Hybrid high-power impulse and dc magnetron co-sputtering (HiPIMS/DCMS) with substrate bias synchronized to the high mass metal-ion fluxes was previously proposed as a solution to reduce energy consumption during physical vapor deposition processing and enable coatings on temperature-sensitive substrates. In this approach, no substrate heating is used (substrate temperature is lower than 150 oC) and the thermally activated adatom mobility, necessary to grow dense films, is substituted by overlapping collision cascades induced by heavy ion bombardment and consisting predominantly of low-energy recoils. Here, we present direct evidence for the crucial role of W+ ion irradiation in the densification of Ti0.31Al0.60W0.09N films grown by the hybrid W-HiPIMS/TiAl-DCMS co-sputtering. The peak target current density Jmax on the W target is varied from 0.06 to 0.78 A/cm2 resulting in more than fivefold increase in the number of W+ ions per deposited metal atom, η = W+/(W + Al + Ti) determined by time-resolved ion mass spectrometry analyses performed at the substrate plane under conditions identical to those during film growth. The DCMS is adjusted appropriately to maintain the W content in the films constant at Ti0.31Al0.60W0.09N. The degree of porosity, assessed qualitatively from cross-sectional SEM images and quantitatively from oxygen concentration profiles as well as nanoindentation hardness, is a strong function of . Layers grown with low η values are porous and soft, while those deposited under conditions of high η are dense and hard. Nanoindentation hardness of Ti0.31Al0.60W0.09N films with the highest density is ∼33 GPa, which is very similar to values reported for layers deposited at much higher temperatures (420–500 oC) by conventional metal-ion-based techniques. These results prove that the hybrid HiPIMS/DCMS co-sputtering with bias pulses synchronized to high mass metal ion irradiation can be successfully used to replace conventional solutions. The large energy losses associated with heating of the entire vacuum chamber are avoided, by focusing the energy input to where it is in fact needed, i.e., the workpiece to be coated.
Growth of thin films by conventional physical vapor deposition (PVD) techniques requires extensive substrate heating to ensure sufficient adatom mobility, necessary for reaching near-bulk densities. Layers grown with no external heating are typically under dense, leading to poor mechanical and optical properties, as well as high electrical resistivity.1 In addition, a typical deposition process is preceded by several hours of preheating the vacuum chamber to reach substrate temperatures Ts in the range of 400–600 oC, with the intention to achieve the sufficiently low background pressure and, hence, minimize the contamination levels in deposited films. As a result of the above, the total electricity consumption due to heating often exceeds that due to the operation of sputtering sources. Another disadvantage of this classical approach is that temperature-sensitive substrates such as polymers, lightweight Al-, and/or Mg-based alloys are excluded from applications.
It is important to note that the temperature necessary to provide sufficiently high adatom mobility during growth is much higher than what is required to ensure clean deposition conditions. To effectively reduce the water vapor content in the residual atmosphere, it is enough to maintain vacuum chamber walls and components above 100 oC. Thus, the overall energy consumption during PVD processing could be significantly reduced if one could find alternative means of providing high adatom mobility that would not involve extensive substrate heating.
One attractive solution is offered by a hybrid high-power impulse and dc magnetron co-sputtering (HiPIMS/DCMS) involving a high mass (m > 180 amu) HiPIMS target and metal-ion-synchronized bias pulses.2,3 Due to the effective low-energy recoil generation caused by large mass mismatch between the incident ion and film constituents,4 the high-mass target driven by HiPIMS provides energetic metal-ion fluxes that densify species deposited between HiPIMS pulses, while the metal targets operate in a DCMS mode and guarantee a high deposition rate.5 Based on time-resolved ion mass spectrometry analyses performed at the substrate position,6,7 the synchronized substrate bias voltage Vs acts selectively on the metal-ion-rich portions of the pulsed HiPIMS fluxes, so that metal-ion energy and momentum can be controlled without significantly affecting Ts as the duty cycle does not exceed 2%. With a high enough metal/gas atom mass ratio mMe/mg, gas rarefaction in front of the HiPIMS target is prominent allowing for more flexibility in the choice of substrate bias length and offset.8 Densification of transition metal (TM) nitride films induced by a high-mass Ta+ (mTa = 181.0 amu) ion irradiation have been demonstrated for Ti1−xTaxN (Ref. 2) and Ti1−x−yAlxTayN (Ref. 3) films. In each case, the substrate temperature was lower than 150 oC.
More recently, we initiated systematic studies addressing the role of metal-ion-induced densification in a metastable Ti-Al-N-based system, that is highly relevant to industrial applications9 and presents challenges for phase stability control.10–12 The selection of metal-ions was guided by the time-resolved ion mass spectrometry analyses, which revealed that gas rarefaction effects during HiPIMS sputtering in Ar/N2 mixtures are more severe for group VIB TM targets (Cr, Mo, and Ta).8,13 All experiments were performed with Ts < 130 oC. While studying the effects of the incident ion mass, we concluded that (Ti1−yAly)1−xCrxN layers grown by Cr-HiPIMS/TiAl-DCMS (irradiation by light Cr+ ions, mCr = 52.0 amu) were porous irrespective of the Cr content. In contrast, (Ti1−yAly)1−xWxN films deposited by W-HiPIMS/TiAl-DCMS (irradiation by heavy W+ ions, mW = 183.8 amu) were dense even with the lowest W concentration, x = 0.09. These films show no evidence of hexagonal wurtzite AlN precipitation and exhibit state-of the-art mechanical properties typical of Ti0.50Al0.50N grown at 500 oC.14 By independently controlling the W+ concentration and incident energy, we identified the process parameter window that allows to grow dense (Ti1−yAly)1−xWxN films with low residual stress.15
Further study of the W+ densification effects as a function of increasing Al content in (Ti1−yAly)1−xWxN films covering the entire range up to the practical solubility limits (y ∼ 0.67) revealed that the key parameter that governs the growth under the conditions of low substrate temperature and effective low-energy recoil production is the average momentum transfer supplied by W+ irradiation per deposited metal atom .16 The latter quantity is proportional to the product of the number of W+ ions per deposited metal atom (in which α is the ionized fraction of the W flux) and the square root of the applied substrate bias . We demonstrated that under optimized conditions, the energy consumption during heating/coating phase of the process can be lowered by ∼64% with respect to the conventional DCMS-based processing without sacrificing coating quality.
In the previous experiments, was varied by controlling x and Vs. An independent way to vary the average momentum transfer supplied by W+ irradiation per deposited metal atom is to alter the ionized fraction α, while keeping both x and Vs constant (which is equivalent to varying ). This can be realized by varying the peak target current density Jmax, thereby changing the plasma density in front of the W target and, hence, the probability for the ionization of the sputtered W flux.
The present study provides direct evidence that the densification effects in TiAlN-based films scale up with the number of W+ ions per deposited metal atom, η = W+/(W + Al + Ti). While using the hybrid W-HiPIMS/TiAl-DCMS film growth scenario with metal-ion-synchronized substrate bias and no external heating, we varied the latter parameter by changing the peak target current density Jmax on the W target operated in the HiPIMS mode in the wide range from 0.06 A/cm2 (corresponding to DCMS-like conditions) to 0.78 A/cm2, while maintaining the substrate bias amplitude Vs constant. Film composition is kept constant at Ti0.31Al0.60W0.09N by controlling average power on HiPIMS and DCMS sources. The intensity of W+ fluxes is directly assessed by time-resolved ion mass spectrometry analyses performed at the substrate plane under conditions identical to those during film growth. We show that for the identical biasing conditions layers grown at lower η values (thus predominantly with W neutral flux) are porous and soft, while those deposited with the highest η (corresponding to high W+ content in the incident ion flux) are dense and hard. The strong dependence of the film porosity on W+/(W + Al + Ti) ratio serves as a direct proof that W+ ions are crucial for providing adatom mobility that is sufficient to densify TiAlN-based films, in the absence of thermally activated surface mobility.
An industrial magnetron sputtering system from CemeCon AG (CC800/9) is employed to grow (Ti1−yAly)1−xWxN films in a hybrid co-sputtering configuration such that the W target (8.8 × 50 cm2) operates in the HiPIMS mode, while two TiAl targets are used as DC magnetrons (W-HiPIMS/TiAl-DCMS).14–16 The W target is mounted symmetrically in the middle between two TiAl targets and its surface is parallel to the substrate holder plate. The target-to-substrate distance is 18 cm and the angle between substrates and both DCMS targets is 21°. A calibrated thermocouple positioned near the substrates is used to monitor the temperature during the entire process. The schematic diagram of the deposition system can be found in Ref. 14.
All films are grown onto Si (001) substrates (1.5 × 1.5 cm2) cleaned in acetone and isopropanol. Gentle substrate heating aiming to degas the chamber before film growth is accomplished by resistance heaters mounted symmetrically on the front and back sides of the vacuum chamber. Each heater is powered to 2 kW during the 1h heating phase, which results in that the highest temperature reached during the heating is 110 °C, i.e., sufficient to efficiently desorb water vapor from chamber walls. After that the heaters are turned off and the system is allowed to cool down to 65 °C for another hour reaching the base pressure of 0.3 mPa (2.25 × 10−6 Torr). During the film growth phase, the substrate temperature increases again due to the plasma heating and reaches ∼130–150 °C in the end of the deposition sequence (cf. Table I). The total pressure Ptot during deposition is 0.4 Pa (3 mTorr) with a constant N2 and Ar flow at 148 and 360 SCCM, respectively.
|Films .||PHIP (kW) .||PDC (kW) .||Jmax (A/cm2) .||Ep (J/pulse) .||Ds (nm/min) .||tD (min) .||Tmax (°C) .|
|DC Ti0.4Al0.6N reference||10||0.05||124.2||12||158|
|Films .||PHIP (kW) .||PDC (kW) .||Jmax (A/cm2) .||Ep (J/pulse) .||Ds (nm/min) .||tD (min) .||Tmax (°C) .|
|DC Ti0.4Al0.6N reference||10||0.05||124.2||12||158|
The HiPIMS pulse length is constant at 100 μs, and the HiPIMS frequency is 200 Hz. A 200 μs long substrate bias pulse synchronized to the HiPIMS cathode at the value of Vs = 240 V is applied. The bias pulse offset is 30 μs (with respect to the cathode pulse) based on the outcome of the ion mass spectrometry analyses performed at the substrate plane.17 By varying the average HiPIMS power applied to the W target, PHIP, while keeping pulse length and frequency constant, the peak target current density, Jmax varies in a wide range from 0.06 to 0.78 A/cm2. The DC power applied to both TiAl targets, PDC, is adjusted for each Jmax value such that all films maintain the same composition—Ti0.31Al0.60W0.09N. To keep the film thickness in the range from 1.4 to 1.9 μm, the deposition time is varied from 10 to 55 min. Detailed settings are summarized in Table I.
The Ti0.4Al0.6N film is also grown by DCMS at the floating potential of Vf = −10 V, with the DC power of 10 kW, and the average current on the TiAl targets of 20 A (corresponding to the current density of 0.05 A/cm2). Such a reference layer, grown in the absence of metal ion irradiation, while maintaining the DCMS conditions identical to those during the DCMS phase of the hybrid HiPIMS/DCMS growth, is necessary to evaluate the effects of W+ irradiation.
Time-resolved mass- and energy-spectroscopy analyses of ion fluxes during W-HiPIMS/TiAl-DCMS hybrid sputtering are conducted with the Hiden Analytical EQP9 instrument equipped with a quadrupole mass analyzer. The 50 μm diameter orifice of the spectrometer is grounded and aligned along the W target surface normal. The target-to-orifice distance is the same as the target-substrate separation during film growth to ensure that the intensities and time evolutions of ion fluxes are representative of the situation during deposition. No heating is applied so that the temperature during analyses is similar to that during film growth. The measurement conditions (gas composition, total pressure, power, HiPIMS frequency, and pulse length) are the same as used during deposition. Time-resolved ion energy distribution functions (IEDFs) of W+, are recorded for each combination of HiPIMS peak target current density and DCMS power in the W-HiPIMS/TiAl-DCMS hybrid mode. The scanned range is from 1 to 60 eV, with a 1 eV step. In order to obtain the plasma composition at the front end of the mass spectrometer rather than at the detector, all data were compensated for the time-of-flight (TOF) of ions through the instrument using the same approach as described in Ref. 18. In order to eliminate the risk of detector saturation, the 184W isotope with lower abundance is selected. The gate time window is set at 10 μs, and the data are recorded with a 10 μs steps starting from the ignition of the HiPIMS pulse and up to the delay time of 230 μs, to cover the entire period Vs is applied during the film growth. The dwell time of the detector is set at 1 ms, implying that each plot is an average of 100 HiPIMS pulses. More details concerning the IEDF measurements can be found in Ref. 19.
Time-of-flight elastic recoil detection analysis (ToF-ERDA) is used to measure the elemental compositions of all Ti0.31Al0.60W0.09N films. For that, the 36 MeV127I8+ probe beam is incident at 67.5° with respect to the surface normal, and the recoils are detected at 45°.20
An Axis Ultra DLD instrument (Kratos Analytical, UK) is used for x-ray photoelectron spectroscopy (XPS) analyses. The excitation source is a monochromatized Al Kα radiation (hν = 1486.70 eV) and measurements are conducted at the base pressure better than 1.1 × 10−9 Torr (1.5 × 10−7 Pa). Concentration depth profiles are obtained by etching with 0.5 keV Ar+ ions incident at an angle of 70° with respect to the surface normal.21–23 Information is obtained from an area of 0.3 × 0.7 mm2 centered in the middle of the ion-etched crater (with the size of 3 × 3 mm2). Quantification analysis with sensitivity factors supplied by instrument manufacturer is performed in the CasaXPS software package.24
The thickness as well as the plan-view and cross-sectional morphology of the films are determined by scanning electron microscopy (SEM) with a Zeiss Sigma 300 instrument.
A Philips X’Pert MRD system with a point-focus Cu Kα radiation source is used for θ−2θ x-ray diffraction (XRD). For more accurate phase determination,25 analyses are performed at the sample tilt angle ψ (the angle between the scattering plane defined by the incoming and outgoing x-ray beams and the sample normal) of 0°, 18.4°, 26.6°, 33.2°, 39.2°, 45.0°, 50.8°, 56.8°, 63.4°, and 71.6°. The residual stresses are evaluated by the substrate curvature method in a Panalytical Empyrean XRD system. The curvature of the samples is obtained from rocking curve measurements of Si (400) reflections. Based on Stoney’s equation,26 the stress σ of the film is then calculated as27
where hS = 525 μm is the thickness of Si substrate and hf is the film thickness from cross-sectional SEM analyses. The biaxial modulus of the substrate is 181 GPa.28 R0 and R correspond to the substrate curvature before and after film deposition, respectively. Since R >> R0, R0 is neglected. The final result is the average value from measurements conducted along several directions, so that the error caused by the uneven substrate bending is taken into account.
The nanoindentation tests of all films were carried out on a TriboIndenter of Hysitron Ti 950. Trapezoid course of the load was used for the tests and the device was operated in force feedback-control mode. Maximum load of 18 mN was used with a dwell time of 5 s. The load was selected so that the indentation depth of the indenter did not exceed 10% of the samples thickness, thus reducing a potential elastic interaction from the substrate. 25 indentations were performed during nanoindentation hardness tests within a square grid, built of 5 impressions per row and 5 impressions per column, while minimum distance between them was fixed to 20 μm. Standard deviation errors were included to the obtained results as well, in accordance with the Oliver and Pharr rule.29
III. RESULTS AND DISCUSSION
Figure 1 shows (a) target current density JT(t) and (b) target voltage VT(t) waveforms recorded during W-HiPIMS/TiAl-DCMS depositions conducted with different peak target current density values Jmax from 0.06 to 0.78 A/cm2. In the limit of highest current, the discharge becomes power supply limited resulting in the characteristic shape of the current pulse: following the steep increase in the initial phase (0–10 μs), JT(t) reaches the maximum value at t = 40 μs, and after that decays to ca. 0.07 A/cm2 toward the end of the pulse. The target voltage waveform VT(t) reveals a significant drop from the initial value of −940 V to −910 V at t = 0 μs to −544 V at t = 100 μs. This indicates that a significant portion of the charge stored in the capacitor bank has been withdrawn during the high-energy pulse (EP = 14 J) so that the constant voltage cannot be maintained throughout the entire pulse duration.
The above effects are to some extent visible also in the waveforms recorded for the Jmax = 0.35 A/cm2 case. The magnitude of variation in both JT(t) and VT(t) is, however, lower due to lower pulse energy of 6 J. JT(t) reaches the maximum value at t = 30 μs, and after that decays to ca. 0.11 A/cm2 at t = 100 μs, while VT(t) decreases from −750 to −730 V in the beginning of the pulse to −100 V at t = 100 μs.
In the case of discharge operated with the two lowest peak target current values of 0.11 and 0.06 A/cm2, both JT(t) and VT(t) exhibit little variation during the pulse (apart from the initial 20 μs phase necessary for the target current to reach the stable value). This is because the energy per pulse is relatively low at 1.5 and 0.75 J, respectively, thus not setting particular demand on the capacitor bank.
Figure 2 consists of four sets of 3D plots showing 184W+ IEDFs recorded during hybrid W-HiPIMS/TiAl-DCMS sputtering with Jmax varying from 0.06 to 0.78 A/cm2. Both DC magnetrons operate at the power listed in Table I. Data are recorded at the substrate plane starting with t = 0 μs corresponding to the ignition of 100 μs HiPIMS pulses and up to t = 230 μs, i.e., the end of the substrate bias pulse.
With higher peak target current density, Jmax ≥ 0.35 A/cm2 [cf. Figs. 2(a) and 2(b)], the first W+ ions are detected at t = 40 μs (low energy peak at t = 0 μs is due to thermalized ions from the preceding pulse). These are the fastest species that are sputter-ejected in the beginning of the pulse and travel to the orifice with no collisions. The IEDFs for 40 ≤ t ≲ 120 μs exhibit broad Sigmund–Thompson-type sputtered-species energy distributions with the average energy of 20–30 eV and high energy tails extending up and beyond 60 eV.30,31 The maximum intensities are of the order of 2−5 × 105 cps. After the HiPIMS pulse is turned off (t = 100 μs), the IEDFs maxima gradually shift toward 10–12 eV, corresponding to the plasma potential during hybrid co-sputtering, while maximum intensities reach 1−2 × 106 cps. This indicates partial thermalization of the sputtered flux and is in contrast to previous results where no low-energy peak was observed,8,17 which was assigned to the severe gas rarefaction taking place in front of the sputtering target. However, in contrast to results reported in those papers, the present analyses are conducted in a hybrid configuration, i.e., with two additional cathodes operating in the DCMS mode. An additional difference is that all three magnetrons are physically at the same positions as during film growth, hence the original magnetic field configuration is preserved. Taking that into account, the observed W thermalization seems to stem primary from the collisions with high Al and Ti fluxes from DC sources [note that the W/(W + Al + Ti) fraction is relatively low]. In the context of Ti0.31Al0.60W0.09N film growth conducted with the Vs = 240 V synchronous substrate bias, IEDFs shown in Figs. 2(a) and 2(b) reveal that ∼73% of W+ ions arrive at the film surface with the energy of 250 ± 10 eV, while the remaining ∼27% has an energy higher than 260 eV.
The appearance of time-resolved W+ IEDF’s changing dramatically for Jmax < 0.35 A/cm2 [cf. Figs. 2(c) and 2(d)]. First, the time delay between the start of the HiPIMS pulse and the time first W+ ions are detected increases to 70–80 μs. This is predominantly the consequence of the fact that high-energy tails are absent: only 3% of all W+ ions have energies higher than 30 eV. Even during the most intense phase of the discharge the average ion energy is only ∼12 eV, i.e., similar to the plasma potential. The maximum intensities of W+ ion fluxes are at 2−6 × 104 cps, i.e., about an order of magnitude lower as compared to those recorded during sputtering with Jmax ≥ 0.35 A/cm2. These pronounced changes are explained by the condition that the plasma density in front of the W target decreases with decreasing peak target current density, thus lowering the probability for the electron-impact ionization of the sputtered W flux.32,33 As a consequence of that, the ionized fraction of the W flux to the substrate decreases with decreasing Jmax.
The 184W+ IEDFs shown in Fig. 2 can be integrated for each delay time to obtain the relative measure of W+ ion flux intensity arriving at the growing film surface within a 10 μs time window at a given time during and after the W-HiPIMS pulse. Such plots are shown in Fig. 3 for all four values of peak target current density. The length of HiPIMS and substrate bias pulses is also indicated to facilitate results interpretation. are, in general, much broader than the width of HiPIMS pulses, revealing that the contribution of W+ ions that arrive at the substrate after the cathode pulse is turned off (t > 100 μs) is dominant. This is due to two reasons: (1) even for those ions that do not undergo collisions on the way to the substrate, the time of flight is on the order of 60–80 μs for the 18 cm separation,8 (2) majority of W+ ions loses energy on the way to the substrate, which further increases their transit time.
Highly relevant for the scope of this paper is the observation that the intensity rapidly decreases with decreasing peak target current density such that the maximum W+ ion count is ca. 25 times lower for the case of Jmax = 0.06 A/cm2 with respect to the value obtained for Jmax = 0.78 A/cm2. Even more interesting is the comparison of the total W+ ion count [corresponding to the area under the curves] for 30 ≤ t ≤ 230 μs, i.e., during the time the substrate is biased at −240 V. decreases from 2.02 × 109 with Jmax = 0.78 A/cm2 to 5.92 × 107 with Jmax = 0.06 A/cm2. This means a factor of 34 times decrease in the number of W+ ions that are accelerated in the electric field of the substrate to produce low-energy recoils that provide adatom mobility in the absence of substrate heating. Since the film growth rate decreases by a factor of ∼6 (cf. Table I) and the film composition is maintained constant for all peak target current values, we can conclude in relative terms that the number of W+ ions per deposited metal atom, η = W+/(W + Al + Ti), is 5.7 × higher during Ti0.31Al0.60W0.09N growth with Jmax = 0.78 A/cm2 than with Jmax = 0.06 A/cm2. This number, which in fact reflects the relative change in the ionization degree of the W flux incident on the film surface, should be in fact considered the lower limit as the actual drop in the growth rate upon increasing Jmax is also caused by an increasing film density [see Figs. 6–8 and discussion below].
An independent evidence for the substantial change in the ionization degree of the sputtered W flux with varying Jmax is provided by the PHIP/PDC average power ratio (cf. Table I). It is evident that in order to maintain the same film composition, PHIP/PDC has to increase from 0.05 with Jmax = 0.06 A/cm2 to 0.29 with Jmax = 0.78 A/cm2. This is because an increasing fraction of the W flux is ionized and attracted back to the target leading to the well-known drop in the HiPIMS deposition rate.5,34–36
Thus, the time-resolved mass spectrometry analyses conducted at the substrate plane in the hybrid W-HiPIMS/TiAl-DCMS configuration reveal two essential differences between Ti0.31Al0.60W0.09N film growth with high and low peak target current density: (1) η is significantly higher for Jmax ≥ 0.35 A/cm2, and (2) the W+ ions with energies higher than 260 eV are essentially absent during growth with Jmax < 0.35 A/cm2.
The ToF-ERDA-derived W, Al, Ti, and N atomic compositions are plotted in Fig. 4 as a function of the peak target current density Jmax. Results for the DCMS Ti0.4Al0.6N reference layer are also added for comparison. For all four Jmax settings, the atomic compositions are fairly constant with [W] at ∼4.3 at. %, [Al] at ∼27 ± 1 at. %, [Ti] at ∼14 ± 1 at. %, and [N] at ∼50 ± 1 at. %, so that all films can be denoted Ti0.31Al0.60W0.09N. The Ar content obtained from ToF-ERDA is 0.3 ± 0.1 at. % for all Ti0.31Al0.60W0.09N layers as well as the DCMS Ti0.4Al0.6N reference layer. The substantial oxygen content in the films is indicative of their porosity14–16 and is therefore discussed separately below.
The θ–2θ XRD scans recorded as a function of the sample tilt angle ψ for Ti0.31Al0.60W0.09N films grown with different values of the peak target current density are shown in Fig. 5. All films, including the DCMS Ti0.4Al0.6N reference, are single cubic phase with strong 111 orientation. With low peak current intensity (Jmax < 0.35 A/cm2), they have complete 111 orientation; while with Jmax ≥ 0.35 A/cm2, a low intensity 002 peak is also detected. The 111 peak position in all films is at 37.2 ± 0.2°. The peak shifts as a function of tilt angle ψ are less than 0.3°, indicating that the stress levels in all films are low [see Fig. 8 and related discussion below]. The full-width-at-half-maximum (FWHM) of the 111 reflection from the DCMS Ti0.4Al0.6N reference sample is 0.79°. For Ti0.31Al0.60W0.09N films, the FWHM at ψ = 0° decreases from 0.89° with Jmax = 0.78 A/cm2 to 0.72° with Jmax = 0.06 A/cm2, revealing that growth with higher peak target current density, hence under more intense W+ irradiation [cf. Figs. 2 and 3], results in grain refinement.
The cross-sectional SEM (XSEM) and plan view SEM images acquired from all Ti0.31Al0.60W0.09N films grown with the peak target current density Jmax varying in the range of 0.06–0.78 A/cm2 are shown in Fig. 6. DCMS Ti0.4Al0.6N reference film (Jmax = 0.05 A/cm2) is also included for comparison. Large degree of both the intercolumnar and intracolumnar porosity is visible for films grown with Jmax ≤ 0.11 A/cm2. This is a consequence of low adatom mobility in the absence of substrate heating. Despite the fact that the high-amplitude bias pulses are applied during deposition, the mechanism of metal-ion-induced mobility through effective generation of low energy recoils is very limited as the number of W+ ions per deposited metal atom is relatively low. Films grown under these conditions are also characterized by conical column tops resulting in larger surface roughness. The column diameter is in the range of 80–200 nm.
The pronounced change in the film microstructure takes place for layers grown with Jmax = 0.35 A/cm2. The intercolumnar voids are much smaller, the column width is significantly larger, and the surface roughness is decreased. This effect stems from the rapid increase in the intensity of the W+ flux incident at the growing film surface [cf. Figs. 2 and 3], resulting in higher η. Ti0.31Al0.60W0.09N layers grown with the highest Jmax value of 0.78 A/cm2, thus under the most intense W+ bombardment have much smaller columns (<100 nm) that are closely packed resulting in that the intercolumnar porosity is essentially eliminated. This densification cannot be ascribed to higher growth rate as this typically leads to lower density, as proven by the reference Ti0.4Al0.6N film (cf. Table I), which is clearly porous. Likewise, the effects of more intense plasma heating due to higher powers used in the Jmax = 0.78 A/cm2 case are compensated by shorter deposition time such that the substrate temperature in the end of the coating phase is essentially the same as during growth with low Jmax values (see Table I).
SEM results are fully confirmed by the XPS depth profiles depicted in Fig. 7, which are used to assess the oxygen content in the top-most layer extending from 2 to 20 nm from the surface (thus excluding the first 2 nm dominated by native surface oxide). We established in our earlier papers that the O concentration in the films following the prolonged air exposure (several days or so) correlates closely to mechanical properties and provides a good measure of film porosity.14–16 Underdense films exhibit very high oxygen levels (bulk content of 2–5 at. % according to ToF-ERDA) due to inward O diffusion along grain boundaries, while layers with high density are characterized by significantly lower O content (<0.5 at. %), which is determined by the O pickup during film growth under high vacuum conditions. XPS depth profiles of the type shown in Fig. 7 provide good qualitative comparison between the samples, they are, however, overestimating the absolute O content due to inward sputtering and redeposition during Ar+ etch.
Clearly, the data in Fig. 7 show that the oxygen content in Ti0.31Al0.60W0.09N films strongly depends on the peak target current density. Films grown with higher Jmax values, thus under more intense W+ irradiation (higher η), contain less oxygen. This is an independent piece of evidence that film density increases with increasing Jmax. The highest O levels are measured for the DCMS Ti0.4Al0.6N reference film (Jmax = 0.05 A/cm2), which also shows the slowest decay of O content with depth. Noticeable decrease in the overall O content is applied to Ti0.31Al0.60W0.09N films grown with Jmax ≤ 0.11 A/cm2. Both, Jmax = 0.06 A/cm2 and Jmax = 0.11 A/cm2 layers, show essentially identical O profiles, which agrees very well with their open structure best visible in plan view images in Fig. 6. A pronounced drop in O concentration takes place for the Jmax = 0.35 A/cm2 film. The lowest O levels, however, are measured for the sample grown with the highest Jmax of 0.78 A/cm2, again in excellent agreement with the SEM studies.
The above conclusions based on XPS depth profiles are fully confirmed by bulk O concentrations cO obtained from ToF-ERDA. As depicted in Fig. 8, cO is highest for the DCMS Ti0.4Al0.6N reference film, at 4.6 ± 0.1 at. %. Ti0.31Al0.60W0.09N films grown with Jmax = 0.06 and 0.11 posses 3.4 ± 0.1 and 2.4 ± 0.1 at. % oxygen, respectively. Increasing the target peak current to 0.35 A/cm2 results in denser films with only 0.7 ± 0.1 at. % oxygen. The lowest cO of only 0.4 ± 0.1 at. % is detected in the layer grown with Jmax = 0.78 A/cm2. The latter value is determined by the residual water vapor pressure during film growth (ca. 0.3 mPa).
Figure 8 shows the nanoindentation hardness H, oxygen content cO obtained from ToF-ERDA, as well as residual stress σ measured by XRD curvature method for all Ti0.31Al0.60W0.09N films plotted as a function of the peak target current density Jmax. Corresponding results for the DCMS Ti0.4Al0.6N reference film grown at high substrate temperature and DC bias are also included. As discussed above, the trend in cO (Jmax) is very clear: the higher the peak target current the lower the oxygen content in the films. This is explained by the fact that the number of W+ ions per deposited metal atom, η, increases with increasing Jmax, resulting in higher density of collision cascades, hence, more efficient production of low-energy recoils, which provides mobility necessary to grow dense layers and, hence, substitutes for the lack of thermally-activated adatom mobility. The H(Jmax) plot fully confirms this interpretation. H is lowest at only 8.8 ± 0.8 GPa for the DCMS reference film grown with Jmax = 0.05 A/cm2. Somewhat harder Ti0.31Al0.60W0.09N films with H∼ 15.0 GPa are obtained for the hybrid growth with Jmax ≤ 0.11 A/cm2. A rapid increase in hardness to 25.9 ± 1.4 GPa takes place as Jmax increases to 0.35 A/cm2. Finally, films grown with the highest peak target current density of 0.78 A/cm2 exhibit the highest H values of 32.6 ± 1.0 GPa. Thus, there is a very strong correlation between the H(Jmax) and cO(Jmax) trends, as indicated in our earlier papers.14–16
The hardness values obtained here for Ti0.31Al0.60W0.09N films grown with no substrate heating, at the substrate temperature lower than 150 oC, are, in fact, very similar to those reported for layers grown at 420 oC by HiPIMS with floating substrates37 or at 500 oC by cathodic arc evaporation with DC bias.38 While both techniques also employ highly ionized metal fluxes, the effect of selective biasing on heavy ion component of the ionized flux was not exploited in those experiments.
The residual stress levels σ in all films are low. All layers grown with Jmax ≤ 0.11 A/cm2 are essentially stress-free. Ti0.31Al0.60W0.09N films grown with higher peak target current density are under slight compressive stress state, with σ = −0.24 ± 0.06 and −0.71 ± 0.14 GPa for Jmax = 0.35 and 0.78 A/cm2, respectively.
We have studied the Ti0.31Al0.60W0.09N film growth using the hybrid W-HiPIMS/TiAl-DCMS co-sputtering with metal-ion-synchronized substrate bias and no external heating. The key parameter in experiments was the peak target current density Jmax on the W target that was varied in the wide range from 0.06 A/cm2 (corresponding to DCMS-like condition) to 0.78 A/cm2. That led to large changes in the ionization of the sputtered W flux, and thus, in the number of W+ ions per deposited metal atom, η = W+/(W + Al + Ti). The intention was to investigate the extent of W + -induced densification in Ti0.31Al0.60W0.09N layers grown in the absence of thermally-activated adatom mobility (substrate temperature did not exceed 150 oC).
As evident from time-resolved ion mass spectrometry analyses conducted at the substrate plane during hybrid W-HiPIMS/TiAl-DCMS co-sputtering under conditions identical to those during film growth, large increase in Jmax (from 0.06 to 0.78 A/cm2) implies large increase in the plasma density in front of the sputtering target and, hence, increased probability for the ionization of the sputtered W flux. η is ∼5.7 × higher during Ti0.31Al0.60W0.09N growth with Jmax = 0.78 A/cm2 than it is with Jmax = 0.06 A/cm2. In the former case, ∼73% of W+ ions arrive at the negatively biased film surface with the energy of 250 ± 10 eV, while the remaining ∼27% has an energy higher than 260 eV. During growth with low peak target current values, the high energy tail is essentially absent.
We showed that for the identical biasing conditions (200 μs-long bias pulses with the amplitude of −240 V synchronized to W+-rich portions of HiPIMS pulses) the porosity in Ti0.31Al0.60W0.09N films, assessed qualitatively from XSEM images and quantitatively from the oxygen concentration profiles and nanoindentation hardness, is a strong function of HiPIMS peak target current density Jmax (thus the η ratio). Layers grown at lower Jmax values (low η, i.e., predominantly with W neutral flux) are porous and soft, while those deposited with the highest peak target current density (high η corresponding to the high fraction of W+ ions in the incident ion flux) are dense and hard.
The observed strong dependence of the film porosity on W+/(W + Al + Ti) ratios is a direct proof that W+ ions are crucial for providing adatom mobility that is sufficient to densify TiAlN-based films, in the absence of thermally activated surface diffusion. Nanoindentation hardness of Ti0.31Al0.60W0.09N films grown with Jmax = 0.78 A/cm2 is ∼33 GPa, which is very similar to state-of-the-art values reported for layers deposited at much higher temperatures (420–500 oC) by conventional metal-ion-based techniques such as HiPIMS or cathodic arc evaporation. The additional advantage is that residual stresses in Ti0.31Al0.60W0.09N layers grown by the hybrid method are very low not exceeding −0.7 GPa, thus resulting in layers with unique combination of properties rarely achieved by other techniques.
The above results prove that the hybrid HiPIMS/DCMS co-sputtering with bias pulses synchronized to high mass metal ion irradiation can be successfully used to replace conventional solutions. This novel approach, that aims at significantly reduces energy consumption during PVD growth and expansion of the PVD process envelope to enable coatings on temperature-sensitive substrates, is fully competitive against the state of the art high-temperature processing. The large energy losses associated with heating of the entire vacuum chamber are avoided, by focusing the energy input to where it is needed most, i.e., the workpiece to be coated. As a result of that, the energy consumption of the complete coating process is significantly reduced.
The help of Dr. Babak Bakhit with ToF-ERDA analyses and Dr. Bartosz Wicher with nanoindentation hardness measurements is gratefully acknowledged. The authors also acknowledge financial support of the Swedish Research Council VR Grant No. 2018-03957, the Swedish Energy Agency under Project No. 51201-1, the Åforsk Foundation Grant No. 22-4, the Knut and Alice Wallenberg Foundation Scholar Grant Nos. KAW2016.0358 and KAW2019.0290, the Carl Tryggers Stiftelse contract CTS 20:150, and the Competence Center Functional Nanoscale Materials (FunMat-II) VINNOVA Grant No. 2016-05156. Support from the Swedish Research Council VR-RFI (No. 2017-00646_9) for the Accelerator based ion-technological center and from the Swedish Foundation for Strategic Research (Contract No. RIF14-0053) for the Tandem Accelerator Laboratory in Uppsala University is acknowledged.
Conflict of Interest
The authors have no conflicts to disclose.
Xiao Li: Data curation (lead); Formal analysis (equal); Investigation (lead); Writing – original draft (lead). Ivan Petrov: Conceptualization (equal); Formal analysis (equal); Writing – review & editing (equal). Lars Hultman: Formal analysis (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Grzegorz Greczynski: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal).
The data that support the findings of this study are available from the corresponding author upon reasonable request.