Materials Science and Engineering, 25 ( 1 9 7 6 ) 23 - 28

Elsevier S e q u o i a S.A., L a u s a n n e - - P r i n t e d in the N e t h e r l a n d s

23

Wear

ERNEST RABINOWICZ

Professor of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Mass. 02139 (U.S.A.)

SUMMARY

ADHESIVE WEAR

The various forms of wear are considered,

and the fundamental aspects of each which
are not well understood are discussed. In adhesive wear, we do not know the factors
which determine the wear coefficient, perhaps
because we do not know how adhesive wear
particles are formed. Abrasive wear is in rather
good shape, and corrosive wear may also be,
but the resulting equations are so complex
that they are hard to test. Our understanding
of surface fatigue wear will probably always
lag behind our understanding of the fatigue
phenomenon. The large statistical scatter of
wear tests impedes critical testing of all wear
theories, both with regard to the various
forms of wear and where to draw the boundaries between them.

Adhesive wear arises as a result of a process

by which isolated spots on two sliding surfaces
adhere together momentarily so that, when
shear occurs, the shear takes place at some
point other than the original interface. In consequence, an adhering particle is formed, and
this often comes loose at a later stage. Adhesive wear is considered the purest and most
important form of wear because it is the only
one which is always present, and, unlike the
others, cannot be eliminated.
A quantitative expression for adhesive wear
was derived by Archard more than 20 years
ago [3]. Archard used a model for the wear
process involving a non-dimensional constant,
k, to represent the probability that a junction
between two surfaces would lead to the formation of a wear particle. Making this assumption, he was able to derive an expression for
the wear volume, V, formed after sliding
through a distance, x, in the form

INTRODUCTION

Wear is the removal of material from a solid

surface as a result of mechanical action. Following a scheme of classification first outlined
by Burwell [ 1 ], and later followed in a formal
glossary [2] by a committee of the OECD
{Organization for Economic Cooperation and
Development) we may divide wear into four
separate types, namely adhesive wear, abrasive
wear, corrosive wear and surface fracture wear.
In addition, there are several c o m p o u n d types
of wear, like erosion, fretting, and cavitation
erosion which we shall avoid in this presentation. Thus, when considering what we d o n ' t
know about wear, we have to take each of the
forms of wear separately and, in addition,
examine how reliable is our system of deciding
under which conditions which form of wear is
operative.

V -

kLx

(1)

3p

where L is the normal load and p the penetration hardness of the material at the surface.
A more recent paper critically discusses the
other assumptions of the Archard model [4].
Given the relationship of eqn. (1), which
involves only known parameters except for
the constant k, it is clear, as it was clear 23
years ago, that the key problem in analyzing
adhesive wear is to derive a relationship for k,
presumably in terms of the basic material
properties of the contacting materials and any
contaminants or lubricants at the interface.
In fact, over the past 23 years no progress
whatever has been made in this direction, either
because the problem is inherently a difficult

24
I

io -t

10-2
o~

10 -2

._u
10-3

c
o

Q.

~ 10-3
o
I0 -4

c:

g 10-4

io-S i_~

~, i0-5

lO-6 ~
i0 "6

10-7

Cleon
High Vacuum

No lubricant
in air

?oar lubricant
Water
Gasoline
Non - wetting
liquid metal

Fair lubricant
Pure Mineral oil
Molten gloss
Wetting liquid
metal

Good lubricant
M~nerol oil with
lubricity additive
Fatty oil
Good synthetic
lubricant

Fig. 1. Typical values o f t h e wear c o e f f i c i e n t , k, as a f u n c t i o n o f the s t a t e o f lubrication o f t h e surfaces. Similar

metal pairs are m u t u a l l y soluble, while dissimilar metals have low m u t u a l solid solubility. The values o f k differ
by a b o u t a f a c t o r o f t h r e e d e p e n d i n g o n w h e t h e r we are measuring wear by t r a n s f e r t o the o t h e r surface, or by
loose particle f o r m a t i o n .

one or else because the numbers or abilities of

the research workers who have studied wear
have been inadequate.
What little has been done has been in the
mundane activity of collecting values of k
applicable to various sliding situations [5] and
making appropriate generalizations [6]. The
result of one such study is shown in Fig. 1.
These data have also been published in tabular
form and reproduced several times [7, 8]. In
addition, there have been several suggestive
discoveries with regard to material combinations which give exceptional k values. For
example, it has been found that hexagonal
metals with large c/a ratios (like cobalt and
rhenium) give very low values of k [9]. However, no one has yet been able to integrate
these empirical observations and figure out
which parameters determine k under which
circumstances.
Before leaving this problem, I should perhaps c o m m e n t that perhaps one reason why
it is so difficult to analyse the adhesive wear
process is because the process itself is so implausible. If Fig. 2 represents a junction between a top and a b o t t o m sliding surface, then
an adhesive wear particle forms when the junction shears, n o t at the original interface {path 1),
but within one of the two bodies, say along

Path 2

Fig. 2. S c h e m a t i c illustration o f a j u n c t i o n b e t w e e n
asperities o n t w o c o n t a c t i n g surfaces. Shearing along
p a t h 2 will result in a wear particle f r o m t h e t o p surface being t r a n s f e r r e d to the b o t t o m surface.

path 2. Since path 2 is, of course, longer than

path 1, and furthermore the interface is generally contaminated so that it is weak mechanically, it seems impossible that shear will
ever occur along path 2, but occasionally it
does. Furthermore, if the b o t t o m surface has
greater mechanical strength than the top surface, then it is found that most of the shear
that does not occur at the interface occurs
within the weaker material, as we would expect, b u t every once in a while shear occurs
within the b o t t o m stronger materials. Thus,
when sliding a polymer against steel, occasionally steel particles are found adhering to the
polymer surface [10].
It should perhaps be mentioned that there
are a number of other general problems in
adhesive wear theory. The first one we might
consider is that of predicting the size of wear
particles. There has been one solution to this

25

problem, which yields a value for the diameter,

d, of loose wear particles in the simplified
form
d -

60,000

Wab

(2)
P
where Wab is the surface energy of adhesion at
the interface [11]. However, although this
equation gives values that are rather well
o b e y e d in practice, it is appropriate to point
o u t that the derivation has n o t been generally
accepted, perhaps because it seems t o o simple,
and also because it cannot be readily extended
to give the size of adherent wear particles.
Two other adhesive wear problems await
solutions. One is with regard to the observation that in some sliding situations, generally
at low interfacial pressures and in the presence
of good lubricants, no wear particles form, b u t
instead the surfaces become burnished. There
has been very little study of this problem,
apart from one isolated effort [ 12 ], and no
attempts have been made to determine wear
rates to be expected for systems sliding in the
burnishing regime.
The last general problem is a very practical
one, namely, what is the least k value that can
be expected in any sliding system, and under
what conditions will it be obtained. To give
an example, are there any situations in which
k = 0, so that no adhesive wear occurs? Surely
many design engineers would like to have the
answer to this question.

of wear volume, load, hardness, etc., can be

arranged in a dimensionally consistent way.
Values of tan0 to be expected in various
circumstances have been tabulated [ 14] and,
in addition, a number of subsidiary relationships has been clarified. For example, the
reduction in wear rate which occurs when the
abrasive hardness and the workpiece hardness
are nearly equal [ 1 5 ] , and the reduction in
wear rate which is observed when small, abrasive particles are used [16], are also reasonably well understood.
Perhaps the main puzzling aspect of abrasive wear theory is trying to come up with a
good explanation of the fact that a heattreated form of a metal gives higher values of
tan0 than does an annealed form of the same
material (Fig. 3). An explanation has been
suggested in terms of the different shapes of
the grooves produced by the abrasive in the
two cases, b u t it is not clear that the explanation is correct.

/
/
l

w
u

,.o

X12

40

ABRASIVE WEAR

This form of wear is perhaps the only one

which is reasonably well understood. A simple
evaluation of the wear produced by a hard
protuberance plowing a groove in a softer surface so that the material in the groove is worn
away, yields a value for the wear volume in
the form
-

3p

/
0

tanOLx
y

/-

20

(3)

where tan0 is the tangent of the average effective roughness angle for the hard surface, or
for hard abrasive grains at an interface [13].
This, of course, is similar in form to Archard's
equation, which is perhaps inevitable in view
of the limited number of ways the quantities

200

400
600
HARDNESS~HD--Kg/mrn~

e00

Fig. 3. Data o f K r u s c h o v [17 ] o n t h e abrasive w e a r

resistance ( t h e reciprocal o f the abrasive w e a r rate) as
a f u n c t i o n o f material hardness. Lines o f c o n s t a n t
w e a r c o e f f i c i e n t pass t h r o u g h t h e origin. T h e f a c t t h a t
for h e a t - t r e a t e d steels t h e wear resistance-hardness
line does not pass through the origin shows that tan0
is a f u n c t i o n o f hardness.

26

Another problem area in the abrasive wear

field is that of deriving a comprehensive model
for the polishing phenomenon. There are possible explanations for polishing in terms of
the difficulty of forming very fine, abrasive
chips [18], but a lot of quantitative and
qualitative aspects of polishing still remain to
be unravelled.

CORROSIVE WEAR

This is a form of wear in which corrosion

of a surface occurs, then sliding occurs and
removes the corrosive product, and this allows
further corrosion to occur.
Quantitative information on this form of
wear is not readily available, because the wear
rate is governed by so m a n y factors, namely,
by the rate of corrosion, the rate of removal
of corrosive product, the geometry of the
sliding surfaces, whether the whole of the
corrosive product is removed all at once or in
part, etc. What quantitative expressions have
been derived [19] are so complex and have so
many adjustable constants, that there are no
ready means of testing them.
In spite of this, I do not regard this as a
form of wear about which we know little,
since the qualitative description of corrosive
wear matches the experimental results reasonably well. It is just that the p h e n o m e n o n is
inherently a complicated one.

SURFACE FRACTURE WEAR

There are two sub-divisions of this type of

wear. One of them is the wear of brittle materials like glass, which proceeds via the formation of large, chevron-shaped cracks during
sliding. This form of wear is not well understood in a quantitative way, and indeed, I do
not think attempts to analyse it quantitatively
have been frequent.
The other sub-division is surface fatigue
wear, a c o m m o n form of wear of rolling elements like ball- and roller bearings, wheels and
rails, and gears. In this case, as the stresses
come and go during continued rolling service,
cracks grow in or below the surface of the
element, and eventually a spalled particle is
produced. This phenomenon is closely related
to the fatigue failure p h e n o m e n o n (for exam-

ple, in the relationship between stress and

life) but there are a number of differences,
which are n o t yet explained.
Thus, there is in normal fatigue situations
a fatigue limit, namely, a stress equal to about
half the yield stress such that no fatigue
failure occurs if the stress is always kept
below the limit. No such limit appears to
exist in surface fatigue wear.
In ordinary fatigue, if a hundred specimens
are tested, the range in fatigue lives is about a
factor of ten. In surface fatigue failures the
ratio is about a factor of 100, and no obvious
reason for this greater scatter has been determined.

STATISTICAL SCATTER DURING WEAR

Now that we have mentioned the matter of

scatter in wear values, we should perhaps discuss this matter specifically. In surface fatigue
wear, we can carry out identical repeat tests
and find that the lives (defined as the periods
from the start of testing until the first spalled
particle is formed) vary by up to two orders
of magnitude. We console ourselves for this
very wide scatter with the thought t h a t the
fatigue p h e n o m e n o n is always associated with
large statistical fluctuations.
In abrasive wear testing, the experimental
scatter in the wear rates is quite small, less
than 2% in carefully-controlled repeat experiments [20], and, indeed, abrasive wear can be
used as a reliable materials test, for example,
in the Taber tester [21]. In the case of adhesive wear, however, carefully controlled repeat experiments frequently give wear rates
that vary by factors of two, five and even ten
(Fig. 4). During one continuous test, the wear
rates also seem to fluctuate by up to factors
of ten from one period of time to another.
14-

C o p p e r specimens

12-=" 1 0 - E
E 8--

_--

._I-.-

4
0

Fig. 4. R e p r e s e n t a t i o n o f t h e results o f f o u r t e e n rep e a t e d a d h e s i v e wear tests, as alike as we could m a k e

t h e m , using c o p p e r s p e c i m e n s . T h e wear r a t e s c o v e r a
r a n g e o f a f a c t o r o f four.

27

No satisfactory general explanation of this

p h e n o m e n o n has been suggested. A further
point is that this scatter makes it difficult to
devise critical experiments for testing theories
of adhesive wear. If y o u r theory is n o t outlandish, and y o u run tests for long enough,
y o u are bound, eventually, to come up with a
data point which just fits your theory.

W H A T F O R M O F W E A R IS W H A T

In most cases, it is relatively easy to decide

which form of wear is applicable in any given
circumstance. One obvious consideration is
the nature of the situation. (Is it rolling or
sliding? Is there a corrosive environment? Are
abrasive particles present?) An i m p o r t a n t
diagnostic procedure consists of an examination of the surfaces and the wear debris, while
in some cases, experimental testing, such as
determining the effect on the wear rate of
changing the variables (for example, using a
better lubricant or increasing the temperature),
usually resolves the question of which type of
wear is present. However, currently, a great
debate is raging in the literature, essentially on
where to put the boundary between adhesive
wear and the fatigue form of surface-fracture
wear.

Briefly, in the last fifteen or so years, and

especially in the last five, a number of research
workers has come to feel that m a n y wear phenomena that are classically categorized as adhesive wear are actually manifestations of
surface fatigue wear [22 - 26]. The factors
that lead to these conclusions are first an
intuitive appeal to experience from the general field of solid mechanics (i.e., instead of
postulating a unique phenomenon, namely
adhesion-caused material transfer, it seems
more sensible to use the well-established
notion of crack growth during cyclic stressing);
secondly, an examination of surfaces after
sliding often shows surface and sub-surface
cracks and analogous surface features and it
seems plausible to relate them to the wear
process. Third, an evaluation of typical wear
rates and wear particle shapes and sizes seems
to m a n y workers to be explained more readily
by a fatigue theory than by an adhesion theory.
If one went merely by the volume of recent
research publications on this topic, one would
suppose that these fatigue theories have by

now become a majority point of view among

workers in the wear field, but this is probably
misleading. First, the various fatigue theories
do n o t agree with each other, so t h a t it is hard
to say that they form a majority at all. Second,
most workers on wear problems are very loath
to abandon the classical adhesive wear theory
and the Archard wear equation because it does
provide a priceless bonus -- namely, an estimate of the wear rate to be expected in any
sliding situation. {This estimate is likely to be
off by about a factor of two or three from
the actually observed wear rate, but an agreement of this type is a lot better than nothing.)
Few quantitative estimates of expected wear
rates are available from the surface fatigue
theorists, though a recent paper [27] has
tried to show that something resembling
Archard's equation actually can be derived
from a fatigue model. Progress along these
lines, allowing for critical tests of the new
wear models, would be very welcome.

EVALUATION

It is clear t h a t most aspects of wear are still

in the "we d o n ' t k n o w " category. As one who
has worked in this field for nearly thirty years
I am rather embarassed about it, feeling that
m y colleagues and I have been supported by
public funds (in one way or another} for all
that length of time, and have little to show
for the resources invested in us.
The other side, however, should be emphasized also. The number of research workers
studying wear in general, rather than wear of
some specific device like a sewing machine or
a specific material like reinforced Teflon, has
never been large. I d o u b t if even t o d a y there
are as m a n y as fifty people in the world involved in basic wear studies, and for most of
the past generation the number has been
smaller than twenty. Given the long neglect
of wear problems, their complexity, and the
difficulties in obtaining reproducible wear
measurements, our record of achievement is
nothing of which to be ashamed.

REFERENCES
1 J. T. Burwell, Survey o f possible wear m e c h a n i s m s ,
Wear, 1 ( 1 9 5 7 ) 1 1 9 - 141.

28
2 Friction, Wear and Lubrication, GlOssary, Organization for Economic Cooperation and Development, Paris, 1969.
3 J. F. Archard, Contact and rubbing of flat surfaces,
J. Appl. Phys., 24 (1953) 981 - 988.
4 J. F. Archard, Wear, in P. M. Ku (ed.), Interdisciplinary Approach to Friction and Wear, NASA
Spec. Publ. 181, Washington, D.C., 1968, pp. 267 333.
5 J. F. Archard and W. Hirs.t, The wear of metals
under unlubricated conditions, Proc. R. Soc.
London, Ser. A, 236 (1956) 397 - 410.
E. Rabinowicz, Friction and Wear of Materials,
Wiley, New York, 1965, Table 6.6.
I. V. Kragelskii, Friction and Wear, Butterworths,
London, 1965, Tables 5.1, 5.2.
D. Pavelescu, Friction and Wear (in Roumanian),
Academy of the Roumanian Socialist Republic,
1971, Table 5.2.
9 E. Rabinowicz, Friction and wear properties of
rhenium, Wear, 10 (1967) 313 - 318.
10 E. Rabinowicz and K. V. Shooter, The transfer of
metal to plastics during sliding, Proc. Phys. Soc.,
London, 65B (1952) 671 - 673.
11 E. Rabinowicz, Friction and Wear of Materials,
Wiley, New York, Section 6.14.
12 E. Rabinowicz, The nature of polished and burnished surfaces, Proc. Int. Conf. Surface Technology, S.M.E., Dearborn, Mich., 1973, pp. 23 - 38.
13 E. Rabinowicz, Friction and Wear of Materials,
Wiley, New York, Section 7.2.
14 E. Rabinowicz, New coefficients predict wear of
metal parts, Prod. Eng. (N.Y.), 29 (1958) 71 - 73.
15 R. C. D. Richardson, The abrasive wear of metals

16
17

18
19
20

21

22

23
24

25
26

27

and alloys, in Lubrication and Wear -- Fundamentals and Application to Design, Institution of Mechanical Engineering, London, 1968, pp. 410 - 414.
E. Rabinowicz, Friction and Wear of Materials,
Wiley, New York, Section 7.5.
M. M. Kruschov, Resistance of metals to wear by
abrasion, as related to hardness, Proc. Conf. Lubrication and Wear, Institution of Mechanical Engineers, London, 1957, pp. 655 - 659.
E. Rabinowicz, On the mechanism of polishing
with abrasives, Wear, 18 (1971) 169 - 170.
H. H. Uhlig, Mechanism of fretting corrosion, J.
Appl. Mech., 76 (1954) 401 - 407.
E. Rabinowicz, L. A. Dunn and P. G. Russell, A
study of abrasive wear under 3-body conditions,
Wear, 4 (1961) 345 - 355.
Resistance of transparent plastics to surface abrasion, ASTM Test Method D 1044, American
Society for Testing Materials, Philadelphia, 1973.
R. G. Bayer, W. C. Clinton, C. W. Nelson and R. A.
Schumacher, Engineering model for wear, Wear, 5
(1962) 378 - 391.
L. Rozeanu, Fatigue wear as a rate process, Wear,
6 (1963) 337 - 340.
K. Endo and Y. Fukada, The role of fatigue in
wear of metals, Proc. 8th Japan Congress Testing
Materials, Kyoto, 1965, pp. 69 - 72.
N. P. Suh, The delamination theory of wear, Wear,
25 (1973) 111 - 124.
Y. Kimura, An interpretation of wear as a fatigue
process, Proc. JSLE-ASME Lubrication Conf.,
Tokyo, 1975, pp. 89 - 95.
J. Halling, A contribution to the theory of mechanical wear, Wear, 34 (1975) 239 - 249.