1 Introduction

Nowadays, the mechanical engineering sector needs to meet increasing demands on products in terms of lightweight construction, cost efficiency and performance under the prerequisite of saving resources. In order to meet these requirements, the substitution of metallic materials with polymers is a target-oriented approach. A typical representative of a substitute is poly ether ether ketone (PEEK), which has been established as alternative substrate material for components in a wide range of technical applications like gears, sliding bearings or compressor plate valves under low tribological loads [1]. This is due to a high chemical and temperature resistance as well as promising mechanical properties for technical applications. However, this leads to high costs compared to other conventionally used polymers like polyamide 66 (PA66). This polymer possesses lower mechanical and chemical properties in relation to PEEK. Polyamide is often used for structural components and tanks in the automotive industry. Nevertheless, a disadvantage arises in the use of polymers in tribological systems due to the reduced load carrying capacity compared to metals. This is accompanied by increased wear of the polymers and thus limits the field of application under high loads.

There is a long-term effort to decrease wear of PEEK mainly by addition of various fillers, fibers and solid lubricants to form better performing composites [2,3,4,5,6,7,8,9,10,11,12]. The most often carbon and glass fibers are used for reinforcement and PTFE particles as a solid lubricant. Wear improvement up to three orders of magnitude and friction decrease by up to 75% were reported. Laux et al. showed that wear depends on PEEK molecular weight and that sliding perpendicular to the roughness lay leads to thinner transfer film [13]. Zalaznik et al. pointed out that processing temperature influences hardness and tribological performance of PEEK [14].

A promising approach to increase the wear resistance of polymers is the application of physical vapor deposition (PVD) coatings, which enable a friction and wear reduction in tribological systems among others. Since the fields of application for polymers steadily increased during the 90 s, first analyses on PVD coating of thermoplastics like polycarbonate (PC), polybutylene terephthalate (PBT), polyamide (PA6.6) were conducted by Lugscheider et al. [15]. Therein the authors showed the possibility of depositing TiN on polymers by upscaling from a laboratory scale coating unit on an industrial unit. Riester et al. analyzed the influence of Ar, O2 or N as plasma etching gases on the surface activation of PBT before coating and found the highest potential for surface activation using Ar or O2 in contrast to N2 [16]. Further analysis of Lugscheider et al. using a combined pulsed magnetron sputtering technology for the deposition of TiN on polymers like PA, PEEK amongst others showed an improved compound adhesion between coating and substrate material compared to the sole use of direct current (dc) or pulsed magnetron sputtering [17]. The deposition of a zirconium modified DLC coating on PEEK was analyzed for the targeted application on a PEEK based cage for spindle bearings [18]. Therein, the results demonstrate an improved adhesion between coating and PEEK after plasma etching with argon or alternatively an argon ethyne mixture combined with a radio frequency (RF) pulsed bias voltage. These basic research studies prove the possible application of PVD coatings on polymers by process parameter adjustments.

Also, first tribological analyses were conducted by Kapinski et al., testing a–C:H coated PA66 against uncoated 100Cr6 under dry conditions at an initial Hertzian pressure of pH = 160 MPa [19]. The results prove that a–C:H coatings developed for steel substrates can be transferred to PA66 substrate and are able to reduce friction and wear and enhance lifetime. Kaczorowski et al. analyzed the effect of an O2/CH4 etching plasma compared to a N2/CH4 etching plasma on the surface activation and found, that remains of N2 are integrated into the coating structure [20]. The addition of N2 into the a–C:H coating also led to a friction reduction under dry conditions against zirconium oxide at a Hertzian pressure of pH = 117 MPa compared to the use of an unmodified a–C:H coating. Another approach by Dufils et al. is texturing of a–C:H coated PEEK tested against alumina and lubricated with distilled water at a Hertzian pressure of pH = 53 MPa [21]. It was found that dimples with a size between 12 and 21 µm and a density on the coated surface area between 10 and 30% lead to a significant friction reduction compared to uncoated PEEK. Tribological tests at higher initial Hertzian pressure of pH = 345 MPa under boundary and mixed lubrication against uncoated 100Cr6 also prove a friction and wear reduction compared to uncoated PEEK [22]. The overview of tribological investigations shows that DLC has a high potential to affect the wear and friction behavior of polymers positively. Nevertheless, there are only a few studies in this field so far, so that further investigations regarding the mechanical and chemical interactions during the tribological contact are necessary for a broad knowledge base.

Besides the achieved friction and wear reduction, the DLC microstructure is affected by the tribological contact conditions, which can be analyzed my means of Raman spectroscopy. The effect on the microstructure is called relaxation as described by Kalish et al. [23] and is dominated by the two main mechanisms clustering of graphitic sp2 bindings and the transformation of diamond-like sp3 bindings into sp2-bindings. Furthermore, clustering subdivides itself into the three processes:

  • Transformation of sp2 chain bindings into sp2 ring bindings

  • Reduction of density and rearrangement of sp2 bindings

  • Effusion of hydrogen, which is only valid for hydrogen-containing DLC

The Raman spectrum of carbon is primarily determined by the presence of the G peak, which is based on bond stretching vibrations of chain and ring sp2-bindings [24]. The D peak is solely based on breathing modes of sp2 ring bindings. Ferrari and Robertson analyzed the influences of different bonding states and composition of hydrogen-free amorphous carbon coatings [25]. Thereby, the position and intensity of D and G peak are affected by clustering, bond disorder, share of sp2 chain bindings and sp3 bindings. A qualitative evaluation of the binding states of hydrogen-free DLC is possible through the 3-stage model established by Ferrari and Robertson. Therein, the G peak position and ratio of the intensities of G and D peak are considered to describe the amount of sp3 bindings in DLC. The interpretations of the 3-stage model are extended for hydrogen-containing DLC coatings by Casiraghi et al. [26]. Furthermore, the parameters dispersion of the G peak Disp(G) and the full width half maximum FWHM(G) of the G peak can be determined. The Disp(G) predominantly describes the topological disorder, which combines the information of size and shape distribution of sp2 clusters of chains. At the same time, changes of sp2 ring bindings do not affect Disp(G) based on the results for a hydrogenated DLC coating ta–C:H of Ferrari and Robertson [27]. Furthermore, also the structural disorder of amorphous carbon can be described by determining the full width half maximum of the G peak FWHM(G)UV under UV excitation. Based on the findings of Casiraghi et al., structural disorder arises mainly from bond angle and bond length distortions [26]. It must be noted, that the influence of the structural disorder on the Disp(G) and the topological disorder on the FWHM(G)UV cannot be completely excluded.

In this study, durable DLC coatings applicable on PEEK and PA66 substrates to improve tribological performance are developed. The a–C:H coated specimens were analyzed after tribological testing to identify possible changes in the coating structure and further correlated with findings based on frictional and wear behavior.

2 Methods

2.1 Pretreatment of PEEK and PA66 Specimens

Squared polymer samples of PEEK, TECAPEEK, Ensinger GmbH, Nufringen, Germany, and PA66, TECAMID 66 LA natural, Ensinger GmbH, Nufringen, Germany, with a size of 30 mm × 30 mm and a height of h = 8 mm were grinded and polished by means of Phoenix 4.000 ×, ITW Test & Measurement GmbH, Esslingen am Neckar, Germany, to a roughness of Ra ≈ 0.05 µm. Therefore, SiC paper with a grain size of 1200, 2000 and 4000 and diamond suspension with a grain size of 3 µm was used. The samples were chemically purified in a cleaning unit with the water-soluble basic detergents GALVAPREP SU 2012, NGL Cleaning Technology SA, Nyon, Switzerland, and a subsequent washing with water and isopropanol C3H8O.

2.2 Coating Process

Two a–C:H coatings, called a–C:H-1 and a–C:H-2, were developed and deposited in an industrial scale coating unit CC800/9 Custom, CemeCon AG, Wuerselen, Germany. The coating unit was equipped with two high power pulsed magnetron sputtering (HPPMS) cathodes. The arrangement of the cathodes and the layout of the deposition chamber are depicted in Fig. 1. For the deposition of the a–C:H coatings, four C targets with a purity of C = 99.9% were used. The process and reactive gases were argon (Ar), helium (He) and ethyne (C2H2). The coating processes were adjusted in accordance with mechanical properties, which show higher values for PEEK in comparison to PA66. A reduced helium gas flow lowers the degree of ionization of carbon in the coating process. The resulting structure shows a reduced content of diamond-like sp3 bindings in favor of more graphitic sp2 bindings. The reduction of the bias voltage UB influences the morphology as well as the mechanical properties due to the lower kinetic energy of the ionized particles during the coating growth.

Fig. 1
figure 1

Schematic representation of the industrial coating unit CC800/9 Custom and target

Full size image

The helium gas flow was reduced from Q = 70 sccm to Q = 50 sccm and the bias voltage from UB = − 250 V to UB = − 200 V for the deposition on PA66, Table 1. In addition, the bias voltage UB was pulse-synchronized with the HPPMS cathodes, which reduces the thermal transfer on the substrate material. The PVD coating process is basically divided into the four steps heating, etching, coating and cooling. In between the steps heating and etching there is a process step called off-gassing, where the starting pressure of p = 3 mPa is reached again. In order to analyze the substrate temperature monitored during the coating process, temperature measurements were carried out using Insitemp, which is described in [28]. Therefore, specimen dummies made out of PEEK and PA66 were used.

Table 1 Process parameters of a–C:H-1 on PEEK and a–C:H-2 on PA66 shown in [22]
Full size table

2.3 Coating Analysis

The coating thicknesses and morphology were analyzed by scanning electron microscopy (SEM), Oxford Link ISIS, Oxford Instruments plc, UK, at the Central Facility for Electron Microscopy (GfE), RWTH Aachen University, Aachen, Germany. A nanoindenter of type TI 950 TriboIndenter, Bruker Corporation, Billerica, Massachusetts, USA, with a Berkovich indenter with a nominal edge radius of approximately R = 20 nm was applied for the determination of the indentation hardness HIT and modulus of indentation EIT. The samples were unprepared and the used constant indentation force was F = 2 mN. Thereby, the penetration depth was kept below 10% of the top layer thickness. Thus, the influence of the substrate material can be minimized as explained by Fischer-Cripps [29]. In order to determine HIT and EIT, x = 40 individual measurements were conducted and subsequently averages out of the force–displacement curves were calculated. Calculations of the modulus of indentation are based on Oliver and Pharr’s equations [30]. Further informations on the nanoindentation method are presented in [31]. A constant Poisson's ratio of υ = 0.25 was assumed for the coatings. Nuclear reaction analysis (NRA) were carried out at Helmholtz Zentrum Dresden Rossendorf (HZDR), Dresden Rossendorf, Germany, to determine the hydrogen content of the a–C:H coatings in the as-deposited status. Two positions were analyzed on each coating for the penetration depths γ = 35 nm and γ = 500 nm with 20 repetitions. Thereby, the resulting hydrogen contents had to be evaluated by an exponential fitting method with approximation to a limit value in contrast to the determination by mean value.

2.4 Compound Analysis

The analysis of the compound adhesion between the a–C:H coatings and PEEK respectively PA66 were conducted by means of scratch tester Gesellschaft für Fertigungstechnik und Entwicklung Schmalkalden e.V. (GFE), Schmalkalden, Germany. Scratch tests, with one repetition each, were conducted with a Rockwell C diamond at different constant forces between 10 ≤ FN ≤ 250 N in steps of 10 N. Every scratch was generated with a length of xs = 4 mm and a feed speed vf = 10 mm min−1. The load was applied during the first xs = 0.2 mm of the total scratch length. In accordance to DIN EN ISO 20502, the compound adhesion between coating and substrate was evaluated by scratch tests and confocal laser scanning microscopy (CLSM), VK-X210, Keyence, Neu-Isenburg, Germany. It has to be stated that the scratch testing is not standardized for coated plastics. Nevertheless, the indention and scratching with a Rockwell C diamond through coated plastics at different loads simulates a high stress collective for the compound of coating and substrate material.

2.5 Tribological Analysis by Means of Pin-on-Plate Tribometer

Reciprocating pin-on-plate (PoP) tribological tests were carried out on an UMT Tribolab, Bruker Nano, California, San Jose, USA, under dry and lubricated conditions. In the device, normal load is close-loop controlled and measured simultaneously with friction by two axis sensor. The pin is stationary and the plate performs a harmonic motion. A circular point contact was formed between 10 mm in diameter 100Cr6 bearing steel ball and flat coated and uncoated PEEK and PA66 specimens with a geometry 30 × 30 mm and height of h = 8 mm. Dry and oil lubricated conditions were measured. A mineral base oil FVA3 was used as lubricant, Table 2. Operating conditions of the tests are listed in Table 3. Tribological performance over time was studied in five steps of test duration.

Table 2 Properties of the considered mineral oil FVA3 [32]
Full size table
Table 3 Operating conditions of PoP tribological tests
Full size table

In initial tests, a higher frequency compared to f = 2 Hz was tested. It led to wear tracks that have signs of wear markedly influenced by contact temperature rise, thus the frequency was decreased to f = 2 Hz. The test temperature for oil lubricated conditions was set to ϑOil = 90 °C to decrease oil viscosity and minimize formation of fluid film. The goal was to perform tests at severe mixed and boundary lubrication (see Table 3).

The average coefficient of friction and volume of material removed by wear were evaluated. The wear volume was evaluated from surface topography measurement of the complete wear track after the experiment by optical profilometer Contour GT-X8, Bruker, Arizona, Tucson, USA. To be precise, the wear volume was evaluated as a net missing volume of material from topography data with reference plane established on surface out of the wear track. In this way of evaluation, the transported material on the plate is not included in the wear volume. The volume was calculated from the area of wear track and the closest surroundings by masking out the rest of sample. This approach decreases uncertainties of the volume measurement due to surface geometry imperfections (waviness). The estimated uncertainty of wear volume is around 1 105 µm3. It is mainly given by precision of reference plane establishment and local sample waviness in the area of wear testing.

2.6 Chemical Analysis After Tribological Testing by Means of Raman Spectroscopy

The molecule structure of the hydrogen-containing amorphous carbon coatings before and after tribological testing was analyzed by means of Raman spectroscopy. Therefore, the Raman spectrometer, Renishaw InVia Reflex, Renishaw GmbH, Pliezhausen, Germany, with a λVIS = 532 nm laser with a spot size of d ≈ 1 µm and a diffraction grating gVIS = 1800 l mm−1 was used. In addition, Raman spectra were taken with an ultraviolet (UV) laser λ = 325 nm with a spot size d ≈ 6 µm and a diffraction grating gUV = 2400 l mm−1. The lasers were calibrated before the measurement using a silicon reference sample. To enable an evaluation of statistic valid data, a mapping measuring method was used. Hereby, a matrix with 2 × 3 measurement points with size of 100 × 100 µm and a step width of 50 µm was positioned horizontal over the middle of the wear track on a–C:H coated disk for the analysis by VIS and UV laser. The measurement parameters are shown in Table 4.

Table 4 Raman measurement parameters for chemical analysis of a–C:H coatings before and after tribological testing
Full size table

For visible light (VIS) laser excitation the G peak position ranges between 1560 cm−1 ≤ ῦG,VIS ≤ 1600 cm−1 and in the area between 1600 cm−1 ≤ ῦG,UV ≤ 1690 cm−1 using an ultraviolet (UV) laser. The D peak occurs in the area 1360 cm−1 ≤ ῦD ≤ 1400 cm−1 under VIS excitation, as far as ring bindings are present in the DLC structure. Higher wave numbers of the D peak position are measured by UV laser excitation in the area between 1400 cm−1 ≤ ῦD,UV ≤ 1500 cm−1.

All spectra were taken using the software Wire® 5.4, Renishaw plc, Wotton-under-Edge, United Kingdom. Thereby, the spectra were evaluated section by section of the Raman shift by a combined Gaussian and Lorentzian fitting. Based on the measured spectra, the parameters I(D)/I(G)VIS ratio, I(D)/I(G)UV ratio, FWHM(G)UV and Disp(G) were determined. In case of the characteristic value FWHM(G)UV, the UV excitation enables to exclude the photoluminescence of the spectra as shown by Casiraghi et al. [26]. Furthermore, the dispersion of the G peak Disp(G) can be calculated by determining the position at visible light ῦG and UV ῦG,UV excitation divided by the excitation wavelengths.

The interpretations of the a–C:H coatings were performed in analogy to the findings of Casiraghi et al. [26]. Correspondingly, the coatings were classified as graphite-like a–C:H (GLCH) with a hydrogen content Ψ > 20% and sp2 clustered ring bindings.

$${\text{Disp}}\left( {\text{G}} \right) = \frac{{\tilde{\upsilon }_{{\text{G,UV}}} - \tilde{\upsilon }_{{\text{G}}} }}{{\lambda_{{{\text{UV}}}} - \lambda }} \left[ {\frac{{{\text{cm}}^{ - 1} }}{{{\text{nm}}}}} \right]$$
(G1. 1)

3 Results and Discussion

3.1 Coating and Compound Analysis

3.1.1 Substrate Temperature Measurement During Coating Process

The results of the substrate temperature measurement during the coating processes of PEEK and PA66 are shown in Fig. 2. Thereby, the coating process is divided into the five phases heating, off-gasing, etching, coating and cooling. The highest substrate temperature for a–C:H-1 is reached during the etching phase with Tmax,1 = 174 °C, whereby the maximum substrate temperature for a–C:H-2 was measured at Tmax,2 = 186 °C. It is believed, that the deviation between the maximum substrate temperatures results predominantly from differences of the PEEK and PA66 specimen properties and to a minor extend due to differences in the coating processes.

Fig. 2
figure 2

Substrate temperature during coating process of a–C:H-1 and a–C:H-2

Full size image

In general, the measured substrate temperatures are far below the melting temperatures of PEEK with T = 340 °C and PA66 with T = 255 °C, given by Domininghaus et al. [1]. In contrast thereto, the glass transition temperatures of PEEK with T = 143 °C and PA66 with T = 70 °C are surpassed. On this point, it must be stated that the PEEK and PA66 specimens are mounted using holders which allow movement such that no significant thermal stresses occur during coating process.

3.1.2 Morphology of a–C:H Coatings by SEM Analysis

The cross-sectional images and top layer images of a–C:H-1 and a–C:H-2 are shown in Fig. 3. Thereby a columnar structure with the transition into an amorphous character of both coatings a–C:H-1 and a–C:H-2 can be observed.

Fig. 3
figure 3

Cross sectional images and topography images of a–C:H-1 on PEEK, shown in [22], and a–C:H-2 on PA66

Full size image

The cross-sectional image for a–C:H-1 shows a wavy structure and it can be assumed that this is caused by the preparation of the specimens for the microscopy. Hereby, it is necessary to apply high mechanical forces to break the specimens. Even though there are high plastic deformations of the substrate material visible, no cracking or delamination at the interface between coating and substrate can be observed. At the top view of a–C:H-1 interspaces can be seen in between the cauliflower structure, which can be a result of the sample preparation during grinding and polishing. A coarse columnar structure can be observed for a–C:H-2 on PA66. Furthermore, a very rough structure can be seen at the top layer of a–C:H-2, where the columns show locally heightened areas. It can be assumed, that among others the structural differences between a–C:H-1 and a–C:H-2 result from the growth process on the ring binding structure of PEEK and chain binding structure of PA66. The coatings should be predominantly composed of sp2 ring bindings due to the HPPMS sputtering of graphite in contrast to the share of sp2 chain bindings resulting out of the low ethyne gas flow with \(\dot{\text{Q}}\) C2H2 = 10 sccm. Thus, the “epitaxial” coating growth of a–C:H-2 on PA66 is affected in comparison to the coating growth of a–C:H-1 on PEEK and leads to the observed uneven structure. Due to the small differences of bias voltage with ∆UB = 50 V and helium gas flow with ∆QHe = 20 sccm compared to a–C:H-1, it is not believed, that the process parameters lead to the observed changes of the coating structure. Catena et al. came to a similar conclusion by analyzing the structure of a–C:H on different thermoplastic substrate materials like polyoxymethylene (POM), high density polyethylene (HDPE) and polyethylene terephthalate (PET) [33].

3.1.3 Roughness of Uncoated and Coated Samples

The results of the roughness measurements of the arithmetical mean roughness Ra and root mean square roughness Rq before and after coating of PEEK and PA66 are shown in Fig. 4. The increase in roughness for the as-deposited state of a–C:H-1 compared to uncoated PEEK is moderate.

Fig. 4
figure 4

Roughness measurement of uncoated and coated PEEK and PA66

Full size image

An explanation can be given by small defects on the a–C:H-1 coating, which result out of the coating growth process. In contrast, the roughness of the PA66 specimen increases significantly after coating deposition. The measurements of Rq show a similar trend to the changes of Ra. Overall, the roughness measurements confirm quantitatively the observations of the SEM images.

3.1.4 Chemical Analysis of Hydrogen Content of a–C:H Coatings

The hydrogen content Ψ of a–C:H-1 and a–C:H-2 at γ = 35 nm and γ = 500 nm depths from the top layer are shown in Fig. 5. It can be observed, that the hydrogen content at the top layer in γ = 35 nm depth varies between Ψ ≈ 5.0 at.% by comparing a–C:H-1 and a–C:H-2. By increasing the penetration depth to γ = 500 nm the hydrogen contents of both coatings tend to equalize. It can be assumed, that the deviations of the hydrogen content at the top layer result from the differences of the coating process parameters as well as the substrate material properties of PEEK and PA66 on the coating growth process.

Fig. 5
figure 5

Hydrogen content of a–C:H-1 and a–C:H-2 at different depths

Full size image

Information about the binding structure of hydrogen in the a–C:H coatings could be obtained out of the evaluation of the hydrogen content. Based on extrapolation method against zero, the absolute hydrogen contents could be calculated. The reason for the exponential fitting method is due to an effusion of hydrogen during the measurement, which indicates that hydrogen is primarily interstitial embedded in the a–C:H coating structure. It can be assumed, that this enables an easier effusion to the surface and affects the tribological behavior.

3.1.5 Indentation Hardness and Modulus of Indentation of a–C:H Coatings

The indentation hardness HIT and the modulus of indentation EIT of the DLC coatings are shown in Fig. 6. In order to minimize an influence of the substrate material on the hardness measurement, the penetration depth of the Berkovich indeter was kept at 10% of the individual coating thickness. The difference between the indentation hardness and the modulus of indentation of a–C:H-2 compared to a–C:H-1 can be explained based on the reduced helium gas flow and changed bias voltage UB during the coating process. The changes lead to a lower ionization energy of the C–C and C–H clusters in the a–C:H-2 coating process, which negatively affects the formation of sp3 diamond bindings and reduces the share in relation to the sp2 bindings. In addition, the hydrogen content at the top layer is higher for a–C:H-2 in contrast to a–C:H-1, which also reduces the hardness of the coating.

Fig. 6
figure 6

Indentation hardness HIT and modulus of indentation EIT of a–C:H-1 on PEEK and a–C:H-2 on PA66 shown in [22]

Full size image

3.1.6 Compound Adhesion of a–C:H on Polymers

The compound adhesion between a–C:H-1 and PEEK as well as a–C:H-2 and PA66 was tested by means of scratch testing, Fig. 7. Even though the scratch test is not standardized for the analysis of coated polymers, important insights can be gained. The results show a high deformation of ∆h = − 64.3 µm of the a–C:H-1 on PEEK compound at a critical load Lc = 150 N in the center of the scratch without a coating delamination. In the edge regions of the scratch, high plastic deformation up to a height of h = 21.4 µm only show partial cracks of a–C:H-1.

Fig. 7
figure 7

Compound adhesion between a–C:H-1 on PEEK and a–C:H-2 on PA66

Full size image

With regard to the a–C:H-2 on PA66 compound, higher critical loads up to Lc = 250 N could be applied without coating delamination. Higher critical loads also result in increased plastic deformation up to ∆h = − 132.7 µm. At Lc = 250 N, the maximum load of the analyzing unit was achieved. Nevertheless, the elevations at the edge of the scratch remain at a comparable level around ∆h ≈ 20 µm compared to the a–C:H-1 on PEEK compound. In the case of a–C:H-2 on PA66 compound, cracks form at the top layer within the left side of the scratch and not in the edge region. Both a–C:H coatings are able to undergo high deformation without delaminating from the polymers and significant differences in the hardness and modulus of indentation. This offers a high potential for a subsequent tribological analysis under high loads, to investigate the effect on friction and wear in highly stressed sliding and rolling-sliding contacts.

3.2 Tribological Behavior and Evaluation

3.2.1 Friction and Wear Measurements

The results of wear volume and the average coefficient of friction µ depending on the number of passes for uncoated PEEK and a–C:H-1 on PEEK under dry conditions at T = 60 °C are shown in Fig. 8. Solid and dashed lines show the linear fits with the wear rates to the data of PEEK and a–C:H-1 respectively, whereby the y axis is in log scale.

Fig. 8
figure 8

a Wear volume and b coefficient of friction of PEEK/100Cr6 and a–C:H-1/100Cr6 under dry conditions

Full size image

Wear is approximately one order of magnitude lower for coated than for uncoated PEEK for all measured passes. The coefficient of friction is at an almost constant level for PEEK/100Cr6, whereas an increase with the number of passes for a–C:H-1/100Cr6 can be observed. As the number of passes N exceeds 8000, the coefficient of friction of a–C:H-1/100Cr6 is lightly increased compared to PEEK/100Cr6. Due to high hardness of a–C:H-1 coating and 100Cr6 ball, both surfaces undergo transformations during wear process under dry lubrication. Therefore, a possible explanation for increase of friction can be an increase of abrasive wear. The latter would be connected with change in tribofilm chemistry. No dedicated analyses were made to chemistry of tribofilms in this study. Figure 9 shows the wear volume and coefficient of friction for lubricated conditions at T = 90 °C. The pairing of a–C:H-1/100Cr6 features lower wear by approximately one order of magnitude compared to PEEK/100Cr6. When compared to dry conditions, the wear volumes are about one and a half orders of magnitude lower. The coefficient of friction is stable for a–C:H-1/100Cr6 and there can be seen a slight increase and leveling off for PEEK/100Cr6. In this pairing, the wear occurs primarily at the PEEK specimen. A significant wear track with scratch topography (Fig. 10c) is produced for high number of passes. The decrease of friction can be connected with improved lubrication due to PEEK surface adaptation to the steel pin, and thus, operation of the contact with less solid contacts.

Fig. 9
figure 9

a Wear volume and b coefficient of friction of PEEK/100Cr6 and a–C:H-1/100Cr6 under lubricated conditions

Full size image
Fig. 10
figure 10

Wear tracks on uncoated PEEK and a–C:H-1 after N = 16,000 passes; a and b for dry conditions; c and d for lubricated conditions

Full size image

Fig. 10 shows surface height images of uncoated and coated PEEK after wear testing for N = 16,000. The reason for the selection of N = 16,000 passes is that this is a representative mid-point of the analyses, as Figs. 8 and 9 show that an increase in the number of passes lead to comparable or minimally increased values. The wear track on uncoated PEEK shows high plastic deformation and a transport of material towards the edge regions under dry conditions. The average rise in contact temperature due to friction is higher under dry conditions compared to lubricated conditions. The uncoated PEEK sample has the highest wear scar depth in the center while the coated sample a–C:H-1 shows no major signs of deformations. Visible abrasion marks can be found on a–C:H-1 after dry testing and uncoated PEEK after lubricated testing. The a–C:H-1 coating decreases wear significantly under both lubrication methods.

The results of wear volume and the average coefficient of friction depending on the number of passes for PA66/100Cr6 and a–C:H-2/100Cr6 under dry conditions at T = 60 °C are shown in Fig. 11. Plotted solid and dashed lines are best fit linear trend lines of the wear data. Both samples follow an almost linear wear trend, note that y axis is in log scale. The wear volume for a–C:H-2 coated sample is significantly lower than for uncoated counterpart. The uncoated and coated configurations have a similar progressive trend of the coefficient of friction over the number of passes, whereas friction for a–C:H-2/100Cr6 is slightly lower compared to PA66/100Cr6.

Fig. 11
figure 11

a Wear volume and b coefficient of friction of PA66/100Cr6 and a–C:H-2/100Cr6 under dry conditions

Full size image

The wear and friction performance is different under lubricated conditions, see Fig. 12. The wear volume of uncoated PA66 and a–C:H-2 is almost identical. However, the coefficient of friction of a–C:H-2/100Cr6 is higher but stable compared to PA66/100Cr6. The trend of increasing friction for a–C:H-2/100Cr6 with increasing number of passes is similar to a–C:H-1/100Cr6, see Fig. 9. Thus, similar explanation can be expected.

Fig. 12
figure 12

a Wear volume and b coefficient of friction of PA66/100Cr6 and a–C:H-2/100Cr6 under lubricated conditions

Full size image

The surface height images of wear tracks after N = 16,000 passes are shown in Fig. 13. For dry conditions, both wear tracks show elevated edge regions formed by wear debris. Some cracking and partial coating removal was visible for a–C:H-2/100Cr6 operating under dry conditions. It seems that surface defects and grooves support initiation of cracking but it is not possible to exclude also initiation on a regular surface. Since the cracking was not observed for a–C:H-1 coated PEEK and a–C:H-2 coated PA66 under lubricated conditions it is assumed that the cracking and partial coating removal is related to thermal properties of PA66. Abrasion scratch marks are visible dominantly for uncoated PA66 under lubricated conditions.

Fig. 13
figure 13

Wear tracks on uncoated PA66 and a–C:H-2 after N = 16,000 passes; a and b for dry conditions; c and d for lubricated conditions

Full size image

The DLC coating a–C:H-1 under dry conditions was exemplarily chosen for further analysis, due to the significant wear reduction compared to uncoated PEEK. Therefore, SEM and Raman analysis of the a–C:H-1 coating after N = 16,000 passes were conducted.

3.2.2 SEM and Raman Analysis of a–C:H-1

SEM topography and cross-sectional images of a–C:H-1 on PEEK after N = 16,000 passes are shown in Fig. 14. Several cracks can be identified inside the wear track on the topography images. Here, smaller detachments of the coating from the substrate material can also be seen in individual cases. The single columns of the cauliflower structure and the PEEK seem to have formed a composite layer. Changes in the mechanical properties could result from this, which favor the formation of cracks. This requires further investigation to gain a deeper understanding of the tribological behavior of the DLC coated thermoplastics.

Fig. 14
figure 14

Topography and cross section of wear track on a–C:H-1 on PEEK under dry conditions after N = 16,000 passes

Full size image

With regard to the cross-sectional micrographs, this hypothesis can be confirmed: A lens shaped zone of impact with an amorphous structure can be identified within the wear track. Here, the interface between a–C:H-1 and PEEK can not longer be clearly differentiated from each other. The transition from the unloaded zone to the wear track can be observed in the left edge region. Raman measurements inside the wear track on a–C:H-1 prove, that the position and shape of the G peak changes significantly, Fig. 15. Evaluated data from the Raman spectra of a–C:H-1 as-deposited and a–C:H-1 after N = 16,000 passes are shown in Table 5. In general, the spectrum after N = 16,000 passes differs significantly from the uncoated PEEK spectrum and is more related to the a–C:H-1 as-deposited spectrum.

Fig. 15
figure 15

Raman spectra of uncoated PEEK and a–C:H-1 on PEEK in different states

Full size image
Table 5 Results of Raman spectra evaluation
Full size table

The G peak shifts about ∆ῦG = 115 cm−1 from the as-deposited state to the status after N = 16,000 passes, which can be assigned to an increased clustering of the C–C and C–H bindings in the a–C:H-1 composite layer. The reason for this is the high Hertzian pressure and high frictional heat during dry contact conditions, which was calculated up to T = 152 °C based on the average temperature rise in accordance with [34]. Thus, the sp2 chain bindings are transformed into sp2 ring bindings, whereby the effusion of hydrogen additionally occurs. Furthermore, metastable sp3 bindings transform into sp2 ring bindings. The increasing share of sp2 ring bindings can be additionally confirmed by the growing pronunciation of the D peak, which is solely based on vibrations of sp2 ring bindings. However, the effect of increasing D peak can not be observed in the I(D)/I(G) ratios whether under visible light or UV excitation. With regard to the shown Raman spectra, an increase of the FWHM would have been expected. However, the FWHM reduces from the as-deposited state compared to the after N = 16,000 passes.

The increase of the Disp(G) is a result of the increased share of sp2 ring bindings, whereby the size and shape distribution of the sp2 clusters is changed as a consequence of the tribological load. An explanation for the unexpected observations of the I(D)/I(G) ratios and the FWHM(G)UV could be a mutual neutralization of the described effects. A further explanation, leading to possible misinterpretations of the considered Raman values could be the formation of the composite layer between a–C:H-1 and PEEK in the wear track due to high energy input. However, the application of a–C:H-1 still enables wear protection of PEEK. So far, these changes cannot be interpreted in terms of a positive or negative influence on tribological behavior, thus further studies are required.

4 Conclusions

In this study two a–C:H coatings were developed and tailored to the application on PEEK and PA66. Tribological analyses under dry and boundary lubricated conditions were conducted with focus on the friction and wear behavior. Finally, SEM and Raman analysis enabled a deeper understanding of the effects of dry contact conditions on the a–C:H coatings. The results show:

  • High plastic deformation of a–C:H coatings without delamination from the thermoplastic substrates PEEK and PA66

  • Significant wear reduction under highly stressed dry and boundary lubricated conditions due to the application of tailored a–C:H coatings

  • Possible formation of a composite layer of a–C:H-1 and PEEK in the wear track due to high energy input under dry conditions

  • Transformation from sp2 chain into sp2 ring bindings as well as change of size and distribution of the sp2 clusters during dry conditions

This study underlines the potential of DLC on thermoplastics to enhance the tribological performance. The tribological analysis in a rolling-sliding contact and subsequent Raman analysis are presented in future studies. Future work will focus on application of DLC coatings on gears made of polymers.