Mechanisms for transitional friction and wear behavior of sliding metals

Mechanisms for transitional friction and wear behavior of sliding metals

8 Elsevier Sequoia &A., Lawanne - Printed in The Netherlands MECHANISMS FOR TRANSITIONAL BEHAVIOR OF SLIDING METALS PETER FRICTION AND WEAR J. B...

1MB Sizes 0 Downloads 100 Views

8 Elsevier Sequoia &A., Lawanne

-

Printed in The Netherlands

MECHANISMS FOR TRANSITIONAL BEHAVIOR OF SLIDING METALS

PETER

FRICTION AND WEAR

J. BLAU

Metallurgy (Received

Division,

National

Bureau of Standards,

Washington,

DC 20234 (U.S.A.)

Fefwuary 19,198l)

Four processes which may cause transitions in the unlubricated sliding behavior of metals are described. These processes are (1) metal transfer, (2) film formation and removal, (3) debris generation and (4) cyclic surface deterioration, Although these are not the only processes which may cause unexpected sudden friction and wear transitions, they may contribute significantly to many frequently observed sliding phenomena. It is possible that some combination of these mechanisms may act to effect changes in the wear and friction behavior of a given tribologieal system. The cyclic dete~o~tion model involves four sequential stages in each cycle: (1) plastic deformation to &t&n a highly deformed layer (HDL); (2) debris generation by crack initiation and growth in the HDL to produce flake-like debris; (3).debris removal through the shearing-off of load-bearing plateaus which slide along on their fracture surfaces to produce deep grooves; (4) plateau removal followed by a smoothing of the deep wear grooves (a process similar to the break-in of a rough surface with changes in the friction and the rate of wear). This process, when stages are occurring simultaneously on a large portion of the wear surface, could be observed as a cyclic variation in the friction and/or the wear rate.

1. Introduction The determination of the moment when a sliding contact has reached a “steady state” continues to be elusive because quite often sudden transitions in the friction force and/or the wear rate are observed without any apparent change in the externally imposed sliding conditions. Traditionally, friction and wear rate transitions are generally thought of in terms of the early break-in or run-in stages of sliding in which the system may be approaching some steady state running condition or, in other cases, the sudden sliding behavior transitions which may signal the end of the useful service life of a component. However, there is a third case in which the friction force andfor wear rates may start to fluctuate periodically; in this case the system never

56

reaches a steady state sliding condition. This type of behavior has often been observed in unlubricated metal sliding; this is reviewed in subsequent sections.

2. Background In the technical literature on the subject of friction and wear, the term “transitions” usually refers to changes in sliding behavior which are observed when one or more external variables are purposely changed by the investigators. For example, Whitehead [l] dealt with changes in the sliding friction of silver, aluminum and copper with increases in the test load, and Ling and Werner [2] examined self-mated lead under similar conditions. Rabinowicz [ 31 discussed decreases in the friction of copper at increasing temperatures and Rozeanu [4] related thermally induced frictional transients to the “frictional failures” of high velocity sliding systems. In 1971, Eyre and Maynard [ 51 published an interesting study of mild-to-severe-to-mild wear transitions as functions of sliding velocity for gray flake cast iron. The sliding velocity was also the main variable in the work of Nakajima and Mitzutani [6] ; they presented data showing maxima in wear rates for self-mated low and medium carbon steels at low sliding velocities. Begelinger and de Gee [ 71 showed transitions to lower wear rates for silver on electroplated iron as the oxygen content in Ar-Oz mixtures was decreased. In the foregoing examples, the experimenters induced the changes in the test behavior by altering the external sliding conditions. The question which is to be considered further here is that of the way in which various tribological systems can and do effect internal alterations in the sliding contact zone so as to induce transitions in friction and wear behavior. If the mechanisms which produce these transitions can be identified unambiguously, the feasibility of suppressing or eliminating undesirable (possibly catastrophic) “surprises” in the operating conditions of sliding components could be examined more clearly. Even if the transitions can be delayed or extended in duration through improved knowledge of the dominant mechanisms, warning signs of impending component failure may be more likely to be detected so that the potential for serious machine failure can be reduced.

3. Internal

transitions

in friction

and wear

Although it is quite probable that a considerable number of processes may exist in metal sliding situations, the following four specific types are addressed here: (1) metal transfer; (2) film formation and removal; (3) debris generation; (4) cyclic surface deterioration. The first three processes have been well documented in the literature (they are discussed below); however, the fourth process is a model which is proposed in order to encompass some of the more recent concepts in surface deformation, crack nucleation and crack growth. The applicability of these processes to the induction of sudden

transitions in friction or wear rate depends greatly on the sliding materials, the contact geometry, the thermal and chemical environment of the contact and the mechanical parameters of the system. 3.1. Metal transfer The role of metal transfer between sliding surfaces has been discussed in numerous papers (e.g. refs. 7 - 14). This role involves the changes in the sliding characteristics which occur when material from the bearing surface of one member of a sliding couple is transferred to the bearing surface of the other member. Thus before transfer we may observe the sliding behavior of a pair of diss~il~ materials, and after transfer self-mated sliding behavior may be observed. Beyond this somewhat simplistic situation, there are several variations in the transfer which may be more representative of the complexity of the process in many sliding systems. One variation involves retransfer of material back to the surface from which it came [ 151. During this process, the material may become mechanically mixed with material of the other surface, as demonstrated for Cu-Zn couples by Sasada et al. [ 151. Another variation involves the mixture of transferred material with oxides (e.g. ref. 16). Such a metal-oxide mixture is shown in Fig. 1 [ 171. This kind of “marble cake” structure has been observed frequently in polished sections of both ferrous and non-ferrous wear surfaces. The transfer process can affect wear and friction transient behavior in several ways, including the following: (a) by changing the bearing area of the sliding contact (e.g. the sudden loss of a transferred deposit resulting in a rapid change in the bearing pressure); (b) by changing the relative hardness, ductility and composition of the materials in the sliding interface; (c) by covering sharp hard asperities on one surface with a more ductile deposit from the opposite surface, Often the nature of the transferred deposit may be the factor which determines whether friction and/or wear transitions occur suddenly or over a

Fig. 1. Cross section of a wear track in Cu-7wt.%Al used in a block-on-ring wear test; dark deposits of the oxidemetal mixture can be seen (temperature, 400 “C).

period of time (i.e. sliding cycles). If the transfer takes place by the gradual build-up of a continuous film of one metal on the mating surface, a less abrupt transition might be expected than if a large clump of debris material became attached to the mating surface during the course of a few sliding cycles. In order to correlate changes in wear and friction with transfer, not only post-test but also in situ examination of one or both sliding members is desirable. If the facilities for this kind of in situ examination are lacking, indirect evidence based on “negative wear” (i.e. surface build-up) indications from specimen position sensors may demonstrate the presence of transfer. Transfer may invalidate assessments of wear test data which rely only on sample weight changes to determine wear rates. 3.2. Film formation and removal The formation of oxide films which alter the sliding behavior with time has been the subject of numerous reviews (e.g. refs. 3,18 and 19) and will not be discussed further here except to point out that both the chemical reactivity of sliding surfaces and the kinetics of oxide film formation should be evaluated before ascribing an oxidative mechanism to friction and wear transients. For example, a high sliding rate may preclude the formation between subsequent passes of sufficient oxide to affect the sliding behavior. Another related factor to be considered, however, is the higher oxidation rates reported for surfaces (e.g. steel [ 161) which were deformed by sliding. In general, we might expect that transitions due to surface oxidation would be much less sudden than those from sources such as the loss of lumps of transferred debris out of the sliding contact zone. Another type of film has received less attention in the literature, namely surface films which have chemical compositions different from the bulk sliding material and which may have formed by solid state diffusion of one or more alloying elements in the material near the sliding interface. An example of this type of film as related to periodic changes in friction of Cu-Al was discussed by Amsellem and Caubet [20] . The presence of a copper-colored film was correlated by these workers with lower wear; they recommended that such a film should be created during run-in to give an improved wear resistance. An example of such a copper film on a Cu-3.2wt.%Al pin-on-disk wear track (with a fixed steel 52100 ball bearing rider) is shown in Fig. 2 [ 171 . In this instance periodic positive disturbances in the friction traces similar to those seen by Amsellem and Caubet [20] were observed. It seems likely that near-surface diffusion and possibly selective solute oxidation may play a part in forming surface films during sliding. Darkening and further discoloration of the copper-colored films on golden Cu-Al debris flakes were observed during prolonged examinations with the optical microscope. Surface segregation of aluminum and tin in copper alloys has received considerable study [21], and in fact this segregation was identified as a contributor to the wear mechanisms of Cu-Al alloys in aircraft fuel systems [22] . Taga et al. [23] have correlated both friction and wear transitions

59

Fig. 2. Copper film formed on the wear track of a Cu-3.2wt.%Al sliding by a steel 52100 ball bearing rider (load, 2.45 N).

disk specimen owing to

with surface chemistry changes in Cu-Sn pins sliding on stainless steel disks under atmospheric conditions. The role of fihn formation in wear and friction clearly requires more detailed study, particularly with respect to the following: (a) the composition and mechanical properties of thin surface films; (b) the abrasiveness of debris from the wear of surface films; (c) the oxidation and chemical reactivity of surface films; (d) the possible role of surface films in promoting or reducing material transfer in specific sliding systems. 3.3. Debris generation Two types of debris are produced during metal wear: these are (1) passive debris which is removed from the contact zone immediately after it leaves the site of its origin and (2) active debris which participates further in the wear process as an additional inter-facial abrasive (or lubricant [24] ). Passive debris is generally composed of relatively large flat particles which may contain features of the original sliding surface on one face, while active debris may be much finer powdery material intermixed with oxides and other interface constituents (e.g. ref. 25). During sliding, the type and amount of debris may change causing drastic changes in both friction and wear. This phenomenon was exemplified in the Cu-lOwt.%Zn test data shown in Fig. 3 [17]. For the first 15 min of sliding, no debris was observed beneath the pin-on-disk contact. The friction remained at about 0.35. Suddenly, two or three sharp friction peaks were observed and the average value and the amplitude of the friction trace rose rapidly. Metallic flake debris was observed immediately. Friction remained at the higher level (about 1.1) for the rest of the test (about 1 h). The conversion of a sliding situation from a two-body wear situation to a three-body (abrasive) wear situation was the major cause of this transition. Numerous examples exist of sudden catastrophic changes in wear rates due to debris generation in the use of precision ball bearings. Close tolerances of mating components do not permit the introduction of even relatively fine

60

I

t--

I4

I

-4

i

16 1440 cycles

(minf

Fig. 3. A sudden transition in the friction accompanied by the onset of an audible grinding noise and debris generation (material, Cu-lOwt.%Zn; load, 250 gf; sliding velocity, 5 cm s-l ; atmosphere, argon).

particles; if such particles are introduced, the wear of the component (e.g. the baB, surface or bearing race) is seriously affected. In such cases, the near-surface fatigue properties of the race and bearings largely determine the incubation period before wear and friction transitions occur. The me~h~ism of fatigue crack incubation periods has been discussed in relation to wear transitions in copper alloys [ 261. Evidence which supports a faction-reduces role of debris in other types of wear situations also exists, Hahiday and Hirst [ 243 demonstrated how the removal of debris from a fretting steel contact increased friction. This was explained in part in terms of the action of debris fragments, as tiny roller bearings, in reducing metallic contact. Therefore the conditions of sliding must be examined carefully before either a positive or a negative influence on transitions from the onset of debris generation can be ascribed. A model for the surface deterioration by cyclic wear of single-phase metals was proposed in a previous investigation by the present worker [ 1’71 I Each fuli cycle involves four sequential stages which repeat periodically. The model was based both on microst~c~~l observations of wear tracks from Blau’s [ 1’71 work (Appendix A) and on examinations of numerous published photomicrographs of the wear surfaces of unlubricated metals. The mechanisms responsible for the observed features are based on two of the more recent wear models; these are. (a} the concepts of Rigney and Hirth [ 271 on the role of the highly deformed layer (HDL) in friction and wear and (b) certain aspects of the de&&nation theory for wear proposed by Sub [ZS] , In addition, the rn~h~~rn seems to explain the origin of certain surface and subsurface micro~ctu~l features observed in severely worn metals; these features are described below.

61

The model basically involves four stages in each repeating cycle. (1) The first stage consists of plastic deformation (including microgeometric conformation of opposing surfaces), crystallographic texturing and generation of a near-surface HDL whose average thickness and structure is determined by the imposed shear stress level and the material properties of the sliders. (2) The second stage is debris generation. This occurs after the subsurface can no longer sustain additional work hardening. Cracks initiate and propagate in the textured HDL so as to form flake-like debris. Cracks may initiate at various sites in the HDL such as dislocation cell walls, deformation twin boundaries or other localized crystal structure discontinuities. (3) Debris removal is the third stage. The highest remaining plateaus of the original surface must bear more and more load. These plateaus shear off and slide away on their fracture surfaces producing deep grooves below the level of the tops of the remaining plateaus. (4) The fourth stage is plateau removal. The last plateaus are rapidly fractured and the mating surface is lowered onto the deeply grooved surface where some smoothing (similar to a break-in period) occurs before the cycle repeats. Figure 4 illustrates schematically the features of the model. At any given time, various portions of the wear surface may be experiencing different stages in the process, and it is only in the specific case when the majority of the sliding surface is simultaneously progressing through the four-stage cyclic deterioration sequence that periodic friction and wear rate changes may be most apparent in external measurements. The chronological progress of cyclic surface deterioration might also be applied to a sliding contact whose nominal contact area increases with time. For example, in a rounded pin-on-disk contact, the center of the wear track may be at a more advanced stage than the track edges. This effect would be further compounded because the hertzian stress under the spherical pin is at

Fig. 4. Illustrations of the four proposed stages of surface deterioration; these are discussed in the text: HDL, highly deformed layer; P, plateau; S, sliding direction.

62

a maximum at the center of the track. The center of the track may thus be deeply grooved with few remaining plateaus, while nearer the sides the bearing surface may show less deterioration. Figure 5 shows how the center of a pin-on-flat track on copper is much more worn than the edges of the track. In fact, rather continuous ridges of smoother bearing surface material are parallel to the central worn region. Microstructural evidence for the various stages in the model is provided in Figs. 6 - 8. In addition, the taper section in Fig. 9 shows how the tops of some plateaus on the wear track of a Cu-7wt.%Al disk conform to the spherical rider shape. In view of the microstructural evidence from examinations of severely worn metal surfaces, there appears to be considerable support for the cyclic surface deterioration model in the context of sliding wear and friction transitions. Further work, ideally involving in situ observations of wear surfaces correlated with friction and wear measurements, must be undertaken if the range of applicability of this model is to be established. The model must be placed in the proper perspective with regard to other processes which could also induce sliding behavior transitions in metallic systems. Finally, it must be recognized that many of the types of transitions which occur in sliding behavior without upparen t changes in the externally imposed parameters may be closely related to the results of previous studies of transitions in which the investigators purposely changed the sliding condi-

(a)

(b)

Fig. 5. (a) Copper wear track and (b) tapered section through the wear track of Fig. 5(a) showing the severe wear at the center of the track and the milder wear at the edges of the track: a, edges of the track; b, a raised lip of smeared appearance which flanks the central portion of the track; S, sliding direction.

Fig. 6. An example of stage 1 deterioration in the wear track of a Cu-lOwt.%Zn (ioad, 9.98 N; sliding velocity, 5 cm s-l).

disk

Fig. 7. An example of stage 2 deterioration in the wear track of a Cu-lDwt.%Zn (load, 2.49 N; sliding velocity, 5 cm s-l).

disk

Fig, 3. An example of stage 3 deterioration in the wear track of a ~u-~~~.%Zn disk (load, 2.49 N; sliding velocity, 5 cm s -I ); the deep grooves G lie below the ‘level of the remaining plateaus P. Fig. 9. Tapered section of a wear track in Gu-7wt.%Al showing plateaus P which support the spherical rider tip of the steel pin, There is deep grooving below the tops of the plateaus and, underneath the grooving, there is extensive rn~&rost~~tur~ damage.

tions, To clarify this point, let us consider the previously cited cases (refs. 1 and 2) in which friction underwent a transition over a range of test loads. The more useful variable would seem to be, rather than the load, the load per unit bearing area (i.e. the contact stress). Therefore, as the bearing area of a system changes during the progress of wear by any one of many processes, the contact stress level of the system is changing internally and the system may therefore pass from one short-term steady state to another and, perhaps, back again. In this way, previous studies which involved variables such as the test load may be directly related to the unde~~d~~ of transitions under conditions which might initially seem to be constant sliding con-

64

ditions. The instantaneous velocity at various locations on the sliding surface may not always equal the nominal sliding velocity owing to accelerations and decelerations of portions of debris patches on the surface; thus velocity effects on sliding processes may alter the dominant wear and friction mechanisms at particular regions of the contact zone.

Acknowledgments The present worker wishes to acknowledge helpful discussions with A. W. Ruff and L. K. Ives of the National Bureau of Standards. Some of his work referred to in the present paper was supported by grants from the Office of Naval Research and the National Science Foundation.

References 1 J. R. Whitehead, Surface deformation and friction of metals at light loads, Proc. R. Sec. London, Ser. A, 201 (1950) 109 - 124. 2 F. F. Ling and R. S. Weiner, A bifurcation phenomenon of static friction, J. Appl. Mech., 28 (1961) 213 - 217. 3 E. Rabinowicz, Fiction and Wear ofMaterials, Wiley, New York, 1965. 4 L. Rozeanu, Friction transients, ASLE Trans., 19 (4) (1976) 257 - 266. 5 T. S. Eyre and D. Maynard, Surface aspects of unlubricated metal-to-metal wear, Wear, I8 (1971) 301 - 310. 6 K. Nakajima and Y. Mitzutani, Structural change of the surface layer of low carbon steels due to abrading, Wear, 13 (1969) 283 - 292. 7 A. Begelinger and A. W. J. de Gee, Sliding characteristics of silver against iron as influenced by oxygen concentration, ASLE Trans., 10 (1967) 124 - 133. 8 I.-M. Feng, Metal transfer and wear, J. Appl. Phys., 23 (9) (1952) 1011. 9 M. Cocks, Interaction of sliding metal surfaces, J. Appl. Phys., 33 (7) (1962) 2152 2161. 10 M. Antler, Processes of metal transfer and wear, Wear, 7 (1964) 181 - 204. 11 A. W. J. de Gee, Friction and wear as related to the composition, structure, and properties of metals, Znt. Metall. Rev., 24 (2) (1979) 57 - 67. 12 J. T. Burwell and C. D. Strang, On the empirical law of adhesive wear, J. Appi. Phys., 23 (1) (1952) 18 - 28. 13 T. Shirakashi, R. Komanduri and M. C. Shaw, On friction and metal transfer of sliding surfaces undergoing subsurface plastic deformation, Wear, 48 (1978) 191 - 199. 14 T. Sasada, S. Norose and J. S. Thian, Birth and growth process of wear particles observed through relative traversal movement of the rubbing surfaces, Proc. 18th Jpn. Conf. on Materials Research, 1975, Japan Society of Mechanical Engineers, pp. 72 - 76. 15 T, Sasada, S. Norose and H. Mishina, The behavior of adhered fragments interposed between sliding surfaces and the formation process of wear particles, Proc. Znt. Conf. on Wear of Materials, Dearborn, MZ, 19 79, American Society of Mechanical Engineers, New York, 1979, pp. 72 - 80. 16 H. Krause, Tribochemical reactions in the friction and wearing processes of iron, Wear, 18 (1971) 403 - 412. 17 P. J. BIau, A study of the interrelationships among wear, friction and microstructure in the unlubricated sliding of copper and several single-phase binary copper alloys, Ph.D. Dissertation, Ohio State University, 1979.

65 18 I. V. Kragelskii, Friction and Wear, Butterworth, London, 1965. 19 T. F. J. Quinn, J. L. Sullivan and D. M. Rowson, New developments in the oxidational theory of the mild wear of metals, Proc. Znt. Conf. on Wear of Materials, Dearborn, MZ, 1979, American Society of Mechanical Engineers, New York, 1979, pp. 1 - 11. 20 C. Amsallem and J. J. Caubet, Quelques observations sur le comportement en frottement du cupro-aluminum, Wear, 12 (1968) 257 - 276. 21 J. Ferrante and D. H. Buckley, A review of surface segregation, adhesion and friction studies performed on copper-aluminum, copper-tin and iron-aluminum alloys, ASLE Trans., 15 (1) (1972) 18 - 24. 22 W. Poole and J. L. Sullivan, The role of aluminum segregation in the wear of aluminum/ bronze-steel interfaces under conditions of boundary lubrication, ASLE 34th Annu. Meet., St. Louis, MO, 1979. 23 Y. Taga, A. Isogai and N. Nakajima, The role of alloying elements in the friction and wear of copper alloys, Wear, 44 (1977) 377 - 391. 24 J. S. Halliday and W. Hirst, The fretting corrosion of mild steel, Proc. R. Sot. London, Ser. A, 236 (1956) 411 - 425. 25 A. W. Ruff and P. J. Blau, Studies of microscopic aspects of wear processes in metals, NBS Znt. Rep. 80-2058,198O (National Bureau of Standards). 26 G. H. G. Vaessen and A. W. J. de Gee, Influence of water vapour on the wear of lightly loaded contacts, Wear, 18 (1971) 325 - 332. 27 D. A. Rigney and J. P. Hirth, Plastic deformation and the sliding friction of metals, Wear, 53 (1979) 345 - 370. 28 N. P. Suh, An overview of the delamination theory of wear, Wear, 44 (1977) 1 - 16.

Appendix A The wear and friction

test procedure

In this Appendix, which is based on the work of Blau [Al], more detailed information is provided about the conditions used to produce the microstructures in Figs. 6 - 9. The geometry was that of a horizontally loaded pin (a steel 52100 ball bearing 0.635 cm in diameter fixed on the end of a brass rider stub) on a flat disk whose contact surface was oriented vertically. In this way, loose debris would fall from the contact region and not remain on the wear track as it might if the disk surface were horizontal with a vertical rider pin on top. Not only did this contact orientation provide a means of reducing the contribution of debris to the sliding wear process but it also facilitated the collection of passive debris which had not been further deformed by repeated passage through the sliding interface. Observations which led to the formation of the deterioration model were made on a series of disks consisting of the following copper alloys: high purity copper; Cu-3.2wt.%Al; Cu-7.5wt.%Al; Cu-lOwt.%Zn; Cu-30wt.%Zn; Cu-5wt.%Sn0.3wt.%P. The disks were polished metallographically and then tested in flowing argon at a sliding velocity of 5 cm s-l and under loads of 0.98 - 3.9 N. The test length was typically a sliding distance of 200 m. Observations of wear track microstructures (e.g. Figs. 6 - 9) were made at the conclusion of each test run, and it was common to find examples of several stages of deterioration on different portions of the same wear track. This latter observation led to the conclusion that, in the absence of other dominant transition processes (such as those mentioned in the text of this paper), cyclic surface

66

deterioration effects on friction and wear transitions would only be observable when a major portion of the contact surface was passing simultaneously through the same stages of the process. Reference Al

for Appendix

A

P. J. Blau, A study of the interrelationships among wear, friction and microstructure in the unlubricated sliding of copper and several single-phase copper alloys, Ph.D. Dissertation, Ohio State University, 1979.