Introduction

There are a number of applications where operation in a temperature range from 200 to 400 K or larger is required for device success [1–3]. These extreme conditions are often the motivation for variable temperature studies in tribology [4–7], but there is a paucity of relevant tribology data available for temperatures below 273 K. In the range from 300 to 400 K the friction coefficient of PTFE increases with decreased temperature and increased speed. This behavior has been interpreted as evidence for a viscoelastic friction dependence of PTFE since McLaren and Tabor [8] first proposed the hypothesis in 1963. It is well-known that PTFE films transfer and adhere strongly to the counterface, and a modern hypothesis is that both friction and wear of PTFE are dominated by the interactions of PTFE chains sliding past one another at weak self-mated interfaces, often within transfer films [9–15]. A recent paper by McCook et al. [4] found that friction of PTFE matrix composites continued to increase in the cryogenic regime down to 200 K, and the notion of a thermally activated friction coefficient for PTFE was proposed (analysis of an activation energy gave E a = 3.7 kJ/mol).

A recent molecular scale study of graphite by Zhao et al. [16] provided evidence that the temperature dependent friction found for PTFE may be a general result of interfacial sliding. In these studies, an atomic force microscope was used to collect friction data on molecularly smooth terraces of graphite over a temperature range from 140 to 750 K at a vacuum level of 2 × 10−10 torr. The friction coefficient increased with decreasing temperature, and the data collected followed an Arrhenius dependence with an activation energy of E a = 9.6 kJ/mol. The molecular scale experiments by Zhao et al. [16] addressed many of the uncertainties raised in the macroscopic experiments conducted by McCook et al. [4]; namely, the sliding interface was well-characterized, interfacial sliding was confirmed, and the experiments were run in ultra-high-vacuum at temperatures well above the temperature for equilibrium ice formation on the surfaces (frost-point).

It is difficult, if not truly impossible, to prescribe interfacial sliding and zero-wear in macroscopic tribological testing. Under conditions where wear, viscoelastic losses, and plastic deformation are negligible, interfacial sliding forces may be captured in the global force measurements. Solid lubricant films and composites of materials such as MoS2, PTFE, and graphite, operating in their lowest wear environments and preparation, provide the best opportunity to make fundamental measurements of interfacial sliding forces in macroscale tribology testing.

A molecular view of thermally activated friction is one where the thermal energy blurs the potential surfaces giving rise to a reduction in the effective barriers for sliding with increased temperature and vice versa. In systems with wear, athermal behavior of friction is often reported; this may result from a cancellation of thermally sensitive material properties that govern both the deformation resulting in wear and friction or temperature independent mechanics. In the schematic shown in Fig. 1, the steep curve from low to high friction is indicative of thermally activated behavior. At elevated temperatures, thermodynamic, chemical, and physical properties of materials will dictate changes in the friction behavior from that predicted by the Arrhenius relationship. Suggested transitions from Arrhenius behavior in the realm of rapidly increasing friction at low temperature may exist due to interfacial forces that reach such a high level that alternative athermal mechanisms of motion become more favorable. In macroscopic testing, the solid lubricant systems appear to transition into a mode of fine wear debris generation at the lowest temperatures examined, which is coincident with the greatest deviation from an activated trend.

Fig. 1
figure 1

A schematic for the thermally activated hypothesis of friction coefficient as applied to macroscopic tribology experiments for materials that accommodate slip through interfacial sliding. The temperature independent friction coefficient and macroscale limits are likely the results of opportunistic modes of accommodating slip in tribological systems, which prevent direct assessment of interfacial sliding forces. Upper temperature limits and theoretical molecular scale postulated limits are noted.

Full size image

The goal of this study was to improve our understanding of thermally activated friction of PTFE. In the experiments by McCook et al. [4], the role of ice could not be quantified, and the effectiveness of the liquid nitrogen boil off to prevent ice deposition at cryogenic temperatures was uncertain. Data were collected here at elevated humidities over a similar temperature range to study the potential role of ice on cryogenic PTFE sliding interfaces. The results demonstrate that the experiments by McCook et al. were likely free of ice contamination at the tribological interface, and provide further evidence that interfacial sliding is a thermally activated process.

Results and Discussion

The reciprocating tribology experiments described here (6.3 MPa contact pressure, 50 mm/s sliding speed) were run in an open gaseous environment of nitrogen and water vapor with the temperatures of the steel counterface being prescribed by a copper heating and cooling block. The experimental details are described in the methods section and are fundamentally different than the experiments published in McCook et al. [4], which used liquid nitrogen gas boil-off in an impinging jet configuration to cool and inert the tribological surfaces. The results of friction experiments run at 2% and 6% relative humidity are given in Tables 1 and 2 respectively. In this experiment no effort to inert the environment below the ambient water vapor content was attempted, and the cooling of the sample did not rely on an impinging jet of volatilized liquid nitrogen (i.e., the temperature and environmental control are decoupled). These data are also plotted in Fig. 2, where the lines and arrows denote the order in which the data has been collected (the sequential data is given in Tables 1 and 2). For cryogenic experiments, temperatures were ramped from ambient to sub-ambient and back.

Table 1 Results of 2% relative humidity variable temperature experiments
Full size table
Table 2 Results of 6% relative humidity variable temperature experiments
Full size table
Fig. 2
figure 2

Friction coefficient plotted versus interface temperature for unfilled PTFE in tests varying interface temperature. Tests were conducted on a 304 stainless steel counterfaces in a nitrogen environment at two humidity levels. Sliding speed was 50 mm/s over a 25.4 mm reciprocation length. The normal load and humidity were held constant at 275 ± 2 N and 1.85 ± 0.13% at the low humidity condition and 273 ± 2 N and 5.5 ± 1.1% at the high humidity condition. The background temperature inside the glove box was held constant at 296 ± 1 K. The dew points of the 2% and 6% relative humidity experiments were calculated to be 243 K and 255 K, respectively.

Full size image

For the case of 2% RH, the friction coefficient climbs as the temperature is reduced to 225 K, which corresponded to visible ice formation on the counterface (initially shiny but becoming hazy); deviations from this trend were observed during the room temperature phase transitions [17–19] and during a hypothesized glass transition near 400 K [17, 20–22]. At 2% RH with a background temperature of 296 K, the frost-point is approximately 243 K. As the temperature was further reduced below 225 K, the friction coefficient continues to fall, which is presumably due to a more contiguous film of ice on the surface. Upon heating from 188 K, the friction coefficient exhibits significant hysteresis, presumably because the ice film is intact and the trend is that of PTFE on ice—at a temperature above the frost-point (250 K) the friction coefficient rejoins the original trend.

For measurements at 6% RH, values of friction coefficient followed the previous experiments conducted at 2% RH until crossing the 6% RH frost-point at 255 K, where a sharp drop in the friction coefficient was observed. Upon further cooling, the friction coefficient resumed a monotonic increase, eventually overlapping the heating data (thought to be PTFE/ice) collected at 2% RH. During heating from 225 K the friction coefficient showed little hysteresis (a deviation thought to be associated with ice melting was observed just below 273 K). Collectively these results suggest that below the frost-points ice formed on the wear tracks—perhaps in a competitive rates fashion [23, 24]—and that the presence of ice provides a weaker pathway to interfacial slip.

The relative change in friction coefficient, μ*, is computed by normalization with the ambient temperature datum. The relative change in friction coefficient from the variable temperature experiments of PTFE presented above and those of previous studies have been calculated and plotted versus temperature in Fig. 3. It should be mentioned that a recent study by Burton et al. [26] completely spanned the range from 4 to 450 K, but due to transients in setting up transfer films and the velocity dependence of PTFE, this data is not included as the single pass experiments were done under variable speed.

Fig. 3
figure 3

Normalized friction coefficient plotted versus interface temperature for variable temperature studies of PTFE in the literature: (a) McCook et al. [4] (b) Bowden [25] (c) Pooley and Tabor [10] (d) Blanchet and Kennedy [13] (e) McLaren and Tabor [8] (f) Tanaka and Uchiyama [12] (g) Pleskachevsky and Smurugov [14]. This dataset suggests that the friction of PTFE is a thermally activated process with an activation energy near 5 kJ/mol over the temperature range from 200 to 425 K. The results collected in the current study are connected by lines.

Full size image

The effect of ice formation on the frictional behavior in macroscopic PTFE sliding experiments is clear—deviations from Arrhenius behavior are observed when the surface temperature is decreased below the temperature for ice formation. The temperature at which such deviations occur is related to the kinetics of equilibrium ice formation. Based upon these observations, the data of McCook et al. [4] can be seen to be free of ice contamination down to 200 K. A subtle deviation from the more global trend is observed at 173 K. Any future macroscopic studies designed to interrogate this regime will likely require cryogenic vacuum instrumentation. This combined data set is well fit to a thermally activated process with an activation energy of 5 kJ/mol, as shown in Fig. 3.

Conclusions

  1. (1)

    In the absence of ice, the friction coefficient of PTFE increases monotonically with decreasing temperature.

  2. (2)

    The friction results collected here are in excellent agreement with data collected during previous investigations of PTFE under different experimental setups. A fit of activation energy E a = 5 kJ/mol over the collective dataset suggests thermally activated friction.

  3. (3)

    The presence of ice at a nascent PTFE interface provides a lower energy pathway to interfacial sliding. The friction coefficient of this interface also tended to increase with decreasing temperature.