Dry bearings: a survey of materials and factors affecting their
Transcription
Dry bearings: a survey of materials and factors affecting their
Dry bearings: a survey of materials and factors affecting their performance J. K. Lancaster* Following a general discussion of performance criteria and testing of dry bearings, the various materials currently available commercially are classified into four general groups - polymers, carbons-graphites, solid film lubricants and composites, and ceramics-cermets. The material properties relevant to bearings are discussed in detail for each group, and the special features required in design are noted. Finally, an attempt is made to develop a selection procedure for dry bearings, based on identifying the major requirements in a given application. Applications ofunlubricated 'dry' bearings have been expanding rapidly in recent years, and there are three main areas in which their use is indicated: a where fluids are ineffective, as at low or high temperatures, or in reactive environments, b where fluids cannot be tolerated because of the possibility of contamination of the product or the environment, c where fluids are undesirable because of lack of opportunity for. or the impossibility of, maintenance. Cost, although always an important consideration, is not usually the decisive reason for choosing a dry bearing. Some of the most successful dry bearing fornmlations are significantly more expensive than their mass produced metallic counterparts intended for lubricated service. Most of the discussion in this survey is concerned with sliding bearings, and for these there are two primary property requirements. Firstly. the materials must be able to support an applied load in the environment concerned without significant distortion, deformation or loss in strength. Secondly, both the coefficient o f friction and the rate of wear must be acceptably low and preferably also insensitive to minor changes in the conditions of sliding likely to be encountered, eg temperature, humidity, contamination, etc. Four groups of materials satisfy all, or most of these requirements. The largest group consists of materials based on synthetic polymers to which are added various fillers or reinforcements intended to enhance particular properties. The second group, in order of general usage, comprises carbons and graphites, together with additives, and these are particularly important for applications at temperatures above those which can be tolerated by most of the common polymer-based products. Thirdly, for applications in which a low coefficient of friction is the main requirement, solid film lubricants, based on ptfe or lamellar solids such as graphite or MoS2, can be used in * Materials Department, Procurement Executive, Ministry of Defenee, Royal Aircraft Establishment, Farnborough, Hampshire, England conjunction with suitable metallic or non-metallic substrates. Finally, for high temperature applications (above about 400°C) where the magnitude of the wear rate is usually the major consideration, a variety of hard metals, ceramics and cermets is available, either in bulk form, or as coatings on a metallic substrate. The properties of a variety of materials falling within these four groups are discussed in detail in this survey and their advantages and limitations for dry sliding bearings are compared and contrasted. Some of these materials can also be used as constituents of rolling element bearings intended for operation in the absence of fluid lubricants and this aspect is also examined. First, however, it is pertinent to discuss some of the more general features associated with the operation of dry bearings and, in particular, the problem of wear. Performance criteria Strength There is no single strength parameter which uniquely defines the load-carrying capacity of a dry bearing. During sliding, both tensile and compressive stresses are present within the contact area, and shear stresses within the subsurface layers. A widely-used limit is one-third of the maximum compressive stress, but this is relevant only to those materials which exhibit similar tensile and compressive properties. For carbon/graphites and ceramics, the ultimate tensile strengths may be very much lower than the compressive strength. A further complication which can arise with polymer-based materials is visco-elastic behaviour: the dependence of mechanical properties on time. The stress-strain relationships at any particular rate of loading are non-linear and even under static loading creep or permanent deformation may occur. Ptfe and many of its composites are particularly prone to the latter. Although carbons and graphites do not exhibit visco-elasticity, the stress-strain relationships of these materials also tend to be non-linear. It is therefore impossible, in general, to quote unique values of elastic moduli for many of the materials of interest in dry bearings. TRIBOLOGY December 1973 219 Fortunately, except in a few specialized applications such as air-frame bearings, the stiffness o f a dry bearing assembly is seldom a critical parameter. F o r design purposes it is usually adequate to use tensile or flexurai moduli obtained b y conventional test methods. Wear 13y far tb.e greatest uncertainty associated with the operation o f dry bearings is wear. Eoth simple theory and numerous experiments have shown that once the surface conditions during sliding attain a steady-state the volume of wear. v. is proportional to the distance of sliding d. Further, if changes in the applied load, g', do not cause significant changes in any other variable, and particularly in the surface temperature, then the volume of wear per unit sliding distance is directly proportional to the load. Thus, ,, = k W d (i) The constant k is usual!y called the 'specific wear rate' and is basically a material property. However, because the two assumptions made during the derivation of Equation 1 are not always valid, the value o f k may depend on the conditions o f sliding and on the precise geometrical sliding arrangement used to determine wear. It is this last complication which m a y give rise to uncertainties when attempting to extrapolate the results of accelerated laboratory wear tests to the practical case. The specific wear rate, k = v/Wd, has the dimensions of (stress) -1 , but it is more meaningful to use units whose physical significance is more readily apparent - mm3/N m. However, in the operation o f a journal bearing, it is not usually the volumetric loss of material which is the important factor but the radial wear h, which leads to increased Table 1 Conversion of wear rates and PV factors into Sm units from British, American and non-SI metric units in 3 Specific wear rate 10 - 1 2 10 -13 mm 3 = 1.2 X 10 . 8 - ft - lbf Nm in3 = 1.4 X 10 - 8 ram3 in - lbf Nm cm 3 ~ 1 0 - 8 m m 3 10-12 _ _ cm - kgf Nm Wear factor 10-10 10 - 8 PV factor in3 _ rain mm 3 -~ 1.9X 1 0 - 8 - ft - l b f - h Nm cm 3 - rain mm 3 ~ 1.7X 10 - 8 m - kgf- h Nm lbf ft MN m 1000 i ~ X min-- -- 0.035 ~ m2 X -s kgf m MN m t00-X--~0.17 Xcm 2 rain m-2 s 220 TRIBOLOGY December 1973 Stdtic l o a d - c a r r v i n c D_ CD O d Log V Fig 1 Relationships between P and V for dry bearings clearance. To a first approxhmation, h = via where A !s the projected area of contact, and hence h = kPd where P is the nominal applied pressure. If the assumption is now made that changes in speed. V. do not cause changes in other variabIes, and in particular the surface temperature. then d = Vt and h =kevt (2) The constant k = h/PVt is numerically the same as the specific wear rate defined above, but when used in the context o f radial wear is often referred to in US literature in 3 mm as a w e a r factor' and expressed in units ff I b f - h For convenience, some conversion factors between the various units, Imperial. Metric and SI. are given in Table i. PV factors Expressing the wear rate of a bearing in terms o f radiai wear per unit time, Equation 2 leads directly to the con° cept o f a PV factor as a performance criterion for bearings. PV factors are widely quoted in the literature and m a y take two forms: a the 'limiting PV' above which wear increases rapidly either as a consequence of thermN effects or of stre> sos approaching the elastic limit, b the P V f a c t o r for continuous operation at some arbitrarily specified wear rate, eg 25 # m / 1 0 0 h In neither case is the PV factor a unique criterion of performance because the assumptions made in the derivation of Equations 1 and 2 (changes in P and V do not introduce changes m any other variable) are only usually valid ave: a very restricted range of P and V. In Fig i, curve (a) shows a schematic relationship between P and V fo~ constant, wear rate and it m a y be noted that there is an inverse propor tionality, ie a c o n s t a n t P V factor, only over the central range o f P and V. At low speeds, the m a x i m u m pressure which can be used is limited b y the strength of the material and as this pressure is approached the specific wear ra-Ee no longer remains independent of load but begins to increase. The surface temperature m a y also increase and #lese two factors manifest themselves in Lhe P - V curve for constant wear rate as an asymptotic approach to a limiting P as V decreases. At high speeds, the generation of frictional heat again raises the temperature of the surface layers and this also tends to increase the specific wear rate. Consequently, there is an asymptotic approach to a limiting V as P decreases. Similar considerations apply to the 'limiting P V' relationships. The general shape of the curve is the same, see Fig 1 curve (b) but is displayed to higher values of P and V. The extent of this displacement depends on the particular material involved and the way in which its specific wear rate depends upon temperature and stress. Because of the general non-tinearity of P - V curves, design information for dry bearing materials is best expressed by presenting the complete P - V relationship. However, data for many commercially available materials are insufficient for this purpose and the only information often available is a PV factor at one or two speeds. In the latter circumstances, extrapolation of the data to other sliding conditions can be somewhat uncertain. Temperature As already mentioned, the major complicating factor responsible for the deviations of the P V relationship from linearity, particularly at high speeds, is the temperature rise due to frictional heating. In view of the importance of temperature, it is worth devoting some attention at this stage to ways in which its magnitude can be estimated. The temperature developed at the localized asperity contacts involves two components: Ta, the mean temperature of the interface between the bearing and its journal resulting from the general dissipation of frictional heat around the system. The value of T a depends on the geometrical construction and materials of the whole bearing assembly. Tf, the 'flash' temperature rise at the localized asperity contacts, which is largely independent of geometry. Thus, T = Ta + Tf. Neglecting heat losses, the mean surface temperature Ta is proportional to the total energy dissipated. T~ = To + CuIt'V (3) materials whose thermal conductivities are very much less than those of a mating steel journal, eg polymers, Tf = 1 X 10 2 gp~2w1/2 V for low apeeds, (4) and Tf = 5.7 X 10 -5 p.p~4w1/4V1/2 forhigh speeds.(5) Pm is the flow pressure of the bearing material and is approximately equal to the indentation hardness. W is the absolute load applied. The two separate cases, low and high speeds, arise because at low speeds the frictional heat is shared between the two sliding components and time is available to reach thermal equilibrium. At high speeds, the thermal conditions are transient, and the majority of the heat generated is conducted into the moving shaft. For bearing materials whose thermal properties are similar to those of a steel journal, eg carbons-graphites, tile values of Tfat low speeds are about half those given by Equation 4, t at high speeds they remain the same as those given by Equation 5. As mentioned earlier when discussing the mean surface temperature rise, the thermal conductivity of tile journal is a much more important factor at high speeds than that of the bearing material itself. Finally, it should be noted that the temperatures derived are the maximum possible, assuming that the whole of the applied load is supported by a single contact. In practice, this will not be so and there will be a distribution of lower asperity temperatures depending on their number, size and proportion of load supported by each. It was also assumed in the derivation of Equations 4 and 5 that the hardness of the bearing material did not vary with temperature. ~Nilst this is generally true for carbon/graphites and ceramics and cermets, the hardness of polymers varies appreciably with temperature. To a first approximation, the variation is exponential r m = Po e aT where Po is the room temperature hardness, and a is a material constant of the order of 0.005. The effect of hardness variations will be most pronounced at higll speeds and for this situation Tf = T/oe 0.0038Tf where 1o is the temperature of the environment, # is the coefficient of friction and C is a constant characterizing the thermal properties of the materials and the particular geometrical configuration. Values of C usually lie within the range 0.1 1°C s/N m, and a typical value for 25 m m × 25 mm polymer-based bearings sliding against a steel journal is about 0.5°C s/N m. In general, an increase in the thermal conductivity of the journal material causes a much greater reduction in the value of C than a corresponding increase in the conductivity of the bearing itself. For any geometrical bearing arrangement, the mean surface temperature, Ta, is easily measurable with a surface thermocouple. The flash temperatures at the asperity contacts, however, cannot be measured in this way and it is only possible to make theoretical estimates. A convenient method is to apply Jaeger's 1 analysis for two semiinfinite bodies in relative motion making contact over a small area of square cross-section. Simplification of the relevant fommlae has been discussed elsewhere 2 and it is sufficient here merely to quote the end results. For where Tfo is the temperature derived from Equation 5, assuming no change in hardness with temperature. In summary, Fig 2 gives curves for T¢'/I~ against W 1/2 V for materials of different hardnesses. The full lines are for polymers whose hardness varies with temperature and whose thermal conductivities are negligible in comparison with the steel counter-face. The dotted lines are for carbons-graphites whose hardness is constant and whose thermal conductivities are comparable with those of the steel journal, and the hatched lines are for ceramics and cermets, again with constant hardness, but with negligible thermal conductivities compared to steel. It may be noted that despite the higher thermal conductivities of carbonsgraphites, the values of Tf at high W 1/2 V exceed those for polymers of similar hardness because thermal softening does not occur. Comparison of the expressions for the mean temperature rise, Equation 3, and the flash temperature rise, Equations 4 and 5, leads to the general conclusion that at high speeds and low loads, the temperature limit of TRIBOLOGY December 1973 221 I- ./. 103i . ' / C .... iC cerrnets I , \ "~ ] [0 -3 I / ~ z l [0 .2 PI l IO -I / / /5o0 / ...~ . " / ," . / / ,,// ." I I~ -] I I I I '1 I0 2o~11 ~L4 . " ~----~ ~/ ~ 8 0 20 ~ I I 5~ interest, and this can be measured mos~ conveniently by monitoring the relative movement of the bearing 'Mth respect to a fixed datum. The housing or base-plate supporting the rolling-element bearings of the shaft o h e n fulfils this function. For comparative purposes, it is aIso desirable :o compute the specific wear ra~e e ~ J l e matefiai. With thrust-beariv~gs there is no oroblem, but with jour~,~a] bearings the depth of wear does r.,ot become directly pro° portionat ~o the volume until the width of the wear scar becomes equal to the bear/ng diameter~ Until this stage is reached, the volume of wear, v. is given by: I II 1 10 2 W½V(N½m/s} Fig2 Flash temperatures during sliding on steel. Fulllines, polymers; dotted lines, carbons-graphites; chain lines, ceramics-cermets. FigUres adiacent to each line are hardnesses, VPN a bearing is mainiy the result of the flash temperature, whereas at low speeds and high loads the mean temperature rise may become more significant. Clearly, therefore, the temperature conditions attained during the operation of a particu]ar size of bearing are dependent upon the individual values o f P and V and not direct1?} on the PV product. It is also apparent that increasing the size of a bearing for constant P and V will again affect the temperature conditions because of the changes in absolute load involved in maintaining constant pressure. Testing The most satisfactory way in which to assess the performance o f any dry bearing material is undoubtedly a practical trial in the intended application. However this is often impractical for reasons o f time, cost or ]ack o f opportunity, arid in such circumstances it is necessary to resort to laboratory testing. A variety o f simplified bearing test rigs has been developed, and two examples are shown in Fig 3. Ideally, all test rigs should be able to m o n i t o r continuously the frictional torque, wear and temperature, and should also be capable of operating over a wide range of loads, speeds and ambient temperatures. As yet there has been no agreement on a standard size for dry bearing tests and various sizes between 12.5 and 25 m m bore are used b y manufacturers of different products. P.ecause of the complicating effects o f temperature already described, it is not always valid to compare the reported performance of bearings of different types and sizes. Determination of the limiting P V curve for a partic'alar material is a relatively simple matter. A constant velocity is chosen and the ioad on the bearing is then increased ir~ stages, sufficient time being allowed at each stage for the friction and temperature to stabilize. Fig 4 shows the results of a typical experiment of this type. At high loads it is no longer possible for the friction or temperature to stab~ize, and the load corresponding to the last region of stab~ity is taken as the limiting value. Repetition o f the whole procedure at different speeds enables the limiting P V curve to be built up. The determination of P - V relationships corresponding to a specified wear rate is much more time-consuming as it involves measurements o f the rate of wear over a wide range o f loads and speeds. As already mentioned, the depth or radial wear rate of a dry bearing is of particular 222 TRJBOLOGY December 1973 where D is the shaft diameter, l is the bearing length, h is the maximum de~th of wear and e is the il~itial clearance. Values of the fn (h/e) are shown in Fig 5. Measurements of the depth of wear are sometimes complicated by a deformation componen'; which may be significant for the more elastic materials in the early stages o f sliding when linecontact conditions prevail. Deformation wil! not, however. normaity enter into the determination o f the steadyosta~e wear rate. Because wear-testing of bearings can be very rimeconsuming, a great deal of the available data on the wear of dry bearing materials has been obtained from accelerated laboratory tests with simpler geometrical configurations. In general, these take the form of a smal] rider c f the bearing material loaded against a larger, moving, metal surface. The essential feature of this type of test is the non-conforming geometry, and a number of different T a N e 2 Retationshipsfordetermimngwearvo!umesin accelerated wear tests of the pin-ring and pM-disc type Plane-ended cylinder on ring v~ [ d2 6R - a : cylinder radius R = ring radius d = scar width ~r Crossed cylinders 3a2 "\~ d3 1 v ~ /~- 64R J R .[ d~ i d2 [ 1+--16 ~ b k 7 - a = small cylinder radius R : large cylinder radius d = long dimension of elliptical scar 7i" Cone on V~ d 3 cot 0 24 disc 0 = semi-apex angle of cone d = scar diameter 7/ Sphere on disc V~ d4 + - - 64R R = s~here radius d = scar diameter d2) 12 R ~ o] Fig3 a Journal bearing test r i g - DowCorningLFW5 b Thrust bearing test rig - Dow Coming LFW6 load, speed, temperature, environment, counter-face inetal and counter-face roughness. The major differences in test conditions arise as a consequence of the non-conforming geometry which is deliberately used to increase the sensitivity of the wear measurements. With non-conforming geometry, potentially abrasive wear debris can readily escape from the contact zone without adverse effects, whereas in a bearing the debris may contribute significantly to increased wear. Further, the contact stress for a given load is often greater in non-conforming geometry than in a bearing, and may thus lead to material breakdown in the former but not in the latter. These effects will be discussed in more detail in the various sections devoted to particular groups of materials. It is worth noting here, however that despite these differences there can nevertheless be a very specimen configurations has been used. A selection is shown in Fig 6. The end of the rider may be shaped to enable the volume of wear to be computed from the dimensions of the wear scar developed, and fornmlae for some of the more common arrangements are given in Table 2. It should be mentioned that determinations of the wear volume from the dinnensions o1 wear scars is accurate only if the elastic modulus of the material being worn is sufficiently high to ensure that elastic recovery is negligible following removal of the load. The limiting modulus is about 0.2 GN/m 2. and below this value it becomes necessary for wear measurements to be made from weight losses. Accelerated wear tests are often considered to be an acceptable alternative to full-scale bearing testing in the early stages o f materials selection or a development programme. However, there is still a considerable amount of controversy about the value of accelerated tests in predicting the wear behaviour o f bearings in practical conditions. In an accelerated test there is little difficulty involved in simulating the practical operating parameters of absolute 200 % Pressure 150 ,oo~~, ~ 1- 50 , \ Failure Temperature Friction ___L I 1 2 3 4 5 Time (h) Fig4 Typical variation of friction and temperature during the determination of 'limiting PV' for a polymer-based bearing (data from Willis;) TRIBOLOGY December 1973 223 is too great for accurate predictions of bearing wear to be made from the accelerated tests, the correlatior_ is sad ficiently high to enable the most promising materials to be isolated. !-O O-8~ Z~ t- (I- ¼hI O.Oi O.2![ o i 2 ; 4 h Fig 5 Values of fn(h/e) for calculating the volume of wear of a journal bearing from depth measurements: V = Olh fn(h/e} reasonable correlation between accelerated wear measurements and those on bearings. Fig 7 shows a comparison obtained for a series of polymers containing carbon fibres as reinforcement. Bearing wear measurements were obtained over a period of 500 h in the conditions given. The bearings were then removed from their housings and wear rates determined on their outer surfaces during sliding against mild steel in a crossed-cylinders geometry, see Fig 6a, for a period of about 12 h. Whilst the 'scatter Polymer-based materiNs Of the very large number of polymers available commercially, oniy a stuN] proportion is in signigcant use for dr-f bearing applications. These are listed in Tab]e 3. Nate~4als may conveniently be divided into XhermoptasticF, which soften appreciably or melt at a characteristic cemperamre, and 'thermosets" wl-~ich cross-l~k under the influence of heat and do not subsequently melt. There are two main effects of tempera-rare on mechanical properties. Firstly, there is the gener~ reduction in strengtb and stiffness which results mainly from. a weakening of the interactions between the polymer chains. The temperarare tLrr,.it from this cause is conveniently specified in terms of a 'heat-distortion temperature' at which tl~e deflection of a loaded cantilever beam reaches an arbit rarily defined value (ASTM-D-648). Secondiy~ s~rength changes also occur with it±creasing temperature as a result of chemical reactions within the polymer itself, or between the polymer and the e n w r o n m e ~ , usuatly oxygen° [t is thus possible to identify both a °therm± stability' and an 'oxidatwe stabilityL However, the quota- W w b n W ~L~v v i \ , \ \ \ \ I @ ~2 W CO2v' w ] @V2 i n I C [ E:ZE3 R d Fig 6 Accelerated wear test arrangements for polymer-based materials: a 'Pin and ring', b 'Pin and disc', c 3-body abrasive wear, d 2-body abrasive wear (Taber abraser) 224 TRIBOLOGY December 1973 IO-5 tion of precise values of temperature corresponding to these limits is generally precluded because the extent of chemical reaction is time, as well as temperature, dependent. This may be illustrated by Fig 8 which shows the temperatures for different times of exposure at which various groups of polymers retain 50% of their room temperature strength. It is important to note that there appears to be no polymer currently available commercially which ret~:ins 5(F/c of its room temperature strength for more than ~O 000 h at temperatures above 300°C. To put the thermal and mechanical properties of polymers as a general class of materials into perspective, it is useful to quote order-of-magnitude comparisons of some properties with respect to mild steel. The tensile strengths of polymers are lower by a factor of about 10, their elastic moduli and thermal conductivities are lower by a factor of about 100, and their thermal expansion coefficients are greater by a factor of about 10. These deficiencies can be partially offset by suitable choice of fillers or reinforcing fibres, and a list of the more widely used types is also given in Table 3. Before describing the properties of individual groups of polymers and polymer-based composites, a few general comments on friction and wear may be helpful. o _-g oZ o iO-6 - t-co o ° o t_ o 13~ iO-7- o o o o o o o o o o o / I iO-8 iO-7 I iCr• I io-S iO-4 ~WGor rate in pin and ring tests(mnna/Nrn) Correlation between pin and ring and journal bearing tests for carbon fibre reinforced polymers sliding on m i l d steel. Each p o i n t represents a different material Bearing tests: load = 8.9 N (P = 0.05 M N / m m 2 ) , V = 0.65 m/s counterface - cast iron, 0.3 # m cla Pin and ring tests: load = 11.7 N (P variable), V = 0.54 m/s counterface - mild steel, 0.18/Jm cla Fig 7 Friction It is often assumed that, in comparison to metals, tile coefficients of friction of polymers are relatively low, but this is not generally correct. Because most polymers exhibit visco-elastic behaviour, the magnitude of the coefficient of friction involves an appreciable component arising from deformation and elastic hysteresis loss, and this component varies markedly with the conditions of sliding, and particularly with speed. Thus, although the coefficient of friction of ptfe is commonly quoted as 0.05- O. 1, these levels are obtained only at very heavy loads and low speeds of sliding, or when tile ptfe is present as a thin film on a harder substrate. At high speeds or light loads, ptfe sliding on metals or against itself may exhibit friction coefficients as high as 0.3. Similar, though less marked, effects occur with other polymers. The addition of fillers to polymers also affects the coefficient of friction. If the filler is a solid lubricant, particles may transfer to the counter-face, establish a lubricating film and reduce friction..alternatively with hard, rigid fillers Table 3 o such as glass, mica or asbestos, the load may be preferentially supported by the filler particles and the coefficient of friction is then largely characteristic of the Idler/counterface interactions. For all these reasons, values of the coefficient of friction quoted in the literature, or, in fact, even in this section, should be regarded as order-ofmagnitude values only: there is no such thing as a unique coefficient of friction for a given material. Wear A great deal of the information on the wear properties of polymers and polymer-based composites has been derived from accelerated wear tests of the type already shown in Fig 6. The main objective of these tests has been to determine how wear depends upon the conditions of sliding, such as load, speed, temperature, etc, and it is reasonable to assume that the trends observed will be relevant to bearings. As discussed earlier, however, the absolute magnitudes of the wear rates obtained from accelerated tests Main plastics and fillers of interest for bearings Polymers Fillers and reinforcements Thermosetting resins To improve mechanical properties To reduce friction To improve thermal properties Polyethylene, high molecular phenolics asbestos graphite bronze weight Acetal-homo-, and co-polymer polyesters glass MoS2 silver epoxies ptfe (particles or fibre) carbon/graphite. Thermoplastics Polyamides (Nylon 6, 6.6 and 11) silicones carbon textile fibres Ptfe polyimides mica Polyphenylene oxide metals and oxides Polycarbonate T R I B O L O G Y December 1973 225 ~A O A B E C F- Oq I I0 [O2 tO 3 IO 4 IC Time {h] Fig 8 50% of Area A Area B Area C Area D Temperature-time relationships for the retention of mechanical properties. Polyamide-imide, pNyimide Silicone, polyphenyiene, polybenzimidazo!e Epoxy, polyester, phenol-formaldehyde, ptfe Melamine-formaldehyde, fep, polyphenyJene oxide, polysulfone (data from Ref 4) may not always be sufficiently reliable to be used for the prediction o f wear in bearings. In the early stages of sliding of a polymer composite over a freshly prepared metal counter-face, the wear rate is relatively high. However, during subsequent traversals over the same track on the counter-face, the metat surface can_ be modified by transfer o f polymer or Nler, abrasion by the filler, or abrasion by contaminants from the surrounding environment. In most cases these modifications lead to a smoother counter-face surface and the wear rate then gradually decreases to a constant value. If the initiN counter-face roughness is increased, the magnitude of the initial wear rate increases markedly, but that of the final wear rate is less affected, because of counter-face modification by transfer or abrasion. These effects are illustrated schematically in Fig 9. In service conditions, the wear of bearings is sometimes estimated from single measurements at the end of a given period of operation, and the dotted lines in Fig 9 show how wear rates derived from such single measurements may be greatly in error. At least two wear determinations at different times are required to define the approximate shape of the:_wear volume-time relationship. To a very rough approximation, the initial wear rate of a polymer is proportional to the centre-line-average surface roughness of the counter-face raised to a power of 2 - 3 . The wear occurring in this regime is abrasive in type and similar to that caused by hard abrasive particles. The metal asperities penetrate the softer polymer and produce wear debris by shear or micro-cutting, or by low cycle fatigue. For abrasive wear, there is an inverse proportionality between the wear rate of po!ymers and the product so, the breaking strength times the elongation to break. This result therefore accords with the general observation that the abrasive wear resistance of rubbers is appreciably greater than that o f the more rigid polymers. V,h e n fillers or reinforcing fibres are added to a polymer, the strength increases, typically by a factor of 1.5-4, but the elongation to break may decrease by a factor of up to t 00. Following reinforcement of a polymer therefore, the product s e is frequently reduced, and this, in turn, leads to a reduction in its resistance to abrasive wear. 226 TRIBOLOGY December 1973 As mentioned earlier during repeated traversals of a counter-face beneath a polymer coro_posite~ the meta~ sur~ face may be modified by transfer or by abrasion. When steady state conditions have been reached, the wear process then becomes prLmarily one of fatigue on a locNized scale, the number of cycles to failure depending ".upon the 1ocaiized s~resses occumng, wMch. in turn depend on the t o p o g a p h y generated on the counteroface. Transfer from polymers with high elongations, such as pt.',~e, acetals or nylons, leads to smoother counter-face surfaces and Lence to lower rates o f wear. However~ transfer from brittle polymers such as polyesters, some epoxies or polystyrene. is often in the form o f irregular lumps: the effective surface roughness of the counter-face may then increase during sliding, together with the rate of weal Solid iubricast fitlers, such as ptfe, MoS2, etc. which are intended ~.o re4-ace friction, also contribute to the formation of relative!} smooth transfer films and so assist in reducing wear. tn general, ptfe ~s more effective in this respect than graphite. which in turn is more effective than MoS2. ~n some poiyo mer composites, notabiy those filled witb~ carboa/gra?hite particles or fibres, transfer film formation, an4~hence the magnitude of the wear ra~e, is very susceptible ;o fluid contamination. As an extreme examp!e, FN i0 shows the wear volume-th'ne relationship for a carbon fibre-frilled epoxy resin sliding on stainless steeI. In the early stages of s!iding a transfer film deve!ops on the counter-face and the wear rate gradually decreases to a 1~iting value. When water is added to the system, however, the transfer film is removed almost immediately and the orlginai recta1 surface is exposed, leading to. an increase in wear rate by a factor of 500. Composites containing ptfe show // f/ f (3 i ~6 E 20 > . . . - -- .- j /// / @1 Initial wear //./" ng go%i2e roughness i- 1 Steady-state wear T~m¢ Fig 9 Typicamwear volume-time re~ationsNps for po!yme~ based bearings on journais of increasing roughness Factor increase in wear rate if journal roughness is doubled ~nitiai Steady° state No transfer.film No fil!er or non-abrasive filler X4-X6 X4.-X5 Transfer fimm formation No filler or non-abrasive filler X4--X6 X2 Xt-Xt.6 X I - X ~ o4 Moderately abrasive f[Her (reiative to shaft) tive figures for the abrasiveness of different fillers, based on an arbitrary test in which a bronze ball is oscillated over a filled polymer surface. When the filler is very abrasive, such as glass, the counter-face may be seriously damaged leading to an increased roughness, and in turn, an increased wear rate of tile composite. It is wellestablished that glass-filled polymers should not be used as bearings against a relatively soft shaft material such as aluminium alloy or bronze. Mildly abrasive fillers such as graphite, or high modulus carbon fibres merely remove the peaks of the cotinter-face asperities, reduce the localized stresses, and in turn the wear rate of the composite. The effect of fillers in modifying the counter-face surface may influence the wear properties of a composite far more than the improved mechanical properties of the filled material. 0"4 gEO.3 E "6 o . 2 E -v/ Water added 0.1 I 0 02 I I 04 0-6 A /~. 1 V 2.8 3.0 i0 5 rev Fig 10 Wear volume-time relationship for an epoxy-carbon fibre composite on stainless steel, showing the effect of removing the transfer film by water additions similar effects, though to a less marked extent as ptfe can transfer to metals even in the presence of water. This susceptibility of transfer film formation to fluid contamination may well account for the poor performance of many dry-bearing materials in wet, or alternate wet and dry conditions. It is not generally appreciated that most of the fillers and reinforcing fibres added to polymers are abrasive towards metals. This applies even to lamellar solid lubricant fillers, such as graphite and MoS2, where the abrasiveness is partly intrinsic, resulting from their crystal structure, and partly a consequence of the inevitable impurity content in these materials. Mthough the abrasiveness of lamellar solid fillers is very slight, it is nevertheless sufficient to influence the wear process. Table 4 shows some comparaTable 4 fillers Relative abrasiveness of plastics with various Composite Abrasiveness wear rate of bronze ball (10 7 mm3/N m) Ptfe + 30% + 25% + 25% + 30% + 25% + 25% + 40% + 33% <0.1 620 100 80 31 17 2 0.8 0.5 glass fibre asbestos fibre carbon fibre (high strength) mica coke carbon fibre (high modulus) bronze graphite Phenol-formaldehyde Mineral filled Asbestos paper filled Wood filled Paper filled Cotton cloth reinforced 730 75 17 2.6 1.8 Ptfe and p tfe-based composites The major disadvantages of unfilled ptfe as a bearing material are its poor mechanical properties and a marked tendency to cold-flow under load. Its use in the unfilled form is therefore restricted to a few applications at very light loads, or as thin films bonded to a harder substrate. The mechanical and wear properties of ptfe improve dramatically with the addition of fillers, and a very wide variety of compositions is available; one manufacturer lists 49. The properties of a selection of the more popular are given in Table 5. A particularly widely used composite is one containing bronze, graphite and lead, and one reason for the success of tiffs material appears to be the facility with which it can form a transfer film on a steel cotinter-face. Unfortunately, fillers which are most effective in reducing deformation under load are not necessarily file most effective in reducing wear. The deformation under load for mica-filled ptfe, for example is only about 1% and thus very much lower than that of most other filled ptfe formulations. However, the specific wear rate of this composite, ~ 7 0 X 10 7 mm3/Nm is much higher than that of the glass, bronze or graphite-filled materials, see Table 5. The ultimate selection of a particular filled composite for a specific application is therefore almost always a compromise. Most filled ptfe compositions are available either as moulding powders or as bar-stock from which bearings can be machined. Ptfe cannot be injection-moulded and the usual production technique is compression moulding followed by free-sintering in a vacuum or inert gas atmosphere. Material containing small proportions of fillers can be extruded. An alternative, but more expensive, method for composite preparation is hot compression moulding followed by slow cooling under pressure. There is some evidence to suggest that materials prepared in this way exhibit superior mechanical and wear properties to those produced by free-sintering. Even when filled or reinforced, however, tile mechanical properties of ptfe composites still remain relatively poor in comparison to those of other filled polymers, see Tables 8 and 9. This situation can be improved by using ptfe fornmlations in thin layers attached to a harder backing. The most common of these materials, and one of the most successful, is produced by incorporating a ptfe-Pb mixture into a layer of porous bronze sintered on to a steel backing. The maximum load-carrying-capacity of this type of bearing is of the order of 150 MN/m 2 and the bronze content greatly increases the thermal conductivity. This bearing material has been extensively described and tested, TRIBOLOGY December 1973 227 Table 6 Changesin iife of porous bronze-ptfe-Pb bushes produced by Ntering sliding conditions, Relative to mlid steet shafts of 0.4 #m cta roughness, and continuous operation at 20°C (Data from Pratt s) Bronze gridceramic filled # Factor c~.ange Sliding conditions i :o.,_ 5i OO ~ ~ [~" Y_ / , S ~ 1 Shaft roughness ~ Porous bF6nze-ptfePb bearings A Shaft material ,# decreased to 0.2 #zm increased to 0.6 g m X2 t X½ I stainless steel I chromium plate anodized aluminium ×2 Temperature increased to f 120°C 1 200°C Intermittent operation at high PVs Time Fig 11 Typical variations of depth of wear with time for a Porous bronze - ptfe - Pb bearings b Bronze grid - ceramic filled ptfe bearings and exhibits a characteristic variation of wear with time as shown in Fig 11. The initial period of rapid wear is conffmed to an overlay of ptfe approximately 12 #m thick, and bronze begins to appear at the sliding surface as soon as this overtay is removed. There is then a period of low wear at a constant rate until, after removal of a further depth of about 50/~m, the bronze content of the surface increases to an unacceptably high level and the wear rate begins to increase rapidly. Because of this characteristic pattern of behaviour these bearings possess a well-defined !ife, and the relationship between life and P V factor is shown in Fig 12. The approximate way in which the life X~A x ~/_,~ ×2 is affected by various changes in the sliding conditions is given in Table 6. Bearings with essentiaily the same ingredients, but of different construction are also available. Ceramic-fil!ed ptfe tape can be pressure-rolled into a bronze grid to a depth of about three-quarters of the thickness of the grid. and the layer is then cemented on to a metal backing. This bearing exhibits a similar characteristic wear-time relationship to the porous bronze-ptfe-Pb type, but with greater 'bedding-m wear' and a somewhat higher steady.state wear rate, see Fig 11, cu.we B. To compensate in part for the latter, however, We wear rate remains constant until a depth of about 200 #m has been removed. It has been known for some time that ptfe in fibrous form exhibits much greater strength than in bulk, possibty because of a greater degree of chain orientation. Severn bearing con- Table 5 Properties of ptfe and filled ptfe composites F ler, Properties ~ Specific gravity Tensile strength MN/m 2 Elongation % Ftexural modulus GN/m 2 Deformation under % load at 25°C and 14 MN/m 2 Expansion coef10-5/°C ficient 2 5 100°C Thermal conducW/m°C tivity Specific wear rate 10 -7 mm3/N m Friction coefficient on steel at 0.01 m/s [ 0.05 m/s Limiting y~0.5 m/s MN/m z X m/s P V at t 5 m/s 228 TRIBOLOGY December 1973 glass 12½% weight glass 12½% weight 15% weight MoS 2 graphite 20% weight carbon 5% weight graphite 55~ weight bronze 5% weight MoS 2 2.19 t7.5 300 1.1 11 2.3 13 230 1.1 4 2.12 9.5 130 1.4 8.1 2.1 11.6 70 1.2 2°9 3.9 i3.0 90 15 4.6 12.1 11 12.5 8.4 10.1 None 2.2 9 400 0.6 Continuous flow 17 0.43 0.5 t 0.45 0.44 0.72 @.25 1.4 0.09 1.2 0.09 6.8 0.12 L2 0.12 1 o0 0.13 0.33 0.39 0.5 0.5 0.35 0.53 0.44 0.04 0.62 0.62 0.6 0.95 0.53 0.42 0.44 0.44 0.06 0,09 4000 0.! view of the earlier remarks about the abrasiveness of fillers on the counter-face, it is important to note that the provision of a smooth counter-face is much more important when the fillers are non-abrasive. This is illustrated in Fig 13 which compares the wear-time relationships of ptfe/carbon fibre (high modulus) with those of ptfe/glass fibre on mild steel counter-faces of different roughnesses. In the case of the ptfe/glass, the composite is sufficiently abrasive to smooth tile steel surface and the steady state wear rate is virtually independent of counter-face roughness. Increasing the counter-face hardness, however, will tend to offset this smoothing effect, and careful attention should always be paid to minimizing the roughness of very hard counter-faces for all types of composites. In summary, Fig 14 gives the P V relationship corresponding to a wear rate of 25 #m/100 h for many of the ptfe-based formulations discussed. 5OOC 2000 "Z ..J I000 o 500 o 200 IOC 5C 20 OI l 0.2 I I 0.5 I PV(MN/m2x I l 2 5 m/s} Fig 12 Life - PV relationship for porous bronze-ptfe-Pb bushes on mild steel shafts. 15.8 mm bore, 19 mm long. Speeds 0.62, 1.25 and 2.5 m/s (750, 1500 and 300 rpm) (from Pratt s ) structions utilize ptfe fibre. In one, fibre/phenolic resin mixtures are attached to a woven polyamide backing, and in turn to metal strip. Another version involves ptfe fibre interwoven with glass fibre in such a way that the rubbing surface is largely ptfe and the opposite surface is mainly glass. The fabric is then cemented with a phenolic resin to a metal substrate. This particular type of bearing is capable of supporting extremely high loads (up to about 350 MN/m 2) at low speeds and is particularly suitable for applications involving oscillating motion. The wear-time curves are of similar shape to those in Fig 11 but a unique feature is that at light or moderate loads several distinct 'plateaus' may occur in sequence, separated by short periods of more rapid wear. Performance tends to deteriorate rapidly at high speeds~ partly because of the low thermal conductivity. For less arduous applications, a variety of filled ptfe liners and tapes is marketed. Some ceramic-filled tapes are provided with a self-adhesive backing, and graphitefilled materials are available as liners already cemented to steel or reinforced phenolic resin backings. Separate liners. 0.4 0.8 mm thick have the advantage that they can be allowed to 'float', which facilitates heat-dissipation because some heat is then liberated directly at the housingliner interface. The counter-face roughness generally recommended for use with ptfe composite bearing materials is within the range 0.2 0.4/xm cla. Improvements to the shaft roughness usually reduce the wear rate, particularly in the early stages of sliding as already mentioned. However, there is some evidence to suggest that there are few advantages to be gained by bnproving surface finish beyond about 0.05/~m cla. In any case, manufacturing to roughnesses below this level can become inordinately expensive. In Other thermoplastics The particular advantage of thermoplastics over thermosetting resins is that they can generally be injection moulded: mass-produced bearings are thus potentially extremely cheap. The most widely used thermoplastic bearing materials are polyacetals and polyamides (nylons) and different types of each exist. Polyacetals may be either a homopolymer of formaldehyde or a copolymer of formaldehyde and acetaldehyde, but the differences in properties are slight. At room temperature, the homopolymer is about 10% stronger and stiffer than the cope lymer, but the position is reversed when the temperature exceeds about 100°C. Some typical properties of the homopolymer are given in Table 7. The polyamides are a complex family: nylons 6 and 11 are self-condensed amino acids, the number corresponding to the number of carbon atoms in the parent acid, and nylons 6.6 and 6.10 are reaction products of hexamethylene diamine (with 6 carbon atoms) and dibasic acids with 6 and 10 carbon atoms respectively. Nylon 6.6 is tile most widely used of all types, and has the higl, est strength and stiff ness. Some properties are given in Table 7. Its main disadvantage is tile relatively high moisture absorption, 3 2 A B c 'S b I 2 3 4 5 I 2 3 4 S IO5 revolutions of disc Fig 13 Wear volume-time relationships on mild steel a. 0 . 0 4 / l m cla, b. 0 . 1 8 # m cla, c. 0.5/*m cla A ptfe-carbon fibre (high modulus) - transfer to steel B ptfe-glass fibre. Abrasion of steel T R I B O L O G Y December 1973 229 I02,~. \-~. Wnvcn nff~/nln~,c, fibre' I E Z Q.,. o-ol 04 { IO V(rn/s) Fig 14 P-V retationships for ptfe-based bearings at a depth wear rate of 25 # m / l O 0 h (mainly from Ref 6) and consequent low dimensional stability, and nylons 6.!0 and 11 are better in this respect (24 h cnoisture absorption ~0.4%). The thermal expansion coefficient of nylon i i, however, (~18 X 10-5°C 1) is about twice that of the other types. Very large bearings have been made from cast nylon monomers which are subsequently polymerized in situ and the properties of this materiai are very similar to those of nylon 6.6. gearings fon-ned from sintered nylon powder of high molecular weight and crystallinity are also available and appear to show some advantages from the wear resistance standpoint. By controlling the sintering conditions, the degree of porosity can be increased and the material impregnated with oil for use as an alternative to oil-filled porous TaMe 7 metals or for rolling-element bearing cages. £{.thoag£ nylon 6.6 has a higher melting point and room~.;ernpe<ature: tensile strength than acetai, the s,rength decreases much more rapidly with temperature, as shown by compam~g the values for the heat-distortion temperatures m 7abie Where frictional heating is likely tc be appreciabie aceta is therefore a preferred choice Some properties of ? olycarbonate are a.tso gl,reu it:,' Table 7, and the main claim tc distinction of this po_~y. mar is its extremely high Lmoact strength Poiycarbonatc is therefore a potentially val~able material for abrasive or erosive wear situations, although its bearing properues in clean environments are nor remarkab]e The outstand.ing temperature stabiIity of me polyimides has a!ready been noted in Fig 8. These polymers are a~ain availao!e in several different types, some of which are p:-oeessed as thermoplastics by hot-comoression moulding and some as thermosetting resins. Table 7 includes data for fi~e sore> pression moulded type. [n gener£ the rr'_,ecaanicai ~roparties are similar at room ten~ perature to d?ose of nv!ons and acetats, but the heat distortion ~empera~ure is ,~crv much higher, and because of ti~is, tt~e !irnitm.g PV ~ac :ors are also higher. Whilst the specific wea, rare of poiyo imide at room temperature is not particuiar!y ;.ow. > e r e is iitt!e or no increase with temperature until abo~t ?CO°C. It has aiso been reported that ttse wear rate (n n,_t: . . . .Oh,A1 . is much lower than in air. by a Nctor of aboux ~O0. As with ptfe, ti~e mechanical properties of other ;:herrno plastics are also improved by the addition of fillers o~ reinforcing fibres. Some typical values for glass :fibre reinforced and solid-lubricant filied cornposr~es are givers in Table 8. Provided that the filler contents renta~ be}ow about 20% by volume, there ~s no serious rmsedimem ~s injection moulding. Some of the improverr~en[s obtained in the friction and wear properties by ? fie/glass or -~tfe additions can be very large, eg with polycarbonate ar~d nylon 6.6. With acetaL however, ptfe/giass causes a significant increase in tLe specific wear rate and ptfe alone is a more effective addition, despite the slight impair ment of the mechanical properties. The wear grepert~es of polyimide are improved by graphite additions° our MoS2 is ineffective in air and on!y becomes a useful lubricant for applications in ssace or ultraoh~gh vacuum~ Additions of si~nail amounts of MoS2 ,.2 5%)i to nylon 6.6 Properties of some unfilled thermoplastics Potycarbonate Acetal (homopolymer) 1.2 65 2.34 225 130 1 o4 ] ,i 69 2,8 175 125 79 2o9 260 70 Polymer Properties Specific gravity Tensile strength Fiexural modulus Softening/melting point Heat distortion temperature at 1.8 MN/m 2 [zod notched impact strength Moisture absorption, 24 h Specific wear rate Friction coefficient on steei MN/m 2 GN/m 2 °C °C J/cm % 10 - 7 mm3/N m { 0.05 m/s Limiting ~ 0.5 m/s P V at [ 5 m/s 230 TR1BOLOGY December t973 MN/m 2 X rots 8.7 0.35 480 0.35 0.03 0.01 < 0.01 0.76 0.25 12.5 0.2 0.14 0.12 0.09 Nylon 6.6 0.49 .5 38 0.25 Po!yimide 1,43 ~7q 3.2 No melt >26,0 ,% A ~.~;-8 0,32 30 0°42 0.6 4 0:11 0°09 < 0.09 - Table 8 Properties of thermoplastics with lubricating fillers Polycarbonate ~f i latieps rn c )d p e ~r e r s Specific gravity Tensile strength Flexural modulus Heat distortion temperature at 1.8 MN/m 2 Izod notched impact strength Moisture absorption, 24 h Specific wear rate Friction coefficient on steel 0.0S m/s Limiting ~ 0.5 m/s PV at ! 5 m/s 22~,4 ptfe Acetal Nylon 6.6 15% 15% 15% ptfe 3(~: glass ptfe 3~4 glass 44% ptfe ptfe 3~:~ glass 22% ptfe Polyimide 15% 15% graphite MoS 2 1.33 1.55 1.43 1.4o 1.51 1.50 45 1.3 130 120 8.3 145 1.5 40 2.1 100 1.75 MN/m 2 GN/m 2 °C 107 0.7 160 38 2.1 82 163 ~).3 250 45 3.8 >260 41 3.5 >260 J/cm 1.1 0.14 1.1 0.06 5.8 0.2 0.97 I .05 0.46 0.27 0.25 3.2 0.15 >1.4 0.4 0.17 0.38 0.2 38 0.28 0.44 0.42 0.28 0.27 0.55 2.3 0.18 >1.4 0.95 0.28 0.98 0.5 3.1 0.26 0.61 0.7 0.46 % 10 7 mm3/N m 0.15 MN/m 2 X m/s 0.06 are marginal in their effects on friction, but tend to improve wear resistance by affecting the hardness and crystallinity developed in the polymer during processing. Several thermoplastics, notably nylon 6.6 and acetals, can benefit greatly from marginal lubrication with fhfids, and in these conditions their performance often exceeds that of conventional metallic bearing materials. To optimize the load-carrying-capacity, several types of bearings are available comprising thin layers of nylon or acetal bonded to a steel backing. A more sophisticated construction is a porous bronze layer on a steel backing, impregnated with acetal, and leaving a layer of polymer about 100/Ira thick over the surface of the bronze. Regularly spaced recesses on the surface are provided to retain small amounts of lubricant. This type of bearing can exhibit a life of above 1000 h at PV factors up to about 2 MN/m 2 X m/s. A summary of the available P V relationships for the various thermoplastics, unfilled and filled, is given in Fig 15. 0.32 0.32 5 50 0.1 0.3 0.1 0.3 6 5 bearing is almost inevitable, unit costs tend to be rather high. A characteristic feature of reinforced thermosetting laminates is their anisotropy in mechanical properties. Both tensile and compressive strengths may vary by factors of up to 5 depending on the direction of testing relative to that of the laminations. Because of th is, the data given in Table 9 only provides order-of-magnitude '°°I Thern?osetting rosins Although the raw material costs of the more common thermosetting resins, such as phenol-fornlaldehyde, are less than those of most thermoplastics, fabrication costs tend to offset this advantage. The precursors are usually fluid and the addition of hardeners and a catalyst induces cross-linking of the molecular chains leading to an irreversible transformation to the solid state. The cheaper phenolformaldehyde and polyester resins are somewhat brittle and are almost always used in fibre-reinforced form. Epoxies and silicones, on the other hand, can be produced with a whole spectrum of properties by the addition of suitable flexibilizing agents. ~fhe simplest method of producing reinforced thermosetting bearing materials is by impregnation of fibrous mat or cloth cellulose, cotton, asbestos, glass, carbon etc by the liquid resin, or a solution thereof, pressing into an appropriate shape and curing at elevated temperature. Alternatively, tubes may be produced by filament winding techniques. Neither method is amenable to rapid, large scale production, and since some machining to produce a finished E g EL nrnide "aphit¢ 0' \ \ 0"0 \ 0.01 O4 I IO V(m/s) Fig 15 P-Vrelationships for thermoplastic and filled thermoplastic bearings at a depth wear rate of 25/lm/1 O0 h (partly from Ref 6) TRIBOLOGY December 1973 231 TaNe 9 Properties of reinforced thermosetting resins MelaminePhenolformaldehyde formaldehyde Silicone + cellulose + cellulose + asbestos Properties Specific gravity Tensile strength Flexural modulus Izod notched impact strength Water absorption, 24 h Maximum allowable temperature MN/m 2 GN/m 2 J/cm 1.4 104 6.9 0.68 % °C 1.5 130 Specific wear rate 10 - 7 mm3/N m i .35 76 8.3 i .! 1.5 i50 1.75 62 I4 I .d 0.8 250 Epoxy + cellulose 1.2 90 3.5 1.1 1.5 175 Polyester +celluiose 1.25 69 4.8 t .6 t .5 140 With solid lubricants added, values normally within the range 3 - 5 0 Polyimide + carbo~ fibre 1.6 450 40 0~5 2 300 0.5 Vaiues may vary appreciably depending on form of reinforcement (cloth, chopped fibre mat, unidirectional ~bres); direction of testing relative to iaminations; degree of resin cure. properties. The figures are nevertheless sufficient to show that the strengths and stiffnesses of some composites can greatly exceed those of filled or reinforced thermoplastics. In bearings, advantage can only be taken o f these high strengths if some means are made available either for dissipating the frictional heat effectively, or minimizing its generation by reducing the coefficient of friction. A major area o f application of reinforced thermosets is thus for water4ubricated bearings in marine engineering, rolling mills, etc. For operation as dry bearings, the addition of solid lubricants such as ptfe, graphite or MoS2 is virtually essential, but even with these additions the specific wear rates seldom fall as low as those obtainable with ,°2t :~ ~ "~ I- graphite, MoS2,pt f¢ \\ × \ the better filled ptfe composites. Because reinforced thermosets do no~ soften appreciably on heating, they continue to be useable up to the temperatures ar which thermal or oxidative degradation of the resin or reinforcement begin to be significant. The rates of wear then tend to increase, but not so rapidly as for thermoptastgcs near their softening points, t~einforced thermosets will continue to operate for short periods at temperatures wetl in excess of those quoted in Table 9 because therma! or oxidative degradation are bo-th time and temperature dependent. Epoxy resins are considerably more expensive than phenolics, and as shown in Table 9 the mechanical properties of reinforced composites of both types are broadly shmilar. However, the fact that some epoxy formulations can be made less brittle than phenoiics has led ro their use in a filled rather than reinforced form, which partly offsets the cost. Epoxies containing graphite or MoS2 are available as solid bars for machining into bearings, or as a two-component fluid for casting. Filled epoxies can also be sprayed as thin layers on to a metat backing. Bearkr~gs based on reinforced p olyhqqides are st~ll a com~ parative rarity, partly due to the expense of the po!ymer and partly to the difficulties in fabrication. The introduction of carbon fibre as reinforcement further increases cost, but this material is neverthe!ess of interest for specialized aircraft applications because o f the combination of very high strength and a specific wear rate as tow. or lower, than those obtained with ptfe composites, see Table 5. There is little information avai!able on the F - V re!atmno ships o f many of the reinforced therm_osetting resin composites described, but a generalized grouping is given in Fig 16, based, as in the earlier results, on an arbitra~ wear rate of 25 gin/100 h. Design points o.oJ 0 •~01 0 ' II ~,,"/ I - : V{m/s} Fig 36 P-V relationships for filled and reinforced therrnosetting resin bearings at a depth wear rate of 25/~mtl00 h (partly from Ref 6) 232 TRIBOLOGY December 1973 ~0 It is convenient here to summarize some of t2c~especific design factors which are invo!ved in using polymer-based dry bearings. Counter-face roughness As already discussed, this is one of the most [mportar-t parameters affecting the magnitude of the wear, particulariy in the early stages of sliding. Values of 0.2 0.4/am cla, with an upper limit of 0.7/am cla, are usually recommended, but reductions down to 0.05/am cla will almost always reduce wear unless abrasive fillers, such as glass, are present in the composite. Abrasive fillers necessitate harder counterfaces. It is important to note that the magnitude of the cla roughness is not a unique guide to the effect of the counter-face on wear of the bearing. A ground surface of 0.2/am cla will generally cause less wear than an abraded surface of the same value, and in turn an abraded surface is preferable to one of the same roughness produced by grit-blasting. The reasons for these differences lie in the detailed nature of the topographies produced by each finishing treatment. Aspect ratk) In theory, the performance of a dry hearing should be independent of the aspect ratio length/internal diameterbut in practice there are two complicating features. With large aspect ratios, distortions or misalignment may cause stress concentrations and excessive localized heating. In such cases a short period of 'running-in' at reduced loads, or initial marginal lubrication, may be helpful. Short aspect ratios introduce problems associated with the location of bearings in their housing. In general, the optimum ratio appears to be of the order of unity. Wall thickness This factor is frequently dictated by the overall design of the component. However, where a choice is at all possible, a typical value of thickness is of the order of one tenth of the shaft diameter. This value may usefully be increased if shock-loadings are anticipated, or decreased if the pressure approaches the limit at which deformation under load becomes significant. A reduced wall-thickness also facilitates the dissipation of frictional heat. Clearance Insufficient initial clearance has probably been responsible for more dry bearing failures than any other single cause. The clearance required for polymer-based materials is much greater than that typical of lubricated metallic bearings and for most applications is typically 5/am/mm with a minimum of 125/am. The necessity for this large value arises from the combined effects of dimensional instability, arising from expansion and moisture absorption, and the development of transfer films of debris on the shaft surface. With certain proprietary ptfe-based bearing assemblies, the initial clearance can be greatly reduced by minimizing the counter-face roughness, and in some cases reduced to zero by preloading the bearing. In the latter situation, however, starting-torque may present problems. Fitting The clearance may also be affected by the particular method used for fitting bearings into their housings. The most popular method for polymer-based bearings is press-fitting with a degree of interference ranging from about 7/am/ram for unfilled ptfe, through 5/am/mm for nylons and acetals, to 2-3/am/mm for reinforced thermosets. The closure in bore resulting from interference fits is very roughly equal to the interference itself and should be allowed for when defining the clearance required. Where temperature fluctuations are likely to be encountered, interference fits in housings are inadvisable and alternative methods such as keyways, flanges or adhesives are preferable. Commercial products A partial list of proprietary dry bearing materials with trade names and suppliers is given in Table 10, together with a brief description of each. The precise compositions of many of the more complex bearing constructions are not always divulged by the manufacturers. Carbons and graphites Manufactured carbons-graphites comprise a complex series of materials whose mechanical and tribological properties can vary very widely. The raw materials take many forms, as shown in Table 11, and particles of one or more of these are mixed with binders, pressed into a solid, and finally heat-treated. A typical composition could be 20% petroleum coke. 6~} of a mixture of other cokes, natural graphite and lamp black and 20°~ of pitch binder. The pressure during consolidation affects the density and porosity, the latter being typically within the range 3 l 5%. Heat treatment up to 1000°C removes volatile constituents, particularly in the binder, and above 2100°C the amorphous carbon constituents begin to be reordered into a graphitic structure. The degree of graphitization is a function of both the temperature and of the structure of the starting material. Cokes, for example, graphitize more readily than carbon blacks at a given temperature. Following heat-treatment any residual porosity may be impregnated with resins or solid lubricants. Alternatively, if high proportions of solid lubricants are required, carbon/ graphite powders prepared as above may be mixed with the lubricant powders and resins and hot-pressed at a temperature sufficiently high to cure the resin. It follows from the above oversimplified, and necessarily brief, description that a whole spectrum of materials can readily be prepared. One of the major difficulties, in fact, is reproducibility of a desired product. For convenience, the various materials available can be divided into four main categories as shown in Table 12. The properties given are typical of each class, and, as usual, are significant only as orders of magnitude. However, several characteristic features may be noted: a the thermal conductivities are of the same order as that of mild steel, and therefore a factor of about 100 times greater than those of unfilled polymers, b the coefficients of thermal expansion are only about ¼ ½ those of steel, and thus about 20 times lower than the values for unfilled polymers, c the moduli of elasticity are about 10 20 times lower than that of steel, and roughly of the same order of magnitude as those of reinforced polymers, d the tensile strengths are substantially lower than the compressive strengths, sometimes by a factor of 10. Carbons-graphites do not exhibit plasticity in their stress-strain behaviour. The elastic modulus decreases with increasing strain, and failure occurs by brittle fracture at strains which are typically 0.5 3%. There are no large changes in mechanical properties over the temperature range up to 1000°C in vacuum or inert gas atmospheres. In air, however, oxidation begins to become significant at around 350 500°C, depending on the type of carbon and Table 12 shows that the amorphous materials oxidize more rapidly at a given temperature than the more graphitic ones. For practical purposes, the temperature limitation of bearings is determined by the rate of oxidation in relation to the rate of mechanical wear; the oxidation TRIBOLOGY December 1973 233 TaNe 10 Trade names and suppliers of some polymer-based bearing materials Trade name Material Manufacturer Plaslubes High lubricity FRTP Nylatron GS, GSM Nylasint 2G, 6G Fulton 404 Delrin AF (Fortified polymers; no specNc trade names) Rulon Fluorosint CF2 Station M7 DQ2 DQi DQ3 Fibreglide unreinforced thermoplastics with ptfe o r M o S 2 glass reinforced thermoplastics with ptfe or MoS2 nylon + MoS2 nylon + graphite acetal + ptfe acetal + ptfe glass reinforced thermoplastics with ptfe Fiberfit Inc Fiberfil inc ',?olypenco Ltd Polypenco Ltd LNP Corp Du Pore LNP Corp DNon Corp Po!ypenco Lid Crane Packing C o Morganite Carbon Ltd Nobrac Carbon Glacier Metals Glacier Metals Glacier Metals Ampep Ltc Motynium ptfe with various fillers ptfe with inca ptfe with glass ptfe with carbon ptfe with graphite ptfe with graphite ptfe with bronze and graphite ptfe with bronze and lead oxide interwoven ptfe/cotton cloth with thermosetting resin and graphite bonded to metal interwoven ptfe/glass cloth + thermoset bonded to metal interwoven ptfe/glass cloth + thermoset bonded to metal filled ptfe on metal backing porous bronze impregnated with ptfe/Pb on steel backing porous bronze impregnated with aceta! on a steel backing filled ptfe on metal backing ptfeowoven bronze mesh on metal backing polyimide polyimide + ] 5% graphite polyimide + 15% MoS2 polyimide + z~netal and solid lubricant fillers porous metal impregnated with polyimide and solid lubricants reinforced po!yimides with sotid lubricants reinforced polyesters with graphite or MoS 2 thermosetting resins with cotton or cellulose reinforcement and graphite, MoS2 or ptfe additives thermosetting resins with cotton or cellulose reinforcement and graphite, MoS 2 or ptfe additives thermosetting resins with cotton or celhilose reinforcement and graphite, MoS2 or ptfe additives epoxy with MoS2, graphite or ptfe fiilers epoxy with MoS2, graphite or ptfe fillers Eccoslip epoxy with MoS2, graphite or ptfe fillers Fibreslip Fabroid Vandry DU DX Unifton Unimesh Vespel SPi Vespel SP21 Vespel SP3 Feuraton Feuralloys Kerimid Orkot RL, RH, TL ~afnol Ferrobestos Raitko ERB 14J TaMe 11 Raw materials used in the production of carbonsgraphites Ampep L;:d Transport Dynamics !nc Vandervelt Products Glacier Metals Glacier Metals RoseoForgrove Ltd Rose-Forgrove Ltd ©u Pont Du Pont Du Pont Berno] inc Bemoi !nc Rhone-Pouienc British Steel Corp qp, ~ ;r /[ ![~Iflsoi zX ~= J. W. Roberts Ltd Raitko Lid Nobrac Carbon Lie Gredimex AG (Molykote G,K Lid) Emerson and Cuming products, being gaseous, do no;: interfere with the wear process. F r i c t i o n and wear Carbons {petroleum pitch h retort Cokes Binders Additives pitch white metals tar bad bronze J ~, metallurgical resins Graphites natural artificial lamp ~ channel ( furnace Carbon blacks Charcoal 234 T R I B O L O G Y December 1973 silver ptfe MoS 2 resins Fig t7 gives some values for the coefficients of fricuon and rates of wear of a variety of types of carbons°graphites during sliding against hardened steel. The coesideraNe variations may be noted. In genera] the friction is lowest for the most graphitic materiais, but the wear rates do not appear to depend so much on the grap_hite content The iowest wear rare of N1, in fact, is that ef an amorphous carbon. The ciassical explanation for the low friction of graphite relates friction to the cG~stal s~mcture. Graphite possesses a layer-lattice structure in which networks of" hexagonally arranged carbon atoms are separated from each other by a distance much !argot than the interatom~e spacb.g within the layers. The binding forces between layers are comparatively weak; little energy is therefore Table 12 Properties of carbons-graphites ,,,• Properties Specific gravity Tensile strength Compressive strength Elastic modulus Expansion coefficient Thermal conductivity Oxidation rate at 500°C Maximum temperature for COlltinuous Metallized carbon Carbon-graphite Amorphous carbon of carbon MN/m 2 MN/m 2 GN/m 2 10 5/°C W/m°C g/m 2h °C Low graphite High graphite Electrographite 1.6 1.6 25 180 11 0.5 7.1 200 300 23 160 1I 0.2 q.2 80 350 I .65 21 83 9.5 0.3 35 15 350 14 65 7 0.4 55 2 500 0.2 0.35 0.4 1.65 White metal Pb-Cu 2.8 45 230 16 0.35 - 55 260 18 0.5 25 130 300 Mild Steel 7.85 560 410 183 1.1 46 2.7 use Limiting PV at low speeds MN/m 2 X m/s Specific wear rate 10 7 mm_~/N m 0.7 0.7 Generally within the range 5 - 5 0 needed to induce cleavage or shear, and so the coefficient of friction is low. ttowever, the cleavage energy is low only when water or other condensable vapours are present in the environment, and this observation correlates with tile fact that the coefficient of friction is low only when the water vapour content exceeds a critical pressure (of the order of 3 inmHg partial pressure in air). The presence of water vapour also influences the magnitude of the rate of wear, and catastrophic increases can occur, by factors of up to 103 or 104, when the water vapour content falls below the critical value. In sliding conditions it should be realised that the relevant concentration of water vaponr is not that in the environment as a whole, but that in the vicinity o f the actual sliding surfaces. If there is a significant rise in surface temperatures, the concentration of water vapour relative to the saturation concentration decreases and the carbon surface may begin to lose its physically adsorbed water vapour, leading to an increase in both the coefficient of friction and the rate of wear. As the temperature increases further the rate of adsorption of water no longer suffices to satisfy the fresh carbon surfaces produced as a result of mechanical wear, and both the friction and wear rate increase dramatically. These effects are illustrated schematically in Fig 18 which shows typical variations of friction and wear with temperature for carbons-graphites sliding on metals. High friction and wear may also be induced at ambient room temperatures if the combination of load and speed becomes sufticient to raise the surface temperatures to tire order of 500°C or greater. Apart from the water vapour concentration, there are two other factors which can influence the magnitude of the coefficient of friction. Finely divided debris from tile wear process may become consolidated on the surface of the carbon or its counter-face and exhibit a preferred crystallographic orientation and direction of easy shear thus reducing friction. Alternatively these layers may fill the surface irregularities and increase tile real area o f J O-4 ~.r o.2 0 u O-I I I IOO 2OO I 300 I 4OO 5OO Temperature (°C) IO2 'E ,o _u .~ .-=_~ Specific wear rate {mm3/Nm) 10-7 Compacted natural gCaphlt¢ ~ eiatin-e - ~ o nde~ n aTu~T 9r ~pNt ~ Copper - ~ (a ~ - ~ - ~ 9 t ~ b ~ Copper - g r a phite (low copper) Spectroscopic st andard ~sg~hLt e' Am~r Phous carbon ]E/ectrogr~h~te (brush grade) ~ b o n 2 ~ b J ~ l T o w~ _ ~ ~ a ~ ~ ~ I0 6 io-S Friction coefficient 10-4 10-3 0.4 0.8 1:2 16 g.o or) . " m 12 C a r b o n - rag5~hire ~ C a r b on - copp¢ r - ~ _ 1 5 ¢ 5 ] r ,n 91~a87 ! N a t u r a l qraphite - h i g h temp treated £Sr-bdn- white m e t a l - ~ Carbon lead bronze E J ¢ ¢ t r ~ r ~ h i t l t (bearing qrade) Amorphous c a r b o n (electrode grade~ ..... ~ . ~ . . . . . . I IOO I m Fig 17 Friction and wear of various carbon/graphites sliding on hardened 1% carbon steel of 0.025/lm cla roughness. Load = ION speed = 18 m/s 2OO 3OO 400 Temperature (°C) 500 Fig 18 Schematic variation of friction and wear rate of carbons/graphites with temperature during sliding against steel TRIBOLOGY December 1 9 7 3 235 ×/ d initial ~O'ml wear z gG:~E E Q fd ~_ P a Natural graphite x ESectrographit¢ e Hard carbon u ¢J / h Steady- state wear ~ / ~~'x~ uransTer {~ Transfer --xNo~-----~-------...~e o abrasion Abrasion lC?6L 0-0t O-i I Io t IOO Surface roughness (gin cla) Fig ! 9 Variation of wear rate with counterface roughness for carbons/graphites sliding on copper contact which tends to increase the coefficient of friction° Nddgley and co-workers ~ have observed the latter effect occurring in a cyclic manner during the sliding o f carbon thrust washers. When the friction rises to a high value, the surface stresses become sufficient to disrupt the layer of consolidated debris, the surface is roughened, the real area of contact decreases and the friction again decreases. Because the moduli of elasticity of carbons and graphites are relatively low and comparable with those of reinforced polymers, elastic deformation and fatigue play an important role in the wear process. As with polymers, the wear rates in the early stages of sliding are very dependent on the surface roughness of the counter-face and Fig 19 shows typical variations for three types of carbons. In the final stages of sliding, when steady-state conditions are attained, there is much less dependence of wear rate on initial counter-face roughness, see Fig 19, for the same reasons as discussed earlier for filled polymers; the counter-face surface is modified either by abrasion or by transfer. Abrasion of a counter-face by carbons-graphites arises from two causes. Firstly, since graphite is anisotropic in structure, it is also anisotropic in properties, and the maximum hardness in the direction parallel to the basal planes is of the order o f 1500VPN. Secondly, carbons and graphites are seldom pure and contain small quantities of abrasive materials such as metal oxides. These impufites tend to volatilize during heat-treatment, so that the degree of purity depends not only on the type of starting materiai but also on the heat-treatment temperature. In very general terms the degree of abrasiveness from both causes, intrinsic and impurities, increases in the order: electrographite, carbon-graphite (high graphite), carbon-graphite (low graphite), natural graphite and amorphous carbon. 236 TRIBOLOGY December 1973 The formation of transfer films on the counter-face revolves structural breakdown of the 5carbon into units of the order of 5 50 nm m size. Breakdown of the more graphitic materials to t.his level is relatively easy and most etectrographites, for example~ wN readily form transfer Films ever, on counter-face metals as soft as copper. For the stronger, non-graphitic carbons, however, locNized stresses sufficiently high to achieve breakdown are oPJy possible when sliding against very hard counter-face materials. Thus amorphous carbons or low graphitecarbons will generate a tra~sfer ~irn on ceramics or tungsten carbide but not on steels or copper. As a genera] conclusion, each carbon-grapNte generates its own characteristic topography on the counter-face by abrasion or transfer, or a combination o f the ~wo It is this factor which is primarily responsible for most of the difficulty in predicting the wear rates of carbons-graphites directl] from a knowledge of their structure, composition and mechanical properties. Despite "chair long history o f use, the P - V properties of carbons-graphites do no~ appear ,~o be as well categorized as those o f polymer-based bearing materials. Some approximate values of the limiting P V factors at tow speeds are given in Table t2, and estimates o f the P - Y relationships for continuous operation are shown in Fig 20. As for rei~> forced &errnosetting resin bearings, the !LmRing PV factors are greatly increased if fluids are available to increase the rate of heat dissipation. Carbons-graphites are particularly suitable for operation in fluids because not only are t~hey chemically compatible with most types, other than strong oxidizing agents, but they are not prone ~o dimensional changes ~o the'sm~ne extent as polymers. However, fluids may often result in increased wear o f the more graphitic grades which in dry conditions exhibit low wear as a consequence of transfer fitm formation. Grades contNning little or no graphite are less sensitive in this respect~ Addi.. !o2[ tO jBronz ' ~ e -graphite MoS2 E Z O_ Carbongraphite 4-Sn-Pb-Sb Carbon Cu/Pb low g r,~phit~\~N~ carb°n 04High greph~ -carbon O.Oi O!O~ O.I v [ra/s) Fig 20 P-V relationships for soma carbon/graphite bearings at a depth wear rate of 25 #m/100 m (from Ref 6) tions of ptfe to carbon-graphites may be helpful in maintalning transfer films in wet, or alternate wet and dry, conditions. Ptfe/MoS2 additions also tend to reduce the dependence of friction and wear on the environmental humidity during dry operation. Design points The counter-face roughness usually recommended for carbon-graphite bearings is 0.2-- 0.4 ~um cla, but as for polymers, a further reduction in roughness will generally lead to lower wear rates, particularly in the early stages of sliding. Mild steel is not a particularly good choice of counterface metal for operation against carbons-graphites because of its lack of corrosion resistance in humid environments. Corrosion may also be enhanced electrochemically because carbon is strongly electro-negative with respect to iron. If a relatively soft and inexpensive shaft material is essential austenitic cast iron is probably the best. However, harder materials are generally preferred to minimize serious abrasion by the carbons-graphites and stainless steels and stellites with hardnesses within the range 4 0 0 600 VPN are most suitable. Very hard counter-faces, such as chromium plate or ceramics and cermets are also satisfactory but because they cannot easily be polished by carbons during sliding, care must be taken to minimize the initial surface roughness. Carbon bearings are readily available in finished form suitable for installation into housings. Machining, however, presents few difficulties provided that cemented carbide tools are used to offset the abrasiveness of carbon debris. Because of poor impact resistance and tensile properties, the wall thicknesses of carbon bearings need to be somewhat greater than those used with polymer-based bearings. Typical values range from about 3 mm for a 12 mm bore bush to 12 mm for 100 mm bore. Flanges on bushes intended to support axial loads should also be avoided because of the risk of tensile stress concentrations at the neck of the flange. Bearing aspect ratios are preferably kept below 2, again to minimize the effects of any tensile stresses arising from distortion or misalignment. It is acceptable to press-fit bushes into housings with an interference of about 2 ~m/mm maximum, provided that the outer edges of the bearings are chamfered. This technique, however, is mainly restricted to relatively small bushes, of less than about 25 mm in bore, intended for operation over a restricted temperature range. For larger bushes, or when wide temperature fluctuations are expected. shrink-fitting is to be preferred, the relative dimensions of the bush and housing being chosen so that interference will still be maintained at the maximum temperature encountered. Typical values are 4 6 #m/ram. If the housing has a very large expansion coefficient relative to the carbon bush, eg bronze or aluminium alloy, it may be advantageous to shrink-fit the bush into a steel sleeve which is then attached to the housing by any conventional technique. Preformed metal-backed carbon bearings are available commercially. Because carbon bearings are much more dimensionally stable than those based on polymers, running clearances can be significantly lower. Allowance is mainly required to offset the build-up of consolidated wear debris between the sliding surfaces, and typical clearances are about 3 # m / m m with a minimum of 50/am. Finally, it should be mentioned that because carbons do not exhibit any plasticity and are weak in tension, their tolerance towards abrasive contamination is generally lower than that of polymers. To avoid high wear rates in dirty environments, therefore, some form of shielding for carbon bearings is essential. Solid-film lubricants The advantages of solid lubricants as additives to polymers have already been mentioned. Lubricants such as MoS 2 or ptfe function essentially by forming a film over the surface of the composite or counter-face during sliding so that the friction becomes characteristic of this film rather than of the matrix material. In principle, the lubricating film is self-replenishing and should provide lubrication throughout the life of the bearing. It is obvious that an alternative method to provide lubrication is to preform the lubricating film on the surfaces before assembly By adopting this technique, the properties of the film can be optimized to achieve the lowest coefficient of friction, or the longest life, or any suitable compromize, and the substrate materials can be chosen to obtain maximum load-carrying-capacity. These preformed fdms, however, inevitably have a finite life and it is the difficulty of predicting this life in practical conditions which is responsible for most of the uncertainties associated with this type of lubrication. Table 13 gives a list of the more common solid-film lubricants, together with estimates of their temperature limit in air. This limit is determined primarily by the oxidation characteristics, and the values quoted are clearly arbitrary because oxidation is time as well as temperature dependent. Thermal stabilities of lamellar solid lubricants in vacuum or inert gas are much higher than the oxidative stabilities, and for MoS2 and similar materials are of the order of 1000°C. MoS 2 is by far file most widely used lubricant. The simplest method of applying a film of a lamellar solid to a substrate is by burnishing of dry powder on to a clean surface with a soft cloth. Particles adhere locally to the surface, mainly by mechanical interlocking with surface defects or depressions, and a coherent film builds up via the cohesive forces between the crystallites within the particles. There is some evidence to suggest that softer substrates give more adherent films whilst harder ones produce more coherent films. As might be expected in these circumstances, therefore, the effects of substrate hardness on friction and wear-life of rubbed films tends to be ill-defined. Film thicknesses for MoS 2 are usually less than about 1 #m, and increase with relative humidity of the environment and with time of rubbing. Practical applications of rubbed films, however, are few and confined mainly to assembly of components or for lubrication of precision parts when the amount of sliding involved is relatively small. The most common type of preformed solid film lubricant is a 'bonded-coating' in which particles of the lubricant are cemented together and to the substrate by some type of binder, usually an organic resin. Some of the most widely used binders are listed in Table 13, together with an estimate of their temperature limit. Because the temperature stabilities o f organic resins are generally less than those of the solid lubricants themselves, the temperature limits of the coating are determined primarily by the resin properties. For temperatures in excess of about 400°C, inorganic binders have been developed based on sodium silicate, aluminium phosphate, or various mixed oxides. The choice of binder is also influenced by a number of other factors in addition to temperature, such as mechanical properties, compatibility with the environment, ease of processing, and cost. The last two can often be of major importance TRIBOLOGY December 1973 237 TaNe 13 Types of solid lubricants and surface pretreatme~qts Maximum Temperature (°C) Lubricants Lamd|ar solids MoS2 WS 2 GrapNte TaS2 CaF 2 Other solids ptfe phthalocyanine B203-PbS SiO2oPbO Na2WO4 MoO 3 350 400 500 550 1000 300 400 550 650 700 800 Binders Maximum Temperature (°C) acry~cs cellulose any& phonetics epoxies silicones 65 65 95 160 200 300 grit -blast add-etch phosphate (steels) anodize (At, Ti) dichromate (Mg) phosphate-fluoride (Ti) oxalate (Cu) polyimides silicates phosphates vitreous 350 450 500 550 porous sintered or sprayed iavers and there is an increasing interest in the use of air-curing resins, such as acrylics and cellulose-based materia/s, which are applied, together with the lubricant itself, from pressurized aerosol containers. The ratio of lubricant to binder varies with the materials invoNed, but is usually within the range 1 : I - 4 : t. Tge higher ratios generally minimize ~ e coefficient of friction, whilst the lower ones maxLmize the wear-life. Apart from lubricant and binder, however, other additives are often incorporated to enhance one or more aspects of performance. Soft metals may facilitate re-adhesion of debris to the substrate during sliding, smalt additions of graphite and other metal sulphides to MoS2 films enhance wear-life, and inhibitors prevent corrosion of the substrate ;! by the oxidation products of MoS2' in humid eaviro~ments eg H2SO4. Sb203 is a widely-used additive with MoS~ films to increase wear-Ere, but its mode of action still remains obscure. ~n general, the precise folxnulations G~ commercially-available coatings are no~ avNlable to the user. Most coatings have been developed in the USA under the stimulus of military and aerospace requirements° anc~ because of this a greater degree of standardization has been achieved in the US than elsewhere. Four specifications are relevanu a air-drying lubricant MIL L-23398B. b general purpose, heat-cured, bonded solid fiim lubricant M i L L 8937A, c corrosion resisting, heat-cured, bonaed solid film iubacan~ MIL-46010A, d extreme enviromnem 300°F 750°F. bonded solid f t m h b r i c a n t M I L L 81329ASG. A partial list of some of the commerciN oroducts safisfyin~ these specifications is given in Table 14. There are~ of course numerous other products which, whilst not satisfying the specifications in one or more respects, may nevertheless be quite suitable for particular applications. i App]ica tfon ¢- % d~ .w d 0-25 0.5 O'75 !.0 ~'25 1.5 Substrate roughr~ess {gin de) I 75 Ficj 21 Life-surface roughness relationships for Falex tests on one particular bonded sblid film lubricant {from Peterson and Finkin 6 ) 238 Substrate pretreatments T R I B O L O G Y December 1973 The most effective method of application of bonded solid £im lubricant coatings is by spraying on to a carefully cleaned and roughened metal substrate. For small nurabers of components, brushing or dipping is sometimes used_ but it is then much more difficult to control film thickness and quality to the required standard. Even spraying is best carried out, if possible, on an automated or semiautomated basis to ensure consistency. In addition ~o cleaNng of the substrate, various types of pre-~reatment can also be "used to enhance the wear-life of bonded coatings, and some of these are ~isted in Table i3. The mos~ knportant facet of surface pre~reatment is roughening ~e increase the~egree of mechanical 'keying' of ~ie N m to the surface. Fig 21 shows how roughening produced by grit-blasting is more effective than grinding to the same numerical cla roughness. Wet git-hlasting is also more effective than dry. The essential feature reG;aired in ~ e Table 14 Partial list of some commercial solid lubricant coating formulations satisfying US specifications (From Peterson and Finkin 8) MI L-L-8937A MIL-23398B MIL-L-46010 (A) Acheson Colloids, Dag 254 Electrofitm, Lubri-Bond A Dow ('orning 3400A Dow Coming, Molykote X106 Hohman, Surfkote A1625 Everlube, Ecolube 642 Electrofihn, Lub-Lok 5306 Lubrifthn, 600A Fel-Pro C651 A Everlube 620 Electrofihn, Lub-Lok 2109 Fel-Pro (%40 Lubrifihn LF710A Hohman, Surfkote M1284 Sandstrom 9A M! L-L-8132q (ASG) Dow Corning, Molykote X1 5 Lubrifihn, LF700 Table15 Effect of coating different parts in Falextests with a bonded MoS 2 coating to MIL-L-8937 (From McCain 9) Coating applied to Wear life (min) V-blocks only Pin only Pin and V-blocks 10 958 965 roughening process is the production of as uniform a distribution of surface depressions as possible, see Fig 22a. Techniques other than grit-blasting either give anisotropic topographies, non-uniform depths of depressions, or both, see Figs 22b and c. Further improvements in wear-life can usually be obtained by phosphating of steel surfaces, as shown in Fig 21, but in practice the advantages may not always be considered to be sufficiently great to justify the additional processing cost. In addition, phosphate coatings break down thermally above about 300°C, and are therefore unsuitable for use with the higher temperature lubricant formulations. After spraying, the coating should be carefully examined to check uniformity and then cured for a time and temperature appropriate to the binder; these values are usually specified by the manufacturer. The coating thickness may be estimated by weighing or, on steel surfaces, measured with a magnetic gauge. The thickness normally recommended is from 7 17/am, and the solid content of tire lubricant-resin dispersion is, in fact, often adjusted to give fihn thicknesses of this order during a single spraying operation. Thicker coatings (up to about 50/am) may sometimes give improved life in low stress conditions and for these it is preferable to build-up the coating gradually from successive spraying/curing cycles. a variant thereof (LFW 1, one block; Macmillan, one block; '.tohman, two blocks, Dual Rub Shoe, two blocks). Both of these tests are used to obtain wear-lives during continuous running, or load-carrying-capacity by increasing the load in stages as shown previously (Fig 4) When time and opportunity permit, these accelerated tests are supplemented by plain bearing assessments, either with continuous rotation or more usually, with oscillatory motion, see Fig 23c. In all three tests it is usual to coat both of the rubbing surfaces with solid film lubricant, but the layer on the rotating surface is by far the most critical as is shown in Table 15 for Falex tests. Similar experiments with plain bearings have shown that coating of the staaft surface is responsible for about two-thirds of the wear-life and coating of the bush for about one-third. A major difficulty with all testing of solid film lubricants, both in accelerated or service conditions, is lack of reproducibility of wear-life. Even under very carefully controlled conditions a scatter of +-50% is common for Falex tests and for Timken-type tests the scatter is even worse, -+100%. Fig 24 illustrates the variations observed in the wear-life of one particular film between different laboratories using identical testing conditions and apparatus. a G r i t - blasted Across grinding marks Along grinding marks b Ground Testing The development of solid film lubricant formulations has been, and still is largely an empirical process in which friction and wear testing plays the dominant role. Because the possible combinations of materials are numerous, tests are frequently made in apparatus specifically designed to produce data in short time intervals. Fig 23 shows two of the most common tests; the Falex, and the Timken, or C Randomly abraded Fig 22 Profiles of mild steel surfaces prepared in different ways to a roughness of 0.5--0.62/am cla ~ X 5000 4+ X 100 TRIBOLOGY December 1973 239 Load a Faiex Lood Load b Timken -type Fi~ 23 C Oscillating plain b¢(lrirEj Tests for bonded solid film lubricants Falex V-blocks: free-cutting stee~, 250 VPN 12.7 mm diameter 10 mm long, 91 ° angle. Pin: Ni-Cr steel, 150 VPN 6.35 mm diameter, 31.8 mm long Typical operating conditions: V = 290 rev/min (~0.1 m/s), W=upto13MN T~mken-type Ring: C-Cr steel, 750 VPN 35 mm diameter, 8.9 mm long Block: Mo-steel, 620 VPN 15.4 mm X6.35 mm X 10 mm TvpicN operating conditions: V = 72 rev/min (~0.13 m/s}, W=3MN Oscillating plain bearing Shaft: Cr-ptated Mo steel, 1000 VPN 15.8 mm diameter Bush: C-Cr steel, 750 VPN 22 mm od 16 mm wide, 0.1 mm clearance Typical Operating conditions: 64 ° arc, 10 cycles/rain, W = 70 MN Because each type of test uses a different geometrical sliding arrangement as welt as different loads and speeds, it is obvious that there is unlikely to be any correlation o f weardives when these are expressed in terms o f time or distance o f sliding. However, a reasonable correlation has been shown to exist if the life o f the coating is expressed in terms of the number o f cycles o f compression/ flexure to which each element of the film has been subjected 9. [n a Timken-type test with a coated ring, there is one such cycle per revolution o f the ring, whereas in the Falex test there are 4 cycles/revolution. There are complications involved in extending this concept to the case of oscillating plain bearings because in these conditions the load m a y never be removed from part of the coated shaft. In these situations o f conforming geometry, there is also some evidence to suggest that wear-life is lower than with nominal point or line contact geometries. The reduced life has been attributed to the fact that debris cannot readily escape from the contact areas, and this explanation appears to be confirmed by the observation that the wear-life increases if grooves or depressions are provided in one or both of the sliding surfaces. F r i c t i o n a n d wear Wear volume-time relationships for b o n d e d solid film lubricants are generally o f the type already shown in Fig t 1. In the early stages o f sliding, loose material is rapidly removed and the film is consolidated, the thickness being reduced b y as m u c h as 50% in the first few hundred revo- 240 T R I B O L O G Y December 1973 lutions. The wear rate then decreases to an extremely iow value, sometimes virtually to zero, as a consequence of sintering of the particles together ~ d the formation of a smooth, c~stallographicaliy orientated surface layer~ Failure ultimately occurs either as a resui~ of continuous wear down to the substrate or to the gradual build-up o f compressive stresses within the coating leading ~o 'bl~stering' and a loss o f adhesion to the substrate, i n practice It is oRen difficult, if not impossible, to differentiate between these ~wo failure modes. Although the wear-life o f a bonded coating thus comprises two par~s, there is little or no information available about the effect of the conditions of sliding on each part separately. It is consequently difficult to apply the concept of a specific wear rate to solid fitm lubricants except as a mean value over the whole wear-life. WbJtst such values lack precision, they may nevertheless be useful for preliminary design purposes, and Tabie t 6 shows estimates for several types of solid f~.m lubricants based on accelerated tests, tt m a y be noted that these mean wear-rates are not, in general, partic~Aarly iow in comparison ro the steady-state wear-rates of fitled ptfe's (Table 5) and some carbons (Fig 17). Both the wear tile and the coefficient o f friction of bonded coatings are svrongly affected by the conditions of sliding. The general trends are shown in Fig 25, and it can be seen that some parameters produce opposing effects on friction and on wear F o r example, h e a ~ loads and !ow fi!m thicknesses reduce the coefficient of fr/ction, but they also tend ~o reduce the wear life. The magnitude of "the coefficient o f friction obviously depends on the particular coating formulation used and the conditions of sliding. Values as iow as 0.03 are not u n c o m m o n for MoS2 f:flms o f the order o f 5 g m thick a{ heavy loads m conditions o f nominal point or !ine contact. With decreas= ing load, increasing Film thickness, or an increase Ln_the apparent area of the surfaces towards a conforming geometry, the coe~]cient of friction m a y rise to values of ~ e order of 0.15 0.2. It is often observed that the coefficient of friction o f MoS2 films also tends to decrease with increasing time of sliding, ~ypically by factors of abo~t 1.5 2. This is partly attributable to the reduction of fitrn thickness which occurs during the early stages o f sliding, together with the development o f a preferred orientation o f the crystallites, and partly to an increase in surface temperature which leads to loss of physicaily adsorbed water vapour and a weakening of the [nterparticie bonds° q x R u i e I; ooo ] .,".:o : .... fi O 3 5 7 Fdex Fig 24 me~ me em U t v I - 9 A C E G ~ K Im i3 Different laborGtori~s LFW-~ Reproducibility of wear Hfe of one bonded solid film lubricant in two tests in different laboratories {from McCain 9) Table 16 Order of magnitude values of the mean specific wear rate for various types of MoS 2 bonded coatings at 20°C (From Finkin 1°) Mean specific wear rate (mm3/N m) Coating Ceramic MoS 2 MoS2 MoS2MoS2 MoS2 - 10 4 10-5 10-5 10 6 10-6 10 7 bonded oxides ceramic bonded (no metals) thermosetting resin binders (except polyimide) Sb203 - polyimide binders graphite - sodium silicate binders (above 200°C) graphite - metals - ceramic/glass binders The effects of substrate hardness on the wear life of solid film lubricants are conflicting, but do not, in general, appear to be very great. Softer substrate metals have the advantage that the film may sometimes be able to repair itself following a localized penetration, and failure is not then so abrupt as on the harder substrates. Greater effects on the wear life of MoS2 coatings have been observed by changing the type of substrate metal from steel to molybdenum, or a high molybdenum alloy such as TZM (Mo + small proportions of Ti and Zr). The increase in life obtained with Mo substrates is greatest in conditions of sliding where the localized flash temperatures become sufficiently high (>1000°C) to dissociate the MoS 2 thermally. Sulphur may then react with the Mo to reform a lubricating film. Summary The main advantages of solid film lubricants for dry bearings, compared to polymer- and carbon-based materials are that: a it is possible to use very high loadings, up to the yield stress of the substrate metal, b high speeds can be tolerated because the films are comparatively thin and of thermal conductivity comparable to that of steels, c formulations are available for use at temperatures up to 1000°C (CaF2/BaF 2 eutectic coatings) d very stiff bearing assemblies can be obtained by using thin films, and little or no back-lash is introduced as a result of wear, e coatings are able to provide the lowest coefficients of friction of any sliding dry-bearing system, except perhaps those based on ptfe fibre at heavy loads and low speeds. Against these advantages must be offset the facts that: a the prediction of the wear life of a coating is only possible in order of magnitude terms and tests are virtually mandatory, b the performance of the great majority of coatings is extremely sensitive to fluid contamination during operation, leading to greatly reduced life, c careful attention to detail is essential at all stages during processing, and in particular all forms of fluid contamination must be avoided, d even with die processing conditions optimized it is not usually possible to prepare coatings with wear lives reproducible to better than +50%, e the ultimate failure of solid film lubricants at heavy loads may be rapid and catastrophic, leading to seizure. Increase Mo substrotes High speeds Substrate pretreotments Increased film thickness I High humidities I Friction coefficient Wear life Heavy loads High temperatures High humidities Rough counterfaces Abrasive contamination Fluid contamination Heavy loads High substrate hardness Reduced film thickness High temperatures Line or point contracts Decrease Fig 25 Factors affecting the wear life of bonded solid film lubricants Little or no mention has been made of the various hightemperature solid film lubricants in Table 13. Apart from formulations using MoS 2 sodium silicate (Molykote X15) and glass-bonded 1~oS2 (Vitrolube), these are not generally available commercially and their use for any given application normally involves a development and test programme by the user. Pecause of the major part played by processing in the performance of solid film lubricants, there is a growing trend towards custom-coating of parts by specialized processors. Most coatings can be obtained in this way, sometimes from the manufacturer of the coating itself. There are also a number of specialized coating treatments which are only available on a custom-coating basis. These include aluminium or titanium alloys with anodized surfaces impregnated with ptfe ('Canadizing' "Tufram') and coatings of MoS2, graphite etc, produced by particle impingement, electrostatic spraying or sputtering. At the present time there is insufficient data available on these specialized coatings to evaluate their performance generally in comparison to the more conventional resin-bonded types. TRIBOLOGY December 1973 241 TaMe 17 Some commercially availabfe composites of metals-!ametlar solid ~ubricants Material [steaded. application Tra de name/S~pp~iez Ag-ptfe Ag-NbSe 2 Ag-Cu-MoS2 Small bearings Polymer AG~ Polypenco ~ Ag-ptfe-NbSe 2 I Ag-WSe 2 Ni.WS 2 a Ni-CaF 2 l Ag-WS2-polyi~Jde Metats-polyimide-lameliar solids Ta-MoS2 Ta-Mo-MoS 2 Ni-NiO-CaF 2 Bronzes 1 ken - Graphite Ni Bronze-soIid lubricant inserts Bronze-MoS2 1 Bronze-ptfe-MoS2 Ag-porous carbon vacuum, electrical contacts Slfding electricalcontacts Self-lubricating bearing cages Self-lubricating alloys Berne! High temperature bearings Self-lubricating bearing cages vacuum High toad bearings vacuum Self-lubricating bearing cages - high temperature High !oad beatings High temperature bearir~gs Feuralons Feuralloys 1 Molalioys, P~re Carbon Co, ,TX4organiteo553 General applications. Moderate ~emperamre Deva Metals General applications, particularly at heavy loads Lubrite, Fraaerlube Sinite Sinitex Grapha!loy Self-lubricating retainers Metat-soJid lubricant mixtures As an Nternative to preformed solid lubricant Nms on metals, numerous attempts have been made to incorporate the solid lubricant within the structure of the bearing metal itself to provide lubrication continuously. One of the earliest ways of achieving t_his was to machine grooves or holes into conventional bearing altoys and frill the recesses with a solid lubricant originally based on graphite. Bearings of this type have now been available for many years, eg 'Lubrite' and Franertube', and have become increasingly sop~sticated with a wide range of different alloys and types of lubricants. Unfortunately details of the various lubricants are not divulged by the manufacturers. P V factors range from 0.5 MN/m 2 X m/s for continuous operation to about 3 MN/m 2 X m/s for low speed, intermittent service. The maximum speed is about 2.5 m/s. In contrast to all the materials so far discussed it is claimed that optimum performance is obtained with rough shafts of the order of 1.5 3 #mcta. Recommended clearances are 2 #m/mrn with a minimum of 75 #m and are thus very similar to those for carbon/graphite bearings The coefficient of friction decreases with increasing load from about 0.15-0.05. There is no information available on wear. More recent developments have concentrated on the production of more uniform mixtures of metals and solid lubricants, and one product of this type, porous bronze/ ptfe/Pb, has already been described. Mixtures of metals and graphite are produced by powder-metallurgical techniques (Deva metals) and the graphite contents range from 6-25% wt (~12 50% vot). Metal matrices include bronze, leaded bronze, brass, iron and nickel, the particular choice being dictated by the temperature of operation, tn general, the composite strengths tend to decrease with increasing graphite content, but the coefficients of friction and wear rates also decrease. Friction coefficients of the bronze-graphites during dry operation against steel are typically 0.15-0.3. The particular advantages of these 242 ) TR~BOLOGY December 1973 materials over plain carbon/graphites and carbon/graphites containing small proportions of metal are a grea~er resist.ance to impact and shock-ioadings, and a higher load carrying capacity of !5 35 MN/m 2 for the bronzegraphites° The specific wear rates, however~ are somewhat greater than those of carbon/grapNtes and are typically in the range10 5 t0 4 m m 3 / N m . Numerous types of other metai-lm~etlar solid rmxtures have been developed for particular applicationso mostly m aerospace, and some of those which have reached commercial exploitation are listed in Table 17 tn connection with all these materials, it is worth making the genersJ point that lubrication is provided v/a the development of a transfer N m of the so!id lubricant o~ the counier-~ace surface. Transfer of lameHar solids is an inefficiei~t process and. in general, a considerable excess of wear debris _~.s required to form a sufficiently coherent fi!m to reduce friction and wear. There is consequently a tendency for the wear rate to correlate inversely with the soefficient of friction, and this is illustrated in Fig 26 for a number of materials for which the larnei!ar so!id is a minor constb men~. Ptfe, however, transfers readily to rnetals and no~ oNy functions as a lubricant in Rs own right but may also facilitate transfer of any lamellar solid present° The ~mproved performance of the metea-ptfe-lametlar solid mixtures over those without ptfe may be noted° V e ~ recently, a group of composites has been developed based on Ta-Mo-MoS2 and containing >5£~o of MoS2, ie 'Molalloys'. These exhibit reasonably tow wear rates and coefficients of fliction, as shown in Fig 26. and are suitable for operation a~. temperatures up to ~500°C in air or ~1000°C in inert atmospheres or vacuum. Despite the higg~ content of MoS2, the compressive strengths can be as higiq as 700 MN/m 2, but the tensile strengths are at least a factor of t0 lower, and the materials are relatively brittleo Similar assembly techniques are used to those already described for carbon/graphites. FinNly, it may be noted that the specific wear rates of all the metai/lameltar solid mixtures in Fig 26 are significantly greater than the porous bronzeptfe-Pb composite already described. Their main area of application as bearings, therefore, lies at a temperature in excess of that pemrissible for ptfe, ie above about 275°C. 10-2 l\\ \ --Cu-lamellar solids Solid lubrication of rolling-element bearings All discussion so far has been concerned with materials for plain bearings where the coefficients of friction are of the order of 0.05 or greater. In some applications, the magnitude of the coefficient of friction may be a critical parameter because of restrictions on the power available or because of difficulties in heat dissipation. The use of rolling element bearings is then indicated. Such bearings are "also preferred for many instrument applications where precise location of shafts may be critical. Conventional rolling bearings operated without any fornr of lubrication whatsoever exhibit relatively short lives as a result of retainer, ball and race wear. The problem of dry operation thus becomes one of selecting the appropriate nrethod of solid lubrication, and there are three possibilities: continuous supply of solid lubricant powders (MoS> graphite/CdO mixtures, phthalocyanine) in an inert carrier gas preformed fihns of lubricants on the surfaces of the retainer, balls and races self-lubricating retainers which provide lubrication by transfer t~ the balls and/or races. The first technique has so far been used only for feasibility studies on the lubrication o f high temperature bearings and has not achieved commercial exploitation. Attention will therefore be concentrated on the remaining two methods. II is extremely difficult to reach general conclusions about the performance of solid lubricants in rolling bearings. The problem is not simply one of the correct choice of solid lubricant because design, type and materials of the bearing itself play an equal or even more important role in performance. It is widely accepted that the optimum lubricant/bearing combination for one application is seldom the same as for another, and in the space available here it is only possible to outline a few of the combinations which have shown promise in particular applications. Solid films All the types of bonded solid film coatings described earlier are potentially suitable for use in rolling element bearings. Because of the limited amounts of sliding involved, it is also possible to use thin films of lamellar solids produced by burnishing of dry powder, sputtering, or from the conversion of Me coatings to MoS 2 in H2S. Most of the interest in solid lubricant coatings has arisen from space applications where finite, and relatively short, bearing lives can often be tolerated, provided that test programmes are able to establish these lives with a reasonable degree of certainty. For vacuum use, MoS 2 is normally the preferred lubricant over other types of lamellar solids: as nrentioned earlier, graphite is ineffective in these conditions. In one particular example, a life o f > 1 0 8 cycles has been obtained for MoS2-Na2SiO 3 coatings on the retainer and races of 440C stainless steel bearings of R4 size (~(~.3 mm bore) o]~erating in a vacuum of l0 9 torr (133 × 10 9 N/nrZ) at light loads (~500 g) and low speeds (~400 rev/min) 11. Other work with similar sized bearings has shown, however, that the lives of various MoS 2 coatings m vacuunr are very irreproducible. By far the most success- \ \ Mixtures of grephil¢ Ag MO S2 Cu WSe2 Ni & NbS¢2 Co MoSe 2 F¢ MoTe 2 Cu -Ni CoF2/Bal O ~. \ °0 o E" I O - 4 Z 0 71~ 0 0 ~q o\ ,.g \ o iO-s ~a • o "- .. \ , "=- ." . . . . U~lO-6 t o\ \ ~ graphite '\\ Bronze \graphite o ......." Ta_Mo_MoS2 ~-,.\~.~ ~Metals -ptfe lamellar solids \o~ o \ \ \ 1(7-7 Porous b r o n z e - p t f e -Pb w o v e n p t f e f i b r e / g l a s s fibre iO-8 0.2 0-4 O.6 0.8 I.O 1.2 Coefficient o f f r i c t i o n Fig 26 Relationship between wear rate and coefficient of friction for metal-lamellar solid composites sliding on metals (from various sources) ful vacuum lubricant was a thin electroplated lead coating on the races. Lives in excess of 1010 cycles at 3000 rev/ rain were obtained with full-complement bearings and the addition of a retainer reduced the life somewhat, although the torque was "also reduced. The optimum retainer material was a leaded bronze. Other soft metal fihns as lubricants for rolling bearings in vacuum have also been examined (Ba, Ga, In, Sn, Ag and Au) and promising results are reported for Au and Ag in lightly loaded R2 bearings (~3 mm bore) at 10000 rev/min t3. The deposition of these films, however appears to be more critical than for Pb; gold plated balls in conjunction with silver plated races are much more effective than the opposite arrangement. Self-lubricating retainers The solution most widely adopted for vacuum lubrication of rolling bearings is to use a self-lubricating retainer fabricated frona a composite of ptfe/MoS2/glass fibre (Duroid 5813, Bartemp). Bearings of this type are available commercially in sizes up to 12.5 mm bore, together with a limited amount of design data. Compared with fluid-lubricated bearings, the ratio of the total dynamic load to the total static load is very small and ranges from about 4% with 2.36 mm bore to about 0.3% with 12.5 mln bore. Within the recommended limits of maximum load (~2 kg for 12.5 mm bore), lives in excess of 109 revolutions have repeatedly been obtained in vacuum. Some success has also been reported in extending the use of T R I B O L O G Y December 1973 243 filled ptfe retainers to larger bearings ~4. Lives exceeding 108 revolutions were obtained with 204 bearings (20 mm bore) operating with radial loads of about 5 kg in a vacuum of 10- 5 tort at room temperature. The life decreased by a factor of about 10, however, when the temperature increased to 150°C. One difficulty which arises in attempting to collate information on the performance of different lubricants in different bearings is the dependence on bearing design. This has been demonstrated very forcibly in some recent work where the performance o f 6 different types of 204 size bearings was compared in vacuum in the absence of any form of lubrication ~s. The bearings involved different degrees of precision, different retainer materials and designs, and different numbers of balls. The average lives showed a variation of 400 000 to i, ranging from 3 rain to 20 000 h, the most successful being the bearing with highest precision, a greater number of balls than usual for this size (11 c f 7) and a leaded-bronze retainer. This bearing also performed well in air without lubrication, although the life was a factor of 1 0 - 2 0 times lower than in vacuum. Bearings with ptfe/MoS2/glass fibre retainers are also effective in air over the temperature range - 185°C to 300°C, provided that there is no condensation of vapours to interfere with the transfer process of ptfe to the balls and races. A major area of application of these bearings is, in fact, at temperatures above !50°C where the performance of miniature bearings lubricated by even the best synthetic high temperature fluid lubricants begins to deteriorate markedly. At 200°C, the life of fluid-lubricated miniature bearings seldom exceeds 108 cycles, whereas dry bearings with ptfe/MoS2/glass fibre cages wi!l give lives of this order at 300°C. As an alternative to fabricating retainers from filled ptfe composites, and for operation at temperatures exceeding 300°C, metallic retainers can be made selflubricating by suitable modifications, eg machining holes or grooves in critical areas, which are then filled with solid lubricant. Some examples of the various designs which have been investigated are given in Fig 27. The most successful appears to have cylindrical reservoirs in the ball-pockets together with rectangular reservoirs on the inner surface of the retainer and on the lands of the inner race. MoS2graphite-sodium silicate has been the solid lubricant formulation most widely used with these designs and some results obtained with 204 size bearings at 400°C are shown in Table 18. The improvements obtained by providing lubrication in the lands as well as in the bali-pockets are clearly evident, and it may also be noted that molybdenum is the rnost effective retainer material. In connection with the solid lubrication of relatively large ball or roller bearings at high temperature, two recent developments in retainer materials are worth noting. Boes and co-workers iv have produced a series of self-lubricating composites in which the lamellar sotid hibricant VTSe2 is bonded together with a gallium-indium alloy. A typical composition is 80% wt WSe2 and 20% Ga-In (75/25), with a compressive strength of ~15 MN/m 3. Because the tensile strength is tow, the most satisfactory design of retainer is one in which the composite is shielded by a metal of similar coefficient of thermal expansion, such as titanium. With this retainer material, lives of up to 108 cycles have been obtained for 207 size bearings (35 mm bore) at 420°C &qd loads of 45 kg and speeds of 244 TRIBOLOGY December 1973 10 000 rev/min. The second development is by Van Wyk and co-workers ~s who have prepared a series of composites based on MoS2 with minor proportions of refractory metals, such as Ta or Mo. The compressive strengths are similar to those of the WSe2/Ga4n composites, but the tensile strengths are significant!y higher, which fac~ities the fabrication of retainers. Some form of metallic support for the larger sizes is still, however, desirab!e: One potential application is for 'fail-safe' conditions in helicopter rotor bearings in the event of failure o f the main oit suFpiy. The target life is 30 rain at 12 000 roy/rain with a radial load of 1350 kg, and conventional metalIic retainers normatty ey&ibit lives of less than 5 rain. A retainer of MoS 2-Ta-Mo, however, has given lives in excess of 60 rain, and in the complete absence of oil, where transfer film formation can become more uniform,, lives of up to 22 ir have been obtained. Summary An attempt to summarize the above data in rerms o f a tileload reiationship is shown in Fig 28. The individual lines refer to different sizes of bearings operating at different speeds and represent the best result achieved with any one particular system. For comparison, two lines are drawn for the El0 Iives of conventional bearings lubricated by ~uids, and the difference in slope may be noted. Failure under lubricated conditions results from fatigue o f the bails or races and in these conditions iife is inversely proportional to the third power of the load. With self-lubricating bearings, however, life is determined either by cage wear or by wear of the lubricating film, and ~o a first approximation, iife is therefore inversely proport!onal ro load !t is a~so instructive to compare the life of self-lubricated rolling bearings with that of a porous bronze/ptfe/Pb sliding bearingo Data for the latter has been derived from Fig i2, assuming a speed of 1500 rev/min, and is shown by the dotted ?hqe in Fig 28 The lives of roiling bearings with self-lubricating cages are clearly of the same order of magnitude as those of the ptfe-based sliding bearing, and the roiling bearings have the additional advantages of a hig~her limiting speed and temperature, and a lower coefficient: of friction. The major factor Precluding the more widespread use of selflubricating rolling bearings a~ present is the cost. Table 18 Effect of retainer material and position of soiid lubricant reservoirs (Lubricant 71% weight MoS2, 7 % weight graphite, 22 % weig,~t Na2SiO 3. Tool steel (M10) balls and races, t 0 0 0 0 rev/min, 400°C, Radial load = 1.4 kg, Thrust load 2.3 kg) {From Devine et at/% Life(h~ Retainer material Reservoirs in cage only Reservoirs in cage ane lands Fe-Si-bronze Tt Tool steel ~W-Cr-V) M10 Tool steei (Cr-Mo-V) Mo-0.5% Ti-0.08% Zr 25 32 61 [ 07 ! 50 ~39 300 ~~48 Fig 27 Reservoir designs and locations for self-lubricating retainers in rolling element bearings (from Devine et al, Ref 16) TRIBOLOGY December 1973 245 Leed film~ m} R416-3mm) 207 (35ram) IolO Table 19 'Hard metals', and 'super a!~oys' suitabie for nigh tern perature bearings tO 4 IX'o',;;~lass :~';~ MOS -Te-MO I,;ooI~o,,:'\ ,,~;,°o, , r:-X lub icat¢ Material ~ II09_ e0 w $¢2 -G~ - I n I~°8 [42OOC) p t f ¢ - MoS2-gless iTs= , 50°<} -..... X Iolr) =107 Io Limiting :emperature ea i g ,oo= lo2 ~1~3 184 Lood,IN) Fig 28 Life-load relationships for rolling element bearings w i t h solid lubricant films or retainers (from various sources) High temperature materials The choice of bearing materials for operation at temperatures in excess of 500°C is somewhat restricted and it is usually necessary to compromise on the conflicting requirements of low friction and low wear. Many hard metals, and superaltoys exhibit wear rates which tend to d rcrease with increasing temperature as a result of reaction with the environment and the formation of protective oxide layers: however, the coefficients of friction of these iayers seldom fall below about 0.2-0.3. Some materials which have been used for high temperature bearings are listed in Table t9, together with an estimate of their limiting temperature. The latter arises partly as a result of increasing oxidation and loss of material during sliding from this cause, and partly because of loss of strength and elastic modulus. For temperatures in excess of about 800°C, or for applications at very high speeds of sliding where the localized flash-temperatures are sufficiently high to me!t most metals, a variety of ceramics and cermets are available. Some of the more Mo alloys (TZM) Mo t o o l steels Nitrided steels 500°C HastetIoy C (57% NL 17% Mo, 16% Cr, 5%Fe TMn, Si, C) Stetlite d Stellite Star J (43% Co, 32% Cr, 17% W, 3% Fe, +Ni, C. Mn. Si) 750°C [ncone! X (73% Ni, ! 5% Cr, 7% Fe, 2½% Ti + Mm Si. Nb,0 Stellite ] 9 Rend 4! (55% Ni, t9% Cr, 10% Co, t0% No, 3% Ti + At, Fe. Si, Mn, C, B) 850°C widely used compositions are shown m Table 20. The advantages of cermets (metal-bonded ceramics) over ceramics alone are increased toughness, ductility and resistance to shock loads. However, with increasing metai content the overatl hardness decreases and the wear rate tends to increase, as shown in Fig 29a for tungsten carbide-cobalt mixtures. The two cermets containing A1203 have been found parficu!arly suitable for high temperature bearing applications: LTIB 19% A1203.59% Cr, 20% No and 2% TiO2; LT2 --- 15% A1203. 60% W and 25% Cr There are few general guide lines from which to predict 1 --.~ZE I0-'~ I Z 0 IO-S[ E 0 ~9 0 E? t_ IO~L- OI u u u u a_ IO-~ o_ uO U3 I 4 8 I 8 I I 12 16 O/o C o b a l t o-81 R - - 20 24 Fig 29 a Variation of wear rate w i t h cobalt content for tungsten carbide-cobalt m ixtu res sliding on hard 18% W tool steel. W = l O 0 - 5 0 0 N , V=O.2-3.2m/s 246 T R I B O L O G Y December t 9 7 3 i0_71 b 0 lC !O2 03 L o a d tNi b Variation of wear rate w i t h toad for tun£sten carbide/12% cobalt sliding on tself. V = 0,7 2,6 m/s Table 20 Types and properties of some ceramics and cermets Ceramics Cermets a-A12° 3 B4C TiC I Cr3C2 SiC WC Si3N4 A1203 Cr -- Mo ZrO 2 (MgO-stabilized) A1203 W Ni Coatings A1203 + TiO2 Ni Co Mo + Cr2C 3 + Ni/Cr/Co Cr WC + Ni/Fe/Co/Cr Cr Cr203 + Cr/A1203 NiO Typical properties Thermal stress resistance factor]" (°C) Materials UTS (MN/m 2) E (GN/m 2) Thermal diffusivity* (10 4 m2/s) a-Al20 3 240 360 0.08 60 B4C 240 450 0.3 100 170 70 0.1 50 LTIB (19% A1203, 59% Cr, 20% Mo) 280 260 0.1 150 K162B (64% TiC, 25% Ni, 5% Mo} 800 380 0.15 250 20 8 Si3N 4 SiC Graphite 1 >1000 * Thermal diffusivity = thermal conductivity/density x specific heat t l'hermat stress resistance factor = o(1 v)/o.E. IO -~ the wear of ceramics and cermets from a knowledge of their properties and composition. One major difficulty is the marked dependence of properties on minor changes ira composition or in methods of manufacture. The wear process at high speeds has been attributed to thermal fatigue on an asperity scale due to repeated cycles of localized heating and cooling. This concept is supported by tile fact that there appears to be a significant correlation between the wear rate and a thermal stress resistance factor, as shown in Fig 30. The scatter is too great, however, for the results to be of much value for design or prediction purposes. It may be noted from Table 20 that graphite has by l;ar the, highest themral stress resistance factor of any material. Its use for high temperature or high speed bearings is limited, however, by oxidation above about 5000( ` and by the fact that adhesive wear and high friction become important in the absence of condensable vapours for adsorption. Various techniques for producing ceramic and cermet coatings up to 0.5 nrm thick on metal substrates offer a convenient way of utilizing the wear resistance of these materials with a minimum of processing cost. Plasma spraying is the most widely used of these techniques but more recent developments include impingement coatings from a detonation gun ('Linde flame plating') and electrolyric co-deposition from an electrolyte containing ceramic particles ('Tribomet coatings'). The advantages o f ' f l a m e plating" are an improved adhesion to the substrate metal and a lower porosity, whereas the advantage of electrolytic co-deposition lies in its ability to coat small internal surfaces inaccessible by any olher technique. In each case the c C' O E O E =L O O O -6 IO o o E i- o o° o o 0 o_ i 0 O oO 0 0 O °o° o o o O o© IO O O I I O -I O O I I I IO I IO 2 Wear rote {t~m/min) - theoreticol, derived from W = 1,5 p.R -p25 D -O'75 Fig 30 C o r r e l a t i o n of the wear rates o f ceramics and cermets at high speeds w i t h an empirical r e l a t i o n s h i p involving their thermal and mechanical properties ( f r o m Sibley and A l l e n 6 } TRIBOLOGY D e c e m b e r 1973 247 Z + @ Z ~ 6 8 I 9 Sp¢c.ific wcee F=%¢ ( ~ m ' / N t ~ ) 0"~ LWt Coatin9~ QgainsC themselves 540oC ~0 - ~ ] 0o~ LWI N30 LC4 LCRH ©.?_~ O,Z7 LA2 Coatinqs against ~hcmselv¢~ 7~0oC 0.~I LW5 LClC O.l@ Lr.,5 - - O.ZZ 0.@ - 0 . 7 LA7 0.2.¢ LCC;A 0"f9 LCIC C o ~ i ncjs __ LW 5 Qgainst Haynes c111oy 0'~9 O-Z3 LA7 ~ LC5 Z5 ?(~OOC 0-17 LAZ C o e f f i c i e n t of frictJom i At. -tO0 ° c Composition and properties of coatings Hardness VPN LW1 LW1N30 LC4 LC9H LA2 LW5 LC1C LC5 LA7 LC9A WC - 9% Co W C - 13%Co Cr203 Cr203 2O% Cr (heat-treated) A1203 WC 5%Ni Cr2C 3 - 15% Ni-Cr Cr203 20% A1203 A1203 40% Ti 02 Cr203 40% Cr 1300 1150 1300 i !00 1075 800 925 950 T max (°C) 540 540 540 870 000 760 000 870 700 870 Expansion coefficient ( t 0 -6 °C-1) 218 218 8.1 8.1 !75 !75 84 85 56 127 56 77 6.8 8.3 - 70 140 ~25 70 63 Apart from the coatings themselves, other suitable counterfaces are: Haynes alloy 25; 20% Cr, 15% W, 10% Ni, 3% Fe-Co, 1 °5% Mn, 0.1% C Hardened stainless steels (EN 59, 440 C) Haynes LT-1B; 19% A1203, 59%, Cr, 20% Mo, 2% TiO2 TiC cermets (K161, 162 and 163) and for light loads, carbon/graphites Thrust bearing tests, P = 3.5 MN/m 2, V = 0.05 m]s, surfaces ground and lapped to 0.025 0.05 #m cla Fult fines - rotating member, dotted lines stationary member (Data from Union Carbide 2°) Fig 31 248 Specific wear rates and friction of various ceramic and cermet coatings at high temperatures TRIBOLOGY December 1973 Bond strength to substrate Elastic modulus (GN/m 2) (MN!m2) properties of a coating depend at least as much on the processing conditions as on its nominal composition. Generalizations about specific types of materials are therefore very difficult to make; each coating is unique. The friction and wear properties of a series of ceramic and cermet coatings from one particular manufacturer are shown in Fig 31, based on thrust washer tests. The most suitable mating surface is usually another coating of either the same or a similar composition, but where this is not possible hardened stainless steels may be satisfactory. The wear process of ceramics and cermets often exhibits discontinuities analogous to the transitions between mild and severe wear found with many metals, and an example is shown in Fig 29b. Above the critical load, the localized stresses within the cermet are sufficiently high to cause failure of the ceramic-matrix bond, and the size of the wear debris is therefore comparable to the size of the ceramic particles in the composite. In escaping from within tile contact zone these particles cause further stress concentrations and damage leading to roughened surfaces which help to maintain high stresses and a high rate of wear. On the other hand at light loads, and with surfaces which are initially smooth, the wear process is one of gradual attrition of the ceramic particles rather than complete detachment. The surfaces then remain relatively smooth, the localized stresses are low, and the wear rate, in turn. also remains low. It follows that the avoidance of localized stress concentrations is a major requirement in maintaining low wear of ceramics and cermets, and for this reason, the initial surface finish should always be as smooth as possible. In obtaining these smooth surfaces, however, care must be taken to ensure that the finishing process itself does not introduce surface or sub-surface defects which weaken the bonding of the ceramic particles. The wear resistance of many ceramics and cennets can be ruined irretrievably by excessively severe grinding operations. Taking into account all the complicating factors discussed above, it is clear that attempts to generalize the specific wear rates of all the different hard metals, ceramics and cermets are fraught with even more uncertainty than for other materials. However, Table 21 shows tile orders of magnitude to be ¢- ~oo u expected in conditions o f relatively high stress at 500°C. In addition to their use for high temperature plain bearings, some hard metals and cermets are suitable for rolling element bearings and gas bearings. The main advantages for these applications are a high hot-hardness and dimensional stability. Fig 32 shows how the load-carrying capacity of some materials fabricated into rolling element bearings varies with temperature in comparison to a standard carbon steel bearing. The temperature limit is largely the result of thermal softening. For even higher temperatures, ~1100°C, rolling bearings have been fabricated from sintered a-A1203 and ZrO2. The difficulty with most ceramic or cermet rolling elements, however, is their tendency to failure by chipping under the high stress concentrations at the ball-race contacts. ZrO2 is reported to be less prone to this defect than At203, possibly because its lower elastic modulus may help to relieve the contact stress via elastic deformation. The lives of all these bearings tend to be very short, however, unless some form of solid lubrication is provided, either by coatings or by selflubricating retainers as discussed earlier. Conclusions Despite the very large number of dry bearing materials discussed in this survey, the choice of the most suitable material type for a particular bearing application is usually not too difficult. As already mentioned, prediction of the wear rates to be expected often gives rise to most uncertainty, and it is therefore useful to condense in a single chart all the data on specific wear rates given earlier. Fig 33 shows this summary. Once again, it must be strongly emphasized that the specific wear rate cannot be regarded as a unique material property because its value will depend on the particular conditions of sliding involved. No precise predictions are therefore possible, and tests under service conditions, or close laboratory simulations thereof, should always be made wherever feasible. If the only major material requirement for a bearing application were a low rate of wear, the choice of material would be obvious from Fig 33. However, this is seldom the case because numerous other factors are normally involved. By identifying these factors separately, it is possible to draw up short lists of the most suitable types of materials and from comparison of these lists, the appropriate compromise choice can then be made. A selection chart o f this type is given in Table 22. Within a given group of materials, the final choice can be more difficult because 80 o_ i -Mo) 6o Table 21 Order of magnitude of the specific wear rates for various high temperature materials sliding against themselves at 500°C. Based on pin/disc type tests 40 Specific wear rate (mm3/N m) u I io/~ r- ~.~, 2o \ O o 3- 2dO ......... 400 \ ~A~ \'~st¢~'~' \ \ 6dO 8bO cast N stellit¢ I000 Temperature (°C) Fig 32 Limiting load capacity of various rolling-element bearing materials at high temperatures (from Glaeser 21 ) Material Ceramics (A1203, ZrO2, SiC) Nickel-base alloys Tool steels Co-base alloys Cermets (WC-Co, TiC-Ni-Mo; Cr3C2-Ni-Cr; A1203-Cr-Mo) 10 10 10 10 3 3 4 5 10--5 10 5 10 5 10.-6 10-5 10-7 T R I B O L O G Y December 1973 249 Table 22 Selection of bearing materials for various conditions Operating requirement Decreasing suitability > Low wear/long life Low friction High temperatures Low temperatures High loads High speeds High'stiffness Dimensional stability Compatibility with fluid lubricants Corrosive environments Compatibility with abrasives Tolerance to soft counter-faces Compatibility with radiation Space/vacuum Minimum cost 5 11 10 3 9 11 11 10 7 10 1 1 7 11 1 3 9 9 tt 10 9 9 11 10 7 3 9 4 9 2 KEY 1 Unfilled thermoplastics 2 Filled/reinforced thermoplastics 3 Filled/reinforced ptfe 4 Filled/reinforced thermosetting resins 5 Ptfe impregnated porous metals 6 Woven ptfe/glass fibre 7 8 9 10 11 6 5 11 9 6 8 5 9 4 3 2 2 9 4 3 7 3 ? 4 5 5 4 7 8 4 4 3 10 3 4 do J g ~' 5 g 2 2 < /~ 2 2 Carbons-graphites Metal-grap.hite mixtures Solid film lubricants Ceramics, cermets, hard metats Rolling bearings with self-lubricating cages the detailed properties o f individual materials are either not always k n o w n or are not available to the user. This situation applies particularly to carbons-graphites, ceramiccermet coatings, and some o f the recently developed polymermetalhamellar solid composites. In such cases, final selection is often possible on the basis of past experience in similar applications, and most manufacturers o f dry bearings and materials are able to provide information of this type. Midland, Michigan, and Fig 27 by the American Society of Lubrication Engineers. This paper is Crown Copyright and is reproduced b y permission o f the Controller, Her .Majesty's Stationery Office. Acknowledgements The follovAng short ~ist of review papers and book chap ~ers maybe helpful in providing further detaited information on some of the materia!s described in this survey. A number o f figures have been redrafted from published data and in each such case, the original source is quoted. Figs 3 and 27, however, are direct reproductions by kind permission o f the copyright holders; Fig 3 by Dow Coming, Further reading Polymer-based materials Pratt, G. C. , 'Plastic-basedbearings', Lubrication and Lubricm-~ts, edited by E. R. Braithwaite, Elsevier. Amsterdam (1967) Lancaster° J. K., 'Friction and wear (of polymers)', Polymer Science: edited by A. D. Jenkins, North Holland Publishing Co, Amsterdam (1972) Carbons and graphites Badami D. V. and Wiggs, P. K. C. 'Friction and wear (of carbons and graphites)*, Modem Aspects of Graphite Technology, edited by L. C. F. Btaekman, Academic Press~ London (1970) Mild steel Metals - lametlar solid l u b r i c a n t s I Ceramics ptfe Solid lubricants Unfilled t h e r m o p l a s t i c s Reinforced t h e r m o s e t s + SOlid l u b r i c a n t s F i l l e d and reinforced t hermopIostk:s Metais-lam¢lfor solids-ptfe I- B o n d e d solid film ~bricants CampbeU, M. E., Loser, J: B. and Sneegas, E. "Solid Lubricants'. NASA SP 5059 (1966) Benzing, R. J. 'Solid lubricants', Modem Material~ Vol 5, edited by B. W. Gonser and H. H. Hausner, Academic Press, London (1964) 1 C a r b o n s - gcaphites k Cermets Filled p t f ¢ POROUS b r o n z e - P b - p t fe woven p t f ¢/91ass f i b r e Against t hemselves "I High temperature materials" I IO-8 t IO-7 I I I [O-6 IO-5 $O-4 Specific wear rate ( mmS/Nm ) I IO-S I0-2 Fig 33 Order of, magnitude values of wear rates for various groups of materials during sliding against steel at room temperature 250 TRIBOLOGY December 1973 Amatean, M. F. and Glaeser, W. A. 'Survey of materials for high temperature bea6:ng and sliding applications', Wear, Vol 7, (1964) p 385 Peterson, M. B. 'High temperature lubrication', Proceedings of the [ntemationN Symposium on Lubrication and Wear, edited by D. Muster and B. Sternlicht, McCutchan Publishing Corporation, Berkeley (1964) General 'Dry rubbing bearings - a guide to design and material selection', Engineering Science Data Item 68018. lnstn mech Engrs, London (1968) Lancaster, J. K. 'Composite self-lubricating bearing materials', Proc Instn mech Engrs, Vol 182, Part 1 (2) (1967168) p 33 Bisson, E. E. and Anderson, W. J. 'Advanced bearing technology', NASA SP (1964) 10 Finkin, E. F. 'A wear equation for bonded solid lubricant films: estimating film wear life', TransASME, JLub Tech, Vol 92 (1970) p 274 11 Kirkpatrick, D. L. and Young, W. C. 'Solid lubricants for instrument bearings', Proceedings of AFML-MR1 Conference on Solid Lubricants. Kansas City, Mo (1969). AFML-TR70-127 (1970) p 407 12 Harris, C. L. and Warwick, M. J. 'Lubrication of bearings and gears for operation in a space environment', Symposium on Lubrication in Hostile Environments, London, lnstn mech Engrs, Paper 6 (1969) 13 Flatley, T. W. 'High speed vacuum performance of miniature ball bearings lubricated with combinations of Ba, Au and Ag films', NASA TN-D-2304 (1964) 14 Boes, D. J. 'Long term operation and practical limitations of dry, self-lubricated bearings from 1 x 10 - 5 torr to atmospheric', Lubrication Engineering, Vol 19 (1963) p 137 15 Mecklenburg, K. R. 'Selection of bearings for lubrication research', Proceedings AFML-MR1 Conference on Solid Lubricants, Kansas City, Mo (1969). AFML-TR-70-127, 451 (1970 Devine, M. J., Cerini, J. P. and Stallings, L. 'Improving frictional behaviour with solid film lubricants', Metals Eng Quarterly, Vol 7(2) (1967) p 33 References l Jaeger, J. C. 'Moving sources of heat and the temperature at sliding contacts', Proc R Soc, NSW, Vol 76 (1942) p 203 2 Lancaster, J. K. 'Estimation of the limiting PV relationships for thermoplastic bearing materials', Tribology, Vol 4 (1971) p 82 3 Willis, D. P., and O'Rourke, J. T. 'Evaluation of selflubricating bearing materials made from phenolic moulding compounds', Plastics design and processing (April 1964) p 14 4 'Polyimide laminate resins are setting a hot pace', The Engineer (19 March 1970) p 41 5 Pratt, G. C. 'Plastic-based bearings', Lubrication and Lubricants, edited by E. R. Braithwaite, Elsevier, Amsterdam (1967) 6 16 17 'Dry rubbing bearings - a guide to design and materials selection', Engineering Sciences Data Item 68018. Inst mech Engrs, London (1968) Boes, D. J., Cunningham, J. S. and Chasman, M. R. 'The solid lubrication of ball bearings under high speed - high load conditions from - 2 2 5 ° F to +1000°F ', Proceedings of AFML-MRI Conference on Solid Lubricants, Kansas City, Mo (1969). AFML-TR-70-127 (1970) p 257 18 7 Midgley, J. W., and Teer, G. D. 'An investigation of the mechanism of the friction and wear of carbon', ASME Paper 62-Lub-15 (1962) van Wyk, J. 'MoS2 solid lubricant composites', Proceedings of AFML-MR1 conference on Solid Lubricants, Kansas City, Mo (1969). AFML-TR-70-127 (1970) p 290 19 8 Peterson, M. B., and Finkin, E. F. 'Application of new and improved solid lubricant materials and processes to naval aircraft', MTI Report 71 TR 48 (1971) Sibley, L. A. and Allen, C. M. 'Friction and wear behaviour of refractory materials at high sliding velocities and temperatures', Wear, Vol 5 (1962)p 312 20 'UCAR - metal and ceramic coatings: mating materials', Union Carbide Coatings Service Booklet 9 McCain, J. W. 'A theory and tester measurement correlation about MoS 2 dry film lubricant wear', SAMPE Journal (Feb/ Mar 1970) p 17 21 Glaeser, W. A. 'High temperature bearing materials', Metals EngQuarterly, Vol 7 (2) (1967)p 53 TRIBOLOGY December 1973 251