Wear, 41 (1977) 351 - 363 0 Elsevier Sequoia S.A., Lausanne
351 -
Printed
in the Netherlands
RECRYSTALLIZATION AS A CONTROLLING WEAR OF SOME F.C.C. METALS
ROBERT
C. BILL
and DONALD
May 28,
IN THE
WISANDER
National Aeronautics and Space Administration, Cleveland, Ohio 44135 (U.S.A.) (Received
PROCESS
Seals Section,
Lewis Research
Center,
1976)
Summary Detailed examination of copper specimens after sliding against 440 C steel in liquid methane at speeds up to 25 m s-l and loads of up to 2 kg showed the metal comprising the wear surface to possess a fine cell recrystallized structure. Wear proceeded by the plastic shearing of metal in this near surface region without the occurrence of visible metal transfer. A dynamic balance between the intense shear process at the surface and the nucleation of recrystallized grains was proposed to account for the behavior of the metal at the wear surface. Sliding wear experiments were also conducted on Ag, Cu-10% Al, Cu-10% Sn, Ni and Al. The results were correlated with published hot-working observations and recrystallization kinetics. It was found that low wear and the absence of heavy metal transfer were associated with those metals observed to undergo recrystallization nucleation without prior recovery.
Introduction It has been demonstrated that under certain circumstances the wear of metals in sliding contact occurs by the mechanism of plastic flow of the surface and near subsurface regions. The details of such mechanisms have been the subject of numerous studies. Antler [ 1 - 31 and Cocks [ 4, 51 described a “wedge flow” mechanism applicable to sliding combinations of soft metal pairs. In an investigation of experimental parameters affecting adhesive metal transfer, Landheer and Zaat [6] pointed out that the wedge flow mechanism might be suppressed when a soft metal is slid against a hard flat metal surface. This is in agreement with earlier observations for the case of relatively soft copper in sliding contact with a 440 C steel dish [ 71. Wear was observed to occur by the incremental shearing of copper in a series of plastic- “waves” across the wear surface. A recent contribution is the delamination theory of wear developed by Suh et al. [8 - lo], which is in essence a wear model based on dislocation
352
_-Dome
Friction
dynamometer
specimen J Section A-A
k
Liquid level I ./-Test
Fig. 1. Cryogenic
fuel friction
apparatus
with specimen
chamber
loading system.
dynamics and which accounts for wear phenomena observed in low energy sliding situations. Wear is envisioned to occur by the coalescence of voids at a critical distance below the surface. This is the distance over which dislocation image forces caused by the proximity of a free surface overcome the internal dislocation frictional forces. In this paper wear phenomena observed during relatively high energy sliding situations in which copper and other face-centered cubic metals are slid against 440 C steel are described. The processes that control wear are identified and discussed in the light of microstructural details observed on the sliding surface and near subsurface regions of the copper. A generalized theory is developed and discussed with respect to the results obtained here and to observations made by others. Wear experiments were conducted in a liquid methane environment at sliding velocities up to 25 m s-r, in support of a cryogenic fuel pump vane wear study.
353
.6
.I
,2
.4
.6 .8
1
2
4
A\ 1
2
Normal load, L, kg
46810 20 Slidi& velocity, V. mlsec
40
6oEillM
Fig. 2. Wear rate of copper as a function of normal load. Fig. 3. Wear rate of copper as a function of sliding velocity.
Apparatus and procedure
The apparatus used is shown in Fig. 1 and is described elsewhere [ II]. The basic elements consisted of a hemispherically tipped rider specimen of radius 4.76 mm, composed of 99.95% purity copper (Rz = 85), held in sliding contact with the lower flat surface of a rotating disk of diameter C3.5 mm, composed of 440 C steel (Rc = 58). The experiments were conducted with the specimens completely submerged in liquified methane. The sliding speed was varied from 3.1 to 25 m s-r (1000 - 8500 rev min- ’ ). The rider was loaded against the disk by a helium-pressurized bellows assembly. Frictional force and normal load were measured by strain-gage dynamometer rings. Wear was continuously measured by a linear voltage differential transformer (LVDT). Load, wear and friction coefficient were continuously recorded during a sliding experiment. The test chamber was cleaned with 90% ethanol before each run. After cleaning and the installation of the specimens, the test chamber was closed, purged for 15 min with helium gas and then filled with liquified methane. After the test chamber was full and the liquid boiling had stabilized, the rider was loaded against the disk and the disk was rotated at the desired speed. The duration of the runs was ‘/z h. The disk specimen surfaces were finish ground and lapped to 5 X 10m2 pm r.m.a, scrubbed with moist levigated alumina, washed in tap water, washed in distilled water and air-dried.
354
Fig. 4. Wear scar surface details seen on copper after sliding against 440 C at 12.4 m s-“ under a % kg normal load in liquid methane.
Transmission electron microscopy specimens were prepared from a 99.95% copper rider in the following manner. After being subjected to sliding wear, an axial circular cut circumscribing the wear scar on the nose of the rider was electron-discharge machined to a depth of about 6 mm below the wear scar surface. The nose of the rider was then cut off 6 mm below the wear surface, and the cylinder generated by the circular cut of an electron discharge machine (EDM) was extracted. Slices about 200 pm thick were then cut by EDM from the cylinder. The first slice thus cut had the wear scar as one of its surfaces. The disk-shaped slices were then electrolytically thinned to about 500 a and prepared for transmission electron microscope studies. Comparison of the micro~aphs from different depths revealed no obvious artifacts due to the EDM process. In the case of the first slice, with the wear
355
Fig. 5. Section of a layered stack of wear material attached to the traiting edge of a wear scar. The “turbufent” material on the left was formed from copper that comprised the bond between and the trailing edge of the wear scar.
scar as one of its surfaces, thinning was carried out from the inner surface and the oxide film was removed from the wear scar surface. In this manner, a transmission electron microscopy specimen was obtained that was as close as practicable to the wear surface itself. The rate of wear of the copper rider was calculated from the continuously monitored and plotted LVDT measurements. The wear rate at any instant is given by
where 1 is the change in length of the rider due to wear of the hemispherical tip. Since the value of r is constant, the instantaneous wear rate may be calculated by measuring the slope of the I uersus time curve and the value of 1 at the instant in question. In all cases, the wear rates were calculated after 65 pm of wear (i.e. 1 = 65 pm) had occurred to the rider. Results Wear 0 f copper
Figures 2 and 3 show the wear rate of copper as a function of load and sliding velocity, respectively. The wear rate varies approximately as the
(a)
Fig. 6. Metallographic sections cut paraliel to the sliding direction showing a recrystallized layer on the copper wear surface after sliding against 440 C steel in liquid methane, under a load of 1 kg at 12.4 m s-l: (a) entrance region, (b) exit region.
velocity squared and is roughly proportional to the normal load. No measurable wear to the 440 C disk occurred, the only evidence of sliding being a very slight discoloration of the disk wear track surface. The wear surfaces on the copper riders were studied using scanning electron microscopy. It was observed that wear proceeded by the transportation of thin layers of copper across the wear surface to form a layered build-
357
up on the exit side. Figure 4 shows an overview of the trailing edge of such a wear surface, and a micrograph of a section of the layered buildup is shown in Fig. 5. It is estimated that each layer is of the order of 1 pm in thickness. The material comprising the layered buildup attached to one of the high purity copper specimens was carefully removed and weighed. It was found that the weight of this material very nearly matched the theoretical weight of the material originally making up the wear volume. From this it was concluded that the mechanism causing the layered formations was the dominant wear mechanism, with the effects of metal transfer and any conceivable oxidative wear that might have occurred in the liquid methane env~onment being secondary. Selected high purity copper specimens were sectioned, polished and etched to reveal the grain structure beneath the wear scar surface after % h of sliding against 440 C steel. Typical micrographs are shown in Fig. 6, The material beneath the wear scar surface is composed of two distinct layers. The upper layer appears to be very fine-grained recrystallized copper. The depth of this layer varies uniformly from zero at the entrance edge to a maximum thickness near the exit edge of the wear scars. The thicknesses of the fine-grained structure at the midpoint for the three sliding conditions examined are summarized in Table 1. Directly beneath the fine-structured layer is a region of severely cold-worked copper.
TABLE 1 Summary of recrystallized layer thickness Normal load (kg)
Sliding speed (m s-1)
Thickness of layer (Pm)
2
12.4 12.4 6.2
12 10 3
1 %
Figures 7 and 8 are tr~smi~ion electron micrographs made from thin film specimens of copper showing microstructural features at various depths below the wear surface. The microstructure of the region nearest the wear surface is revealed in Fig. 7. Here a very fine cell structure may be seen, the average cell diameter being 0.3 - 0.4 pm. The cell walls are sharply defined with the individual dislocations that form the walls being clearly discernible. The interiors of these cells are nearly dislocation free. The microstructure of the copper at a depth of 200 pm below the wear surface is shown in Fig. 8. In contrast to Fig. 7, extremely dense dislocation entanglements may be seen and some evidence of ill-defined cell walls is apparent. These features are typical of heavily cold-worked copper.
Fig. 7. Microstructure of copper adjacent to the wear scar surface 440 C steel in liquid methane under a 1 kg load at 6.2 m s-l.
Wear of other f.c.c.
after sliding against
metals
Final wear measurements made after ‘/z h of sliding against 440 C steel under a 1 kg load at 12.4 m s-’ are summarized in Table 2 for Cu-lO%Al, Cu-lO%Sn, 99.9%Ag, 99.9%Ni, high purity aluminum and 2024-T4 aluminum alloy, in addition to high purity copper. Macroscopic examination showed that layered wear formations similar to those seen on copper were present on the silver, Cu-lO%Al and Cu-10% Sn. Also the wear surface of the metals showed the same polished appearance as was seen on copper. The nickel, aluminum and aluminum alloy, however, exhibited rough heavily disrupted wear surfaces with heavy plow marks caused by metal that had transferred to the 440 C steel disk. The transferred material seen on the 440 C steel disk after sliding against nickel and aluminum is shown in Figs. 9(a) and (b) respectively. For comparison purposes the wear track on 440 C steel after sliding against copper is shown in Fig. 9(c).
359
Fig. 8. Electron micrograph methane. The microstructure scar surface.
TABLE
of copper subjected to sliding against 440 C steel in liquid is representative of material from 200 Mm below the wear
2
Summary
of wear results
Metal
Wear volume after % h Stacking fault energy* (erg em-*) (cm3 X 10s)
& cu Cu--lO%Sn Cu-- 1 O%Al Ni
250 900 40 1500 30000
150
Al
10000
200
9000
2024T4
*From
ref. 12.
25 70
unknown
Condition of wear track on 440 C disk Light discoloration Light discoloration Light discoloration Very heavy galling of Ni to 440 c Heavy galling of Al to 440 C Heavy galling of alloy to 440 c
360
(a)
(b)
Fig. 9. The wear tracks on 440 C steel disk surfaces after sliding against riders of various materials at 12.4 m s-l under a load of 1 kg in liquid methane: (a) nickel rider, (b) aluminum rider and (c) copper rider (magnification 16x ).
Discussion The microscope studies performed on the copper show that wear has taken place by the plastic shear of a fine-grained recrys~lized surface layer. The sheetlike morphology of the copper wear material is suggestive of a
361
delamination-type of mechanism, as proposed by Suh, but none of the other ingredients of the delamination model are present. The structure of the surface layer as revealed in Fig. 7 is far from being dislocation free. The fine cell structure is indicative of initial recrystallization proceeding at a very high nucleation rate. The observation that the thickness of the sheets that comprise the wear material is much less than the overall recrystallized layer thickness means that the wear-controlling processes take place entirely within this layer. Furthermore the SEM photographs indicate that, rather than being continuous, wear of the copper proceeded by a sequence of discrete events. The load and speed dependence of the wear of copper may be explained in terms of such events. When one of the wear events takes place, on a microscopic scale it is envisioned that the intense local plastic deformation in the presence of frictional surface heating leads to a dynamic balance between plastic strain and the formation of recrystallization nuclei throughout 100% of the affected volume of material. The balance between the processes of deformation and the nucleation of new crystallites permits very extensive plastic deformation to take place in the surface layers without the occurrence of strain-hardening effects and reduced ductility. Thus a deformation wear mechanism may proceed without the incidence of fracture and the associated generation of transferred debris. In this situation the adhesion between copper and the 440 C steel disk results in a very thin uniform transfer film indicated by the slightly discolored wear track shown in Fig. 9(c). Large particles of transferred copper are absent. With the above mechanism as a basis, it is not difficult to explain the observed relationship
in which l@ is the wear rate, K a proportionality constant, L the normal load and V the sliding velocity. Regardless of the condition of the surface layer (provided it does not grow to too great a thickness) the actual contact area and hence the number of events taking place at any instant is directly proportional to the normal load. It was shown earlier [7] that the volume of material sheared during the lifespan of such a wear event is expected to be proportional to the thickness of the recrystallized “superplastic” layer. Inspection of Table 1 reveals that the thickness of this layer is roughly proportional to the sliding velocity and is apparently not very sensitive to the normal load. Finally, if the rate at which shear occurs during a wear event, or equivalently the flux of events across the wear surface, is also proportional to the sliding velocity, the LV’ proportionality emerges. The recrystallization model may be extended to account for the wear observed in the other metals, summarized in Table 2. The metals may be divided into two categories based on their wear behavior. First there are those that underwent wear by the intense plastic shearing of a thin surface layer, with very small surface disruption; included in this category are high purity copper, Cu-lO%Al, Cu-lO%Sn and silver. In the second category are
362
nickel, aluminum and 2024T4 aluminum alloy, all of which show severe wear with an extensively disrupted wear surface and the occurrence of heavy transfer to the 440 C steel disk. A fundamental material parameter that is consistent with the wear results summarized in Table 2 and of primary importance to the recrystallization process is the stacking fault energy. The stacking fault energy values listed in Table 2 are taken from ref. 12. From Table 2, the first category metals are those with the lower values of stacking fault energy. The recrystallization of these metals tends to proceed without prior recovery [ 131. A large amount of strain energy is therefore available in the lattice to promote rapid nucleation once recrystallization begins. This is consistent with obse~ations made in hot-working experiments ]14] _ Thus it is hypothesized that in the low stacking fault energy metals recrystallization nucleation can keep pace with the wear process and be a controlling factor in promoting low wear. In the case of metals with a higher stacking fault energy, recovery occurs more easily and much of the internal strain energy is relaxed prior to recrystallization. The nucleation step of recrystallization begins at fewer sites and requires more time to go to completion or impingement. The net effect is that either larger recrystallized grains result or islands of recrystallized material are present in a cold-worked matrix at the wear surface. For either situation, intense plastic strain may extend to a considerable depth below the wear surface ultimately resulting in fracture and transfer of material to the disk surface. The importance of the stacking fault energy on the frictional characteristics of copper-aluminum alloys under low speed sliding conditions was described by Buckley [ 151. The role of the stacking fault energy was different under those circumstances; recrystallization did not occur to any significant degree. Rather, the increased strain-hardening rate associated with low stacking fault energy resulted in higher friction for those alloys having low stacking fault energy values.
Conclusions The results of this wear study conducted on Cu, Cu-lO%Al, Cu-lO%Sn, Ag, Ni and aluminum show that: (1) those f.c.c. metals (Cu, Cu-lO%Al, Cu-lO%Sn, Ag) that generally undergo recrystallization nucleation without extensive prior recovery are characterized by low stacking fault energy values and exhibit low wear and very little metal transfer; (2) the wear of the low stacking fault energy metals occurs by the intense plastic shearing of a recrystallized layer of metal comprising the wear surface; (3) the recrystallization model is consistent with quantitative wear trends observed for the case of copper.
363
Acknowledgments The authors express their appreciation to Mr. B. Buzek for the preparation and examination of the transmission electron microscope specimens.
References 1 M. Antler, The lubrication of gold, Wear, 6 (1963) 44. 2 M. Antler, Processes of metaf transfer and wear, Wear, 7 (1964) 181. 3 M. Antler, Wear, friction and electrical noise phenomena in severe sliding systems, ASLE Trans., 5 (1962) 297. 4 M. Cocks, Shearing of junctions between metalsurfaces, Wear, 9 (1966) 320. 5 M. Cocks, Role of displaced metal in the sliding of flat metal surfaces, J. Appl. Phys., 35 (1964) 1807. 6 D. Landheer and J. H. Zaat, The mechanism of metal transfer in sliding friction, Wear, 27 (1974) 129. 7 R. C. Bill and D. W. Wisander, Role of plastic deformation in wear of copper and copper-lo-percent aluminum in cryogenic fuels, NASA Tech. Note, NASA TN D7253, 1973. 8 N. P. Suh, The delamination theory of wear, Wear, 25 (1973) 111. 9 N. P. Suh, S. Jahanmir and E. P. Abrahamson, The delamination-theory of wear, Contract N00014-67-A-0204-0080, NR-229-011, Massachusetts Institute of Technology, Cambridge, Mass., September, 1974. 10 N. P. Suh and P. Sridharan, Relationship between the coefficient of friction and the wear rate of metals, Wear, 34 (1975) 291. 11 D. W. Wisander, Friction and wear of selected metals and of carbons in liquid natural gas, NASA Tech. Note, NASA TN D-6613, 1971. 12 C. Chalmers and T. B. Massaiski (eds.), Progress in Materials Science, Vol. 17, Pergamon Press, New York, 1973, p. 155. 13 D. McLean, Mechanical Properties of Metals, Wiley-Interscience, New York, 1962. 14 J. J. Jones, C. M. Sellars and W. J. Tegart, Strength and structure under hot-working conditions, Metall. Rev., 14 (1969) 1. 15 D. H. Buckley, Possible relation of friction of copper-aluminum alloys with decreasing stacking-fault energy, NASA Tech. Note, NASA TN D-3864, 1967.