1 Introduction

Polytetrafluoroethylene (PTFE) is a chemically resistant, high-temperature (upper use temperature ~260 °C) polymer with a very low friction coefficient (µ ~ 0.1) [1]. For tribological applications at sliding velocities greater than 10 mm/s, the wear of unfilled PTFE becomes unacceptably high (K ~ 1–8 × 10–4 mm3/Nm) [2,3,4]. This high wear rate has been attributed to the formation of subsurface cracks in the amorphous region within the microstructure of the highly crystalline PTFE [2, 4,5,6]. These cracks cause unfilled PTFE to have large (~ 1–20 µm) thick wear debris that delaminate from the surface [2]. Unfilled PTFE under these conditions will also form patchy, unstable tribofilms on the counterbody material (typically low carbon or stainless steel). To reduce delamination, secondary phases (particles, fibers, and platelets) of ceramics, metals, and other polymers have been added to PTFE [2, 7,8,9,10]. Secondary phases provide load support and crack arrestation mechanisms [2, 7, 11], which can improve the wear of unfilled PTFE by 10–1,000 times. These mechanisms reduce the size of the PTFE wear debris significantly and may improve transfer film adhesion of the debris onto the counterbody material [2, 7].

An improved class of PTFE composites, often designated “ultralow” wear PTFE composites, were developed in the early 2000s and exhibited wear up to 10,000 × less than unfilled PTFE. Though PTFE filled with nanostructured-α-Al2O3 (to the authors’ knowledge) is the most studied ultralow wear PTFE composite [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27], additional filler materials (GeO, BeO [28], carbon nanoscale fillers [29,30,31], Mn, Ti, and Cr [32], FeCo [33]) have produced similar ultralow wear rates when sliding against stainless steel as well (K ~ 2 × 10–8–3 × 10–7 mm3/Nm). All these tribological systems form thin continuous tribofilms on both the composite (i.e., the running film) and on the counterbody (i.e., the transfer film) [19, 24].

The importance of the transfer film formation on ultralow wear PTFE composites was evaluated over the last fifteen years. In the initial few kilometers of sliding, PTFE composite wear debris begins to attach to the counterbody (often a polished or lapped stainless steel surface). The initial patches of the transfer film are typically larger, and slowly decrease in size with increasing sliding cycles [13, 34]. Refinement of the transfer film corresponds with reduction in wear rate which initially starts at ~ 3 × 10–4 mm3/Nm and decreases to ~ 1 × 10–7 mm3/Nm [15]. The formation of transfer films may be affected by modifying the surface finish (polishing and directionality) [16, 32, 35] or counterbody material [16, 32, 35]. Transfer film coverage and quality may be characterized using the mean-free space length metric, defined by Ye et al. [15]. The mean-free-space length (Lf) is the size of the transfer film-free region on the counterbody [15]. As sliding progresses, the transfer film covers more of the counterbody and the size of the film-free regions on the counterbody decreases. Ye et al. proposed that as the Lf decreases, direct polymer composite-on-counterbody contact is lowered, which corresponds with observed reduction in the wear rate of the system [21].

Additional experiments on ultralow wear PTFE composites have shown tribofilm formation and wear performance is dependent on environment. Tests on PTFE-α-Al2O3 composites slid in vacuum and dry nitrogen environments have shown about 100 × increase in wear compared to samples tested in laboratory air [22, 23, 36, 37]. X-Ray photoelectron spectroscopy (XPS) and infrared spectroscopy (IR) showed critical chemical differences on the worn surface of the composite and counterbody material when tested in dry environments versus humid air environments. The spectra of samples tested in humid air exhibited carboxylate salt groups, while the spectra of both the vacuum and dry nitrogen experiments showed no evidence of carboxylate salt formation. A tribochemical mechanism for the formation of carboxylate salts, developed by Harris et al., proposes shear stress caused by sliding leads to chain scission of carbon bonds within the PTFE backbone [25].The broken carbon bonds expose free radicals at the chain ends, which react with both ambient oxygen and water in the environment to form carboxylic acid endgroups. These endgroups bond to nearby metal and metal oxide surfaces, such as alumina or other particles within the PTFE composite or iron within the counterbody, to form carboxylate salts observed on the infrared spectra of both tribofilms. These carboxylate salt groups are hypothesized to increase cohesiveness and adhesion between the PTFE matrix and filler (running film) as well as the wear debris and counterbody material (transfer film). These bonds lead to improved mechanical properties of the worn composite surface and resilience of the transfer film and worn counterbody surface against wear [21]. Additional molecular dynamics simulations have studied the degradation of the PTFE backbone leading to bonding with alumina [38, 39].

To the authors’ knowledge, the all previous IR and XPS measurements of worn PTFE-α-Al2O3 composite systems have only looked at the worn composite or transfer film before and after testing (often after 25 km). The only outlier to this trend was Pitenis et al., who did use XPS to evaluate the transfer film as a function of reciprocating cycles at 100, 1 k, 10 k, 100 k, and 1 M cycles [24]. The aim of this work was to provide the physical and chemical evolution worn surface of PTFE-α-Al2O3 composites and 304 SS counterbody from moderate wear (~ 105 mm3/Nm) to ultralow wear (< 5 × 10–7 mm3/Nm). In particular, this study analyzed the worn polymer surface via attenuated total reflectance and transmission infrared spectroscopy throughout the transition from moderate to ultralow wear for PTFE-αAl2O3, which has not been published in the literature to the best of the authors’ knowledge. These measurements should promote understanding of how tribochemistry and wear correlate are during the run-in period of ultralow wear fluoropolymer materials.

2 Materials and Sample Preparation

Compression molding, sintering, and machining methods of the PTFE-α-Al2O3 composite samples were similar to past procedures [32]. First, 5 wt.% α-Al2O3 particles (Nanostructured and Amorphous Materials, stock # 1015, BET estimated specific surface area = 41.4 m2/g, d50 = 3.95 µm) and PTFE 7CX (Chemours, average particle size 31 µm) resin were massed, manually mixed, and submerged in isopropyl alcohol (IPA) in 2:1 volume. The powder-alcohol mixture was sonicated using an ultrasonicating horn (Branson Digital Sonifier SFX 550, Emerson Electric Company, St. Louis, MO, USA) with 3.125 mm microtip for five minutes. After sonication, the mixture was dried overnight in an oven at 50–60 °C to remove the alcohol. The dried powder mixture was formed into a ~ 12.7 mm diameter cylinder using a hydraulic press (50 MPa pressure). The compressed samples were free-sintered in laboratory air a furnace under the following heating profile: slow temperature rise (2 °C/min) to 380 °C, hold at 380 °C for three hours, slow cooldown (2 °C/min) to room temperature. The sintered PTFE-α-Al2O3 cylinders were machined into 6.3 × 6.3 × 15 mm3 rectangular pins. The pins were polished using a Grinder-Polisher (Buehler, Ecomet® 250, Lake Bluff, IL USA) with 800 grit SiC paper with water as a lubricant. Prior to testing, polymer pin dimensions were taken using digital calipers with a resolution of 0.01 mm. The pins were sonicated in a methanol bath for thirty minutes to remove any contamination during machining/polishing and dried in lab air for at least 24 h before testing began.

304 Stainless Steel (SS) was used as the counterbody material (mirror finish, Ra ~ 0.04 μm, annealed, Rockwell B80, purchased from McMaster Carr, Elmhurst, IL, USA). Counterbody samples were cut into their final size (25.4 mm × 38.1 mm × 3.2 mm) using a waterjet cutter, then cleaned with soap and water (Alconox Powder Precision Cleaner) to remove any contamination. Each counterbody was sprayed with methanol and dried using a lint-free cloth at least 30 min before testing to ensure a clean surface before testing began.

3 Experimental Methods

3.1 Tribological Measurements

Friction coefficient and wear rate of the PTFE-α-Al2O3 composites was evaluated using a custom made flat-on-flat tribometer. Similar testing parameters to past experiments on PTFE-α-Al2O3 composites were used [32]. These parameters are: normal load (Fn = 250 N, ~ 6.3 MPa nominal contact pressure), stroke length (25 mm), and sliding velocity (50 mm/s). The contact pressure was less than half of the yield strength of the PTFE matrix (12–15 MPa) [21] and the low sliding velocity ensured insignificant heat generation at the interface [25]. All experiments were completed in a laboratory air environment (20–21 °C and 37% relative humidity). Mass measurements were taken with a precision balance (Metler Toledo XSR205) with 0.01 mg precision. Nine PTFE-α-Al2O3 samples were tested and each were removed between 100 and 100 k cycles (Table 1). All samples followed the same incremental increase in sliding distance until the final sliding distance was reached (i.e., mass was recorded after 5 m, 20 m, 35 m,...) to allow direct comparison between each sample.

Table 1 Sample name, corresponding total sliding distance, and its corresponding distance of the experiment cycles
Full size table

To avoid the effects of transducer misalignment on friction coefficient measurements, the reversal technique introduced by Burris et al. was used to calculate the friction coefficient [40]. Modified Archard’s equation was used to calculate the wear rate (K):

$${\text{K}} = \frac{{\text{Volume Lost}}}{{{\text{Normal Force}} \cdot {\text{Sliding Distance}}}} = \frac{{V_{lost} }}{{F_{n} \cdot d}} = \frac{{\left( {\frac{\Delta m}{\rho }} \right)}}{{F_{n} \cdot \left( {2SN} \right)}} = \frac{{\Delta mL_{1} L_{2} L_{3} }}{{2m_{i} SNF_{n} }}$$

where, Fn = Normal Force (N), Vlost = volume lost (mm3), d = sliding distance (m), Δm = mass loss (g), L1, L2, and L3 are the lengths of the three sides of the PTFE-α-Al2O3 pin (mm), mi = initial sample mass (g), S = stroke length (m), N = number of cycles. Steady-state wear rate was defined using the last three available data points and run with a three-point Monte Carlo test. Uncertainty of wear rate was analyzed using the formula from Schmitz et al. [41].

3.2 Tribofilm Analysis

A KLA-Tencor P-10 stylus profilometer (Milpitas, California, USA) was used to evaluate the height of the transfer film. Three 10 mm scans were taken of each countersurface (one at the center of the wear track and two scans 3 mm away from the center of the wear track). The uncovered surface served as a reference plane and was used to level the data before transfer film thickness estimates were determined. Additionally, optical micrographs from a Bruker Contour GT-I Scanning White Light Interferometer with a 5 × objective were used to determine the coverage of the transfer film using the mean-free-space-length (Lf) method outlined by Ye et al. [15].

The worn PTFE-α-Al2O3 surface was removed from the rest of the pin using a sledge microtome instrument (AO Instrument Co. Model 860, Sr # 48,296, Buffalo, NY, US) for infrared spectroscopic analysis. A microtome blade (Lipshaw, A4333, Detroit, MI, USA) was used to cut 160–200 µm thickness slices of the worn polymer surface, which were used for the ATR-IR and transmission IR spectroscopy analyses. ATR-IR and transmission spectroscopy measurements were completed using a Thermo-Scientific Nicolet iS50 spectrometer. Each individual spectrum is an average of 32 individual scans between 450 and 4000 cm−1. The crystal accessory used for the ATR mode was made of diamond. For transmission IR measurements, the microtomed slices of worn PTFE-α-Al2O3 were mounted on blank IR sample cards to allow the IR laser to pass through the sample to the detector.

4 Results & Discussion

4.1 Tribological Results

Friction behavior for the PTFE-α-Al2O3 composites tested is shown in Fig. 1b. Initially, the friction coefficient was near 0.17 and then varied between 0.19 and 0.22 during the remaining experiments. The observed friction coefficient range along with the trend of slightly increasing friction during sliding agrees with previous experimental results on PTFE-α-Al2O3 systems [21].

Fig. 1
figure 1

a Diagram showing typical tribofilms that form for PTFE-Al2O3 vs 304 SS tribosystem. Tribological results b friction coefficient, and c total wear rate (mm3/Nm) over sliding distance (m). Data points provided are averages of all available samples. Error bars represent the greater value of either variation over the test interval or variation between test samples

Full size image

Initially, the wear rate of the PTFE-α-Al2O3 composites was 3.5 × 10–5 mm3/Nm during the first 100 m of sliding. Increasing sliding distance corresponds to a decrease in the wear rate of the composite, as shown in Fig. 1c. After 5 km of sliding, the total wear rate dropped to 2.0 ± 0.13 × 10–7 mm3/Nm, which is within the range of wear rates reported for PTFE-α-Al2O3 composite materials [21, 35, 42].

4.2 Tribofilm Observations

Initially, there was no transfer film on the 304 SS surface (Fig. 2). As sliding progressed, the tribofilm starts to form and gradually covers the whole wear track. From Fig. 2, the samples with less than 250 m of sliding have very thin, patchy tribofilms. However, on samples with greater than 250 m of sliding distance, we can see a continuous brown tribofilm on the counter surface. This film formation was validated by the mean-free-space-length measurements done on the counterbodies (Fig. 3a). The mean-free-space-length decreases gradually (Lf) as the sliding progresses from 113 ± 52 µm after 5 m of sliding down to 39 ± 11 µm after 250 m. The average mean free space length was in the same range and showed a similar decreasing trend as found by Ye et al. [21]. Stylus profilometry confirmed the existence of the stable film formation (76–148 nm thick) on the counterbody (Fig. 3b).

Fig. 2
figure 2

Images of transfer films on top 304 SS counterbodies after sliding from 5 m to 5 km of sliding. The size of the transfer films is approximately 6.3 mm × 31.3 mm

Full size image
Fig. 3
figure 3

a Mean-free-space-length (Lf, µm) over sliding distance (m). Measurements were taken from one counterbody with 15 representative images used to determine the standard deviation, which were used for the error bars. The reader should note the very high standard deviations before 100 m of sliding are due to statistical variation of the transfer film but at higher sliding distance the uncertainty is due to the range of microscope objectives used in this study. b: Stylus profilometry scans of transfer film height (nm) profile across the wear track. Plots shown are the numerical average of three scans. Legend includes average height value across the transfer film

Full size image

Similar to the transfer film, brown discoloration was observed on the worn surface of the PTFE-α-Al2O3 composite samples (Fig. 4). During the early stage of the experiment (< 250 m of sliding) the surface of the PTFE-α-Al2O3 composite changes color slightly from white to beige. As sliding progresses, we can see the formation of a darker brown, continuous film on the worn composite. After 250 m (5 k cycles), this color remains on the samples for the duration of testing. After 250 m, the wear rate of the composites drops (Fig. 1c) and corresponds both stable and continuous tribofilm present on the worn composite and counterbody.

Fig. 4
figure 4

Images of PTFE-α-Al2O3 contact surface as a function of sliding distance. Each image was taken after the last test increment was completed for that sample

Full size image

4.3 ATR-IR Spectroscopy of Worn PTFE-α-Al 2 O 3 Composite Surfaces

The ATR-IR spectra of worn composite surfaces is shown in Fig. 5a. As sliding progressed, carboxylate salt generation increases. Carboxylate salts are represented by a strong peak at 1641 cm−1 and a weaker peak at 1430 cm−1. Another broad peak in the 3000–3700 cm−1 range is attributed to the waters of hydration (-OH bonds). These peaks, with the exception of CF2 (two peaks at 1144 cm−1& 1200 cm−1), are non-existent in the unworn PTFE-α-Al2O3 [32] and weaker during the early stages of the experiment. This shows the lack of chemistry on the worn polymer surface during the first 250 m of sliding.

Fig. 5
figure 5

a ATR IR intensity of the worn PTFE-Al2O3 surface after 5 m, 50 m, 500 m, and 5 km of sliding. b COOM/CF2 area ratio and its relation to total wear rate as a function of sliding distance

Full size image

Initially, the carboxylate salt (COOM) peaks are very small and become more prominent in the later cycles (peaks @ 1430 cm−1 and 1641 cm−1). This increase in concentration can be tracked by dividing the area underneath of the carboxylate group peaks over the area of the CF2 peaks (1148 cm−1 and 1206 cm−1) as a function of sliding distance (Fig. 5b). Figure 5b also includes plots the total wear rate as a function of sliding distance for the polymer composite. The COOM/CF2 ratio increases as sliding progresses which means more carboxylate groups are present on the worn surface. As the concentration of the carboxylate salt group gets higher, the wear rate starts to decrease. We can also notice a sudden drop in the ratio at 1000 m, which corresponds to an increase in total wear rate. The increased volume lost during this period likely caused the COOM-rich tribofilm to partially wear off which resulted in less carboxylate salt in the spectra. This direct observation in the transition between lower and higher wear corresponding directly to changes in the chemical composition in the spectra has not been reported previously in other work on ultralow wear PTFE systems. This trend supports the importance of tribochemistry to reduce the wear rate of the PTFE composite.

4.4 Transmission IR Spectroscopy

Transmission infrared spectra of the different PTFE-α-Al2O3 samples are shown in Fig. 6. Figure 6a shows the entire spectra from 400 to 4000 cm−1. The broad flat plateau for most spectra between 1100 and 1300 cm−1 represents CF2 bonds within the PTFE matrix. The signal in these cases was saturated due to the thickness of the microtomed samples. The absorbance of the additional peak at 2356 cm−1, often referred to as a “thickness band,” represents the approximate thickness of the microtomed sample [43]. The broad plateau below 900 cm−1 on all spectra is due to metal oxides within the sample. Figure 6b highlights wavenumbers between 1300 and 2000 cm−1, where carboxylate salts and carboxylic acids typically appear within PTFE-α-Al2O3 composites. Around 1790 cm−1, there is a clear peak in all spectra regardless of sliding distance. This peak represents bonded carboxylic acids, which have been attributed to the endgroups of fluoropolymer chains in previous transmission infrared experiments [35]. Past studies on worn unfilled fluoropolymer materials have shown the development of free carboxylic acids, which often occur during sliding [35]. It is likely that these free carboxylic acids still form in PTFE-α-Al2O3 during sliding but bond immediately to the alumina within the surface, forming carboxylate salts. Additionally, the peaks associated with carboxylate salts between 1400 and 1650 cm−1 do not show an increase in strength in the transmission infrared experiments as seen in the ATR-IR measurements. This is likely due to the thickness of the microtomed samples (160–200 µm) probed by the infrared beam. In comparison, the ATR-IR spectra only measures the chemical groups on the first few microns of the surface. It is therefore likely, tribochemical bonds are limited to near the worn surface. This agrees with past literature that used nanoindentation to probe the surface [36]. In that work, the worn surface was found to have increased hardness (250 MPa) at 75 nm below the surface. Deeper indents (~190 nm) of the worn polymer surface showed the hardness decreasing (~87 MPa) to values near those found for the unworn PTFE-α-Al2O3 (85 MPa).

Fig. 6
figure 6

Transmission IR results of the running films

Full size image

4.5 Mechanistic Discussion

The PTFE-α-Al2O3 vs 304 SS tribosystem is remarkable in how it evolves to an ultralow wear rate from moderate wear after a few thousand reciprocating cycles. This system has a variety of factors at play that allow it to reduce its wear rate by 104 times. The PTFE-α-Al2O3 worn surface shows an increase in the concentration of fillers near the surface, as shown by TEM [26] and infrared measurements (Fig. 5a). Additionally, XPS and IR have shown the formation of carboxylate salts at the worn surface, which this study showed correlates well with wear performance (Fig. 5b). This was especially apparent between 500 and 1000 m of sliding, where the drop in carboxylate salt/ CF2 ratio from 0.58 to 0.9 corresponded with an increase in total wear from 3.5 × 10–7 mm3/Nm to 5.9 × 10–7 mm3/Nm. This direct observation of the correlation between the chemical spectra of the worn polymer surface and wear results during the run-in period is novel and illustrates how critical the chemistry of the worn polymer surface is to wear performance. Additional past experiments have shown increased hardness and elasticity of the worn surface [36]. Transmission infrared and attenuated-total-reflectance infrared measurements from this study helped prove that changes to the worn surface are likely limited to the first few hundred nanometers of the composite and sliding does not appear to cause significant chemical changes to the first 100 µm of the worn polymer composite (Fig. 6). In addition to the worn polymer surface, the transfer film is critical to system performance. Even coverage of a thin transfer film has been shown to correlate with lower wear. Infrared and X-ray Photoelectron spectroscopy measurements have shown similar chemistry on the transfer film as observed on the worn polymer surface [24]. Though robust enough to last up to one million cycles, the tribosystem is sensitive to certain factors including surface roughness [19, 45,46,47], filer size [26, 48], filler mechanics [36], and environment [22, 23, 49]. Therefore, this system requires a balance of both chemistry and mechanics to achieve its ultralow wear performance.

5 Conclusion

Nine sample of PTFE-5 wt%-α-Al2O3 were slid against 304 SS countersamples between 5 m and 5 km of sliding. Wear and friction results agreed with values reported previously (K ~ 2 × 10–7 mm3/Nm, µ ~ 0.19–0.22). The transfer film on top of the 304 SS composite showed thin (100–500 nm thick) brown film that became finer (lower Lf) with increasing sliding distance. The worn polymer composite surface showed brown discoloration with increasing sliding distance. Infrared spectra of worn PTFE-α-Al2O3 surface showed formation of carboxylate salts, with an increased concentration after 250 m of sliding. This increase in COOM/CF2 ratio directly correlates with the drop-in wear rate. Transmission infrared confirmed that these chemical changes were local to the worn surface. ATR-IR spectroscopy and transfer film measurements highlight the importance of tribofilm formation on both surfaces in lowering wear of the tribosystem. These results and conclusions of this work provide new findings that show direct correlation between tribochemistry and wear performance during run-in, and emphasize how monitoring the physical and chemical properties during sliding help identify the critical mechanisms for ultralow wear performance..