Effect of Al alloying on cavitation erosion behavior of TaSi2 Effect of Al alloying on cavitation erosion behavior of TaSi2 nanocrystalline coatings nanocrystalline coatings

To broaden the scope of non-aerospace applications for titanium-based alloys, both hexagonal C40 binary TaSi 2 and ternary Al alloyed TaSi 2 nanocrystalline coatings were exploited to enhance the cavitation erosion resistance of Ti-6Al-4V alloy in acidic environments. To begin with, the roles of Al addition in influencing the structural stability and mechanical properties of hexagonal C40 Ta(Si 1-x Al x ) 2 compounds were modelled using first-principles calculations. The calculated key parameters, such as Pugh's index (B/G ratio), Poisson's ratio, and Cauchy pressures, indicated that there was athreshold valueforAladdition, belowwhich theincreaseof Alcontentwouldrenderthe Ta(Si 1-x Al x ) 2 compounds more ductile, but above which no obvious change would occur. Subsequently, the TaSi 2 and Ta(Si 0.875 Al 0.125 ) 2 coatings were preparedand theirmicrostructure andphase compositionwere characterizedby X-ray diffraction(XRD),scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Both the two coatings exhibited a uniform thickness of 15μm and a densely packed structure mainly composed of spherically shaped nanocrystallites with an average diameter of about 5nm. Nanoindentation measurements revealed that Al alloying reduced the hardness (H) and elastic modulus (E) values of the TaSi 2 coating. Ultrasonic cavitation erosion tests were carried out by immersing coated and uncoated samples in a 0.5M HCl solution. The cavitation-erosion analysis of the tested samples was in- vestigated by various electrochemical techniques, mass loss weight and SEM observation. The results suggested that both coated samples provided a better protection for Ti-6Al-4V against the cavitation-erosion damage in acidic en- vironments, but the addition of Al further improved the cavitation-erosion resistance of the TaSi 2 coating.


Introduction
Through their unique combination of high strength-to-weight ratio and excellent resistance to corrosion, titanium alloys are increasingly finding new industrial application outside military and commercial aerospace fields, for example in the petrochemical and chemical industries [1]. The low dissolution rate of titanium alloys under most corrosive media is due to the presence of stable, continuous and protective passive films that form on the alloy surface. Despite its high corrosion resistance in various media, the naturally formed passive film lacks sufficient mechanical stability and is readily removed in some harsh environments, especially where the occurrence of corrosion is directly related to mechanical degradation processes (e.g. erosion, cavitation, abrasion, etc.). Therefore, much attention has been paid to improving the resistance of titanium alloys against a combined erosioncorrosion attack.
Cavitation erosion is the most common type of material degradation in overflow components in hydrodynamic systems, for example marine propellers or hydraulic pumps or turbines [2]. The formation of cavities stems from the tensile stress imposed on the liquid as the local pressure drops below the vapor pressure of the fluid produced by flow fluctuation (vibration) or motion of the solid boundary [3]. When these cavities migrate to higher pressure zones, they collapse violently, exerting pressure shock waves and liquid micro-jets on to adjacent solid surfaces. The combined action of pressure pulses and fatigue processes incurred from the repetitive shock waves and micro jets causes severe cavitation damage of flow-handling components and leads to premature failure [4]. If the cavitating liquid is an aggressive aqueous solution, cavitation erosion and corrosion can take place simultaneously. The corrosion-enhanced cavitation-erosion aggravates the removal of the protective oxide film on the solid surface, since the mass loss can increase significantly beyond that due to either cavitation-erosion or corrosion alone [5]. The resistance to cavitation erosion for titanium alloys depends largely on their microstructure and mechanical properties, as well as the cavitation medium etc. Hou et al. [6] found that the microstructure of Ti-6Al-4V plays a significant part in its cavitation erosion behavior in deionized water. When Ti-6Al-4V was exposed to cavitation, induced by the repeated bubble collapse, the bcc structured β-phase was more readily removed with respect to the hcp structured αphase. This resulted in fatigue fracture of the remnant protuberant αlathes and led to signnificant cavitation erosion damage. Mochizuki et al. [7] investigated the cavitation erosion behavior of various titanium alloys using a rotating disk method in seawater. They showed that the cavitation erosion resistance was proportional to the Vickers hardness of the tested samples. Further, Li et al. [8] studied the effects of SiO 2 particles on ultrasonic cavitation erosion of Ti-6Al-4V in a 0.1 M H 2 SO 4 solution. It was shown that the mean depth of erosion of the alloy sharply decreased with increasing particle concentration and size.
For either corrosion or cavitation erosion in corrosive liquids, material degradation primarily occurs at the liquid/solid interface, that is dependent on surface, rather than bulk properties. Recently, much research has focused on improvement in the surface properties of titanium alloys under cavitation, through exploitation of surface engineering techniques. Mann et al. [9] reported that a laser-treated twin wire arc-sprayed SHS 7170 coating significantly improved the cavitation erosion resistance of Ti-6Al-4V. Further, Man et al. [10] examined the cavitation erosion behavior of laser gas nitrided commercial titanium alloys in a 3.5% NaCl solution. Their results showed that the cavitation erosion resistance was improved by an order of magnitude through surface treatment. Higher cavitation erosion resistance of the nitrided layer was attributed to an increase in surface hardness as well as the fine, dendritic microstructure of the TiN layer. Amongst the various surface modification technologies currently available, the double cathode glow discharge deposition method has been proven to be an effective approach to improve the wear resistance, corrosion resistance and oxidation resistance of titanium alloys through the deposition of nanocrystalline silicides coatings [11].
Transition metal silicides (TMS) possess unique combinations of outstanding physical and chemical properties, such as excellent resistance to high temperature oxidation, low density, high melting points and creep strength, endowing them with potential as candidate materials for high temperature applications [12]. Among these silicides, TaSi 2 , that exhibits a hexagonal structure with a P6 2 22 space group, has found wide applications in the semiconductor industry, because of its superconductivity and good contact ability with silicon. In particular, TaSi 2 shows great potential for use as a wear-corrosion resistant material, due to its high hardness and good corrosion resistance. However, similar to other metal disilicides, its poor room temperature toughness inhibits its widespread application. Usually, alloying with certain alloy elements to replace metal and Si sites in disilicides has been shown a fruitful path to increase toughness [12]. Our recent work has shown that the Al alloyed β-Ta 5 Si 3 has a higher resistance to erosion-corrosion and ultrasonic cavitation erosion as compared to the binary β-Ta 5 Si 3 coating and 316L stainless steel [13].
In this work, the influence of Al additions on the structural stability and elastic properties of TaSi 2 were systematically studied using firstprinciple calculations, in order to improve the understanding of the Al alloying on the brittle-or-ductile behavior of TaSi 2 at an atomic level. These theoretical results are valuable for the optimization of Al content in this compound. As a consequence, the binary TaSi 2 and ternary Ta (Si 0.875 Al 0.125 ) 2 coatings were deposited onto Ti-6Al-4V substrates using the double cathode glow discharge technique. The mechanical behavior and cavitation erosion behavior in a 0.5 M HCl solution for the two coatings were investigated and compared to the bare substrate.

Computational method
In this work, all the first-principles calculations for binary TaSi 2 and ternary Al alloyed TaSi 2 were conducted by using the density functional theory (DFT), as implemented in the Cambridge serial total energy package (CASTEP) code [14]. To compare the calculated results, the exchange correlation functional was treated by the generalized gradient approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) functional [15]. Ultrasoft pseudopotentials were used to describe the interactions between the ion core and valance electrons. The electronic configurations for the pseudo-atoms were Ta 5d 3 6s 2 , Si 3s 2 3p 2 and Al 3s 2 3p 1 , respectively. As seen from Fig. 1, a 2 × 2 × 1 supercell consisting of 12 Ta atoms and 24 Si atoms was constructed for the binary C40-structured TaSi 2 and a series of ternary C40-structured Ta(Si 1-x Al x ) 2 supercells were built by replacing Si atoms by 1, 2, 3 and 4 Al atoms, where the corresponding Al concentrations (i.e., the value of subscript x) were 0.042, 0.083, 0.125 and 0.167, respectively. Based on the convergence testing for the total energy at ground state, a plane-wave cutoff energy of 400 eV for both the binary and ternary Ta(Si 1-x Al x ) 2 were deemed adequate to achieve good convergence. According to the Monkhorst-Pack scheme, Brillouin zone integration was performed with a 4 × 4 × 4 k-point mesh for the Ta(Si 1-x Al x ) 2 supercell model. The tolerances of the geometrical optimization met the following conditions: total energy changes less than 2 × 10 −6 eV/atom, maximum ionic Hellmann-Feynman force below 0.01 eV/Å, maximum stress lower than 0.02 GPa and maximum ionic displacement within 2 × 10 −4 Å. The elastic constants (C ij ) for Ta(Si 1- x Al x ) 2 were determined from evaluation of stress tensor generated small strains [16]. According to the calculated C ij values, mechanical properties such as bulk modulus (B), Young's modulus (E), shear modulus (G) and Poisson's ratio (υ) could be derived using the Voigt-Reuss-Hill approximations method [17].

Specimen preparation
Ti-6Al-4V discs (Ø40 mm × 3 mm), with the nominal composition in wt%: Al, 6.04; V, 4.03; Fe, 0.3; O, 0.1; C, 0.1; N, 0.05; H, 0.015 and the balance, Ti, were used as substrates. Prior to deposition, the substrate samples were ground with different grades of silicon carbide papers and finally polished using 0.05 μm diamond pastes. Subsequently, the polished substrate samples were ultrasonically cleaned in ethanol and finally dried in cold air. A double cathode glow discharge system (Wuhan Shoufar Surface Engineering Ltd.) was employed to prepare both the binary and ternary Ta(Si 1-x Al x ) 2 coatings using two targets with different stoichiometric ratios (Ta 20 Si 80 and Ta 20 Si 70 Al 10 ), respectively. The chamber was evacuated to a base pressure of 5 × 10 −4 Pa. To remove any contamination from the surface of the substrates prior to deposition, the samples were cleaned by Ar ion bombardment at a potential of −650 V for 20 min. During sputter deposition, the pressure in the chamber was maintained at 35 Pa with an Ar gas flow rate of 80 sccm. The deposition conditions used were as follows: target electrode bias voltage (−900 V) with direct current (3.1 × 10 −2 A/cm 2 ); substrate bias voltage (−300 V) with impulse current (a pulse frequency and a duty cycle set at 1100 Hz and 85%, respectively); target/substrate distance of 10 mm; a deposition temperature of 800 • C and a deposition time of 3 h.

Microstructure characterization and mechanical tests
The phase composition of the as-deposited coatings was analyzed with a D8 ADVANCE X-ray diffractometer (XRD) using a Cu-Kα (λ = 1.54056 Å) X-ray source (potential 35 kV and current 40 mA). The diffraction spectra were collected using Bragg-Brentano geometry. The tested specimens were examined with a scanning speed of 0.05°per second in a 2θ range of 20°to 80°. The surface and cross-sectional microstructures of the as-prepared coatings were studied using a field emission scanning electron microscope (FE-SEM, Hitachi, S-4800, Japan), equipped with an energy-dispersive X-ray analyzer (EDX-4; Philips). Detailed microstructural examination of the coatings was performed using a JEOL JEM-2010 transmission electron microscope (TEM) operating at an accelerating voltage of 200 kV. Plan-view TEM specimens were mechanically ground only from the untreated side of the substrate and then thinned to electron transparency using a singlejet electrochemical polisher.
Hardness and elastic modulus were measured on the coatings' surfaces using UMIS (Ultra-Micro Indentation System 2000, CSIRO, Sydney, Australia) workstation, A Berkovich diamond indenter was employed, which was applied at a constant loading rate of 20 mN/min with a maximum applied load of 20 mN. Hardness and the elastic modulus were calculated using the Oliver-Pharr method [18] derived from the measured load-displacement curves. For each specimen, ten measurements were conducted and the presented values are an average of 10 indentations. To examine the contact damage resistance of the coatings, the cross-sections of the indentations, generated by a spherical indenter with a radius of 5 μm, were examined using a dual beam focused ion beam (FIB) microscope. The adhesion strength between the coating and the substrate was measured quantitatively by means of a WS-97 scratch tester, which was equipped with an acoustic emissions (AE) sensor. During scratch tests, a conical Rockwell C diamond stylus with a 0.2 mm radius was drawn across the coating surface at a constant loading rate of 20 N/min and a scratching speed of 1 mm/min. The critical load (Lc), defined as the minimum normal load required for starting adhesion failure, was determined by both acoustic emission fluctuations and through SEM observation of the scratch tracks.

Electrochemical measurements and XPS analysis
Electrochemical measurements were undertaken in a standard threeelectrode arrangement using a CHI 660C electrochemical workstation (Shanghai ChenHua Instruments Inc., China). The tests were carried out in a 0.5 M HCl solution using a saturated calomel reference electrode (SCE) and platinum plate counter electrode. Current-time response tests were conducted using a potential of +0.6 V SCE . Potentiodynamic polarization measurements were undertaken with an electrode potential range from −0.5 to +1.5 V SCE with a scan rate of 1 mV/s. The electrochemical impedance spectroscopy (EIS) experiments were performed at open circuit potential (OCP) using a perturbation amplitude of 10 mV over a frequency range from 100 kHz to 10 mHz. Then, the data acquired from EIS were analysed and modeled via an appropriate electrical equivalent circuit (EEC) using ZSimpWin software.
The composition and chemical valence states of the passive films formed on samples following corrosion testing were characterized by Xray photoelectron spectroscopy (XPS) using a Kratos AXIS Ultra ESCA System, operating with an Al Kα X-ray source, with a 12 kV accelerating voltage and a 12 mA emission current. The base pressure of the sample chamber was~10 −10 Torr or better. The pass energy was chosen to be 80 eV for survey scans and 10 eV for feature scans. After subtraction of the background signals, the spectra were fitted using both Gaussian and mixed Gaussian/Lorentzian functions. Binding energies were calibrated using carbon contamination to the C1s peak value at 284.8 eV. The standard binding energies of the analyzed elements were determined from the NIST XPS database (V4.0).

Cavitation-erosion tests
The cavitation erosion tests were performed using a JY92-IIDN magnetostrictive-induced cavitation apparatus (NingBo Scientz Biotechnology Co., Ltd.) with a maximum output power from the ultrasonic generator of 1000 W. Tests were performed according to ASTM G32-16 [19]. Fig. 2 shows a schematic diagram of cavitation erosion equipment. During cavitation erosion testing, the samples were mounted on a support table that was was positioned 1 mm below the tip of a vibrating horn made of titanium alloy. Both the vibrating horn and the test samples were immersed in the test liquid in a 1000 ml glass beaker. This was maintained at 20 ± 1°C by flowing cooling. The vibratory frequency and double vibratory amplitude were kept at 20 kHz and 60 μm, respectively. The total test period of each sample was 9 h and the tested samples were weighed periodically by an analytical balance with a sensitivity of 0.01 mg to determine mass loss.

Calculation results
The structural stability of a compound, which is closely associated with its formation enthalpy, is of great importance to evaluate its alloying ability. The formation enthalpy (ΔH f ) was determined by subtracting the total energies of individual constituent atoms in their bulk structures from the total energy of the compounds. ΔH f for Ta(Si 1-x Al x ) 2 where E tot [Ta(Si 1-x Al x ) 2 ] refers to the total energy of the super cell for the Ta(Si 1-x Al x ) 2 compounds; E(Ta), E(Si) and E(Al) signify the energy per atom for Ta, Si and Al atoms in the solid state, respectively. The equilibrium lattice constants and formation enthalpies for C40 structured Ta (Si 1-x Al x ) 2 determined as a function of Al concentration are tabulated in Table 1, along with other available theoretical data [20]. From Table 1, it is clear that the calculated lattice constants and formation enthalpy for binary TaSi 2 are close to previously generated theoretical results, suggesting that the present calculations are both reasonable and reliable. As expected, both the calculated lattice parameters a and c increase simultaneously with increasing substitution of Al atoms for Si, as the atomic radius of aluminum is slightly larger than that of silicon. All of Ta (Si 1-x Al x ) 2 compounds have a negative formation enthalpy, showing that they are thermodynamically stable in the ground state. Moreover, the formation enthalpy calculations suggest that the addition of aluminum lowers the structural stability of TaSi 2 , because the higher aluminum concentration has a more positive ΔH f value. Elastic properties of a solid, such as bulk modulus (B), Young's modulus (E), Shear modulus (G) and Poisson's ratio (υ), can help gain a deep understanding of its mechanical properties [21]. The elastic constants (C ij ), reflecting a solid's response to an external pressure, provide critical information to understand the elastic properties of compounds. According to the symmetry of hexagonal structured TaSi 2 , there are five independent elastic constants: C 11 , C 12 , C 13 , C 33 , and C 44 . Further, the elastic constant C 66 has a connection with other constants as C 66 = (C 11 − C 12 )/2. Table 2 gives the calculated elastic constants of binary and ternary TaSi 2 , together with the available theoretical data. Based on Born-Huang stability criteria, the elastic constants of hexagonal crystals need to satisfy the following inequalities [22]: Clearly, the calculated elastic constants for these compounds meet the Born-Huang stability criteria, indicating that both binary and ternary TaSi 2 are mechanically stable at the ground state. Furthermore, the calculated elastic constants of binary TaSi 2 are in strong agreement with the experimental and theoretical data reported by others [20,23]. It is obvious that the C 33 values for all of the Ta(Si 1-x Al x ) 2 compounds are larger than the C 11 values, suggesting that these compounds exhibit the higher deformation resistance along the c-axis directions in comparison with the a-axis direction under uniaxial stress. This may have resulted from the presence of strong Si-Si covalent bonds along the caxis . With increasing Al concentration, the C 33 values slightly decrease, implying that the Al additions weaken rigidity along the c-axis directions. For Al alloyed TaSi 2 , a component of the Si-Si covalent bonds is replaced by Al-Si covalent bonds. Hence, this result also indicates that the strength of the formed Si-Al covalent bonds in Al alloyed TaSi 2 is weaker than that of the Si-Si covalent bonds along the c-axis. In addition, the elastic constant, C 44 , indicates the resistance to monoclinic shear along the (1 0 0) plane. From Table 2, it is evident that the C 44 values decrease with Al alloying, denoting a decrease in shear resistance along the (1 0 0) plane.
In terms of the obtained C ij values, the mechanical moduli of Ta(Si 1- x Al x ) 2 compounds with differing Al contents can be determined using the Voigt-Reuss-Hill approach, as shown in Fig. 3. Our calculated data for binary TaSi 2 are consistent with calculations by others [20], further confirming the reliability of our calculations. Bulk modulus (B) is associated with the ability of a material to resist crystal volume changes under hydrostatic pressure, whereas the shear modulus (G) describes the ability of a solid to resist shape change caused by a shearing force, whilst Young's modulus (E) reflects the degree of elastic stiffness for a solid [24]. From Fig. 3, it is clear that the values for bulk modulus (B), shear modulus (G) and Young's modulus (E) all decrease with increasing Al additions. This indicates that the addition of Al into TaSi 2 not only weakens the deformation resistance to volume and shear deformation, but also reduces elastic stiffness. It is important to note that Young's modulus is also regarded as the indicator of the nature of chemical bonding of a compound [25]. Predictably, the increase of Al addition reduces the covalent feature of these Ta(Si 1-x Al x ) 2 compounds. Vickers hardness (H V ) represents the ability of a material to resist plastic deformation and thus plays an important role in material's selection, for example against wear. An empirical relationship between Vickers hardness and elastic modulus was proposed to estimate the hardness of polycrystalline materials through the following equation [26]: where k refers to the G/B ratio. This equation signifies that that Vickers hardness is more closely related to G/B ratio than either E or G alone, due to the G/B power index of 2. As shown in Fig. 3 To better understand the influence of Al alloying on the brittle-orductile behavior of TaSi 2 , we examined variations of Pugh ratio (B/G), Poisson's ratio (ν) and Cauchy pressure with Al concentrations in TaSi 2 , as presented in Fig. 4. According to Pugh's rule, the ratio of the bulk modulus to shear modulus (B/G ratio) can acts as a parameter to indicate whether a material exhibits brittle or ductile behavior. In general, the larger the B/G ratio, the greater the degree of ductility. The critical value of B/G ratio, separating ductility from brittleness, is approximately 1.75. If B/G ratio > 1.75, a solid is expected to exhibit ductile behavior, and vice versa; if B/G ratio < 1.75, then a solid shows brittle behavior. As shown in Fig. 4, the B/G ratios for all the Ta(Si 1-x Al x ) 2 compounds are lower than the critical value, suggesting that those compounds exhibit brittle behavior. A clear increase in B/G ratio is observed when the Al concentration increases from 0 to 12.5 at.%, and the B/G ratio changes little with further increases in Al addition. This is due to the fact that shear modulus decreases much faster than bulk modulus when the Al concentration increases from 0 to 12.5 at.%, and the decreasing trend of shear modulus is comparable with that for bulk modulus, with a further increase in Al addition. Poisson's ratio (υ) is the rate of transversal   bonding features of a solid and this parameter has been also extensively used to characterize the brittleness or ductility of the solid [27]. A crystal that exhibits a positive Cauchy pressure is ductile with metallic bonding, otherwise it is brittle with directional covalent bonding. For the five Ta (Si 1-x Al x ) 2 compounds, the values of Cauchy pressures are negative, but increase with increasing Al concentration, indicating an enhanced ductility. Clearly, the variation of the B/G ratios, Poisson's ratios and Cauchy pressures of the Ta(Si 1-x Al x ) 2 compounds with Al concentration all shows a similar trend: a significant increase in plasticity as the Al content increases from 0 to 12.5 at.%, but a negligible change in plasticity with further additions of Al. Therefore, we chose Ta(Si 0.875 Al 0.125 ) 2 as an appropriate Al addition with which to prepare an alloyed TaSi 2 coating and then compare its mechanical behavior and corrosion response under cavitation erosion conditions with those of the binary compound and Ti-6Al-4V substrate.  (1 0 1), (1 0 2), (0 0 3), (2 1 1) and (3 0 2) planes, respectively, that were consistent with the expected diffraction peaks for C40 structured TaSi 2 . For the ternary Ta(Si 0.875 Al 0.125 ) 2 coating, no characteristic peaks of other Ta silicides are detected, indicating that no phase transition occurred after substitutional alloying with Al in this compound. Compared with the binary TaSi 2 coating, all the diffraction peaks for the Ta(Si 0.875 Al 0.125 ) 2 coating shift towards lower 2θ angles, presumably because the atomic radius of Al is bigger than that of Si. In addition, both patterns displayed relatively weak and broad peaks, arising from the very fine grain size of the two as-deposited coatings. Fig. 6 shows typical cross-sectional SEM micrographs of the two asdeposited Ta(Si 1-x Al x ) 2 (x = 0 and 0.125) coatings, together with EDS elemental maps and energy dispersive X-ray point spectra from the coatings. From the cross-sectional SEM morphologies shown in Fig. 6, three distinct zones namely the Ta(Si 1-x Al x ) 2 coating, the substratecoating interface and Ti-6Al-4V substrate could be distinctly observed along the depth from the top to the bottom of cross-sectional coated samples. Both coatings have a uniform thickness of nearly 15 μm and exhibited a dense and homogeneous microstructure without cracking and delamination at the interface. Such a thick Ta(Si 1-x Al x ) 2 coating of good quality is desirable for protection of the metal substrate from mechanical and corrosion damage in the adverse circumstances. The elemental maps verify that Ta, Si and Al elements are homogeneously distributed throughout the coatings. EDS analysis of the coatings further revealed the presence of Ta, Si and Al (for the ternary coating) elements and Ta/Si or Ta/(Si + Al) atomic ratio was measured to be close to 0.5, consistent with the stoichiometry ratios for TaSi 2 and Ta(Si 0.875 Al 0.125 ) 2 . Fig. 7 presents typical planar view bright-field TEM and HRTEM images recorded from the Ta(Si 0.875 Al 0.125 ) 2 coating, along with corresponding selected area diffraction (SAD) pattern presented as an inset in Fig. 7(a). It is evident from Fig. 7(a) and (b) that the Ta(Si 0.875 Al 0.125 ) 2 coating is characterized by a uniform nanocrystalline grain structure. The grains are round in shape and appear equiaxed with an average diameter of~5 nm, as evidenced by the high-resolution TEM image (Fig. 7(c)). As can be seen from the SAD pattern, the almost continuous rings with uniform intensity suggest that the coating exhibits an essentially random crystallographic texture. The three Debye-Scherrer rings (marked in the inset of Fig. 7(a)) were identified as the (1 0 2), (0 0 3) and (2 1 1) Bragg reflections of hexagonal TaSi 2 , and no other crystalline phases were detected. This is consistent with the data obtained from XRD. We have employed an image quality enhancement program using Digital Micrograph software to analyze the high-resolution TEM image, as displayed on both the left and right sides in Fig. 7(c). It is evident from Fig. 7(c) that many spherical grains overlap with each other and the interlayer spacings of 2.56 Å and 2.19 Å, measured by Digital Micrograph software, correspond to the (1 0 2) and (0 0 3) planes, respectively, of hexagonal C40 structured TaSi 2 . A similar microstructure also exhibiting a hexagonal structure was observed for the TaSi 2 coating. It is apparent that the maximum penetration depths of the two coatings are significantly lower than 10% of their coating thickness and thus the contribution from the underlying substrate to the measured mechanical properties is negligible. As expected, under the same maximum load, both coatings exhibit smaller indentation depths and less steep unloading slopes than the uncoated Ti-6Al-4V. This suggests that the uncoated alloy undergoes significantly more plastic deformation than the two Ta(Si 1-x Al x ) 2 coatings. The hardness values of the TaSi 2 and Ta(Si 0.875 Al 0.125 ) 2 coatings measured from the P-h curves, determined by Oliver and Pharr method, are 24.92 ± 0.82GPa and 23.09 ± 0.75GPa, respectively, which are about 4.4 and 4.1 times, respectively, greater than that of uncoated Ti-6Al-4V which exhibited a hardness value of 5.69 ± 0.41GPa. Likewise, the values for the Young's modulus of the TaSi 2 and Ta(Si 0.875 Al 0.125 ) 2 coatings are about 3.3 and 2.9 times higher than that of uncoated Ti-6Al-4V alloy, respectively. It is important to note that the values for hardness and Young's modulus for the binary TaSi 2 coating are much higher than those reported elsewhere. For example, the hardness of nanostructured TaSi 2 , with an average grain size of 60 nm, was reported to be 13GPa) [28] and a bent TaSi 2 fiber was observed to exhibit a hardness of 11 GPa and a Young's modulus of 251 GPa) [29]. This higher hardness was also observed on other nanostructured metal silicides coatings reported in our previous studies, which was attributed to the finer grain size and denser microstructure [30]. Al alloying lowers the values of both hardness and Young's modulus of the TaSi 2 coating. This is consistent with the data obtained from first-principles calculations. During nanoindentation, the total deformation energy (W t ) induced by the indenter can be divided into two parts: energy dissipation of plastic deformation (W p ) and elastic recovery energy (W e ). The total deformation energy can be evaluated by the area enclosed by the loading plot and the maximum penetration depth, while the elastic recovery deformation energy can be evaluated by the area enclosed by the unloading plot and the maximum penetration depth. The ratios of W e /W t and W p /W t represent the contribution of elasticity and plasticity to total deformation, respectively [31]. As shown in Fig. 8(d), the Ta(Si 0.875 Al 0.125 ) 2 coating has a higher W p /W t ratio than the TaSi 2 coating, indicating that Al addition enhances the plasticity of the TaSi 2 coating. Fig. 8(e) shows the H/E and H 3 /E 2 ratios calculated using the measured values of E and H for the two Ta(Si 1-x Al x ) 2 coatings. Clearly, these ratios are approximately equal for the two coatings.

Mechanical properties
To investigate the influence of Al addition on the deformation behavior of the TaSi 2 coating, Vickers indentations were carried out on both coatings using different indentation loads ranging from 100 g to 1000 g with a 15 s dwell time, as shown in Fig. 9(a) and (b). Apparently, both coatings exhibit a plasticity-dominated response to indentation, with no evidence of radial cracking emanating from the indentation corners under an applied load. Owing to the absence of similar reported studies for TaSi 2 we compared the results of the two Ta(Si 1-x Al x ) 2 coatings with other metal silicides. Previous studies have shown that cracks were present at the corners of the indent impression for coarse grained binary silicides, such as MoSi 2 [32], Mo 5 Si 3 [33] and Cr 3 Si [34], and a MoSi 2 -SiC nanocomposite [35], when they are subjected to indentation loads less than or equal to 1000 g. Hence, in comparison with these metal silicides, the two Ta(Si 1-x Al x ) 2 coatings studied here show higher toughness, since they require larger deformation energy for crack formation. In addition, after applying an indentation load of 1000 g, the same characteristic circular patterns surrounding the edges of the indent impression were seen on both coatings. Such patterns are actually slipstep patterns from shear bands, associated with the circular stress distribution induced by a Vickers indenter [36]. This deformation mode is frequently found on bulk metallic glasses (BMG) materials subjected to indentation [37]. As described earlier, both coatings exhibit ultrafine crystallites with a mean grain size of~5 nm. Because of the lack of dislocation activity in such fine grain materials, the dominant deformation mechanism at room temperature is through shear sliding between the Ta(Si 1-x Al x ) 2 grains, which to some extent resembles deformation in BMG's. Fig. 9(c)-(f) shows plan-view images FIB cross-sections through indents on both coatings made under a normal load of 500 mN using a spherical indenter with a radius of 5 μm. The shape of the indentations on both coatings shows a round geometry in the plan-view images with no cracking observable on either the surface or cross-section of the indents. The Ta(Si 0.875 Al 0.125 ) 2 coatings exhibit larger projected indentation area and deeper non-recoverable indentation depth than those for the binary TaSi 2 coating, indicating that the binary TaSi 2 coating is harder. This is consistent with the values obtained using nanoindentation (Fig. 8). A number of researchers [38][39][40] have reported that sub-surface cracks can be observed underneath spherical indents applied using a 500 mN load on monolithic TiN coatings or TiSiN/TiN multilayer coatings. Consequently, these Ta(Si 1-x Al x ) 2 coatings show higher capability to accommodate deformation as compared to these TiN-based coatings. Fig. 10 shows the acoustic emission signal intensity of the two Ta (Si 1-x Al x ) 2 coatings as a function of applied normal load, together with the corresponding scanning electron microscope images of the scratch tracks (see inset). During the scratch tests, the critical load (L c ) is defined as the minimum value load needed to remove the as-deposited coating from the substrate. As shown in Fig. 10, it is evident that the starting point for the occurrence of continuous acoustic emission is observed at applied loads of 46 N and 57 N for the TaSi 2 coating and Ta (Si 0.875 Al 0.125 ) 2 coating, respectively. The SEM images of the scratch morphologies suggest that the width of the scratch tracks increases with increasing the applied normal load, and large-scale spalling on both sides of the scratch tracks is observed once the critical load is attained. These results demonstrate that the Ta(Si 0.875 Al 0.125 ) 2 coating exhibits superior in adhesion strength to the TaSi 2 coating. This is crucial to the reliability of the coating under hostile working conditions.

Open circuit potential (OCP)/current-time responses
Under quiescent and ultrasonic cavitation conditions, the changes of the open circuit potential (E OCP ) registered for the two Ta(Si 1-x Al x ) 2 coatings and uncoated Ti-6Al-4V at different output powers in a 0.5 M HCl solution are shown in Fig. 11(a), where the output powers versus cavitation exposure time plot is displayed in Fig. 11(b). Under quiescent condition, the E OCP for the three samples moves slowly to the noble direction. This shift of E OCP is usually attributed to the formation of protective passive films on those tested samples. In contrast, when a cavitation measurement is to be initiated, the E OCP of the tested samples shifts rapidly towards the cathodic direction. For the uncoated alloy, with increasing output powers, the E OCP instantly decreases. In the case of the two coatings, the E OCP almost maintains constant, being independently of the output powers. The influences of cavitation erosion on the E OCP are determined primarily by two competing processes. The formation and subsequent collapse of cavitation bubbles induced by the transducer vibrations increase mass transport of oxygen, leading to a noble shift of the E OCP . On the other hand, the repeated collapse of cavitation bubbles may cause the rupture of any spontaneously formed passive films on the tested samples and, thus, the underlying fresh surfaces are exposed to corrosive medium by the mechanically depassivation, resulting in a drop in E OCP . In our case, the latter process plays a predominant role in cavitation tests at applied output powers ranging from 400 to 800 W. Such an active shift of the E OCP , created by cavitation bubbles, has also been observed by other researchers [41]. Alternatively, an obvious E OCP oscillation can be clearly observed on the E OCP versus time curves for the tested samples. This is generally attributed to the depassivation/repassivation events, which occurred on the small regions of the tested samples [42]. Compared with the uncoated alloy, the amplitude of vertical oscillations in the E OCP is markedly reduced, suggesting that the passive films formed on the two coatings exhibit a higher self-healing capability. The E OCP of the Ta(Si 0.875 Al 0.125 ) 2 coating is greater than that of the TaSi 2 coating, indicating that Al addition further enhances the thermodynamic stability of the passive film present on the TaSi 2 coating. Fig. 11(c) shows the corresponding currenttime transient plots recorded for the two Ta(Si 1-x Al x ) 2 coatings and uncoated Ti-6Al-4V potentiostatically polarized at potential of +0.6 V SCE in a 0.5 M HCl solution under different output powers. The potential of +0.6 V SCE employed here is located in the passive region observed from potentiodynamic polarization curves of the tested sample, as described in the following sections. For the uncoated alloy, in absence of cavitation, its current density is very stable; when cavitation is present, its current density rises sharply with increasing output powers, accompanied by large and continuous fluctuations in current. On the contrary, the application of cavitation does not appreciably affect the current density of the two coatings, especially for the Ta(Si 0.875 Al 0.125 ) 2 coating, which has a smoother current-time plot as compared to the TaSi 2 coating.  Fig. 12 shows typical potentiodynamic polarization curves of the two Ta(Si 1-x Al x ) 2 coatings together with bare Ti-6Al-4V in a 0.5 M HCl solution after different cavitation exposure times under a given output power of 800 W. All of the tested samples were polarized up to an anodic potential of +1.5 V SCE to examine their passive nature. Table 3 gives the electrochemical parameters describing the polarization behavior, including corrosion potential (E corr ), corrosion current density (I corr ), and anodic/ cathodic Tafel slopes (βa and −βc) extracted from polarization curves. As shown in Fig. 12, with increasing cavitation exposure time, a marked upwards displacement in cathodic branches is observed for the tested samples, meaning that the cathodic current densities rise continuously by prolonging cavitation time. For the hydrochloric acid medium used here, the cathodic reactions are mainly controlled by reduction of hydrogen ions and dissolved oxygen at the sample/solution interface. Hence, an increase of cavitation time accelerates the cathodic reactions process and speeds up the cathodic reaction rate. The anodic branches of the tested samples exhibit spontaneous passivation after different cavitation time, without exhibiting any active-passive transition. Analogous to the E OCP /currenttime curves, all the anodic branches show wide current oscillations, and gradually drift to higher anodic current density zones, with increasing cavitation time. Upon anodic polarization, the tested samples' surfaces undergo electrochemical dissolution and corrosion products form by anodic oxidization. With increasing cavitation time, the formed passive film becomes less protective and thinner due to the mechanical impact caused by the collapse of bubbles and the anodic dissolution was accelerated by more active sites exposed for corrosive attack [43]. As a result, the passivation current densities of the tested samples increase due to the prolonged interaction of the H + and Cl − ions present in the 0.5 M HCl solution with the passive films formed on the tested specimens. From Table 3, it is observed that the anodic Tafel slopes (βa) are bigger than the cathodic counterparts (−βc) for the tested samples at a given cavitation time. This denotes that the overall kinetics of cavitation corrosion for the two Ta(Si 1-x Al x ) 2 coatings and uncoated alloy in a 0.5 M HCl solution are under anodic control, that is, the existence of the passive films grown on the tested samples. As shown in Table 3, with an increase of cavitation time, the E corr values for the two coatings shifts towards more negative values, while the E corr for the uncoated alloy tends to move towards more positive values. As the cavitation time is extended from 0 to 9 h, the anodic and cathodic branches of the tested samples shift following the order a 1 → a 2 → a 3 → a 4 and c 1 → c 2 → c 3 → c 4 , respectively, and the corresponding E corr changes in the following order E1 corr → E2 corr → E3 corr → E4 corr. For the uncoated alloy, the displacement of the cathodic branches to more positive potential region is more pronounced than that of the anodic branches moving to the lower potential region. Contrary to the uncoated alloy, a marked leftward displacement to lower potential region in the anodic branches of the two coatings is observed. Such variations explain why different E corr shifts are observed on both coatings and bare Ti-6Al-4V. The values of i corr for the tested samples increase with cavitation time, suggesting a reduction in the protective properties of the passive films  Fig. 13 displays the impedance spectra of the two Ta(Si 1-x Al x ) 2 coatings and uncoated Ti-6Al-4V as Nyquist plots/Bode plots versus cavitation time in a 0.5 M HCl solution under a given output power of 800 W. Under the quiescence condition, the Nyquist plots of the test samples exhibit single, large unfinished capacitive semi-circle over the entire frequency range and the corresponding Bode-phase plots are typical of passive character, characterized by a broad frequency range with a maximum near −90° [44]. These results indicate that the passive films formed on the tested samples under static conditions show a strong capacitive behavior and endow the tested specimens with large electrochemical reaction resistance. In this case, impedance spectra registered for the tested samples is composed of only single time constant, and thus a modified Randles circuit (R s ((Q p R p )) was proposed to fit the impedance data, as shown in Fig. 14(a). In this equivalent circuit, a parallel combination of a resistor, R p , representing the resistance of passive films, and a capacitor, Q p , representing the capacitance of the passive films, are in series with a resistor, R s , representing the solution resistance. For ultrasonic cavitation conditions, the capacitive loops in the Nyquist plots become more depressed and the diameters of the capacitive arcs decrease with increases in cavitation exposure time, implying that the protective ability of the passive films formed on the tested samples is decreased with increasing cavitation exposure time. From the Bode-phase plots, it is evident that there are two distinct phase angle peaks located in the high and low frequency parts for the uncoated alloy and binary TaSi 2 coating under ultrasonic cavitation erosion conditions. An equivalent circuit with two time constants, given in Fig. 14(b), was used to fit the experimental EIS data obtained from uncoated alloy and binary TaSi 2 coating after different cavitation times. From the literature, the two time constants are relevant to the generation of a passive film with bilayer architecture, that consists of a more protective, inner, dense layer with a lower frequency time constant and a porous and non-protective outer layer with a higher frequency time constant [45]. In this equivalent circuit (R s (Q p R p ) (Q b R b )) comprises two capacitive/resistive elements in series, among which R b and Q b represent the resistance and capacitance of the inner dense layer, and R p and Q p represent the resistance and capacitance of the outer porous layer. In the case of the Ta(Si 0.875 Al 0.125 ) 2 coating, the impedance spectra obtained from 3 h of exposure were fitted using a (R s ((Q p R p )) model shown in Fig. 14(a) and the (R s (Q p R p )(Q b R b )) model shown in Fig. 14(b) was employed to simulate the experimental impedance data obtained from 6 h and 9 h of exposure. Table 4 summarizes the calculated equivalent circuit parameters derived from fitting the EIS experimental data of the tested samples. The chi-square (χ 2 ) values were of the order of 10 −3 -10 −4 that suggests the validity of the proposed circuit models. Under the quiescence condition, the R p value increases from 4.62 × 10 6 Ω cm 2 for the uncoated alloy to 8.52 × 10 6 Ω cm 2 for the binary TaSi 2 coating and further to 1.17 × 10 7 Ω cm 2 for the Ta(Si 0.875 Al 0.125 ) 2 coating. This increasing order of R p values is in agreement with the increased protective capability of the passive film grown on the tested samples. As shown in Table 4, for all the tested samples under cavitation erosion conditions, the resistance (R p ) values of the outer layer are negligible as compared to the resistance (R b ) of the inner layer. This is because the porous outer layer contains highly defective or microscopic pores, acting as shortcircuiting paths for the penetration of species from the electrolyte. Therefore, the inner dense layer is predominantly responsible for corrosion resistance of the tested specimens under cavitation erosion conditions. Moreover, the exponent, n, of a constant phase element (CPE) reflects the heterogeneity of the passive films due to the presence of defects and is regarded as an indicator of passive film quality [46]. The lower n values usually mean that a passive film exhibits a rough and porous surface. With increasing ultrasonic cavitation time, both the resistance (R p and R b ) and n (n p and n b ) values decrease. These results indicate that an increase in the ultrasonic cavitation time leads not only to a significant increase the number of defects for the porous outer layer, but also defects extending towards the dense outer layer. At a given cavitation time, the resistance values for the Ta(Si 0.875 Al 0.125 ) 2 coating are apparently higher than those of the binary TaSi 2 coating and the bare Ti-6Al-4V alloy, denoting the beneficial effect of Al alloying on the cavitation corrosion performance of the TaSi 2 coating.

XPS analysis
To gain a deeper understanding of the effect of Al alloying on the composition and chemical states of the elements present in passive films formed during cavitation tests, XPS analysis was carried out on the surface of the two Ta(Si 1-x Al x ) 2 coatings after 3 h of cavitation erosion testing in a 0.5 M HCl solution under a given output power of 800 W. The XPS survey spectrum for the Ta(Si 0.875 Al 0.125 ) 2 coating (Fig. 15(a)) shows the Ta 4f, Si 2p, O1s and Al 2p core level principal peaks, whereas for the TaSi 2 coating, no Al 2p peaks were determined, as expected. A small C 1s peak appeared at 284.7 eV with negligible intensity that originated from a contaminant hydrocarbon layer covering the specimen surface. From Fig. 15(b), the measured Ta 4f photoelectron spectrum for the binary TaSi 2 coating was composed of one doublet peaks from 4f 7/2 and 4f 5/2 located at 26.3 and 28.2 eV, which corresponds to the binding energy of Ta in stoichiometric Ta 2 O 5 [47]. In Fig. 15(c), the measured Si 2p spectrum for the TaSi 2 coating shows one component with a binding energy at about 103.6 eV, corresponding to Si in the form of fully oxidized SiO 2 [48]. In Fig. 15(d), an O 1s signal was recorded for the TaSi 2 coating that exhibited a broad peak, which can be fitted with two overlapped peaks: the lower binding energy peak, centered at 530.5 eV, can be assigned to O 2− ions, the higher binding energy peak, centered at 531.4 eV, can be attributed to adsorbed hydroxyl groups [49]. For the Ta(Si 0.875 Al 0.125 ) 2 coating, the Al 2p spectrum, shown in Fig. 15(e), appears as single peak centered at a binding energy of 74.4 eV, corresponding to oxidized aluminum (Al 2 O 3 ). It is noteworthy that the Ta 4f peaks for the Ta (Si 0.875 Al 0.125 ) 2 coating (Fig. 15(b)) shifted towards higher binding energy as compared to the binary TaSi 2 coating. This is due to the interaction between Ta and Al, leading to charge transfer from Al 3+ to Ta 5+ ions in the passive films [50]. Fig. 15(f) shows the content of the main components present in the passive films on the two coatings. As presented in Fig. 15(f), the passive film grown on the Ta(Si 0.875 Al 0.125 ) 2 coating has lower Ta 2 O 5 concentrations and higher SiO 2 concentrations as compared to the passive film present on the binary TaSi 2 coating. Compared with Ta 2 O 5 , SiO 2 exhibits higher dielectric characteristics together with a smaller oxygen diffusion coefficient [51]. Hence, Al additions promote SiO 2 enrichment in passive film, which is conducive to the improvement in the corrosion resistance for the Ta(Si 1-x Al x ) 2 coatings. Fig. 16 shows the cumulative weight loss of the two Ta(Si 1-x Al x ) 2 coatings and uncoated Ti-6Al-4V as a function of cavitation time in distilled water and in a 0.5 M HCl solution under a given output power of 800 W. For the cavitation erosion tests in distilled water, the variation of cumulative weight losses with cavitation time can be divided into two distinct stages: an incubation stage with a negligible weight loss and subsequently a slight increase stage. It is obvious that the incubation period for the uncoated alloy is shorter than the two coatings in distilled water. However, no incubation period was observed for the samples tested in a 0.5 M HCl solution, and the cumulative weight losses increase steadily with increases in cavitation time. A discrepancy in cavitation erosion properties in the two cavitating solutions can be directly traceable to the synergy between the cavitation erosion and corrosion processes that play an important role in controlling the cavitation erosion resistance of the tested samples. Whether in a 0.5 M HCl solution or in distilled water, the weight loss of the TaSi 2 coating is markedly decreased as compared to the uncoated counterpart. A further decrease is achieved by the Ta(Si 0.875 Al 0.125 ) 2 coating. Fig. 17 presents the evolution of the eroded-corroded surface morphologies of the two Ta(Si 1-x Al x ) 2 coatings and bare Ti-6Al-4V after different cavitation exposure times in a 0.5 M HCl solution at a given output power of 800 W. As shown in Fig. 17 (b), for the uncoated alloy after 3 h of cavitation erosion, the original smooth surface becomes broken up and material removal is characterized by ductile fracture, accompanied by the roughened surface appearance containing numerous micro-craters. With increasing the cavitation time to 6 h and 9 h (Fig. 17 (c) and (d)), the uncoated alloy suffered more severe cavitation damage with some coalescence of micro-craters that developed into larger and deeper craters induced by micro-plastic deformation, yielding a honeycomb appearance. Ti-6Al-4V is a two-phase alloy composed of α-Ti and intergranular β-Ti, where there is disparity in strength between the harder α-Ti and the softer β-Ti phases [1]. Because cavitation erosion is a fatigue process, the selective erosion damage is more likely to occur at the α/β grain boundaries and this subsequently propagates into the α-Ti grains [10]. Meanwhile, the surface passive film formed on the alloy is easily disrupted by mechanical shock generated from bubble collapse. As the cavitation erosion process continues, the Ti-6Al-4V alloy sample contains a high density of defects, such as pores and cracks. Such a rough and porous surface not only increases the number of nucleation sites for bubbles, but also is detrimental to surface repassivation [52]. By comparison, as shown in Fig. 17 (f), after 3 h of the cavitation erosion test, the TaSi 2 coating still exhibited the asdeposited surface morphology, except a small number of shallow craters with an average diameter of~13 μm. These craters originated from brittle delamination of the outermost layer of the TaSi 2 coating. With increasing cavitation time (Fig. 17 (g) and (h)), the number and size of these craters increased, although much of the as-deposited surface remained relatively smooth. For the Ta(Si 0.875 Al 0.125 ) 2 coating, the entire eroded surface exhibited a smooth and compact appearance with almost no visible cavitation craters, after testing for three hours (Fig. 17 (j)). With prolonging cavitation exposure time (Fig. 17 (k) and (l)), a number of small and

Discussion
Cavitation erosion resistance of a material is affected by several factors, which depend, not only upon mechanical properties, including hardness, fracture energy, tensile strength, Young's modulus, fatigue strength, but also on the cavitation experimental conditions, such as the aggressiveness, flow velocity and temperature of cavitating liquid, gas content and the intensity of cavitation [53]. Understanding the key parameter responsible for the cavitation erosion resistance of a specific material is essential in explaining its degradation mechanism. Kaspar et al. [54] studied the cavitation erosion resistance of laser nitrided Ti-6Al-4V under different gas atmospheres in deionized water. Their results showed an inversely proportional relationship between cavitation erosion rate and the hardness of laser nitride layer is tenable under gas atmospheres with N contents between 0 and 13%. However, the erosion rate did not reduce further with greater increases in surface hardness. In our case, the reverse tendency seems to occur, i.e. the cavitation erosion rate of the Ta(Si 1-x Al x ) 2 coatings was proportional to the coating hardness. H/E and H 3 /E 2 ratios have been previously used as suitable indicators to predict the tribological performance of materials. Recently, several studies have also attempted to correlate cavitation erosion resistance to these ratios [55,56]. However, apparent contradictions involving the relationship among those ratios and erosion rates have been observed [57,58]. Our experimental results also indicate that these ratios bear no clear relation to cavitation erosion resistance of the two Ta(Si 1-x Al x ) 2 coatings. The H/E and H 3 /E 2 ratios for the TaSi 2 and the Ta(Si 0.875 Al 0.125 ) 2 coatings are approximately equal, as presented in Fig. 8(e). However, from the eroded-corroded surface morphologies of the two Ta(Si 1-x Al x ) 2 coatings (Fig. 17), cavitation craters appeared to result from the fracture and delamination of the coatings. This suggests the surface degradation of these coatings occurred in a brittle manner. Therefore, it appears that fracture toughness is the most relevant mechanical property which governs the cavitation erosion resistance of these Ta(Si 1-x Al x ) 2 coatings. According to the results obtained from first-principle calculations, Al additions can improve the toughness of TaSi 2 , which is consistent with the observations from cavitation erosion testing. Moreover, in the case of the hard Ta(Si 1-x Al x ) 2 coating-soft Ti-6Al-4V substrate system, there is remarkable difference in mechanical properties between the coatings and Ti-6Al-4V substrate. For example, the two Ta(Si 1-x Al x ) 2 coatings have approximately three times greater Young's modulus than the Ti-6Al-4V substrate. During the cavitation erosion process, the Ta(Si 1-x Al x ) 2 coating and Ti-6Al-4V substrate show different responses to impact loading exerted by the cavitation pulses. During the unloading process, the Ta(Si 1-x Al x ) 2 coatings exhibit a higher relaxation rate than the Ti-6Al-4V substrate, but the coating/ substrate adhesion requires the coating to adjust to the deformed Ti-6Al-4V substrate to ensure they relax together. This leads to increased residual stresses in the coating and the decreased adhesion of the coating to the substrate [59]. Compared with the binary TaSi 2 coating, the ternary Ta(Si 0.875 Al 0.125 ) 2 coating possesses a higher adhesion strength, which is beneficial in reducing coating delamination during cavitation erosion process. Alternatively, when the tested samples are exposed to a corrosive cavitating liquid, the surface passive film and its repair capacity play a crucial role in their cavitation erosion resistance, due to the presence of the synergetic effect between mechanical erosion and electrochemical corrosion. The mechanical action of bubble implosion can accelerate removal of an existing passive film or reduce its thickness, thereby exposing the underlying fresh surface. This phenomenon is further supported by the obvious rise in current density under the hydrodynamic conditions that is registered with respect to the static conditions. Subsequently, the galvanic coupling between the mechanically depassivated areas and the surrounding passive areas is created, resulting in an increment of I corr value and an active shift of E corr as shown in Fig. 12. The damaged passive film can be subsequently reformed due to re-passivation and the erosion rate of the tested samples depends on the competition between the thinning or removal of passive films and its self-healing capability [60]. According to the above electrochemical measurements, Al additions enhance the thermodynamic stability and the self-healing capability of the passive film on the TaSi 2 coating, thus improving its cavitation erosion resistance.

Table 4
Electrochemical parameters obtained from numerical fitting for the uncoated Ti-6Al-4V and two coatings after cavitated for different time at an output power of 800 W.

Conclusions
The influence of Al alloying on the structural stability and mechanical behavior of C40-structured TaSi 2 was investigated by first-principles methods. In the light of these theoretical calculations, a Ta(Si 0.875 Al 0.125 ) 2 compound was selected as an appropriate Al addition to prepare an Alalloyed TaSi 2 coating using a double cathode glow discharge plasma method. Following detailed microstructural characterization and mechanical property measurements, the electrochemical behavior of the binary TaSi 2 and ternary Ta(Si 0.875 Al 0.125 ) 2   solution was determined by means of various electrochemical analytical techniques under ultrasonic cavitation erosion conditions. The following conclusions can be drawn: (1) Theoretical studies on the mechanical properties of Ta(Si 1-x Al x ) 2 compounds show that a significant increase in plasticity as Al content increases from 0 to 12.5 at.%, but with a negligible change in plasticity after further additions of Al. (2) The Ta(Si 1-x Al x ) 2 coatings exhibit a uniform thickness and dense structure, composed of spherically shaped nanocrystallites with an average diameter of about 5 nm. The hardness values of the TaSi 2 and Ta(Si 0.875 Al 0.125 ) 2 coatings are about 4.4 and 4.1 times greater, respectively, than that of uncoated Ti-6Al-4V substrate. The Al addition increases the ratio of W p /W t of the Ta(Si 1-x Al x ) 2 coating, suggesting that Al alloying enhances the plastic deformation capability of the TaSi 2 coating. (3) The results of electrochemical measurements indicate that the Al addition enhances the thermodynamic stability and the self-healing capability of the passive film on the Ta(Si 1-x Al x ) 2 coating, thus improving its cavitation erosion resistance.