Most studies on the friction of PTFE have been carried out with unfilled PTFE. Whilst the exact mechanisms involved are still not fully understood a picture emerges in which the ‘dry’ coefficient of friction is dependent upon the pressure, the speed, the temperature, the mating surface, the orientation of the PTFE, the environment and the time of running. Coefficients from 0.016 to 0.36 have been quoted and while this work is discussed in detail on page 27 it may be summarised (with some additional information) as follows:

1. Load and pressure :

   The classical laws of dry friction state that the friction force is independent of the apparent area of contact, making the friction force proportional to load rather than pressure. Many investigators quote the coefficient of friction of PTFE as a function of load and show it to rise steeply at very light loads (below 5 Ib) and decrease with increasing load. R B Lewis does not support this, but suggests the coefficient of friction µ is proportional to the applied pressure P (Ibf / in²) according to the formulae:
   µ = CP^-0,2 where C = 0.12 ± 0.03 at velocities below
   10mm/s (2 ft / min)
   C = 0.35 ± 0.10 at 50mm / s (10 ft / min)
   C = 0.45 ± 0.15 at 500mm / s (100 ft / min)
   and above

2. Speed :

   The coefficient of friction falls markedly at low speeds (below 50 mm / s; 10 ft / min) and increases with increasing speed.

3. Temprature :

   The coefficient of friction appears to be stable over the range -45 to 100°C (-49 to 212°F) but to rise at lower temperatures and fall at higher temperatures.(see page 25).

4. Mating surface :

   Work by Steijn showed that sliding of PTFE against steel gave lower coefficients of friction than sliding bulk PTFE against bulk PTFE. He suggests that when mating areas are large, friction is primarily due to adhesion

5. Orientation :

   It has been shown (see page 27) that the coefficient of friction can be affected by up to 30% depending upon the orientation of the PTFE molecules.

6. Environment :

   Steijn showed that prolonged and continuous running under dry nitrogen (5-10 parts per million of water) gave rise to intermittently high coefficients of friction but this was alleviated as soon as normally moist air (50% r.h.) was admitted. The short term tests at temperatures from -1 to + 60°C (30 to 140°F) in helium, oxygen, nitrogen and air showed no such effect and neither did tests in air at room temperature with relative humidities in the range 12 to 54%. The friction of PTFE in vacuum (10^-9 mm Hg) was studied by Buckley and Johnson who obtained coefficients of friction of 0.25 with a load of 1 kg. They also report the coefficient to be constant over the speed range < 50mm-5 m/s (< 10-1000 ft/min). This high figure could well be attributed to the relatively small loads applied, but may be linked with Steijn’s observations regarding very dry atmospheres. Several investigators have shown that the coefficient of friction is decreased dramatically by the addition of lubricants. This is not surprising since, if a full film of oil is present, the friction is virtually independent of the mating surfaces.

7. Time of running :

   The work of Steijn shows that the coefficient of friction for PTFE on PTFE is influenced by the number of traverses, the time lapse between runs, the nature (especially velocity) of the preceding sliding and the thermal history of the sliding components. Mitchell and Pratt demonstrated a similar increase in friction with time for PTFE on steel, up to a steady level (from 0.05 to 0.20 in 4 hours), and showed this to be due to a change in the surface of the PTFE rather than a change in the surface of the steel (i.e. the transfer of PTFE to the steel).

8. Filler type and volume :

   Thompson et al. suggest that when using molybdenum disulphide (MoS₂), asbestos, carbon, graphite, and copper as fillers, as the volume of filler increases the coefficient of friction increases from 0.016 (no filler) to ~ 0.030 (30% of filler), but that there is little difference in this effect between the various fillers. For a similar range of fillers Milz and Sargent showed the coefficient to be independent both of the type of filler and its volume addition. In particular, MoS₂ and graphite showed no advantage over glass fibre, asbestos and copper. Their results for all types ranged from 0.09 to 0.22 depending on velocity, load, etc. They concluded that the filler was effectively encapsulated and the friction was that of PTFE only. O’Rourke originally came to the same conclusion but later states that friction is dependent more upon the volume than the type of filler, although cadmium oxide is claimed to be an exception. At the very low temperatures of liquid oxygen and nitrogen and under conditions of high vacuum there is considerable variation in the coefficient, but this does not appear to be correlated with either filler type or volume. In practical tests with the Wankel engine using various grades of PTFE as a seal, the coefficient was again found to be independent of the filler, whilst in a laboratory test, Ganz and Parkhomenko state that the type of filler is important; however, they appear to quote the filler content as % by weight so that filler type and filler volume are not separable. They again found MoS₂ and graphite fillers to give high coefficients of 0.26 to 0.34. The evidence of Mitchell and Pratt is that filler type has a greater effect than filler volume, with MoS₂ giving a lower coefficient than unfilled PTFE. They found bronze had little effect and kieselguhr increased it by 25%. Work done on behalf of AG Fluoropolymers (see Table 11) suggests that volume of filler is not directly related to friction coefficient but fillers in general raise the coefficient under these particular test conditions by a factor of about two. It has also been suggested that the addition of MoS₂ and carbon to glass fibre compounds reduces the coefficient of friction, although figures quoted show only a marginal decrease. Tests carried out on behalf of AG Fluoropolymers have not confirmed this and Buckley et al. found no improvement when working under vacuum. Similarly, practical tests showed no advantages for adding MoS₂ to glass although this combination was suggested as a possible means of reducing the scoring of shafts, and for use in very dry gases. It is conceivable that after prolonged continuous running under dry conditions, the MoS₂ is not subject to the rise in friction reported for PTFE. There is therefore conflicting evidence as to the effect of filler type and volume upon the coefficient of friction of PTFE

9. Filler particle size and shape :

   It is difficult to separate the effects of particle size and shape from those of filler type, since specific forms of particle tend to be used with specific types of filler (e.g. glass fibre, irregular particles of graphite and MoS₂, spherical-bronze, etc.). Moreover, in much of the published work no details of filler particle are given. The most explicit work in this field is that of Speerschneider and Li where, with the very hard particles of alumina (Al₂O₃), they found spherical particles gave coefficients of friction similar to that of unfilled PTFE (0.05-0.08) whereas irregular particles gave significantly higher results (0.14-0.15). They attributed this increase to cleavage of the irregular Al₂O₃ which saturates the surface until the coefficient is that of Al₂O₃ on steel. The abrasive nature of the filler also gives a ‘rough’ surface finish to the steel, thereby giving a coefficient approximately double that of a ‘smooth’ steel surface. This effect is less likely to occur with softer fillers, and this has been found true with bronze, where no difference in friction has been found between spherical and irregular particles, although Thompson et al. suggest that particle size can have an effect in extreme cases.

Table 11. Dynamic coefficient of friction for various Filled PTFE grades

Grade	63% Bronze+Graphite Filled PTFE	15% Glass Filled PTFE	25% Glass Filled PTFE	60% Braonze Filled PTFE	25% Carbon Filled PTFE	15% Graphite Filled PTFE	15% Glass+5% MoS2 Filled PTFE
Coefficient of friction	0.20	0.10	0.11	0.16	0.17	0.22	0.19

Test conditions
Mating surface: 420 S 37 steel in T condition (BS 970:Part 4:1970)
Surface finish: 0.3 µ m R (CLA) BS 1134:1972
Pressure: 20 kgf / cm² (300 Ibf / in²)
Speed: 0.02 m/s (50 ft / min)

10.  Other parameters (load, speed etc.) :

    The statements made in the first part of this section for the effects of load, speed, temperature, etc. upon the coefficient of friction of unfilled PTFE in general hold good for filled PTFE, although O’Rourke shows that the coefficient of glass-filled PTFE does not rise at low loads, whilst other investigations suggest that it does.

11.  Environments :

    Work with filled PTFE at low temperature and in contact with liquid oxygen and nitrogen shows the coefficient to rise with the passage of time (e.g. 0.18 to 0.43 in 23 hours), which tends to confirm the work of Steijn with unfilled PTFE. This same effect at room temperature has been found by work done on behalf of AG Fluoropolymers and by Mitchell and Pratt although actual coefficients are lower (0.07 to 0.20 in 20 hours). There is some evidence therefore that the coefficient of friction increases in the presence of liquid oxygen or nitrogen. High coefficients (0.2 to 0.4) were also found by Buckley et al. for filled PTFE under high vacuum, but some of the fillers, notably copper, silver and powdered coke gave coefficients lower than for unfilled PTFE under the same conditions. The reasons for these effects are not known: the effects may be due to temperature or environment, or the mechanisms may be similar to that experienced with graphite where the low coefficient of friction is attributed to the presence of adsorbed gases at the crystallite interfaces where cleavage occurs.

1. Wear mechanisms :

   The mechanisms responsible for the wear of PTFE are not fully understood, but it is generally thought that adhesion and the freeing of transferred wear fragments, either in terms of surface energy or by virtue of fatigue, are of major importance. It is known that when PTFE is rubbed against other materials a transfer takes place and it is believed that the wear process involves the laying down and subsequent removal of such transferred layers. An ideal situation is given as having a highly oriented mono-molecular layer of PTFE bonded to the metal surface which then rubs against as smooth a mating surface of PTFE as possible. What is not clear is exactly how and why fillers and conditions affect both the initial laying down and subsequent removal of the PTFE particles. It is suggested that a minimum temperature at the interface is required to promote adequate bonding and that certain fillers function by causing frictional heat. It is also clear that surface finish will affect this transfer, and whilst there is wide agreement that too rough a mating surface will cause rapid wear, one school of thought suggests that too smooth a surface finish leads to high wear rates, while others suggest that is not so. The answer may be that although too fine a finish may well inhibit good transfer, many filled compounds are sufficiently abrasive to roughen the mating surface adequately. However, if the filler or environmental conditions are too abrasive, rapid wear will occur through ploughing. The entrapment of wear debris can have a similar affect,. It has been suggested that chemical reactions at the interface may be important. Buckley and Johnson consider that wear is related to the decomposition mechanism and hence to the temperature at the interface, while Hargreaves and Tantam suggest lead oxide can be an oxygen carrier to other metals, giving selective oxidation of roughnesses on the mating surface. Mitchell and Pratt have noted the formation of copper fluoride at the interface of bronze-filled PTFE, presumably caused by local degradation of the PTFE and bronze. They do not, however, attribute the reduction in wear accompanying the formation of copper fluoride to the chemical action, but rather to the fact that the area of contact at the interface increases with time, which reduces the interface temperature. Vinogradov did however attribute a reduction in friction between copper and PTFE to the formation of the solid lubricant copper fluoride.

2. Load and speed (Pressure x Velocity) :

   The most widely quoted formula for the wear of filled PTFE is that of Archard and Hirst which states that the volume wear (W) is proportional to the relative speed at the interface (V), the load supported (M) and time run (T), ie,
   W ∝ MVT
   or using the ‘wear factor’ K
   W= KMVT (Equation 1)
   By dividing by an area (A), a linear wear (R) is obtained such that
   R = W / A = KVT M / A ;
   but M / A is the pressure applied (P)
   
   R = KTPV (Equation 2)
   Equation 2 suggests that the wear rates of materials can be classified in at least three ways:
   (i) By quoting the maximum PV value the material can withstand (termed limiting PV value). This is found experimentally either by determining the maximum load that can be applied at constant velocity while still maintaining temperature and/or frictional torque equilibrium, or by determining the PV value at which the wear rate suddenly increases. (ii) By quoting the PV value which gives a specified wear rate. This is generally quoted as the PV value to give 0.127mm (0.005 inch) wear in 1000 hours, and is determined from specific wear tests. (iii) By determining the constant K in Equation 2. This is known variously as the ‘wear factor’ or ‘K-value’. Work has been carried out to determine these factors for various filled PTFE materials. However, during these and subsequent investigations it has been shown that there are two major errors in accepting results derived from Equation 2: (a) K is not necessarily a constant for a given material, but will vary with the load applied, the velocity, the length of time run, the temperature and other factors such as clearance and conditions at the interface. (b) The method of determining the factors is invariably to test specimens. The conditions of test have a considerable effect upon the results and universal values cannot be obtained from one series of tests.

3. Temperature :

   R B Lewis suggests that each material has two wear rates (metric), mild wear (K~ _ 7) and severe wear (K > 35) which is attributed to a rise in temperature at the interface. The actual temperature at which the transition occurs is reported to depend upon the load. He concludes that the PV value at which transition begins depends upon the application geometry, ambient temperature, and manner and amount of cooling, whilst the slope of the transition depends upon the application parameters and properties of the compositions. The mild wear is reported to be characterised by wear of the surface layers whilst severe wear is characterised by bulk removal of material. Similar conclusions were drawn by Summers. Smith who considered the composite to be hard granules in a ‘cement’ of softer materials suggests that the mild wear region corresponds to a gradual attrition of the hard granules whilst the change to severe wear occurs when the ‘cement’ becomes softened by heat and the granules are plucked bodily out of the matrix. From work carried out by Mitchell and Pratt and work done on behalf of AG Fluoropolymers, it is concluded that the entrapment of wear debris as well as surface temperature is a very important factor in determining whether severe wear occurs or not. For example, although the difference in running conditions between a thrust washer and a piston ring is mainly considered to be one of interface temperature, it is also true that wear debris is far less likely to become entrapped in the piston ring. It is also true that differences in wear rate can be attributed to differences in behaviour (abrasive or otherwise) when trapped wear debris is present. Nevertheless, whatever the mechanisms, it is generally accepted that an increase in interface temperature increases the wear rate.

4. Mating surface and material :

   At room temperatures and above it is generally agreed that a hard, approximately 900 VPN (Vickers Pyramid Number), mating surface is beneficial. Softer materials can be used providing the filler will not abrade them. The materials with good dry bearing properties of their own (e.g. bronze) are preferred to the softer more easily damaged materials (e.g. aluminium). There is some divergence of opinion as to the suitability of chromium plating. Prattshows chromium plating to be advantageous whereas O’Rourke et al. show it to give poor results. The answer might well be that the fillers used by Pratt were less abrasive than those used by O’Rourke The surface finish of a material is generally quoted as a mean centre line average - CLA - or Ra (BS1134-2: 1990) of the ‘peaks and valleys’ of its surface as detected by traversing a diamond stylus across it. This does not fully specify a surface however, since a turned and a ground surface of the same Ra value will be different. It is now generally accepted that a ground surface is superior to a turned surface and that above 0.75 µ m the wear rate of the filled PTFE will increase. The existence of a lower limit is still in dispute and so the best compromise is to use a ground surface finish of 0.2-0.4 µ m.

5. Lubrication :

   ‘Lubricant’ is a very general term and it used to be stated that any liquid will act as a lubricant and be beneficial to PTFE. To some extent this is true in that, if hydrodynamic conditions are established, no wear will take place, but filled PTFE may run under conditions of boundary lubrication. Hydrocarbon oils are generally advantageous, with a significant reduction in wear rates. This is not so with water. O’Rourke confirmed that the wear factor increased for unfilled and various filled PTFE compounds when running against steel with water boundary lubrication. Work done on behalf of AG Fluoropolymers has shown that boundary lubrication with water gave a reduction of wear life of 50% when filled PTFE ran against steel.