STP780 Eb.1415051 1 PDF

Copyright by ASTM Int'l (all rights reserved); Fri Jan 27 22:01:44 EST 2012
Downloaded/printed by
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions authorized.
MATERIALS EVALUATION
UNDER FRETTING
CONDITIONS
A symposium
sponsored by
ASTM Committee G-2
on Erosion and Wear
Warminster, Pa., 3 June 1981
ASTM SPECIAL TECHNICAL PUBLICATION 780

S. R. Brown, Naval Air Development
Center, symposium chairman
ASTM Publication Code Number (PCN)

04-780000-29
iSlb 1916 Race Street, Philadelphia, Pa. 19103
Copyright © by AMERICAN SOCIETY FOR TESTING AND MATERIALS 1982
Library of Congress Catalog Card Number: 81-71869
NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication.
Printed in Ann Arbor. Mich.

August 1982
Foreword
The symposium on Materials Evaluation Under Fretting Conditions was
held on 3 June 1981 in Warminster, Pa. The sponsors for the event were
ASTM Committee G-2 on Erosion and Wear and its Subcommittee G02.30
on Wear. The symposium chairman was S. R. Brown of the Naval Air
Development Center.
Related
ASTM Publications
Erosion: Prevention and Useful Applications, STP 664 (1979), 04-664000-29
Erosion, Wear, and Interfaces with Corrosion, STP 567 (1974),

04-567000-29
Characterization and Determination of Erosion Resistance, STP 474 (1970),

04-474000-29
Corrosion and Degradation of Implant Materials, STP 684 (1979),

04-684000-27
Selection and Use of Wear Tests for Metals, STP 615 (1977), 04-615000-23
Statistical Analysis of Fatigue Data, STP 744 (1981), 04-744000-30
Tables for Estimating Median Fatigue Limits, STP 731 (1981), 04-731000-30
A Note of Appreciation
to Reviewers
This publication is made possible by the authors and, also, the unheralded
efforts of the reviewers. This body of technical experts whose dedication,
sacrifice of time and effort, and collective wisdom in reviewing the papers
must be acknowledged. The quality level of ASTM publications is a direct
function of then- respected opinions. On behalf of ASTM we acknowledge
with appreciation their contribution.
ASTM Committee on Publications
Editorial Staff
Jane B. Wheeler, Managing Editor
Helen M. Hoersch, Senior Associate Editor
Helen P. Mahy, Senior Assistant Editor
Allan S. Kleinberg, Assistant Editor
Virginia M. Barishek, Assistant Editor
Contents
Introduction 1
Occurrence of Fretting in Practice and Its Simulation in the

Laboratory—R. B. WATERHOUSE 3
A New Macliine for Studying Surface Damage Due to Wear and

Fretting—D. KUSNER, C. POON, AND D. W. HOEPPNER 17
An Evaluation of Fretting at Small Slip Amplitudes—p. KENNEDY,

M. B. PETERSON, AND L. STALLINGS 30
Evaluation of Fretting Corrosion by Means of a New Device for the
Control of Oscillation Amplitude—K. G. BUDINSKI 49
Fretting of Electrical Contacts—MORTON ANTLER 68
Heat Exchanger Tube Fretting Wear: Correlation of Tube Motion and

Wear—p. L. KO, J. H. TROMP, AND M. K. WECKWERTH 86
An Electrochemical Method for Investigating Corrosion in Rubbing
Surfaces—V. s. AGARWALA, K. Y. KIM, AND S. BHATTACHARYYA 106
Fretting-Resistant Ion-Plated Coatings—M. G. EL-SHERBINY AND

F. B . SALEM 125
Effect of Ion Implantation on Fretting Fatigue in Ti-6Al-4y Alloy—

R. G. VARDIMAN, D. CREIGHTON, G. SALIVAR, A. EFFATIAN, AND
B. B. RATH 138
Fretting Fatigue of Ti-6A1-4V Alloy—c. LUTYNSKI, G. SIMANSKY, AND

A. J. MCEVILY 150
Review of Factors That Influence Fretting Wear—R. C. BILL 165
Copyright by ASTM Int'l (all rights

(PDVSA Los Teques) pursuant to
STP780-EB/Aug. 1982
Introduction
The term fretting refers to small-amplitude oscillating motion between two

solid surfaces in contact and is often used to denote fretting corrosion, fret-
ting wear, fretting fatigue, and other damage that may ensue. This ASTM
definition is both descriptive and complete in setting limits for fretting condi-
tions. Fretting damage occurs in a wide range of items, such as household
appliances, automobiles, aircraft, electrical equipment, and even human
body implants. As a result, investigations to prevent fretting and to develop
fretting tests take place in many different areas of science and industry.
The aim of this symposium was to bring together active contributors and
others concerned with test methods for evaluating materials and equipment
designs under fretting conditions. While standard test procedures and prac-
tices for most aspects of wear testing are developing at a reasonable rate, the
same cannot be said for fretting test methods. The last ASTM symposium on
the subject of fretting was in 1952,^ and it has been 10 years since R. B.
Waterhouse documented in Fretting Corrosion the wide variations in equip-
ment used for experimental studies of fretting wear and corrosion.^ This
symposium was intended to be a state-of-the-art review of the equipment,
devices, and techniques available for controlling fretting test conditions and
for assessing damage.
In the keynote paper, Waterhouse traces the systematic investigation of
fretting phenomena and presents classic illustrations of fretting damage. He
points out that test environments in laboratory studies must be controlled as
well as important mechanical variables such as the amplitude of slip between
the two surfaces and the normal load applied to the surfaces. Allowances
must be made to prevent linkages and elastic deformation of load arms from
becoming sources of testing error. Waterhouse observes that practical situa-
tions in which fretting may be experienced are so varied that the investigator
must decide which apparatus or method most closely approximates the prac-
tical application.
Three papers describe general purpose fretting test devices. One new test
machine was capable of simultaneously testing eight replicate specimens.
Another device was a speaker drive mechanism for controling the yaw motion
of a flat specimen against a fixed steel ball. A simple slider-crank tester was
^Symposium on Fretting Corrosion, ASTM STP 144, American Society for Testing and
Materials, PhUadelphia, 1952.
^Waterliouse, R. B., Fretting Corrosion, Pergamon Press, Oxford, 1972.
Copyright by ASTM Int'l (all rights reserved); Fri Jan 27

Copyright' 1982 b y AS FM International www.astm.org
(PDVSA Los Teques) pursuant to License Agreement. No f
MATERIALS EVALUATION UNDER FRETTING CONDITIONS
used for close control of fretting variables in solving fretting problems of in-
dustrial equipment.
Two papers discuss the involvement of computers in controlling conditions
and correlating results. A paper on the simulation of electrical contact fret-
ting describes the instrumentation for contact resistance measurements and
equipment which provides precise computer control of fretting movements.
A computer is used to correlate fretting parameters and predict heat ex-
changer tube wear from the data generated by a fretting apparatus. Another
paper deals with an electrochemical polarization cell combined with a pin-
on-disk device to establish a correlation among corrosion, wear, and the
nature of rubbing surfaces.
Three papers are concerned with methods of preventing fretting. A special
fretting test was used to evaluate ion-plated films for fretting prevention. Ion
implantation was used to improve fretting fatigue life of titanium alloy
specimens. Fretting fatigue experiments were carried out on shot-peened
specimens to study the effects of surface conditions on resistance to fretting.
The final paper in this volume is a critical review by Bill of the effects of
various factors on fretting wear. He divides these factors into the three broad
categories of (1) contact conditions, (2) environmental conditions, and (3)
materials properties and behavior. More fretting studies are clearly needed to
provide a better understanding of these factors, especially their interactions.
The information in this volume should be useful to scientists and engineers
concerned with the development and testing of fretting-resistant materials.
Test equipment, procedures, and damage assessment methods are explained
in detail and will be helpful to others planning evaluations to be conducted
under fretting conditions.
S. R. Brown
Naval Air Development Center, Warminster,
Pa. 18974; symposium chairman.
R. B. Waterhouse^
Occurrence of Fretting in
Practice and Its Simulation
in the Laboratory
REFERENCE: Waterhouse, R. B., "Occurrence of Fretting in Practice and Its Simula-

tion in the Laboratoiy," Materials Evaluation Under Fretting Conditions, ASTM STP
780, American Society for Testing and Materials, 1982, pp. 3-16.
ABSTRACT: Fretting arises in practice where a structure or multicomponent system in-

volving contacting surfaces experiences vibration. Examples are the false brinelling of
wheel bearings when automobiles are transported by road or the damage which can arise
in gun mountings on military vechicles. Fretting also occurs when one of the members of
a contact is subjected to cyclic stressing, as in the contact region between the hub and
wheel seat of a wheel-on-axle or flywheel-on-shaft assembly. In attempting to simulate
fretting situations in the laboratory, attention must be given to the contact configuration,
the mode of exciting oscillatory slip, the control and measurement of the amplitude of
slip, alignment and application of the normal load, and control of the environment. Fi-
nally, the resultant damage must be assessed and the method used should take into ac-
count not only the material removed from the surfaces but the material transferred be-
tween the surfaces as well as surface deformation and cracks, which in practice could
have a profound influence on the life of the component.
KEY WORDS: fretting, wear, materials fatigue, testing machines, slip amplitude, fre-
quency, environment, high temperature, measurement, material transfer
The first systematic investigation of fretting phenomena was made in 1927

by Tomlinson [1],^ who described the essential features of its occurrence,
namely, that it only occurred where there was relative movement or "slip"
between the surfaces; that it occurred at very small amplitudes of movement,
even as low as 0.0025 ^m, that is, of the order of interatomic distances; and
that the environment had a significant effect on the nature of the debris.
Although one might be somewhat skeptical about Tomlinson's ability to pro-
duce such small movements mechanically, his observations are still relevant
'Reader in metallurgy. Department of Metallurgy and Materials Science, University of Not-

tingham, Nottingham NG7 2RD, United Kingdom.
2The italic numbers in brackets refer to the list of references appended to this paper.

Copyright 1982 b y AS FM International www.astm.org
4 MATERIALS EVALUATION UNDER FRETTING CONDITIONS
today and his papers are worth studying. Since World War II, the number
and variety of reported cases of fretting has grown steadily, particularly with
the development of nuclear power engineering and the gas turbine engine. A
less-publicized area of cases has been in the medical field where fretting has
been observed in orthopedic implants and fracture fkation devices. Recently
the author has seen examples of fretting damage on the contacting surfaces
of contiguous teeth, which destroys the surface of the tooth enamel and pro-
duces a typical wear scar. This area is a common site for the development of
caries.
The Origin of Fretting Movement in Practice
Cyclic Stress
In practice fretting movement often arises from an assembly or its com-
ponents being cyclically stressed. Very often if the two components forming
the contact were fabricated as one unit there would be a severe stress concen-
tration at the point where the contact region is situated, for example, a
flywheel on a shaft (Fig. 1). In the single-component unit, stress concentra-
tion would result at the circumferential points marked A in Fig. 1. In the
assembly, slip at the edge of the contact region would relieve this stress con-
centration but would result in fretting at the circumferential areas marked F.
Fatigue cracks are commonly initiated in this region but they may not
necessarily continue to propagate to failure, as the detailed work of Horger
on railway axles has demonstrated [2].
Other common sites of fretting arising from the cyclic stressing of one of
SINGLE COMPONENT ASSEMBLY
w w w w
•SHAFT-
-FLY WHEEL-
W w
FIG. 1—Schematic diagram of a flywheel on a shaft indicating the regions of stress concentra-
tion for a single component (welded) and the regions of fretting for an assembly.
WATERHOUSE ON FRETTING IN PRACTICE AND THE LABORATORY 5
the members of a contact area are bolted and riveted joints, flexible cou-
plings such as splines, toothed flange fittings and keyways, the fir-tree or
dovetail fitting of blades to a disk in turbines, the area where steel ropes flex
as they pass over guides or pulleys, particularly those of the locked coil va-
riety, and screw-countersink hole contacts in orthopedic fracture fixation
devices. This list is by no means exhaustive, but it gives some idea of the wide
variety of applications and environments in which fretting may occur, with
the possible effect of considerable reduction in the fatigue strength of the
components concerned.
Vibration
Another common situation in which fretting damage occurs is when con-
tacting surfaces are subjected to vibration. The classic example is the so-
called false brinelling which has occurred in the bearings of automobiles
transported by road from Detroit to the West Coast. Similar problems have
arisen in the roller bearings of gun mountings in military vehicles. Other ex-
amples of fretting occurring during transportation are surface damage to
stacks of mirror-finished rolled aluminum sheet and fretting of aluminum in-
gots transported by rail, which gives rise to serious blistering on subsequent
extrusion [3]. Fretting damage due to vibration has been reported on the
hydraulically operated rams of mobile heavy-duty earth-moving devices and
between the cylinder liners and cylinder blocks of large diesel engines [4].
Fretting also occurs in heat exchangers of the type consisting of packs of em-
bossed stainless steel sheets. The liquid to be cooled passes through alternate
compartments and the cooling liquid through the intervening compartments.
Pumping the liquid gives rise to vibration, which results in fretting at the
points of contact between adjacent sheets. A similar problem is encountered
in the supports or baffles through which condenser tubes pass. A tight fit is
not advisable since latitude is required for thermal expansion. Vibration aris-
ing from fluid flow can cause fretting between the outer surface of the con-
denser tubes and the edges of the holes in the baffle. Fretting problems have
developed in the nuclear power industry because of vibration caused by the
flow of the cooling medium, whether water or a gas such as helium or carbon
dioxide. Fretting has been reported between the canned fuel elements and
the surrounding pressure tubes, between overlapping sheets of the stainless
steel shielding, and in control mechanisms for raising and lowering the fuel
elements or graphite moderator rods [5].
A form of vibration-induced fretting is experienced in electrical contacts,
and considerable research effort is being devoted to minimizing its effects
[6]. Again, this list of occurrences is only meant as a guide to the diversity of
situations in which fretting may be found when a structure is subjected to
vibration.
Deliberately Induced Fretting

Although in most cases fretting is regarded as damaging and something to
be avoided, there are a few applications where it has a beneficial effect.
Energy is consumed in the fretting process, and so it can be employed to
damp unwanted vibration. This is the purpose of the snubbers on turbine
blades and also of the wires sometimes threaded through the outer ends of
such blades [7]. A recent paper advocates frictional damping in joints and
expresses confidence that the resultant deleterious effects of fretting can be
controlled by suitable surface treatments [8]. However, it should be pointed
out that such treatments could have an adverse effect on the damping capac-
ity of the contact.
Simulation of Fratting in the Laboratoiy

The most important mechanical variables which must be controlled in an
investigation into fretting are the amplitude of slip between the surfaces and
the normal load applied to the surfaces. Frequency is of less importance, ex-
cept perhaps in aqueous environments such as saline solutions. The methods
chosen to control and monitor these two variables depend on the aspect of
fretting which is of major interest; that is, is it the removal of material by
fretting wear which is to be investigated or the initiation of fatigue cracks by
fretting and the subsequent effect on fatigue life?
Fretting Wear
Geometry of the Contact—The most frequently encountered geometry in
fretting is the sphere on a flat. If the amplitude of movement, which can be
either rotational or linear, is kept sufficiently small, the condition of
microslip can be achieved; this condition occurs when there is a central, cir-
cular nonslip region surrounded by an annulus where slip occurs according
to the Mindlin analysis [9] and where the fretting damage results. However,
in most reported investigations the amplitude is sufficiently great to result in
slip over the whole of the circular contact area. The disadvantage of this type
of geometry is that as wear occurs the contact area grows, and therefore if a
dead-loading method is used to apply the normal load, the nominal pressure
in the contact zone will be reduced although the actual pressure at the real
areas of contact is constant. However, if the nominal pressure is required to
be kept constant, it should be possible to devise a mechanical system where
the applied load is automatically increased, for example, by a lever system,
as material is removed from the surfaces. This disadvantage also applies to a
cylinder-on-flat geometry or cross cylinders.
The advantage of these geometries is that the wear scar is of a well-defined
shape and amenable to measurement both in area and depth, which can be
achieved by a system of surveying with a profilometer. The other advantage

in the case of a sphere-on-flat, configuration or crossed cyHnder is that align-
ment is not a particular problem.
Another common geometry consists of two flats in contact. Again the
movement can be linear or rotational. In the latter case, a common geometty
is an annulus on a flat. This has the advantage that a range of amplitude is
obtained from the inner to the outer edge of the annulus. The major diffi-
culty with the flat-on-flat configuration is that of ensuring alignment so that
the pressure is evenly distributed over the contacting surfaces. One method is
to have one of the surfaces connected to the applied load by means of a
universal joint such as a ball in a conical seating. The friction in such a joint
must be such that alignment occurs under a light load but no slip occurs
under the operating load. If constraints have to be applied, these may in-
terfere with the alignment. Another method is to mount one of the surfaces
on a flexible metal diaphragm which allows a certain amount of tilting but is
very rigid in the plane of the diaphragm. The load is applied from the other
side of the diaphragm either directly or by means of the universal joint men-
tioned previously. This system has been used successfully in the torsional rig
illustrated in Fig. 2. However, some difficulty was experienced with this ar-
rangement at temperatures above 3(X)°C because of thermal expansion of the
diaphragm, with the result that the ball-in-socket joint became the sole
source of alignment under these conditions.
The mating surface must, of course, remain with its surface in a fixed
plane. If the movement is rotational, the rotation must be carefully main-
tained about an axis normal to the surface. This situation may be achieved by
a type of sleeve bearing or, as in Fig. 2a and b, by vertical suspension vanes
which are able to deform elastically to accommodate the small amplitude of
movement. This deformation is related to the amplitude of movement of the
contacting surfaces and can be measured by applying strain gages.
Method of Actuation—The most common methods of producing a small-
amplitude movement over a wide range of frequency are (a) by mechanical
means, where the movement is produced by an eccentric or cam and may
subsequently be reduced by a lever system; (b) by means of an elec-
tromagnetic vibrator, where the movement is controlled by the power sup-
plied to the vibrator; or (c) with a system of rotating out-of-balance weights.
Whichever method is used, the tangential force applied should be sufficiently
high to overcome frictional resistance in the contact. It must be remembered
that the latter can change considerably in a fretting experiment.
If linkages are used between the actuator and the contact, there is a strong
possibility of looseness occurring in such links or of looseness developing
because of fretting in the links. For this reason a rig has been developed in
the author's laboratory which uses a rigid arm between the electromagnetic
vibrator and the upper of the two fretting surfaces, which is part of a sphere
(Fig. 3) [10]. The electromagnetic vibrator is mounted in a trunnion, and the
.^
SI
I!
1}
Q-CD E tl
i
(J
o
a:
o
COJ•c
U Q
>
FLEXIBLE BELLOWS
PRESSURE
ESSEL
VIBRATOR DRIVE tt
ARM
LOWER
PRESSURE
BALANCING SUSPENSION
SPRING VANES
INNER HEAT
SHIELD
10cm DYNAMIC BALANCING

ARMS
FIG. 2b—Plan of the high-temperature fretting rig in Fig. 2a.
normal load is applied by tilting it and locking it in position. The magnitude

of the load is measured by strain gages on the arm. Other strain gages
monitor the tangential force and hence the coefficient of friction. The arm
must be sufficiently massive to make elastic deformation negligible. To gain
maximum rigidity with as low a weight as possible, the main part of the arm
is tubular. This also allows it to be water cooled in high-temperature ex-
periments, since otherwise the strain gages would be adversely affected.
In some instances a linkage is unavoidable, particularly when the ex-
perimenter wishes to lift the surfaces apart to measure any adhesion that has
developed between the surfaces, which, under certain circumstances, can be
considerable [11]. Figure 4 shows a mechanically driven device with a hinge
incorporated between the sleeve bearing and the rider support. This ap-
paratus was designed to investigate fretting between orthopedic device
materials at low frequencies of about 1 Hz in artificial body fluid.
In devices that produce vibration by the use of rotating out-of-balance
weights, the force is that of inertia, and therefore the amplitude increases
with the speed of rotation and hence the frequency. The author has used such
a machine, which consisted of two horizontal engagmg gear wheels, each
with symmetrically disposed out-of-balance weights. When they are rotated a
vibration is produced in a direction perpendicular to their center line (Fig. 5)
[12].
In choosing the method of actuation, the required frequency is an impor-
tant factor. For low frequencies the mechanical method based on eccentrics
or cams and levers is to be preferred. At frequencies above 10 Hz up to
several hundred hertz, the electromagnetic vibrator is most convenient. The
o
I/)
lU UJrv-
o oq
UJ U - J
Q- C L O
«>
.^
.8
•2*)
I
i
I'
d
i
t
!
•a
f
!
't.
I
I
d
FIG. 5—Schematic diagram of engaging gear wheels {00') with symmetrically arranged out-
of-balance weights (GG') to cause vibration along a line (YY) perpendicular to the center line
(XX).
out-of-balance weight method is best suited to moderate frequencies, that is,

25 to 50 Hz.
Measurement of Amplitude of Slip—Following from the previous remarks
concerning possible looseness in linkages, it is important that the amplitude
of slip is measured as close to the contacting surfaces as possible. This is
often achieved by attaching one plate of a capacitance gage to the end of the
rider arm (Fig. 4). In the apparatus in Fig. 3 the arm is sufficiently rigid to
allow the capacitance gage to be attached to the vibrator end. The apparatus
shown in Fig. 2 utilizes a split laser beam reflected from mirrors on the riser
to the annulus and the edge of the lower specimen acting as optical levers.
The amplitude is determined by passing the beams through a filter of graded
absorptivity.
Continuous monitoring of the amplitude is recommended, particularly
when an electromagnetic vibrator is used, since the signal can be used as a
feedback device to maintain the amplitude constant.
Control of Environment—Since the amplitude of motion at the contact
end of a fretting wear apparatus is very small, there is usually no difficulty in
enclosing that part of the rig in a closed transparent box which can be flushed
out with any desired atmosphere. For high-temperature work the contact can
be surrounded by a furnace of the top-hat variety. For investigations in
aqueous solutions the contact can be immersed in the solution contained in a
suitable cell, as shown in Fig. 4. The cell and any screws for fixing the lower
specimen should be of an inactive material such as Plexiglas or nylon.
Assessment of Damage—One of the more difficult problems in fretting
wear experiments is that of quantitatively determining the damage. The mat-
ter can be further complicated by the transfer of material from one surface to
the other and back again. If one is concerned only about removal of material,
without considering its ultimate state or destination, weight loss
measurements may be applied. The weight of material removed in fretting is
usually many orders of magnitude less than that of the component. One
method used to magnify this has been, first, to assess the area of the damage,
A, and then to machine the surface carefully until all traces of the damage
have been removed. The loss in weight is now related to the depth of the pits,
d, which can be calculated from the dimensions of the surface and the den-
sity of the material. The function A X d is related to the magnitude of the
damage, although its accuracy depends on the damaged areas having a more
or less uniform depth.
If the resultant state of the surfaces and the extent of material transfer is a
relevant consideration, a systematic survey of the surfaces with a pro-
filometer is more informative and, now that this can be controlled and
analyzed with the computer, a relatively simple task.
For visual appraisal of the damage, the scanning electron microscope is
the most useful instrument. If the two surfaces are of different composition,
the transfer of material can be readily identified by use of X-ray energy
dispersive analysis. The presence of cracks, the shape of wear particles about
to detach themselves from the surface, and the general morphology of the
damage, particularly at the boundary of the damaged region, can be studied
over a wide range of magnifications.
Indirect methods of investigating the fretting process are sometimes possi-
ble in particular applications. The author has found electrochemical mea-
surements particularly informative in studying the fretting of corrosion-
resistant materials such as stainless steels and titanium alloys. By using
potentiostatic methods, the increase in corrosion current can be determined
as a function of amplitude or frequency. These measurements have the ad-
vantage of being recorded continuously throughout an experiment.
Another indirect investigation method is to measure the energy consumed
in the fretting action by its damping effect on the imposed vibration. This
method has been used by Fricker to study fretting in the tube supports of
heat exchangers [13],
Fretting Fatigue
In studying fretting fatigue it is usual to work with a fatigue specimen of
conventional pattern and to clamp bridges onto it so that the surface strains
produced by the cyclic stressing result in the required small amplitudes of
slip. Since these strains are small in laboratory-sized fatigue specimens, the
result is that microslip is easily achieved in such experiments.
Geometry of the Contact—The geometry of the contact is dictated by the
type of fatigue specimen. Most investigators have chosen to use either flat
specimens, fabricated from sheet or plate, or cylindrical specimens, turned
from rod or bar, onto which parallel flats are machined. The flat specimens
are most suitable for testing in plane-bending or push-pull machines. The
cylindrical specimens can also be tested in push-pull machines but may be
used in rotating-bending machines as well. Fretting is applied to the
specimen by clamping onto the flats a pair of bridges with a proving ring,
which enables the normal load to be adjusted. The bridges must be suffi-
ciently rigid so that, as the specimen is stressed, movement occurs between
the feet of the bridges and the surface of the specimen.
The geometry of the contact at the feet of bridges is usually either flat, flat
with a chamfer, or cylindrical.
Control of Amplitude of Slip—Since the amplitude of slip depends on the
surface strains, which in turn depend on the applied cyclic stress, it is evident
that when a fatigue curve is determined using a given size of specimen and
size of clampmg bridges, the amplitude of slip will vary as the stress is varied.
It may appear that this is a serious disadvantage, but this is what actually
happens in practice. To vary the amplitude independently of the changes in
stress, bridges with different distances between the feet can be used [14],
Nishioka and Hirakawa devised a mechanical system of levers to enable the
amplitude of slip to be maintained constant as the applied cyclic stress was
varied [15].
The amplitude of slip is usually calculated from consideration of the elastic
deformation at the surface of the specimens.
Control of the Environment—With fatigue machines in which the
specimen is more or less stationary, such as push-pull or plane-bending
machines, it is relatively simple to surround the specimen and bridges with a
cell for containing a gaseous or liquid environment or a furnace. It is ad-
visable to have the proving ring which applies the clamping force to the
bridges outside the enclosure. This involves the use of long prongs which
pass through the cell or furnace wall. These should be of the same material as
the specimen if an electrolyte is being used, or of creep-resistant material for
high-temperature work. An example of a furnace used with a push-pull fatigue
machine is shown in Fig. 6. In studies in an aqueous electrolyte, it is, of
course, possible to make electrochemical measurements or, more important,
to control the potential of the surface and thereby to investigate the effects of
cathodic or anodic protection on the fretting fatigue process.
Assessment of Damage—Since the object of fretting fatigue experiments is
to investigate the fretting effect on fatigue life, it is usual to determine a
fatigue curve (applied alternating stress versus the log number of cycles to
failure) for the specimens with and without fretting. It may be argued that
the stresses induced by the clamped bridges will have an effect on the fatigue
life, but again this situation applies in practical cases of fretting fatigue. Ex-
periments carried out where the fretting damage is applied to fatigue
specimens before fatigue testing usually show a smaller effect than when the
fretting and fatigue occur simultaneously. The latter occurrence is usually at-
ADJUSTING SCREW PROVING RING
• RECORDING THERMOCOUPLE CONTROL

THERMOCOUPLE
SPECIMEN
FIG. 6—A furnace suitable for investigating fretting fatigue at high temperatures with a
push-pull fatigue machine.
tributed to a fonn of conjoint action similar to that which is experienced

when fatigue tests are carried out in the presence of a corrosive atmosphere.
The reduction in fatigue strength produced by fretting is a useful quan-
titative assessment of the damage. However, the specimens are available for
visual or profilometer examination, as in the case of fretting wear specimens.
Conclnsion
The practical situations where fretting may be experienced are so varied
with respect to the contacting materials, their geometries, the amplitude and
nature of the movement, and the character of the environment that the in-
vestigator, who is usually more interested in palliatives for a particular prob-
lem, must decide which of the apparatuses or methods discussed in this
paper wUl approximate the practical case most closely. At times it may be ad-
visable to conduct experiments with the actual contact which is creating the
problem, or where the scale does not allow it, such as with a massive
flywheel, a model may be the next best thing. However, the author has fdund
that controlling important variables is often difficult in such situations, and
recourse has to be made to the simpler laboratory apparatuses.
Refeiences
[7] Tomlinson, G. A., Proceedings of the Royal Society (London), Vol. USA, 1927, pp.
472-483.
[2] Horger, O. J. in Symposium on Fretting Corrosion, ASTM STP144, 1953, pp. 40-5L
[3] Waterhouse, R. B., Proceedings, Sixth European Congress on Metallic Corrosion, Society
of Chemical Industry, London, Sept, 1977, pp. 615-622.
[4] Waterhouse, R. B., Wear, Vol. 34, No. 3, Sept. 1975, pp. 301-309.
[5] Gadda, F. and Ronchetti, C , Energia Nucleare, Vol. 24, No. 6, June 1977, pp. 297-305.
[6] Antler, M. and Drozdowicz, M. H., Wear of Materials 1981, New York, American Society
of Mechanical Engineers, 1981, pp. 259-270.
[7] Johnson, R. L. and Swikert, M. A., "Friction and Wear Under Fretting Conditions of
Materials for Use as Wire Friction Dampers of Compressor Blade Vibration," NASA-TN-
D-4630, National Aeronautics and Space Administration, Lewis Research Center,
Cleveland, Ohio, July 1968.
[8] Beards, C. F., Journal of the Society of Environmental Engineers, Vol. 19-2, No. 85, June
1980, pp. 23-27,
[9] Mindlin, R. D., Journal of Applied Mechanics, Vol. 16, No. 3, Sept. 1949, pp. 259-268.
[10] Hamdy, M. M., Overs, M. P., and Waterhouse, R. B., Journal of Physics E: Scientific In-
struments, Vol. 14, No. 6, June 1981, pp. 889-894.
[;;] Bethune, B. and Waterhouse, R. B., Wear, Vol. 12, 1968, pp. 289-296.
[12] Goto, S. and Waterhouse, R. B., "Titanium '80, Science and Technology," Proceedings,
Fourth International Conference on Titanium, Kyoto, Japan, 19-22 May 1980, H. Kimura
O. Izumi, Eds., Metallurgical Society of the American Institute of Metallurgical and
Petroleum Engineers, 1980, Vol. 3, pp. 1837-1844.
[13] Fricker, H. W., Proceedings, Conference on Component Design in High Temperature
Reactors Using Helium as a Coolant, 3-4 May 1972, Nuclear Engineering Group, Institu-
tion of Mechanical Engineers, pp. 213-233.
[14] Gaul, D. J. and Duquette, D. J., Metallurgical Transactions, Vol. IIA, No. 9, Sept. 1980,
pp. 1555-1561.
[15] Nishioka, K. and Hirakawa, K., Bulletin of the Japanese Society of Mechanical Engineers,
Vol. 12, No. 50, 1969, pp. 180-187.
D. Kusner, ^ C. Poon, ^ and D. W. Hoeppner^
A New Machine for Studying

Surface Dannage Due to Wear
and Fretting
REFERENCE: Kusner, D., Poon, C , and Hoeppner, D. W., "A New Machine for
Studying Surface Damage Due to Wear and Fretting," Materials Evaluation Under Fret-
ting Conditions, ASTM STP 780, American Society for Testing and Materials, 1982, pp.
17-29.
ABSTRACT: A test machine was designed and constructed for investigating the effects
on wear rate of various parameters such as contact load, relative slip amplitude, and fre-
quency. The test machine consisted of a rigid load frame and a flexible platform onto
which specimen coupons were mounted. Vertical wear pads were held stationary in
holders that allowed contact between the surfaces of the wear pad and the specimen
coupon. Normal contact pressures between the wear pads and the specimen were ob-
tained by using spring-loaded cap screws. The specimen platform was designed to move
horizontally at high frequency (100 to 200 Hz) with the amplitude of motion under con-
trol. The new test machine allows up to eight specimens to be tested at a time and is ex-
tremely useful for statistical investigation of wear behavior in which a large number of
specimens is expected to be tested. Tests were performed to study the wear behavior of
plain magnesium and magnesium with different coatings against steel. Wear rate curves
were presented which were used to compare the performance of the different coatings
against wear.
This paper discusses the operation and typical results of this unit.
KEY WORDS: fretting, fretting wear, delamination, fretting
The construction of turboprop engine gearboxes often necessitates the use

of steel-magnesium interfaces. Because of manufacturing tolerances and
vibration loads within engines, a condition for fretting wear often exists. A
program was undertaken to evaluate the various fretting wear parameters as
applied to this magnesium-steel interface. In addition a capability was
desired for evaluation of coatings for the prevention of fretting wear.
'Graduate student, professional engineering officer, and Cockbum professor and director.
Structural Integrity Laboratories, Department of Mechanical Engineering, University of To-
ronto, Toronto, Canada M5S 1A4.
17
Copyright® 1982 b y AS FM International www.astm.org
(PDVSA Los Teques) pursuant to License Agreement. No fu
The wear research program was divided into two major tasks. Task 1 in-
volved the design of a suitable wear machine, while Task 2 included the test
program. A review of the dominant parameters associated with wear was
made. The number of parameters which might influence the wear rate at the
magnesium-steel interface was reduced to three: the contact load, relative
slip amplitude, and frequency. An analysis of engine operating conditions
highlighted the possible ranges for each of these parameters. A wear machine
was then designed and built which had the capability of runnmg under these
conditions. A test matrix was then laid out which included the effect of three
surface conditions as well as the three aforementioned parameters.
Test Rig
A study of possible wear machine configurations indicated that because of
the high frequency required in the test program, a closed-loop elec-
trohydraulic system would be best. Because of the statistical nature of wear
testing, a capability of testing up to eight distinctive specimen coupons was
included.
As seen in Fig. 1, the actuator is connected directly to a platform to which
the eight magnesium specimens are attached. (The cross section in Fig. 1
passes through two of the eight test stations.) The small slip amplitude al-
lowed the use of flex plates to support the oscillating platform. These flex
plates are able to resist all moments, the vertical force, and lateral motion
and yet are very flexible in the slip direction. They have the advantage over
linear bearings of not needing any lubrication, and they are immune to the
wear particles produced during the test. In addition, with strain gages
mounted on the flex plates, the movement of the platform may be monitored
after suitable calibration.
The vertical steel specimens against which the magnesium specimens wear
are attached to eight individual specimen holders. These holders are, in turn,
connected to load cells that allow the normal loads to be monitored. The
loads are applied by means of springs that are compressed by cap screws.
The load cells were previously calibrated to 1500 N or 210 MPa at the wear
surface. The springs serve two purposes. They provide a resilient load train
so that as the specimens wear they do not unload themselves; in addition the
adjustment of the load is made finer by the screws so that the loads may be
set accurately. Movement of the upper load trains is constrained by spherical
bearings at the top and by flex bars at the bottom. The specimen holders
were designed so that all specimens can be removed without disassembling
any parts of the wear machine. This saves operator time and also maintains
the alignment of the various components.
A schematic drawing of the electronic and hydraulic layout is shown in
Fig. 2. The eight load cells are monitored by means of a multichannel bridge
KUSNER ET AL ON MACHINE FOR WEAR DAMAGE STUDY 19
zm
MICROMETER
HEAD
LOAD SCREWS
LOAD SPRING
SPHERICAL
BEARING
LOAD CELL
OSCILLATING
PLATFORM
UPPER SPECIMEN
HOLDER
STEEL
SPECIMEN
MAGNESIUM
SPECIMEN
FLEX-PLATES
0 ACTUATOR
STRAIN GAGE
LOCATION
FIXED BASE
91 2 3 4 5 CM.
FIG. 1—Cross section of the test rig.
amplifier. The actuator is controlled by an MTS 436 controller. A d-c signal

conditioner on the controller is used for the "slip gages." A function
generator is used for the program signal and a digital counter for the cycle
count. The wear machine has the capability of running all eight specimens
for any combination of the designed operating parameter values. The system
characteristics, however, allow the values of these parameters to be extended.
Higher frequencies may be run at the expense of either the normal load or the
slip amplitude. Conversely, higher slip amplitudes, up to about 500 /xm, may
be run at the expense of higher loads or higher frequencies. The load cells are
limited to a maximum load capacity of 1500 N. The wear machine has now
accumulated over 3000 h with no major problems.
Z uj
< K
I- < c O
UJ "•'
I
"a
UJ —
zt ^ a: s
ii oo
99 u
O UJ
-UJ
CD V
I
a:
UJ
>o
.cb
UJ
CC IT
Ui 1 -
(0 2
o
o
I
O
Q
< 0
O
•Dt]
Oo
oc
Q.
s
5
t
The specimen design was determined by a number of factors. It was de-

cided to go with the present configuration to reduce machining costs. While
this complicates the mounting fixtures, it reduces the machining time to a
minimum. The size of the magnesium specimen allows up to four tests to be
performed per specimen. The dimensions of the steel pin and the magnesium
specimen are shown in Fig. 3.
Method of Wear Analysis

It was originally proposed to measure the rate of wear by weight loss.
Based on the available laboratory balances and the specimen dimensions, it
was calculated that the resolution of the wear depth was about 8 /xm for the
magnesium and 2 ^m for the steel. However, in practice this method of wear
measurement was found to be insufficient. This was due to a number of fac-
tors: material transfer between the specimens and holders, chipped-off
coatings, and attached wear debris. An alternative method was to use a
Talysurf profilometer for depths of less than 50 /^m and a dial gage for
0.100 ± .002
NORMAL
STEEL (AMS6265) LOAD
SPECIMEN
\
0.930 f .002
SLIP
DIRECTION
MAGNESIUM
(AMS4439) s.
0.100 ±.005
FIG. 3—The specimen dimensions, given in inches (1 in. = 25.4 mm).

greater depths. Multiple readings or passes could be made and the average
volume calculated. Prior to taking the readings, a light blast of compressed
air could be used to clear lightly attached debris. (The definition of what con-
stitutes debris is sometimes arbitrary because of material transfer and loosely
attached debris particles.)
Making measurements on the basis of an end point is only useful if one
may assume a linear wear rate. This may not be the case for nonho-
mogeneous materials such as coated specimens. Multiple readings may be
taken if one is willing to return the specimen at the end of each reading.
However, severe transient effects would probably be introduced. In order to
overcome this obstacle, it was decided to measure the displacement of the
steel pin continuously as it wore into the magnesium during a test. Since the
steel specimen showed no signs of measurable wear, one may assume that the
wear is entirely of the magnesium specimen. (The steel specimen actually had
a negative wear rate because of material transfer.) Micrometre heads were at-
tached to the wear rig for this purpose. They allowed a resolution of approx-
imately 3fim.
A typical wear plot is shown in Fig. 4. There are six specimens in all: two
were coated with a polymer-type coating, ^ two had a hard surface treat-
ment,'' and two were left plain. The polymer coating was applied as a baked-
on paint. It consists of polymide binders and fluorocarbons. The coating
functions by reducing the coefficient of friction. The hard coating was ob-
tained by an anodizing process which improves the hardness of the
magnesium surface. The porous anodized surface is subsequently im-
pregnated with molybdenum disulfide to reduce the coefficient of friction.
Both coatings are approximately 20 fim thick.
Test Results
A total of 96 tests, with 16 runs of six specimen pairs, have been made so
far. An initial worst-case matrix indicated that frequency had little effect on
a per cycle basis over the range tested (80 to 290 Hz). The effect of slip
amplitude and normal load had quite unexpected results. Figures 5 and 6
show the wear rates obtained by performing a linear regression on the linear
portions of the wear data. These are the sections of the wear depth plots
where the steel pin is fretting into the magnesium substrate. The effect of slip
amplitude shows a definite nonlinear relationship. Waterhouse [1]^ and
Halliday and Hirst [2] have reported similar results with "mild steel" and
AISI 9310 steel. It was noted during the course of the tests that with a slip
amplitude of 50 jum a fine white oxide debris was produced in contrast to a
^The italic numbers in brackets refer to the list of references appended to this paper.
^Technical literature available from Whiteford Corp., manufacturers of Xylan.
""Technical literature from General Magnaplate, llinden, N.I., patent owners of Magnadize
application.
250
T«it CondiMon»
Fraquancy 100 Hz.
Slip AmpL 200 ^ m
Norm Lead 35 MNAn^
02 04
CYCLES (MILLIONS)
FIG. 4—The wear depth plots from a typical run of six specimens.
Xlb'
1 1 1
5— NORMAL LOAD
FREQUENCY
3S MN/m^
100 Hz
A ~
—
\ '
s.
1= — / —
^X, 100
SUP AMPLITUDE
200
()lin)
1 __
300
FIG. 5—The effect of slip amplitude on the wear rate.
larger dark particle for slip amplitudes of 100 /im or greater. This seems to
indicate that a different wear mechanism occurs for slip amplitudes of
greater than 50 /^m.
The effect of normal load showed an intriguing result. One normally ex-
xio
2.5
2.0 —
1 1.5
SLIP 200Mni
•T-
A
K -12
< XIO
1.5
1.0
SLIP SOiim
0.5
0 50 100 150
FIG. (>~The effect of normal load on the wear rate.
pects that an increase in load would result in a proportional increase in the

wear rate. The wear data for slip amplitudes for 200 and 50 pim indicate that,
for the range tested, a doubling or even a tripling of the normal load results
in very little increase in the wear rate. A possible explanation is that the
greater load makes it more difficult for wear particles to leave the contact
zone.
The effect of the coatings was determined by extending the linear portion
of the wear depth plots beyond the thickness of the coating to the abscissas.
The mtersection of this line with the abscissas then gives some idea of the ef-
fect of the coating in terms of the life extension in cycles. This method is
based on the observation for the coatings tested that there does not appear to
be any residual effect of the coating on the subsequent wear of the
magnesium substrate.
It was found using this method that the polymer coating had no discernible
effect and appeared to wear at the same rate as the magnesium. Its only
usefulness would be as a sacrificial coating. The hard coating had a definite
beneficial effect. The life extension was quite considerable, although there
was a large amount of scatter. As was the case for the wear rate of the
magnesium substrate, the life extension was found to be dependent on the
test conditions (see Table 1).
TABLE 1—Life extension of hard coated specimens.
Life Extension, millions of cycles

Normal Load,
MPa Slip, ^m x' s"
35 50 n.A 29.1
35 100 0.75 0.20
35 150 0.48 0.16
35 200 0.38 0.25
35 300 0.15 0.04
"Mean life of coating.

*Standard deviation.
Observations
On removal of the wear specimens from the test rig, it was always im-
mediately obvious that extensive surface and subsurface damage was occur-
ring. Often, because of "cold welding," a reasonable amount of force was
needed to separate the steel specimen from the magnesium. Material transfer
was observed from the magnesium to the steel pin in amounts ranging from
10 to 35 percent of the steel pin surface area. The thickness ranged up to
about 200 ^m.
The wear debris which was steadily produced during the tests was black for
the slip amplitudes of 100 fira and greater and consisted of a very fine white
powder for the slip amplitude of 50 nm. Metallic magnesium particles were
observed in the black oxide debris in increasing proportions for greater slip
amplitude. Because of the test geometry there was a large amount of wear
debris present in the contact zone.
Figure 7 shows a normal view of the magnesium wear scar for a slip condi-
tion of 50 /xm. In the central region is a large shallow pit where the
magnesium has transferred to the steel specimen when it was removed.
Toward the edges of the pit area are long thin lamina which at one end re-
main attached to the magnesium, as shown in the enlargement, and at the
other end were probably attached to the steel pin. The formation of these
lamina is indicative of subsurface crack growth. The large size of these
lamina in comparison with the actual debris powder extruded from the edge
of the contact zone would seem to indicate that the conforming test geometry
has a large effect on the breakup and removal of wear debris.
Duquette [3] has reported the observation of similar platelike morphology
for AISI 4130 steel fretting under similar conditions. He also observed the
generation of a very fine "metal powder." He attributes the formation of this
powder to the breakup of flakes into smaller brittle flakes, which are then
reduced to a powder under the action of continued faying.
1.0 m m t
. 0.1 mm.
FIG. 7—A normal view of the wear scar, showing a pit in the central region. The enlargement
at the bottom reveals long, narrow lamina platelets. Test conditions: frequency, 100 Hz; load,
34.5 MPa; slip, SO iim.
Cross sections of the wear scar indicate that wear particles are generated in
the central third of the contact region by delamination and adhesive wear.
The theory of delamination has been well expounded by Suh et al [4]. As this
wear debris migrates toward the edges of the contact region, it causes
abrasive wear and the generation of more wear particles. Although for the
smallest slip, 50 /tm, delamination was noticed toward the edges of the con-
tact region and not just in the central third region.
Figures 8 through 10 show examples of subsurface cracking of the
magnesium specimen. It was observed on several specimens that the cracks
nucleated below and propagated parallel to the surface. In some regions a
number of parallel subsurfaces cracks were observed. (It should be pointed
out that these cracks are not related to fretting fatigue. In fretting fatigue it is
generally accepted that cracks are initiated at or near the contact zone and
then propagate as a result of an alternating stress in the base material, usu-
ally in a du:ection perpendicular to the surface.)
From these observations (macro and micro), it is apparent that one of the
rate-limiting processes for this type of wear is the disposition of the wear par-
ticles. The decrease in the size of the particles as the slip amplitude was
decreased was probably due to the greater amount of time the particles spend
between the faying surfaces.
L
50 iJin J
FIG. 8—A cross-sectional view, showing the subsurface crack growth and the formation of
wear platelets. The geometric constraint prevented these particles from being removed. Test
conditions: frequeruy. 100 Hz: load. 34.5 MPa: slip, 50 fim.
». t
, 50 j j i n
FIG. 9—A cross section showing large amounts of plastic deformation and subsurface crack
growth. Test conditions: frequency, 100 Hz; load, 34.5 MPa: slip, 50 /im.
, 50MIW ,
FIG. 10—An optical photomicrographic cross section showing multiple parallel crack growth.
Test conditions: frequency, 100 Hz: load, 34.5 MPa: slip, 50 nm.
Conclusions
A closed-loop electrohydraulic machine has been designed and built. The
machine is capable of testing up to eight specimen coupons simultaneously.
A wear-monitoring technique was developed in order to evaluate the effec-
tiveness of various wear coatings. The initial results indicate that the hard
coating tested was beneficial, while the soft polymer coating had little effect
on life. The increase in life resulting from the hard coating, however, shows a
wide scatter in values.
The effect of slip amplitude, frequency of oscillation, and normal load
were also studied. For the range of frequencies studied, it appears that the
frequency has no effect on a per cycle basis. The effect of slip amplitude,
however, is great and bears a distinct nonlinear relationship for the values
tested. For the values studied, the wear rate shows no correlation with the
normal loads tested.
The fretting wear mechanism under steady-state conditions appears to in-
volve adhesive wear, delamination, and abrasive wear. Because of the entrap-
ment of wear particles, the conforming test geometry affects the wear rate.
Acknowledgments
The authors are indebted to the following for financial support: the
Wallace G. Chalmers Fellowship Fund, Department of Mechanical
Engineering, University of Toronto; the Natural Sciences and Engineering
Research Council-Canada; the Department of Energy, Mines, and
Resources-Canada; and Pratt and Whitney Aircraft of Canada Ltd.
We are also grateful to Cliff Brownridge, John Harrop, and R. Marks for
stimulating discussion and interaction.
References
[1] Waterhouse, R. B., Fretting Corrosion, Pergamon, New York, 1972.
12] Halliday, J. S. and Hirst, W., "The Fretting Corrosion of Mild Steel," Proceedings of the
Royal Society of London, Vol. A-236, 1956, pp. 411-425.
[3] Duquette, D. J., "The Role of Cyclic Wear (Fretting) in Fatigue Crack Nucleation in
Steels," Research Report, Materials Research Center, Rensselaer Polytechnic Institute,
Troy, N.Y.
[4] Suh, N. P., Wear, Vol. 44, 1977, pp. 1-16.
p. Kennedy,^ M, B. Peterson,^ andL. Stallings^
An Evaluation of Fretting at
Small Slip Amplitudes
REFERENCE: Kennedy, P., Peterson, M. B., and Stallings, L., "An Evaluation of Fret-
ting at Small Slip AmpUtudes," Materials Evaluation Under Fretting Conditions, ASTM
STP 780, American Society for Testing and Materials, 1982, pp. 30-48.
ABSTRACT; Fretting wear, produced by the twisting motion of a flat steel specimen
against a fixed 12.7-mm steel ball, has been investigated in the limited microslip region
ranging from 0.05 to 5.0 /»m. The construction of the test rig and the type of fretting wear
associated with this particular specimen configuration and motion are discussed. This
discussion is exemplified by data obtained either with SAE 52100 steel or Hastelloy B
alloy ball specimens in combination with SAE 1018 steel flat specimens. The first indica-
tion of metal oxidation normally associated with the onset of fretting occurred at a slip
amplitude of about 0.075 ^m. Severe damage was noted at 2.5 fim.
KEY WORDS: fretting, wear, wear tests, fretting wear tests, friction, metal, carbon
steel, microslip
Fretting, particularly in its relationship to reduced fatigue strength and in-

creased wear, constitutes a serious problem in both fixed and rotary-winged
aircraft. The list of aircraft components subjected to fretting damage is
lengthy and includes parts such as splines, cables, bearings, hinges, seals,
and actuating devices. The list of proposed solutions for fretting problems is
also extensive and involves techniques such as the use of special surface
finishes, shims, adhesives, lubricants, coatings, and, if necessary, design
changes to control fretting. To find the correct solution to a given fretting
problem, one usually relies fairly heavily on the trial-and-error approach.
This is primarily a result of insufficiencies in the understanding of the fret-
ting process. This work was undertaken in order to clarify the effect of sev-
eral operational parameters on the fretting mechanism, which would in turn
facilitate the selection of appropriate countermeasures against fretting.
' Chemists, Naval Air Development Center, Warminster, Pa. 18974.

^Physicist, Wear Science Corp., Arnold, Md. 21012.
30
Copyright® 1982 b y A S T M International www.astm.org
KENNEDY ET AL ON FRETTING AT SMALL SLIP AMPLITUDES 31
It has been suggested that fretting is dependent on the simultaneous or

subsequential occurrence of two or more different wear processes, such as
adhesion, oxidation, fatigue, and abrasion [1].^ The possibility of different
wear mechanisms operating under different conditions could help to explain
why a certain fretting countermeasure might be effective under one set of
conditions but detrimental under a different set of conditions [1,2]. This
would also explain the relationship of wear rate to slip amplitude. For exam-
ple, several investigators who had studied the effect of slip amplitude on fret-
ting wear in the 10 to 1000-fim range reported the existence of characteristic
slip amplitudes, usually somewhere between 30 and 70 ^m, at which the fret-
ting wear rate may increase by several orders of magnitude [3-5]. Below this
critical amplitude, wear rates were found to be small and normally difficult
to measure. This study is primarily concerned with the 0 to 10-;im range,
where surface damage rather than wear occurs. Surface damage is important
since it is associated with the initiation of the fatigue process. Another area of
interest is the mechanical differences in the fretting process at small and
large slip amplitudes. If significant differences were found, it would follow
that different material and lubricant solutions might apply at different slip
amplitudes or, conversely, that an understanding of this effect could possibly
lead to the development of lubricant solutions dependent on the amplitude of
microslip.
Slip amplitude is an excellent investigational parameter. In addition to be-
ing a sensitive indicator of changes in the fretting wear process, this type of
data can also serve as an aid in the development of design criteria for compo-
nent parts. For example, data showing the relationship between the ampli-
tude of microslip and the amount of surface damage could be either applied
in the design of components for applications where microslip would be
limited to certain amplitudes or used to specify the maximum amount of
microslip that could be tolerated in a given application. Furthermore, where
fretting is a problem, the amplitude could be measured and, with the rela-
tionship of amplitude to wear known, an appropriate solution proposed.
Other factors, such as the relative hardness of the rubbing surfaces, were also
investigated to some extent, but the primary aim of this work was to investi-
gate the effect of slip amplitudes in the 0 to 5-/im range on fretting damage.
Procedure
Test Apparatus
A test apparatus previously used in pivot fretting studies [6] was adapted
for this investigation. A schematic drawing of the test rig is shown in Fig. 1.
This apparatus was mounted on a heavy steel base plate in order to eliminate
AIR-
PRESSURE
BASE PUTE
FIG. 1—Schematic drawing of a fretting test rig.
extraneous vibrations. This test rig provides for the mounting of a ball
against a flat test specimen. For the sake of clarity, certain components such
as labyrinth seals and a special holder for the flat specimen were omitted
from this sketch. These components, along with a typical ball and flat
specimen, are shown in Fig. 2. In this test rig, the ball is attached to the
metal shaft in a special labyrinth seal arrangement, and the flat is attached
to the gas bearing using the holder shown in Fig. 2.
The metal shaft shown in Fig. 1 is held in place by means of two thin metal
diaphragms. This shaft is rigidly attached to a metal diaphragm immediately
below the bellows-type load cell. The lower metal diaphragm serves mostly as
a guide for the shaft and also to prevent lateral movement. Application of air
pressure to the load cell results in a downward flexing of the upper metal
diaphragm, which in turn produces a downward movement of the ball.
Under load conditions, the ball is held rigidly stationary against the flat
specimen.
The flat specimen in Fig. 1 is rigidly attached to the gas bearing. This gas
bearing is essentially a metal hemisphere. A top view of the gas bearing with
the flat specimen attached is shown in Fig. 2. This gas bearing fits into a
spherical seat in the base plate that conforms to the outer geometry of the gas
.^
I
•a
bearing. In operation, the gas bearing floats on a thin film of air at 35 per-
cent relative humidity, which separates the bearing from its spherical seat.
This floating or lifting of the gas bearing is achieved by forcing air through
small orifices located on the surface of the spherical seat. Application of air
pressure to the base of this rig results in an upward movement of the flat
specimen. Loading of the ball against the flat specimen is achieved through
balancing the air pressure applied to the load cell against the air pressure
supplied to the gas bearing so that the amount of gas bearing lift will remain
at some preselected value.
The primary type of motion studied was a twisting or oscillatory rotational
movement. Controlled movement of the flat specimen was achieved through
two electromagnetic drive units attached to the base of the gas bearing. A top
view of the gas bearing and the arrangement of the two electromagnetic drive
units is shown in Fig. 3. These drive units are operated in phase in order to
achieve the desired twisting motion about the ball-flat pivot point. Operation
of the drive units in an out-of-phase relationship will produce a pitch or rock-
ing motion. These drive units consist of two low internal mass 75-W audio
speakers modified by fastening an aluminum cone and steel push rod assem-
bly onto the surface of the speaker cone. The amplitude produced by these
electromagnetic drive units at the point of attachment of the push rod to the
gas bearing base plate is measured by a 250-jum capacitance probe. The am-
plitudes at this point range from 2.5 to 150 jum. Because of geometric consid-
erations, the actual amount of slip amplitude in the ball-flat contact area
would be much less.
Wear Scar Analysis

An analysis of a typical wear scar is presented in Fig. 4. This analysis con-
siders a ball resting on a rigid plate. The ball, in this case, is subjected to a
normal load, W, which would determine the radius of contact, r, and a tan-
gential or twisting force acting normal to the load or in the plane of the flat
plate. A value for the maximum pressure or normal stress, TN, max, can be
calculated from the hertz equations. These equations also indicate that the
normal stress distribution would be at a maximum at the pivot point and
would approach zero at the contact edge. A graphical representation of a
possible normal stress distribution along the contact area is shown. The
shear stress, TJ, also graphically represented, would be zero in the center or
pivot point and would approach a maximum value at the edge.
A sketch of a typical wear scar is shown at the bottom of Fig. 4. No wear
damage can occur in the central unshaded region. This region is normally de-
fined as a locked region. Microslip and surface damage is limited to the outer
shaded region, which can be called the slipped region. This type of wear pat-
tern is a result of the distribution of shear stress relative to normal stress. In
order to have slippage or relative motion of the ball against the flat, a neces-
I
C
^
TWIST ( F )
/OR
SHEER STRESS ( T j )
STATIC COEF.
OF FRICTION
SLIPPED REGION
FIG. 4—Ball-on-flat configuration.
sary condition would be that the surface shear stress exceed the frictional re-
sistance; that is, the ratio of the two stresses would be T^/T^ > f^ (where/s is
the static coefficient of friction for the material pair). At the center or pivot
point, the ratio of shear to normal stress would be at a minimum. Proceeding
outward from the center along the radius, r, this ratio would be expected to
increase until, at some point, this ratio would equal the static coefficient of
friction for the pair. This point would define the boundary between the
locked and slipped regions. Prior to this point or when rs /TN < /s, the two
surfaces would be locked together and would only be able to undergo elastic
deformations. Beyond this point where rs /TN > /s, microslip will occur; if a
load is applied to the contact area and a torque is also applied, slip will first
occur at the outer region since the shear stress is high and the pressure is ex-
tremely small. Thus, slip must result from an applied twisting movement no
matter how small. Slip would start at the outer edge of the contact and pro-
gress radially inward as the amplitude is increased. At some point where the
input torque F > fsW, slip will occur over the entire region of contact.
The assumption of a sharp line of demarcation between the locked and
slipped regions may not be entirely justified. This theoretical wear scar can
be looked upon as containing three distinct wear regimes. The first region
would be characterized by elastic deformation without slip and would corre-
spond to the locked region. The second region would be characterized by the
occurrence of microslip, but the amplitude of this microslip would not be suf-
ficient to produce any measurable surface damage. The last region would be
characterized by microslip of sufficient amplitude to produce measurable
surface damage. Other factors, such as material hardness and surface irregu-
larities, should also be considered in interpreting the significance of this type
of wear scar.
Calculation of Slip Amplitude

In Fig. 3, it can be seen that the in-phase operation of the electromagnetic
drive unit produces a twisting motion about the center of the flat specimen.
Motion about this center is indicated since the point of contact of the ball
with the flat specimen acts as a pivotal point. In this figure, input amplitude
is represented by a double-headed arrow, and the maximum displacement of
the base of the gas bearing is represented by the dashed Une. The relative mo-
tion that occurs on the perimeter of the flat specimen is represented by a
curved arrow. It can be seen that the amplitude at this point on the flat speci-
men can be determined from the amplitude measured by the capacitance
probe and the length of the gas bearing arm. This length represents the dis-
tance between that point at which the push rod is connected to the base of the
gas bearing and the pivot axis. Since the area of interest is the actual contact
area between the ball and flat, as shown in Fig. 4, it should be possible to
determine or measure the radius of this area. For the case of a 12.7-mm ball
under an 88-N load resting against a flat plate, the radius of the associated
contact area was found to be 0.16 mm. Taking this value, together with a
value of 7.9 mm for the length of the gas bearing arm, it can be shown that
the amount of microslip occurring at the perimeter of the contact area would
be equal to 0.02 times the input amplitude measured at the capacitance
probe.
Slip amplitudes calculated by this procedure also represent an average
value for the amount of microslip occurring on the specimen when there is
wear and damage. For example, in a typical case of severe wear, measure-
ments on the specimen indicated that the radius of the locked region was only
0.08 mm, and the radius of the wear scar was approximately 0.23 mm. It can
be seen that the average slip amplitude of 0.16 mm, calculated on the basis of
the wear scar diameter, still represents a good estimate of average microslip.
Thus, the slip amplitude used in these experiments is that which occurs at
r — a where a is a load radius.
Materials and Operating Procedures

The flat specimens used in this study were machined from SAE 1018 steel
rod stock. Specimens were received with a 0.2-/im ground surface finish.
Some of these specimens were later hand polished to a mirror finish using 0.3
and 0.05-/xm aluminum polishing wheels. Testing was done on both types of
surface finishes. The ball specimens were made from either SAE 52100 steel
or Hastelloy B alloy.'' The SAE 52100 steel specimens were standard
12.7-mm grade 25 chrome balls. Prior to use, the specimens were cleaned by
washing in Stoddard solvent and rinsing in petroleum ether.
The arrangement of the ball and flat specimens in the test rig is shown in
Fig. 1. Prior to the attachment of air lines, the clearance between the two
specimens was on the order of 0.8 mm. Since the gas bearing is usually oper-
ated at a lift of 0.025 mm, loading of the ball against the flat was achieved
through increased pressure in the load cell. After balancing the air pressure
supplied to the base plate against that supplied to the load cell, a twisting
type of motion was imparted to the flat specimen by means of the electromag-
netic drive units. In testing, the ball was stationary and motion was limited to
the flat specimen. The amplitude of microslip was controlled by an oscillator
and amplifier attached to the electromagnetic drive units. Testing was nor-
mally done for 3 h. After testing, specimens were microscopically examined,
and some of these specimens were later subjected to surface profile measure-
ments and microprobe analysis.
Results and Discussion
SAE 52100-SAE 1018 Steel Combination

Typical wear scars, obtained at various fretting amplitudes, are shown in
Fig. 5. In this part of the investigation, the test specimen configuration con-
sisted of a 12.7-mm-diameter ball against a highly polished SAE 1018 steel
flat specimen. Fretting experiments were run at an 88-N load and 210 Hz for
3 h. The photomicrographs in Fig. 5 depict the wear scars on the surface of
the flat specimens. The corresponding wear scars formed on the balls were
identical and are not shown. The photomicrographs in Fig. 5 were taken at
X80 magnification since these wear scars typically have diameters ranging
from 0.3 to 0.5 mm. Average microslip amplitudes are reported under each
wear scar.
Visual inspection of the photomicrographs in Fig. 5 indicates that a defi-
''Registered trademark of Cabot Corp.
1
.o
••5.
c
2P
¥(S^'
2>
=0
M •*•/'
5
e4
O
^'h^0^ l^.^-'---^^^^. ••'^••^-- • • i ' ' ^ ' -
i
nite, although minimal, amount of fretting damage can occur at slip ampli-
tudes as low as 0.1 urn. The wear scar diameter under these conditions was
the same as that predicted using the hertz equations. Two tests made on a
polished and an unpolished specimen at an amplitude of 0.05 /tm did not
suggest any trace of surface damage. Photographs of the flats used on these
latter tests are not included in Fig. 5. Surface damage appears to increase
dramatically with amplitude in this series. An estimate of 0.075 /xm for the
minimum amount of microslip necessary to produce surface damage would
seem reasonable.
Additional testing, basically under identical conditions, was done on a
series of unpolished specimens in order to determine if fretting damage was
significantly affected by the surface finish. The wear scars on the unpolished
specimens were found to be for all practical purposes identical to the wear
scars formed on the polished specimens. No surface finish effect was found.
Visual inspection of the photomicrographs suggested an exponential type of
wear growth with regard to slip amplitude. In order to investigate this possi-
bility graphically, data from both poHshed and unpolished specimens were
combined for use in plotting Figs. 6, 7, and 8. The effect of microslip on the
outer diameter of the wear scar is shown in Fig. 6. Although these data were
.7
• UNPOLISHED SPECIMENS
A POLISHED SPECIMENS
.6
E
E
BC .5
1=
S
^
4
a .4
<
.2
— I — — I —
1.0 2.0 3.0 4.0
MICROSLIP, urn
FIG. 6-i-Effect of the slip amplitude on the diameter of the wear scar for the SAE 1018-SAE
52100 steel combination.
collected under conditions of constant load, there is a significant increase in

the outside diameter of the wear scar with increasing slip amplitude. This
may represent surface damage and wear. However, theoretical explanations
are possible. Another parameter investigated was the width or the thickness
of the slipped region. This width would be equal to the difference in length of
the radius of the wear scar less the radius of the locked region. These data are
shown in Fig. 7. Each point represents the average of two determinations
since there is variation in the width of this region around the periphery of the
wear scar. The effect of a small change in amplitude on the width of this area
or the amount of surface damage was found to be very pronounced. This
relationship is linear in the region ranging from 0.5 to 2.5 jxm.
A peculiarity in the fretting wear process was noted at a microslip value of
approximately 2.5 /im. In the normal operation of this test rig in the 0 to
2.5-/im region, the power requirements for the electromagnetic drive units
were found to be in direct proportion to the slip amplitude. Also, the power
requirement needed to maintain a given slip amplitude did not noticably
change during the 3-h test period. For amplitudes greater than 2.5 nm, it was
found that the power supplied to the electromagnetic drive units needed to be
constantly increased throughout the test period in order to maintain the
selected slip amplitude. This type of behavior is suggestive of a severe adhe-
sion type of wear process and may represent the transition from fatigue wear
to adhesive wear. This factor would help to explain the failure of the data
0.30 n
• UNPOLISHED SPECIMENS
A POLISHED SPECIMENS
at
0.20
a
»
O.IO
MICROSLIP, urn
FIG. 7—Effect of the slip amplitude on the width of the slipped region for the SAE1018-SAE
52100 steel combination.
1.8n
1.4
i '•'
0.6
0.2
—r-
.04 0.8 1.20 1.60 2.00
RADIUS OF LOCKED REGION, mm
FIG. 8—The minimum slip amplitude needed to produce surface damage on an SAE 1018
steel flat specimen under a 9-kg load as a function of the radius of the locked region.
point corresponding to 3.2 /im in Fig. 7 and particularly in Fig. 8 to fall on

the experimental line.
In Fig. 5, it can be seen that the diameter of the locked region decreases as
microslip increases. Theory suggests that the diameter of the locked region
can be defined by those points at which the ratio of shear to normal stress is
equal to the coefficient of friction of the metal pair and this, in itself, offers
an explanation of the observed behavior. An alternative explanation for this
behavior might possibly be that fretting damage requires a certain minimum
amount of microslip before it can occur and that this factor is primarily re-
sponsible for determining the diameter of the locked region. Both explana-
tions are consistent with each other if it is proposed that a certain region ex-
ists outside of the true locked region in which microslip occurs without any
resulting damage. It should be possible to determine whether the diameter of
the locked region is primarily determined by stress distribution or the
amount of microslip, through an analysis of the minimum microslip that oc-
curs at the circumference of the locked region.
In this study, the radius of the locked region varied from 0.07 mm in the
case of severe fretting damage to 0.16 mm in the case of minimal damage. If
the calculated minimum slip amplitude at the circumference of the locked

region for these specimens is found to be essentially constant and indepen-
dent of slip amplitude or wear, then this would support the assumption of a
certain minimum threshold value for severe fretting damage. If some other
type of functionality exists between these two parameters, this would not be
true. In Fig. 8, the radii of various wear scars are plotted as a function of the
amount of calculated microslip based upon rigid body analysis occurring at
the leading edge of the wear scar. In the interpretation of this plot, the
smaller radii correspond to greater damage. There appears to be some evi-
dence from this plot that a certain minimum amplitude is necessary to pro-
duce fretting damage.
In order to determine an absolute value of the minimum microslip, micro-
slip must be based on an elastic analysis of the contact behavior. In order to
check the experimental data, a theoretical estimate of the radius of the
locked region as a function of amplitude was obtained from Deresiewicz's
equation for elastic spheres under an oscillating torsional couple [7]. In this
analysis, the radius of the locked region can be defined by the equation
fN 16 V 8 64
where K^ = I — c^/a?-, n is the shear modulus, a is the initial load radius, j8

is the amplitude in radians,/ is the frictional coefficient, and N is the applied
load. For these calculations, the shear modulus was taken at 10^, the fric-
tional coefficient was assigned a value of 0.8, the load was 7 kg, and the ini-
tial load radius was 0.152 mm. These data are shown in Table 1. There is ex-
cellent agreement between theory and experiment, considering the damaged
TABLE 1—A comparison of the theoretical versus the measured radii of the
locked region as a function of amplitude.
Theoretical Measured
Relative" Average Slip Radius of the Radius of the
Amplitude, Amplitude, Locked Region, Locked Region, Percent Relative
fim ^m mm mm Error
0 0 0.152 0.159 4.6

5.1 0.11 0.150 0.159 6.0
12.7 0.26 0.145 0.162 11.7
25.4 0.53 0.137 0.157 14.6
50.8 1.06 0.117 0.142 21.4
76.2 1.59 0.104 0.110 5.8
101.3 2.12 0.084 0.086 2.5
114.3 2.38 0.081 0.072 -11.1
152.4 3.18 0.066 0.082 24.2
"Amplitude measured at the capacitance probe.

^^^îm£:..
FIG. 'iz—Photomicrograph of a wear scar formed in an SAE 1018 steel flat specimen at X55
magnification. The wear scar was formed using a Hastelloy B ball under a 9-kg load.
FIG. 9b—Expanded view of the slipped region in Fig. 9a at X275 magnification.
nature of the surface. These data mean that an elastic analysis can be used to
predict microslip within the interface.
Hastelloy B Alloy-SAE 1018 Steel Combinations

The SAE 52100-SAE 1018 steel combination was initially studied because
it is a relatively common material combination and because it had been im-
plicated in a U.S. Navy-related fretting problem. Further studies were car-
ried out using a Hastelloy B alloy ball and SAE 1018 steel flats. A typical
photomicrograph of a wear scar with an expanded view of the wear track is
shown in Fig. 9. In this part of the study, photomicrographs were obtained
using an electron microscope rather than the previously used optical micro-
scope. The wear scar in Fig. 9 was formed on an SAE 1018 steel flat specimen
using a Hastelloy B ball. This latter material is a nickel alloy containing a
high concentration of molybdenum. As a result of the significant difference
in the metallurgy, flat specimens were subjected to microprobe analysis in
order to determine if metal transfer processes had occurred. The results of
this study indicated the presence of iron, nickel, and molybdenum in the
slipped region in the flat, but no nickel or molybdenum was detected in the
locked region or outside this contact area. The only element found in these
latter areas was iron. This type of data helps to define the actual wear pro-
cesses that occur under fretting conditions. In this case, microprobe analysis
indicated that actual metal transfer from the ball to the flat did occur.
A knowledge of the morphology of the wear scar is important in order to
validate that the test rig is operating properly and also to gain an insight into
the type of damage occurring in the slipped region. Traces of the surface pro-
files for several different wear scars are shown in Fig. 10. These traces illus-
trate the type of damage that would be found on the SAE 1018 steel flat
specimens. The data do not suggest any permanent deformation of the flat
that might be expected as a result of loading the Hastelloy B ball against the
SAE 1018 steel flat. In these traces, the elevation of the locked region ap-
pears to be at the same height as the surface of the flat outside the damaged
region. Any waviness in the locked region could possibly be attributed to sur-
face deformation during polishing. If surface deformation had occurred, this
would have required a modification of the assumptions concerning the distri-
bution of normal stress in the contact area.
Surface profiles are presented as a function of slip amplitude in Fig. 10.
All the tests were run for 3 h except for one test which was run for an hour at
0.53 iim. This latter test was included since it shows some evidence of detach-
ment of metal from the flat specimen itself. As an approach to monitoring
fretting damage, wear scar volumes were determined from these traces by
first determining an average cross-sectional area and then multiplying this
value by the length of the average circumference of the slipped region. These
data are shown in Fig. 11. The calculated wear scar volumes were found to be
SUP AMPLITUDE
.08 tim
.53 fum
1.33 um
1.48 fifn
125 //ffl
FIG. 10—The effect of slip amplitude on the surface profile of wear scars formed using the
SAE 1018-Hastelloy B material combination.
extremely small, but these data did indicate an exponential relationship be-
tween the wear scar volume and the slip amplitude. Because of a lack of data
in the critical region between 0.4 and 1.2 /xm, it is difficult to say with cer-
tainty if any mechanical changes occurred in this slip amplitude region; how-
ever, it is apparent that surface damage began in the 0.08 to 0.53-/Lim range.
A value of 0.075 ixm was found for the SAE 52100-SAE 1018 combination in
the previous section.
Conclusions
These data are, of course, preliminary. However, they show that this test-
ing device can generate highly reproducible wear scars which have character-
istics highly dependent on slip amplitudes. Important parameters related to
the fretting process, such as a characteristic slip amplitude associated with
the first evidence of mild oxidative fretting and what appears to be a charac-
teristic slip amplitude associated with the onset of severe adhesive wear, have
been identified and determined from wear data. In the first case, this charac-
teristic slip amplitude can be easily determined through visual inspection of
wear scars. For example, in the case of the SAE 52100-SAE 1018 steel com-
bination, the first indication of red oxidative wear would be expected at
7TX10-4
SLIP AMPLITUDE, Mm
FIG. 11—Wear volume as a function of slip amplitude for the SAE 1018-Hastelloy B material
combination.
around 0.75 ptm. A second limiting parameter of slip amplitude, which is

probably more associated with specimen configuration than the fretting pro-
cess, was found at 2.5 ixta. Below 2.5 ixm, the frictional characteristics of the
metal pair remained essentially constant throughout the test. Above this
value, the frictional characteristics tended to increase rapidly.
References
[/] Begelinger, A. and DeGee, A. W. I. in AGARD Conference Proceedings, No. 161, National
Technical Information Service, Springfield, Va., 1974, pp. 9.1-9.7.
[2] Peterson, M. B. and Gabel, M. B. in AGARD Conference Proceedings, No. 161, National
Technical Information Service, Springfield, Va., 1974, pp. 12.16-12.17.
[3] Ohmae, N. and Tskizow, T., Wear, Vol. 27, 1973. pp. 281-294.
[4] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, New York, 1972.
[5] Waterhouse, R. B. in Proceedings of the 1981 International Wear Conference, San Fran-
cisco, Calif., American Society of Mechanical Engineers, April 1981, pp. il-ll.
[6] Peterson, M. B., Geren, B. F., Arwas, E. B., Gray, S., Murray, S. F., Lund, I. W., and
Ling, F. F., "Analytical and Experimental Investigation of Gas Bearing Tilting Pad Pivots,"
NASA CR 72609 (MTI TR-32), Mechanical Technology Incorporated, Latham, N.Y., 1969.
[7] Deresiewicz, M., Journal of Applied Mechanics, Transactions of A.S.M.E., Vol. 76, 1954,
pp. 52-56.
K. G. Budinski^
Evaluation of Fretting Corrosion by

IVIeans of a New Device for tlie
Control of Oscillation Annplitude
REFERENCE: Budinski, K. G., "Evahifitioii of Fretting CoRoskm b;; Means of a New

Device for the Control of Oscillation Amplitude," Materials Evaluation Under Fretting
Conditions, ASTM STP 780, American Society for Testing and Materials, 1982, pp. 49-
67.
ABSTRACT; Many studies of fretting corrosion and wear are conducted on production
machinery or on fatigue testing devices. Such devices often rely on elastic strains in loaded
members as the source of the fretting motion. This makes measurement and control of
slip amplitude and frequency and of normal force difficult. Assessment of the damage is
also often difficult.
This paper discusses problems in studying fretting phenomena and describes the use of
a slider-crank tester that allows control of most fretting variables, including slip
amplitude. Fretting resistance test data are presented on a variety of coatings and metal
couples. Recommendations are made on the use of various metal couples and selected
surface treatments.
KEY WORDS: fretting, fretting wear, fretting corrosion, surface coatings
Fretting motion has been recognized in the literature as a source of equip-

ment damage for at least 70 years. Researchers have studied many aspects of
fretting corrosion and fretting wear with almost as many different test rigs as
there are researchers; there is no standard test or standard test rig. In this
paper we shall review the various parameters that affect fretting corrosion
and describe yet another device for studying fretting systems.
This new device was only reluctantly developed after unfavorable ex-
periences with rigs patterned after those used by other researchers. The
primary purpose of this paper is to describe this new device and to show why
it was necessary. The secondary purpose of this paper is to present the results
obtained to date in a program of continuing studies on the fretting corrosion
'Senior metallurgist, Eastman Kodak Co., Rochester, N.Y. 14650.
49
tendencies of various metal-to-metal couples and on the effectiveness of

coating and surface treatments in preventing fretting corrosion. The objec-
tive of these continuing studies on this new device is to find ways of reducing
the cost of fretting damage in high-speed manufacturing equipment.
Review of Fretting Tests and Measured Parameters

Oscillatory Sliding—This is possibly the only parameter that all in-
vestigators agree is important. The word fretting in current tribology ver-
nacular means "oscillatory motion." It is derived from the word fret, which
was derived from the Anglo-Saxon word fretan which means "to gnaw."
Thus fretting is an appropriate term for the situation that occurs when
materials experience oscillatory sliding. A fretting test rig must produce
oscillatory sliding or it is not a fretting test.
Amplitude—When does fretting become reciprocating sliding? Fretting
studies have been reported with slip amplitudes as low as 2 ixm [1]^ and as
high as 25 mm [2]. Ohmae and Tsukizoe [3] have shown that material
removal increases with slip amplitude and peaks at about 70 ^m. From a
practical standpoint, fretting becomes reciprocating sliding and the damage
becomes sliding wear when the oscillatory motion becomes macroscopic.
Movements in excess of 300 /im are certainly macroscopic in most systems.
Most studies are conducted in the amplitude range of 10 to 300 ^m, and this
range seems to define the typical fretting amplitude reasonably well.
Normal Force—Without surveying the literature, it can be stated that fret-
ting corrosion and fretting wear occur with normal forces as small as 0.01 N
and as high as 3 X 10* N. Fretting corrosion is common in electrical contacts
with miniscule loads and in the forces of injection molding cavities where
clamping loads are in the high normal force range. Damage can occur at
almost any normal force.
Frequency—A situation similar to normal force exists with frequency. In-
dustrial experience has shown that fretting corrosion is prevalent on the fay-
ing surfaces of ultrasonic horns. Slip occurs because of magnetostrictive ef-
fects, and the frequency of slip can be as high as 20 X lO-' Hz [4]. At the
other extreme, the frequency of clamping in plastic injection molding is
seldom less than once every 10 s, but fretting corrosion still occurs.
Velocity—Sliding velocity is not important in producing fretting damage.
The slip in ultrasonic devices is at extremely high velocities. Fretting is also
common on stainless steel racks in drying vessels where the oscillatory move-
ment is due only to thermal expansion and contraction; this is definitely a
low-velocity slip since one process cycle may take 8 h.
Friction—Should a fretting test rig include a means of measuring and
recording friction? A significant number of investigators have included fric-
BUDINSKI ON OSCILLATION AMPLITUDE CONTROL DEVICE 51
tion measuring devices in fretting test rigs [3,5-8]. The value of friction
measurements appears to be that such measurements allow one to study the
progression of surface damage. The absolute values do not seem to yield in-
formation on a system's resistance to damage under fretting conditions.
Therefore, the ability to study forces at the fretting interface is not essential
in material evaluation studies. A test rig that allows friction measurements is
preferred but not necessary.
Environment—Since fretting corrosion involves an oxidation process, the
ability to control the environment is indeed desirable. It is well known that
fretting in a vacuum or in an inert atmosphere reduces the oxidation compo-
nent of fretting corrosion damage [4,9]. Some investigators have found that
even moisture in the air can affect fretting damage [1,10]. From a practical
standpoint, inclusion of environment control in a fretting rig need only be
done if special environmental conditions apply to a system under study. If
elevated temperature is involved in a particular application, it also should be
included in the control of the environment.
Damage Measurement—The degree of damage produced by a particular
fretting system can be measured in several ways (Fig. 1). Measuring weight
loss, the traditional wear measurement, may be impractical if the fretting
specimen is large. A fretting bridge on a bending fatigue specimen produces
a very small amount of material removal on what are usually large tensile-
type specimens. Such measurements may be imprecise because of the load
limits on analytical balances. Fretting tests have even been conducted on
350-mm-diameter axle shafts [11] and on whole wings of aircraft [12]; again,
gravimetric damage measurements are impractical in these situations.
Surface profilometers can be very useful in measuring weight loss. They
can be used to calculate volume losses on bulky specimens and on full-size
parts. They can even be appropriate for small specimens [5,13].
When fretting tests are performed on fatigue specimens, the measured
fretting parameter is often the reduction in fatigue strength [8,10,13-15].
Measurement of the scar area can be used to get around some of the
drawbacks of measuring small wear volumes. The measurement can be made
as a percentage of the apparent area of contact [16], or it can be of the entire
scar area. With hemispherical riders, the area of contact increases with the
amount of damage [1,4,8,10,17,18].
Some more creative measurements of fretting damage include such things
as contact resistance, frictional energy, and wear debris.
Scanning electron microscope (SEM) and transmission electron microscope
(TEM) photomicrographs are frequently used to rate fretting damage [10,13,
15,17,19]. Usually such measurements are qualitative, but they provide much
information on surface deformation, pitting, and oxide formation.
The subject of damage measurements can be summarized by saying that
there is little agreement on the best way to make measurements.
We have summarized some of the trends in fretting test rigs, the parame-
(0
Ul
u
z
lU
o
z Ul
lU
oo H >
O <
z
1° cu. < Ul
X S § i
OQUI
z i* c
lU O M
E
3
± O U s (0
H^ lUrr C U.
<
(0 uj
Ml > C
S oeO s
Ill >• >• >•
UJ Q Ul UJ U
U o ga gg •- 2 J
O < < < St <
I J lU 111 Ui
. K oe E - • z S zu
H OE
^<<< Ul Ul «pt-
O E E E O
3S E
wooo I - Ul Ul E - I K-
<a. W U. Ul Ul
NO S M (0 M U. (0
i.o^ •
Q. S
M E
z
z
ters that should be measured and controlled, and the measurement of

damage. With these factors in mind, what is the best type of fretting test rig?
The obvious answer is one that measures everything. Unfortunately, such a
device may be beyond the economic means of most researchers. In an effort
to analyze what a proper rig should look like, it is possible to do some
categorization. There are basically two type of devices: (1) testers that rely on
elastic strains for generation of oscillatory slip and (2) testers that directly
oscillate one member in contact with a fixed member. Figure 2 illustrates
what some of the specimens used in each of these categories look like. The
American Society of Lubrication Engineers' Catalog of Friction and Wear
Devices [20] shows many additional specimen configurations that can be
used for fretting studies.
Unfortunately, no individual is in a position to state that all fretting tests
should be conducted with a rig selected from one of those listed here or
elsewhere in the literature. The purpose of this discussion is to show that
there is a need for some agreement on fretting test rigs and measurements of
wear damage. Without such agreement, intralaboratory comparisons will
never be possible. The author hopes that a test standards organization such
as ASTM could conceivably arrive at a standard at least on how to run fret-
ting tests, that is, on what to control, what to measure, and how to report the
results.
Development of a New Fietting Test

As was pointed out in our review of various fretting rigs, there is no test
that is agreed upon as a standard or even preferred test. After the author's
laboratory had conducted fretting corrosion tests for several years on a rig
that relied on elastic deflection as the source of oscillatory movement [13], it
was decided that an "ideal" test rig should have the capability of controlling
and easily measuring slip amplitude. Some of the other criteria considered
important in a new design were the following:
(a) the ability to use small specimens for accuracy in weight loss
measurements,
(Z>) the ability to use specimens that could be easily finished and not re-
quire close machining tolerances,
(c) the ability to use specimens that could be removed during a test and
examined,
(d) the ability to control the speed of oscillation,
(e) the ability to control the environment, and
(/) the ability to control the normal force.
The device previously used for our laboratory tests (Fig. 3) consisted of a
cantilevered shaft rotating in a standard ball bearing. The fretting surfaces
were the outside diameter of the shaft and the inner race of the bearing. This
^ It
/
X ^
0f ^
4 n
• SiS
<
1a
I Q. 10
«Q O—
III Z %
»
£5
>^ 1*^
" n
- U)
?"
5>
<
i"
"3
.0
0 S,
£>
"S^
o«
.c
I
o
TEST BEARING
• TEST SURFACE
FIG. 3—Test rig that requires elastic strains to produce oscillatory motion.
rig produced fretting conditions that simulated many service conditions but
did not meet many of the previously mentioned criteria.
The test shaft weighed over 1 kg, and this precluded the use of an
analytical balance to measure weight loss. The test specimens were expensive
because of the close tolerances required at the bearing diameters. The slip at
the bearing interface could not be easily measured, and it was calculated
from measured elastic deflection of the test shaft. The rate of material
degradation could not be determined from intermittent inspections of the
faying surface; it was too difficult to put the shaft back in its exact location.
Finally, the use of a ball bearing and a shaft as faying surfaces made it dif-
ficult to control the normal force.
Many different approaches were examined in trying to arrive at a fretting
test rig that would eliminate all of the problems that existed with the deflect-
ing shaft rig. Fretting bridges on elastically strained specimens in a fatigue
tester were decided against because they make it difficult to measure slip
amplitude [21,22], and it is difficult to make periodic inspections of the fret-
ting surfaces. It was decided hemispherical riders would be avoided because
it is difficult to control the surface microtopography and shape. Accurate
spheres are difficult to obtain unless ball grinding facilities are available. In
addition, hertzian loading is the exception rather than the rule in many
machine components that are subject to fretting conditions. Oscillatory mo-
tion produced by an ultrasonic device was not desired because this creates a
very special fretting condition, atypical of most machines. Electrical
vibrators were considered as a source of fretting motion, but they were de-
cided against because in continued service the vibrating member in the elec-
trical device can fail because of the fretting motion.
Based on these opinions on other fretting rigs and consideration of the
stated design criteria, a rig was designed with mechanical actuation using a
combination cam, toggle, and slider-crank mechanism. A schematic of the
device arrived at is shown in Fig. 4. A variable-speed electric motor rotates a
cam with variable eccentricity. The cam has a shoulder screw protruding
from it which is coupled with a rod-end bushing to a slider mechanism. Since
the vertical translation of the slider link is small, there is a toggle action, pro-
.^frrmTm>^ ADJUSTABLE CAM
DEAD WEIGHT
FIXED SPECIMEN
(RING)
MOVING SPECIMEN
(BLOCK)
FIG. 4—Schematic drawing of a slider-crank fretting tester.
viding significant mechanical advantage in the axial direction (oscillation

direction). The movable member in the fretting couple is clamped in the
slider block, and the slider block moves on preloaded ball bushings on
hardened steel pins. The stationary specimen is held on a platen which, in
turn, is held fixed in the horizontal direction by four guide pins. The platen
weight provides the normal force at the fretting couple. It is free to move
downward on preloaded ball bearings. The test specimen's dimensions are
shown in Fig. 5. The movable specimen is a block with four usable test sur-
faces. This configuration was chosen because it is the same size as the
specimen that fits on existing metal-to-metal test rigs (block-on-cylinder
design). The fixed-specimen ring configuration was arrived at after tests
showed that other configurations did not work. Hemispherical riders showed
too much variation in spherical form for repeatability. Flat-ended pins would
invariably make contact in only a small fraction of their apparent area. It was
difficult to get the end of a cylinder free of waviness and of errors of form and
also perpendicular to the outside diameter of the cylinder. The rings could be
lapped to reasonable flatness on a lapping plate, and they were made self-
aligning to the block by locating them on a cemented carbide ball attached to
the platen weight. The ball fits into the chamfered annulus of the ring.
The capabilities of the new rig are summarized in the following list:
(a) an oscillatory amplitude variable from 1 to 1000 ^m,
(b) a normal force from 30 to 100 N,
(c) sliding velocities from 4 X 10"* to 2 X lO'^ m/s,

(d) a frequency of oscillation 1 to 1 500 Hz, and
(e) the capability of running in an environmental chamber.
The only desired parameter that was neglected on this basic rig was a fric-
tion measurement. It will be added in the future with a strain gage on one of
the slider links. This design was not trouble-free, however; the initial version
did not work. The first version of the device employed porous bronze plain
bearings in the link members. The running clearances on standard porous
bronze bearings were large enough to preclude oscillatory amplitudes below
about 1000 ^m. The unit had to be rebuilt with preloaded precision needle
bearings in the links. The present design does produce reproducible
oscillatory motion; the fretting specimens are easily prepared and low in cost;
fretting corrosion can be produced in untreated steel surfaces in short test
times (as low as 10 min). Finally, the simple specimen configurations lend
themselves to the application and testing of coatings.
Fretting Resistance Tests

The primary purpose for developing the new testing rig was to try to find
coating or material combination solutions to recurring fretting corrosion
problems in production machinery. Previous studies [23] on bearing-shaft
combinations showed that controlling the diametral clearances was the most
effective means of preventing fretting corrosion. The systems aimed at with
this test rig were those where unavoidable elastic deflections produced fret-
ting motion and, subsequently, fretting corrosion. One example of such a
system is rotary slitting knives on a common knife bar. Such systems are used
for slitting steel and countless other web-type products. These knives are
usually made from hardened tool steels, and they fit on shafts with diametral
clearances usually in the range of 2 to 3 /xm. The knives are spaced with
hardened steel rings, and they are clamped to the slitter shaft with various
"friction" devices. Invariably, the knife shaft deflects from the slitting
forces, and the shaft bow causes a small amplitude slip at the knife-shaft fay-
ing surfaces. Fretting corrosion occurs on both members, and frequently the
shafts are destroyed in trying to remove the knife sets for periodic sharpen-
ing. If a material combination or shaft coating could be found to prevent the
fretting corrosion, a significant savings in cost could be realized.
With these types of problems in mind, tests were initiated on the newly
developed test rig. Debugging tests indicated that fretting corrosion could be
easily produced in a very short running time under most operating condi-
tions. Debugging tests were conducted with blocks and rings of Type 1020
steel with a hardness of about 90 Rockwell hardness, B scale, and a ground
surface finish of 0.5 to 1 nm roughness average (Ra). Specimens were loaded
with the minimum possible normal force, 30 N, and the specimens were run
BLOCK SPECIMEN
FIG. 5—Fretting test specimens.
for varying times at a number of different sliding velocities in an air environ-

ment. Slip amplitude was held constant at 150 fim because this was a value
that some investigators have shown produces significant fretting corrosion.
At sliding speeds ranging from 0.000025 to 0.002 m/s, fretting corrosion
was visible on at least some portion of the apparent contact area after run-
ning times as short as 5 min. Fretting corrosion occurred in as few as 10^
cycles of oscillation. The apparent contact area of most specimens was com-
pletely covered with oxide and pitting damage after 10^ cycles (about 24 h).
Since the goal was to find a fretting corrosion adjuvant that would work
under severe conditions, it was decided to standardize a test of 10^ cycles with
a slip amplitude of 150 /xm and a sliding velocity of 0.0017 m/s.
Tests of 10* cycles were conducted on a range of metal combinations com-
mon in slitter knife applications:
Type 4130 steel at 25 HRC versus Type 02 tool steel at 60 HRC,
440C stainless steel at 59 HRC versus Type 02 tool steel at 60 HRC,
M2 tool steel at 63 HRC versus Type 02 tool steel at 60 HRC,
high-carbon-high-chromium hardfacing at 61 HRC versus Type 02 tool
steel at 60 HRC,
25 percent lead bronze versus Type 02 tool steel at 60 HRC, and
beryllium copper at 38 HRC versus Type 02 tool steel at 60 HRC.
BUDINSKI ON OSCILUTION AMPLITUDE CONTROL DEVICE 59
Simple visual examination showed whether fretting corrosion had occurred

in the test, and little would be gained by weight loss measurements since the
goal was to produce "no fretting corrosion." The test was quantified
somewhat by using areal analysis to measure the percentage of the apparent
area of contact that showed damage on both the ring and the block. A
transparent grid with ten lines per centimetre was used for the areal analysis.
Four different test blocks and rings were used for each material combination,
and the results were averaged for the four replicates.
A variety of "antifretting" coatings were tested under the same conditions:
lithium grease
mineral oil
polytetrafluoroethylene (PTFE) solid particle spray
antiseize compound
graphite
silver electroplate
lead electroplate
The fretting damage was measured by the same areal analysis technique used
on the metal-to-metal combinations.
Test Results
The results of the unlubricated metal-to-metal tests are presented in Fig.
6. All of the steel-steel combinations showed severe fretting corrosion. The
damage, expressed as a percentage of the apparent area of contact, was
essentially the same on both members of the contact. Typical specimen
damage is shown in Fig. 7. After 24 h of oscillation, the damaged areas con-
sisted of tightly adherent oxide and shallow pitting. The damage starts on the
crests of the scratch pattern produced by surface grinding (Fig. 8). Eventu-
ally the crests are removed by attrition, and the entire surface is covered with
the oxidation product of the fretting corrosion. The soft steel. Type 4130 at
25 HRC, exhibited more damage than the hard-hard combination which in-
dicates that a hard-hard combination has better fretting resistance. When
the test times were increased to 100 h the hard-hard and soft-hard combina-
tions showed the same degree of damage—100 percent of the apparent area
of contact on both members of the couple. Thus in long-term applications,
both soft-hard and hard-hard steel combinations will show substantial
damage if the use conditions are similar to those of this laboratory test.
The test results on the copper alloys were quite different. The steel
member in the fretting couple, the ring, showed no fretting corrosion m the
tests of high-lead bronze and beryllium copper versus Type 02 tool steel.
There was a slight yellow discoloration on the steel ring in the leaded bronze
test and gross transfer of copper to the steel ring in the beryllium copper test.
In both cases the steel was protected from damage by the copper alloy
DAMAGED DAMAGED
BLOCK MAT'L AREA RING MAT'L AREA
% %
4130 STEEL 02 TOOL STEEL
@ 25 HRC @ 60 HRC
440C STAINLESS • • • • 02 TOOL STEEL | M H |

@ 59 HRC • • • (a>60HRC ! • •
M2 TOOL STEEL b B B | 02 TOOL STEEL ^ B H

@ 63 HRC • • • (3) 60 HRC p i l l
HIGH C HIGH Cr ^ ^ ^ 02 TOOL STEEL

HARD FACING ^^M @ 60 HRC
@ 61 HRC ^ ^ ^
HIGH Pb BRONZE • • • • • • 02 TOOL STEEL NO

@150HB • H I H H ^ @ 60 HRC PITTING
BERYLLIUM Cu • • • • • • • 02 TOOL STEEL Cu TRANSFER

(3)38 HRC P H H H (i>60HRC TO STEEL
FIG. 6—Fretting damage produced with various dry metal couples (SO percent relative
humidity, 20°C, 150-nm slip amplitude, 30 N normal force, 10 Hz, 10^ cycles, velocity 0.0017
m/s).
FIG. 7—Fretting corrosion on test samples.
transfer. The damage on the copper alloys was similar for the leaded bronze
and the beryllium copper: discoloration over the entire area of contact and a
measureable scar depth. Figure 9 shows typical damage on the beryllium
copper.
The coating test results are shown in Fig. 10. The lithium grease did not
prevent fretting damage on either member. A one-time application of light
mineral oil (220 SSU) reduced the damage on both members to an extremely
small area (about 5 percent of the apparent area of contact). When both
members were sprayed with a commercially available PTFE mold-release
agent, fretting damage to both the steel members was undetectable with op-
tical microscopic examination. The particle size of the agent was about 25
nm, and application was made by three passes from an aerosol spray device.
When the same spray was used, but the coating thickness was reduced by
burnishing, the effectiveness of the coating was greatly reduced. Fretting cor-
rosion occurred over 90 percent of the apparent contact area of both
members.
A 25-/im thick application of a commercial copper-lead antiseize grease
reduced the fretting damage on both members to undetectable levels. A
25-/im thick coating of graphite had the same effect. Damage to both
FIG. 8—Fretting damage on a ground steel surface (X400 magnification).
FIG. 9—Fretting damage on beryllium copper after testing versus Type 02 tool steel (under
the same testing conditions as in Fig. 6).
members was nil after the 24-h test. The graphite coating was produced by
dipping the test ring four times m a colloidial dispersion of graphite in
alcohol. The dispersion was about 5 percent by weight graphite. Each im-
mersion of a test ring in the dispersion resulted in a coating thickness of be-
tween 5 and 10 /^m. The coatings were air dried until they were dry to the
touch and did not wipe off in handling. Only the rings were coated. In the
tests on the rings with 5 and lO-jiim thick coatings (one or two immersions)
fretting damage was not prevented.
Of the two electroplates tested, the l-^tm thick lead coating was the only
one that appeared to have a palliative effect. The silver electroplate was
almost completely transformed into an oxide form in the area of contact. The
lead coating remained intact but did show some oxide debris. Fretting corro-
sion on the 440C stainless steel block was only 11 percent.
Discussion of Test Results

This study was not aimed at understanding the mechanisms of fretting cor-
rosion and wear. It was aimed at finding usable solutions to fretting prob-
lems in manufacturing equipment. The exercise of determining load effects,
o
UJ 1
<
2«
"^ a. z
UJ I
1- u z UJ UJ
ZBC
Ul lu
Sw
u. <
k & O
o oc
oc
3x t I.I
o '*'
u.
Ouj
u.
O UJ
LL
<
_J
<
a.J
s
UJ o oc O ui UJ O
u so
< =! 0. < o
-1 < to
S UJ
u. J. u. a.
dz
u. a. il
u. < -J
oc u. < oc u. «UJj tC
(0 z o z m | ! i i c
3 UI UI
o
e < u- <
hi I* Id Id
h-
« " û. X»
1 N a N N in ^
in o'' N ^ f
°t
<oeo
o<'-
11.
-1 -J
-1
Z
-1
1•
_i
-J
Z
-1 ^
-1
z • •
WLM M JB.
-J
1
-1
UJ UI UJ UJ u UJ UJ UJ
-J -UJ
J UJ
_j UJ UI UJ UJ UJ UJ UI UJ UI
u
<
S -Ji 'J
-ll -Jl
.oc Sg
-Jl
5o
.oc fey Sg
J l J X
:^z
"oc
ss Sg 11 "oc
-JX -II J l
a
z
£
Is Is is Is Is °s Is is ll N® N®
J- iv
o o o o o o Oo o o
$1 • • it
Zuo • J
z
•• _l
z
-i
z • • •
^ ^ H
it
<a:o
0<r-
_i
•. tn
u>
UJ
•.
(A
U
UJ
CO
UJ
M
(A
V>
UJ
M
UJ
H
<
sÔ
uju iuu
UjOC UjE
I--Z i - z
WlO WtO
uju
lilOC
i-i
««
UJU
UJOC
t-z
«I0
-J
UlU
UIOC
HZ
«I0
_i
2u
<GC
»-Z
«0)
-1
So
<oc
(-Z
«<»
_l
2y
<cc
i-i
«0>
-1
?o
<
««
a
l-X
U
5o
<
KZ
a
«0)
fj
o ON ON OM ON ON um oin um um
CD w® n® W® CO® P® o® o® o® o®
6®
<» * * * * * * « « *
velocity effects, amplitude effects, and so on was forgone. These things have
been done many times before by many investigators [17,20,23]. The test used
was very severe. If a material combination or surface coating could make it
through this test, that is, if both members of the fretting couple were free of
damage after the test, it is very likely that the couple or coating would pre-
vent fretting damage in long-term service.
Wear depth measurements on 100-h tests on soft-hard and hard-hard
steel combinations indicate that a hard-hard combination produces less at-
trition but that the damaged area will be about the same. Thus using various
combinations of material hardness does not appear to be a method for long-
term elimination of fretting damage.
The alloy content did not show any effects in preventing damage in these
tests. The Type 4130 steel had less than 3 percent alloy, and the M2 tool steel
had an alloy content in excess of 20 percent, yet they behaved about the
same; both alloys showed about the same degree of damage after the 24-h
fretting tests.
A wide variety of microstructures were used: bainite was the major compo-
nent in the Type 4130 steel, martensite plus chromium carbide was the
microstructure of the 440C stainless steel, martensite plus mixed carbides
was present in the M2 steel, and a dendritic structure of iron carbide and
chromium carbides existed in the hard-faced alloy. Once again, these alloys
behaved about the same. This may not be the case with all counterfaces, but
it was true with a counterface of Type 02 tool steel. The explanation of this
may be as simple as the fact that all of these microstructures are iron or iron
compounds. If fretting corrosion is a mechanically assisted oxidation of iron
[22], then it is reasonable to assume that iron-base alloys would behave in
about the same way. Microscopic examination of fretted areas did not pro-
duce any evidence that carbides reduce the severity of the fretting damage.
The tests with the copper alloys were an attempt to see if the use of a
significantly dissimilar metal couple would prevent fretting damage. The
nonferrous metals in general appear to be equally susceptible to fretting cor-
rosion, as are the ferrous materials. The oxides and debris, of course, do not
look like red rust, and the damage often appears to be less severe. For exam-
ple, the copper-steel tests undertaken in this study prevented weight loss of
the steel member (ring). The results could have been interpreted as 100 per-
cent damage on the steel member if metal transfer was included m the fret-
ting wear or fretting corrosion. Thus the author concludes that the use of a
copper alloy-steel couple did not prevent damage to either member under
our testing conditions.
The results of the lubricant tests can be understood quite easily. If a lubri-
cant can prevent contact of the faying surfaces, it will prevent fretting
damage [24]. Graphite did not work until the thickness was such that there
was no possibility of asperity interactions between the metal pair. Similarly,
the FIFE spray only worked when a thick macroscopic coating was used. A
thin coating of PTFE particles burnished into the surface roughness did not
work.
The antiseize compound was effective because it contained large particles
of copper and lead (about 25 jitm), and these maintained an effective separa-
tion of the faying surfaces.
A machine oil or grease will only prevent fretting damage if it stays on the
faying surface and prevents contact. It is doubtful if any one-time application
of a petroleum-based grease or oil will prevent fretting damage in a long-term
apphcation.
The behavior of the silver and lead electroplates was similar to that of the
lubricants. If the coating stays intact, it will protect the oscillating couple.
Both the lead and the silver electroplate are susceptible to fretting corrosion.
They suffer fretting damage and thus are not a long-range preventive of
substrate damage; they merely delay it. Lead performed better than silver
probably because it is softer and behaves more like a solid film lubricant than
does silver. (It has a lower shear strength than silver.)
Conclusions
Material and Coating Tests

1. The test rig that was developed for these studies is easy to use, and it
produces an extremely aggressive fretting condition. If a particular coating
or combination of materials is going to be susceptible to fretting corrosion, a
24-h test on this rig will provide an answer.
2. The metal-to-metal fretting tests conducted to date on this new tester
were not comprehensive, but the work to date indicates that all uncoated
steel combinations—soft-soft, soft-hard, and hard-hard—have essentially
equal susceptibilities to fretting corrosion. The dissimilar metal couples and
structures tested in this study did not prevent damage.
3. Nonferrous metals suffer substantial damage (oxidation and pitting)
under fretting conditions, but they may minimize damage to a steel counter-
face by transfer to the steel surface.
4. Petroleum lubricants only prevent fretting corrosion if they can main-
tain complete separation; boundary lubrication is inadequate.
5. Dry film lubricants such as PTFE and graphite will prevent fretting
damage only when the surface film is thicker than about 25 jum.
6. Soft platings do not prevent fretting corrosion. They accept the damage
and protect the substrates while they are mtact.
General
The laboratory test results that were obtained on the newly developed
test rig can be used to solve fretting problems in machine situations using
similar couples and conditions. The most useful result of industrial impor-
tance was the finding that a 25-tim dry film graphite coating will prevent
long-term fretting damage. This coating has been used on production
machinery, and the laboratory test results were confirmed. The author
recommends that such a coating be considered for use in fretting-prone
systems that can tolerate a substantial film thickness.
It is hoped that the discussion of fretting test parameters and fretting test
rigs has pointed out the need for agreement on how to simulate fretting con-
ditions and how to obtain usable information. The test apparatus that was
developed in this study may not have been necessary if a commercially
available test apparatus could be identified as a reliable fretting device.
Similarly, every investigator who ventures into the area of fretting wear
and fretting corrosion reinvents the wheel by studying the effects of
operating parameters on results. Progress in the mvestigation of frettmg
damage hinges on a combined effort on the part of researchers in the field to
agree on (1) the slip amplitude range that constitutes a fretting condition, (2)
how to measure fretting damage, and (3) how to simulate fretting condi-
tions m the laboratory.
References
[1] Higman, P. A., Scott, F. H., and Bethune, B., Wear. Vol. 47, 1978, pp. 71-80.
[2] Gensheimer, J. and Friedrich, G., Wear, VoL 17, 1971, pp. 407-419.
[3] Ohmae, N. and Tsukizoe, T., Wear. VoL 27, 1974, pp. 281-294.
[4] Wayson, A. R., Wear. VoL 7, 1969, pp. 435-450.
[5] Halliday, L S. and Hirst, W., Proceedings of the Royal Society. Vol. 236, 1956, pp.
411-425.
[6] Miller, K., Tribology. April 1975, pp. 57-64.
[7] Campbell, W. E. in Symposium on Fretting Corrosion. ASTM STP 144. American Socie-
ty for Testing and Materials, Philadelphia, 1952, pp. 4-18.
[8\ Endo, K. and Goto, H., Wear. Vol. 43, 1978, pp. 347-367.
[9] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, New York, 1972, p. 129.
[10] Poon, C , and Hoeppner, D. W., Wear. VoL 52, 1979, pp. 175-191.
[//] Horger, O. J. in Symposium on Fretting Corrosion, ASTM STP 144, American Society
for Testing and Materials, Philadelphia, 1952, pp. 41-52.
[12] VonTem, V., Proceedings of the Agard Conference, No. 161, North Atlantic Treaty
Organization, 1974, p. 201.
[13] Starkey, W. L., Product Engineering, July 1971, pp. 642-645.
[14] Bolin, E. L. and Hanik, D. K., Proceedings of the A.S.L.E. Conference on Solid Lubri-
cants, 9-11 Sept. 1969, Kansas City, Mo., pp. 12-16.
[15] Sproles, E. S. and Duquette, D. S., Wear. VoL 49, 1978, pp. 339-352.
[16] Godfrey, D., Journal of A.S.L.E.. Feb. 1973, pp. 43-44.
[17] Benzing, R. I. and McConnell, B. D., "Wear Behavior of an Air Drying Methyl Phenyl
Silicon Bonded MoS2 Lubricating Film," Air Force Materials Lab Technical Report 70179,
Wright-Patterson Air Force Base, Dayton, Ohio, Oct. 1976.
[18] Waterhouse, R. B., Wear. Vol. 45, 1977, pp. 355-364.
[19] Reeves, R. K. and Hoeppner, D. W., Wear. Vol. 48, 1978, pp. 87-92.
[20] Catalog of Friction and Wear Devices. American Society of Lubrication Engineers, Park
Ridge, 111., 1971.
[21] Uhlig, H. H., Tiemey, W. D., and McClellan, A. in Symposium on Fretting Corrosion,
ASTM STP 144. American Society for Testing and Materials, Philadelphia, 1952, pp.
71-81.
[22] Sproles, E. S. and Duquette, D. S., Wear, Vol. 52, 1979, pp. 95-109.
[23] Budinski, K. G., Thin Solid Films, Vol. 64, 1979, pp. 359-363.
[24] Bill, R. C , A.S.L.E. Transactions, Vol. 24, 8 Nov. 1977, pp. 226-242.
Morton Antler^
Fretting of Electrical Contacts
REFERENCE: Antler, Morton, "Fretting of Electrical Contacts," Materials Evaluation

Under Fretting Conditions, ASTM STP 780, American Society for Testing and Mate-
rials, 1982, pp. 68-85.
ABSTRACT: Fretting can cause the contact resistance of electrical components to

become unstable. Fretting originates in vibration, mechanical shock, thermal fluctua-
tions, and magnetic forces to which conductors are subjected. Gold is a preferred contact
material because it is inert and therefore has a contact resistance which remains nearly
constant during fretting. However, insulating oxide films appear at the interface of base
metal contacts and gold-coated base metal contacts when the gold wears out. Platinum
group metal contacts catalyze the formation of frictional polymers from organic mate-
rials, such as air pollutants, in their vicinity. Thus, fretting corrosion and friction poly-
merization can increase contact resistance and thereby cause electrical failure.
This paper describes a new research apparatus for studying the fretting of electrical
contacts, in which a stepping motor causes small oscillatory movements of a slide, table.
One specimen is mounted on the table, and the stationary contact is loaded against it in
the force range of 5 to 500 g. Contact resistance is determined at 1.6-fim intervals along
the wear track by a data acquisition system computer. Examples of the use of this equip-
ment to model fretting of electrical contacts are given.
KEY WORDS: connectors, contact resistance, electrical contacts, fretting, fretting cor-
rosion, fretting wear, friction polymerization
Electrical connections consist of two or more current-carrying members

with a common interface or contact surface. Components that have contacts
include separable connectors, which are designed to allow conductors to be
taken apart and joined, and relays, switches, and circuit breakers, whose
function it is to interrupt or to establish the current flow in active circuits.
Since the primary requirement of nearly all such devices is low and stable
contact resistance, their contact surfaces must be relatively free of insulating
films, especially when low forces are used and low voltage-low current cir-
cuits are served.
Fretting is a significant factor in the reliability of electrical contacts.
ASTM Committees B-4 on Metallic Materials for Thermostats and for Elec-
'Member of the technical staff. Bell Telephone Laboratories, Inc., Columbus, Ohio 43213.
68
ANTLER ON FRETTING OF ELECTRICAL CONTACTS 69
trical Resistance, Heating, and Contacts and G-2 on Erosion and Wear, in
recognition of the importance of this phenomenon, are cooperating in the
definition of fretting and related terms for ASTM lists of standard terminol-
ogy. For example, the following are some of the definitions being considered
for mclusion in the ASTM Terminology Relating to Erosion and Wear
(G40-77):
Fretting—small-amplitude oscillatory motion, usually tangential, between
two solid surfaces in contact.
Fretting wear—wear arising as a result of fretting.
Fretting corrosion—a form of fretting wear in which corrosion plays a
significant role.^
Another effect of fretting is friction polymerization. This is the formation
at the interface of polymeric organic insulating solids which originate in air
pollutants that adsorb on contact surfaces. Contact lubricants may also pro-
duce frictional polymers. Palladium, rhodium, and other platinum group
metals catalyze this transformation.
Fretting is inherent in the operation of some electrical components, such
as wire spring relays that have contacts that wipe an area of some tens of mi-
crometres during operation. Separable connectors, although nominally at
rest during service, may be subjected to vibrations and mechanical shock
from nearby equipment and during transport. Magnetic forces induced by
a-c currents may produce displacements at connector-bus interfaces [1]?
Thermal excursions can cause fretting in connectors of as much as 100 ^m
due to differential thermal expansions and contractions of the structures to
which contacts are joined [2].
The frequency of fretting varies greatly. For example, a daily thermal ex-
cursion through a maximum and a minimum, as can occur in areas which
are poorly controlled environmentally, is 10~^ Hz. Magnetically generated
forces correspond to the a-c current in the circuit, a cycle producing two
pulses, such as 120 oscillations/s (240 wipes) for 60-cycle service. Mechanical
vibrations often occur at still higher frequencies.
Gold and high-gold alloys, such as solid 95Au-5Ni alloy and electrodepos-
its hardened with from 0.1 to 0.3 percent cobalt, are preferred contact mate-
rials because they are nearly inactive as catalysts and are chemically inert in
virtually all environments. The surfaces and wear products of such materials
remain metallic and conducting. However, wear-out of thin gold coatings
leads to high contact resistance when the base substrate is exposed.
The practical effects of fretting on electrical contacts seem first to have
been discovered in the telecommunications industry, which uses a very large
ÂSTM Committee B-4 would expand its definition to include the note that, in electrical con-
tacts involving nonnoble metals, fretting corrosion can cause rapid and substantial increases in
contact resistance.
•'The italic numbers in brackets refer to the list of references appended to this paper.
number of contacts. Frictional polymers were described in 1956 by Her-

mance and Egan [3] of Bell Laboratories, after they had been isolated from
palladium relay contacts. Fairweather et al [4] at the British Post Office
fretted nickel silver in air and found that contact resistance increased. At
about the same time, contact resistance measurement techniques began to be
used as a research tool for studying surface changes due to fretting corrosion
[5,6]. More recently, electrical effects of fretting in connectors have been
observed. For example, separable connectors with tin-plated contacts began
to be used commercially in television sets in the early 1970s, and difficulties
occurred because of fretting from thermal cycling when the equipment was
turned on and off. Escalation in the price of gold has stimulated the use of
alternative materials for contacts in electronic connectors and printed circuit
boards. In addition, the application of thm gold coatings where thick depos-
its formerly were standard is another recent development. In both cases there
is increased concern about the possible effects on contact resistance caused
by fretting [7-16].
Experimental Apparatus
Research studies of electrical contact fretting are facilitated by equipment
which permits materials and experimental parameters to be varied readily
and with which contact resistance may be determined. A new apparatus was
designed that has proven to be versatile, simple to use, and particularly
capable of operating at conditions like those which occur in separable elec-
tronic connectors. The unique features of this equipment are its ability to fret
small distances reproducibly and to measure contact resistance along the
wear track. The latter feature is important because, as will be shown later,
contact resistance change generally is not uniform from one end of the track
to the other, and such information is essential to permit the electrical conse-
quences of fretting to be interpreted correctly. The specimens are of a size
and shape which allow ready surface examinations.
Figure 1 illustrates the test apparatus. A stationary rider, mounted on a
leaf spring, is placed against a moving flat at loads from 5 to 500 g. The flat,
on a high-resolution slide table, is driven by a d-c stepping motor through a
2.5-cm-long micrometer screw having 16 turns per centimetre. The motor is
interfaced to a computer which controls the speed of the table and track
length with a resolution of 1.6 /urn. The entire apparatus is mounted on an air
table for isolation from external vibrations.
Figure 2 illustrates operation of the apparatus. The user gives commands
to a data acquisition system (DAS) computer by means of a teletypewriter
(TTY) as to the velocity, track length, and number of cycles in a run. This in-
formation is transmitted by the computer to a translator module circuit
which provides sequencing and switching logic for bidirectional control of the
stepping motor. The velocity is controlled within the range of 1.6 to 320 fim/s
Rider Specimen
Flat Specimen
Vise
Stepping Motor
FIG. lÂ fretting test apparatus.
Teletypewriter
Data Acquisition
System Computer
FIG. 2—A block diagram of the fretting apparatus.
in increments of 1.6 ^m for track lengths of 1.6 /^m to 13 mm. The stepping
motor moves in discrete identical steps. Thus, the motion of the table can be
intermittent or nearly continuous, depending on how fast the motor is pulsed,
as shown in Fig. 3.
Contact resistance determinations are made using a d-c four-wire dry cir-
t(msec)
Typical Step Response

FIG. 3—A typical step response of the slide table in the fretting apparatus.
cuit technique in accordance with the ASTM Standard for Measuring Con-
tact Resistance of Electrical Connections (Static Contacts) [B 539-70 (1975)].
The current and voltage leads are clamped to the rider, and a similar set of
contacts is wired to the flat. The bulk resistance of the volume of metal be-
tween the leads and the rider-flat interface is very small for metals, so only
contact resistance can be measured effectively. The open circuit voltage is 20
mV, with the current limited to 100 mA. These conditions do not cause phys-
ical changes of the junction, such as breakdown of insulating films or soften-
ing of contact asperities which might affect fretting. Higher-circuit voltages
can be used, but it has been shown [7] that under such conditions, the con-
tact resistance changes tend to be less.
The voltage drop of the contact is monitored by a programmable digital
voltmeter through an operational amplifier in the contact resistance cttcuit
(Fig. 2). The operational amplifier provides for zeroing and a tenfold in-
crease in the signal. The DAS samples contact resistance at preprogrammed
numbers of cycles. At these times the table is stopped at the beginning of a
wear track, and the DAS takes four contact resistance measurements and
stores them. The table then moves one step and another four measurements
are made. This procedure continues for the length of the track. Each set of
measurements is averaged by the computer and printed out on the TTY, as
shown in Fig. 4.
The rider specimens are hemispherically ended rods with a radius usually
of 1.6 mm, and the flats can be from about 5 by 5 mm to 5 by 10 cm in size. It
has been convenient to employ riders in the form of commercially available
mushroom-shaped small contact rivets, which can be fabricated from nearly
any ductile metal. Rod-shaped samples can also be fixed in the apparatus
and mated in a crossed position, with one member stationary and the other in
motion. The specimens may be coated by electrodeposition or other methods.
Investigations of Contact Fretting

Studies of the fretting of contact materials with the stepping motor-slide
table apparatus illustrate phenomena having considerable interest in the
field.
Friction Polymerization
Palladium rivets mated to a 5-^m-thick clad palladium flat were fretted
with a wipe amplitude of 80 /xm. The normal load was 50 g, and the flat was
coated with a thin layer of a polyphenyl ether contact lubricant. Lubricants
are employed with separable connectors for the purposes of reducing wear
during insertion and withdrawal and lowering friction. After only 10^ cycles,
contact resistance became unstable and rose rapidly as fretting continued.
MAY 8, 1980
RUN #140...SOLID PD RIDER VS 5UM PD FLAT

SAMPLE #7534... 20UM TRACK LENGTH. .. ir>9UM/SEC VELOCITY
40%RH...25 DEGREES C...TIME 1:23 PM
#STEPS 13...PULSES/SEC 100
CONTACT RESISTANCE FOR CYCLE # 1
3.83 3.87 5.01 5.78 5.26 5.75 5.82 5.82 5.67 6.13 6.54 6.93 7.02 7.16
CONTACT RESISTANCE FOR CYCLE t 10
3.49 3.61 3.64 3.68 3.63 3.71 3.75 3.91 3.91 4.00 3.97 4.08 4.22 4.61
3.65 3.74 3.74 3.81 3.79 3.87 3.83 3.93 4.03 4.15 4.26 4.50 4.65 5.02
4.07 4.17 3.96 3.96 3.91 4.08 4.09 4.17 4.16 4.29 4.37 4.52 4.77 5.28
CONTACT RESISTANCE FOR CYCLE » 100
4.51 4.68 4.73 4.89 4.90 5.06 5.13 5.19 5.16 5.46 5.82 6.76 7.97 9.17
FIG. 4—An example of a computer printout of contact resistance from a run.

exceeding 1 Q by W cycles. This is shown in Fig. 5 where contact resistance

is plotted at 1.6-/xm increments along the track. Although contact resistance
rose in all places as fretting continued, the increase was greatest at or near
the ends.
Optical examination after a run showed that a considerable amount of
black solid had formed on and surrounding the track. Under the scanning
electron microscope, observed at 20 kV beam energy (Fig. 6), this material
was highly charged, which indicated that it was insulating. The droplets on
the surface of the flat are the contact lubricant. Energy dispersive X-ray
analysis showed that the palladium on the flat was not worn through. The
foreign substance can therefore be inferred to have been a frictional polymer
which probably originated, at least in part, in the contact lubricant. Repeat
runs with samples that had been thoroughly degreased showed similar or
worse contact resistance behavior [17\, although the volume of polymer was
less. In this case, it is evident that there was sufficient organic pollution in
the room air to form the polymer which led to early contact failure. Pollu-
tants could have come from outgassing of surface finishes on furniture and
10
iAa^ SOg
^ i
SOuDi Wipe
A
1Hz
-DDDD D ^
Lubricated (Thin Coating)
«• 1 D
E
£
D
o a
O
a ,, D
^n
a -0°°
o
Â Og 105 C y c l e s o"
'^ 1^ ODOnDDODDDDnO
cc '^ A
A
A ^
•s A
a
A .AJO'Cycles ^..-
3 'âÂA
8 01 ÂAAA'^
^ o o OOOQ I D ' Cycles

„o°°°"°oooo
°oooooooOOOoooooooooo„„oooooooO°"
.001 1 1 1 1 1 r 1 r-*
0 10 20 30 40 50 60 70 80
Distance Along Track (^m)
FIG. 5—Variation of contact resistance along a track for increasing numbers of fretting
cycles. A solid palladium rider versus a clad palladium flat.
100 fim
^
100 jKm
FIG. ()—A worn rider (bottom) and flat (top) from a run with palladium contacts under the
test conditions given in Fig. 5 after 10^ cycles. The friction polymer is the charged (light colored)
material on and surrounding the wear spots. The scanning electron microscope numbers are the
distances between the adjacent white markers (20 kV).
walls or from electrical and mechanical apparatuses, which contained wire

insulation, potting compounds, and plastics.
The rise of contact resistance comes from the pileup of the frictional poly-
mer that is being continuously scraped from the surface as it is generated.
The role of wiping has been postulated [18] to create fresh surfaces, which
are necessary for rapid transformation of organic compounds that adsorb on
them.
Virtually no frictional polymer appears when gold-gold contacts are
fretted [18], and contact resistance remains low and stable. This is illustrated
in Fig. 7, which is a plot of the maximum contact resistance determined at
each measurement interval for unlubricated gold-gold (Curve d) and for pal-
ladium-palladium (Curve a) contacts fretted with a 20-jum wipe. The palla-
dium-gold (Curve c) and gold-palladium (Curve b) systems are also shown,
and their contact resistance is similar to that of the all-gold pair. The reason
for this behavior is that gold transfers readily to the harder palladium mate-
100
Material and Thickness (fxrii)
Rider Flat Underplate
10 Pd (Solid) Pd.clad(5)
Co-Au plate (9) on Cu Pd, clad (5)
Pd (Solid) Co-Au plate (3.3) Ni (2.5)
(A
Au (Solid) Co-Au plate (0.6) Ni (2.5)
E
1 -
0)
u
c
flj
(fl .1
50g
w
20|:im wipe
cc Not lubricated
o
™ .01
c
o
O
.001 -
10^ 10^ 10^ 10-^ 10' 10- 10'

Cycles
FIG. 7—The contact resistance versus the fretting cycles for palladium-palladium, gold-
gold, and combinations of gold with palladium contacts.
rials, thereby effectively transforming them into a gold-gold system. It has

empirically been found that palladium-palladium contacts m hardware are
much more likely to give unstable contact resistance than palladium-gold
contacts, and examples of the successful use of mixed material pairs have
been described for relay contacts [19] and for separable connectors [20].
Since frictional polymer tends to pile up at the ends of the track, the track
length would be expected to be a factor in the reliability of fretting contacts.
This is borne out by a study summarized in Fig. 8, which gives the numbers
of cycles required to attain 100 mfi from an initial value of about 5 mQ as a
function of track length in a series of runs with polyphenyl ether-lubricated
contacts. The maximum values of contact resistance are plotted. ReUability
is dependent on the wipe distance. The rise of contact resistance occurred
after only about one hundredth the number of cycles as the track length was
increased from 10 to 100 /im. A similar tendency has been found [17] for un-
lubricated contacts. Thus, fretting over large amplitudes is worse than fret-
ting on small tracks.
>10^^ Oo
10^ \ 50g
Lubricated (Thin Coating)
o\
a
E
8
^
\°
c
"<5
C ,
<10*
0
*^
(A
0)
u
>• \ Q
o
CO o\.
10=* 1' i 1 1 • 1 1 1 1 1 1 1 1 1 p f 1 1 1
50 100 150 200

Track Length (jitm)
FIG. 8—The fretting cycles required to attain 100 mG versus the track length for palladium-
palladium contacts coated with a thin lubricant film.
Fretting Corrosion of Base Metal Contacts

Base metals are prone to a rapid increase in contact resistance. This char-
acteristic is illustrated in Fig. 9 for systems of nickel and 60Sn-40Pb alloy
fretting on themselves, where the maximum values of contact resistance
along the track are plotted. The fall in contact resistance immediately after
the commencement of movement is due to the wear-out of natural oxide films
on these materials, but after only a few tens to hundreds of cycles, the contact
resistance begins to increase, attaming levels in excess of 10 Q within several
thousand cycles. This condition is due to the rapid formation of insulating
oxides at the interfaces, particularly near the ends of the tracks.
Nickel is used in screw terminals for high-temperature applications, which
are successful only when the screw terminals are securely bolted to preclude
frettmg movements. Solder-plated contacts are employed on the contacts of
low-cost separable connectors and printed circuit boards where high reliabil-
ity and long life are not prime requirements, as in consumer applications.
100 1 ,
50g ojL j
2Q\im wipe ^ fyj i
E
4Hz
Not lubricated /
/\W
V /I
1
/ 1/
o
0) 1-
o
c
(0
A /^b
(A
0)
K
*^
o
10
J .01 S. "'^fmô—y
^N- J^ Material and Thickness (f^m)
Rider vs Flat Substrate

.001 a Ni (Solid) Ni. Plate (2.5) Cu
b 60Sn40Pb. 60Sn40Pb. Cu
Plate (25) Plate (25)
1' 1 1 1 1 1 1
10' 10 10^ 10- 10' 10^ 10'
Cycles
FIG. 9—The contact resistance versus the fretting cycles for typical base metal contacts,
nickel-nickel, and 60Sn-40Pb-60Sn-40Pb alloy.
High-contact forces and other design features are employed to minimize the
possibility of fretting.
Metallic wear debris from fretting does not contribute significantly to con-
tact resistance. The increase of contact resistance is due to oxidation of the
debris, and since this is a kinetic process, it might be expected that contact
resistance changes would be related to fretting frequency. Such relationships
have been found, as with copper versus copper in Fig. 10, which is a plot of
(0
E
|.01
0)
o
c
m
DC
u
(0
C
O
U
.001
100g
20fim wipe
Not lubricated
Cu-Cu specimens
1
100 1000 10,000
Cycles
FIG. 10—The effect of the cycle rate on the contact resistance in the fretting corrosion of solid
copper-copper contacts.
the maximum values of contact resistance in the tracks from runs at several
frequencies. The lower the cycle rate, the smaller the numbers of movements
needed to attain a given increase of contact resistance. More time is available
for oxidation of copper exposed by wear when fretting is slow than when it is
fast.
Wear-Out of Noble Metal Contacts

It has been shown that thick gold-coated contacts maintam low and stable
contact resistance during fretting, while the contact resistance of base metals
quickly rises to high values under the same conditions. Obviously, the inter-
mediate case is that of thin gold coatings, from which contact resistance is
stable until the coating wears through. This is illustrated in Fig. 11, which
compares the contact resistance from three systems with (a) solid copper
members, (b) solid gold versus 0.05-/im cobalt-gold-plated copper, and (c)
solid gold versus 0.6-/xm cobalt-gold plate with a nickel underplate on cop-
100
Material and Thickness ( f i m )
Rider
10 Cu (Solid) Cu (Solid)
Au (Solid) 0>-Au plate (.05) on Cu
Au (Solid) Co-Au plate (0.6)
on Ni underplate (2.S)
E onCu
£ 1-
o
4)
U 50g
C
(0 20^m wipe
I 4 Hz
Not lubricated
s
2 .01
c
o
u
.001
, 1 1 1
10' 10' 10"
10° 10^ 10^ 10^
Cycles
FIG. 11—The contact resistance versus the cycles of movement for (a) base metal and for (b)
thin and (c) thick gold contact pairs.
per. In the second material combination, (b), wear-through of the thin co-
balt-gold plating occurs at about 1000 cycles. However, contact resistance
does not rise very high, but remains at an intermediate level.
The explanation for this behavior can be seen in Fig. 12, in which worn
0.05-^m cobalt-gold-plated copper flats were observed with the scanning
electron microscope from runs at the same condition at successively greater
numbers of cycles of fretting. After 10^ cycles the surface is intact, but at lO"*
and 10^ cycles wear has progressed to the point where some copper appears
on the track, in loose debris, and transferred to the rider. Finally, after 10^
cycles, an equilibrium condition is achieved in which the surface is relatively
smooth and consists of a mixture of gold, copper, and copper oxide. Both the
rider and the flat are the same, and an estimated 10 percent of the area of the
fretted surfaces is coated with gold. Higher power examination shows cracks
to be present in the copper surface, which probably is due to delamination
wear [21].
In this example, contact resistance did not rise catastrophically to high
values when the gold on the flat wore through. This is attributable to the
solid gold rider, which continued to provide gold to the flat by transfer, al-
though it is clear that copper transferred to the rider as well. Had the rider
been copper with a thin coating of gold, the contact resistance behavior
would have been intermediate to that in Curves a and b of Fig. 11. An under-
standing of contact resistance changes can be obtained by surface analysis in
which wear, transfer, and the composition of the interface material is deter-
mined.
Smninaiy and Conclasions

Fretting can cause undesirable increases of contact resistance in separable
connectors, relays, switches, circuit breakers, and other contact-containing
components. Fretting is caused by vibration, mechanical shock, thermal
fluctuations, and magnetic forces to which conductors are subjected. Small
displacements on contact closure in some relays produce frettinglike degra-
dations. Contact materials in common use include gold and gold alloys; some
base metals such as tm plate, tin-lead coatings, and nickel; and platinum
group metals, especially palladium. Noble metals are commonly employed as
thin coatings on base substrates.
The contact resistance of gold contacts remains substantially constant dur-
ing fretting because of its nobility. However, insulating oxide films appear at
the interface of base metals and gold-coated base metal substrates when the
gold wear out. Platinum group metals catalyze the formation of frictional
polymers from organic materials in their vicinity, such as air pollutants. The
rise of contact resistance originates in these organic and inorganic insulating
surface-contaminating solids.
mm
c
CO
(PDVSA Los Teques) pursuant to License Agreement. No further reproductions auth
A new research apparatus for studying fretting has been described in

which a stepping motor causes small oscillatory movements of a slide table to
which a specimen is mounted and against which the opposing contact is
loaded. Contact resistance is determined at intervals along the wear track by
a data acquisition system computer. Examples of the use of this equipment
to model fretting of electrical contacts have been given, including friction
polymerization, fretting corrosion of base metals, and wear-out of gold
coatings on base substrates. Determination of the surface changes of contact
materials caused by fretting, using optical, scanning electron microscope,
and energy dispersive X-ray analysis techniques, has been useful in explain-
ing the causes of contact resistance change.
Important findings from this study include the following. Contact resis-
tance increases caused by fretting are at a maximum at the ends of the track
because of the pileup of frictional polymer or oxide debris. The longer the
track is, the smaller is the number of cycles needed to produce a given in-
crease in contact resistance. The slower the frequency of movement, the
fewer the cycles needed to cause the contact to fail. Metallic transfer from
fretting and changes in the surface composition, which occur as a result of in-
teraction with the environment, control the contact resistance.
Acknowledgment
M. H. Drozdowicz designed and constructed the fretting apparatus and
conducted some of the studies described in this paper. The assistance of N.
Cochran in this work is appreciated.
Discussions with E. S. Sproles and F. E. Bader have been helpful.
References
[1] Johnson, J. L. and Moberly, L. E., Proceedings of the Holm Conference on Electrical Con-
tacts, Illinois Institute of Technology, Chicago, 111., 1975, pp. 53-59.
[2] Antler, M. and Drozdowicz, M. H., Wear, Vol. 74, No. 1, 1982, pp. 27-50.
[3] Hermance, H. W. and Egan, T. F., Proceedings of the Electronic Components Sympo-
sium, Washington, D.C., Engineering Publishers, New York, 1956, pp. 85-93.
[4] Fairweather, A., Lazenby, F., and Parker, A. E., Research Report. No. 20986, Engineer-
ing Department, British Post Office Research Station, DoUis Hill, England, 1964.
[5] Fenner, A. ] . , Wright, K. H. R., and Mann, J. Y., Proceedings of the International Con-
ference on Fatigue of Metals, 1956, p. 388.
[6] Halliday, J. S. and Hirst, W., Proceedings of the Royal Society, Vol. A236, 1956, p. 411.
[7\ Bock, E. M. and Whitley, J. H., Proceedings of the Holm Conference on Electrical Con-
[8] Crossland, W. A. and Murphy, P. M. K., I.E.E.E. Transactions on Parts, Hybrids, and
Packaging. Vol. PHP-10, No. 1, 1974, pp. 64-73.
[9] Freitag, W. O., Proceedings of the Holm Conference on Electrical Contacts, Illinois Insti-
tute of Technology, Chicago, 111., 1976, pp. 57-63.
[10} Garte, S. M., Proceedings of the Holm Conference on Electrical Contacts, Illinois Institute
of Technology, Chicago, III., 1976, pp. 65-70.
[7/] Braunovic, M., Proceedings of the Holm Conference on Electrical Contacts, Illinois Insti-
[12] Kongsjorden, H., Kulsetas, J., and Sletbak, J., Proceedings of the Holm Conference on
Electrical Contacts, Illinois Institute of Technology, Chicago, 111., 1978, pp. 87-92.
[13] Abbott, W. H. and Campbell, W. E., Proceedings of the Holm Conference on Electrical
Contacts, Illinois Institute of Technology, Chicago, 111., 1978, pp. 359-362.
[14] Abbott, W. H., Proceedings of the Holm Conference on Electrical Contacts, Illinois Insti-
[15] Theisen, P. J. and Forsell, K. A., Proceedings of the Holm Conference on Electrical Con-
[16] Gyurina, D. and Smith, E. F., III, Proceedings of the Holm Conference on Electrical Con-
[17] Antler, M. and Sproles, E. S., "Effect of Fretting on the Contact Resistance of Palla-
dium," J.E.E.E. Transactions on Components, Hybrids, and Manufacturing Technology,
Vol. 5, No. 1, 1982, pp. 158-166.
[18] Hermance, H. W. and Egan, T. F., The Bell System Technical Journal, Vol. 37, 1958, pp.
739-777.
[19] Keefer, H. J. and Gumley, R. H., The Bell System TechnicalJoumal, Vol. 37, 1958, pp.
778-814.
[20] Napp, D. T., Proceedings of the Fifth Symposium on Plating in the Electronics Industry,
American Electroplaters' Society, New York, 24-25 March 1975, pp. 5-27.
[21] Waterhouse, R. B., Wear, Vol. 45, 1977, pp. 355-364.
p. L. Ko, 1 /. H. Tromp, ^ and M. K. Weckwerth^
Heat Exchanger Tube Fretting

Wear: Correlation of
Tube Motion and Wear
REFERENCE: Ko, P. L., Tromp, J. H., and Weckwerth, M. K., "Heat Exchanger Tube
Fretting Wear; Conelation of Tube Motion and Wear," Materials Evaluation Under
Fretting Conditions, ASTM STP 780, American Society for Testing and Materials, 1982,
pp. 000-000.
ABSTRACT: A typical steam generator in a nuclear power station consists of about 4000
tubes which form the boundary separating the light water in the secondary circuit from
the primary circuit water. The integrity and the life expectancy of these tubes are
therefore of prime concern to the designers.
One of the several tube failure mechanisms, which include corrosion and fatigue, is
fretting wear due to flow-induced tube vibrations. The tube-tube support clearaftce
needed for design considerations allows periodic separation and impacting between the
tube and the tube support hole and hence ready removal of loose wear particles by the
fluid flow. It has been found that the combined rubbing and impacting motion, together
with the periodic separation of contacting surfaces, all contribute to an accelerated wear
process.
A tube-fretting test apparatus has been developed to study the effect on tube wear of
various parameters, such as tube-tube support interaction, materials, temperature, and
clearance. Tests have been conducted in water at room temperature and at steam-
generator operating temperature (265°C). Recently the authors have looked into the
more fundamental aspect of tube fretting wear mechanisms. Some of the worn surfaces
were further studied by a surface analyzer interfaced with a computer and by a scanning
electron microscope. The results from these analyses were correlated to other tube fret-
ting parameters, namely, tube motion, clearance, and impact forces at the support. We
have found that the effects of these parameters on wear are interrelated. The understand-
ing of these interrelated effects is of great importance in predicting long-term tube wear.
This paper will describe our fretting test facilities and present the results from recent
studies on wear correlations.
KEY WORDS: fretting, wear, heat exchanger tube, flow-induced vibration, impact
force, surface analysis
'Research engineer, technologist, and technician, respectively, Atomic Energy of Canada

Limited, Chalk River Nuclear Laboratories, Chalk River, Ontario, Canada, KOJ IJO.
86
KO ET AL ON HEAT EXCHANGER TUBE FRETTING WEAR 87
In recent publications [1-3],^ the importance of vibration analyses of heat

exchangers and the understanding of tube fretting mechanisms have been
amply stated. This is particularly true for nuclear steam generators. In Cana-
dian Deuterium Uranium (CANDU) nuclear power plants, the tubes sepa-
rate the primary circuit heavy water from the secondary circuit light water.
The failure of one of these tubes would result in a power station outage and
costly repairs.
Tube failures in steam generators or heat exchangers may be caused by
one or a combination of several mechanisms, namely, corrosion, fatigue, and
impact fretting. The last of these mechanisms is a result of the tube impact-
mg and rubbing on the tube supports or, in severe cases, on adjacent tubes.
Flow-induced tube vibration can cause dynamic interactions between the
tube and the tube support, such as reciprocating rubbing motion. These in-
teractions can also be in the form of periodic impacting, since tube-tube sup-
port clearances are needed for assembly considerations.
Flow-induced vibration mechanisms and the application of vibration
analyses to steam generator and heat exchanger designs are discussed in the
open literature, for example, by Chenoweth and Kistler [4], Connors [5],
Shin and Wambsganss [6], and Pettigrew et al [1]. However, knowing the
vibration amplitudes at the design stage or from field measurements cannot
answer the questions of how much vibration is acceptable or how long the
tubes can last. The answers to these questions become less predictable when
new materials and new tube-support geometries are considered, as com-
parison with proven designs will not be possible.
To answer these questions, we need a better understanding of wear
mechanisms and wear rates so that some relationship between vibration and
wear damage can be established. Although the topics of fretting wear and
impact wear have been studied for a long time [7,8], it is only during the past
few years that papers concerning tube fretting wear have been published.
Tube fretting studies have been underway at the Chalk River Nuclear
Laboratories for nearly 10 years. Basically, our studies may be divided into
three parts: (1) experimental investigation to collect fretting wear data and to
understand the tube fretting mechanism; (2) development of analytical
techniques to predict tube-tube support dynamic interaction, that is, the im-
pact forces at the supports; and (3) correlation of the tube wear and tube mo-
tion parameters, such as the wear-impact force relationship. The objective is
to use the support impact force as the parameter to link the predicted wear
damage to the vibration amplitude calculated from a vibration analysis.
The present paper (1) reviews the tube fretting experiments and some of
the results; (2) describes techniques to analyze worn surfaces and support im-
pact forces; (3) discusses the relationship of some mechanical parameters,
namely, the type of tube motion and the tube-tube support clearance and ex-
citation force, with support impact forces and tube wear; and finally (4)
shows the correlation between tube wear and support impact forces.
Tabe Fretting Experiments

Fretting is generally considered to be one form of the wear process, which
consists of the following stages: dispersion of surface film by the oscillating
movement; adhesion, plastic deformation, and metal transfer between the
contacting surfaces; for some materials, oxidation of the wear particles and
formation of an intermediate zone; production of loose wear particles due to
the fretting action; and, finally, repetition of the cycle.
In tube fretting the contacting surfaces may separate periodically because
of the existence of a clearance between the tube and the tube support.
Depending on the flow conditions and, hence, on the type of vibration
mechanisms—for example, fluidelastic instability, periodic wake shedding,
or random turbulence—the tube response may be periodic or narrow-band
random. The tube may rub and impact against the tube support, or it may
simply rub with very little or no separation, as in true fretting. The latter is
the result of high-static normal force and low-amplitude tube vibration, such
as when the tube is quite firmly pressed against one side of the tube support
hole because of a combination of tube-tube support eccentricity, tube bow-
ing and cross-flow drag. However, it is the combined rubbing and impacting
tube motion that usually causes severe wear damage and is therefore of
special concern.
So far we have considered only the periodic type of tube response in tube
fretting studies. Tube motions are described by the ratio of two orthogonal
displacement or force components. The motion may vary from unidirectional
to whirling. Experiments on a tube bank subjected to cross-flow show that
the ratio of these orthogonal components is in the range of 2 to 6 [5]. During
interaction between a tube and its supports, the tube motion would also de-
pend on the tube-tube support clearance, the magnitude of the excitation
force, and the tube-tube support eccentricity.
Tube Fretting Apparatus

In some fretting apparatus, the specimen is excited either by a cam-type
mechanical device or by an electromagnetic shaker. Both of these devices,
when used in tube fretting experiments, have the disadvantage of restricting
the free movement of the tube during the tube-tube support interaction.
A simple excitation device, consisting of a light-weight motor housing and
two stepper motors driving two out-of-balance masses, was developed at the
Chalk River Nuclear Laboratories for tube fretting studies (Fig. 1). A de-
tailed description of this exciter has been given elsewhere [9]. It is versatile;
and it can be attached to single-span or multispan tubes. Figure 2 shows a
OUT-OF-BALANCE STEPPER
MASSES MOTORS
VOIDS
f. = (m , + nij ) ru;^
'max
fx = ( m I - m ,) r o ; ^
max
f = m , ro)^ s i nojt - m 2 r^^
s i n(cut-180 )
FIG. 1—Schematic view of a vibration generator.
single-span tube fretting test rig. It consists of a single-span cantilevered tube

which is vibrated against a tube-support: specimen. Some test rigs are open to
the atmosphere for operation at room temperature. Others are enclosed in
autoclaves for operation at high temperatures in controlled environments.
Figure 3 shows our high-temperature test facility. Excitation is provided by a
vibration generator attached to one end of the tube. When the two motor-
driven eccentric masses are adjusted, the ratio of the two orthogonal excita-
tion force components can be controlled. The maximum values of these two
components are/x^^x and/^„^, where/^âx ^ /xmax- Thus, for/j,„^//:t™ax - °°.
the tube motion is of the impact type, whereas, when/y„ax//a:„aj — 1, the tube
has the tendency to whirl and the tube-to-tube-support contact would consist
TUBE AND TUBE

SUPPORT SPECIMENS
FIG. 2—Room-temperature test rig.
primarily of rubbing. Again, the actual tube motion during tube-tube sup-
port interaction is influenced by the tube-tube support clearance and the
magnitude of the excitation forces.
Effect of Type of Tube Motion on Wear

It has been shown [3] that a unidirectional impact force alone would cause
very little tube wear. In these earlier tests, the vector sum of the two or-
thogonal excitation force components, fy^^^ and fxmax< was kept constant
while the ratio/yn,aj//c„ax was varied. The tube wear from these tests was then
normalized with the root-mean-square value of support impact forces and
plotted against the ratio of the two orthogonal force components, as shown in
Fig. 4. The trend shows that the wear rate decreases as the tube motion
changes toward the impact type.
Effect of Tube-Tube Support Clearance on Wear

Tube wear has been found to increase as the tube-tube support clearance
is increased and the excitation level is kept constant [3]. Figure 5 shows an
example of this. The tests were conducted in pressurized water at 265°C and
were subjected to the same excitation force. Further investigation revealed
that the increase in clearance was usually accompanied by an increase in the
support impact force and, under some conditions, a change in the type of
50 X 10-3
20.0 -
10.0
SUPPORT IMPACT FORCE RATIO, F / F ,
FIG. 4—Effect of the ratio of two orthogonal force components on wear. All data are for 10^
cycles; excitation frequency is 30 Hz.
0,1 0.2 0.3 0.4
TUBE /SUPPORT RADIAL CLEARANCE mm
FIG. 5—Effect of tube-tube support clearance on wear for Incoloy 800 tubing versus carbon
steel support specimens in pressurized water at 265 °C.
tube motion. These changes are illustrated in Fig. 6, which shows a series of
oscilloscope traces of the x-y displacements and of the filtered impact forces
of a single-span tube subjected to the same excitation forces but with dif-
ferent tube-tube support clearances. It is quite clear from these pictures that
the effect of these parameters on wear ought to be considered collectively.
This will be discussed in a later section.
Surface Analysis of Worn Areas
Surface Profile
A surface analyzer is used to study the surface profile of worn areas. A
computer is used to obtain a statistical distribution of wear depths as well as
an average wear depth. The pattern of wear depth around the tube cir-
cumference can be traced by taking multiple longitudinal surface profiles of
z
in
f
•s
V9 C
h-
^
3
a.
• >
—
u.>^
1a,
-o .1
,
V3
1 a,
3
O
QC ^
•a
u. ^^
5 ^
<
a. b
1
X
• "
i 1
a.
a. •
.>
— 3
rsi
s
DC
lU
t-
_i •
>.
the worn tube. The total volume of wear can also be estimated from these
surface profiles. If the density of the material is known, this volume of wear
may be used to verify the weight-loss measurements or vice versa.
Figure 7 shows examples of some of these surface profiles and the cor-
responding plots of the wear depth distribution. It appears that those with
impact-type motion show a relatively smoother worn surface. However,
secondary electron photomicrographs of the worn surface (Fig. 8) reveal little
difference between wear produced primarily by rubbing motion and that pro-
duced by impacting motion; both have a matte appearance. This suggests
that the primary cause of wear is the same; the principal mechanism is shear.
In the case of impact motion, which includes oblique impact, only a small
sector around the tube circumference would be affected, and the shear com-
ponent would also be smaller. Hence the wear rate is lower.
Cross-Sectional Profile
Figure 9 shows several cross-sectional profiles of some worn tube
specimens. These were plotted from a sequence of average depth of wear
from the longitudinal surface profile scans. Inside each computer plot is
shown the corresponding oscilloscope trace of an Fy-F^ plot of filtered sup-
port impact forces. It is interesting to note that the depth of wear around the
tube circumference varies according to the shape described by the two or-
thogonal force components. For a nearly circular motion with the tube sup-
port impact force distributed as shown in Fig. 9a, the wear is nearly uniform
around the circumference. On the other hand, when the tube motion is
almost unidirectional with tube-tube support interaction limited to a narrow
sector, as shown in Fig. 9d, wear occurs only along two opposite sectors. The
maximum wear depth is much less than those shown in Fig. 9a and b,
although the magnitude of the larger impact force component, that is, Fy, is
about the same in all four cases. This shows that tube fretting wear is not
so much affected by the peak force component as it is by the type of tube
motion.
In our earlier studies, wear was expressed as an equivalent uniform wear
depth from weight-loss measurement. However, in real tube wear, mostly
because of tube eccentricity and flow conditions, the wear is not uniform; it is
generally deeper on one side than on the other. The present analysis using
surface profiles, therefore, provides a more realistic interpretation of the
wear pattern and hence a better understanding of the relationship between
tube wear and other motion parameters.
Support Impact Force
Measurement of Support Impact Forces

The stationary tube support specimen in the form of an annular ring is
Copyright by by
held fourInt'l
ASTM piezoelectric-type force
(all rights reserved); transducers
Fri Jan 27 22:01:44with
EST a2012
preload (Fig. 10).
m •9
-M
CD 3
3
a: C!
•a
V •a
-J -c
2 a.
< •§
s )Q<
^ s^
o«
?
•3
c
0
&
!»
fc
o
*iiSN3a AiniBvaoud "" Nl Hid30 B»3« o
<a
«
•a
c
a
VI
^
ja
>.
C9
z
I-
o
2
2
>•
-J
I
i
A1ISN30 lini "i«i Nl Bidao W3ti
These transducers are located 90 deg apart on the outside surface of the ring
specimen. Excitation of the test tube assembly causes the tube specimen to
impact or rub against the inside surface of the annular support specimen.
The impact forces, which include components resulting from rubbing and
normal or oblique impacting, are picked up by the four transducers. The
force signals are read out as root-mean-square values and are also recorded
on tape for further analysis.
This technique for impact force measurement has been verified. A small
stainless steel sphere was dropped onto the inside surface of an annular sup-
port specimen. The measured impact force was compared with the force
predicted by an impact force equation derived from Hertz's elastic theory.
The results have been reported elsewhere [10].
Analytical Prediction of Impact Forces

Support impact forces can be measured during fretting tests. In steam
generators and heat exchangers, however, analytical techniques are needed
to estimate these forces from flow and vibration information. A finite-
element computer program called VIBIC [11] was developed to predict tube
motions and support impact forces of a single-span tube, and later extended
to simulate multispan tubes [12]. The model is essentially a time-domain
numerical simulation of the dynamics of heat exchanger tube-tube support
mteractions. Comparisons between the experimental and the simulated
results show that the computation technique is able to predict fairly accu-
rately the tube-tube support interaction and dynamic response of a multi-
span tube with intermittent contacts at the supports.
Statistical Analysis of Force Signals

The root-mean-square value of the support impact forces is easy to obtain
and has been found to be quite adequate for some comparison purposes,
such as the material combinations and short-term wear rates versus the im-
pact forces, while other parameters remain constant. For instance, it has
been shown [9] that for short-duration tests where the tube-tube support
clearance does not change significantly during the test, the wear rate in-
creases fairly linearly with impact forces. However, when other parameters,
such as tube-tube support clearance and tube motion, start to change, the
root-mean-square value alone is not sufficient to correlate accurately the
wear rate, and a correlation involving all of these parameters as well as a
more detailed analysis of the force signal is needed.
Figure 11 shows an example of a force histogram obtained using a digital
computer. The percentage calculation was obtained by dividing the full force
range (including both positive and negative) into 256 points. The number of
bars in the histogram, and hence the size of individual force component
-s:
a.
O
Si
3
•I
00
(0)
EXCITATION FORCE
FIG. <)—Cross-sectional profiles of four worn tube specimens. The large circular traces are a
cross-sectional view of the worn tube specimens, and the inside oscilloscope traces are filtered
F-F^ plots of tube support impact forces.
ranges, can be arbitrarily chosen. The histogram provides a statistical indica-

tion of the time distribution of various force levels during tube-tube support
interaction. This approach can be used to isolate the force components that
may be the primary cause of wear.
Wear-Impact Force Correlation

Figure 12 shows the probability density and hence the shape of the
distribution of four impact force signals corresponding to those shown in Fig.
9. Each impact force signal is the vector sum of two instantaneous impact
forces recorded orthogonally at the support. The resultant force signal is then
FIG. 10—Close-up view of a tube support specimen and transducers.
digitized and statistically analyzed. The percentage calculation is based on a

full range of 128 points. It is quite understandable that an impact motion or
a near impact motion (Fig. 12c and d) would have a large distribution of near
zero force components, as the time of impacting contact would constitute
only a small fraction of the vibration cycle. By subsequently plotting the
probability density of various force ranges against the wear rate, a trend was
obtained. Figure 13 shows an example of this type of plot. The wear rate ap-
pears to increase fairly linearly with the increase in the percentage of 10 to
20-N force components and decreases with the 0 to 4-N force components. A
least squares fit of these points gives a correlation coefficient of about 0.93 in
both cases. The correlation has encompassed various type of excitation rang-
ing from pure impacting to pure rubbing. Thus it appears that the present
technique, when perfected, would take into consideration the effect of tube
motion on wear. Previously, when the root-mean-square values of the two
orthogonal forces were used, only those data with a similar type of excitation
motion could be correlated.
The analysis has shown that high wear is not caused by some high-force
-36 -24 -12 0 \2 24 6D
IMPACT FORCE L E V E L , N
FIG. 11—An impact force histogram.
components that have a low probability of occurrence, such as in the case of

impact motion; rather, the high probability density of some intermediate-
range force components, which is usually the result of rubbing motion, ap-
pear to be the primary cause for high wear.
The effect of tube-tube support clearances on wear has been observed ex-
perimentally. There the type of tube motion changes as well as the support
impact forces. Although we have not yet analyzed these force signals
statistically, we believe that the same technique could be applied here. It is
likely that this technique will eliminate the complicated interrelationship
among the changes due to clearance, tube motion, and excitation force.
Conclusion
Fretting wear due to flow-induced vibration is one of the mechanisms
which could cause tube failures in steam generators or heat exchangers.
While a properly chosen material combination would minimize the wear
damage, the support impact force due to the dynamics between tube-tube
support interaction may be the key to translating the results of tube vibration
analyses into wear prediction and hence the prediction of tube life. The type
of tube motion, the tube-tube support clearance, and the excitation force all
affect wear rate. These parameters also interact with each other. A statistical
(a) MAINLY RUBBING (b)
fy/fx = 1 70_ EXCITATION FORCE

COMPONENTS f y / f x = 3
3.0 -
6 0
t
- 50
f
2.0 40 1
'1
1 3.0 J
II 1,, I ,. ,'V..
L II
J 1 II 1
1 0 1
1 "i; 1 , 2.0 1
1 ',; 1
1
1 Oj-
0
1 , 5.0
1 I 1
10.0
1 1 .
15 0 20 0
1 , '•
25 .0 0
1 . 1 1 1 . 1 >
5.0 10.0 15 0 20 0 25 0
IMPACT FORCE LEVEL,N IMPACT FORCE LEVEL, N
12.0
=> 3.0
S 3.0
0 5 0 10.0 15.0 20.0 25.0 0 5.0 10 0 15 0 20 0 25.0
IMPACT FORCE LEVEL, N IMPACT FORCE LEVEL, N
(C) (d) MAINLY IMPACTING
FIG. 12—Comparison of the support-impact-force distributions due to four types of tube mo-
tion; the vector sum of the excitation force components, F^ and Fy, remained constant.
2^
2 -z.
o w
o CO
o £
o <
IjJ Z
o—
< ~
^2
Q.
WEAR,/im/IO* CYCLES
FIG. 13—The correlation between wear and the distribution of force components in a force
signal for Incoloy 800 tubing versus Inconel 600 support in water at room temperature (10 to 20
N), Each number in a circle indicates the ratio of two orthogonal excitation force components.
analysis of the resultant impact force signal would include the effects of these
interactions. Therefore, a simple correlation of the force components and the
wear rate is possible.
Secondary electron photomicrographs of the worn surfaces show that the
primary cause of wear is the same in all types of tube motion; that is, the
principal mechanism is shear. The results from the surface analysis and the
force analysis show that high wear is not caused by some high force com-
ponents that have a low probability of occurrence, such as in the case of
impact-type motion; rather, the high probability density of some inter-
mediate-range force components, which is usually the result of rubbing mo-
tion, is the primary cause of high wear.
Acknowledgment
The authors would like to thank P. J. Smith of the Chalk River Nuclear
Laboratories for his assistance in performing the experimental tests. They
are also grateful to M. J. Pettigrew for very helpful discussions.
References
[1] Pettigrew, M. J., Sylvestre, Y., and Campagna, A. O., Nuclear Engineering and Design,
Vol. 48, 1978 pp. 95-115 (also Atomic Energy of Canada Limited Report AECL-6106).
[2] Pettigrew, M. J. and Ko, P. L., "A Comprehensive Approach to Avoid Vibration and Fret-
ting in Heat Exchangers," ASME PVP-41, presented at the 1980 American Society of
Mechanical Engineers Pressure Vessel and Piping Conference, San Francisco, 13-15 Aug.
1980.
[3] Ko, P. L., Proceeding, ASME Heat Transfer Division, Vol. 9, American Society of
Mechancial Engineers Winter Annual Meeting, Chicago, Nov. 1980, pp. 11-18.
[4] Chenoweth, J. M. and Kistler, R. S., "Research on Heat Exchanger Tube Vibration,"
USERDA-HIRI Report, SAN/1273-1, UC-93, prepared for Energy Research and Develop-
ment Administration, Heat Transfer Research Inc., Alhambra, Calif., Feb. 1977.
[5] Connors, H. J., Proceedings, Symposium on Flow-Induced Vibration in Heat Exchangers,
American Society of Mechanical Engineers Winter Annual Meeting, New York, 1 Dec.
1970, pp. 42-56.
[6] Shin, Y. S. and Wambsganss, M. W., "Flow-Induced Vibration in LMFBR Steam
Generators: A State-of-the-Art Review," Report ANL-75-16, Argonne National Labora-
tory, Argonne, 111., May 1975.
[7] Waterhouse, R. B., Fretting Corrosion, Pergamon, Oxford, 1972.
[8] Engel, P. A., Impact Wear of Materials, Elsevier, The Netherlands, 1976.
[9] Ko, P. L., A.S.M.E. Transactions, 7o«ma/o//Vrarare Vessel Technology, Vol.-101, May
1979, pp. 125-133.
[10] Smith, P. J., "A Study of the Impact Response of a Piezo-Electric Type Miniature Force
Transducer," Atomic Energy of Canada Limited Internal Report, CRNL-2133, Ontario,
Canada, Feb. 1981.
[11] Rogers, R. J. and Pick, R. J., Nuclear Engineering and Design, Vol. 44, No. 2, Nov. 1977,
pp. 247-253.
[12] Ko, P. L. and Rogers, R. J., "Analytical and Experimental Studies of Tube/Support In-
teraction in Multi-Span Heat Exchanger Tubes," Paper F9/4, Transactions, Fifth Interna-
tional Conference on Structural Mechanics in Reactor Technology, West Berlin, Germany,
13-17 Aug. 1979. A detailed version of this paper is published in Journal of Nuclear
Engineering and Design, Vol. 65, 1981, pp. 399-409.
V. S. Agarwala, ^ K. Y. Kim, ^ and S. Bhattacharyya^
An Electrochemical Method for

Investigating Corrosion in Rubbing
Surfaces
REFERENCE; Agarwala, V. S., Kim, K. Y., and Bhattacharyya, S., "An Electrochem-
ical Method for Investigating Corrosion in Rubbing Surfaces," Materials Evaluation
Under Fretting Conditions, ASTM STP 780, American Society for Testing and Materi-
als, 1982, pp. 106-124.
ABSTRACT; An electrochemical polarization cell has been designed in which the test
electrode is a rotating disk with a pin for the load contact to simulate a rubbing surface
and a chamber for the electrolyte to create a diverse corrosive environment. Polarization
measurements are performed on several bearing alloys with various corrosion properties
(both the active and the passive type). The effects of the load, the speed, and the nature
of the environment are studied to establish a correlation between corrosion, wear, and the
nature of protective films formed on rubbing surfaces.
KEY WORDS; rubbing surfaces, wear, corrosion, electrochemical polarization, open-

circuit potential, steels, fretting, inhibitors
The damage that occurs at the interface of two surfaces in contact, of

which at least one is metal and rubbing against the other surface under some
load, is termed wear. Under extreme conditions of mechanical stress or envi-
ronmental stress, or both, wear may lead to failure by fatigue or corrosion
fatigue. Generally, oscillatory motions accompanied by friction at relatively
few asperities of the surface describe wear under fretting. Even in perfectly
lubricated ball bearings a similar situation results when particulate matter of
the environment contaminates the system. Pits usually nucleate from such
mechanical action, leading to faster accumulation of metal oxide debris and
thus causing an increase in friction or seizing in machines. Failures of
hinges, fasteners, and bearings in an aircraft structure are examples of this
phenomenon.
'Research scientist. Naval Air Development Center, Warminster, Pa. 18974.
^Senior engineer and research metallurgist, respectively, I.I.T. Research Institute, Chicago,
111.
106
Copyright'^^ 1982 b y A S T M International www.astm.org
AGARWALA ET AL ON CORROSION IN RUBBING SURFACES 107
The other type of failure in metals is corrosion where the electrochemical

action of the environment, such as water (H2O), salts, and gaseous matter,
also leads to the formation of oxide or corrosion products. Particularly when
the metal surface is nascent, as under rubbing conditions, the corrosion ef-
fects are much more accelerated. In frictional wear, the environment is an in-
tegral part of the destruction phenomenon as it interacts with the metal,
causing mechanical stresses by increasing friction.
A mechanistic study of the wear phenomenon was probably best approached
by Uhlig [1],^ who introduced an interplay of corrosion into his model on
fretting corrosion. Earlier, the influence of chemical factors on wear was also
noted by Fink [2] and Fink and Hoffman [3], who reported a frequency
dependence of wear losses in their tests. Since then many studies [4-7] have
shown that metallic surfaces under stress and motion become deformed and
active, whereby their chemical activity is accelerated considerably. However,
if a protective film is formed on such a surface by means of certain chemicals
[8-10] present (added) in the system, such as lubricants, the wear losses tan
also be decreased. It is believed that during rubbing, the protective oxide
film, which is always present on metallic surfaces, is ruptured and removed,
thus leading to exposure of the active (bare) metal surface to the environ-
ment. If this oxide film is not repaired immediately, corrosion occurs, caus-
ing the formation of corrosion product, which becomes the precursor of ac-
celerated wear. Thus, it becomes obvious that wear and corrosion are related
phenomena and probably work conjointly, with corrosion playing an impor-
tant role. Consequently, in the present work an electrochemical approach
was employed to investigate the role of corrosion in the wear process and
possibly to develop an understanding of the interactive mechanism involved
between corrosion and wear.
Electrochemical methods [11-13] have been used extensively to explore the
mechanism of corrosion and passivation. Among these methods, the poten-
tiostatic-potentiodynamic polarization techniques have been most helpful in
determining the protective nature of the surface film formed on ac-
tive-passive metal surfaces. Under certain chemical actions or in the
presence of a damaging species such as chloride, the protective films are
susceptible to breakdown, which causes accelerated environmental attack
(corrosion). In such a case, the electrochemical polarization measurements
become much more significant as they provide more specific information
regarding the stability and repairability of such films when ruptured [14]. A
similar situation probably also exists when the rubbing of two metallic sur-
faces causes the rupture of the oxide film, which protects active metals such
as iron and titanium, to generate new metal surfaces of considerably high
chemical reactivity. Electrochemical polarization methods applied to a newly
generated surface or a crack tip have been studied in the past in phenomena
The italic numbers in brackets refer to the list of references appended to this paper.
such as stress corrosion cracking and corrosion fatigue [15,16]. Since corro-
sion and wear are both surface phenomena involving the environment, it can
be assumed that the mechanisms mvolved must have some basic governing
rules in common. If the mechanisms are of an electrochemical nature, it is only
proper to investigate them by electrochemical procedures. So far, the wear
phenomenon has seldom been studied with an electrochemical approach
[17], and probably no studies have been done in situ on rubbing surfaces.
Experimental Procedure
The most important aspect of electrochemical measurements in the wear
process is the design of a system in which a rubbing motion under stress can
be applied and a simultaneous measurement of electrochemical parameters
can be made. Thus, a dynamic wear-corrosion apparatus was designed, as
shown in Fig. 1. The major parts of the apparatus (Fig. la) are the electroly-
tic cell (E) which contains a rotating disk specimen (A) in radial contact with
a rubbing pin (B) under load, using a precalibrated constant tension spring.
The specimen was rotated by a synchronous motor, and its electrical contact
to a measuring system was made through a wire in the hollow shaft leading to
the current collecting mercury pool. Two platinum counterelectrodes used
for polarization of the working electrode (A) were put on either side of the
specimen (shown in Fig. ib) facing each other. The standard reference elec-
trode used was a saturated calomel electrode (SCE), which was connected
electrolytically through a Luggin capillary (C) and placed very near to the
metal-metal contact surface in the cell containing an electrolyte. The details
of the apparatus design and assembly are better described elsewhere [18].
The materials studied in this work varied from very basic to those used in
bearing systems. They are Armco iron, AISI M2, and Type 52100 steels. A
Type 304 stainless steel was also used for a comparative evaluation as it has
been well characterized electrochemically and is mostly passive under normal
conditions. The disk electrode (A in Fig. la) was approximately 10 mm thick
and 38 mm in diameter. The pin (B in Fig. la) was about 38 mm long and 6
mm in diameter with a hemispherical end. The disk and pin electrodes were
made of the same materials. Except for the total cylindrical surface area of
the disk electrode and the contact area of the pin electrode, the electrodes'
surfaces were coated with epoxy. Care was taken to ensure that the cylindri-
cal surface of the disk electrode was polished (600 grit), degreased, cleaned
of any debris before installation, and mated with a pin electrode prepared in
a similar manner. The electrodes were stored in a desiccator before use and
handled with clean gloves.
The electrolytes were made from distilled water containing various concen-
trations of reagent grade sodium chloride (NaCl) or sodium hydroxide
(NaOH), as needed for the experiments. The chemical additives used as inhib-
itors of corrosion were reagent grade sodium dichromate (Na2Cr207'2H20),
Load Insulating coating

Clear plastic
1..—
Mercury Pool
Electrolyte Cell
A. Lower disc electrode D. C u r r e n t c o l l e c t i n g p i n in a p l a s t i c h o l d e r

B. Upper pin electrode £. E l e c t r o l y t e w i t h stirrer and t e m p e r a t u r e
C. Salt b r i d g e to S C E compensator
Copper-plated ring for current transmission
A. Disc Working Electrode

B. Platinum Counter Electrodes
C. Luggin Capillary Salt Bridge
D. Combination pH Electrode
G. Automatic Temperature Compensator
P, Thermometer
G. Load Spring, Scale, and Mounting Frame
H, Plastic Container, 3.5 liter Capacity
with Plastic Cover
I. Saturated Calomel Reference Electrode
J. Current Collecting Brush
(bl
FIG. 1—An electrochemical test apparatus for measuring corrosion-wear during the rubbing
of two surfaces: (a) a schematic drawing of a corrosion-wear cell and (b) a photograph of the
whole apparatus.
sodium molybdate (Na2Mo04-21120), and sodium borate (Na2B407-101120).

The electrolyte was continuously stirred during the test, and all the tests
were performed at ambient temperature and pressure. The instruments used
for electrochemical measurements were a Wenking potentiostat with a scan-
ner, a Kiethley electrometer, H-P logarithmic converter, and a Houston A"-y
recorder. The procedures used for making potentiostatic-potentiodynamic
anodic and cathodic polarization measurements were similar to those of
ASTM Recommended Practice for Standard Reference Method for Mak-
ing Potentiostatic and Potentiodynamic Anodic Polarization Measurements
(G 5-78) and of other mvestigators [12-13]. The open-circuit potential-time
measurements for each system were made using an SCE as a reference elec-
trode through a Kiethley electrometer (a high-impedance meter) to a strip
chart recorder.
Initially, the disk electrode was mounted on the shaft immersed in the elec-
trolyte. While the electrode was rotating without any contact with the pin,
the open-circuit potential was determined. Next, the pin electrode, mounted
on a hinged arm, was spring loaded on the disk electrode which was stopped
temporarily. Then, the disk electrode was rotated at the desired speed. All
potential measurements were made under dynamic conditions—in other
words, while the disk electrode was in motion.
All polarization measurements were started from the steady-state open-
circuit potentials of the specimens and in steps of 2 mV/s in either direction
(anodic or cathodic). Under rubbing conditions, the wear-corroding system
generally is not stable, therefore, small variations of up to 10 mV/h were con-
sidered steady state for open-circuit potentials. Total immersion tests were
made on materials to determine their weight loss behavior under both wear
and no-wear conditions. From current-potential (polarization) diagrams,
corrosion losses were also measured in terms of current density, and the data
were collected from both the wear and the no-wear tests. The purpose of
these tests was to determine the synergistic effect of corrosion on wear and
vice versa.
Metallurgical Examination
Analytical scanning electron microscopy (SEM), which combines the ex-
amination of surface morphology with element analysis by energy-dispersive
X-ray (EDX) or wavelength-dispersive X-ray (WDX) techniques, or both,
was used for the examination of wear surfaces of disk and pin electrodes. The
metallurgical evaluation included low-magnification (X20) SEM examina-
tion of wear surfaces to select typical as well as nontypical damaged areas for
detailed wear surface morphological examination. EDX spectra were obtained
to identify the various elements present in the corrosion-wear debris, and
then elemental maps showing the relative distribution of various elements in
the area of interest were generated. These X-ray distribution maps, along
with the SEM photomicrographs of the same, provided very useful informa-
tion regarding surface conditions and the corrosion-wear product present.
This information also helped in predicting whether the film would be protec-
tive or not.
Results and Discussion
Gravimetric Study
The weight loss data from corrosion-wear tests of a bearing steel (M2)
polarized in various environments were analyzed and are reported in Table 1.
The wear condition for this study was a 25.4-kg (56-lb) load and a speed of
7r/30 rad/s (1 rpm). Disk electrodes were primarily used for weight loss anal-
ysis, and they were weighed before and after each test in an analytical
balance. After each corrosion-wear test, the disk electrode was taken out of
the test solution, washed in running water, and then cleaned along with a
control sample using National Association of Corrosion Engineers (NACE)
solution [50 g/litre of stannous chloride (SnCl2) and 20 g/litre of antimony
trichloride (SbCla) in concentrated hydrochloride acid (HCl)]. The specimens
were finally washed with distilled water, dried in acetone, and weighed.
In the corrosion-wear system, electrode weight loss may arise from three
different mechanisms: (1) general corrosion of the unprotected noncontact-
ing surface of the electrode in the electrolyte, (2) enhanced corrosion due to
the breakdown of the electrode surface film under wear, and (3) mechanical
wear due to rubbing contact. To determine the weight loss from each mecha-
nism, the corrosion current density data obtained from polarization under
no-wear and wear conditions (shown in Table 2) were utilized to calculate the
weight loss by general corrosion (Mechanism 1) and enhanced corrosion
(Mechanism 2), respectively. The mechanical wear loss (Mechanism 3) was
determined by subtracting the weight loss by both general and enhanced cor-
rosion (Mechanism 1 + 2) from the total weight loss of the disk electrode in
Table 1. The total weight loss by corrosion only indicates the sum of general
and enhanced corrosion weight losses.
The results summarized in Table 1 show a general synergism between wear
and corrosion. It can be observed that the total weight losses by corrosion only
are always smaller than those from wear, shown in the final columns of Table
1. Usually, under active environmental conditions where weight losses due to
corrosion were high, the wear losses were high, too. Under test conditions
where corrosion losses were almost insignificant (Rows 1 and 2), the losses
due to wear remained 1 to 2 orders of magnitude higher. The rubbing action
under load destroys the passive (protective) film on M2 steel, which does not
repair itself readily; therefore, corrosive wear increases. While a small di-
chromate addition (5.7 ppm. Row 3) may improve the passivity of the surface
•a- rt 00 00
t/i -^ TT
t~-- I/) O^
1 d o o
OS a; ^ r)
-H 04 o
6
•a
c
•a g 8
W
s
6^
0> O (N
ov o d «• d
r~ o 5 S 8
I d Vd
W 2»
vO O TT
od d d R -^
o
I g
g' <N «
R «
u+
u d as a ^
•S E
S V)
I
X "^ *^. "1
o. Lo o i/i uS
^*
I
.2 ^
la
1^
1«
uig
|3|s|i
•o II -S
<
•it
2 S2 5 5 I«
^^ DC ^^ ^^ ^^ 2l
a
•s
u Oo 0 0
Qj 03 n cd cs
Copyright by ASTM Int'l (all rights reserved); Fri Jan 27z22:01:44
z z EST z 2012
z
TABLE 2—Electrochemicatly determined rate of loss data for M2 steel under wear
and no-wear conditions."
Chemical Rate of Loss,* mA/cm^ (mpy)

Electrolyte Additive Wear -
(ppm) (ppm) pH Time, h No Wear Wear
NaCl (1000) 5.6 40 7.0 X 10"^ (32) 6.4 X 10"^ (29)

NaCl (1000) NaOH 10 4 9.8 X 10"2 (45) 1.0 X 1 0 ' ' (46)
NaOH (4) 10 4.5 1.7 X 10"5 (<0.01) 1.3 X 10"^ (0.6)
NaCl (10) NajCfjOy 5.4 5.8 X 10"" (0.26)
(2.8)
NaCl (10) NajCrjOy 5.4 10 5.0 X lO^"* (0.23) 9.0 X 10"^ (4.1)
(5.7)
NaCl (1000) NajCtjOy 5.5 8 4.2 X 10"2 (19) 8.2 X 10"2 (36)
(5.7)
NaCl (1000) NajCrjOT 4.9 8 2.0 X 10"^ (0.9) 2.6 X 10"^ (1.2)
(100)
"All wear tests were performed at a load of 25.4 kg and a speed of i/30 rad/s (1 rpm).
*Rate of loss determined in terms of current density, mA/cm^, or calculated in mils per year
(mpy).
and reduce the corrosion and wear losses compared with those under condi-
tions in Row 1, this small dichromate addition is not effective under a higher
chloride condition (Row 4). In a high chloride solution, a larger dichromate
level (100 ppm) is needed to reduce the corrosion loss by more than 1 order of
magnitude and the wear loss by a factor of 4. The comparative weight loss
data in Rows 4 and 5 support this conclusion. It must be noted that,
although dichromates are excellent corrosion inhibitors (passivators) in both
neutral and acid solutions, they become less effective when chloride ions are
present, since it is well known that chlorides break down the passivity in
metals, thereby increasing their susceptibility to pitting. Similar results were
found when an electrochemical method was used to determine losses under
both wear and no-wear conditions (Table 2). In a highly active (corrosive)
medium, the difference between the wear and corrosion components of the
losses is very little, but it becomes greater as the medium becomes less ag-
gressive (or more passive). This effect was also evident in active metals such
as Armco iron (Table 3) but to a lesser extent. Armco iron does not ordinarily
passivate in chloride medium unless a large amount of passivating agent (di-
chromate) is present. In 100 ppm NaCl solution, at least 500 ppm of dichro-
mate is essential to incorporate passivity.
Electrochemical Study
The open-circuit (corrosion) potentials of a metal-environment system are
a measure of the degree by which oxidation-reduction occurs. These poten-
tials are expressed in terms of the Nemst equation as
where E is the electrode potential (corrosion); IP is the standard redox poten-

tial; n is the number of electrons exchanged in the reaction; [OX] and [RED]
are the concentration terms of oxidized and reduced species, respectively;
and R, T, and F are the gas constant, temperature, and Faraday constant,
respectively. A typical reaction for oxidation of iron could be written as
(Anodic) Fe -^ FE++ + 2 e (2)
Here, if the E is negative (measured as an open-circuit potential with respect

to a standard reference electrode), Reaction 2, as written, will proceed, and
iron will corrode. In contrast, with noble metals such as platinum and gold,
this electrode potential is positive, which indicates that material will remain
in the metallic state.
In any case, when a reaction proceeds as in Reaction 2, a corresponding
opposite reaction will also occur. If the medium contains oxygen and is
almost neutral, the reaction product is hydroxide
(Cathodic) O2 + 2H2O + 4e *- 40H- (3)
Thus a corrosion product, hydrated metal oxide, is formed which is generally

a mixed composition of several oxides. The nature of this oxide depends on
the constituents present in the medium, the alloy, and the reaction kinetics
with which they are formed. If this hydrated oxide is compact and rigidly
bound with the surface, a layer of it performs as a resistance barrier, signifi-
cantly reducing further oxidation of the metal. Sometimes other elements
TABLE 3—Electrochemically determined rate of loss data for Armco iron under wear and
no-wear conditions."
NaCl Rate of Loss,* mA/cm^ (mpy)

Solution, Chemical Wear
ppm Additive (ppm) Time, h Under No Wear Under Wear
1000 NajCfjOy (500) 8 7.8 X 10"^ (0.36) 1.4 X 10-3 (0.64)

NajMoO^ (500)
100 NajCrjOy (500) 27 1 X 10-3 (0 45) 1.4 X 10-3 (0.64)
Na2Mo04 (500)
10 NajCrjO, (500) 24.5 2.5 X lO-"* (0.11) 8 X 10-" (0.36)
Na2Mo04 (500)
10 NajCrjO, (500) 31.5 3 X 10-" (0.13) 3 X 10-3 (1.37)
"AH wear tests were performed at a load of 7.26 kg and a speed of 1/30 rad/s (1 rpm).
*Rate of loss determined in terms of current density, mA/cm^, or calculated in mils per year
(mpy).
present in the system (alloy or medium) also participate and help reinforce or
form a highly adherent corrosion product film (as in the presence of
chromium or dichromates). The reduction of corrosion by such a method is
called passivation, and the protective layer, which is usually very thin in
nature, is called passive film. This phenomenon occurs mostly at potentials
that are more anodic (noble) than the open-circuit (corrosion) potentials,
that is, when a metal is polarized in the anodic direction. Correspondingly,
when a metal is polarized to potentials that are more negative than the open-
circuit value, Reaction 3 is subsequently followed by the hydrogen evolution
reaction (HER)
H,0+ + e H2O + V2H2 (4)
Generally the production of gaseous hydrogen (H2) is preceded by the forma-

tion of atomic hydrogen which, being very nascent, is absorbed by the metal.
Particularly in high-strength alloys, this leads to embrittlement.
Effect on Open-Circuit Potentials

The steady-state potentials of the materials in the active state (not forming
a passive film) showed only a negligible dependence on the degree of wear.
Figure 2 shows transient behavior of the potential for M2 steel in an active
environment (0.1 percent NaCl). At steady state, the application of a load
(up to 25.4 g) only influenced the potential temporarily but reached almost
-100 -^h
— • — M 2 Steel in 0.1?. HaCl solution
with 100 ppm of Na2Cr207 v/ith and
r 3.7 lb
without load at 1 rpm.

-200 O M2 steel in HaOH solution of 16 lb
pH 10 with load at 1 rpm.
41 lb
r
-300
\
No load With
load
r 56 l b -
\. ^ \
-400
\.
'^.,
-500-^ -/^
40
Time, hr
FIG. 2—Transient behavior of the open-circuit potential for M2 steel under wear and no-wear
conditions.
the same value of —0.45 V (versus SCE) afterwards. In the passive environ-
ment (pH = 10), the open-circuit potential of M2 steel at steady state (Fig. 1)
was more noble, but the application of load shifted the value to more
negative potentials. The subsequent increment of load decreased the open-
circuit potential only slightly. This is interpreted as indicating that the initial
large change in the potential was due to the disruption of a protective film
whereas the small changes thereafter were due to additional structural defor-
mation of the surface. It must be realized that the maximum hertzian contact
stress between the pin and disk at a 7.26-g load was calculated to be approxi-
mately 2690 MPa (390 ksi). Types 52100 and 304 stainless steel behaved in a
similar way, although Type 304 stainless steel is also passive in several other
environments. The effects of speed on the potential shifts were much less sig-
nificant than those of load variations. Speed probably has more influence on
the current density (the reaction kinetics), which will be discussed later. An
earlier study reported elsewhere [18] gives a more detailed discussion of the
open-circuit potential aspects.
Effect on Polarization Behavior

A theoretical polarization diagram for an active metal-electrolyte system
with both the anodic and cathodic components of the reactions occurring is
illustrated in Fig. 3a. The diagram also demonstrates the method by which
corrosion current, i^.^^, and corrosion potential, £'corr. can be calculated. The
tangents drawn from the linear parts of the anodic oxidation and cathodic
reduction curves represent Tafel slopes, and their point of intersection repre-
sents the ^corr and icorr- In an active-passive system where either the metal or
medium forms a passive film, the anodic polarization curve takes the form
shown in Fig. 3b with a distinct separation of active and passive regions in-
dicative of high and low current densities. Using Faraday's law, these current
densities can be converted into corrosion rates. In alloys such as Type 304
stainless steel, the passive film forms very naturally as oxides of chromium,
nickel, and iron. This film is very stable until the environment becomes very
aggressive, as in the presence of chloride when film breakdown occurs and
pitting initiates. The dotted loop shown in Fig. 3b, produced from a cyclic
(forward and reverse) potentiodynamic scan during anodic polarization, is
very characteristic of pitting and crevice corrosion. The susceptibility of the
system to pitting or crevice corrosion increases as the area in the loop in-
creases, or Ep > Ef, where Ep and E^ are pitting (or breakdown) and repas-
sivation (or protection) potentials, respectively. When Ep < E^, the system
does not undergo pitting or crevice corrosion; in other words the repassiva-
tion kinetics is so fast that the protective film repair is achieved instantly. In
rubbing surfaces it is this property which is most essential for reduction of
wear and corrosion. Nucleation of a pit in bearing elements is one of the
foremost reasons for failure.
.
- .<r^ (a)
•v «." 'Hj' . ' >
. v | , ' ^l /ANODIC 0XID4TI0N CUBVl
* •«EV " ' V , - ^ . ^ y^ C M _ M " . J,"I
= \
X
CORROSION
CURRENT
s \ KNSIiy
s Vs '" " ' ^ ^>"
1
£
— ^ - N*/"
OPfN CIRCUIT
1 >^ ** »c-r'
1 OCHANCt
cwRtm /
^'^~~~^v
^v
1
1
MUSHY. /_! .. J >
E «IVf»Sl»lI • / • * 1 ^ ' 1
3 (OUIUIRIUM V ^ ' i' N.
u ROnmiAl • N o»idJtion frtoclion \^J^T„Qp,(• REoucTION CURVE
V O H ' • 2t' Hjl
S
5 RICION \ \ v^
\4.
<
1
\\ 1 LIMITINC DIFFUSION
^^ 1 CURRtMT HNSinr I I
1 1 1 1 1 1 I "
10 100 lOGO
CURREW OENSIIY t fitkmh
'°^ ^^^--'''''''^No p i t t i n g 1
f rutin, ^ ^
L
t
p
-
X* Paaitv*
y
I
r ^
o.c.
^"•"••-N,^^ Act lv»
L09 Current Dcnilty
FIG. 3—Illustrations of potentiostatic and potentiodynamic polarization behavior of an (a)

active and (b) active-passive metal-environment systems.
The polarization behavior of active systems is not significantly affected in

either wear or no-wear conditions. Figure 4a, obtained for Armco iron in
1000 ppm NaCl solution, shows a typical plot of such behavior. However,
when the test medium was changed to a 100-ppm NaCl with 500 ppm each of
Na2Cr207 and Na2Mo04 as inhibitors of corrosion, the effects of wear began
to show up, as shown in Fig. 4b. It may be observed that the open-circuit
potential, Eoc, as well as the corrosion current have shifted to more active
1 r
-800
o No load, 1 rpin
-111 00 A 16 lb, 1 rpm
Log Current Density, mA/cm
n r 1 r
'O.C.
\
-600
-800
— Armco Iron
o No l o a d , 1 rpm
A l6 l b , I rpm
:^-.
J L
Log Current Density, mA/cm
FIG. 4—The polarization behavior of Armco iron in (a) 1000 ppm NaCl solution and (b) 100
ppm NaCl solution containing 500 ppm each of sodium dichromate and sodium molybdate
under load and no-load conditions.
potentials and higher magnitude, respectively, under wear conditions. The

curves under load conditions were similar to those in Fig. 3a in terms of coor-
dinates, which indicates a situation of active corrosion. It must be noted that,
m 100 ppm of chloride with 500 ppm each of dichromate and molybdate
Armco, iron is less corrosive, and when the chloride concentration is further
reduced to 10 ppm, the metal becomes passive and shifts the ôc to a nobler
potential (approximately 0.05 V versus SCE). However, under a 7.26-kg load
and a speed of 7r/30 rad/s (1 rpm), the polarization curves again shift down
to the position of active corrosion (Fig. 4b). Data obtained from these
polarization measurements for Armco iron are summarized in Table 3,
where corrosion current densities have been used to calculate the rate of loss
in terms of mils per year. A comparative evaluation of results in Table 3
shows that as the test sohition was made less corrosive, the losses due to cor-
rosion and wear were also reduced. At the 10-ppm NaCl level, the effect of
molybdate in solution was not quite clear in terms of corrosion rate but was
very significant from 35 to 9 /xm per year (from 1.37 to 0.36 mils per year) in
decreasing wear rate.
In active-passive metals such as Type 304, 52100, and M2 steels, the ef-
fects of load and speed are more obvious. Under passive conditions, the EQC
shifts from the passive to the active potential region as the surface is rubbed
and the passive film is destroyed. A typical example of this type of behavior is
shown in Fig. 5 for Type 52100 steel in distilled water. It must be observed
that wear conditions also shifted the polarization curves to higher current
densities, indicating accelerated corrosion and wear. A similar study of other
passive alloys (M2 and Type 304 stainless steel) was reported earlier by the
authors [18]. However, it was generally found that if the system (metal-envi-
ronment) did not have the ability to re-form a passive film instantly after rup-
ture, the wear process, like corrosion, was also accelerated and the alloy
behaved like an active metal. Tomashov and Kershinina [19] have reported a
3 to 4 order of magnitude change in ôrr of titanium, chromium, and nickel
from their abrasion tests. In highly passivating solutions containing di-
chromate, the effect of wear (rubbing) on M2 steel was found only on the
anodic component of the polarization curve, the cathodic curve showing no
significant change. This is simply because rubbing action removes the pro-
tective film from the surface, causing the bare metal to corrode. The cathodic
processes were the least affected, probably because of the limiting factors of
oxygen transportation. The results of the polarization measurements on M2
steel are summarized in Table 2. Here the data for wear and no-wear tests
represent the ico„ values determined under load and no-load conditions, re-
spectively. According to the data in Table 2, as the medium becomes less cor-
rosive, the losses from both the wear and corrosion components also
decrease, which is in complete agreement with the weight loss data deter-
mined gravimetrically (Table 1).
52100 Steel In Distilled Water

Q No load and no motion
0 36 lb at 10 rpa
+200
-1000
\
J_ _L
\
-4 -3 -2 .1
Log Current D e n s i t y , mA/cm^
FIG. 5 - -The polarization behavior of Type 52100 steel in distilled water under various wear
conditions.
The most interesting aspect of the rubbing action on polarization behavior

of M2 steel was observed in the passive region of the anodic polarization
curve. As illustrated in Fig. 6, when an M2 steel specimen was polarized
anodically in NaOH solution of pH 10, a passive region of low current densi-
ties (in the microampere-per-square-centimeter range) was found under no-
rubbing (load) condition. When load was applied, the shift of the passive
region occurred toward the higher current densities as expected in wear but
their magnitude was still low, which suggested a passive region of up to +200
mV (versus SCE). It must be noted that M2 steel is spontaneously passive in
a pH 10 solution; that is, the corrosion potential, -Ecorr- shifts instantly to the
passive potential region (compare with Fig. 3) of very low current densities.
The existence of a passive region under wear shows that, although the rub-
400 T—r T r 1 r
200
-200
—0
-400
M2 Steel in NaOH Solution

(pH 10). Ho NaCl
- -600 tlo load at 1 rpm
36 lb at 1 rpn
Current reading at a given overpotentlal
-800 when wear was stopped during anodic
polarization.
\
0
-1000 O- I 11
-5 -4 -3 .2
-1
Log Current Density, mA/cm'
FIG. 6—A potentiodynamk polarization scan of M2 steel in a passive state showing an in-
crease in the current density but still maintaining passivity.
bing action has destroyed a major part of the passive film, a part of it re-
mains to protect the surface. The possibilities are that either this film was not
completely ruptured, as happens in less protective systems, or it was re-
formed instantly because of the fast-reaction kinetics of the metal-solution
interface. A further study to that effect was made in which the specimen dur-
ing anodic polarization under load was intermittently held at a few fixed
passive potentials (potentiostatically) to measure the cell current density with
time. At this point the rotation of the specimen was stopped, but the load was
not removed. The solid circles in Fig. 6, connected through the dotted lines
from the curve under dynamic load, indicate the reduction of the cell current
densities so obtained. The magnitude of this drop was approximately to a
level which would be obtained without wear (compare the open and solid
circles in Fig. 6). These observations indicate that during rubbing in this test
the metal surface was not free of film, and given time the film grew back to
its normal no-wear condition level. If the metal surface had been completely
bare, additional oxidation would have been required to produce an oxide
film. In other words, the current density should increase initially at those
points in the anodic curve (with A) when rotation is stopped and then de-
crease to the levels of the solid dots. Since this did not happen, it can be in-
terpreted that, in spite of the load and rubbing, some passivity was retained
on the surface. This kind of information could only be derived by electro-

chemical means.
Surface Characterization Study

The surface features of the rubbing interfaces were examined by SEM and
found to be typical of wear surfaces of soft material (in the case of Armco
iron). In general, rough surfaces with heavy wear in some areas were
recognized on the disks as well as on the pins. The pin surfaces were compar-
atively rougher than the disk surfaces because, during rotation, the disk
specimen may not be in continuous rubbing contact with the pin. The repre-
sentative SEM photomicrographs of the disk surface in Fig. 7 (Plates a and
b) show low and high magnification views of the debris in the wear track. The
pin surface (not shown) was rougher. EDX spectra of the disk wear surface
(Fig. 7c) were very similar to those of the pin surface and showed the presence
of molybdenum, chromium, and iron only. The test medium in this case con-
tained 100 ppm sodium chloride, 500 ppm sodium dichromate, and 500 ppm
sodium molybdate in solution. Inside the damaged areas entrapped fine
granular wear-corrosion debris were more visible in Fig. Id. The X-ray dis-
tribution maps of this enlarged view, (Fig. Id) for molybdenum and
chromium show the presence of both of these elements in wear debris, with
the molybdenum concentration being lower than that of chromium (Figs. 7e
and/, respectively). The X-ray maps of the pin surface also confirmed the
same phenomenon with the debris in much higher densities than in the disks.
It seems that the higher corrosive activity of the wear area enhances reactions
so that molybdenum and chromium deposit selectively and in higher density.
It should also be noted that the immersed iron surface in this test medium
without wear showed no detectable presence of molybdenum or chromium in
EDX spectra.
Conclusions
It has been shown that corrosion and wear occur simultaneously and con-
jointly. If one is controlled, the other is correspondingly controlled as well.
Electrochemical measurements have shown that the rubbing action of two
metals simulates the active polarization behavior of a corroding metal. It was
found that attainment of passivity to control corrosion was also synergisti-
cally beneficial in controlling wear. In passive conditions, the losses due to
mechanical action were reduced by a greater magnitude than those due to
corrosion. In an actively corroding medium, the distinction between rubbing
wear and corrosion was not significant in electrochemical measurements;
neither the polarization behavior nor the open-circuit potentials seemed to
change. In active-passive systems, the effect of wear on the cell current den-
sity and the open-circuit potential was easily measurable by electrochemical
FIG. 7—A scanning electron microscopic (SEM) examination of an Armco iron disk speci-
men after the test in lOOppm NoCl solution containing 500 ppm each of sodium dichromate and
sodium molyhdate, showing (a) the wear track, X15: (b) the surface damage at high magnifica-
tion X50: (c) EDX spectra from (a); (d) the appearance of corrosion-wear debris on the wear
track, XI75: and elemental X-ray distribution maps of (e> molybdenum X175, and (f)
chromium, y~175.
techniques. In the final analysis it can be affirmed that in rubbing processes

the ability to re-form a passive film on the contact surfaces is an important
factor in reducing wear and corrosion. This can be achieved externally by the
addition of passivating inhibitors.
Acknowledgments
This work was supported by the Naval Air Systems Command through the
Naval Au- Development Center under Contract No. N62269-79-C-0702. The
authors wish to thank Dr. J. J. De Luccia and S. J. Ketcham for their helpful
suggestions and support.
References
[/] Uhlig, H. U., Journal of Applied Mechanics, Vol. 21, 1954, p. 401.
[2] Fink, M., Transactions of the American Society of Steel Treating, VoL 18, 1930, p. 1026.
[3] Fink, M. and Hoffman, U., Archiv fiir Eisenhuttenwesen, Vol. 6, 1932, p. 161.
[4\ Hirst, W. and Lancaster, J. K., Journal of Applied Physics, Vol. 27, 1956, pp. 1057-1065.
[5] Quinn, T. F. J. and Sullivan, J. L., Wear of Materials, American Society of Mechanical
Engineers, New York, 1977.
[6\ Kerridge, M., Proceedings of Physical Society (London), Vol. B68, 1955, p. 400.
[7\ Sakurai, T., Okabe, H., and Tomaru, M., Bulletin of Japan Petroleum Institute. Vol. 14,
1972, p. 161.
[8] Tripathi, K. C , Proceedings of the Sixth International Conference of Surface Active Sub-
stance, Zurich, 11-15 Sept. 1972, p. 491.
[9] Tripathi, K. C. and Groszek, A. J., Aluminum, Vol. 49, 1973, p. 612.
[10] Felsen, I. M., McQuaid, T. W., and Marzani, J. A., A.S.L.E. Transactions, Vol. 15,
1972, p. 8.
[//] Baboian, R., Ed., Electrochemical Techniques for Corrosion, National Association of Cor-
rosion Engineers, Houston, Tex., 1977.
[12] Gileadi, E., Kirowa-Eisner, E., and Penciner, J., IntetfacialElectrochemistry, An Experi-
mental Approach, Addison-Wesley, Reading, Mass., 1975.
[13] Mansfeld, F. and Bertocci, U., Eds., Electrochemical Corrosion Testing, ASTM STP 727,
American Society for Testing and Materials, Philadelphia, 1979.
[14] Starr, K. K., Verink, E. D., and Pourbak, M., Corrosion, National Association of Corro-
sion Engineers, Vol. 32, No. 2, 1976, pp. 47-51.
[15] Beck, T. R., Electrochemical Techniques for Corrosion, R. Baboian, Ed,, National Asso-
ciation of Corrosion Engineers, Houston, Tex., 1977, p. 27.
[16] Diegle, R. B. and Vermilyea, D. K., Journal of Electrochemical Society, Vol. 122, 1975, p.
180.
[17] Ashworth, V., Grant, W. A., and Proctor R. P. M., Corrosion Science, Vol. 17, 1977, p.
947.
[18] Kim, K. Y., Bhattacharyya, S., and Agarwala, V., Proceedings of the International Con-
ference on Wear of Materials, American Society of Mechanical Engineers, Houston, Tex.,
March-April 1981, p. 772.
[19] Tomashov, N. D. and Kershinina, L. P., Electrochimicha Acta, Vol. 15, 1970, p. 501.
M. G. El-Sherbiny^ andF. B. Salem^
Fretting-Resistant Ion-Plated
Coatings
REFERENCE: El-Sherbiny, M. G. and Salem, F. B., "Fietting-Resistant Ion-Plated

Coatings," Materials Evaluation Under Fretting Conditions. ASTM STP 780, American
Society for Testing and Materials, 1982, pp. 125-137.
ABSTRACT: This paper presents experimental results on fretting testing of various sur-
face coatings, with particular reference to ion-plated films. A special test setup is described,
and some peculiar testing conditions are discussed. The results indicate that ion-plated
films are superior to vacuum-deposited, gas-deposited, and sputtered-surface coatings.
The effectiveness in alleviating fretting of four ion-plated soft metallic films—namely,
lead, indium, tin, and silver—is evaluated. Lead showed the best fretting resistance at
small slip amplitudes, while tin showed a good resistance at high slip amplitudes. Ion-
plated titanium carbide films were also examined. The wear-resistant hard carbides
showed outstanding fretting resistance.
KEY WORDS: ion-plated films, fretting, surface coatings, friction, fatigue wear
In most mechanical systems and engineering appliances, there exist me-

chanical vibrations which are generated by the various dynamic actions of the
system. Mechanical vibrations result in a minute reciprocating sliding mo-
tion between the adjacent machine elements. These elements may nominally
be at rest with respect to each other, such as in bolted and riveted joints, but
they undergo a small-amplitude vibrational slip. The result of this slip is fret-
ting, which is defined as the wear mechanism occurring between two mating
surfaces having an oscillatory relative motion of small amplitude (less than
250 ftm) [1].^ Fretting damage may lead to weakening of the structure and
result in possible premature failures. Several techniques to minimize fretting
damage have been developed and are reported in the literature. These in-
clude interposing a noncorrosive lubricant at the contact interface, inserting
soft bushings, applying coatings and surface treatments, and, finally,
minimizing the slip amplitude [1-4]. The effect of the slip amplitude on wear
'Associate professor and assistant professor, respectively. Department of Mechanical Design

and Production, Faculty of Engineering, Cairo University, Cairo, Egypt.
125
volume was examined [5], and the results indicate that relative slip
amplitudes below 70 /xm make a very small contribution to mechanical wear.
Interposing a noncorrosive lubricant at the friction interface [4] reduced the
wear rate to one tenth of that of dry fretting in a normal atmosphere. From
other studies of coating techniques [2,4], results indicate that hard-wear
resistant coatings are superior to soft lubricant coatings, but there is little
published in this area.
The objectives of the present investigation are to introduce a new design of
fretting test setup, and hence to use the proposed testing facilities in eval-
uating the effectiveness and usefulness of several surface coatings deposited
by different coating techniques. Obviously, the selected coatings should not
impair the mechanical properties of the basic metal and should remain
strongly adherent to the substrate under such cyclic operating conditions.
Ion plating was shown [6-9] to be successful in practically all applications
where surface films are needed. These include corrosion protection [10],
wear reduction [7,11], and thin metallic film lubrication [8,12], The success
of the ion-plated coatings is attributed to their outstanding adhesion, unique
film thickness uniformity, fine dense grain structure, and low level of
residual stresses. Apart from the film characteristics, the technique itself is
very versatile in depositing any metal, alloys, and compounds on any metallic
or nonmetallic substrates. Special attention is, therefore, given to evaluation
of the potential of the ion-plated coatings in fretting prevention.
Experimental Procedure
Fretting Test Setup

The basic mechanism of the fretting test setup consists of four leaf springs
which are assembled to form a diamond structure, as shown m Fig. 1. The
spring, diamond-shaped, structure supports two cam followers (6) at two op-
posite comers, while two specimen holders are attached to the other comers.
The central gear (2) is mounted on the shaft of a variable speed motor and
drives two pinions (5). Each pinion rotates an eccentric circular cam (4). The
horizontal relative slip of the contact interface is generated by the structural
deflections affected by the positive displacements of the two cam followers.
Each cam consists of two annular eccentric disks, which can be arranged
at different angular positions. This enables the adjustment of the total dif-
ferential eccentricity of the two disks to produce a preselected fretting
amplitude within the contact interface.
The specimen holders are made adjustable for leveling purposes. A spirit
level (17) and counterweight (18) are provided for a zero load initial setting.
Nominal contact pressures are applied by dead weights (11) which are
mounted onto a hollow loading pin (10). The loading pin allows a thermocou-
ple to be inserted into the bore of the specimen holder and contact the cylin-
EL-SHERBINY AND SALEM ON ION-PLATED COATINGS 127
&
• ^
drical specimen 1 mm below the contact interface. The cylindrical test

specimen, however, is a hollow pin with a hemispherical head of 10-mm
diameter. Observation of the initiation of fretting damage is possible through
a 6-mm thick glass plate (14) and an optical microscope. Finally, the number
of cycles to fretting damage is measured by a digital counter (not shown),
while the dynamic friction forces are measured by strain gages (9).
Coating Facilities
A standard coating unit IPAC-35 is used for vacuum coating, physical
vapor deposition in an inert gas, and ion plating. Figure 2 shows a schematic
layout of the ion-plating arrangement. Both sputtering and ion-plating pro-
cessing can be carried out simultaneously or separately on the same facility.
pirani gauge
Kigh v o l t a g e
D.C. supply S'.ibstrate
( 0 - 6 Kv)
,- metallic shield
glass jar
^ m e t a l l i c screen
rubber s e a l
Penning
Ionization gauge
Argon
a^ v a r i a b l e Y^^"* - i n l e t
leak valve
air inlet Cneedle}
Q valve
diffusion
pun^
a ®=
two way
valve
thermocouple
gauge
Qesxcant vapour t r a p
trap
[jff ^^^ i n l e t
valve
rotary pump
S^
FIG. 1—Schematic layout of an ion-plating unit.
EL-SHERBINY AND SALEM ON lON-PUTED COATINGS 129
The depositon of high-melting-point metal coatings was therefore made by

sputtering the coating material from a separate target and depositing the
sputtered particles onto the substrate by ion plating.
The ion plating procedure is to evacuate the conventional vacuum system
to 0.0667 N/m^ (5 X 10"'' torr). Argon gas is then introduced through the
variable leak valve to give a pressure approximately 1.33 to 6.67 N/m^ (0.01
to 0.05 torr). Sputter etching is carried out to clean the substrate surface
prior to and during the film deposition. As the film atoms are injected into
the gas plasma, they undergo ionization, collision, and energy sharing, and
finally are deposited on the substrate with enough energy to cause mobile
vacancies, resulting in extensive graded interfaces.
The ion-plated coatings in this study were produced at 30-)Lim-mercury
argon (4 N/m^) and a bias voltage of 4 kV, unless otherwise stated; the film
thickness was kept constant throughout at 1.5 ftm ± 10 percent.
Materials Used
Steel substrates meeting the ASTM Specifications for Mild- to Medium-
Strength Carbon-Steel Castings for General Application (A 27-79) were used
throughout the present investigation, the composition of which is given in
Table 1. Before plating, the test specimens were finely polished, cleaned in
alcohol, and dried in air. The coating materials (silver, lead, indium, tin,
titanium), however, were pure metals of 99.5 percent purity.
Results
Tests were made for both coated and uncoated steel substrates. The fol-
lowing are the testing conditions used in the present mvestigation: Load, 0.5
to 4 kg (= 5 to 40 N); speed frequency, 10 to 300 rpm/Hz; relative humidity,
60 ± 5 percent; environmental temperature, 26 ± 2°C.
Uncoated Steel
Figure 3 shows the measured weight loss against the fretting frequency, / ,
at constant slip amplitude, a, of 200 /nm and a constant test duration, n, of
80 000 cycles. It is shown that the fretting weight loss decreases as the fre-
quency increases. An asymptotic value, however, is reached at a test fre-
quency of 15 Hz. Low-frequency tests (below 10 Hz) indicate that the cycle
TABLE 1—Composition of the steel substrates.
c, % Mn, % Si, % P, % S, %
0.08 0.8 0.6 0.04 0.05
40 Co i * a c t P r e siw ' • p
0 10 MN/m'
• 30
a 200
30 n 80 Kcyc'es
•^^— •
I * •
20
,^^ 0
10 o 0 0
•O"—
0 o
"•
1 1 1 1 1 1
10 20 30
Frequency, Hz
FIG. 3—Weight loss-frequency relationship.
duration is sufficiently long for a surface oxide film to be generated. Further-

more, since the reciprocating motion is made at small amplitude and low fre-
quency, the generated oxide wear debris remains trapped within the contact
interface, thereby resulting in subsequent abrasive action.
Figure 4 shows the weight loss-fretting amplitude relationship at different
contact pressures, p (the loads are 4, 2.4, and 0.8 kg). In these tests the fret-
ting frequency was kept constant at the previously obtained critical frequency
of 15 Hz. Repeated tests with different amplitudes were made for a constant
duration of 40 000 cycles. The results indicate that fretting amplitudes below
40 nm do not cause significant wear loss. Higher amplitudes, however, result
in fretting damage which is proportional to the amplitude, particularly within
the range of 80 to 240 nm. Further increases in fretting amplitudes tend to
stabilize the weight loss and finally result in an independent wear loss, which
is believed to be due to a fatigue wear mechanism.
The number of cycles measured at the initiation of fretting between steel
and glass plate is shown for different frettmg amplitudes in Fig. 5. It is
shown that high-amplitude values tend to accelerate the onset of fretting
damage.
Comparisons Between Various Coating Techniques

In these tests silver is used to compare the effectiveness of vacuum-
deposited coatings, coatings deposited in an inert gas (such as argon), sput-
tered coatings, and ion-plated coatings. Figure 6 shows the number of cycles
measured at the initial damage of the film, during testing at contact
0 80 160 240
Amplitude ^m
FIG. 4—Weight loss-amplitude relationships at different contact pressures.
>0 10
Number of Cyc I e s
FIG. S—The surface life in fretting testing at different amplitudes.

10 10"
Film life. Cycles
FIG. 6—Comparison between various coating techniques.
pressures of 30 MN/m^ (the following results were obtained from an average

of ten tests). The vacuum-deposited coatings did not last long and in most
cases were torn away or peeled off before the number of cycles required to
fret the uncoated specimen was reached. Gas scattering resulting from the
admission of an inert gas into the coating chamber was apparently responsi-
ble for better film uniformity and possibly for slightly better adhesion. This is
reflected in the slightly longer life shown for the gas-deposited films. The
ion-plated films, however, showed considerably longer life at high slip
amplitudes. At low amplitude values, the differences in life between the sput-
tered and the ion-plated films were negligible.
The friction results of these tests are illustrated in Fig. 7. The general
trend is that the friction increases during the test, approaching an asymptotic
value of 0.5 for adherent coatings, while continuing to increase for poorly
adherent coatings up to a value of 0.8. It should be noted, however, that the
rate of increase in friction is very smooth, resulting in gentle slopes for
adherent coatings in comparison with the very steep slopes of the poorly
adherent coatings.
Ion-Plated Lead, Tin, and Indium Films

Ion-plated soft metallic coatings are proposed in this section to act as solid
lubricants to minimize the fretting wear of steel components. In the present
investigation lead, tin, silver, and indium films were tested under a nominal
contact pressure of 30 MN/m^ (with a load of 2.4 kg) and constant frequency
of 15 Hz. Figure 8 shows the useful life, before fretting damage takes place,
of the different ion-plated films at different slip amplitudes. Ion-plated lead
films demonstrated superior fretting resistance at low slip amplitudes, while
ion-plated tin showed the best resistance at high slip amplitudes. The
measured shear strengths of the four pure metals are compared in Table 2.
Note the very low shear strength of both indium and tin. The antifretting ef-
fectiveness of tin and indium at high amplitudes can, therefore, be explained
10 ?0 10 10
Number of Cycles n, Cycle
FIG. 7—Frictional behavior of the coatings tested.
P ; 3 0 MN/m
I IJ Hz
ION fiitlD
* *9
E * Pb
O In
A Sn
-L
10 10 10"
Useful Life, Cycle
FIG. 8—Comparison between the useful life of the ion-plated films at different amplitudes.
TABLE 2—Shear strength of the

pure metals.
Shear Value,
Metal MN/m^
Ag 37.77
Pb 26.49
Sn 12.6
In 11.77
by their superior lubrication properties at low shear strength. Because of the

limited motion at low slip amplitudes, metal oxides and oxide wear debris
tend to be trapped at the contact interface. The effectiveness of lead films
under such operating condition is, therefore, related to the good lubrication
properties of lead oxides (PbO) [13].
Titanium Carbide Coatings

Ion-plated titanium carbides [//] show very low wear coefficients in sliding
motions. Evaluation of their fretting resistance is therefore of special in-
terest. Previous results on fretting resistance of boron carbide [4] showed ef-
fective prevention of fretting up to 500 000 cycles. In these experiments the
film thickness of the boron carbide coatings was less than 1 /xm. It is,
therefore, of interest to examine the effect of film thickness on the useful life
of the film. In the present work the fretting life test was made for three
groups of coatings, each having a different film thickness. Figure 9 shows the
effect of slip amplitude on the useful life of the films. The increase in film
thickness of hard ion-plated coatings, unlike that of soft films, greatly im-
proved their useful life. The relationship between film thickness and life,
however, is not linear. Furthermore, the useful life of titanium carbide (TiC)
coatings of a given film thickness is at least 2 orders of magnitude higher
than the life of soft films having the same film thickness. The total fretting
weight loss of a TiC film, however, is considerably smaller and can be further
reduced by using thicker films. Figure 10 shows the measured weight loss at
different slip amplitudes for different film thicknesses. It is also interesting
to point out that for hard coatings, the critical amplitude at which significant
fretting weight loss takes place is slightly higher than that of soft lubricant
films.
Discassion
During the present investigation several attempts were made to separate
the contributions of mechanical and chemical fretting wear. Uhlig et al [14]
proposed an empirical equation of the form
W = (Ko^ - K^) — + K2apn
They evaluated the constants KQ, KX, and K2 for steel-on-steel tests and
found that
ATo = 5.05 X 10-^
K^ = 1.51 X 10-8
Kj = 4.16 X lO"*-
The proposed equation predicts wear losses having the same order of
magnitude as the values quoted in the present investigations, though the ac-
curacy of predictions is rather poor, and the equations appear to be oversen-
sitive to the effect of frequency on weight loss.
Equations like that of Uhlig et al were then used, but again the evaluated
constants failed to give consistent and reliable predictions of weight loss. The
reason is obviously related to the sensitivity of the two competing mechan-
isms—namely, fatigue and oxidation—to many operating conditions, system
properties, surface characteristics, and environmental conditions.
Ohmae and Tsukizoe [4] showed that boron carbide films provide the best
resistance to fretting at small slip amplitudes. Although the adhesion was low
( = 2.5 MN/m^), the film was effective in preventing fretting up to 500 000
- 4 ftm
0 1 ^m
140 \ m 0.5 fim
100 —
P 30 AIN/m2
t 15 Hz
-
60 1 1 1 ^"""^^
10 10 10
Useful Life, Cycles
FIG. 9—The amplitude—life relationship of ion-plated titanium carbides.
• 0.6 jum /
_ 0 1 urn L
A 4 tim 1 ^
n 100 Kcyc/es j l/
— f 25 "' fi
P 30
" 40 80 120 160
Amplitude , ^ni
FIG. 10—The weight loss of titanium carbides at different amplitudes.
cycles. In the present investigation, better adhesion of the titanium carbide

coatings was obtained by using an intermediate layer of titanium (0.1 /xm
thick) on steel substrates, followed by the deposition of titanium carbide up
to the required thickness. Ultrasonic, shear, pull, and centrifugal adhesion
tests showed no interfacial failure of the titanium carbide films. The useful
life in fretting prevention of the TiC film produced was, therefore, found to
be at least an order of magnitude higher than the quoted values of boron car-
bides [4]. As a result of this, the difficulties experienced in earlier investiga-
tions [4] on the adhesion of supremely wear-resistant films and the subse-
quent effects on film life can be resolved by using compatible interfacial
layers.
Conclusions
An experimental study of the fretting resistance of different surface
coatings has been presented. A new fretting test facility has been described
and the effectiveness of coatings produced by a number of coating techniques
demonstrated. The results illustrate the potential of ion-plated hard carbide
coatings in fretting alleviation and prevention. Predictions based on the ex-
isting empirical formula were somewhat different from the measured values;
therefore, development of realistic fretting wear models is badly needed.
References
[/] O'Connor, ]. J. and Boyd, J. J., Standard Handbook of Lubrication Engineering,
McGraw-Hill, New York, 1968, pp. 4-21.
[2] Waterhouse, R. B., Fretting Corrosion, Pergamon Press, Oxford, 1972, p. 197.
[3] Sarker, A. D., Wear of Metals. Pergamon Press, Oxford, 1976, p. 116.
[4] Ohmae, N. and Tsukizoe, T., Wear, Vol. 30, 1974, p. 299.
[5] Ohmae, N. and Tsukizoe, T., Wear. Vol. 27, No, 3, 1974, p. 281.
[6] El-Sherbiny, M. and Salem, F., Wear, Vol. 54, 1979, pp. 391-400.
[7] Tear, D. G. and Salem, F., "The Formation of Low Friction Wear Resistant Surface on
Titanium by Ion Plating," Proceedings. Conference on Ion Plating and Allied Techniques,
Edinburgh, England, June 1977, p. 126.
[8] El-Sherbiny, M. and Hailing, J., Wear, Vol. 45, 1977, pp. 211-220.
[9] El-Sherbiny, M. and Salem, F., Wear, Vol. 66, 1981, pp. 101-110.
[70] El-Sherbiny, M. and Salem, F., "Ion-Plated Coatings for Corrosion Prevention," Second
Metallurgical Conference, El-Tabin Institute, Cairo, Egypt, 9-12 Dec. 1978.
[//] Salem, F. and El-Sherbiny, M., "Wear Resistant Finishes," Paper No. MTSC-9, First In-
ternational Conference on Production, Design, and Control, Alexandria University, Egypt,
27-29 Dec. 1980.
[12] Hailing, J. and El-Sherbiny, M., "The Role of Surface Topography in the Friction of Soft
Metallic Films," Paper C46/78, presented in the Tribology Convention, Swansea,
England, April 1978.
[13] Tsuya, Y. and Takagi, R., Wear, Vol. 7, 1964, pp. 131-143.
[14] Uhlig, H., Feng, I., Tiemey, W., and McClellan, A., "Fundamental Investigation of Fret-
ting Corrosion," N.A.CA. Technical Note, No. 3029, Washington, D.C., Dec. 1953.
R. G. Vardiman, ^ D. Creighton, ^ G. Salivar, ^-^ A. Effatian, ^'^
andB. B. Rath^
Effect of Ion Implantation on

Fretting Fatigue in Ti-6AI-4V Alloy
REFERENCE: Vardiman, R. G., Creighton, D., Salivar, G., Effatian, A., and Rath, B.
B., "Effect of Ion Implantation on Fretting Fatigue in Ti-6A1-4V AJloy," Materials
Evaluation Under Fretting Conditions, ASTM STP 780, American Society for Testing
and Materials, 1982, pp. 138-149.
ABSTRACT: The effect of fretting on the fatigue life of titanium alloys is known to be
severely deleterious. A preliminary study shows ion implantation to hold great promise
for improving the fretting fatigue life in Ti-6A1-4V alloy.
Carbon implanted at 75 keV to a dose of 2 X l o " atoms/cm^ gives a peak concentra-
tion of 30 atomic percent at a depth of approximately 1700 A. Incoherent particles of
titanium carbide averaging 100 A in size were found by transmission electron micros-
copy. Fatigue specimens were subjected to fretting on one surface at a normal stress of
20.7 MPa. A significant improvement in lifetime was found. No difference in the mode of
failure was observed between the implanted and the unimplanted specimens.
KEY WORDS; materials fatigue, fretting, ion implantation, titanium alloys, electron
microscopy
The reduction of fatigue life due to fretting has been identified as a serious
problem in a wide variety of structural alloys and components [1].^ Par-
ticularly in the case of titanium alloys, the reduction in the lifetime and en-
durance limit has been shown to be severe [2]. Although various methods
have been proposed to reduce the effect of fretting [1], none has proved satis-
factory in all respects.
Ion implantation has been shown to improve the resistance of metals and
'Metallurgist and head. Physical Metallurgy Branch, respectively. Material Science and
Technology Division, Naval Research Laboratory, Washington, D.C. 20375.
^Professor and graduate students, respectively, Department of Mechanical and Aerospace
Engineering, University of Missouri-Columbia, Columbia, Mo. 65201.
^Present address: senior materials engineer, Pratt & Whitney, West Palm Beach, Fla. 33402.
''Present address: project engineer, G and H Technology, Inc., Santa Monica, Calif. 91311.
138
VARDIMAN ET AL ON EFFECT OF ION IMPLANTATION 139
alloys to surface-sensitive phenomena such as wear, fatigue, and corrosion

[3]. Since all three of these phenomena occur concurrently during fretting
fatigue, implantation of suitable ions would appear to be an excellent pros-
pect for improving material lifetime under these conditions.
Ion implantation begins with the formation of an ionized plasma from
either a sputtered source or a dissociated gas. The ions are passed through a
magnet, and the isotope of interest is selected and then accelerated to
energies typically in the range of 25 to 200 keV. Figure 1 illustrates this pro-
cedure for implantation of chromium in any target using the Naval Research
Laboratory (NRL) implanter. Following a series of successive collisions with
the lattice atoms, the ions become embedded in the target forming roughly a
Gaussian distribution of the implanted species at depths ranging from hun-
dreds of angstroms to several thousand angstroms. The incident ions also
knock target atoms out of their lattice positions, creating a damage layer
consisting of interstitials, vacancies, and dislocation loops. This damage can
become extremely heavy at ion doses of 10^^ atoms/cm^ or greater, which are
typically used for mechanical property improvements.
For implants of light atoms such as carbon or nitrogen, sputtering of the
target surface is negligible, and peak concentrations of 40 atomic percent or
greater can be obtained, most of which is retained in solution. Such metasta-
ble surface alloys are unique and have exhibited effects on surface-sensitive
properties quite out of proportion to the thickness of the layer directly af-
fected by the implantation.
HIGH-VOLTAGE TERMINAL
TARGET CHAMBER
ION SOURCE
ENCLOSURE AT GROUND POTENTIAL
FIG. 1—Schematic drawing of the Naval Research Laboratory ion implanter, illustrating
isotope selection using a chromium beam.
At the present time, no implanted species is known to give substantial im-

provement in both oxidation and wear resistance of structural alloys. The ox-
idation of titanium at 600° C was improved by implantation of alkaline earths
and divalent lanthanides [4]. Among these, barium was reported to improve
fretting fatigue in Ti-6A1-4V alloy [5]. No wear studies on ion-implanted
titanium have been published, but carbon implantation has been shown to
give considerable improvement in abrasive wear resistance.* Carbon also
gives a strong increase in fatigue life in Ti-6A1-4V alloy [6].
In the present study, carbon was selected as the implanted species. The
fretting fatigue life was measured on both implanted and unimplanted speci-
mens. A substantial improvement in lifetime was found for the implanted
specimens.
Experimental Methods
The alloy Ti-6A1-4V was used for both the fretting pads and the fatigue
specimens. The microstructure of the fatigue specimens is shown in Fig. 2.
The yield strength of this material was measured as 993 MPa. Each specimen
was polished to 600 grit paper lengthwise on each side. The fretted side was
further polished to less than 0.1 ;um rms with 1-^m alumina, as was the sur-
face of the fretting pad.
Since only one side of the fatigue sample was fretted, only this side was im-
planted, in the necked region. Carbon ions were implanted at an energy of 75
keV to a dose of 2 X 10'^ atoms/cm^. The temperature was estimated not to
exceed 100°C during implantation. These conditions produced a concentra-
tion profile of carbon shown in Fig. 3, as determined by nuclear reaction
analysis [7]. The fretting pads were not implanted.
The microstructure developed in the implanted layer was characterized by
transmission electron microscopy (TEM). Three-millimetre disks about 100
/im thick were partially thinned on one side, implanted on this electropol-
ished side, and then thinned through from the opposite side. A 200-keV
transmission electron microscope was used.
The fretting fatigue apparatus used in this study has been described
elsewhere [8,9]. Briefly, it is a closed-loop electrohydraulic servo-controlled
system, with the fatigue stress applied axially on the sample as a sine wave of
frequency 10 Hz and R ~ + 0 . 1 . A normal stress of 20.7 MPa measured
through a load cell was applied to the fretting pad for all tests. Changing the
level of the normal stress to 41.4 MPa has been found to have a negligible ef-
fect on the lifetime for Ti-6A1-4V alloy [10]. The fatigue specimen and fret-
ting pad dimensions are shown in Fig. 4.
After fracture, both the fretted surface and the fracture surface of the
fatigue specimens were examined by scanning electron microscopy (SEM).
^Bolster, R. N., Naval Research Laboratory, personal communication, 1981.
FIG. 2—Fatigue specimen microstructure. rolling plane.
Results
The layer damaged because of implantation appears in transmission
microscopy as a very dense, unresolvable array, as seen in Fig. 5, (top). Elec-
tron diffraction patterns show rings superimposed on the spot pattern of the
titanium grains (insert of Fig. 5, bottom). These rings correspond to a face-
2 X 1 0 " C^/cm^
IN Ti-6AI-4V
ELECTROPOLISHED
100 150 200

DEPTH BELOW SURFACE (nm)
FIG. 3—Concentration of carbon near the electropolished surface of a Ti-6A1-4V alloy speci-
men for a dose o / 2 X 10'^ atoms/cm^ at 75 keV. Obtained by nuclear reaction analysis using
'^C(^He, ao>"C.
centered cubic phase with a lattice parameter of 4.24 A. This may be com-
pared with the lattice parameter of face-centered cubic titanium carbide,
which is 4.33 A [11]. Dark-field imaging using segments of the two innermost
rings shows a fine dispersion of precipitates with a maximum size of about
200 A (Fig. 5, bottom). Although the total volume of precipitate is difficuh to
quantify, it would appear that a large portion of the implanted carbon re-
mains in solution.
The results of the fretting fatigue tests are shown in Fig. 6. The baseline
fatigue data are from Ref 2 and are shown for comparison purposes. The
relatively high tensile strength of our fatigue specimens may be reflected in a
higher fatigue strength. The unimplanted fretting curve, in fact, lies slightly
higher than that found in Ref 2. The main point, however, is that significant
improvement in fretting fatigue lifetime is found with carbon implantation.
Examination of the fretted and fractured surfaces in the scanning electron
microscope showed no differences between the implanted and unimplanted
specimens. The general aspect is typical of fretting fatigue in this material
[2]. In Fig. 7 (left), score marks on the polished surface lead to a fretting pit.
In Fig. 7 {right), debris and small cracks are found at the edge of a fretting
pit. Figure 8 {left) shows the intersection of the wear and fracture surfaces,
.. 0.25
— -« 0.25
1 H «-O.I25
1 - >
0.2S
m t
V--
1
1
1
0.750
J^ '----' ,
0.125DIA —» 1
1
# /*""
-«—0.750 - -0.25
FIG. 4—Size and configuration of the fatigue specimens and the fretting pad. The measure-
ments are in inches: 1 in. = 25.4 mm.
again with debris and secondary cracking. A view of the fracture surface near
the wear surface in Fig. 8 (right) shows secondary cracks and some fatigue
striations. Failure seems to be definitely associated with the surface damage.
Discussion
Although the mechanism of failure does not appear to be affected by im-
plantation, there is an obvious increase in the lifetime. It is unlikely that this
is due simply to an increase in fatigue strength, as only one side of the fatigue
specimen was implanted. Increases in fatigue strength would be expected
1 -a
00 a
3 -a
a =
• ^ ^
a »>
•a ^
ê
B-SS
>e
^ (J
-s Si
II = s
a =
u .a
-a •-
1^ u
5 'S
£^
a a
a »j
S 5
O Oc
S
g.a
I "
160
140 v,^
BASELINE FATIGUE.
NO FRETTING (2)
- 120 \ ^ XK
(0
M
111
cc
!« 100 XX X X
^ ^
s \ ^ X X
s 80 - ©•^-^p
o- UNIMPLANTED
X- IMPLANTED,
2 X 10"c+/cm^ o*
1
40 - 1
10' 10* 10' 10'

CYCLES TO FAILURE
FIG. 6—Cycles to failure versus the maximum fatigue stress for carbon-implanted and unim-
planted Ti-6Al-4V alloy (100 ksi = 689 MPa; normal stress, 20.7 MPa). The baseline fatigue is
taken from Ref 2.
only when the entire potential fracture-originating surface was implanted.

The increase in lifetime could reasonably be ascribed to a slowing of debris
formation and damage on the hardened, implanted surface. Eventually the
implanted layer is breached, and a fracture develops in the normal manner.
We have chosen carbon as an implant species in this initial study basically
because of its hardening properties. It would be very interesting to compare
the effects of oxidation-inhibiting implants, as we hope to do in future work.
Also, there are a number of parameters, such as dose, energy, and postim-
plant heat treatment, which could be explored to optimize the increase in
lifetime. Increased understanding of fretting fatigue mechanisms should
help point the way to improved implantations. We may conclude that ion im-
plantation is an extremely promising technique for alleviating the problem of
fretting fatigue.
Acknowledgment
Dr. R. A. Kant was very helpful in obtaining the implantations. The
nuclear reaction analysis was performed by C. R. Gossett.
•a
a
•a
a
•^
I
to
•a
s
3
II
II
ft
t
I!
II
It
t-g=
f^
it
References
[1] "Control of Fretting Fatigue," Publication NMAB-333, Report of the Committee on Con-
trol of Fretting-Initiated Fatigue, National Materials Advisory Board, Washington, D.C.,
1977.
[2] Goss, G. L. and Hoeppner, D. W., Wear, Vol. 24, 1973, p. 77.
[3] Hirvonen, J. K., Journal of Vacuum Science and Technology, Vol. 15, 1978, p. 1662.
[4] Benjamin, J. D. and Deamaley, G., Institute of Physics and Physical Society, Conference
Series, Vol. 28, 1976, p. 141.
[5] Deamaley, G.,I.E.E.E. Transactions on Nuclear Science, Vol. NS-28, 1981, p. 1808.
[6] Vardiman, R. G. and Kant, R. A., Journal of Applied Physics, in press.
[7) Gossett, C. R., Proceedings of the Fifth International Conference on Ion Beam Analysis, in
press.
[8] Reeves, R. K. and Hoeppner, D. W., Wear, Vol. 45, 1977, p. 127.
[9] Reeves, R. K., "Micro Structural and Environmental Effects on Fretting Fatigue," Ph.D.
thesis, University of Missouri-Columbia, Columbia, Mo., 1977.
[10] Hoeppner, D. W. and Goss, G. L., Wear, Vol. 27, 1974, p. 61.
[11] Krikorian, N. H., Wallace, T. C , and Anderson, J. L., Journal of the Electrochemical
Society, Vol. 110, 1963, p. 587.
C. Lutynski, ^ G. Simansky, ^ and A. J. McEvily ^
Fretting Fatigue of Ti-6AI-4V Alloy
REFERENCE: Lutynski, C, Simansky, G., and McEvily, A. J., "Fretting Fatigue of Ti-
6AI-4V Steel," Materials Evaluation Under Fretting Conditions, ASTM STP 780,
American Society for Testing and Materials, 1982, pp. 150-164.
ABSTRACT: A study of the fretting fatigue process in sheet specimens of Ti-6A1-4V alloy
has been carried out. In the fretting experiments two fretting pads, also of Ti-6A1-4V,
were clamped to the fatigue specimen by a calibrated proving ring. In the experiments
the fretting pressure and the fatigue stress amplitude were varied. Specimens in the as-re-
ceived condition as well as shot-peened specimens were tested to study the effect of sur-
face condition on resistance to fretting. An approximate analysis of the shear transfer
process was developed, and predictions are compared with the measured size of the fret-
ting scars. The fatigue results of this investigation are discussed in terms of test variables
as well as the predictions of the approximate theoretical treatment.
KEY WORDS; materials fatigue, fretting fatigue, titanium alloys, shot peening, fretting
As the use of titanium alloys in applications involving fretting fatigue in-

creases, the need for a clear understanding of the fretting fatigue process
becomes necessary [1].^ Some information which contributes to our under-
standing of the importance of factors such as surface hardness, surface
roughness, clamping pressure and the hardness of the clamps, shot peening,
and the amplitude of cyclic stressing has already been obtained and reviewed
by Waterhouse [2]. However, our overall understanding of the nature of the
fretting fatigue process is far from complete, and since, for example, the pro-
cess can be so dominant as to minimize differences in fatigue strength of the
alloys which would otherwise exist [3], additional work is warranted. There-
fore, as a step toward furthering the understanding of the fretting fatigue
process, the present investigation was undertaken. The study focuses atten-
' Graduate student, former graduate assistant, and professor, respectively. Metallurgy De-
partment and Institute of Materials Science, University of Connecticut, Storrs, Conn. 06268.
150
Copyright® 1982 b y AS FM International www.astm.org
LUTYNSKI ET AL ON FRETTING FATIGUE OF Ti-6AI-4V 151
tion on the effect of clamping behavior on fretting scar characteristics as well

as on the effect of fretting on shot-peened and non-shot-peened specimens.
Apparatus
Fatigue tests were carried out under axial load under R = 0.05 test condi-
tions (R is the ratio of the minimum to the maximum stress in a loading cy-
cle). In order to study the combined effect of fatigue and fretting, a fretting
device capable of measuring the normal or clamping load was constructed.
This device utilized a proving ring to clamp two fretting pads to the fatigue
specimen. The proving ring is an elastic steel ring modeled after the devices
of Waterhouse [4], Fenner and Field [5], and Cook and Rabinowicz [6] and
constructed of Type 4340 steel. Two steel belts threaded through the ring
were used to press the two' fretting pads—one on each side of the specimen
and opposite to each other, against the specimen surface. The clamping force
was reacted by the ring, and a strain in the ring developed. This strain was
measured by four wire strain gages cemented to the ring in a Wheatstone
bridge circuit. An Instron machine was used to calibrate the strain gage
readings as a function of the clamping force. The weight of the unit was ap-
proximately 27 N. During testing the output of the bridge circuit could be
monitored directly as well as continuously recorded. The appearance of the
proving ring as mounted in the fatigue test machine is shown in Fig. 1. In the
lower left of this figure is a device which was used to hold the fretting pads
during surface finishing on a metallographic polishing wheel. The material
chosen for these tests was the alloy Ti-6A1-4V in 1.27-mm thick sheet form.
The test specimens were tensile strips 9.53 mm wide. Prior to testing, the sur-
faces of all of the specimens were metallographically polished in the 9.53-mm
wide test section to a 0.25-ixm. rms finish, as determined by a Talysurf 10
unit. The fretting pads were made of forged Ti-6A1-4V alloy and were also
polished to a 0.25-^m finish. The contact area of the pads was square in
shape and 6.3 by 6.3 mm in size. The hardness of the polished specimen was
350 VHN (50-g load). The corresponding value for the shot-peened surface
was 430 VHN. The shot-peening process utilized 110-size shot. The peening
intensity was 4 to 6 Aimers, with the peening procedures in accordance with
specification MILS 13165-B.
Fatigue tests were carried out under axial load at a test frequency of 30 Hz,
with clamping pressure up to 31.05 MN/m^ as a test variable. The test envi-
ronment consisted of laboratory air at a temperature of 20°C and 50 percent
relative humidity. The fretting device was clamped in place after the mean
load had been applied to the specimen. The minimum clamping force was on
the order of 444.8 N, which was sufficient to hold the 27 N proving ring in
place. The slip amplitude was not measured in these tests. Additional fret-
ting tests at a variety of amplitudes and clamping pressures were run not to
failure but to 10^ cycles to develop fretting scars.
FIG. 1—The fretting device mounted on a specimen in the fatigue machine. The apparatus at
the tower left is used in fretting pads during polishing.
Results
Table 1 provides a list of the fatigue test results. These results are also
plotted in Fig. 2. The lengths of the fretting scars are given in Table 2 for the
various fretting conditions. It is noted that fretting drastically reduced the fa-
tigue strength. For example, at 10^ cycles the fatigue strength of a nonfretted
specimen is approximately 620 MN/m^ (90 ksi), whereas the corresponding
value for a fretted specimen is around 310 MN/m^ (45 ksi), a 50 percent re-
TABLE I—Fatigue test results (R = 0.05).
Peak Fatigue Clamping

Stress, MN/m^ Pressure, MN/m^ Nf, Cycles
Polished Specimens
%6 0 1.4 X lO''
897 0 1.4 X lO''
828 0 3.6 X lO"*
828 0 2.5 X IC*
828 0 5.5 X IC*
759 0 7.3 X 10''
690 0 5.0 X 10''
621 0 5.144 X 10*
828 10.35 4.1 X 10"
759 10.35 5.4 X 10''
690 10.35 7.8 X 10"
690 10.35 1.11 X 10-^
690 10.35 6.9 X 10"
621 10.35 1.05 X 10^
621 10.35 7.7 X 10"
586 10.35 6.3 X 10"
552 10.35 1.05 X 10^
483 10.35 1.28 X 10*
483 10.35 8.09 X 10=
414 10.35 3.557 X 10"
345 10.35 1.090 X lO*"
828 20.7 3.7 X 10"
759 20.7 4.4 X 10"
690 20.7 6.8 X 10"
690 20.7 5.8 X 10"
621 20.7 1.33 X 10^
483 20.7 1.66 X 10^
414 20.7 2.08 X 10^
345 20.7 2.754 X 10'
276 20.7 1.027 X 10'
483 31.1 7.8 X 10"
414 31.1 1.94 X 10^
345 31.1 1.5109 X 10'
Shot-peened Specimen s
828 0 1.8 X 10"
828 0 6.3 X 10"
828 0 3.6 X 10"
828 0 9.6 X 10"
828 0 3.6 X 10"
828 0 3.1 X 10"
759 0 8.6 X 10"
759 0 2.12 X 10*
759 0 4.2 X 10^
724 0 5.59 X 10^
690 0 >10*
690 0 1.783 X 10*'
621 0 <10''
828 10.35 6.2 X 10"
759 10.35 5.5 X 10"
TABLE I—Continued.
Peak Fatigue Clamping

Stress, MN/m-^ Pressure, MN/m^ iVf, Cycles
690 10.35 1.52 X 10^

621 10.35 4.36 X 10^
828 20.7 3.4 X 10"
759 20.7 8.8 X 10"
690 20.7 2.02 X 10^
690 20.7 1.5 X 10^
621 20.7 1.307 X lO'
552 20.7 1.39 X 10'
TABLE 2—Fretting scar lengths.
Maximum Clamping
Fatigue Stress, Pressure, Scar Lengtli,
MN/m^ MN/m^ cm
103.5 10.35 0.55

172.5 10.35 0.81
207.0 10.35 1.17
207.0 10.'35 1.05
241.5 10.35 1.40
310.5 10.35 1.70
317.4 10.35 1.57
414.0 10.35 1.93
103.5 20.7 0.45
172.5 20.7 0.46
207.0 20.7 0.66
207.0 20.7 0.50
262.2 20.7 0,76
310.5 20.7 1.25
317.4 20.7 1.02
414.0 20.7 1.25
414.0 20.7 1.37
483.0 20.7 1.55
621.0 20.7 2.05
634.8 20.7 1.96
duction. On the other hand, shot-peened specimens show a remarkable resis-

tance to fretting fatigue, with test results in the high cycle region being only
slightly below those of specimens that are polished but not fretted. It is also
interesting to note that under the test conditions employed, the highest
clamping pressure, 31.05 MN/m^, does not appear to be significantly more
damaging than the lightest pressure, 10.35 MN/m^. If there are any overall
trends as to the influence of pressure on fatigue life, they are difficult to
detect with certainty within the scatter in fatigue test results.
o
o
(Nl
% o
c
tn
ui
o >-
u. c
z o
•J! 13
S ^
s
«) . a; j
u
'?-'^
•KJ
o
•«:
65 u
o c
'^ •3
(•nS
•a •S
a -S!
2
c s
• ^
« ^ 1v
a
•Sf «J
IS
^ .*:
h^
g1
1 .•C
r^ •3
o s
d8
»—( 2?
a.
(90o=a) YsH 'ss3dis anoiivd wniNixvw •S
The typical appearance of a fretting scar is shown in Fig. 3. It is noted that

despite the 0.25-^m finish on both the specimen and pads, the degree of fret-
ting damage is not uniform in appearance. It is also noted that the intensity
of the fretting pattern decreases toward the center of the contact region and
that, in fact, a no-slip region is present. A Talysurf trace through the contact
region in the longitudinal direction of a specimen is shown in Fig. 4. The
roughness in the no-slip region is representative of the degree of surface
roughness initially present after polishing. Considerably greater roughness
develops during the fretting process in the slip region. Additional observa-
tions of this nature revealed that the extent of the slipped region increased
with fatigue stress amplitude and decreased with an increase in the clamping
pressure, as shown in Fig. 5. In view of the irregular nature of the scars, some
judgment had to be exercised in preparing this figure, but it is considered
that the trends are representative. Limited information obtained with the
Talysurf unit indicated that at a given low fatigue amplitude (114.5 MN/m^)
the degree of roughness appeared to decrease with an increase in the clamp-
ing pressure. However, at a high peak fatigue amplitude (170.4 MN/m^) the
degree of roughness appeared to be insensitive to the clamping pressure (Fig.
6). For a given clamping pressure the roughness increased with the fatigue
stress amplitude.
Figure 7 shows the appearance of a fretting scar in greater detail. In this
figure the no-slip region is at the bottom, and a fatigue crack is seen running
across the top of the scar. This was a typical location for crack formation in
most of these tests. A closer view of a region in which a layer of material has
been removed from the specimen surface by fretting fatigue action is shown
in Fig. 8. Such an effect is consistent with subsurface crack formation (de-
lamination) in a plane parallel to the surface. Another close-up of the same
fretting scar is shown in Fig. 9. In this case, a number of microcracks are
present. Figure 10 provides an example of the fretting damage for a shot-
peened surface. It is seen that the degree of surface roughness developed is
much less than in the case of the polished surfaces. Clearly the difficulties in-
volved in making a quantitative analysis of fatigue life in the presence of such
surface damage are evident.
Discussion
In this discussion we will consider two aspects of the fretting fatigue pro-
cess, namely, the factors influencing the size of the slip region and the influ-
ence of clamping pressure on fatigue life.
If we assume that in the no-slip region there is no relative displacement be-
tween pad and specimen, then we can put the strain at the boundary of the
no-slip region at the surface of the specimen equal to that at the surface of
the pad. Some shear strain of an elastic nature may still exist at the interface
vCvS''*-*'^-^'"^:'' '*''-*'*^^'*aiW^
I
I
.1
"s^
I
&
1^5
O 3
NO SLIP
REGION
K H i 1.0 Mm
K -0.25 in.— M
(6.36 mm.)
FIG. 4—A Talysurf trace through the contact region in the longitudinal direction of the speci-
men. Fatigue maximum stress = 634.8 MN/m^; clamping pressure = 13.8 MN/m^; R = 0.05.
100
in
o 600
o
CLAMPING PRESSURE
3.0 ksi
m= 45.26
AOO
CO
LU
cr
S
CLAMPING PRESSURE 2
1.5 ksi
200
m= 28.09
1.0 1.5 2.5

SCAR LENGTH, mm
FIG. S—The length of the fretting scar as a function of fatigue stress amplitude and clamping
pressure. The slope, m, of each line was determined by linear regression analysis.
in the no-slip region, but this is neglected in this analysis. This shear transfer
due to interfacial friction leads to an increase of interfacial stress in the axial
direction of the pad. This interfacial stress is zero at the edge of the pad and
increases to a maximum value at the edge of the no-slip region. As a result
there is a transfer of axial load from the specimen to the pad. This transfer
reduces the interfacial stress in the specimen to a minimum value at the edge
LUTYNSKI ET AL ON FRETTING FATIGUE OF TI-6AI-4V 159
— - ^ 1 . 0 Mm
Y^^vrfA^- -^||>^TVV^>— -MAIV^T
FIG. 6—Talysurf traces in the transverse direction of the specimen through the midregion of
the fretting scars, (a) Fatigue stress maximum = 241.5 MN/m^; clamping pressure = 6.9
MN/m^; (b) fatigue stress maximum = 241.5 MN/m^; clamping pressure = 13.8 MN/m^; (c)
fatigue stress maximum = 241.5 MN/m^; clamping pressure = 20.70 MN/m : (d) fatigue
stress maximum — 635 MN/m^; clamping pressure = 6.9 MN/m^; (e) fatigue stress maxi-
mum = 635 MN/m^: clamping pressure = 13.8 MN/m^: (f) fatigue stress maximum = 635
MN/m ; clamping pressure = 20.70 MN/m .
of the no-slip region. At this point, since there is assumed to be no relative

motion between the pad and the specimen, the stress in the pad and the spec-
imen are the same. If the transfer of stress is linear over the transfer length,
i x . the interfacial stress in the pad builds up to one half of the applied stress
in the specimen, while the interfacial stress in the specimen decreases to this
same level, that is, one half of the axial stress away from the pad. At the in-
terface in the fretting region, the load is transferred from the specimen to the
pad. If we take the friction developed in the transfer region to be iioj^, where
H is the coefficient of friction andCTNis the normal stress, that is, the force on
the pad divided by the area of the pad, then at the no-slip boundary
Cs = Ep (1)
where e, is the strain in the specimen surface, and ep is the strain in the pad
surface. Also
ffs = ap = a^/l
where a^ is the fatigue stress amplitude. We can then obtain the following ex-
pression involving the transfer length, Xx
where 5 is a constant. Therefore
XT «— (2)
ffN
^ ^ ?'/ s
•-^^^ X
'": ."^.-^^
i^
o
•^
'^^^n^h '^H
X
J' ' ^ ^ ^ * ^r
^SCf:u II
^J
i S i ^ jStrjk,
1
H^-- 5
^
I
I
tS
I
r-- •
i5i
FIG. 8—Delaminated regions. Fatigue stress maximum = 759 MN/m^; clamping pressure =
10.35 MN/m^; Nf = 5.4 X Iff' cycles. The no-slip region below a fretted area is shown fX125
magnification/.
This result indicates that the length of the fretting scar initially should be in
proportion to the stress amplitude and inversely dependent upon the normal
pressure. In Fig. 5 it is apparent that there is indeed a linear dependency of
the transfer length on the stress amplitude for a given clamping pressure.
However, the length as a function of the clamping pressure at a given stress
amplitude does not vary as greatly as predicted. For example, in comparing
the scar width for 10.35 and 20.7 MN/m^ clamping pressures, the ratio
should be 2:1, whereas it is more like 1.6:1. The reason for this discrepancy
has not been established. Further, once fretting debris has developed, the no-
slip region may in fact become a no-contact region, if the debris serves to
wedge the pad away from the surface, as suggested by the Talysurf trace
shown m Fig. 4.
Next, we turn our attention to the more difficult task of trying to assess the
effect of clamping pressure on fatigue. Again, a simplifying assumption is
needed. It is noted that an increase in pressure reduces the length of the
fretted region, but at the same time the shear stress in the interfacial transfer
region increases since we have taken it to be proportional to the normal pres-
sure. We take as an indication of damage the shear stress, r, multiplied by
FIG. 9—Fretted region. Fatigue stress maximum = 759 MN/m^; clamping pressure — 10.35
MN/m^; Nf = 5.4 X Iff' cycles. Note the fine cracks and no-slip region below the fretted area
shown ex 125 magnification).
displacement, A R Q . This product, TARD, is a measure of the wear seen at a

point in the interfacial region. The average stress in the fretted region in the
specimen at the interface is 3/4(TA! that in the pad is 1/4a^- Therefore, the
relative displacement, ARD , is given by the equation
Ok CA'
ARD = (3/4 - 1/4) XLT = • a • (3)
IE IEOK
vifhere i x is the transfer length, that is, the length of the fretting scar. The
product of the shear stress multiplied by this displacement is
<JK'
fia^ X (4)
2Ea^ 2E
which indicates that at a given fatigue stress amplitude there should be little
effect of the normal stress on fatigue for the particular test conditions utilized
in this study. This conclusion appears to be in agreement with the results of
FIG. 10 The appearance of the fretted region on a shot-peened specimen. Fatigi

gue stress
MN/m^; : clamping pressure = 20.7 MN/m. The no-slip region is: above
maximum = 759 MN/m a the
area shown ('X125 magnification).
the fretting fatigue tests shown in Fig. 2, where an increase in stress ampli-
tude is much more damaging in fretting fatigue than is an increase in normal
pressure. It is noted that Eq 4 is analogous to Archard's law of adhesive wear
[7] and implies that, for a given fatigue stress amplitude, equal wear will lead
to equal lifetimes. It is of interest that the wear process in the absence of
cyclic loading has been considered to be a fatigue process in that the stresses
at any point are cyclic with repeated passage of the slider. In addition, cyclic
stresses on a finer scale can arise due to the fluctuation of stresses associated
with asperities. If the wear process is a form of fatigue, then the application
of cyclic loading, as in fretting fatigue, may merely serve to compound a
cyclic-damage process.
Conclusions
The fretting fatigue behavior of sheet specimens of Ti-6A1-4V alloy has
been investigated, and the following conclusions have been reached:
1. Fretting fatigue drastically reduces the fatigue strength.

2. The extent of the fretted region varies with the external parameters in
accordance with considerations of shear transfer from the specimen to the
pad. That is, an increase in the stress amplitude or a decrease in the normal
pressure will result in a larger fretting region.
3. A variation in normal pressure under the conditions investigated has
relatively little effect on fatigue life.
4. Shot peening markedly improves the resistance to fretting fatigue. This
improvement is due to an increase in the hardness of the specimen surface in
relation to the pad which was not shot peened and also due to the introduc-
tion of beneficial compressive residual stresses which would serve to retard
the propagation of incipient fatigue cracks.
Acknowledgment
The authors express their appreciation to the Air Force Office of Scientific
Research for support of this investigation. The assistance of Fabio Hernan-
dez of the Metal Improvement Company in shot peening the samples is also
gratefully acknowledged.
References
[/] Wayne, S. F., Nowotny, H., and Wee, S. L., Titanium 1980, Proceedings of the Fourth In-
ternational Conference on Titanium, H. Kimura, Ed., the Metallurgical Society, American
Institute of Mechanical Engineers, New York, 1980, p. 1295.
[2] Goto, S. and Waterhouse, R. B., Titanium 1980, Proceedings of the Fourth International
Conference on Titanium , H. Kimura, Ed., the Metallurgical Society, American Institute of
Mechanical Engineers, New York, 1980, p. 1837.
[3] Waterhouse, R. B., Fretting Corrosion, Pergamon, Oxford, 1972.
[4] Waterhouse, R. B., in Corrosion Fatigue, National Association of Corrosion Engineers,
Houston, 1971, p. 608.
[5] Fenner, A. J. and Field, J. E., North East Coast Institute of Engineers and Shipbuilders,
Vol. 76, 1960, p. 183.
[6] Cook, N. H. and Rabinowicz, E., Physical Measurement and Analysis, Addison-Wesley,
Boston, 1963.
[7\ Archard, I. ¥., Journal of Applied Physics, Vol. 24, 1953, pp. 981-988.
[5] Hirth, J. P. and Rlgney, D. A., "Microstructural Models for Friction and Wear," Proceed-
ings of the International Conference on Strength of Metals and Alloys, Aachen, Germany,
1979.
R. a Biii^
Review of Factors That Influence

Fretting Wear
REFERENCE: Bill, R. C , "Review of Factors That Influence Fretting Wear," Materials

Evaluation Under Fretting Conditions. ASTM STP 780, American Society for Testing
and Materials, 1982, pp. 165-182.
ABSTRACT: Experimental results obtained by the authors and results from the
literature are summarized to provide a critical review of the effects of various factors on
fretting wear. These factors fall into three broad categories: (1) contact conditions, (2)
environmental conditions, and (3) material properties and behavior. Contact load,
amplitude and frequency of motion, and contact geometry are included under contact
conditions. Environmental conditions include atmosphere, relative humidity, and
temperature. The importance of material properties is discussed on the basis of how
material properties interact with other factors during fretting. Individually, the effect of
contact load, fretting amplitude, temperature, and atmosphere have been fairly well ex-
plored for a limited number of materials (mostly mild steel). More work is clearly needed
to provide basic understanding of the effects of contact geometry and relative humidity
and the interaction of various fretting parameters. A more comprehensive number of
materials should be studied to clarify material property effects.
KEY WORDS: fretting, fretting wear, oxidation, materials fatigue, relative humidity,
temperature, slip amplitude, debris
In its broadest sense, fretting may be defined as surface damage caused by

low-amplitude oscillatory sliding between two contacting surfaces. Within
this definition are included fretting wear, fretting fatigue, and fretting corro-
sion. Distinctions between these three fretting classifications are dictated
more by the nature of a specific problem or objective of a particular study
than by any fundamental difference in the processes taking place at the con-
tacting surfaces.
Fretting wear is the removal of material from contacting surfaces through
fretting action. The practical consequences of this action include the loss of
'Materials engineer. National Aeronautics and Space Administration, Lewis Research

Center, Cleveland, Ohio 44135.
165
Copyright 1982 b y AS FM International www.astm.org
(PDVSA Los Teques) pursuant to License Agreement. No furt
dimensional control at critical interfaces, which can lead to improper func-

tioning of the mechanical components.
Fretting fatigue is the reduction in fatigue life due to fretting surface
damage. It is encountered in cyclically stressed assemblies at locations where
there is an opportunity for the occurrence of slip between assembled com-
ponents. The ultimate consequences of fretting fatigue include fracture of
the most critically loaded element of the assembly.
Fretting corrosion involves strong chemical interaction between the en-
vironment and the fretting process, resulting in accelerated fretting wear or
fretting fatigue. Most commonly, the corrosive process is oxidation, but cor-
rosion in its most general sense may interact with fretting.
Practical fretting problems are especially pernicious. Usually the damage
area is hidden and difficult to inspect without disassembly. There is really no
uniform approach to predicting and circumventing fretting problems in the
design stage, and very likely many fretting problems that do emerge are not
diagnosed as such. Sometimes fretting damage at a given location, for exam-
ple, at a shaft spline, can lead to dramatic changes in the dynamic loading of
the entire mechanical system and to ultimate failure at a location quite
remote from the site of fretting damage.
In any frettmg situation, the extent and consequences of fretting damage
depend on a large number of factors that beg rational consideration. The in-
fluences of various experimental parameters on fretting wear will be con-
sidered in the following sections.
Factors that Affect Fretting

Factors known to influence fretting may be divided into three broad
categories, as shown in Fig. 1. These categories are (1) contact conditions, (2)
environmental conditions, and (3) material properties and behavior. These
factors may interact with one another, as indicated in Fig. 1, to influence
both the nature and the extent of fretting damage. For example, under con-
formal contact conditions certain environmental factors may be excluded
from the contact area and not strongly influence fretting, whereas under con-
centrated contact conditions the same environmental factor may have ready
access to fretting damage sites and strongly influence fretting. Also, many
factors are shown to influence fretting through their effects on material prop-
erties.
The philosophy represented in Fig. 1 is really a specialized case of the
systems approach to wear, proposed and developed by Czichos and Molgaard
[1,2]? Certainly, to simulate adequately any practical fretting wear problem,
contact conditions, fretting load levels and amplitudes, materials, and en-
vironmental conditions must be considered. The fretting process is too com-
BILL O N FACTORS THAT INFLUENCE FRETTING WEAR 167
CONTACT CONDITIONS ENVIRONflEMTAL CONDITIONS

0 LOAD 0 TEMPERATURE
0 AMPLITUDE 0 HUMIDITY
0 FREQUENCY 0 CHEMICAL POTENTIAL
0 DURATION 0 LUBRICANT
0 GEOMETRY
F R E T T I N G RATES
AND
;• E C H A M I S "1 S
W E R I A L PROPERTIES AMD
BEHAVIOR
0 HARDNESS
0 STRENGTH
0 FATIGUE
0 OXIDATION R
CORROSION
0 CRACK PROPAGATION
0 DUCTILITY
0 ADHESIVE PROPERTIES
FIG. 1—Interaction of environmental conditions, contact conditions, and material properties

under fretting conditions.
plex to enable one to extrapolate with confidence from one set of conditions
to another very different set of conditions. Let us consider in detail how some
parameters, including load, frequency, amplitude of fretting motion,
number of fretting cycles, relative humidity, and temperature are observed to
affect frettmg wear under controlled experimental conditions.
Contact Load
The role of contact load on fretting wear appears to be ambiguous. Early
studies by Tomlinson, Thorpe, and Gough [3] indicated no significant
change in the wear rate as the contact stress was increased, and other in-
vestigators actually reported a decrease in fretting wear with increasing load
[4]. More recent studies by Ko [5] and Toth [6] show consistent trends of in-
creased fretting wear with increased contact load for a variety of materials.
The difficulty in reporting the effects due solely to contact load is that, for
most experimental apparatus, on increasing contact load results in more of
the intended interface motion being taken up by rig deflections, with less ac-
tual sliding occurring at the interface. Bearing this consideration in mind, it
is safe to conclude that fretting wear will increase with the contact load in a
nearly proportional manner, provided the interface motion is really held con-
stant.
Amplitude of Motion
The effect of the amplitude of motion on fretting wear is illustrated in Fig.
2, from Ref 7. It is universally agreed that wear increases with increasing
amplitude. However, two questions arise concerning the details of the effect
of amplitude on wear. First, is there a fretting amplitude below which fret-
ting damage does not occur? Second, what is the significance of the apparent
60
I 40
S 30
I
o
>
3 20
•^
^
^l^4*^-»'J I I I L I I
.2 .4 .6 ,8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6x10
in.
L_±_L^,^
5.1 15.3 25.4 35.6 45.6 56 66
10.2 20.4 30.5 40.6 50.8 61
MICROMETERS
FIG. 2—The fretting wear volume versus the fretting amplitude. The fretting exposure is l(r
cycles; normal load, 1.47 N; frequency, 163 Hz; dry air.
BILL ON FACTORS THAT INFLUENCE FRETTING WEAR 169
transition to higher wear rates shown in Fig. 2 for amplitudes greater than
about 25 /xm?
Tomlinson et al in their classic paper [3] conclude that slip amplitudes as
small as 0.025 /nm (10~^ in.) are sufficient to initiate fretting damage. Thus,
from a practical standpoint there appears to be no minimum surface slip
amplitude below which fretting will not occur.
Uncertainty does exist regarding the significance and actual location of the
transition region shown in Fig. 2. According to Waterhouse [8], amplitudes
greater than about 75 /xm result in wear more characteristic of unidirectional
sliding phenomena than fretting wear. Certainly the 75-/im amplitude is
beyond the transition region shown in Fig. 2. However, microscopy studies,
summarized in Fig. 3, indicate no large differences in features observed on
the fretted surfaces as amplitude was varied from 7.5 up to 62 fixn, although
differences in debris distribution were seen after the fretting experiments. At
low amplitude, most of the debris [ferric oxide (Fe203) as will be verified
later] was retained in the contact area, whereas at high amplitudes debris
was forced from the contact area and collected in a ring around the periphery
of the fretted surface. Presumably, debris retention would reduce the in-
cidence of metal-to-metal contact and distribute the load more uniformly
over the apparent contact area, thereby reducing fretting wear rates. If this is
indeed the reason for the apparent transition in wear rate, the amplitude at
which the transition occurs would likely depend on contact geometry details.
The experiments summarized in Fig. 2 were performed using a 4.8-mm
radius hemisphere on flat starting geometry. If a more conformal geometry
were employed, the amplitude required to disperse debris from the contact
area might well be higher. This might account for differences in the
amplitude at which the transition is observed to occur. The suggested in-
teraction between contact geometry and fretting amplitude in determining
the role of fretting debris is a good example of the types of interactions im-
plied in Fig. 1. Interestingly, a transition amplitude is observed in fretting
fatigue experiments, as discussed by Waterhouse [9]. The fatigue life decre-
ment due to fretting increases with the amplitude for slip amplitudes up to
about 20 jum but is relatively insensitive to amplitude increases beyond 20 ^m.
Frequency of Motion
Frequency of fretting motion was one of the parameters investigated by
Feng and Uhlig [10] in their study of fretting of mild steel. Feng and Uhlig
observed that the fretting wear rate (the wear volume per unit sliding
distance) decreased with increasing frequency up to about 30 Hz. Uhlig [//]
developed a general fretting model to account for this observation. He pro-
posed that with each fretting cycle the surface oxide film would grow to a
thickness proportional to the period of the cycle and then would be com-
pletely wiped away. Significantly, no frequency effect was observed when ex-
FIG. 3—The fretting damage features resulting from 10^ fretting cycles in dry air, under the
indicated fretting amplitude conditions. (Top, left) 7.5 to lOiim amplitudes: (top, right) 35 fim
amplitudes; (bottom) 62 jim amplitude.
periments were conducted in nitrogen (N2). Also included in Uhlig's model

are mechanical wear terms in addition to the corrosive wear term that ac-
counts for the observed frequency effects.
Number of Fretting Cycles

One universally accepted observation is that fretting wear increases with
the number of fretting cycles, often in the manner shown in Fig. 4 [12\. There
is an initial "incubation" period, during which the wear rate is quite low,
STAai STAGE II
INITIATION ONSET OF
OF SPALLING
SURFACE
DAMAGE
FREniNG
WEAR
VOLUME 30
(X10-5mm3)
20r-
105
FRETTING CYCLES
FIG. 4—The fretting wear volume versus the number of fretting cycles for AISI 9310 steel in
dry air under a 1.47 N normal load, 35 iim amplitude, 163 Hz.
lasting for 10^ to IC cycles. Beyond 10'' cycles a steady-state condition is

reached in which the wear rate increases markedly. The correlating
photomicrographs shown in Fig. 5 indicate adhesive wear and plastic defor-
mation of the surface during the incubation period. Considerable micropit-
ting, attributed to surface fatigue, and corrosive damage are evident after the
onset of the higher wear rate.
Interestingly, Hoeppner and Goss [13] first observed a similar incubation
period in fretting fatigue studies on titanium alloys. Below a threshold
number of fretting cycles, applied simultaneously with fatigue cycling, the
fatigue life was not reduced. If the fretting action was allowed to continue
beyond the threshold number of cycles, the fatigue life was rapidly reduced.
Relative Humidity
Relatively humidity has a significant and complex effect on the fretting
wear of materials. For example, the fretting of automobile bearings during
shipment has been observed to be more severe in the winter than during other
seasons. The reason for this is believed to be the typically lower levels of
humidity prevalent during the winter. Some fretting results for three bearing
steels—Types 52100, 440C, and M-50—in dry air and in moisture-saturated
air are summarized in Fig. 6. The 440C steel appears to be the least sensitive
of these steels to differences in humidity, although M-50 steel shows the
lowest fretting wear under both dry air and saturated-air conditions.
FIG. 5—Scanning electron photomicrographs of fretted surfaces on titanium after l(r and
Iff' cycles in air. (Top) l(r cycles. The features labeled "A" indicate ductile fracture, and "B"
indicates a plowing event; (botton, left) Iff* cycles, overview; (bottom, right) 10^ cycles, the
enclosed area of the overview, showing striations (labeled "S"J.
Generally, these results are consistent with those of Feng and Uhlig [10] on
mild steel wherein fretting wear was consistently lower in saturated air than
in dry air.
Close inspection of the results obtained from experiments on high-purity
iron over a range of relative humidity (RH) conditions shown in Fig. 7 [14]
reveals the complexity of the role played by relative humidity. Maximum
wear is at about 10 percent relative humidity, and there appears to be a slight
increase as 100 percent relative humidity is approached. Corresponding
I I DRY AIR
1001- ^ SATURATED
AIR
FREHING *0
WEAR
VOLUME
(mm^XlO^) ^
20
m
1
52100 4]0C
1 M-50
WL.
9310
Qw
9310CARBURIZED
FIG. 6—The fretting wear volume of various bearing and gear steels after 10 cycles under a
1.47 N normal load. 50 fim amplitude, and 80 Hz.
izoxw;-
0
lUU
0
<% 80
H
a
=1 6U o
o o
0= 0
o
^ « o
o
o
J \ L
«
J
60
L_L
100
RELATIVE HUMIDITY, »
FIG. 7—The fretting wear volume as a function of the relative humidity for 99.9 percent iron.
photomicrographs of wear surfaces on iron resulting from fretting in dry air,

and in 10, 35, and 100 percent relative humidity conditions are shown in Fig.
8. It is seen that the extent of debris dispersion differs markedly under these
varied conditions. Under dry air conditions virtually all of the debris is re-
tained in the contact area thus probably reducing the extent of the metal-to-
metal contact and therefore the wear, as indicated in Fig. 7. With an increase
• ;,.A*
:j-ai«^ §*
^^
c
Q
:3
s
3
•«
•a
.s
^ *i * Â. S
; £•
^!5
X 1
^:
:f
par, " * • « •• -V •'
in moisture content to 10 percent relative humidity, considerably more

dispersion of debris occurs, and areas of exposed metal may be seen. With
further increases of relative humidity to 35 and 100 percent, more of the
debris is seen to have been extruded from the contact area, leaving a larger
and larger portion of the load-bearing surface area unprotected metal.
Evidence of enhanced surface plastic flow may be seen after fretting in
saturated air.
The results of the transmission electron microscopy examination of debris
generated under dry air and 100 percent relative humidity conditions are
summarized in Fig. 9. Electron diffraction patterns indicated both samples
of debris to be a Fe203. However, there is a great difference in the size and
morphology of the debris particles for the two humidity conditions. Very fine
crystallites of less than 50 A result from dry air fretting, probably because of
continued mulling action. Saturated air fretting debris, on the other hand,
consisted of flat platelike crystallite about 200 A in diameter. Perhaps ad-
sorbed moisture permitted easy shear of these platelets and promoted a high
degree of debris mobility. The effect of relative humidity on mobility of the
fretting debris is, of course, only one aspect of the interaction of moisture
with the fretting process. Superimposed on the mobility effects are elec-
trochemical contributions, effects on the oxidation rates and stages for iron
[15], and possible surface physical effects [16].
Effects of Temperature
There are two ways in which the temperature might interact with the fret-
ting process: first, the rate of oxidation or corrosion increases with
temperature, and, second, the mechanical properties of the materials are
usually affected by temperature. For the case of most metals, the
temperature effects are best understood and explained in terms of surface ox-
idation kinetics. The experiments performed by Hurricks [17,18] on mild
steel in air showed a sudden decrease in the rate of fretting wear at about
200°C. Increasing the temperature from 200 to 500°C resulted in further,
continuously decreasing rates of fretting wear. Hurricks attributed these
results to a transition from slow, logarithmic oxidation kinetics to rapid
parabolic kinetics at about 200°C and accelerating rates of parabolic oxida-
tion as the temperature was increased beyond 200°C. The increasing rate of
oxidation led to the formation of a protective oxide film of sufficient
thickness to prevent direct metal-to-metal contact and to ensure, in effect,
that a "mild wear" regimen prevailed.
Elevated temperature fretting experiments performed on high-purity iron
[14] differ somewhat from those of Hurricks performed on mild steel. It did
turn out that for iron, as for mild steel, the fretting wear rates dropped to a
low value at about 200°C, as shown in Fig. 10. However, for iron the fretting
wear increased monotonically as the temperature was increased from 200 to
I
!>
I
•=,
^
^
li=1
50
«- o
o
30
8
20 o
0
0
f '" 0
X
1 1 1 1 1 1 1
(a) IRON.
l»
CD
«—
30 —
J L J I
0 100 200 300 400 SOO WO 7W
TEMPERATURE. "C
(bl NICKEL
FIG. 10—The fretting wear volume as a function of temperature for high-purity iron (top),
nickel (bottom), and titanium (next page), after 3 X 10^ fretting cycles in dry air.
600°C. Very likely, the discrepancy between the results reported by Hurricks
[18] and those reported in Ref 14 for iron are due to the effect of concen-
trated contact starting geometry employed in Ref 14 and a more marked
decrease in hardness with increasing temperature for iron compared to mild
steel. The fretting results obtained for iron, nickel, and titanium (up to
550°C) shown in Fig. 10 are in qualitative agreement with the oxidational
wear mechanism proposed by Quinn [19,20].
The fretting wear of titanium as a function of temperature requires some
further comment because of the extraordinary decrease in wear that occurs
as the temperature increases beyond 550°C, as shown in Fig. 10. Oxidation
studies performed on titanium [21,22] indicate that the transition from slow.
o
60 — o
J L
0 lOO 200 300 «0 500 600 700
TEMPERATURE, "C
(c) TITANIUM,
FIG. 10—Concluded.
logarithmic oxidation to rapid parabolic oxidation occurs between 500 and

600°C. Hence, the chances of maintaining a hard, protective oxide film
would be substantially increased at temperatures above 550°C, which is con-
sistent with the fretting results shown. Photomicrographs (Fig. 11) do indeed
show evidence that below 500°C, oxide film disruption occurs, with the
generation of quantities of loose debris and pitting of the metal surface. At
650°C, the fretted surface is relatively damage free, the primary evidence of
fretting damage being cracks in the oxidized surface oriented perpendicular
to the sliding db-ection.
The fretting results obtamed for high-temperature oxidation-resistant
superalloys at elevated temperatures are generally like those obtained for
titanium above 550°C. That is, fretting wear rates are very low because of the
in situ formation of protective, adherent oxide films that resist disruption
and spalling [23,24]. Hamdy and Waterhouse [25] observed that under fret-
ting fatigue conditions, the fatigue life decrement due to fretting surface
damage became almost negligible at temperatures of 540°C m experiments
performed in Inconel 718. In comparison, at room temperature and at 280°C
the fatigue strength decrement due to fretting was about 50 percent. They at-
tributed these results to the increased surface protection provided by oxide
film formation at elevated temperatures. Hence, there appears to be some
commonality in the role of elevated temperature in both fretting wear and
fretting fatigue experiments.
FIG. 11—SEM photomicrograph of a fretting wear scar on high-purity titanium after 3 X 10^
cycles at 650°C. (Top) overview: (bottom) central region of the top photograph.
Conclusions
1. The fretting wear rate (per cycle) is virtually independent of amplitude
up to a critical value of amplitude. Beyond the critical value, the wear rate in-
creases almost linearly with the amplitude. The authors suggest that the
critical amplitude value is partly a function of contact geometry.
2. Effects of the frequency of motion have been reported primarily for mild
steel and do not seem to be important above 30 Hz. Up to 30 Hz, the fretting
wear of mild steel decreases with increasing frequency. Further study of fre-
quency effects is in order for other materials and for an expanded range of
experimental conditions.
3. A threshold number of cycles appears to be required for the onset of
steady-state fretting wear rates for a number of materials. The onset of
steady-state fretting wear is marked by the appearance of microspall pits,
which indicates that a surface fatigue mechanism becomes operative.
4. Relative humidity effects are significant and very complex, involving in-
teraction between chemical reaction rates, the effect of moisture on the
physical characteristics of the debris, and possibly effects of moisture on the
surface mechanical properties of contacting material.
5. At elevated temperatures, there appears to be good correlation between
fretting wear resistance and the ability of a metal to form a thick, adherent
oxide film.
References
[/] Czichos, H. and Molgaard, J., Wear. Vol. 44, No. 2, 1977, p. 247.
[2] Molgaard, J. and Czichos, H., Wear of Materials, The American Society of Mechanical
Engineers, St. Louis, Mo., 1977, p. 30.
[3] Tomlinson, G. A., Thorpe, P. L., and Gough, H. C , Proceedings, Institute of Mechanical
Engineers, Vol. 141, 1939, p. 223.
[4] Lipson, C. and Colwell, L. W., Handbook of Mechanical Wear, University of Michigan
Press, Ann Arbor, 1961, p. 238.
[5] Ko, P. L., Experimental Studies of Tube Fretting in Steam Generators and Heat Ex-
changers, 78-PVP-22, The American Society of Mechanical Engineers, St. Louis, Mo., 1978.
[6] Toth, L., Acta Technica, Vol. 74, No. 1-2, 1973, p. 197
[7] Bill, R. C , A.S,L.E. Transactions, Vol. 21, No. 3, 1978, p. 239.
[10] I-Ming Feng and Uhlig, H. 'H., Journal of Applied Mechanics, Vol. 21, No. 4, 1954, p. 395.
[//] Uhlig, H. K., Journal of Applied Mechanics, Vol. 21, No. 4, 1954, p. 401.
[12] Bill, R. C , A.S.L.E. Transactions, Vol. 16, No. 4, 1974, p. 286.
[13] Hoeppner, D. W. and Goss, G. L., Wear. Vol. 27, 1974, p. 61.
[14] Bill, R. C , "Fretting Wear of Iron, Nickel, and Titanium Under Varied Environmental
Conditions," NASA TM-78972, National Aeronautics and Space Administration, 1978.
[15] Waterhouse, R. B., Fretting Corrosion. Pergamon Press, Oxford, 1972, p. 126
[16] Kragelskii, I. V., Friction and Wear. Butterworths, Washington, D.C., 1965, p. 3.
[771 Hurricks, P. L., Wear. Vol. 14, 1972, p. 207.
[18] Hurricks, P. L., Wear, Vol. 30, 1974, p. 189.
[19] Ouinn, T. F. J., Wear, Vol. 18, 1971, p. 413.
[20] Quinn, T. F. J., Sullivan, J. L., and Rowson, D. M., Wear of Materials, American Society
of Mechanical Engineers, Dearborn, Mich., 1979, p. 1.
[21] Hauffe, H., Oxidation of Metals, Plenum Press, New York, 1965, p. 214.
[22] Kofstad, P., High Temperature Oxidation of Metals, Wiley, New York, 1966, p. 169.
[23] Bill, R. C , "Fretting of Secondary-Seal-Ring Candidate Materials in Air at Temperatures
to 816°C," NASA TN D-7073, National Aeronautics and Space Administration, 1972.
[24] Bill, R. C , "Fretting of Nickel-Chromium-Aluminum Alloys at Temperatures to 816°C,"
NASA TN D-7570, National Aeronautics and Space Administration, 1974.
[25] Hamdy, M. M. and Waterhouse, R. B., Wear of Materials, American Society of
Mechanical Engineers, Dearborn, Mich., 1979, p. 351.

STP780 Eb.1415051 1 PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

STP780 Eb.1415051 1 PDF

Загружено:

Авторское право:

Доступные форматы

Copyright by ASTM Int'l (all rights reserved); Fri Jan 27 22:01:44 EST 2012

ASTM SPECIAL TECHNICAL PUBLICATION 780

ASTM Publication Code Number (PCN)

iSlb 1916 Race Street, Philadelphia, Pa. 19103

Printed in Ann Arbor. Mich.

Erosion: Prevention and Useful Applications, STP 664 (1979), 04-664000-29

Erosion, Wear, and Interfaces with Corrosion, STP 567 (1974),

Characterization and Determination of Erosion Resistance, STP 474 (1970),

Corrosion and Degradation of Implant Materials, STP 684 (1979),

Statistical Analysis of Fatigue Data, STP 744 (1981), 04-744000-30

ASTM Committee on Publications

Occurrence of Fretting in Practice and Its Simulation in the

A New Macliine for Studying Surface Damage Due to Wear and

An Evaluation of Fretting at Small Slip Amplitudes—p. KENNEDY,

Evaluation of Fretting Corrosion by Means of a New Device for the

Control of Oscillation Amplitude—K. G. BUDINSKI 49

Fretting of Electrical Contacts—MORTON ANTLER 68

Heat Exchanger Tube Fretting Wear: Correlation of Tube Motion and

Fretting-Resistant Ion-Plated Coatings—M. G. EL-SHERBINY AND

Effect of Ion Implantation on Fretting Fatigue in Ti-6Al-4y Alloy—

Fretting Fatigue of Ti-6A1-4V Alloy—c. LUTYNSKI, G. SIMANSKY, AND

Review of Factors That Influence Fretting Wear—R. C. BILL 165

Copyright by ASTM Int'l (all rights

The term fretting refers to small-amplitude oscillating motion between two

Copyright by ASTM Int'l (all rights reserved); Fri Jan 27

REFERENCE: Waterhouse, R. B., "Occurrence of Fretting in Practice and Its Simula-

ABSTRACT: Fretting arises in practice where a structure or multicomponent system in-

The first systematic investigation of fretting phenomena was made in 1927

'Reader in metallurgy. Department of Metallurgy and Materials Science, University of Not-

Copyright by ASTM Int'l (all rights reserved); Fri Jan 27

The Origin of Fretting Movement in Practice

SINGLE COMPONENT ASSEMBLY

Deliberately Induced Fretting

Simulation of Fratting in the Laboratoiy

achieved by a system of surveying with a profilometer. The other advantage

10cm DYNAMIC BALANCING

FIG. 2b—Plan of the high-temperature fretting rig in Fig. 2a.

normal load is applied by tilting it and locking it in position. The magnitude

out-of-balance weight method is best suited to moderate frequencies, that is,

ADJUSTING SCREW PROVING RING

• RECORDING THERMOCOUPLE CONTROL

tributed to a fonn of conjoint action similar to that which is experienced

A New Machine for Studying

KEY WORDS: fretting, fretting wear, delamination, fretting

The construction of turboprop engine gearboxes often necessitates the use

FIG. 1—Cross section of the test rig.

amplifier. The actuator is controlled by an MTS 436 controller. A d-c signal

The specimen design was determined by a number of factors. It was de-

Method of Wear Analysis

FIG. 3—The specimen dimensions, given in inches (1 in. = 25.4 mm).

FIG. 5—The effect of slip amplitude on the wear rate.

FIG. (>~The effect of normal load on the wear rate.

pects that an increase in load would result in a proportional increase in the

TABLE 1—Life extension of hard coated specimens.

Life Extension, millions of cycles

"Mean life of coating.

Fretting, particularly in its relationship to reduced fatigue strength and in-

' Chemists, Naval Air Development Center, Warminster, Pa. 18974.

It has been suggested that fretting is dependent on the simultaneous or

FIG. 1—Schematic drawing of a fretting test rig.

Wear Scar Analysis

FIG. 4—Ball-on-flat configuration.

Calculation of Slip Amplitude