Surface Roughness and Its Effect on Adhesion and Tribological Performance of Magnetron Sputtered Nitride Coatings

Abstract

Reports of the influence of surface roughness on the adhesion and tribological performance of contemporary nitride coatings with different layer designs are still scarce in the literature. Therefore, in this study, we evaluated the behavior of a single-layer TiAlN, a bilayer TiAlN/CN_x, and a nanolayer AlTiN/TiN coating. Coatings were deposited in an industrial magnetron sputtering unit on the substrates of EN 100Cr6 steel, prepared to four degrees of surface roughness (Sa = 10–550 nm). The coatings’ adhesion was determined by scratch tests performed perpendicular and parallel to the machining marks. Dry reciprocating sliding tests in air were employed to evaluate the coatings’ tribological behavior against an Al₂O₃ ball. Before and after the tests, coating properties were characterized by 3D profilometry, confocal microscopy, and energy dispersive spectroscopy. Deposition of all coatings significantly altered the surface topography and increased the roughness of the samples. No general rule could be established for the effect of surface roughness on tribological behavior and adhesion of different hard coatings. For very fine surface finishes the adhesion and tribological performance of TiAlN and TiAlN/CN_x coatings was independent of the surface roughness. For the roughest surfaces, a decrease in adhesion and an increase in the wear rate were observed. The AlTiN/TiN coating exhibited the largest sensitivity of adhesion to roughness and scratching direction. The coefficient of friction and wear rate increased when AlTiN/TiN roughness exceeded Sa ≈ 100 nm.

1. Introduction

The application of physical vapor-deposited (PVD) hard coatings in various fields of mechanical engineering is rapidly expanding. As such, the development of new and improved wear-resistant coating systems remains both a technological challenge and an active area of scientific research. Optimizing the design of a wear-resistant coating requires a comprehensive understanding of not only the parameters affecting the surface exploitation but also the manufacturing parameters of the part to be coated. This includes the type of machining processes applied in surface preparation, machining regimes, and the resulting surface roughness and texture before the coating deposition.

Wear-resistant PVD coatings are typically very hard and stiff materials [1] that, in tribo-contacts, behave differently from bulk materials. Under certain loads, the plastic deformation at the micro-scale is limited, leading to real contact between rough surfaces occurring in much smaller areas [2]. Consequently, the contact pressure in the asperities increases [2], which can also result in a rise in thermal loading on these asperities [3]. Thermal loading can degrade the mechanical properties of the coating, while increased pressure can lead to local coating cracking, fragmentation, and adhesive failures. Such conditions result in the coated system exhibiting different behavior compared to the same coating deposited on surfaces with different roughness [2,4].

In applications where mechanical loading of the coating system prevails, the effects of the surface roughness should be considered for their impact on the coating adhesion and tribological performance.

The protective effect of a thin ceramic coating relies strongly on its adhesion to the substrate material. Therefore, ensuring high coating-adhesion is of the utmost importance for the practical application of coatings on tools and machine components. The adhesion of thin ceramic coatings is greatly influenced by the surface roughness of the substrate material; coatings deposited on smooth surfaces exhibit better scratch adhesion [5]. Common studies examining the adhesion of PVD coatings typically focus on very smooth surfaces. However, applying these results to real components can be challenging and may lead to miscalculations in adhesion predictions. Polishing is a costly machining operation that is not justified in all cases of coating applications, especially in cases where mirror-surface finishes are not necessary. For some components, like additively manufactured metal tool parts, the reduction of surface roughness beneath some level is a technologically challenging task [6]. Accurate prediction of the coating adhesion is, thus, crucial for appropriate surface design and part–cost estimation. The effects of surface roughness on scratch adhesion were previously elaborated in great detail, but most of these studies concerned monolayer coatings, and predominantly the TiN coating deposited by PVD or CVD [7,8,9,10,11,12,13,14,15]. The relation between scratching direction and surface texture (machining marks) was studied only for TiN coatings in [12,13,14,16], but this influence has not been studied for contemporary coatings with different layer designs, and mechanical and tribological properties.

A lot of work has been done so far in theoretical modeling and simulation of the scratch-test process for the determination of the adhesion of different coated systems based on the coating and substrate properties [17,18,19,20]. However, to accurately predict the stress state in the coated system, these models have to include a vast number of parameters [20]. Besides the coating residual stresses, the other parameters that are difficult to incorporate into the models accurately are the substrate and coating topography, and the coating-growth defects [20,21,22]. Consequently, the precise determination of the scratch adhesion related to surface roughness still needs to be experimentally determined for every specific coating system.

The majority of tribological investigations of sliding contacts concern the coatings on very smooth surfaces where the effects of surface roughness and coating defects on coating performance are usually not determined [2,4]. On the other hand, when considering these effects, they are often evaluated using general rules. For example, it is commonly assumed that friction and wear increase with surface roughness. However, Harlin P. et al. have shown that the steady-state coefficients of friction (COF) of TiN and WC/C coatings do not vary considerably with roughness, unlike their wear rates. Additionally, with the change in surface roughness, these coatings exhibit different behavior in the running-in stage of the sliding process [23]. For the a-C coating reported in [2] and the a-C:H in [24], it has been shown that the wear rate increases with an increase in surface roughness. Svahn F. et al. have found that a-C coatings doped with Cr and W exhibit comparable trends of COF with roughness, but the W-doped coating shows lower sensitivity of wear with change in surface roughness [4]. For TiC/a-C nanocomposite coatings, it was found that the increase in coating nano-roughness leads to a decrease in wear rate and an increase in COF [25,26]. In testing the superlubricity of a-C: H coating, Wang I. et al. found that roughness had a larger influence than the layered design of a-C:H coating [27]. Overall, it was found that different coatings exhibited varying trends in COF and wear rate as surface roughness changes. Furthermore, their behavior depended on the range of the tested surface roughness [13]. A literature review of this field indicated that a limited number of papers have been published in this field and the data regarding the behavior of some contemporary PVD hard coatings are still scarce.

The purpose of this study is to reveal the effects of surface roughness on scratch adhesion, scratching direction, and tribological performance of different hard coatings. As such, this study provides novel information on the behavior of contemporary PVD coatings of different layer designs. This investigation concerned a single-layer TiAlN coating, a bilayer TiAlN/CN_x coating, and a nanolayer (nl) AlTiN/TiN coating, all deposited in industrial magnetron sputtering systems.

2. Experiment

A quenched and tempered EN 100Cr6 steel (55 HRC) was used to produce samples in the shape of Φ 20 × 5 mm disks. To obtain samples with different surface roughness before the coating deposition, the samples were subjected to conventional flat grinding, standard metallographic procedures of grinding with SiC sandpapers, and diamond-paste polishing (3 μm). The roughest surface was produced by conventional flat grinding procedures using regimes for rough grinding. Based on the applied treatment and obtained surface roughness, the samples were divided into four groups: coarse-ground, 400-grit, 1500-grit, and polished, as shown in

Table 1. Applied surface treatment and sample designation.

Surface Treatment Method	Sample Designation
Conventional flat grinding	Coarse-ground
Grinding up to 400 grit	400-grit
Grinding up to 1500 grit	1500-grit
Grinding up to 2000 grit + polishing with 3 µm diamond paste	Polished

. For each coating, only one sample of a specific surface treatment was prepared.

Before deposition, the samples were cleaned in ultrasonic baths with detergents and deionized water, and dried in hot air. Three types of coating were deposited using two industrial deposition units with closed-field unbalanced magnetron sputtering systems equipped with four targets (cathodes). The single-layer TiAlN coating was deposited using the CC800/7 (CemeCon), while the bilayer TiAlN/CN_x and nanolayer AlTiN/TiN coatings were deposited using the CC800/9 (CemeCon) system. The layered design of the investigated coatings is illustrated in Figure 1. Details of the employed deposition system can be found in Refs. [28,29]. In all cases, the substrates were first heated to 450 °C, and then the working gases were introduced, followed by substrate-ion etching. The TiAlN coating was produced using four TiAl targets, while the nanolayer AlTiN/TiN coating was deposited using two TiAl and two Ti targets. In both cases, all four targets were engaged during the whole deposition process. On the other hand, the TiAlN/CN_x coating was deposited, engaging three and one targets at a time. For the deposition of the bottom layer of the TiAlN/CN_x coating, three TiAl-targets were engaged, and for the deposition of the top CN_x layer, one C-target was engaged. In all cases, the TiAl targets contained Al plugs. During all deposition processes, the temperature was kept at 450 °C, and the samples were subjected to 2-fold rotation. Other relevant deposition parameters for different coatings are given in

Table 2. Deposition parameters.

Deposition Unit	Coating	Target Power [kW]			DC Bias [V]	Gas Flow Rates [sccm]			Average Deposition Rate [μm/h]
		TiAl	Ti	C		Ar	N	Kr
CC800/7	TiAlN	8	-	-	−95	100	≈160	80	1.78
CC800/9	TiAlN/CNx	9.5	-	4	−90/−80	100/150	≈170/≈70	90	1.54/0.36
CC800/9	nl-AlTiN/TiN	9.9	4	-	−90	135	≈130	110	2.28

.

The coating thickness was determined by the standard ball-cratering method by repeating it three times. Since all samples during the PVD deposition are subjected to the same conditions, it is expected that the coating thickness does not vary significantly between the samples of different surface roughness. Therefore, the ball-cratering was performed only on the polished-coated samples. The mechanical properties of the coatings were evaluated also on polished samples by the nanoindentation technique using H100C (Fisherscope, Sindelfingen, Germany) device and employing a Vickers prism. The Poisson’s ratio of 0.25 was chosen for indentation modulus estimation, while the Oliver and Phar method was used for the calculation of the coating’s hardness and elastic modulus. For the TiAlN and nl-AlTiN/TiN coatings, the indentations were performed using loads of 50, 100, 200, 300, 500, and 1000 mN. Given that the TiAlN/CN_x is a bilayer coating, additional measurements were performed with loads of 5, 10, and 25 mN. In all cases, measurements performed with loads of 5, 10, 25, 50, 100, and 200 mN were performed with repetitions up to 12 times, and the others were performed with three repetitions.

Surface topography before and after the deposition, on every sample, was acquired by tactile 3D profilometer Talysurf (Taylor Hobson, Leicester, UK). The scans were taken on an area of 1 mm² with high lateral and height resolution (2 µm in the x- and y-directions, and 10 nm in the z-direction). Surface roughness parameters were analyzed and calculated using Scanning Probe Image Processor SPIP (Image metrology) 6.2.0. software. The coating chemical composition was evaluated by a JSM−7600F (JEOL, Tokyo, Japan) scanning electron microscope equipped with energy dispersive spectroscopy (EDS). Measurements were performed on the top of the coated layers, with three repetitions per coating.

The scratch adhesion of the coated samples was evaluated using the Revetest (CSM Instruments, Peseux, Switzerland) instrument. A progressive load test was performed up to a final value of 100 and 150 N over a distance of 3 and 4.5 mm, respectively. The tests were performed by a Rockwell-type indenter with a tip radius of 200 µm and employing a sliding speed of 6 mm/min. The scratch test measurements were taken in the direction parallel and perpendicular to the machining marks. All the tests were repeated three to five times on one sample of a specific surface roughness. Critical forces leading to typical coating failures were identified by visual inspection of the generated scratch tracks by monitoring acoustic signals and by changes in friction coefficient. In this work, we concentrated only on the correlation of the three critical forces used for the determination of coating adhesion: Lc₁—first cohesive spallation (chipping); Lc₂—first adhesive failure; and Lc₃—total coating delamination.

To obtain the coefficient of friction and coatings’ wear resistance, dry reciprocating sliding tests in air were conducted using a standard tribometer (CSM). During the tests, the Al₂O₃ counter-ball (Φ 6 mm) was dead-loaded against the coated surface with a load of 5 N sliding with 5 cm/s velocity. A stroke length of 5 mm was chosen, and the test was performed for 1000 and 2000 sliding cycles without repetitions. Al₂O₃ was selected as a counter material because of its high hardness, wear resistance, and high inertness. As such, consistent and reproducible results of abrasive wear were ensured. Additionally, it is a standard material in coating wear-testing which allows easy comparison with other investigations from the field. During these tests, the coatings were not worn out. Images of the obtained tribo-tracks and counter-balls were taken by a confocal microscope (CFM) Axio 700 (Zeiss, Jena, Germany). The topography of the tribo-tracks was measured in the middle of the track by 3D profilometry on the area of 2 × 1 mm. These measurements served to determine the worn volume and calculation of wear rates.

3. Results and Discussion

3.1. Coating Thickness and Chemical Composition

The properties of the investigated coatings tested on polished samples are presented in Table 3. Results of the chemical compositional analysis show that the TiAlN layer in the TiAlN coating, and in the TiAlN/CN_x coating, had a comparable Al:TiAl ratio. The nl-AlTiN/TiN coating contained more Al in the layers, and had a higher Al:TiAl ratio than the other two investigated coatings.

The TiAlN/CN_x and nl-AlTiN/TiN coatings had the same thicknesses, while TiAlN was somewhat thinner. It has to be noted that the CN_x top layer of the TiAlN/CN_x coating was six times thinner than the underlying TiAlN layer. Details of the microstructure of TiAlN and TiAlN/CN_x coatings can be found in Ref. [29], while for nl-AlTiN/TiN these can be found in the investigation from Ref. [30].

Table 3 presents the average values of the coefficient of friction (COF) determined after the running-in stage of the investigated coatings sliding against the Al₂O₃ counter-ball on polished samples. The values of COF obtained for different coatings differ considerably. High and intermediate values obtained for TiAlN and nl-AlTiN/TiN coatings are in line with those usually obtained for these kinds of nitride coatings reported in the literature for TiAlN [31,32], AlTiN [33,34], and TiN/AlTiN coatings [35]. A low value of COF obtained for the TiAlN/CN_x coating is also typical for this specific kind of diamond-like carbon (DLC) coating produced by magnetron sputtering [36,37]. Considering the range of the COFs obtained for the concerned coatings (0.22–0.87), this investigation covers the most typical range of hard coatings.

Table 3. Coating thickness, chemical composition, and coefficient of friction.

Coating	Properties			Chemical Composition
	Thickness [µm]	Layer Thickness	Average Value of COF	Ti [wt %]	Al [wt %]	C [wt %]	N [wt %]	Al:TiAl
TiAlN	4.9 ± 0.5	4.9	0.87	23.7	23.6	-	52.7	49.9
TiAlN/CNx	6.3 ± 0.6	CNx layer 0.9 µm TiAlN layer 5.4 µm	0.22	10.8	9.7	49.3	30.2	47.3
nl-AlTiN/TiN	6.3 ± 0.6	~10 nm (double layer period 1) [38]	0.44	20.5	27.5	-	51.9	57.2

¹ A thickness of two constituting layers (AlTiN and TiN) in a nanolayer coating.

Table 3. Coating thickness, chemical composition, and coefficient of friction.

Coating	Properties			Chemical Composition
	Thickness [µm]	Layer Thickness	Average Value of COF	Ti [wt %]	Al [wt %]	C [wt %]	N [wt %]	Al:TiAl
TiAlN	4.9 ± 0.5	4.9	0.87	23.7	23.6	-	52.7	49.9
TiAlN/CNx	6.3 ± 0.6	CNx layer 0.9 µm TiAlN layer 5.4 µm	0.22	10.8	9.7	49.3	30.2	47.3
nl-AlTiN/TiN	6.3 ± 0.6	~10 nm (double layer period 1) [38]	0.44	20.5	27.5	-	51.9	57.2

¹ A thickness of two constituting layers (AlTiN and TiN) in a nanolayer coating.

3.2. Mechanical Properties

Figure 2 presents the mechanical properties of investigated coatings in relation to the indentation load, tested on polished samples. At lower indentation loads, all investigated layers exhibited a decrease in hardness. However, for both the TiAlN and nl-AlTiN/TiN coatings, the error bars obtained for the measurements with 50, 100, and 200 mN overlap. This suggests that the differences between the measurements conducted using these loads were insignificant. Generally, among investigated coatings, the TiAlN had the highest hardness (3020 HV0.005), nl-AlTiN/TiN had the second (2250 HV0.005), and TiAlN/CN_x was the softest (1270 HV0.005). In the indentation tests with the lowest loads (5, 10, 25 mN), the TiAlN/CN_x coating exhibited quite low values of hardness. At these loads, the indenter penetrated less than 10% in the CN_x layer, which means that the obtained values corresponded to the hardness of the top layer alone (~1200 HV0.0005). The hardness of ~1800 HV0.02 measured at a higher load (200 mN) only partly corresponded to the hardness of the underlying TiAlN layer of the TiAlN/CN_x coating. The TiAlN layer in TiAlN/CN_x coating can, indeed, have a lower hardness than the single-layer TiAlN coating. This is attributed to the lower ionization and weaker ion bombardment effects during the deposition process [28]. Such conditions occurred due to the coating deposition from only three cathodes (Section 2). The lower hardness of nl-AlTiN/TiN coating, compared to the one produced by the same parameters reported in [39], is attributed to the lower degree of substrate rotation (2-fold) during the deposition process [40]. The modulus of elasticity (E) of TiAlN and nl-AlTiN/TiN coatings is comparable and notably higher than that of the TiAlN/CN_x coating. This implies that these coatings have somewhat higher stiffness. On the other hand, the TiAlN coating exhibited higher n_IT values, which suggests that this coating exhibits more elastic behavior under load than the nl-AlTiN/TiN and TiAlN/CN_x coatings.

3.3. Surface Topography and Surface Roughness

Representative topography images of samples before and after the coating deposition are shown in Figure 3. The coating deposition altered the sample’s surface topography obtained by the machining processes employed in substrate preparation. Besides the machining marks that appear in the form of micro- and nano-grooves, the coated surfaces are dominated by features in the shape of protrusions and dimples. These features are the coating growth defects that typically form in PVD processes, especially in industrial batches [41]. Protruding defects are known as nodular defects, and the dimples are known as crater defects [41]. In most cases, both types of defect are through-thickness, but they can also stretch from a certain depth of the coating layer up to its top. The diameter and height of these defects above the free surface vary from a submicron level to a few tens of micrometer [29,41]. Such surface features affect not just the coating topography but also the coating behavior during exploitation, such as its cracking under load, adhesion, wear, and coefficient of friction in the sliding contact [42].

Surface roughness parameters calculated for the surfaces before and after the coating deposition are presented in

Table 4. Surface roughness parameters and number of defects.

Surface Finish	Coating	Before/After Deposition	Sa [nm]	Sq [nm]	Ssk	Sku	S10z [nm]	Sdq	Density of Islands [mm−2]	Mean Height of Islands [µm]	Mean Surface of Islands [µm2]
polished	TiAlN	before	11.1	14.4	0.2	8.0	420	0.012	-	-	-
		after	71.3	236.7	10.0	136.2	5615	0.079	229	0.522	62.9
	TiAlN/CNx	before	10.6	14.3	0.3	9.3	390	0.013	-	-	-
		after	65.3	174.8	9.0	156.5	5163	0.064	189	0.456	49.1
	nl-AlTiN/TiN	before	10.6	16.9	−3.9	73.1	675	0.018	-	-	-
		after	39.2	116.1	11.8	231.6	4499	0.049	126	-	37.8
1500-grit	TiAlN	before	19.3	25.3	0.1	6.4	477	0.018	-	-	-
		after	58.2	217.9	15.2	326.5	7063	0.063	176	0.613	54.5
	TiAlN/CNx	before	18.0	24.7	0.9	9.2	598	0.016	-	-	-
		after	108.8	266.2	8.5	115.1	6271	0.090	365	0.534	58.3
	nl-AlTiN/TiN	before	15.0	19.6	0.0	9.3	595	0.014	-	-	-
		after	40.2	119.7	7.2	265.5	5523	0.048	134	0.362	35.2
400-grit	TiAlN	before	48.5	64.8	−0.5	10.9	1586	0.028	-	-	-
		after	64.8	116.5	8.3	188.6	3781	0.042	69	0.522	42.1
	TiAlN/CNx	before	46.2	64.0	−1.1	10.4	1222	0.024	-	-	-
		after	99.8	216.3	6.6	70.2	4815	0.080	332	0.467	50.7
	nl-AlTiN/TiN	before	41.8	55.3	0.1	5.6	968	0.021	-	-	-
		after	97.8	270.3	9.2	117.0	6890	0.083	243	0.693	62.9
Coarse ground	TiAlN	before	525.7	682.6	−0.6	3.8	4699	0.142	-	-	-
		after	571.4	733.6	−0.5	3.6	6339	0.148	-	-	-
	TiAlN/CNx	before	467.1	598.3	−0.3	3.5	4732	0.137	-	-	-
		after	493.2	640.7	−0.2	3.9	5956	0.150	-	-	-
	nl-AlTiN/TiN	before	545.2	692.6	−0.2	3.2	5107	0.145	-	-	-
		after	545.9	738.3	0.9	8.8	9948	0.164	-	-	-

. This table also contains the quantitative parameters regarding the coatings’ growth defects. The value of the Sa parameter indicates that the surfaces of polished, 1500-grit, and 400-grit substrates belong to the group of very fine surface finishes, while the coarse-ground substrate is representative of a fine surface finish achieved by the machining process.

The arithmetical roughness (Sa) values and ten-point height (S10z) parameters calculated for all coatings on polished, 1500-grit, and 400-grit samples indicate an increase in surface roughness after the coating deposition. Additionally, in these cases the skewness (Ssk) parameter increased from negative, or approx. zero values, to higher positive values. This means that the surface topography changed from valley-dominated or neutral surfaces to peak-dominated surfaces. The kurtosis (Sku) parameter also increased considerably after the deposition of all coatings. This means that in all these cases the surface became less random, and this is typical for surfaces with uneven and sporadic peaks over. The density of growth defects varied from sample to sample, and none of the investigated coatings displayed a pronounced trend. The density of nodular defects is usually 10- or 20-times larger than the density of crater defects. All these observations indicate that the nodular growth defects had a dominant influence on the change of the surface topography and on the surface roughness parameters. For polished and 1500-grit samples the Sa parameter directly correlates with the density of the defects. The root mean square gradient (Sdq) parameter indicates the steepness of the surfaces. This parameter increased significantly after the coating deposition process, which is linked with the formation of growth defects. In the cases of TiAlN and TiAlN/CN_x coatings, the increase of this parameter was considerably more significant than that for the nl-AlTiN/TiN coating.

The coarse-ground samples did not exhibit any notable change in surface roughness parameters, except the S10z parameter. Before the deposition of the coatings, these samples already had high ridges and deep grooves, whose dimensions were comparable to those of the growth defects. Therefore, in these cases the growth defects could not affect the Sa, Ssk, and Sku surface roughness parameters. The S10z parameter was affected by the growth defects by introducing additional protrusions on already very high surface ridges.

The change of the Sa parameter caused by the deposition process is demonstrated by the diagram shown in Figure 4. Surfaces were prepared to a different degree of surface roughness, with a different Sa parameter, after the coating deposition increased the values of this parameter. Once again, such behavior is linked with the effect of the growth defects on the surface roughness of concerned coatings. After the deposition of TiAlN coating on the substrates with different initial roughness (1500-grit, 400-grit, polished), the roughness of the coated samples converged to a specific range of values (Sa = 58–71 nm). Such an effect was less pronounced for TiAlN/CN_x and nl-AlTiN/TiN coatings, but still observable for a few samples. However, to confirm this finding as certain, a more detailed examination performed on a larger number of samples should be done. All the above-mentioned observations indicate that the coating roughness can considerably differ from the roughness of the coating–substrate interface, which is important to consider in the evaluation of the coating adhesive failure and wear. The roughest samples (coarse-ground) displayed the least changes in surface roughness parameters. This is explained by the dimensions of their initial grooves and ridges that are higher than the sizes of growth defects. It’s noteworthy that, in most of the investigations that concerned the effect of surface roughness on the coating scratch adhesion and tribological performance [2,8,16], the coating almost replicated the substrate roughness (Sa parameter).

3.4. Scratch Adhesion

Representative scratch tracks obtained on polished samples are presented in Figure 5. All investigated coatings display different behavior in scratch testing. The coating failure modes observed in the scratch tracks are classified according to the definitions of the EN ISO 20502:2016 standard [43] and explained in what follows. The TiAlN coating first exhibits longitudinal cracks at the edges of the tracks. This is followed by cohesive spallation (chipping) along the track borders and bucking failures. Short ductile perforation occurs just before the continuous ductile coating perforation typical for complete coating removal. The TiAlN/CN_x starts to fail by longitudinal and forward chevron cracks at the edges of the tracks, and there is a minor appearance of cohesive spallation. This is followed by frequent gross interfacial shell-shaped spallation and ductile perforations. The complete coating removal occurs by the continuous ductile coating perforation mechanism. The nl-AlTiN/TiN first exhibits the formation of forward chevron cracks at the borders of the tracks. This is followed by rare occurrences of cohesive and interfacial spallation along the scratch-track borders. In later stages, ductile perforation of the coating becomes frequent and it is followed by continuous ductile coating perforation. The obtained results indicate that, according to the classification of Bull S. J., all investigated coatings exhibit ductile failure modes [44]. In comparison with TiAlN, a somewhat earlier appearance of adhesive failures (shell-shaped spallation) on the TiAlN/CN_x coating is linked to differences in the deposition process. The TiAlN layer of the TiAlN/CN_x coating is produced with only three active targets of a four-target close field unbalanced magnetron sputtering system. In this configuration, plasma does not overlap the whole rotating sample holder, and consequently reduces the ion bombardment of the growing coating layer. As such, these coatings exhibit a somewhat earlier formation of adhesive failures.

Samples with larger surface roughness exhibited the same failure modes typical for each coating, the representative scratch tracks on rough surfaces are shown in Supplementary Figure S1. However, in certain cases, the machining marks on the surface hampered the visual detection of the first cracks. Therefore, the quantitative values of critical force that led to the formation of the first cracks were not considered in this study. None of the coatings with higher roughness (1500-grit, 400-grit, coarse-ground) exhibited extensive or gross spallation around the scratch track, which is related to the surface grooves and ridges.

Diagrams in Figure 6 present the values of critical normal forces obtained for all investigated samples tested perpendicular and parallel to the machining marks. All Investigated coatings were characterized by high scratch adhesion. The nl-AlTiN/TiN coating exhibited the highest adhesion, the TiAlN/CN_x had intermediate adhesion, while the TiAlN coating exhibited the lowest, performing just slightly below the values obtained for the TiAlN/CN_x coating. This trend was observed for all critical forces concerned (Lc₁, Lc₂ and Lc₃).

Investigated coatings displayed different behavior regarding the effect of surface roughness and scratching direction on the values of critical forces. For TiAlN and TiAlN/CN_x coatings, the increase in substrate roughness, from polished to 400-grit, did not significantly affect the critical forces. But the further increase in roughness, up to the level of the coarse-ground sample (Sa ≈ 450 nm), prominently reduced the critical force needed to induce the coating’s failure, which was in line with the results for TiN from other investigations [9,12]. This effect was more pronounced for both coatings when the scratch test was performed perpendicular to the machining marks. The nl-AlTiN/TiN coating displayed a much more pronounced effect from the surface roughness on the critical forces. For this coating, the increase in roughness altered the onset of coating chipping spallation (Lc₁) and the first adhesive failure (Lc₂) to higher forces. However, this was not so pronounced for the onset of the complete coating removal (Lc₃). This means that for the nl-AlTiN/TiN coating, the increase in surface roughness affects the coating cracking pattern more than its adhesive strength to the substrate. The nl-AlTiN/TiN coating exhibited higher adhesion on the 1500-grit and 400-grit than on the polished samples. Such a result is contrary to most investigations from the field [9,10,16] but agrees well with our previously published results for the TiN coating deposited by ion beam-assisted deposition (IBAD) [13]. However, it is interesting that, among the investigated coatings, only the coating with a nanolayer design displays such behavior. The adhesion of the nl-AlTiN/TiN coating on the coarse-ground sample is significantly lower than on the other samples. Generally, this means that the adhesion of the nl-AlTiN/TiN coating is highly sensitive to the change in the substrate surface-roughness.

In all tested cases the values of critical forces were higher for the scratch tests performed parallel to the machining marks, which is in line with observations for TiN coating evaluated in other investigations [12,13,14,16]. Such behavior can be explained by analyzing the stress field induced around the indenter, explained in [17,20], together with the effects of residual stresses in PVD coatings on rough-coated surfaces elaborated in [45]. Accordingly, it can be postulated that, in this problem, three groups of stresses exist that can cause the coating delamination. The first is the effect of the stresses caused by material pile-up in front of the scratching indenter which causes the coating buckling failure [17,20]. The second is the stress concentration on the tips of the substrate and coating surface ridges, which can cause crack formation, and lead to coating adhesive failure [13,45]. The combined effect of the coating compressive residual stresses and the tensile stresses, generated in the coating by the material pile-up process, induce easier interfacial coating failure on the tips of the substrate ridges of rough samples. The third is the tearing of the coating material on the bottom of the valleys, which occurs due to the drag stresses induced behind the indenter during scratching [17,21]. Therefore, in scratching perpendicular to machining marks, all three mentioned effects manifest. In parallel scratching, only the first and the second effects occur, and consequently the coating adhesive failure is postponed to higher forces.

The difference in critical forces between perpendicular and parallel scratching increases with surface roughness. A somewhat larger difference in critical forces between the tests performed in perpendicular and parallel directions is observed for TiAlN and nl-AlTiN/TiN coatings. The TiAlN/CN_x coating has the lowest sensitivity of critical forces on the scratching direction. A reason for this could lie in the fact that this coating has both the lowest COF and modulus of elasticity among the investigated coatings. Low COF can reduce the shear stress that is transferred from the scratching indenter, through the coating, to the coating/substrate interface [10,17]. As such, the effect of the stresses from the “third group” is reduced. Additionally, a low E value of the CN_x layer in the TiAlN/CN_x coating, governs lower localization of the stresses around contact points and long elastic strain to failure [1,19]. Generally, such behavior of the TiAlN/CN_x coating is beneficial for its practical tribological application.

3.5. Tribological Properties in Reciprocating Sliding

Figure 7 presents the topography images of wear tracks in the central region after 2000 cycle reciprocating sliding tests, with the corresponding CFM images of the counter-ball surfaces after the tests. By analyzing the wear tracks, we found that none of the investigated coatings was worn out in any of the investigated cases, but the coatings displayed quite different behavior. The CFM images of the wear track central regions for the 2000 cycle tests are given in Figure S2.

The topography and CFM images of the wear tracks indicate that the TiAlN coating displays a dominant abrasive wear mechanism. These wear tracks are smooth and characterized by very fine small grooves stretching along the wear track. In the wear tracks of polished, 1500-grit and 400-grit samples, the machining marks are not visible. On the coarse-ground sample, the machining marks are still observable in the track (Figure S2), and this track is the widest. The cross sections of TiAlN wear-tracks have arc-like shapes, their profiles follow the counter-ball profile, except the track on the coarse-ground surface, which is flatter. The morphology of the counter-ball surfaces is mostly comparable between the tests for the TiAlN coating. Only the one worn against the coarse ground sample has a larger worn area and coverage by wear products.

The wear tracks of the nl-AlTiN/TiN coating are smooth, with much more pronounced grooves that stretch along the wear tracks. In the middle of the wear track, a high and wide flat protrusion forms in most cases. The profile of the wear track does not follow the counter-ball profile. Such wear track topographies indicate the operation of abrasive and adhesive wear mechanisms. As the sample roughness increases, the width of the adhesive protrusion inside the track decreases, bringing the counter-ball into better contact with the wear track. This effect is linked to the fact that, on rougher surfaces, the wear debris is easily pushed into the grooves (particle hiding effect). For this coating, the width of the wear tracks considerably increases with an increase in surface roughness. In the case of nl-AlTiN/TiN coating, the appearance of the counter-balls differs significantly between the tests on samples of different roughness, which is linked to the formation of a wide protrusion in the middle of the wear track.

The TiAlN/CN_x coating exhibited the narrowest tracks with small grooves and/or protrusions that stretch along the track. In this case, the coating underwent mostly abrasive wear, with minor adhesion. Inside the wear tracks the machining marks can be distinguished in the cases of 400-grit and coarse-ground samples. This indicates much lower wear than in the case of TiAlN and nl-AlTiN/TiN coatings. Two-dimensional profiles extracted from 3D topographic images indicate that the CN_x layer of the TiAlN/CN_x coating is not worn off. The behavior of the TiAlN/CN_x coating is comparable with the TiAlN coating. The counter-ball sliding against the TiAlN/CN_x coating exhibits only minor wear, which is most pronounced in the case of the coarse-ground sample.

Plots of the friction coefficients are given in the Supplementary Figure S3. The average values of COF were calculated for the period after the running-in stage, i.e., from the 200th to the last sliding cycle. The average COF and wear rate for the investigated coatings tested for 1000 and 2000 sliding cycles are shown in Figure 8. Among the investigated coatings, TiAlN had the highest value of average COF and the highest coefficient of wear, while TiAlN/CN_x exhibited the lowest values of both parameters. The lowest wear of TiAlN/CN_x coating resulted from the low shear-strength and solid, lubricating nature of CN_x graphite-like tribo-film [46,47]. The values of COF obtained for the TiAlN and TiAlN/CN_x coatings did not change significantly with the increase in surface roughness, which is quite a desirable coating property. A slight increase in the TiAlN coating wear occurred for the 400-grit sample and a much more pronounced increase occurred for the coarse-ground sample tested for 2000 cycles. On the other hand, the wear of the TiAlN/CN_x coating increased only in the case of sliding against the coarse-ground sample for 1000 and 2000 cycles. This result was in line with those previously published for different kinds of DLC coatings [2,4]. Since the topography and morphology of TiAlN and TiAlN/CN_x coarse-ground wear tracks are comparable with those of the smoother samples (Figure 7 and Figure S2), it is suggested that the wear mechanism is the same. However, on the rougher surfaces the interaction of the counter-ball and surface ridges occurs over smaller areas which builds up the stress and increases the ploughing effects [2]. These effects increase the abrasive- and fatigue-wear components on high surface ridges, and accelerate the coating wear through coating fragmentation [2,47].

The TiAlN and TiAlN/CN_x coatings show a low sensitivity regarding the change of COF with surface roughness, while the nl-AlTiN/TiN coating is more sensitive in this regard. In the case of the nl-AlTiN/TiN coating, the increase in roughness leads to an increase in both average COF and wear. A pronounced increase in both parameters occurs for the 400-grit sample, with a ~25% increase in COF and ~400% increase in wear. The wear of the coating additionally increases in tests on the coarse-ground sample. It seems that the tribo-film that forms during the wear of the nl-AlTiN/TiN coating has a protective nature against wear, and reduces the COF. However, during sliding on rougher surfaces (400-grit and coarse-ground), the film is easily pushed inside the surface grooves, and its effects on the wear and COF are accordingly reduced. Also, the scratch tracks of nl-AlTiN/TiN coarse-ground samples indicate that this coating is quite susceptible to cohesive cracking on the top of the coating ridges (Figure S1). This means that, during the sliding tests on rougher surfaces, more abrasive particles can be created and introduced into the wear track, which consequently increase the coating wear and COF [47]. Besides the low shear strength of the tribo-layer, its agglomeration in the center of the wear track (polished and 1500-grit) can move the counter-ball away from the tribo-track and consequently reduce the friction, similar to the observations from Ref. [48].

The lower wear rates of TiAlN and nl-AlTiN/TiN coarse-ground samples are observed for the 1000 cycle-tests. This can be attributed to the pronounced effect of wear particles hiding in the deepest surface grooves in shorter tests [23,47], which consequently hampered the precise calculation of the wear rate.

The significance of the obtained results lies in the fact that surface roughness does not have a unique effect on the tribological performance and adhesion of different hard nitride coatings, and that its effects should be separately considered for every specific coating system alone. Also, the effect of the coatings’ layer design is found to be influential in this regard. Given that all investigated coatings from the same group of surface finish had comparable surface roughness parameters, it is indicated that their behavior depends on mechanical properties, residual stress levels, chemical composition, layer design, and the nature of the tribo-film created during the sliding process. Thus, future investigations should establish the exact correlation between the mentioned parameters and the coating behavior. Such results would be beneficial in minimizing the effect of surface roughness in coatings’ practical applications. In addition, to confirm the observed behavior of nanostructured coating on rough surfaces, future investigations should include a few different kinds of these coatings. Finally, in order to reveal the wear mechanism of the coatings with different roughness, the wear tracks should be evaluated by different kinds of high-resolution electron microscopy techniques performed on their cross sections, such as focused ion-beam and transmission electron microscopy. The results of such investigations could lead to the establishment of procedures and treatments that could improve the coating performance on rougher surfaces.

4. Conclusions

In this investigation, we evaluated the scratch adhesion and tribological performance in sliding contact of three contemporary hard coatings prepared by industrial magnetron sputtering devices on substrates with different surface roughness. The investigated coatings were of different layer designs: TiAlN—a single layer, TiAlN/CN_x—a bilayer, and AlTiN/TiN—a nanolayer coating. The mechanical properties and sliding coefficients of friction of investigated coatings covered the most common range of values present in tribological coating applications. From this investigation, the following conclusions can be drawn:

• Surface roughness after the deposition of all investigated coatings increased. The topography and surface roughness of different coatings within the specific roughness group were comparable.
• Substrate roughness had a different effect on the scratch adhesion of different coatings researched in this investigation. In the range of very fine surface finishes (Sa ≈ 10–50 nm), TiAlN and TiAlN/CN_x displayed an insignificant difference in scratch adhesion, while the nanolayer AlTiN/TiN coating exhibited the highest adhesion on substrates with Sa ≈ 15 nm. The adhesion of all coatings significantly decreased on coarse-ground substrates with the Sa ≈ 500 nm, but the largest decrease was found for the AlTiN/TiN nanolayer coating.
• For all coatings, the scratch adhesion was higher in the case of scratching parallel to the machining marks. The TiAlN/CN_x coating showed the lowest differences (sensitivity) in this regard, while the AlTiN/TiN nanolayer coating displayed the largest difference between the critical forces for perpendicular and parallel scratching.
• In reciprocating sliding tests against the Al₂O₃ ball, the TiAlN and TiAlN/CN_x coatings did not exhibit a significant change of friction coefficient over the tested range of surface roughness (Sa = 65–570 nm). The wear of both coatings was constant for the range of very fine surface finishes, but profoundly increased in the tests with the coarse-ground surfaces (Sa ≈ 530 nm).
• The AlTiN/TiN nanolayer coating displayed a higher sensitivity of friction coefficient and wear on the change of surface roughness. Its friction coefficient significantly increased, but the wear rate multiplied on the samples with roughness larger than Sa ≈ 100 nm. This result, together with the results of the scratch tests, indicates that the tribological performance of the AlTiN/TiN nanolayer coating is highly dependent on its surface roughness. The reasons for such behavior were not identified in this study, but a few mechanisms were postulated. These include the specific interaction of the tribo-film with surface asperities, and susceptibility to coating cohesive cracking.
• The results obtained in this study indicate that no general rule can be established about the effect of surface roughness on tribological behavior, or about the adhesion of different hard coatings.