Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Wear, 100 (1984) 533 - 541 533

THE LEAST WEAR

ERNEST RABINOWICZ
Department of Mechanical Engineering, Massachusetts Institute of Technology, Cam-
bridge, MA 02139 (U.S.A.)

Summary

Of the four principal types of wear, adhesive wear is the only one
which can never be eliminated. The three types of adhesive wear, namely
severe wear, moderate wear and burnishing, are described, and the transitions
between them are discussed. Burnishing, or material removal on a molecular
scale, represents the least possible amount of adhesive wear, but we know
little regarding the magnitude of the wear rate and methods of ensuring
that a sliding system will operate in the burnishing regime. This is unfor-
tunate because, for many sliding systems, especially those using unlubricated
surfaces, there is no likelihood of achieving an acceptable life unless opera-
tion of the sliding surfaces in the burnishing regime can be assured.

1. Introduction

It is widely recognized that wear causes great damage to material


objects of all kinds, although the amount of this damage is in some dispute.
A number of published estimates of the annual loss caused by wear to
industrialized nations are in the range 0.5% - 1% of their gross national
product (GNP) [ 1, 21. Values such as this, amounting in the U.S.A. to
U.S. $15 X 10’ - $30 X 10’ year-‘, represent a considerable underestimate,
since the annual damage caused by wear to the American automobile alone
is about U.S. $58 X log [3]. A value for the total wear damage of 6% of the
GNP, currently U.S. $180 X 10’ year-’ in the U.S.A., seems much more
realistic, and moreover is in good agreement with values from the U.S.S.R.
of 9.5% of the GNP attributable to major equipment overhaul costs [4],
a large part of which represents the undoing of the damage done by wear.
Since wear constitutes such a severe practical problem there is of
course great interest in minimizing it. In this note I plan to discuss the least
wear that is feasible in sliding systems, and then to consider what is known
about methods of achieving it.
Now the two most important forms of wear, namely adhesive and
abrasive wear, obey a Holm-Archard relationship [ 5,6] of the type
wear coefficient X load X distance of sliding
wear volume = (1)
hardness

0043-1648/84/$3.00 @ Elsevier Sequoia/Printed in The Netherlands


f

1O-2 - 10-8
1
xlhesive wear abrasive w2ar corrosive wear surface fracture v+ear

10-3 - 10-7 1o-2 - 1o-5 10-2 - 10-5 1o-4 - 10-s


I , r I
solid
karicant

1o-5

brittle surface
fracture fixtime

10-4 18

Fig. 1. Typical wear coefficient values for metal-on-metal sliding systems.


10-2 10-E
.--
I I
1 I
I
I I

zdbesive wear abrasive wear

lo--@ - 10-a I I 10-2 - 10-6 I I 1o-2 - 1o-5 I

polishing fSArface
fatigue

~ lo+ 10-7
~

Fig. 2. Typical wear coefficient values for ceramic-on-ceramic systems.


536

The other principal forms of wear, namely corrosive and surface frac-
ture wear, do not obey eqn. (l), but in many cases they can be assigned
equivalent wear coefficients based on typical observed wear rates. All these
wear coefficients can then be shown on one chart. Figure 1 shows wear
coefficients for metals while Fig. 2 gives the same information for ceram-
ics. Since in sliding systems we generally minimize wear by eliminating
sizeable amounts of abrasive, corrosive and surface fracture wear, all of
which are in principle avoidable, we then only have adhesive wear to contend
with.
Equation (1) indicates that, given that we have a sliding device in which
a specified load must be moved a specified distance, we have just two ways
of minimizing adhesive wear, namely to use hard materials or to achieve a
low wear coefficient. The hardness range which is available in practical
bearing materials is relatively limited, ranging from a reinforced Teflon with
a hardness of 10 kgf mmV2 to boron carbide with a hardness of 3000 kgf
mme2, i.e. over roughly 2.5 orders of magnitude. The wear coefficient
ranges from 10e2 for the adhesive wear of identical metals in a vacuum
environment to values of perhaps 10Pg for very well-lubricated incompatible
metal pairs [ 71. The wear coefficient can thus cover a range of 7 orders of
magnitude, clearly an important fact to bear in mind when planning to
minimize adhesive wear (i.e. changes in bearing materials which reduce the
wear coefficient are much more likely to reduce the wear significantly
than are changes in hardness alone).

2. Types of adhesive wear

In general there are three types of adhesive wear. In metals, severe


galling wear occurs when clean or poorly lubricated metal pairs with a high
degree of metallurgical compatibility are slid over each other. Such sliding
systems generally give wear coefficients in the range 10m2 - 10T4 and wear
particle sizes in the range 200 - 20 pm.
Moderate wear occurs with less compatible or with well-lubricated
surfaces, and it occurs more readily when the pressure at the interface
between the sliding surfaces is low. Typical wear coefficients are around
lo4 - 10-6, while wear particle sizes are in the range 20 - 2 pm.
A third wear regime, burnishing, is encountered in special circum-
stances, e.g. with highly incompatible metal pairs (or with a hexagonal-
structured metal as one of the two surfaces). The pressure should be low
and if possible a very good lubricant should be used. In this case the surfaces
take on a burnished appearance and no sizeable wear particles are observed,
while the wear coefficients are typically in the range 10T6 - 10e8.
With non-metals severe adhesive wear occurs rarely, and in usual prac-
tice only moderate wear and burnishing wear are encountered.
We note with interest that the transition from one wear regime to
another, as the sliding conditions are gradually changed, often occurs quite
537

abruptly. With metals the sudden change from severe to moderate wear,
which is generally accompanied by a change from a system giving large
metallic wear particles to one yielding small oxide particles, has been much
studied [ 8,9]. Such factors as the attainment of a critical temperature [lo}
or oxide thickness [ll, 121 or a balance between oxidation rate and wear
rate [13] have been invoked. In contrast, research on the factors deter-
mining the transition between the low and the burnishing wear regimes
has been far more modest. Before discussing this aspect, it is probably
appropriate to discuss burnishing wear in some detail.

3. The burnishing wear regime

It should be emphasized from the start that there has been little sys-
tematic study of burnishing and the wear coefficients associated with it.
In part this can be explained by the experimental difficulties involved in
studying this wear regime. For example, if in a system giving a wear coef-
ficient of loss, two surfaces of hardness 1000 N mms2 are pressed together
with a force of 1 N and slide at a speed of 100 mm s-l, it will be 3 years
before the wear reaches 1 mg.
Another reason for the neglect of studies of burnishing is that few
people realize that this constitutes a separate and major type of adhesive
wear. In addition, the present author is moved to point out that one reason
why low wear sliding systems are studied so little is that there seems to be
an excessive fascination, quite divorced from reality and in some respects
almost morbid, with studies of such highly wearing sliding systems as un-
lubricated copper sliding on copper. The present author over the years has
carried out his fair share of such studies [ 13,141.
It should be noted that there has as yet been no attempt to predict
the magnitude of the wear coefficient in the bunching regime from more
fundamental parameters. Indeed, it is not even obvious that the Holm-
Archard equation governs burnishing wear (and hence that a wear coeffi-
cient is an appropriate parameter to use in describing this type of wear)
although there is no reason to believe that it does not.
The mechanism of burnishing wear may be deduced if it is assumed
that a burnished surface is smooth on a molecular scale and that burnishing
is a process of single-molecule removal from the peaks of the asperities, so
that eventually a perfectly smooth surface remains. An analogous mechanism
has been postulated for polishing, which is the abrasive wear process for
producing a smooth surface [ 151.
Processes such as burnishing and polishing are thus analogous to mate-
rial removal from a liquid by evaporation. Evaporation is a slow process
which proceeds on a molecular scale, producing a smooth surface. There is
also a rapid material removal process from a liquid which is heated rapidly.
This is boiling, which produces a rough surface.
538

We can make some qualitative statements about the magnitude of the


wear coefficient in the burnishing mode if it is assumed that burnishing
represents evaporation and is stimulated by the sliding action which pro-
vides extra energy to the surface molecules. We would anticipate that a
strong solid such as a ceramic in which the molecules are firmly bonded to
each other would tend to give a low wear coefficient, while a weaker solid
such as a metal would give a higher wear coefficient. (A strong solid is one
with high hardness, high melting temperature, high latent energy of evapo-
ration and low coefficient of expansion.) Figures 1 and 2 give typical bur-
nishing wear coefficient values (10 -’ for ceramics and lo-’ for metals) and
it is seen that ceramics do indeed give lower wear coefficient values than do
metals. However, not enough data are available for a judgment to be made
as to whether, among metals, the stronger metals give lower wear coefficients
than do the weaker metals.

4. Transition between moderate wear and burnishing wear

Indications are that the process of burnishing wear (i.e. the removal
of material on a molecular scale) always occurs during sliding. However, it
is swamped by the process of moderate adhesive wear if this also occurs.
Hence, burnishing wear is only observed when moderate adhesive wear is
absent, and the transition occurs at the point where the occurrence of
moderate adhesive wear becomes minimal.
Two equations for this have been derived. One, applicable to a sin-
gle point contact such as a crossed cylinder system or a pin on a disk,
states that the critical normal force L at the transition obeys the relation-
ship

L = 7r x 10s -W*b2
P
where Wab is the surface energy of adhesion and p is the hardness of the
softer surface [ 161.
This equation is reasonably well obeyed in practice, especially when
noble metals are used. Given clean metal surfaces with values of Wab of
low3 N mm-‘, and p values of lo3 N mm-*, we find a critical normal force
L of 0.31 N. The observed critical forces (i.e. burnishing observed at normal
forces below the critical) are generally rather smaller than this, typically
0.05 N.
For two surfaces with a sizeable apparent area of contact, over which
the junctions are uniformly distributed, the apparent stress u at the transi-
tion is given by
539

where 1z,, is the wear coefficient in the burnishing wear regime and lz, is
the wear coefficient for moderate wear [ 171. For a typical metallic system
where kb = lo-’ and Iz, = 10m5, the critical apparent stress is l/600 of the
hardness stress, and for metals of medium hardness is typically 1.6 N mmm2
(200 lbf ine2).
This equation has not been widely tested, but it does correctly indicate
that it is easier to achieve burnishing with metal pairs such as cobalt against
cobalt or nickel against silver, both of which have exceptionally low values
of k, (around 10e6) and hence exceptionally high critical stresses for opera-
tions in the burnishing regime.
The main problem with this equation is that it gives no indication of
what happens in practical systems which are designed to operate eventually
at low apparent stresses but which start out initially with contacts made
over a limited portion of the total interface, thus constituting a number of
high stress contacts. Will such a system run itself into the burnishing regime,
or will it not?

5. Discussion

In the above sections our limited quantitative knowledge of the low


wear form of adhesive wear, namely burnishing wear, has been summarized.
Surfaces operating in this wear regime are capable of giving very low wear
coefficient values, but we know too little about the circumstances in which
burnishing occurs and these low wear coefficients may be obtained. When
put into these terms, the question of whether or not burnishing occurs
seems to be an academic discussion of an esoteric topic. However, this is a
fallacious way of looking into the situation; given the great economic im-
portance of wear, hardly any discussion of a wear problem is without some
practical value. In this case, the practical use of our discussion lies in the
fact that many sliding systems can only be operated in a commercially
successful manner if the wear coefficient is very low, e.g. lo-’ or less, and
burnishing wear is probably the only way of reaching such low wear values.
For example, let us take the all-ceramic internal combustion engine,
currently under active development in numerous laboratories and institu-
tions around the world. Typically, it is hoped to develop an engine which
needs no lubricant and has a life of about 400 000 km. While an exact wear
calculation of course depends on the detailed design of a specific engine, an
approximate calculation based on the features of typical engines can readily
be carried out.
Let us concentrate on one component, the piston ring, and assume that
the engine is worn out by the time that the radial wear of the ring reaches
0.1 mm.
We may write Archard’s equation in the form
k X apparent stress X distance of sliding
depth of wear = (4)
hardness
540

The average apparent stress at the piston ring-cylinder interface must


be about 0.7 N mm-2, so that the ring can withstand the pressure generated
in the combustion chamber.
The distance of sliding of the piston ring is typically one-quarter that
of the automobile, or 10” mm. The hardness of the ceramic is likely to be
about 2 X lo4 N mmP2.
On substitution in eqn. (4), we find that k must be 3 X lo-* (or less).
As we have seen above (Fig. 2) this value can be achieved in the case of
ceramic sliding systems only in the burnishing regime. We should not be
surprised to learn that unlubricated ceramic engines, which generally do not
operate in the burnishing regime, generally fall far short of meeting their
targeted life performance.
A question which comes up from time to time is whether adhesive wear
is always present when we have sliding, or whether one might have sliding
without wear. The above discussion suggests that severe wear and moderate
wear are indeed preventable but that burnishing wear is inevitable, in the
same way that evaporation from a solid is inevitable. However, some room
temperature evaporation rates of solids are so small as to be negligible and
perhaps some burnishing wear rates are negligible also. Investigation of
this question is a problem for future research.

References
1 Lubrication (Tribology), Education and Research, Her Majesty’s Stationery Office,
London, 1966.
2 J. Hansen, Tribology: economic importance and BMFT sponsorship, Metallurgical
Aspects of Wear, Deutsche Gesellschaft fiir Metallkunde, Oberursel, 1981, pp. 1 - 8.
3 E. Rabinowicz, Tribology II, Video Course Manual, 1983, p. S7.7 (Center for Ad-
vanced Engineering Study, Massachusetts Institute of Technology, Cambridge, MA).
4 I. V. Kragelskii, S. S. Silin and Yu. P. Zampatin, Tribology in the higher school,
Sou. J. Friction Wear, 2 (3) (1982) 31 - 35.
5 R. Holm, Electric Contacts, Almquist and Wiksells, Stockholm, 1946, Section 40.
6 J. F. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys., 24 (1953) 981 -
988.
7 C. N. Rowe, Lubricated wear. In M. B. Peterson and W. 0. Winer (eds.), Wear Control
Handbook, American Society of Mechanical Engineers, New York, 1980, pp. 143 - 160.
8 J. F. Archard and W. Hirst, The wear of metals under unlubricated conditions, Proc.
R. Sot. London, Ser. A, 236 (1956) 397 - 410.
9 M. B. Peterson, J. J. Florek and R. E. Lee, Sliding characteristics of metals at high
temperatures, ASLE Trans., 3 (1960) 101 - 109.
10 T. F. J. Quinn and J. L. Sullivan, A review of oxidational wear. In W. A. Glaeser, K.
C. Ludema and S. K. Rhee (eds.), Proc. Znt. Conf. on Wear ofMaterials, 1977, Amer-
ican Society of Mechanical Engineers, New York, 1977, pp. 110 - 115.
11 R. T. Foley, M. B. Paterson and C. Zapf, Frictional characteristics of cobalt, nickel
and iron as influenced by their surface oxide films, ASLE Trans., 6 (1963) 29 - 39.
12 E. Rabinowicz, Lubrication of metal surfaces by oxide films, ASLE Trans., 10 (1967)
400 - 407.
13 E. Rabinowicz and S. N. Sankaran, Thickness effects using cupric oxide as solid
lubricant for copper, Proc. 3rd Znt. Conf. on Solid Lubricants, American Society of
Lubrication Engineers, Park Ridge, IL, 1984.
541

14 E. Rabinowicz and D. Tabor, Metallic transfer between sliding metals: an autoradio-


graphic study, Proc. R. Sot. London, Ser. A, 208 (1951) 455 - 475.
15 E. Rabinowicz, The nature of polished and burnished surfaces, Proc. Int. Conf. on
Surface Technology, Society of Manufacturing Engineers, Dearborn, MI, 1973, pp.
23 - 38.
16 E. Rabinowicz, Friction and Wear of Materials, Wiley, New York, 1965, Section 6.16.
17 E. Rabinowicz, Discussion. In P. M. Ku (ed.), Interdisciplinary Approach to Friction
and Wear, NASA Spec. Publ. SP-181, 1968, pp. 304 - 308 (National Aeronautics
and Space Administration, Washington, DC).

You might also like