws and 3 Ws, respectively. The comparison of respective IEDFs in Figs. 6(a)–6(d) exemplifies the influences of pulse frequency and average target power on ion intensities and energies; At a pulse frequency of 300 Hz (Figs. 6(a) and 6(c)), the IEDFs show reduced intensities as a result of the comparatively low pulse frequency, which leads to less sputtered material and correspondingly to a lower amount of ions. At the same time, a pronounced high energy tail is observed for Si+ IEDFs, which increases in average energy by approximately 4 eV as the average power, and thus the energy per pulse is raised to 10 Ws (cf. Fig. 6(c)). The inert gas ions, on the other hand, show an increased abundance, but only slightly increased energies as a result of the higher average target power. At a pulse frequency of 1000 Hz and an average target power of 1000 W (cf. Fig. 6(b)), IEDFs of all prevalent ions—although comparatively high in intensity—show mainly low energies. These low energies are attributed to the low initial energy of sputtered ions as a result of the low average target power.56 This is also demonstrated in Fig. 6(d). The increase in the average target power to 3000 W causes mainly an increase of the Si+ energy and additionally an elevated abundance of all ionized plasma species. The comparison of Figs. 6(a) and 6(d) illustrates the comparable ion energies for discharges with equal EpPs. Here, the total ion flux increases for the investigated process window almost linearly with increasing pulse frequencies and average target power.
However, an increased growth rate per pulse as presented in Fig. 5(b) results—similar to increased EpPs—also in increased compressive stresses. Although both parameters are ultimately connected, it can be argued that the increase of the compressive stress is not only caused by the ion impact but also by elevated growth rates per pulse. Enhanced growth rates per pulse reduce the ad-atom mobility at the growing coating surface and can in turn contribute to increased compressive stresses.
The effect of an increased negative bias voltage on the residual stress of coatings grown at 200 mPa, 400 mPa, and 600 mPa is shown in Fig. 7. As expected, the compressive stresses rise as the negative bias voltage, Vb, is increased. However, depending on the applied process pressure, Vb influences the compressive stresses in the coatings to different extents; for coatings deposited at 200 mPa, the comparatively high compressive stress of −1.64 GPa for coatings deposited at − 60 V does not increase significantly and is evaluated to −1.90 GPa as the bias voltage is set to − 200 V. The highest influence of Vb is observed for coatings deposited at 400 mPa. Here, the compressive stress increases from −0.74 GPa to −1.85 GPa for bias voltages of −60 V and −200 V, respectively. The compressive stress for coatings deposited at a process pressure of 600 mPa increase from −0.49 GPa to −0.90 GPa in the range of investigated bias voltages. Increasing compressive stresses with increasing negative bias voltages are consistently reported in literature 16,57,58 and explained with the forward sputtering model sputtering.13–17 The extent to which the compressive stresses are influenced by the bias voltage as different process pressures are applied follows essentially the same model. Here, additionally, the corresponding collision mean free paths (cf. Table I) need to be considered as well. At a comparatively high pressure of 600 mPa, the average charged particle will undergo at least one collision with the chosen substrate-to-target distance of 60 mm, while the collision mean free path at 400 mPa does not facilitate the thermalization of ions to a great extent, thus ions can gain energy corresponding to the set bias voltage. At a deposition pressure of 200 mPa, the ion energies are initially high (cf. Fig. 3(a)) and the additional gain in particle energy due to the increased negative substrate bias does not appear to contribute substantially to the compressive stresses. Here, the onset of the momentum transfer regime, where the compressive stress saturates,58–60 is most likely observed.
Although comparatively high compressive stress values are extracted for coatings deposited using increasing pulse energies or bias voltages, nanoindentation tests showed only a weak and insignificant trend towards higher coating hardnesses and elastic moduli. Here, the hardness of the coatings ranged between 23 and 27 GPa, and the elastic moduli were evaluated to range between 240 and 265 GPa. Keeping the increased hardness values at a reduced process pressure of 400 mPa and with increasing substrate temperatures in mind, the results indicate a dependence of the coating hardness on the origin of the coating stress.
C. Implication of deposition parameters on coating adhesion
Figs. 8(a)–8(d) shows Rockwell indents of SiNx coatings deposited on CoCrMo discs using 1-fold substrate rotation, a Cr seed layer and (a) no additional heating, a process pressure of 600 mPa, an EpP of 2.3 Ws, (b) no additional heating, 600 mPa and 4.5 Ws, (c) a substrate temperature of 340 °C, 600 mPa, 4.5 Ws, and (d) no additional heating, 200 mPa, 4.5 Ws. In order to compare the adhesion of coatings with similar N-contents, the coatings shown in (a), (b), and (c) were grown using a of 0.28, while for the deposition of the coating presented in (d), a of 0.4 was used. As discussed above, the different were necessary as the hysteresis shifts with the process pressure. In Fig. 8, the classification of the adhesion according to Ref. 36, the average roughness (Ra), thickness (THK), and the residual stress (σ) obtained from measurements of the same coatings deposited on Si(001) substrates are indicated.
The Rockwell indent presented in Fig. 8(a) shows no delamination and a very limited network of cracks around the indent (HF1), while an extended network of cracks and an increased Ra are observed in Fig. 8(b). The result exemplifies the influence of a slightly increased EpP from 2.3 to 4.5 Ws, as both coatings were grown at the same Ts and pd. Figs. 8(c) and 8(d) show the effects of increased coating stresses due to substrate heating and a reduced process pressure of 200 mPa, respectively. Here, delamination around the indent and extended cracks are visible. Therefore, the adhesion of coatings shown in (c) and (d) is classified as HF 3. Since the residual coating stress was measured on Si substrates, we infer that the stress induced by substrate heating (cf. Fig. 8(c)) is amplified for SiNx coatings deposited on CoCrMo due to its comparatively high CTE of 13.6 × 10−6 K−1. The Rockwell C test results show that an excellent SiNx adhesion is obtained for coatings with low residual stresses. Here, the process pressure and the substrate temperature possess a higher influence on the coating adhesion than the pulse energy.
IV. CONCLUSIONS
The study investigated the most important deposition parameters to tailor the residual stress of amorphous SiNx coatings deposited by rHiPIMS. The main mechanisms that account for the generation of intrinsic compressive coating stress are
Forward sputtering due to energetic particle bombardment, which, in turn, is mainly controlled by the pulse energy, bias voltage settings, and the process pressure.
Reduced ad-atom mobilities that are apparent at low substrate temperatures and increased deposition rates per pulse.
These mechanisms and additional stresses created by the differences of the thermal expansion coefficients of the substrate and the coating are identified to contribute to increased compressive residual coating stresses.
The rHiPIMS key parameters to reduce the compressive stress from 2.1 GPa to 0.2 GPa in SiNx coatings and to facilitate excellent coating adhesion are the deposition pressure, substrate temperature, pulse energy, and bias voltage. Here, a comparatively high deposition pressure of 600 mPa, substrate temperatures below 200 °C, as well as low pulse energies of <2.5 Ws, and moderate negative bias voltages of up to 100 V were found favorable. Those parameters benefit the coating adhesion also under industrial deposition conditions.
V. SUPPLEMENTARY MATERIAL
See supplementary material for XPS Si2p spectra obtained from coatings deposited using a substrate temperature of 145 °C, a of 0.28, an average target power of 1200 W, and a pulse frequency of 400 Hz at a deposition pressure of (a) 200 mPa, (b) 400 mPa, and (c) 600 mPa.
ACKNOWLEDGMENTS
The research leading to these results has received funding from the European Union's Seventh Framework Program (FP7/2007-2013) under the LifeLongJoints Project, Grant Agreement No. GA-310477. S.Sc. acknowledges the support by the Carl Trygger Foundation for Scientific Research (Grant No. CTS 14:431). L.H. and H.H. acknowledges the Swedish Government Strategic Research Area in Materials Science on Functional Materials at Linköping University (Faculty Grant No. SFO-Mat-LiU No. 2009-00971) for financial support.
References
International Organization for Standardization, “Fine ceramics (advanced ceramics, advanced technical ceramics)—Rockwell indentation test for evaluation of adhesion of ceramic coatings,” ISO 26443:2008.