Improvement of CoCr Alloy Characteristics by Ti-Based Carbonitride Coatings Used in Orthopedic Applications

The response of the human body to implanted biomaterials involves several complex reactions. The potential success of implantation depends on the knowledge of the interaction between the biomaterials and the corrosive environment prior to the implantation. Thus, in the present study, the in vitro corrosion behavior of biocompatible carbonitride-based coatings are discussed, based on microstructure, mechanical properties, roughness and morphology. TiCN and TiSiCN coatings were prepared by the cathodic arc deposition method and were analyzed as a possible solution for load bearing implants. It was found that both coatings have an almost stoichiometric structure, being solid solutions, which consist of a mixture of TiC and TiN, with a face-centered cubic (FCC) structure. The crystallite size decreased with the addition of Si into the TiCN matrix: the crystallite size of TiCN was 16.4 nm, while TiSiCN was 14.6 nm. The addition of Si into TiCN resulted in smaller Ra roughness values, indicating a beneficial effect of Si. All investigated surfaces have positive skewness, being adequate for the load bearing implants, which work in a corrosive environment. The hardness of the TiCN coating was 36.6 ± 2.9 GPa and was significantly increased to 47.4 ± 1 GPa when small amounts of Si were added into the TiCN layer structure. A sharp increase in resistance to plastic deformation (H3/E2 ratio) from 0.63 to 1.1 was found after the addition of Si into the TiCN matrix. The most electropositive value of corrosion potential was found for the TiSiCN coating (−14 mV), as well as the smallest value of corrosion current density (49.6 nA cm2), indicating good corrosion resistance in 90% DMEM + 10% FBS, at 37 ± 0.5 ◦C.


Introduction
In medical applications, especially for hip and knee implants, a CoCr alloy is mostly used, due to good mechanical, anticorrosive and tribological characteristics [1][2][3]. CoCr alloys have also been used as screws in trauma plating systems. In this case, a low osseointegration, compared to Ti alloys, could facilitate an easier removal after the healing of the bone fracture. Due to its higher strength, a CoCr alloy was also used for idiopathic scoliosis applications, where the results proved to be better in the an alternative solution was the addition of Zr and Cr to the Si-N and Si-C-N matrix for severe wear and corrosive applications [33].
In the present study, the coatings were obtained by the cathodic arc evaporation method on the CoCr substrates under a mixture of CH 4 and N 2 gases. The investigation included the examination of elemental and phase composition, texture, structure, morphology and mechanical properties (Young modulus, hardness, roughness, stress). Special attention was devoted to the corrosion resistance performed in a 90% DMEM + 10% FBS solution, at 37 ± 0.5 • C. In order to understand the damaging effect of the corrosion test, the morphology and roughness after corrosion were evaluated. Electrochemical impedance spectroscopy (EIS) was also performed in order to investigate the behavior of the proposed systems. Thus, this method gave an insight into the electrochemical processes which occurred at the material-electrolyte interfaces. The research was conducted in order to find new and improved structures as a better solution to optimize the safety and efficacy of biomaterials.

Coatings Preparation
Both coatings were prepared by the cathodic arc deposition process. For TiCN, one Ti cathode (99.99% purity) was used, while for TiSiCN, a Ti+Si cathode (85 at.% Ti and 15 at.% Si; 99.9% purity) was used. The cathodes were supplied from Cathay Advanced Materials Limited, Guangdong, China. The coatings were prepared using both CH 4 and N 2 as reactive gases (99.999% purity, Linde). The position of each cathode inside the deposition chamber was the same as those described in [33]. In order to guarantee the uniform thickness of the coatings, the substrate holder was rotated by 15 rot/min. The CoCr substrates (ASTM F75 CoCr alloy) were cut into 12 mm discs and polished up to Ra roughness of 46.9 ± 5.9 nm. Each substrate was ultrasonically cleaned in trichloroethylene and flushed with dry nitrogen, and then introduced in the deposition chamber. To eliminate any impurity, the substrates were sputter etched with Ar + ions (1 keV) for 15 min. The deposition parameters are listed in Table 1, and were maintained constant during all deposition runs. The same negative substrate bias voltage was applied on both cathodes and the substrate temperature was around 320 • C. The thickness of the coatings was around 2.5 µm.

Coatings Characterization
Energy dispersive spectrometry (EDS) was used for the analysis of the elemental composition (EDS, Quantax70, Bruker, Tokyo, Japan). The morphology was examined by scanning electron microscopy (SEM, Hitachi TM3030Plus, Tokyo, Japan). Phase composition was studied by the X-ray diffraction method (XRD, SmartLab diffractometer, Rigaku, Tokyo, Japan), using Cu Kα radiation, from 10 • to 100 • with a step size of 0.02 • /min. For the thickness and surface roughness determination, a surface profilometer (Dektak 150, Bruker, Billerica, MA, USA) was used. Surface roughness was measured for each investigated sample on five line-scans, each on a distance of 4000 µm.
The mechanical properties of the coatings were determined using a Hysitron Premier TI nanoindentation unit equipped with a Berkovich indenter tip of 100 nm radius and a total included angle of 142.3 • , respectively. Prior to any sample testing, the Z-axis calibration was performed in the air and the machine compliance was assured using a fused quartz standard calibration sample with known hardness (H = 9.25 GPa ± 10%) and elastic modulus (E = 69.6 GPa ± 10%). In order to perform the nanoindentation experiments, a 15 × 15 µm 2 area was previously scanned using the same Berkovich diamond tip at a normal force of 2 µN to investigate the surface roughness for the subsequent indents. Additionally, the indents were intentionally located at least 5 µm apart from each other, whilst an applied force of 10 mN was employed for every nanoindentation test. The force-displacement curves were recorded using a gradual force increase up to 10 mN in a 7-s time interval, followed by a 2-s dwell time at the maximum force of 10 mN and a gradual force decrease within the next 7 s, until the complete tip retraction from the coatings surface.
The electrochemical behavior of the investigated specimens was analyzed by potentiodynamic polarization and electrochemical impedance spectroscopy (EIS), using a VersaSTAT 3 Potentiostat/ Galvanostat system. The measurements were performed in 90% DMEM + 10% FBS, at 37 ± 0.5 • C, using a typical three electrode setup with a Pt grid counter electrode (CE) and a saturated calomel (saturated KCl) (SCE) as the reference electrode (RE), while the working electrode consisted of uncoated CoCr and TiCN or TiSiCN coated CoCr substrates, respectively.
The open circuit potential (E OC ) was monitored for 1 h, starting after the sample's immersion. Linear polarization, Tafel and potentiodynamic curves were performed by applying a potential of −20 to 20 mV vs. E OC , −50 to 250 mV vs. E OC and −1 V vs. E OC to 2 V vs. RE, respectively, with a scanning rate of 0.167 mV/s. For the linear polarization measurements, the testing conditions were selected based on preliminary results, in such a way that the applied potential gave a linear behavior. The selected value was used to accommodate all the investigated systems. The EIS measurements were performed over a range of frequencies (0.1 ÷ 10 4 Hz), by applying a sinusoidal signal of 10 mV RMS vs. E OC .

Elemental and Phase Composition
The elemental composition of the coatings obtained by the EDS method is presented in Table 2. Both coatings had an almost stoichiometric structure. The amount of Si was low but proved to be sufficient for the goal of this study. Both coatings exhibited a preferred growth orientation after (111) plan ( Figure 1). The TiCN peaks were located between those of TiC and TiN, indicating that the TiCN is a solid solution and consisted of a mixture of TiC and TiN, with a face-centered cubic (FCC) structure (like NaCl). The same results were also reported by other authors regarding the TiCN coatings [21]. The identification of TiCN was performed according to JCPDS no. 042-1488 (red lines in Figure 1). It can be seen that our TiCN is shifted towards a lower Bragg angle compared with the TiCN standard, which is related to the formation of more TiC phase ( Figure 1). This finding is in accordance with the results reported by Wang et al. [34] regarding the TiSiCN coatings. The preferred orientation of both coatings was also confirmed by the texture coefficient T(hkl), a strong texture intensity at (111) plane being found in the case of both coatings. Karlsson et al. reported that the plane with a preferred orientation parallel to the investigated surface has a texture coefficient value higher than 1 [21].  The crystallite size was determined by the Debye-Scherrer equation of the peak (111) and is presented in Table 2. The crystallite size decreased from 16.4 nm for TiCN coatings to 14.6 nm for TiSiCN. The possible explanation is the penetration of impinging Si ions into the lattice of the TiCN coating, and the decreased generation of defects, led to a decreased number of preferential nucleation sites causing reduced grain [35,36]. The strain ε was determined by the Williamson-Hall plot method and is presented in Table 2. In both cases, the strain was low, but after the addition of Si, the strain was significantly reduced. It is reasonable to believe that the observed difference in strain of the TiSiCN coatings is related to the amorphous phases in which Si can be found (Si, SiNx, SiCN) or the C=C phase at grain boundaries, which is detrimental for crystallite development. These effects will be goal of another paper, in which other complex analyses, such as TEM and XPS, will be carried out.
The lattice parameter obtained from 2θ values of the (111), (200), (220) and (311) of TiSiCN was 4.2771, while for TiCN it was 4.2725. According to JCPDS no. 042-1488, TiCN has a lattice parameter equal to 4.2644. When the diffraction peak is shifted towards lower Bragg angle, an increase of lattice parameter is observed. This result is in good agreement with those of Constantin et al. related to TiCN coatings [37]. Both coatings exceeded the size of the standard unit cell parameters. This finding can be a sign that the hardness can be low and, on the other hand, the toughness will be high [38].
Grieveson et al. stated that the TiN and TiC compounds do not combine perfectly; the mixture is far from the ideal Raoultian behavior [39,40]. The same conclusion was also considered by Levi et al., who revealed that the occupation of all sites by Ti, C and N is not random, it is a TiN-based structure with a replacement of the N site by C, forming an FCC or tetragonal structures on the low concentration of vacancies that might be present [41]. All these aspects had an important effect on the characteristics of the TiCN coatings. The crystallite size was determined by the Debye-Scherrer equation of the peak (111) and is presented in Table 2. The crystallite size decreased from 16.4 nm for TiCN coatings to 14.6 nm for TiSiCN. The possible explanation is the penetration of impinging Si ions into the lattice of the TiCN coating, and the decreased generation of defects, led to a decreased number of preferential nucleation sites causing reduced grain [35,36]. The strain ε was determined by the Williamson-Hall plot method and is presented in Table 2. In both cases, the strain was low, but after the addition of Si, the strain was significantly reduced. It is reasonable to believe that the observed difference in strain of the TiSiCN coatings is related to the amorphous phases in which Si can be found (Si, SiNx, SiCN) or the C=C phase at grain boundaries, which is detrimental for crystallite development. These effects will be goal of another paper, in which other complex analyses, such as TEM and XPS, will be carried out.
The lattice parameter obtained from 2θ values of the (111), (200), (220) and (311) of TiSiCN was 4.2771, while for TiCN it was 4.2725. According to JCPDS no. 042-1488, TiCN has a lattice parameter equal to 4.2644. When the diffraction peak is shifted towards lower Bragg angle, an increase of lattice parameter is observed. This result is in good agreement with those of Constantin et al. related to TiCN coatings [37]. Both coatings exceeded the size of the standard unit cell parameters. This finding can be a sign that the hardness can be low and, on the other hand, the toughness will be high [38].
Grieveson et al. stated that the TiN and TiC compounds do not combine perfectly; the mixture is far from the ideal Raoultian behavior [39,40]. The same conclusion was also considered by Levi et al., who revealed that the occupation of all sites by Ti, C and N is not random, it is a TiN-based structure with a replacement of the N site by C, forming an FCC or tetragonal structures on the low concentration of vacancies that might be present [41]. All these aspects had an important effect on the characteristics of the TiCN coatings.

Morphology and Roughness (Before Corrosion)
The surface morphologies of the uncoated CoCr substrate and the TiCN and TiSiCN coatings can be observed in Figure 2. The appearance of the coatings is characterized by a continuous coverage of the substrate with visible individual microparticles distributed on the surface. These droplets are considered defects and are characteristic of the energetic ejection of the particles during the arc deposition process [42,43].

Morphology and Roughness (Before Corrosion)
The surface morphologies of the uncoated CoCr substrate and the TiCN and TiSiCN coatings can be observed in Figure 2. The appearance of the coatings is characterized by a continuous coverage of the substrate with visible individual microparticles distributed on the surface. These droplets are considered defects and are characteristic of the energetic ejection of the particles during the arc deposition process [42,43].

Roughness of Surface
The main surface texture parameters of the investigated surfaces are: Ra: the arithmetic average of the roughness profile, Rq: the root mean square average of the roughness profile, Sk: skewness (according to the ISO 2517 standard [44]), and are presented in Table 3. The uncoated substates had an Ra roughness around 50 μm, as intended. The roughness of the coated surfaces was much higher and it was more evident in the TiCN coated surfaces. Similar values of both Ra and Rq were found for the TiCN and TiSiCN coatings.
A negative value of Sk shows that the surface consists of many valleys, while a surface with a positive Sk contains mainly peaks and asperities. Therefore, a surface with positive Sk has a good tribological performance in lubrication conditions. Taking into account this observation, both investigated coatings exhibited a positive Sk, which is suitable for load bearing implants, which work in a corrosive environment. The uncoated substrate had an Sk close to 0, meaning that the surface was very flat. The TiCN surfaces had a high Sk, indicating many peaks and asperities, whereas Sicontaining coatings had a smaller value, indicating a decrease in peaks and asperities. The dependence of corrosion resistance on these parameters will be discussed below.

Hardness and Elastic Modulus
Nanoindentation is a widely used technique, and it is a powerful tool to investigate the nanomechanical properties of materials, such as hardness and elasticity [45]. In order to overcome the substrate and roughness effects on nanoindentation tests, many studies have pointed out the importance of two basic rules. They refer to the maximum penetration depth, which should not exceed 1/10 of the layer thickness, and should be at least 20 times higher than the average roughness of the indented surfaces [46,47]. Another important limitation that needs to be considered for a reliable nanoindentation test is represented by the tip geometry and radius used, which limits the

Roughness of Surface
The main surface texture parameters of the investigated surfaces are: R a : the arithmetic average of the roughness profile, R q : the root mean square average of the roughness profile, S k : skewness (according to the ISO 2517 standard [44]), and are presented in Table 3. The uncoated substates had an R a roughness around 50 µm, as intended. The roughness of the coated surfaces was much higher and it was more evident in the TiCN coated surfaces. Similar values of both R a and R q were found for the TiCN and TiSiCN coatings. A negative value of S k shows that the surface consists of many valleys, while a surface with a positive S k contains mainly peaks and asperities. Therefore, a surface with positive S k has a good tribological performance in lubrication conditions. Taking into account this observation, both investigated coatings exhibited a positive S k , which is suitable for load bearing implants, which work in a corrosive environment. The uncoated substrate had an S k close to 0, meaning that the surface was very flat. The TiCN surfaces had a high S k , indicating many peaks and asperities, whereas Si-containing coatings had a smaller value, indicating a decrease in peaks and asperities. The dependence of corrosion resistance on these parameters will be discussed below.

Hardness and Elastic Modulus
Nanoindentation is a widely used technique, and it is a powerful tool to investigate the nanomechanical properties of materials, such as hardness and elasticity [45]. In order to overcome the substrate and roughness effects on nanoindentation tests, many studies have pointed out the importance of two basic rules. They refer to the maximum penetration depth, which should not exceed 1/10 of the layer thickness, and should be at least 20 times higher than the average roughness of the indented surfaces [46,47]. Another important limitation that needs to be considered for a reliable nanoindentation test is represented by the tip geometry and radius used, which limits the good testing contact depths, hence, in our case, was necessary to obtain contact depths higher than 40 nm [48].
In view of these testing conditions, the hardness and reduced modulus of the investigated sample coatings were determined according to the Oliver-Pharr method using similar force-displacement curves to the ones presented in Figure 3a. As it can be seen, these curves were obtained in both elastoplastic (CoCr substrate) and nearly elastic regimes (TiCN and TiSiCN coatings) [49], which were used to obtain the mechanical parameters of the samples. Note that all experimental force-displacement curves were qualitatively similar for the TiCN and TiSiCN coatings, providing indentation depth values smaller than 130 nm and exhibiting well-defined edges of the residual indent impressions (inset in Figure 3a). Moreover, the relatively low surface roughness in the investigated regions of the samples (~5 nm) was confirmed by performing several scans on 4 µm 2 areas using the same Berkovich tip. It has to be noted that this roughness corresponds to the flat areas between the microparticles found on the surface and it is therefore much smaller than the values reported for 4000 µm length surface scans. The resulting H = 7 ± 0.2 GPa and E = 97 ± 1.1 GPa values for the CoCr substrate are much smaller than the ones obtained for the investigated coatings. The hardness of the coatings significantly increased from 36.6 ± 2.9 GPa for the TiCN coating to 47.4 ± 1 GPa for TiSiCN, despite the small amounts of Si added into the TiCN layer structure (Figure 3b), while the effective modulus changed from 277 ± 8 GPa to 310 ± 2 GPa. The hardness enhancement originates from the grain size effect (H~d −1/2 , according to the Hall-Petch strengthening mechanism), since the TiSiCN coatings have smaller grain sizes. The hardness and modulus values were consistent with the previous results [50,51]. This implies that the condition H/E > 0.1 [52] was fulfilled for both coatings, as the calculated values for TiCN and TiSiCN were about 0.13 and 0.15, respectively. As the TiCN structure was modified, the sharp increase in the H 3 /E 2 ratio (resistance to plastic deformation) from 0.63 to 1.1, testifies to the superior toughness of the TiSiCN layer [53], and it may be ascribed to the beneficial effect of the Si content. The H 3 /E 2 parameter was considered to be a parameter sensitive to the tribological and corrosion properties of the materials [54].
Coatings 2020, 10, x FOR PEER REVIEW 7 of 17 good testing contact depths, hence, in our case, was necessary to obtain contact depths higher than 40 nm [48].
In view of these testing conditions, the hardness and reduced modulus of the investigated sample coatings were determined according to the Oliver-Pharr method using similar forcedisplacement curves to the ones presented in Figure 3a. As it can be seen, these curves were obtained in both elastoplastic (CoCr substrate) and nearly elastic regimes (TiCN and TiSiCN coatings) [49], which were used to obtain the mechanical parameters of the samples. Note that all experimental force-displacement curves were qualitatively similar for the TiCN and TiSiCN coatings, providing indentation depth values smaller than 130 nm and exhibiting well-defined edges of the residual indent impressions (inset in Figure 3a). Moreover, the relatively low surface roughness in the investigated regions of the samples (~5 nm) was confirmed by performing several scans on 4 μm 2 areas using the same Berkovich tip. It has to be noted that this roughness corresponds to the flat areas between the microparticles found on the surface and it is therefore much smaller than the values reported for 4000 μm length surface scans. The resulting H = 7 ± 0.2 GPa and E = 97 ± 1.1 GPa values for the CoCr substrate are much smaller than the ones obtained for the investigated coatings. The hardness of the coatings significantly increased from 36.6 ± 2.9 GPa for the TiCN coating to 47.4 ± 1 GPa for TiSiCN, despite the small amounts of Si added into the TiCN layer structure (Figure 3b), while the effective modulus changed from 277 ± 8 GPa to 310 ± 2 GPa. The hardness enhancement originates from the grain size effect (H~d −1/2 , according to the Hall-Petch strengthening mechanism), since the TiSiCN coatings have smaller grain sizes. The hardness and modulus values were consistent with the previous results [50,51]. This implies that the condition H/E > 0.1 [52] was fulfilled for both coatings, as the calculated values for TiCN and TiSiCN were about 0.13 and 0.15, respectively. As the TiCN structure was modified, the sharp increase in the H 3 /E 2 ratio (resistance to plastic deformation) from 0.63 to 1.1, testifies to the superior toughness of the TiSiCN layer [53], and it may be ascribed to the beneficial effect of the Si content. The H 3 /E 2 parameter was considered to be a parameter sensitive to the tribological and corrosion properties of the materials [54].

Potentiodynamic Polarization
The open circuit potential (EOC) evolution after 1 h of immersion and the potentiodynamic curves of the investigated systems are presented in Figure 4. During the immersion, the EOC of the TiCN and TiSiCN thin films showed a steady evolution, while the substrate slightly changed its value for half of the time, reaching a stable evolution at the end of the test. The TiSiCN coatings exhibited a positive    Table 4 (all the presented potentials are relative to the SCE value). Polarization resistance (Rp) was determined from linear polarization measurements as the slope of the linear region of the ΔE-Δi curve near Ecorr. Corrosion potential (Ecorr), anodic (βa) and cathodic (βc) slopes were estimated from Tafel plots. The corrosion current density was also calculated based on one form of the Stern-Geary equation (Equation (1)), based on previously determined parameters.
considering the Ecorr parameter, the most electropositive value was demonstrated by the TiSiCN coating (Ecorr TiSiCN = −14 mV), as well as the smallest value of corrosion current density (icorr TiSiCN = 49.6 nAcm −2 ). Taking into account the polarization resistance, the highest value was observed in the case of the CoCr substrate, closely followed by the TiSiCN coating (Rp TiSiCN = 425 kΩ•cm 2 ).

Electrochemical Impedance Spectroscopy (EIS)
It was observed that the electrochemical performance of the investigated specimens was different as a function of their composition. For comparison, Nyquist and Bode plots for the investigated specimens are presented in Figure 5.
The electrochemical parameters were obtained by fitting the data with an equivalent circuit, which took into consideration the phenomenon at the interface of each investigated system with the testing electrolyte (inset Figure 5). Rel represents the electrolyte resistance, CPElayer represents the coating capacitance, Rpore represents the resistance associated with the current flow through the pores generated by the coatings' defects and CPEdl is a double layer capacitance in parallel with a charge transfer resistance -Rct. CPE was used instead of a capacitor due to the non-ideal character of the working electrode. The physical interpretation of a circuit that has a constant phase element (for a better quality fit) depends on the value of α. If the α parameter is 1, then the CPE can be modeled as a capacitor. Since after the fitting, the α parameter showed values less than 1 in both cases (i.e., the α layer The main electrochemical parameters of the investigated specimens calculated based on the Tafel and potentiodynamic curves are presented in Table 4 (all the presented potentials are relative to the SCE value). Polarization resistance (R p ) was determined from linear polarization measurements as the slope of the linear region of the ∆E-∆i curve near E corr . Corrosion potential (E corr ), anodic (β a ) and cathodic (β c ) slopes were estimated from Tafel plots. The corrosion current density was also calculated based on one form of the Stern-Geary equation (Equation (1)), based on previously determined parameters.
considering the E corr parameter, the most electropositive value was demonstrated by the TiSiCN coating (E corr TiSiCN = −14 mV), as well as the smallest value of corrosion current density (i corr TiSiCN = 49.6 nA cm −2 ). Taking into account the polarization resistance, the highest value was observed in the case of the CoCr substrate, closely followed by the TiSiCN coating (R p TiSiCN = 425 kΩ·cm 2 ).

Electrochemical Impedance Spectroscopy (EIS)
It was observed that the electrochemical performance of the investigated specimens was different as a function of their composition. For comparison, Nyquist and Bode plots for the investigated specimens are presented in Figure 5.
The electrochemical parameters were obtained by fitting the data with an equivalent circuit, which took into consideration the phenomenon at the interface of each investigated system with the testing electrolyte (inset Figure 5). R el represents the electrolyte resistance, CPE layer represents the coating capacitance, R pore represents the resistance associated with the current flow through the pores generated by the coatings' defects and CPE dl is a double layer capacitance in parallel with a charge transfer resistance -R ct . CPE was used instead of a capacitor due to the non-ideal character of the working electrode. The physical interpretation of a circuit that has a constant phase element (for a better quality fit) depends on the value of α. If the α parameter is 1, then the CPE can be modeled as a capacitor. Since after the fitting, the α parameter showed values less than 1 in both cases (i.e., the α layer and α dl ), a CPE was used. This can be due to possible deviations from the ideal dielectric behavior and it is usually related to the surface inhomogeneity [55]. According to Hirschorn et al. [56], these deviations arise either from different properties along the surface of an electrode (e.g., roughness), or properties normal for the surface (e.g., thickness).
Coatings 2020, 10, x FOR PEER REVIEW 9 of 17 and αdl), a CPE was used. This can be due to possible deviations from the ideal dielectric behavior and it is usually related to the surface inhomogeneity [55]. According to Hirschorn et al. [56], these deviations arise either from different properties along the surface of an electrode (e.g., roughness), or properties normal for the surface (e.g., thickness). The electrochemical parameters of the investigated systems are presented in Table 5. It can be noticed the low value of the χ 2 parameter, which is an indication of an excellent agreement between the experimental data and those simulated by the equivalent circuit. Taking into account the fitting results for the investigated systems, it can be noticed that the highest pore-associated resistance was obtained for TiSiCN, while the CoCr substrate showed low values. It was stated that CoCr alloys form a passive layer at the surface, which is mainly based on Cr(III), and smaller amounts of Cr(OH)3, Co and Mo oxides [57]. The fitted values associated to the CoCr specimen showed that the formed layer is not as compact and protective as the TiCN and TiSiCN coatings.
The Qdl parameter, which is representative of the substrate-electrolyte interface, indicated a better protection of the deposited/formed layer in the following order: TiSiCN > TiCN > CoCr. Thus, the best protection after immersion in 90 % DMEM + 10 % FBS was observed for the TiSiCN coatings, with the best capacitive character, indicated by the low value of Qlayer and the highest value of Rct.
Considering αlayer, it can be observed that for CoCr, the CPE used for fitting the obtained data were the closest to a capacitor, since in this case, αlayer = 0.96. This could be due to the low roughness measured before the corrosion as compared to the other investigated specimens (Table 3). Similar α layer values were obtained for the coatings and the time-constant dispersion is ascribed to the similar values of roughness. Going deeper, at the interface between the coating and the substrate, another double layer is formed. Qdl and αdl can give an indication of the compactness of the deposited/formed layer and the electrolyte ingress through the defects, which can create pathways for the electrolyte to reach the substrate [58]. It can be observed that even though αdl was higher for TiCN, showing an almost defect-free structure, TiSiCN was the one showing better values of Qdl and Rct. The electrochemical parameters of the investigated systems are presented in Table 5. It can be noticed the low value of the χ 2 parameter, which is an indication of an excellent agreement between the experimental data and those simulated by the equivalent circuit. Taking into account the fitting results for the investigated systems, it can be noticed that the highest pore-associated resistance was obtained for TiSiCN, while the CoCr substrate showed low values. It was stated that CoCr alloys form a passive layer at the surface, which is mainly based on Cr(III), and smaller amounts of Cr(OH) 3 , Co and Mo oxides [57]. The fitted values associated to the CoCr specimen showed that the formed layer is not as compact and protective as the TiCN and TiSiCN coatings.
The Q dl parameter, which is representative of the substrate-electrolyte interface, indicated a better protection of the deposited/formed layer in the following order: TiSiCN > TiCN > CoCr. Thus, the best protection after immersion in 90 % DMEM + 10 % FBS was observed for the TiSiCN coatings, with the best capacitive character, indicated by the low value of Q layer and the highest value of R ct .
Considering α layer , it can be observed that for CoCr, the CPE used for fitting the obtained data were the closest to a capacitor, since in this case, α layer = 0.96. This could be due to the low roughness measured before the corrosion as compared to the other investigated specimens (Table 3). Similar α layer values were obtained for the coatings and the time-constant dispersion is ascribed to the similar values of roughness. Going deeper, at the interface between the coating and the substrate, another double layer is formed. Q dl and α dl can give an indication of the compactness of the deposited/formed layer and the electrolyte ingress through the defects, which can create pathways for the electrolyte to reach the substrate [58]. It can be observed that even though α dl was higher for TiCN, showing an almost defect-free structure, TiSiCN was the one showing better values of Q dl and R ct .

Morphology and Roughness after Corrosion
SEM images after the corrosion tests are presented in Figure 6. It is worth noting that the uncoated substrate was more affected by the corrosion than the coated surfaces. The destruction of the protection layer can be seen on the coated samples, after performing the corrosion tests. Regarding the TiCN coatings, there were various corrosion products on the surface, indicating that this surface was affected by the corrosive solution. Moreover, the coating was partially destroyed in some areas, with the CoCr substrate being visible. The TiSiCN surface also had corrosion products, but there were less compared with the TiCN surface, indicating better anticorrosive properties. All surfaces were affected by the corrosive solution, but TiSiCN was less damaged.

Morphology and Roughness after Corrosion
SEM images after the corrosion tests are presented in Figure 6. It is worth noting that the uncoated substrate was more affected by the corrosion than the coated surfaces. The destruction of the protection layer can be seen on the coated samples, after performing the corrosion tests. Regarding the TiCN coatings, there were various corrosion products on the surface, indicating that this surface was affected by the corrosive solution. Moreover, the coating was partially destroyed in some areas, with the CoCr substrate being visible. The TiSiCN surface also had corrosion products, but there were less compared with the TiCN surface, indicating better anticorrosive properties. All surfaces were affected by the corrosive solution, but TiSiCN was less damaged. The main roughness parameters of the investigated surfaces are presented in Table 6. Comparing these results with the values obtained before the corrosion tests (Table 3), it can be said that all surfaces were significantly damaged after the corrosion tests. The Ra of the uncoated substrates increased from 46.9 ± 5.9 nm to 1342.9 ± 192.4 nm. A significant increase in Ra roughness was found for the TiCN coatings after corrosion (15584.3 ± 7462.8 nm), indicating a major deterioration of the coatings after the corrosion tests. The TiSiCN coatings were also affected by the corrosive process, but they have finer irregularities than TiCN, demonstrating that the addition of Si led to an enhancement of anticorrosive properties. All surfaces showed a negative value of Sk after the corrosion tests, signifying that the surfaces were characterized by many valleys formed during the corrosive processes. The TiSiCN surface had fewer valleys that the TiCN surface, as shown by the smaller absolute value of Sk.

Discussions
The nature of the electrolyte, scanning rate, temperature, impurities, anode material and surface state of the samples are a part of the parameters, which can influence the electrochemical reactions. For example, the surface texture of the working electrode (investigated sample) is one of the most important parameters, which has an influence on the Tafel slopes, and consequently on the corrosion rate. Surface roughness has an important effect on general or pitting corrosion and the nucleation of metastable pitting. Skewness and kurtosis were used to identify the corrosion mechanism. Reid et al. have patented a method and apparatus for the identification of corrosion in metal objects and defining typical values of skewness and kurtosis for the identification of corrosion mechanism [59]. The main roughness parameters of the investigated surfaces are presented in Table 6. Comparing these results with the values obtained before the corrosion tests (Table 3), it can be said that all surfaces were significantly damaged after the corrosion tests. The R a of the uncoated substrates increased from 46.9 ± 5.9 nm to 1342.9 ± 192.4 nm. A significant increase in R a roughness was found for the TiCN coatings after corrosion (15584.3 ± 7462.8 nm), indicating a major deterioration of the coatings after the corrosion tests. The TiSiCN coatings were also affected by the corrosive process, but they have finer irregularities than TiCN, demonstrating that the addition of Si led to an enhancement of anticorrosive properties. All surfaces showed a negative value of S k after the corrosion tests, signifying that the surfaces were characterized by many valleys formed during the corrosive processes. The TiSiCN surface had fewer valleys that the TiCN surface, as shown by the smaller absolute value of S k .

Discussions
The nature of the electrolyte, scanning rate, temperature, impurities, anode material and surface state of the samples are a part of the parameters, which can influence the electrochemical reactions. For example, the surface texture of the working electrode (investigated sample) is one of the most important parameters, which has an influence on the Tafel slopes, and consequently on the corrosion rate. Surface roughness has an important effect on general or pitting corrosion and the nucleation of metastable pitting. Skewness and kurtosis were used to identify the corrosion mechanism. Reid et al. have patented a method and apparatus for the identification of corrosion in metal objects and defining typical values of skewness and kurtosis for the identification of corrosion mechanism [59]. The pitting mechanism appears when the S k < −2. In all our cases, the S k had values greater than −2, indicating that a general corrosion mechanism can be found for all investigated surfaces. Regarding the kurtosis, if the value is < 3, general corrosion can be observed, while for kurtosis > 3, pitting corrosion can be found. Considering this, it can be seen that for all investigated surfaces, a general corrosion mechanism has been identified. If the surface is rough, then a larger area could contribute to the increase in the corrosion current or to the corrosion rate. Therefore, a decreased surface roughness will lead to a better corrosion resistance [60]. According to this statement, the TiCN-coated surface was rougher than TiSiCN, and this is probably the reason for its better corrosion resistance. For the uncoated CoCr, the roughness is not a factor which has a major influence on the corrosion resistance. Compared to the coated samples, the CoCr uncoated substrate had a smaller roughness, while its corrosion resistance was worse than for both coatings. Clearly, the surface roughness affects the corrosive behavior of materials (i.e., metals, alloys, coatings) and the nature of its effects (increase or decrease in the degree of corrosion) depends on the nature of the material.
To the best of our knowledge, there is no direct relation between hardness and corrosion resistance. Hard coatings can be subjected to surface microcracks, and then a localized penetration of corrosive solution will take place, leading to a galvanic cell, which accelerates the corrosive process. Hardness is important for load-bearing implants because a hardened material can have the ability to withstand wear. Taking into account the results of the present study, TiSiCN was harder than CoCr and the TiCN coatings and was more adequate for the proposed application.
For load-bearing implants, resistance to plastic deformation is an important factor and it can be described by the H 3 /E 2 ratio. Moreover, a material with a high H 3 /E 2 ratio resists plastic deformation during low load contact events and exhibits a higher yield strength [61,62]. It is also generally accepted that the H/E ratio can be considered an important indicator of a good wear resistance of the surface [63][64][65]. Thus, the improvement of the H/E ratio and, consequently, of the resistance to plastic deformation (H 3 /E 2 ratio) of the load-bearing implant may offer advantages, such as less surface damage and increased durability. In this study, the TiSiCN coatings have an H 3 /E 2 ratio equal to 1.1, which is higher than the one for TiCN (0.63), indicating that TiSiCN has a superior toughness and it can offer a better resistance to plastic deformation and good wear resistance.
The addition of Si to TiCN coatings leads to a grain refinement, and the crystallite size (d) was decreased to 16.4 nm in the case TiCN and to 14.6 nm in the case of TiSiCN. The formation of new defects, especially dislocations, is also responsible for the reduction in the crystallite size. The strain in the TiSiCN coatings (ε = 0.012) was lower compared to the TiCN coatings (ε = 0.053). The reason for this decrease could be attributed to different factors. One reason could be due to the addition of Si, which has atomic radii (0.111) smaller than that of Ti (0.146 nm), leading to a disorder of the crystal lattice, which is also evident by XRD diffraction (peaks were shifted when compared with the TiCN standard). The second reason could be attributed to amorphous phases in which Si can be found (Si, SiNx, SiCN) or C=C phases at grain boundaries, which are detrimental for crystallite development. TiSiCN has a higher C content than TiSiC. It is difficult to separate these factors and to know their contribution. However, the crystalline disorder becomes more pronounced by an increase in carbon content, which is also suggested by the decrease in the crystallite size and by the decrease in microstrain. Moreover, Franceschini et al. reported a strong dependence of stress on the nitrogen content in a-C:H films; at a low N content, the stress is high [66]. This effect is difficult to see in our coatings, because the N content is reduced after the addition of Si, but it is a minor reduction. This result can also have a major influence on the corrosion resistance of TiSiCN. This coating probably presents fewer defects and it is more compact than TiCN.
When the crystallite size decreases, the corrosion current density decreases and polarization resistance increases, which means that the corrosion resistance of the coatings increases with decreasing grain size. Thus, the TiSiCN coatings, which have the smallest crystalline size, were more resistant to corrosive attack. The dependence of corrosion resistance on crystallite size can be ascribed to the BOLS mechanism [67]. In the grain boundaries, there are undercoordinated atoms with lowered residual cohesive energy which possess high energy, these atoms exist in unstable states and an increase in their percent will lead to an increase of corrosion resistance [68]. This finding is also sustained by the strain ε value. In both cases, the strain was low, but after the addition of Si, the strain was significantly reduced. When the strain decreases, the corrosion resistance of the coatings increases. Thus, the correlation between high corrosion resistance and low strain and small crystallites can be explained in terms of the "bond-order-length-strength correlation mechanism", meaning that the undercoordinated atoms found on the surface or in grain boundaries take the responsibility of the good corrosion resistance. In the current paper, along with the addition of Si, the Q dl parameter was also decreasing, and this result could be due to a smaller crystallite size obtained by the TiSiCN coating. Thus, the decreased generation of defects, in the case of this coating, had a beneficial effect on the protective properties. It was shown that defects within a structure can cause localized corrosion at the coating-substrate interface, due to the electrolyte ingress [69]. In addition, R pore indicated that the resistivity of the electrolyte in the pores had the highest value in this case, which can be also correlated with the lack of defects. The α values ranged from about 0.87 to 0.90, and the deviation from an ideal capacitor was ascribed to differences in roughness, as was shown. The dependence between roughness, capacitance and associated α values was demonstrated [69], although there are also some other factors which can be influences, such as thickness and the dielectric constant of the material.

Conclusions
The present study aimed to find a new and improved possible solution for load-bearing implants. For this purpose, titanium-based carbonitrides with and without the addition of Si were investigated and compared. Both coatings obtained by the cathodic arc deposition method had an almost stoichiometric structure, being solid solution, which consisted of a mixture of TiC and TiN, with a face-centered cubic (FCC) structure. The crystallite size decreased with the addition of Si into the TiCN matrix, the crystallite size of TiCN was 16.4 nm, while for TiSiCN it was 14.6 nm. Both coated surfaces exhibited a uniform coverage, with some microparticles from the ion sputtering and ejection of the particles during the deposition process. After the addition of Si into TiCN, the R a roughness values decreased, indicating a beneficial effect of Si.
All investigated surfaces had positive skewness, which is adequate for load-bearing implants, which work in corrosive environments. The hardness of the TiCN coating was 36.6 ± 2.9 GPa and it was significantly increased to 47.4 ± 1 GPa when small amounts of Si were added into the TiCN layer structure, while the elastic modulus was increased from 277 ± 8 GPa to 310 ± 2 GPa. A significant increase in resistance to plastic deformation (H 3 /E 2 ratio) from 0.63 to 1.1 was found after the addition of Si into the TiCN matrix.
The most electropositive value of corrosion potential was found for the TiSiCN coating (−14 mV), as well as the smallest value of corrosion current density (49.6 nA cm 2 ), indicating good corrosion resistance. The TiSiCN coating exhibited the best protection after immersion in 90% DMEM + 10% FBS, the best capacitive character, indicated by the low value of Q dl , and the highest resistance through the pores generated by the defects of the coatings and the electrolyte ingress, indicated by R pore and R ct .
According to the conducted research, TiSiCN coatings have shown good mechanical properties and high corrosion resistance and are a good alternative for the coating of load-bearing implants.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.