Joseph Martin Cullen A thesis submitted for 'che degree of · PDF file5.17 Dependence of Peak...
Transcript of Joseph Martin Cullen A thesis submitted for 'che degree of · PDF file5.17 Dependence of Peak...
WEAR MECHANISM
OF. SALT BATH CARBONITRiTED STEEL
by
Joseph Martin Cullen
A thesis submitted for 'che degree of
DOCTOR OF PHILOSOPHY
of the University of London
and also for the
DIPLOMA OF MEMBERSHIP OF THE IMPERIAL COLTRGE
JULY 1976
Lubrication Laboratory, Department of Mechanical Engineering,
Imperial College of Science and Technology,
London, 5g7 2BX
"Myself when young did eagerly frequent Doctor and Saint, and heard great argument About it and about: but evermore Came out by the same door as in I went."'
Omar Khayyam
ACKNOWLEDGEMENTS
The author is greatly indebted to Professor Cameron for
his guidance Ana supervision throughout the course of this project;
to Mr. Reg. Dobson PmA Mr. A. Zynark for their encouragement and
invaluable technical assistance; to members of the laboratory
who have assisted in numerous ways; and to Mrs. Sheila Goodwin
for typing the manuscript.
Sincere thanks are also due to the British Leyland
Motor Company Ltd., for providing the funds which made this study
possible. .
i
ABSTRACT
Work has been carried out to discover the mechanism of the wear and
scuffing protection provided by salt bath carbonitriding. Two treatments
have been examined, the sulfinuz and tufftride baths. The majority of
work has been concentrated on the mild wear region.
Wear is found to be highly dependent on the contact temperature in
the range 25°C to 200°C. The dependence of wear on temperature is shown
to be related to the type and rate of oxide formation in the contact.
A transition from the formation of ferric oxide to the formation of
magnetite is found to occur at a constant contact temperature, and to be
associated with a large reduction in year rate.
The transition from mild to severe wear is shown to occur under
conditions of contact pressure, which would be expected from the surface
hardness of the materials.
No evidence is found for the treatments having any effect on the
performance of the steels, which cannot be explained by the increased
hardness which they produce.
Finally, a set of design criteria are put forward for the use of
the sulfinuz and tufftride carbonitriding treatments.
ii
CONTENTS
CHAPTER 1 : INTRODUCTION 1
1.1 On the Wear of Metals
1.1.1 Introduction to Wear 1
1.1.2 Metallic, or Severe, Wear 3
1.1.2.1 Abrasion 3
1.1.2.2 Adhesive tear 7
1.1.2.3 The Delamination Theory of rear 14
1.1.3 Corrosive, or Mild, Wear 19
1.1.4 The Effect of Lubricant Films on 'Wear 22
1.2 Salt Bath Carbonitriding 23
1.2.1 Sulfinuz Treatment 23
1.2.2 Tufftride Treatment 24
1.3 Wear Testing Machines 25
1.3.1 The Pin-on-Disc Machine 25
1.3.2 Disc Machines 27
1.3.3 :Variations on the Four Ball Machine 27
CHAPTER 2 : APPARATUS 30
2.1 The Cone. on Roller System 30
Original Design 30
Cone on Four Roller Rig 32
The Final Solution 32
'2.2 The Heating System
35
. 2.3 The Cooling System 35/
2.4 Measurement of Toraue and Temperature
37
2.5 • Overall Layout of the Test Rig '37
iii
CHATTER 3 : PROCEDURE 40-
3.1 Preparation of the Lubricant
40)
3.2 Preparation of the Specimens 40)
3.2.1 Chemico-Thermal Treatment 40
3.2.2 Cleaaing of Specimens 40
3.3 Cleaning of The Test Rig - 41
3.4 Friction Testing 41
3.5. Wear Testing 42
3.5.1 Running 42
. 3.5.2 The Measurement of Wear 42
3.5.3 Calculation of Wear 44
3.6 Repeatability 46
CHATTER 4 MATERIAL ANALYSIS 4a
4.1 Electron Probe Micro-Analysis (EPMA) 1 48
4.2 Debye-Scherer Analysis 48
4.3 Preparation of Specimens 50..
4.3.1 rear Debris 50
4.3.2 Sections of Specimens 50
4.4 Microhardness Testing 52
CHAPTER_ 5 : RESULTS 55
5.1 Materials 55
5.1.1 Microhardness Measurement • 55
5.1.2 EPLIA of Sectioned Surfaces 55
5.1.2.1 Normalised En8 55
5.1.2.2 Sulfinuzed En8 59
5.1.2.3 Tufftrided En8 .59)
iv
5.2 Friction 59 5.2.1 Slow Speed Friction Tests on Sulfinuzed En8 59.
5.2.2 Slow Speed Friction Tests on Tufftrided En8 59
5.3 Wear 67
5.3.1 The Wear of Tufftrided En8 67-
5.3.2 The Wear of Sulfinuzed En8 70;
5.3.3 The Wear of Normalised En8 83
5.4 The Mild to Severe Transition 86
5.4.1
5.4.2
5.5 Analysis
5.5.1
5.5.2
5.5.3 EPL3A of Wear Scars 88
CHAPTER 6 : DISCUSSION 92
Materials 92
6.1.1 Microhardness Measurements 92
6.1.2 Sulphur Concentration 92
Friction 93
Wear Tests 94
6.3.1 Tufftrided En8 94 6.3.2 Sulfinuzed En8 102
6.3.3 Normalised E11,8 103
6.3.4 The Variations of Wear Rate with Load 104 6.3.5 The Ferric Oxide to Ma petite Transition 106
6.3.6 The Change in Fear Rate at the Transition 10.7
6.1
6.2
6.3
The Mild to Severe Transition for Tufftrided &-18 86
The Mild to Severe Transition for Sulfinuzed En8 86
87
Identification of Debris 87
Sub Surface Appearance of Wear Scars 88
vi
6.4 The Severe Tear Transition 108
6.4.1 The Mechanism of the Transition 108
6.4.2 Sulfirnized En8 110
6.4.3 Tufftrided En8 110
6.5 Possible Lubricant Effects 111
CH CONCLUSIONS' 113
CHAPTER 8 : SUGGESTIONS FOR FUTURE VOTE 114
APPENDIX I : PROGI1-11: FOR THE CALCULATION OF 171^.P. 3.15
118 REFMENCES
LIST OF FIGURES
CHAPTER 1 :
1.1 Abrasion of a Smooth Soft Surface by a Hard Rough Surface 5
1.2 Deformation of an Elastic Sphere on a Hard Plane
1.3 Image of a Dislocation 15
1.4 Strain Gradient of a Surface after Sliding 15
1.5 Pin on Disc Machine 26 1.6 Ball on Triplane Machine 28 1.7 Original Cone on Roller Machine 28
CHAPTER 2 :
2.1 Statics of Cone on Roller Systems 31
2.2 Three Locked Roller System 33
2.3 Cooling System . 36
CHAPTER 3 :
tear Scar Volume
3.2 Dependence of Accuracy on Sample Size 47 •
CHAPTER 4 :
.4.1 Electron Probe
49,
4.2 Debye—Scherer Diffraction Pattern
51
4.3 Method for Sectioning Cones
54
CHATTER 5 :
5.1 Variation of. Hardness with Depth, Normalised En8 56
.5.2 Variation of Hardness with Depth, Sulfinuzed En8 57
5.3 Variation of Hardness with Depth, Tufftrided En8 58
5.4 Dependence of Friction on Temperature, Sulfinuzed En8 60
5.5 Dependence of Friction on Load, Sulfinuzed En8 .61
5-.6 Dependence of Friction on Time, Sulfinuzed En8 62
5.7 Dependence of Friction on Temperature, Tufftllided EnB 64
5.8 Dependence of Friction on Load, Tufftrided En8 65
5.9 Dependence of Friction on Time, Tufftrided En8 66
5.10 Dependence of pear on Time, Variable Load, Tufftrided En8 68
5.11 Dependence of tear on Time, Variable Temperature, Tufftrided En8 69
5.12 Dependence of rear on Temperature, 0.34 kgf, Tufftrided En8 70
Dependence of Fear on Temperature, 1.59 kgf, Tufftrided En8 71
5.14 Dependence of tear on Temperature, 2.95 kgf, Tufftrided En8 72
5.15 Dependence of pear on Temperature, 5.20 kgf, Tufftrided En8 , 73
5.16 Dependence of tear on Temperature, 11.10 kgf, Tufftrided En8 74
5.17 Dependence of Peak tear Rate Temperature on Load, Tufftride::. En8 75
5.18 Dependence of tear on Tine, Variable Load, Sulfinuzed En8 77
5.19 Dependence of tear on Time, Variable Temperature, Sulfinuzed En8 78
5.20 Dependence of Wear on Temperature, 0.34 kgf, Sulfinuzed En8 79
vii
45
viii
5.21 Dependence of Wear on Temperature, 1.59 kgf, Sulfinuzed n8 80
5.22 Dependence of rear on Temperature, 5.20 kgf, Sulfinuzed En8 81
5.23 Dependence of Tic= on Temperature, 11.10 kgf, Sulfinuzed En8 82
5.24 Dependence of Dear on Time, 0.34 kgf, Normalised En8 84
5.25 Dependence of Wear on Temperature, 0.34 kgf, Normalised En8 85
CHM:TM 6
6.1 Dependence of Theoretical Tear on Total Temperature 97
6.2 Dependence of Wear on Total. Temperature, 0.34 kgf, Tufftrided En8 98
6.3. Dependence of rear on Total Temperature, 1.59 kgf, Tufftrided En8 99
6.4 Dependence of rear on Total Temperature, 5.20 kgf, Tufftrided En8 100/
LIST OF PLATES
1 The Cone on Roller Rig Cup and Chuck 34
2 Test Rig with Heating Coil and Spring Balance 38
3 Test Rig with Cooling Tank and Pulley 39
4 Mild Uear Scar on Sulfinuzed En8 43
5 Mild tear Scar on Tufftrided En8 43
6 Mild Wear Scar on Normalised En8 43
7 Section of Sulfinuzed En8 Surface 53
8 Section of Tufftrided En8 Surface 53
9 Section of Normalised En8 Surface 53
10 Section of Mild rear Scar on Sulfinuzed En8 89
11 Section of :aid Wear Scar on Tufftrided En8 89
12 Section of Mild rear Scar on Normalised En8 89
13 Severe Uear Scar on Sulfinuzed En8 90
14 Severe rear Scar on Tufftrided En8 • 9Q,
15. Severe rear Scar on Normalised En8 90
16 Section of Severe Wear Scar on Sulfinuzed En8 91
ix
17 Section of Severe Uear Scar on Tufftridod. En8 91
18 Section of Severe Vrear Scar on Normalised En8 91
19 Particle Flaking Off a Surface 109.
20 Transferred Particle on a Tear Track. 109
CHAPTER 1 : INTRODUCTION
1.1 ON THE MAR OF METALS
1.1.1 Introduction to Wear
The scientific study of friction has a history of about five hundred years,
at least from the time of Leonardo da Vinci. Lubrication has been a subject of
scientific enquiry for only one hundred and twenty years, the first papers being
published in the 1850's. The first research papers on wear appeared in the
early 1950's (1,2,3). It is hardly surprising that, by comparison with its
sister disciplines, the corpus of knowledge concerning wear is rather slender.
It has become clear over the past twenty-five years that the process of
wear may occur by a number of quite separate mechanisms. Fluctuations in the
load on a surface may lead to surface fatigue and the flaking off of material.
The cutting of a surface by a hard abrasive will result in material removal.
Chemical attack either by a lubricant, by moisture or by atmospheric oxygen will
produce the phenomenon of corrosive wear. Metals may wear by an adhesive process
in which the adhesion (4) between contacting surfaces is sufficient to cause the
surfaces to break up and form loose debris. Lastly, over the past few years a
theory has been proposed, to explain the high wear rates which occur when clean
metal surfaces slide together, involving the pile up of dislocations below the
surface of a metal. This last has become known as the delamination theory.
(5, 6, 7, 8)
The earliest work on the wear of metals (1, 2, 3, 9, 10) showed that the
wear rate varied linearly with load except for two transitions. If the load
was such that the contact pressure between two sliding surfaces exceeded one-third
the surface hardness of the softer of the two materials, the wear rate increased
by two orders of magnitude. The electrical resistance of the contact fell
sharply and the surfaces, from being smooth and polished, became bright and
roughened. At higher loads the wear rate was found to fall once more to a low
level, and the surface became smooth and polished (11, 12). The regions of
2
high and low wear have been called respectively, severe and mild. In the
mild region debris is found to be fine with particle sizes less than 0.1/Um,
and composed mainly of oxides. In the severe region debris is upwards of 10/Am
in size and is mainly metallic.
The second transition has been shown to be a temperature affect assoc-
iated with martensitic phase transformations (11). At low loads an oxidised
layer covers the surface which prevents direct metallic contact. When the
contact pressure exceeds one-third the hardness of the softer material the
surface can no longer support the oxide layer, which breaks up allowing metal
to metal contact and greatly increased wear (13). The formation of a hard
martensitic layer at high temperature provides abase which can support an
oxide layer. In a situation where oxidation of the surfaces is impossible,
a. vacuum of over 10 tor, severe wear will occur due to the absence of a
protective film (7).
Recently strong evidence has been produced giving criteria for the mild
to severe transition in terms of shear stress and normal stress on a contact (14).
Severe wear will occur if
Cr-max > Y
( = H/3) •
-rmax ) Y/2
( = R/6)
Where crinax = (0.5 + (0.25 +iteyll A
'rmax = W (0.25 +/MI A
criax is the maximum normal stress, (max the maximum shear stress, Y the
yield stress, H the hardness, W the load, A the real area of contact and/ALthe
coefficient of friction.
Adhesion and delamination are the two mechanisms which have been proposed
for the severe wear which occurs between the two transitions. The adhesion
theory was proposed by Archard in 1953 (1), the delamination theory was proposed
by Suh twenty years later (S, 15, 16, 17). A detailed exposition of both
theories is given later,
3 1.1.2 Metallic, or Severe. Wear
1.1.2.1 Abrasion
Abrasion is the wear of a surface by the cutting action of relatively
hard particles. The particles may be protuberances on a surface, for example
file teeth, or they may be in the form of a loose grit. The farmer case is
referred to as two body abrasion, the latter as three body abrasion (5).
Rabinowitz (5) has given an analysis of the process which occurs when a
hard rough surface abrades a smooth soft surface.
Assume a surface made up of a large number of hard conical asperities
half angle A, fig.1.1. Assume that the asperities are evenly distributed so
that there is one asperity at each of the levels x = h, 2h, 3h etc.
If a single conical indenter penetrates a soft surface a distance x, the
area of contact is given by
A = 'f x2 tan2 g (1.1)
' and if deformation is plastic the load
W = Pm7rx2 tan2g (1.2)
If the surface described above is moved a distance x = Eh into a soft
surface then the asperity in the r th level will penetrate that surface a
distance
Xhz, = (N r)h
The load due to that asperity
r = Pn Irtan29 h2(N r)2
The total load between the surfaces
N =
0
= Pmr tan% h2,2 0
— 02
For the large values of N
W = Pm rtan2g h2.13-13
V = Pm 7/ tan2 gx3 3h
(1.3)
4 If the surfaces are displaced a distance L parallel to each other the
. volume of the groove cut by the asperity in the rth level will be
Vr = L 4 tan
The total volume of the grooves cut will be less than the summed effect
of all the asperities, as some asperities will run in grooves already cut.
Let Ka be the proportion of asperities cutting a new groove at any one time
then the total volume
V = Ka L tang x 0
and as before
V = Ka L tan 9 N3 3
V = Kt, L tan x 3h (1.4)
Equating (1.3) and (1.4) gives
V
Fm'ir tan 9 = Ka L
Ka V L
Pm.rr tan (1..5)
Define K = Ka
r tan 9
Then V = K AL
Pm
(1.6)
Kruschov (18) published a survey of the available knowledge on abrasion,
at the present time. The position can be summarised as follows:
1. Under an established wear regime the volumetric wear (V) is proportional
to load (W) and distance of sliding (L).
V = exPocL
where C is a constant of proportionality.
2. If there is no heating, wear per unit sliding distance (V/L) does not
depend on sliding speed.
5
DIRECTION OF SLIDING
• Figure 1.1 Abrasion of a Smooth Soft Surface by a Hard. Rough Surface
Figure 1.2 Deformation of an Elastic Sphere on a Hard Plate
6 3. Wear rate rises in approximately direct proportion to grain size, until
a cut off point is reached, where grain size has little or no effect.
The wear rate depends on the relative hardnesses of the metal and the
abrasive so that if
Ha 041 lam where 0.7Z(41.4:1.0
there is no wear, and if
Ha > 04. Hm Where 1.3-<01,11Z1.7
the wear rate is not dependent on Ha. Ha and Hm are respectively the
hardness of the abrasive and the metal.
5. ' In addition to the dependence of wear rate on the ratio Ha : Em it
also depends on the nature, shape, grain size and sharpness of the abrasive
material.
6. Crushing of abrasive grains, impregnation of the abrasive by worn metal
may exert an effect on wear rate.
7. Pure metals in the annealed state give wear rates inversely proportional
to their hardness.
a. Heat treated structural steels show a linearrelationship between wear
resistance and hardness of the form
C = Co + G(H - Ho)
Co and Ho are respectively the wear resistance and hardness of the metal
in the annealed state. The wear resistance of a metal in abrasion is
defined as the ratio of its wear rate to that of an arbitrary standard under
identical conditions. The standard is normally chosen to give a much higher
wear rate than any of the materials for comparison.
9. For cold work hardened metals the wear rate is not dependent on the hard-
ness resulting from work hardening.
10. The wear resistance of metallic materials in the annealed state is prop-
ortional to the elastic modulus of the metal to the power 1.3.
11. The wear resistance of metals can be directly related to the metal-metal
bond energy B(M-M), given by
B(M-M) = AMs
CN
7 where QMs is the heat of sublimation of the metal and CN its bulk
coordination number. As the melting point of a metal is closely related
to B(U—M) it follows that the wear resistance is approximately directly
proportional to the melting point. This is an empirical observation,
like most in this list, exceptions are Berrylium and Manganeue (19).
1.1.2.2 Adhesive Wear
The coefficient of adheision was defined by McFarlane and Tabor (4) as
the force required to separate two surfaces to the force applied between them.
This coefficient was found to vary with the nature of the materials and the
state of the surfaces but not with load. Later work by Iwata et alia (20),
produced an expression for the coefficient of adhesion in terms of the yield
stress of the weaker metal, the temperature in the contact, and an energy term.
The validity of the expression was supported by their experiments. A dependence
of adhesion on temperature was shown of the form
of, e- T
where y is the coefficient of adhesion and -r the absolute temperature. This
effectively results in a dramatic increase in adhesion at half the melting point
of the metal. Both. McFarlane and Tabor and Iwata found that adhesion increased
with time in contact, an almost immediate level of adhesion increasing asympto-
tically to about 1.8 times its initial value after approximately 3000 seconds.
In a study of the effect on the structure of the contact zone Dayson and Lowe (21)
repeated some of McFarlane and Tabor's work with satisfactory agreement. In
addition Dayson and Lowe showed large amounts of plastic deformation occurring
in the separation of clean metal surfaces, but if the surfaces were allowed to
oxidise before being brought into contact much smoother surfaces resulted, in
addition to much less adhesion. Some interesting work was carried out in a
vacuum chamber, equipped for LEER and Auger spectroscopy, by D.M. Buckley (22,23).
The loads used were very low (0 - 500 dynes), but they established several imp-
ortant points, that the more reactive metals give higher values of adhesion and
that the cohesively weaker metal will adhere to the cohesively stronger. With
8
oxygen present on the surface, where the bonding energy between metal and
oxygen was weaker than the cohesiveness of the metal the adhesive forces
measured could be correlated with the binding energy of the metal to
oxygen layer.
Bethune and Waterhouse (24, 25) related the adhesion coefficient•to
fretting cycles, and found that for their apparatus it rose up until 4 x 104
cycles and then became constant. At 10. cycles no difference in adhesion
was found between annealed and work hardened brass and between annealed and
work hardened copper. The variation in adhesion with the number of cycles
would seem to be due to work hardening. Marked reductions in adhesion
were found for alloys of copper as opposed to the pure metal, but this is
not related to the concentration of the alloying element. Much lower
adhesion was found between metals of differing crystal structure. Adhesion
was found to be much higher in Nitrogen than in air.
The effect of shear strains on Adhesion was studied by Anderson (26)
using a 'twist compression' technique. This involves rotating one surface
in the plane of the contact, while the surfaces are loaded together.
Anderson used the technique in air, where very large scatters result.
Bradford and Sikorski (27) used the twist compression test in vacuum with
much higher reproducability. Anderson showed a mechanism of ploilering
under of oxides and exposure of nascent metal. Neither of the groups found
any correlation between angle of twist and adhesion, or between surface
roughness and adhesion. Both groups found that the coefficient of adhesion
was constant with respect to both load and nominal area of contact. Anderson
found no evidence of melting at 'hot-spots'. All experiments were carried
out at 20°C. Predmore et alia (13) showed that cold welding can occur between
a. metal-oxide-metal sandwich, if plastic deformation breaks up the oxide layer
and allows metal to metal contact. Both Anderson and Bradford and Sikorski
found that coefficient of adhesion were 2 - 3 times greater with twisting
than without.
9 Buckley and Pepper (28) showed that during sliding in va.clumi, silver,
aluminium and copper will be transferred from a pin onto a steel disc.
Early experimental work on wear (3, 10, 29) showed that volumetric wear
was governed by the equation
V = KWL Pm
where V is the volumetric wear, TI the load on the contact, L the distance of
sliding, K a constant and Pm the flow of pressure of the material.
ArehArd (1) produced a theoretical analysis to explain this result, based
on the assumption that wear was due to the adhesion between metal surfaces.
Consider a deformable sphere, fig. 1.2, radius R pressed against a perfectly
flat non. deformable plate. If the deformation is elastic the radius of the circ-
ular area of contact will be:
a = 1.1 WR 2E
and if the deformation is plastic:
a = W, Trm
(1.7)
(1.8)
Where W is the load between the bodies, E the elastic modulus of the sphere, and
Pm the flow pressure of the sphere.
If the non-deformable surface is moved a distance x, radially towards the
centre of the sphere, the area of contact is given by
A i = 27Rx
a = v4F11
and for plastic deformation from (1.8)
Ni 11 Pm
W = 277-RPmx
and for elastic deformation from (1.7)
WR A Niffx = 1.1 it
= 4.25 ER t x16 •
Put
Then
10, A = b x (1.9)
11 = c xP (1.10)
b = 21TR
and for plastic deformation p = 1 c 27rRPm
for elastic deformation p = 2/3 c = 4.25 ER
If instead of bringing a single sphere into contact with the perfectly flat
surface, referred to above, a nominally flat surface made up of spherical
asperities is postulated. The surface is assumed to contain a large number of
asperities of equal radius, R, which are evenly distributed in depth, so that
there is an asperity at each of the depths given by x = 0, h, 2h, Nh, where h<nl.
The density of asperities per unit depth is then 1/h. For convenience put M = 1/h.
If the surfaces approach each other a distance x = Nh, then the contact area
of the asperity in the rth level SAr will be given by
SAr = b xr
where xr is the deformation of the asperity in the rth level where
xr = (N r)h
The total area of contact between the two surfaces will be given. by
A2 =0 A
r
N = bh (N —
0
which for large values of N gives
A2 = bh (N2 —4-N2)
ibhN2
substituting . x = Nh and M = 1/11
A2 = jMbx2
Similarly, the load on the asperity in the rth level will be
= cxrP
and the total load between the surfaces .N
12 = • dig:: (N r)P 0
13.
which for large values of N gives
W = chP NP+1 (p + 1)
substituting as above
Mc (p+1) 1/2- -(147-13- x •e (1.12)
Pat = B
Then from (1.11) and (1.12)
Mc = C (1.7- )
A2
Bx 2
2 = Cx(INI-1)
(1.13)
(1.14)
In order to produce a model for mechanical wear it is necessary to make four
major assumptions:
a) The model surface just described is used, but asperities are assumed on both
sides of the contact.
b) The contact is assumed to consist of two asperities sliding so that at time
t = 0 they completely cover each other and at time t = 2a/U, where U is the sliding
speed, they are just completely separated. It is assumed that at t = 2a/U two
more asperities have just moved into contact.
c) Two simple assumptions are made about the possible shapes for the resultant
wear particles.
(i) The material is removed as a layer, so that the volume of a particle
v p
where] is a constant equal to the depth of material removed multiplied by 27T.
The implication of this is that the depth of material removed is a constant
independent of the external conditions of sliding.
(ii) The material is removed as lumps so that
6v = of a3
That is the depth of wear is proportional to the radius of the contact area
and the shape of the wear particles is independent of their size.
d) The probability of a wear particle being produced by any one contact is a
constant, K, for a given system.
(1.15)
(1.16)
12
The wear per unit aiding distance for the asperity in the rth level is,
therefore, given by
vz. 2a
for layer removal . Vr =
Vr
For lump removal Vr = sia2 2
_ Vr = aftx
Put _ Vr = erg •
where for layer removal e = and q = 2
.and for lump removal e = ocR and q = 1
The total volume of material removed in unit sliding distance will be
V = K Z Vr 0
Ke hq N(g+1) crri- 1)
Substituting as before V = KM e x(l+q) 1 + q
Pat
KM e 1+ q)
then E x(1+4)
Combining equations (1.14) and (1.17)
(1.17)
1+cl l+p V E (1.18)
The classic statement of this equation is for the plastically deformed case,
assuming'that wear particles are removed as hemispherical lumps.
For plastic deformation P = 1
c = 21TRPm
13
Hence:
For lump removal
For hemispherical lumps
q = I
e = Rot
64 = 2/37r
V = K 3Pm
(1.19)
This is commonly known as Archard's wear law. The load-wear relationships
for different situations are not all linear; they are set out below.
Relationship between Form of Deformation Particle Shape Load and Wear
Elastic layer V 04 17 0.9
lump V.o& 1.2
Plastic layer V oC W 0.75
lump V 0( W
Assuming plastic deformation and lump removal it is possible to produce
expressions similar to (1.19) for surfaces made up of differently shaped asperities,
these differ from (1.19) by a constant factor. Thus, the expression for a surface
where the spherical asperity model is replaced by cones is
V = 21K17 3Pm
(1.20)
and where the model is based on ridges of triangular section, with debris of semi-
circular section,
V = .2Pm
(1.21)
It is clear that, while for the case of plastic deformation and lump asperities,
the size and distribution of asperities cancels from the equation, this is not so
for the other possibilities. It has been shown (Halliday) that the majority of
asperity contacts between two surfaces are likely to be elastic, unless the stress
over the entire contact is sufficient to cause plastic deformation. Recent work
with the ferrograph and scanning electron microscope (30, 31) has shown that
metallic wear particles tend to be flat and platelike. This would seem to challenge
the basis of the theory. However, it has been shown that in practice equation (1.19)
14 gives a reasonably good account of the wear process (3).
The value of K has been shown to vary between 161 and 10 -/ for different
materials (5). K is substantially lower for similar metals in contact than for
dissimilar metals, as would be expected. Over recent years it has become widely
accepted that the production of a wear particle at a contact is not a purely
random event, but the cumulative effect of a large number of encounters (9).
1.1.2.3 The Delamination Theory of Wear
The Burgers vector defines the magnitude and direction of slip of a dis-
location (32). The shear stress produced by an edge dislocation, with its
Burgers vector in the xy plane is, in cartesian co-ordinates
TxY =x = Gb x(x2 2 (x Y
If this stress field interacts with a free surface at a displacement h, the
resulting stress field is equivalent to that produced by a combination of the
original dislocation and a dislocation of opposite sign a distance h beyond the
surface, see fig. 1.3. For the case where the Burgers vector lies normal to the
. surface the stress on the dislocation due to its 'image' is
2-7W = Gb (1.23) 47f(1 - V)h
If the stress field interacts with a hard boundary the image is a dislocation
of like sign and so
't-xY = 7Yx = - Gb (1.24) 411(1 -V)h
where G is the bulk modulus, b the Burgers vector, Poisson's ratio, and 2 the
shear stress (33). It follows that dislocations will be drawn towards a free
surface and repelled from a hard boundary.
Suh (8) suggested that, if the stress on a dislocation were related to the
minimum stress required to move a dislocation, then there should be a low disloc-
ation layer thickness.
h = Gb (1.25) or(i -))Pad-1?
where °I, is the 'friction stress' the minimum stress necessary to move a disloc-
ation.
2 (1.22)
Image of Dislocation
Original 'real' Dislocation
Free Sutface
1\ Direction of Slip
Figure 1.3 Image of a Dislocation
h
x
Plastic Deformation
15.
Direction of Sliding Surface
Figure 1.4 Strain Gradient of .a Surface after Sliding
16 When a hard slider moves over a surface initially dislocations will be
repelled from the surface to a depth h. Once the slider has passed it leaves
a,clean surface free of oxide films and any dislocations less than a distance
h from the surface will be drawn out.
It has been demonstrated by the use of 'Low Energy Electron. Diffraction'
(LEED) and by Auger Spectroscopy, that in sliding between two clean metal
surfaces the cohesively weaker metal will adhere to the cohesively stronger
metal (22, 23). There is, however, little evidence to show whether adhesion
is an important wear process in practical machinery. According to the Archard
theory, wear debris from an adhesive process should be in the form of work
hardened lumps of metal. In practice, work with the Ferrograph and SEM (30)
has shown that metallic wear debris tends to be in the form of flat plates,
which have flaked away from the surface.
The 'Delamination Theory' was produced by Suh (8, 15, 16, 17) in order
tó cope with the new evidence on the nature of metallic wear, which was becoming
available. The theory is set out in the following propositions.
a) Material at and very near to a surface has a low dislocation density during
wear, due to the elimination of dislocations by the image force acting on those
dislocations parallel to the surface.
b) Mith continued sliding there will be pile ups of dislocations a finite
distance from the surface. In time this will lead to the formation of voids.
•• The formation of voids will be enhanced if the metal contains a hard second phase
for dislocations to pile against. Men there are large, bard secondary phase
particles in the metal, voids are primarily formed by plastic flow of the matrix
round the secondary particles.
c) Either by continued growth, or by shearing of the metal surface, the voids
will coalesce. ;Then a critical size is reached the surface above the void will
flake off.
d) The final observed shape of the wear particle will depend upon its size and
internal strains.
17
In order to produce an equation for rear rate in terms of applied
conditions, it is necessary to introduce three assumptions:
(i) Metals wear layer by layer, each layer consisting of N sheets.
(ii) The number of wear sheets per layer is proportional to the
number of encounters in the contact.
(iii) The rate of void and crack nucleation can be taken together with
the critical sliding distance So necessary for the removal of one layer.
The volume worn in each layer will be
CV = NxAxh (1.26)
where A is the area of a delaminated sheet and h its thickness.
Therefore, in sliding a distance S the wear will be
V =/ S)xAxh (1.27) So
It is assumed that the thickness of the delaminated layer, h, is equal to
the thickness of the low dislocation. Zone h is therefore given by the
expression
Gb (1.25) 41T(1
where G is the shear modulus, b the Burgers vector, VPoisson,s ratio and (Tithe
friction stress.
It is also assumed that, for dry contact, the real area of contact (a) is
proportional to the load (U).
A = CA
ThisiS only true if deformation is completely plastic. C is a constant
dependent on surface topography.
Combining equations (1.25) and (1.27) gives
V = (
r S ) x. A x Gb So 41((1 -)))07
Substituting for A •
V = ( S )CW x Gb So Or (1 — Y)cris-
(1.28)
(1.29)
(1.30)
18
Put K = 1xCx . Gb So 4 7(1 - Cr/
then V = KxSxW (1.31)
A major difficulty with most wear theories is their inability to predict the
amount of wear on the basis of the energy supplied in sliding that is the frictional
force and the total path length. Suh (17) has produced a treatment, based on the
delamination theory, which is able to show a good agreement between the properties
of the system, the external forces acting and the resultant wear rate.
The model of the surface shown in fig. 1.4 is used, where the equivalent
plastic strain 611 is assumed to follow the distribution
;P = e0 - cix
0 -4- x Lxc
eP = X X C
xe is the depth at which a and dgP/dx have the same value for both functions et),
es, c( and pare constants, and
1 2e.1)1+p
xe = ot
It is assumed that each asperity interaction produces a loading cycle,
which contributes to the total plastic strain so that N
a A i=1
for N cycles, where ZYjiP is the nett plastic strain generated by the ith cycle.
There is no analytical solution available for the relationship between ,nt eiP
and ei the total equivalent strain per cycle. It is assumed that
ei
•
a constant
A 6iP It is also assumed that the equivalent stress cr" is a constant and that
- = (the stress at x = 0) 1'77;
19
Use of an energy analysis for the work done in sliding leads to the two •
equations:
K = ( ea —
2 if h< xa
1-8
in o-B
(oo xc — (Axe elli° — txc\ =
2h r=15 ( h
if h > xa
where h is the thickness of the delaminated layer, "-the coefficient of friction,
and K is the wear factor
K = V W.S0
This treatment has been shown, empirically, to give results of the correct
order of magnitude.
An interesting implication of this theory is that where a soft metal is
plated on a hard substratE, in order to improve the friction and wear performance,
it is important that the plated layer should be less than the critical thickness,
h, at which dislocations pile up to form voids. Experiments with cadmium on AST
1018 steel (16) have shown this to be the case.
1.1.3 Corrosive, or Mild, Wear
Archard (2) shored that during mild wear a high resistance skin develops on
a metal surface. Kadhim and Earles (34) showed that for copper sliding on steel
a copper oxide film built up on the wear track on the steel. Powell and Earles (35)
showed that the establishment of mild wear was dependent on the establishment of a
sufficiently high flash temperature. They also showed (36) that frictional, heating
can cause rapid oxidation of a surface resulting in a wear rate two or three orders
of magnitude lower than the severe wear rates, which they found if the surface was
not oxidising. Earles and Hayler (37) showed that above 570°C the debris produced
in sliding between low alloy steels changed from magnetite (Fe304) to ferrous
oxide (Fe0). The latter having a much lower wear resistance than the former.
20
Tenwick and Earles (38). developed a theory which sought to relate wear rates to
the rate of surface oxidation. The results produced by this theory were of the
correct order of magnitude, it will be referred to later.
Kasak and Neumeyer (39) found ferric oxide present in the wear debris
• produced by hard steels. Bjerk (40) showed that oxygen could act as an extreme
pressure agent forming a scuff resistant layer on a steel surface. Kawamoto and
Okabayashi (40) shored that for- cast iron sliding at low speeds debris was W..- Fe203
as the speed was increased severe wear was set up at high speed mild wear was
re-established the debris being a combination of ferrous oxide and Magnetite.
Clark et alia (42) shoved that for high hardness steels wear increases to a
maximum with.increasing temperature and then falls sharply. Below the ma,rimum
debris is thought to be ferric oxide and above the maximum magnetite. Quinn (43)
showed that the debris produced in the mild wear of steel consisted of OL Fe203
and Fe304 with traces of otiron and ferrous oxide.
Tao et alia (44, 45, 46,•47) showed how an excessively oxidising atmosphere
can result in very high wear rates, which are destructive of the wearing parts.
Tenwick and Earles (38) produced a theory of oxidative wear based on a
linear oxidation. Standard tests on the oxidation of Metals (49, 50, 51) hold
that a parabolic oxidation rate 'occurs during the oxidation of iron in dry air.
Quinn (52) produced an analysis for the oxidative wear of steel based on a
parabolic oxidation process. Starting from the standard equation
V = KA (1.32
where V is the volume worn per unit distance of sliding, A the real area of contact,
and K is the reciprocal of the number of asperity encounters required to form an
oxide film of critical thickness h, where h is the thickness at which the oxide
layer will flake off the surface.
It is assumed that the oxide film is built up steadily by each additional
encounter. If the average duration of an encounter is ZS.tthen the average time
21
to produce a wear particle will be
t = at K
(1.33)
If the average length of a wearing contact is d and the sliding speed V
= (1.34) V
t = d VK
The mass of oxygen per unit area of the surface will be given by
m = f p h
(1.35)
(1.36)
where f is the mass fraction of oxygen in the oxide, and p is the density of the
oxide.
A parabolic oxidation rate law is assumed on the basis of the literature (49,
50, 51)
M = kp t
(1.37)
where kp is the parabolic rate constant. Putting (1.35) and (1.36) into (1.37)
and rearranging gives:
kp d
f2 p2 hZV
(1.33)
The parabolic rate constant kp is given by
kP = Ap e-QP/11T (1.39)
Ap is the Arrhenius constant and Qp is the activation energy for the reaction.
R is the gas constant and T the absolute temperature. This gives
d Ap e
f2 p2 h2 V
(1.40)
Putting in (1.32)
V = A d Ap e-4 2 2 -f p h2 V
(1.41)
In addition to the possibility of corrosive wear by the atmosphere it has been
shown that lubricants can attack a surface and so cause significant wear.
Goldman (53) showed that under 'non-scuffing' conditions in air the addition of the
22
extreme pressure agent Tricresylphosphate (T.C.P.) increased wear. Goldblatt
and Appeldoorn (54) shored that when T.C.P. reacts to form an iron phosphate this
is rapidly worn away, giving increased corrosive year. Beerbower and Goldblatt (55)
showed that aromatic compounds could react with a surface to form metal organic
compounds of low shear strength, which could then be removed in sliding.
Sakurai (56) showed that the wear rate of copper under non-scuffing conditions
was increased by the addition of sulphur to the lubricant. For low concentrations
the wear rate depended linearly on the sulphur concentration, but for high concen-
trations (greater than 0.5% by weight) the sulphur concentration is no longer the
rate determining process and wear is directly proportional to the contact area.
Buckley (57) has shown that oxygen will displace a sulphide film from an iron
surface, which would suggest that sulphur corrosion of iron in air is unlikely.
1.1.4 The Effect of Lubricant Films
All the situations described up to this point have been under conditions of
dry sliding. The introduction of a lubricant, and the presence of a hydrodynamic
film changes the situation considerably. If a full hydrodynamic film exists
there is no contact between the surfaces and there can be no wear. In the case
of EEL, however, the peak height of the asperities can be sufficient to penetrate
the film so that there is a solid contact between the sliding bodies. This is,
of course, the region of boundary lubrication. (58) 69)
If the greater part of the load in a contact is carried by a fluid film then
the degree of asperity interaction sill be governed by the properties of the film
and not the load carrying properties of the asperities, For an EHL film the film
thickness is related to the load by approximately the 1/20th power, and to velocity
by the power of about 0,7 . The radius of a Hertzian contact is related to the
1/3 power of the load and so the area of contact will be proportional to I/4 The
amount of asperity interaction will be governed by the ratio of asperity height to
film thickness, and by the apparent area of contact (59)0
It has been shown by Johnson (60) that the influence of surface roughness on
23
the build up of an EHL film is small. The removal of the peaks of interacting
asperities should not then result in a reduction of film thickness; once the
peaks have worn down wear will cease. If wear continues in a lubricated contact
it must then be due to a breakdown in the lubricant film and the problem is
essentially one of dry wear.
1.2 SALT BATH CARBOVIARIDING
Salt bath carbonitriding processes involve the heating of ferrous components
in a bath consisting of fused cyanides and/or cyanates, with the intention of
improving the wear and seizure resistance of the component. There are a nnmber
of different systems; the principal systems using sulphur or air to accelerate
the nitriding process. The two treatments dealt with here are the most common
examples.
1.2.1 Sulfinuz Treatment
The sulfinuz process is carried out in a bath of sulphides, cyanides and
cyanates at temperatures from 500°C to 600°C. A typical bath composition would
be NaCN 9.4%, NaCNO 11.7%, Na2SO4.16 the remainder of the bath is made up of
chlorides and carbonates (61). A pickling process can be added to remove the
black iron sulphide formed on the surface of treated parts (62).
The process results in the formation of a sulphurised surface layer
resistant to wear and seizure (63). This layer consists of two parts, a white
unetchable layer and a layer of high nitrogen content. The white layer is made up
of compounds of iron, carbon, nitrogen and sulphur and it has been suggested that
some free sulphur is also present (64). However, X-ray analysis has shown no
sulphur to be present in the supposedly sulphurised surface layer (65). The white
layer is about 0.01-0.015 mm deep, and below it is found a layer of high nitrogen
content, where nitrogen is present both as nitrides and in solid solution. The
nitrogen layer is about 0.5 Eon deep (62).
The treatment has been shown to confer a marked resistance to scuffing and
wear on Falex and Amsler test pieces. . Falex pieces when run dry became red hot
24 and extruded before seizure (61, 64). Other tests have shown a large reduction
in coefficient of friction, again accompanied by improved wear and scuffing
resistance (66). However, the process has been found to soften case hardened
steels and also to roughen their surfaces, so as to cause gross rsar on mating
bronze components, in addition the coefficient of friction of matang parts was
increased after treatment (67).
The mechanism of the protection provided by the sulfinuz layer is not clear.
It has been suggested that the protection is afforded either by free sulphur in
the coating, or by the break up of sulphur compounds. This is given some
support by the fact that a suspension of sulphur in oil has been found to reduce
running in time (68) Rrui that a Falex machine running with powdered sulphur as a
lubricant withstood very high loads without seizure. Against this must be set
the fact that it is not absolutely certain that the sulphur content of steel is
increased by the sulfinuz treatment (65) nor is it certain that any free sulphur
is present. Furthermore, a study of the lubricating effect of sintered iron
impregnated with sulphur showed lubrication to depend on the formation of sulphides,
such as EoS2' which have lubricating properties. Iron journals run in a Falex
machine between sintered iron jaws, impregnated with sulphur, seized very
quickly (66).
1.2.2 Tufftride Treatment
The tufftride process is carried out in a bath of fused cyanides and
cyanates at a temperature between 500°C and 600°C. Air is blown through the
bath to accelerate the reaction of the salts with the metal surface. The bath
should contain 20-4% KC/70 and 30-60;:, NaCN, the remainder of the bath is made up
of alkali metal carbonates. The content by weight of the metal potassium should
not be less than 10 nor greater than 30,11 preferably it should be 18%. The
remaircier of the metal should be sodium (70). The coating is thicker if between
0.5 and 10 of the total cyanate is present as n-cyanate and the remainder is
iso-cyanate (71).
25
A surface layer 6-12/AA-thick consisting of 80% epsilon iron Nitride and
20% of nitrogen containing iron carbide is formed. The brittle iron nitride
Fe2N is almost completely absent. Nitrogen penetrates between 0.4 mm and 1.0 mm
deep, depending on the type of steel and the bath conditions (72).
Tufftride shows a remarkable resistance to bending fatigue. This resistance
is said to be due to compressive stresses induced in the surface by the presence
of nitrogen (73). Tufftrided steels are also claimed to possess considerable
resistance to scuffing and wear (73).
Improvements in wear resistance of 500-1000% have been claimed, and
increases in fatigue life of 100% (74). Improved resistance to thermal softening
in die-casting tools has also been claimed.
There is no widely accepted theory available for the action of the tufftride
coating. Existing information is insufficient for any adequate theory to be
constructed. The similarity in structure and behaviour between the sulfinuz and
tufftride coatings would indicate a similarity in their mechanism. If it were to
be accepted that the protective mechanisms were similar, then the sulphur explan-
ation could not be held.
1.3 VWAR TESTING MACHINES
1.3.1 The Pin-on-Disc Machine
The most common type of wear testing machine is the pin-on-disc machine, fig.1.5
A pin is loaded onto a disc in the same way in which a gramaphone needle rides on a
record. A variation on the system loads the pin onto the circumference of the
disc, as in a phonograph. The disc is normally made of a material which has a
high wear resistance relative to the pin, wear on the pin is then measured when the
disc rotates. Facilities for heating the disc can be provided.
Wear is measured in a variety of ways. Shortening of the pin may be meas-
ured with a displacement transducer. A more common method is to measure the area
of the wear scar on a hemispherical or conically ended pin. Flat ended pins tend
26
PIN LOADING ARM
DISC
HEATER
Figure 1.5 Pin on Disc Machine
27.
to wear unevenly due to errors in alignment.
There was only one such machine available for this project. The minimum
disc diameter was about four inches, which meant that lubricated tests at high
temperature would be very difficult due to the rapid evaporation of lubricant
from the relatively large surface area. The machine available would have
dictated the use of wear scar area measurement. A. pin of sharp enough angle,
or high enough curvature to allow useful wear measurement would have resulted
in an abnormal coating formation. Only a very small amount of wear would have
been possible before penetration of the coating.
1.3.2 Disc Machines
Disc machines have sometimes been used to study wear. These are simply
machines where two, or more, discs rotate in contact, the speeds being adjusted
to give the desired amount of sliding. These machines have the beauty that the
geometry of the contact is not changed by wear so that load and contact pressure
can both be kept constant during a wear test. Unfortunately, in practice, disc
machines are complicated, expensive and use a large amount of lubricant. Such
a machine was not available, nor was the money to construct one.
1.3.3 Variations on the Four Ball Machine
The four ball machine, as originally described, consists of three balls
clamped together to prevent movement and a third rotated in contact with all three.
Uhen this machine is used to measure wear, the wear scars on the fixed balls are
measured. This measurement involves completely dismantling the system and it is
in practice impossible to reassemble it in exactly the original pattern, therefore,
only one wear measurement can be obtained from each test.
A variation on the standard four ball rig is the ball on triplane, fig.1.6.
The bottom three balls are replaced with the flat ends of three rollers. The
rollers are arranged on different wear tracks and are positioned so that the Hertz
contact pressure is the same on all three rollers. This system has the advantage
that there is no interference between the wear tracks produced by the three rollers.
28
Figure 1.6 Ball on Triplane Machine
Figure 1.7 Original Cone on Roller Machine
29
A further variation on the four ball machine is the cone on roller rig,
fig. 1.7. Three rollers are fixed firmly in a cup and a cone is rotated in
contact with their cylindrical sides. This system then has three point contacts
in a plane perpendicular to the axis of the roller. The main advantage of this
system is that it is possible to use materials -which are not available as balls.
As the materials used in this project, carbonitrided and normalised En8 are not
available as balls this was the system used.
30)
CHAPTER 2 : APPARATUS
2.1 THE CONE ON ROLLER SYSTEM
2.1.1 Original Design
The original machine used on this work was the cone on roller system
designed by M. Bailey (76). A section of the system is shown in figure 1.7.
The cone, of one inch base diameter and 45° slope, is held in the chuck and
rotated about the vertical axis. Three rollers, 0.25 inch diameter, are
mounted in the cup so as to be constrained from all movement. The cone is
then supported on the cylindrical sides of the three rollers, whose axes form
the sides of an equilateral triangle. The resulting contact between the cons
and each roller is an elliptical Hertzian point contact. In order for equilibrium
to be achieved the contacts must be in the same horizontal plane.
Define a plane containing the axis of the cone, and perpendicular to the line
between the axis and one contact. Taking movements in this plane only two
contacts will act, fig. 2.1. It is clear that the two contacts will exert a
couple on the cone which is not counterbalanced, unless both lie in the same
plane, perpendicular to the axis of the cane. It is equally clear that it is
possible to restore equilibrium by the addition of a third contact acting in the
plane. Superficially it would seem possible to balance a three roller system
by altering the angular distribution of the rollers in the plane perpendicular to
the axis of the cone. In practice it is not possible to obtain equilibrium in
two planes at right angles to each other,at the same time.
The rollers are retained in position by the small grooves at the base of
the cup. The cone is retained by three grub screws set at 120o round the rim of
the chuck.
This system was originally designed for low speed friction tests at a
rotational speed of 1 rpm. At this speed there was little or no vibration, and
the location of the rollers was perfectly adequate. This system was used for all
the low speed friction tests reported here. At 100 rpm, the speed used for wear
testing, considerable vibration was experienced, and it was found that rollers
31
3 2
Figure 2.1 Statics of Cone on Roller Systems
32
frequently jumped out of their grooves. It was, therefore, decided to redesign
• the system.
2.1.2 Cone on Four Roller Ric
As was stated in the previous section, for a cone to be sim?ly supported
in equilibrium at three points all the support points must lie in the same plane.
However, elementary statics shows that it is possible for a cone to be supported,
in simple equilibrium on four points each in a different plane, perpendicular to
the axis of the cone.
A system was constructed where a cone was supported on four rollers, so that
there was an equal load on each roller. The principle was that for a given
number of revolutions four points would be available for the wear/distance curve
for the system.
Unfortunately, if two points of contact are approximately diametrically
opposite, and if inaccuracy in construction results in either of those two contacts
being closer to the base of the cone than it was intended, or either of the other
two contacts is further from the base than intended, a position of unstable
equilibrium results. The cone then rocks on the two opposed rollers leading to
an uneven wear pattern.
In practice it proved impossible to machine the cup to the required accuracy
to avoid the effect described.
2.1.3 The Final Solution
Men the attempt to produce four points of the wear/distance curve was
found to be impractical it was decided to return to the original system of three
rollers with the contacts in one plane. However, a much more solid system of
• location for the rollers was produced, which owed a great deal to experience
gained with the four roller rig.
Fig. 2.2, shows a section of the three roller cup. The inside of the cup
Figure 2.2 Three Locked Roller System
CHUCK
OIL SHIELD HEATER COIL
CONE
ROLLER CUP
ROLLER SLOTS
MACHINING HOLES ---USED-FOR LOCATION
OF THERMOCOUPLE
33
34
Plate 1 The Cone on Roller Rig Cup and Chuck
35
is a 45° cone, which therefore parallels the test cone. The rollers are
.dropped into slots so that each cuts a cord across a circular section of the
cone, see plate 1. In order to prevent the rollers from rotating in their
slots they are each secured by a grub screw as seen in fig. 2.2. This system
proved completely free from the vibration problems referred to in connection
with the original system. In the course of about four hundred individual runs
there was not a single instance of a roller moving from its original position.
2.2 THE HEATING SYSTEM
If required, the oil shield could be surrounded by a 75017 heating element,
as shown in fig. 1.7. The power for the heater was controlled by a Pye Ether
1990 temperature controller, which proved capable of maintaining a constant
temperature to within one centigrade degree of set point.
203 THE COOLING SYSTEM
In order to limit the effect of bulk heating of the system due to friction,
a cooling system was designed. This system was only required for high loads.
In the great majority of tests the frictional heating was negligible.
The system was surrounded by a perspex tank, fig. 2.3. Uater flowed in
through A and out through B, providing a constant heat. The water supply was
direct from the tap, and discharge to the drain. During a test the bulk
temperature of the lubricant in the test rig rose to an equilibrium within the
first ten minutes, and maintained this equilibrium within plus or minus ten
centigrade degrees thereafter. The equilibrium temperature varied with the
total friction for a given test, and could lie between 30°C and 110°C.
This rather crude system left a lot to be desired, but allowed tests to be
made, which would otherwise have been impossible.
36,
CHUCK
OIL SHIELD
INLET
PERSPEX TANK
CONSTANT HEAD OUTLET
TORQUE TUBE ASSEMBLY 4
Figure 2.3 Cooling System
37 2.4 MRASURMENT OF TORQUE AND TEMTMATURE
Torque was measured as described by Bailey (76). The cup was constrained
from rotation by a torque tube, fitted with strain gauges. The output of the
gauges was transmitted to a chart recorder. The torque tube was regularly
calibrated and over the total period of testing drift was found to be less than
ten per cent.
Temperature was measured by a chrome-alumel thermocouple, which also
provided the signal for the temperature controller. The output of the thermo-
couple was also fed to a chart recorder. Bailey (76) had shown that the
positioning of the thermocouple was not critical. For these tests it was found
convenient to wedge the head of the thermocouple at the rear of one of the roller
slots.
2.5 OVERALL LAYOUT OF THE TEST RIG
An overall view of the rig is shown in plate 2, with the heating coil in
place, and plate 3, with the cooling tank in place. The chuck is driven by a
1.5 kW synchronous motor, through a toothed belt and a gearbox. The motor runs
at 1500 rpm, and the overall reduction is 15:1. A clutch is provided which will
slip if the torque exceeds a pre-set value.
The oil cup (A) is located on top of the torque tube assembly (B) by three
pins. The system (A,B) is mounted on the loading arm (C) which is hinged at (D),
Load is applied, at a mechanical avantage of 5:1, either by a spring balance (E)
on plate 2, if the load is to vary during a test, or by a pulley and deadweight (F)
for constant load tests, plate 3.
The weight of the loading arm, and all the equipment attached to it, is
equivalent to -1.59 kgf applied at the end of the arm.
Plate 2 Test Rig with Heating Coil and Spring Balance
39
Plate 3 Test Rig with Cooling Tank and Pulley
CHAPTER 3 : PROCEDURE
3.1 PREPARATION OF THE LUBRICANT
The lubricant used was highly refined hexadecane (cetane). This is a
saturated paraffin, that is it is hydrocarbon chain with no double bonds and
no elements except carbon and hydrogen. There is, therefore, no way in which
a molecule of cetane can chemically attach itself to a surface. Cetane has
virtually no properties as a boundary lubricant.
A large amount of work has been carried out with cetane in the laboratory.
There is, therefore, a body of knowledge concerning this lubricant readily
available.
In order to ensure that any impurities were removed from the cetane, the
cetane was agitated with activated alumina for two days, and stood over the
alumina for at least two weeks before use. This technique has previously been
Shown to remove all boundary lubricating agents (77, 78).
3.2 PREPARATION OF SPECIMENS
3.2.1 Chemico-Thermal Treatment
Tufftriding was carried out for two hours at 570°C in a bath consisting of
44% potassium cyanide, 53% potassium cyanate and 3% potassium carbonate. The
specimens were water quenched on removal from the bath.
The sulfinuzed specimens were treated, similarly for two hours at 570°C in
a bath containing 9% sodium cynide and 0.24% sulphur. The remainder of the bath
was composed of alkali metal chlorides and carbonates, the proportions of which
are not critical.
3.2.2 Cleaning of Specimens
All specimens were degreased with toluene in a soxhlet extractor for two
hours. The toluene was removed from the specimens by rinsing in acetone. Care
Was taken to avoid contact between the cleaned specimens and any possible source
41
of grease or other contanination. All handling of the specimens was with
.tweezers which had been cleaned in a similar manner.
Some sulfinuzed En8 specimens were cleaned by rinsing for five minutes
in Pyridene, and removing the pyridene with ether. A wear distance curve at
room temperature gave results identical to those for the cleaning pattern
described above. The pyridene technique was not continued due to the toxicity
of pyridene.
3.3 CLEANING OF THE TEST RIG
The oil shield cup and chuck of the test rig were cleaned with toluene
in an ultrasonic cleaner. Five minutes in the cleaner was found to be sufficient
to remove any trace of grease on the surface of the parts. The toluene was
removed from the parts by rinsing in acetone. Any items of equipment which came
into contact with the specimens were also cleaned in this way. These items
were: the thermocouple, tweezers, and the keys used to tighten the grub screws
on both the cup and chuck.
3.4 FRICTION TESTING
The system used was the first cone on roller system described in the last
chapter. The rig was run at a speed of 1 rpm, equal to 5.0 x 10 4 m/s. At
this speed any flash temperature effects are less than 1/10°C, based on Blok's
analysis (80). The amount of wear which occured during these tests was insig-
nificant, the contact pressure remained constant at 166 kgf/mma.
The system was run until the friction became constant and then either load
or temperature was steadily increased. The temperature was increased steadily
at a rate of 2°C per minute; the load was increased by different increments
depending on the test being carried out, and was left to run-in to a constant
friction before the load was increased again.
42
3.5 WAR TESTING
3.5.1 Running
Two types of wear tests were carried out, those to produce a rear/distance
curve, and those to produce a rear/temperature curve. The former tests involved
taking a number of points at different distances and constant temperature, the
latter points required that tests were run for a constant distance at different
temperatures.
The rig was heated to the temperature required for a given test, the full
load required was applied, and the rig was started immediately at full speed.
The test was run at 100 rpm (0.05 m/s). A friction trace was recorded during
each test in order to provide a basis for flash temperature calculations, these
are referred to in the discussion chapter.
nen the test had run for the required time the rig was stopped. For the
speed to drop from 100 rpm to a standstill took less than one second. Timing
- of runs was accurate to within ten seconds. The rig was then dismantled and 1
the specimens set aside in a rack for measurement of the amount of wear during
the test. As a check on repeatability two such tests were made for each point
on the wear graphs.
3.5.2 The Measurement of Wear
The initial, intention was to measure wear by the weight difference of the
rollers before and after a test. The rollers were numbered, cleaned weighed,
cleaned once more run in the rig, cleaned and weighed a second time. Mithin
the errors involved in weighing there was no significant difference. On each
roller there was, however, a significant wear scar. Observation of the maxima on
the wear/temperature curves shows that the highest values were of the order of
400 pg. The wear on the specimens which were weighed turned out to be about
50 pg. The scales were accurate to 0.1 mg and it was, therefore, impossible to
weigh the specimens sufficiently accurately. A more accurate set of scales were
not available.
43
Plate 4 Mild Wear Soar on Sulfinuzed En3 (x75)
Plate 5 Lild fear Scar on Tufftrided En8 (x75)
Plate 6 Mild Wear Scar on Normalised En8 (x75)
44
In view of the impracticality of direct weighing, it was decided to measure
the major and minor axes of the wear scars. The wear scars were elliptical in
plan as shown in the photographs, plates 4, 5, 6. The measurement was carried
out on a Hilger and Vatts projection microscope. The magnification used was one
hundred times and the accuracy was -0.5 x 10 3 inches.
3.5.3 Calculation of gear
The first method used to calculate the volume worn was to assume that the
form of the scar was that of a hemi-ellipsoid. Two of the axes could be measured
with the microscope, the third, the depth of the scar could easily be calculated.
Call the major axis A, the minor axis B and the depth of the scar C. Then for
a roller of radius R, as in fig. 3.1
C = B2/2R
(3.1)
Given C,the volume of a hemi-ellipsoid is
V = 11- A. B C (3.2) 3
This was solved separately for each roller, that is six times for each graph
point.
In order to lighten the workload a computer program was written to handle the
calculation. It was then decided to use a slightly more sophisticated calculation
for the volume of wear. It is assumed that the roller wears to a pattern conformed
to the cone, and that dimensional change on the cone is negligible for the purpose
of the calculation. Then looking at a section of the scar taken parallel to the
axis of the roller and perpendicular to the scar, fig. 3.1, the arc MN has the
same radius of curvature R' as the cone. If a set of right cartesian co-ordinates
are established in the plane of the ellipse such that the ellipse has the equation
2 x2 + 7-- = 1
A2
B2
Then the volume of the scar
=B V )-E-- A By
y=-B
where A is the cross-section area shown in fig. 3.1.
(3.3)
WEAR SCAR
Figure 3.1 Wear Scar Volume
45
46
The volumes of thirty-six scars were calculated taking ten, one hundred,
one thousand, and ten thousand values of y between ±B. Above one thousand values
the calculated value of V remains constant, and this was, therefore, the number
used for future calculations.
For presentation the mean and standard deviation were plotted for the six
values of V required for each data point. The six values come from two sets of
three rollers.
Comparison between this integral technique, and the hemi -ellipsoidal
calculation showed agreement within ten per cent of the mean. There was slightly
less scatter in the results from the integral calculation. The hemi-ellipsoidal
result is very sensitive to variations in the minor axis, which is exclusively
used to calculate C. The integral technique does not stress one piece of
measured date in this way.
The good agreement between'two very different methods of calculation was
taken as confirmation of both. The integral technique was used because of the
smaller scatter, and the more limited assumptions required in the analysis. A
listing of the program is given in Appendix I.
3.6 REPEATABILITY
The decision to use two runs per data point was based on a test where thirty
runs were carried out. Normalised Ena steel was used with an arm load of 0.34 kgf.
The mean was calculated for groups of different sample size. From these figures
the standard error of the mean for samples of different sizes was calculated. A
graph was plotted of the standard error of the mean for groups containing one to
fifteen wear scars; fig. 3.2.
Because the rollers must be run three at a time the only possible sample
sizes for a test program are multiples of three. On the basis of this curve it
was decided that six samples would be sufficient.
Figure 3.2 Dependence of Accuracy on Sample Size
x
Size of Sample
10 15
47
48
CHAPTER 4: MATERIAL 'ANALYSIS
4.1 ELECTRON PROBE MICRO-ANALYSIS (EPMA)
The bombardment of a surface with a stream of electrons results in the
raising of electrons, in the surface and sub-surface atoms, to hiL;her energy
levels. On returning to their original energy levels the excited electrons
emit radiation with a frequency characteristic of the element, of which they
form a part. Much of the emitted radiation is in the X-ray spectrum.
Diffraction of the radiation, by a crystal of known lattice spacing, will give
the frequency of the radiation, from the Bragg equation
n>s = 2d sin A (4.1)
yffiere X is the wavelength of the radiation, n an integer, d the lattice spacing
and 9 the diffraction angle.
The intensity of the radiation can be 'counted' for a given frequency by
means of an ionisation chamber containing a mixture of methane and Argon. By
counting the intensity of the radiation at different frequencies a quantitative
measurement of the elemental composition of the surface can be obtained.
The machine used in these investigations was a Japanese Electron Optics
Laboratory JXA-3A electron probe microanalyser. A diagram of the machine is
*shown in fig. 4.1. The electron beam is provided by a 25 kv tungsten filament
electron gun focussed by two magnetic lenses. The electrons strike normal to
the surface and the X-ray take off is at 20°. The specimen can be traversed at
constant speed below the beam allowing a scan of the concentration of an element,
at various parts, to be obtained.
4.2 DEB7E-SCIERER ANALYSIS
Debye-Scherer is a technique of X-ray diffraction suitable for use with
small quantities of powder. The powder is packed into a thin walled capillary
at the centre of a co-axial cylinder of photographic film. The random distrib-
ution of the particles comprising the powder is equivalent to a single crystal
specimen viewing z
• specimen - lumi nation
-n IW
Yi
0 0
counter objective output amplifier
Ipulse height analyser
f
scaler
rat emeter specimen
recorder
counter output amplifier
tuts e height analyser r 7
x-rays
electron gun
condenser ens
electrons beam current stabilisation
IIIMOM a,
49
Figure 4.1 The Electron Probe
50
rotated about all possible axes. The result of this is shown diagrammatically
in fig. 4.2.
For small particle sizes there is an increasing scatter in'the Bragg angle.
This effect has been quantified in the Scherer formula
)■ B cos 9B
(4.2)
where t is the particle diameter, )c the X—ray wavelength, B is the angular
broadening of a diffraction line and GB the angle of diffraction of the beam.
This formula can then be used to determine the size of wear particles.
4.3 PREPARATION OF SPECS ENS
4.3.1 Wear Debris
During wear tests debris was deposited in the cup. In order to collect
this in a form suitable for analysis it was necessary to separate the debris
from the lubricant. First the debris was left to settle, and as much lubricant
pippetted off as possible without disturbing the sediment. The debris was then
washed into a boiling tube with toluene, and again allowed to settle. There
would still be a significant quantity of cetane present, in solution in toluene.
The toluene, with the cetane in solution was pippetted out and the boiling tube
again filled with toluene. In this way the quantity of cetane present could be
reduced to negligible proportions. This removal of cetane was necessary, because
cetane does not easily evaporate at room temperature. Subsequent handling of the
material was eased considerably if it was held in suspension in a volatile liquid.
The debris was kept for analysis in a sealed specimen bottle of toluene.
If it was allowed to dry out, it tended to adhere very firmly tb the sides of the
bottle, making handling difficult.
4.3.2 Sections of Specimens
Sections through the materials were required for =A, for microhardness
measurements and for inspection of the effect of wear on the subsurface.
p∎ Iint ►►•here inehlvat heam
enters 120 = IMP) —
,
Li 6 (b)
Figure 4.2 Debye-Schere Diffraction Pattern (a) Diffraction Pattern (b) Appearance of Film
(a)
\20 iio
LL___
52
Sections were made of rollers in the plane of the axis. Thus the magnification
was 1 : 1. Sections were also made on the cylindrical side of the cones, .1-3 ),
plates 7, 8, 9. For the latter a small piece was cut from the circumference of
the cone and a flat was then ground on the curved surface, fig.4.3. If the width
of the flat is a, and the radius of the cone R, then the magnification, U, at a
distance x from the centre line of the flat is given by
• M = - a2 — a4 /R — x2/R
a— 2x.
(4.3)
All the sections were mounted in bakelite and polished on a 1/Am diamond lap.
Where etching was required a 2% Nital etching solution was used.
4.4 MICRON. RDNESS TESTING
Microhardness tests were carried out using a Leitz microhardness tester.
Four indentations were made for each value of hardness quoted, and the extreme
values were marked as error limits.
In order to check whether any phase changes occurred with prolonged heating,
which would affect the hardness of the materials, specimens were heated fon
twelve hours at 200°C, in a fluidised bed. There was some surface scratching
from the sand in the fluidised bed, and so repolishing was carried out. The
materials were retested and it was found that there was no significant change
in hardness.
a • I
r . ,. . •
Il•'• .. . • • • • VI ,,,.
L. • 49 L`' - Y i ."
• w • b.: L •
'-"J"411/141 -11r.,1••••"/....:-
. • • . • 11, •
dir gm. -11 • to
4 .f : Lew •••
53
Plate 7 Section of Sulfinuzed En8 Surface (x500)
Plate 8 Section of Tufftrided En8 Surface (x500)
Plate 9 Section of Normalised En8 Surface (x500)
54
.rH . --!-
So,mple
Figure 4.3 Irietho.d for Sectioning Cones
, . I I
"
Bakelite Mounting
Original CiTcumference of' Cone
55? CHAPTER 5 : RESULTS
5.1 MATERIALS
5.1.1 Microhardness Measurements
Fig. 5.1 shows the microhardness profile for normalised EnS steel. It
can be seen that there is a slight, but significant increase in hardness close
to the surface with a 7HN of about 350 at a depth of 0.1 thousandth of an inch
(2.54 11m). The hardness can be seen to drop quickly initially and then level off
at about 270 VHS. The error bounds are for one standard deviation of the
individual measurements from the mean. Figs. 5.2 and 5.3 respectively show the profiles for sulfinuzed and tufftrided EnS. Both curves have the profile of a
rectangular hyperbola, with very high hardness close to the surface fallirg to
approximately 270 VEN in the bulk material. Very close to the surface the
tufftrided steel is significantly harder than sulfinuzed 770 as opposed to 580 VBIL Both are very much harder than normalised EnS. The difference in
hardness between tufftrided and sulfinuzed En8 has almost completely vanished
after the first 7.5)4m after which any differences fall within experimental error.
5.1.2 Ella of Sectioned Surfaces
5.1.2.1 Normalised En8
Table 5.4 gives the range of composition of En8 steel, it is taken from B.S.970
Table 5.4 ELEMENT MAX.% MIN.%
Carbon 0.45 0.35 Silicon ' 0.35 0.05
Manganese 1.00 0.60
' Sulphur 0.06 -
Phosphorous 0.06 -
The electron probe gave a very low count for sulphur indicating about 0.02%
composition of the steel. It was suggested that the sulphur which was present was
in the form of manganous sulphide inclusions.
z 0
56
Figure 5.1 Variation of Hardness with Depth Normalised En8
rf) 0
O 0-
0_ co ■■••■••■••■
0_
57
Figure 5.2 VariatiOn of Hardness with Depth Sulfinuzed En8
114
z cL
0 0 0 0 'St*
O
58
cp O
Figure 5.3 Variation of Hardness, with Depth Tufftrided En8
x C
•
0 co
‘9
59 5.1.2.2 Sulfinuzed EnB'
A. very high count was found in the white layer, that is about the first 10,/Am
from the surface. The count was equivalent to a percentage composition of 0.3%
sulphur. The trace from the probe was very ragged indicating a very uneven
distribution. Once outside the white layer the drop in sulphur concentration was
immediate. In the interior of the steel a level identical to that for normalised
En8 was found.
5.1.2.3 Tufftrided En8
Surprisingly the tufftrided whitelayer also gave a high sulphur count of
between 0.07 — 0.14%. The drop in sulphur concentration is, however, much more
gradual than for sulfinuz and it is not particularly easy to define a precise
high sulphur zone. Whereas for sulfinuz the concentration drops in less than
0.125/ m, for tufftride the sulphur concentration drops evenly from a maximum
half way through the white layer. The minimum value found in the bulk material
is as for the normalised and sulfinuzed En8.
5.2 FRICTION
• 5.2.1 Slow Speed Friction Tests on Sulfinuzed En8
The results of a series of teststo determine the frictional behaviour of
sulfinuzed En8 are given in figures 5.4, 5.5 and 5.6. Figure 5.4 shows the
' variation in friction with temperature for various combinations of surfaces,
figure 5.5 shows the variation of friction with load and figure 5.6 the way in
which friction varies with time, during the running in of a contact.
In all but one case the curves of figure 5.4 show a remarkable drop in
friction, for a sulfinuzed En8 cone running on similar rollers the drop occurs
at about'150°C Ana for a sulfinuzed En8 cone running on En31 rollers the drop
occurs slightly higher at about 170°C. Her:ever, for the case where an En31
cone is run on sulfinuzed En8 rollers no drop is observed. An interesting
feature of the curve for a sulfinuzed cone on En31 rollers, also observed where
60
Figure 5.4 Dependence of Friction on Temperature, Sulfinuzed En8
Sulfinuzed Cone/roller )1( Sulfinuzed Cone/roller, white layer removed
Sulfinuzed Cone/En31 roller C) En31 Cone/Sulfinuzed roller
0.2
O•f
0 4 8 120 16b 20b 240
63.
Figure 5.5 Dependence of Friction on Load, Sulfinuzed En8
0.5
0.4
0.3
0.2
0.I
LB
100 200 300
Figure 5.6 Dependence of Friction on Time, Sulfinuzed En8
20 40
0.5
Min.
Sulfinuzed Cone/roller A Sulfinuzed Cone/roller, white layer removed Q Sulfinuzed Cone/ En31 roller
En31 Cone/Sulfinuzed roller
62
63
a sulfinuzed cone, abraded to remove the white layer, ran on sulfinuzed rollers,
was a short lived drop in friction about 30°C before the final drop. The
- friction remained low through a temperature rise of 6°C to 10°C and then rose
again to approximately its previous value.
Fig. 5.5 shots the variation of friction with load at 70°C and 160°C.
Within the limits of the experiment there appears to be little or no load
dependence at 70°C but a markedly lower friction at high than low loads is
observed when the temperature is raised to 160°C.
The variation of friction with time was plotted for the surface, whose
friction temperature characteristics were shown in fig. 5.4. The sulfinuz on
sulfinuz contacts showed a slightly lower coefficient of friction, than the
sulfinuz on En31 contacts, though the extent of this is hardly significant.
There was in addition a considerable variation in the time taken by the different
materials to run-in. The En31 cone on sulfinuzed rollers ran in very quickly by
comparison with the other systems, the reverse of this with a sulfinuzed cone on
En31 rollers taking eighty minutes to reach a constant friction. The two
sulfinuz on sulfinuz systems took about forty minutes placing them about half way
between the mixed systems.
5.2.2 Slow S eed Friction Tests on Tufftrided En8
The results of tests on tufftrided En8, similar to those carried out on
sulfinuzed En8 are shown in figures 5.7, 5.8 and 5.9. Figure 5.7 shows the
variation of friction with temperature, figure 5.8 of friction with load and
figure 5.9 shows the variation in friction with time during running-in.
The frictional behaviour with temperature is roughly similar to that found
with the sulfinuzed steel. The drops in friction are, relatively gradual, and
for the cases of an En31 cone on tufftrided En8 rollers, and a tufftrided cone on
tufftrided rollers the drop starts at around 90°C but is not complete until over
160°C. The curve for the case where the cone had been abraded to remove the
white layer showed a similar early dip to that observed for the sulfinuzed steel.
64
Figure 5.7 Dependence of Friction on Temperature, Tufftrided En8
$ Tufftrided Cone/roller 41, Tufftrided Cone/roller, white layer removed C) Tufftrided Cone/En31 roller
Tufftrided roller/ En31 Cone
0.2
_0.1
111111 0 40 80 120160200240
65
Figure 5.8 Dependence of Friction on Load, Tufftrided En8
0.5
-0.4
_0.3
_0.2
0.1
LB
O 100 200 300
Figure 5.9 Dependence of Friction on Time, Tufftrided En8
X Tufftrided Cone/roller Tufftrided Cone/roller, white layer removed
'40 Tufftrided Cone/En31 roller Tufftrided roller/En31 Cone
66
67
The variation of friction with load, fig. 5.8, seemed to contradict the
findings for sulfinuzed steel, in that at room temperature a small but steady
fall in coefficient of friction was observed, while at 150°C after a small
initial rise there was no significant change in coefficient of friction with
increasing load.
The cuzves of friction with time showed all the different combinations of
surfaces running in to similar values of the coefficient of friction. The time
taken to reach a steady value varied from 15 to 6o minutes, with the tufftride/
tufftride systems lying between the mixed systems. However, the mixed systems
were reversed in order from the tests on sulfinuzed steel, the tufftrided cone
on En31 rollers running in much faster than the reverse arrangement.
5.3 VEAR
5.3.1 The Tear of Tufftrided En8
The first tests carried out were of the classic wear displacement type;
the results of these tests at two different loads are shown in fig. 5.10. It
can be seen that they follow the classic pattern, of a high initial wear rate
leading to a period of constant wear, which is well established in the first
sixty minutes.
The graph, fig. 5.11, shows a comparison of the wear/displacement curves
at room temperature and at 170°C, with an effective arm load of 0.34 kgf. The
,very low wear rate at high temperature, with the very early establishment of
constant wear rate is the noticeable feature of this graph.
Figure 5.12 shows the variation of wear rate with temperature at the same
load as that used in fig. 5.11 (0.34 kgf arm load). The wear rate rises very
rapidly to a peak at 120°C and then falls to less than one-third that value by 130°C;
by 180°C the wear rate has fallen to one-twentieth its peak value and by 240°C to
about one-hundredth. Fig. 5.13 shows the same type of curve for an arm load of
1.59 kgf. The pattern of wear rate behaviour is identical, except for a lowering
68
Figure 5.10 Dependence of Wear on Time, Variable Load, Tufftrided En8
Cr) •
o C: .
. 69
Figure 5.11 Dependence of Wear on Time, Variable Temperature, Tufftrided En8
oC
T0.5
70
x io-4
2.0
--6-1.5
Figure 5.12 Dependence of Wear on Temperature, 0.34 kgf, Tufftrided En8
1
-1.0
71
Figure 5.13 Dependence of Wear on Temperature, 1.59 kgf, Tufftrided En8
-3.0
•• -2.0
-1.0
0 40 80 120 1-6/6 26e00
72
g x 0-4 Figure 5.14 Dependence of Wear on Temperature, 2.95 kgf, Tufftrided En8
1-0
al.11•11
0 4b 12b 16. 20.240
73
g x Figure. 5.15 Dependence of Wear on Temperature, 5.20 kgf, Tufftrided En8
-2.0
-1.5
-1.0
-0.5
oc
sO 4 (3 i216t20b---2-40-
74
oC
40
Figure 5.16 Dependence of Wear on Temperature, 11.10 kgf, Tufftrided En8
75
Figure 5.17 Dependence'of Peak Wear Rate Temperature on Load, Tufftrided En8
76
of the peak wear rate temperature to approximately 105°C. Fig.5.14shows the
pattern repeated for an arm load of 2.95 kgf, the pattern is repeated with the
peak wear rate coming at approximately 85°C. The fall off in wear rate is now
much more gradual, and the scatter of the results on the downward slope is very
much greater. In fact, the points of the two runs at 92°C would appear to
belong to different distributions. Increasing the arm load to 5.20 kgf, shown
in fig. 5.15, the large variations in wear rate has gone. A relatively small
dip comes between 110°C and 140°C, accompanied by a noticeable reduction in the
scatter of the data for the individual points. Fig.5.16 with a load of 11.10 kgf
gives very much the same pattern as fig. 5.15. A definite drop seems to occur
at about 130°C though the preceding rise is well within the limits of random
errors.
The peak wear rate temperatures (PTRT) at 0.34 kgf, 1.59 kgf and 2.95 kgf
were plotted against load. The result is shown on fig. 5.17, and demonstrates
an 71most linear relationship between loadAFWRT.
5.3.2 The Tear of Sulfinuzed En8
As was the case for the tufftrided steel the distance wear curves for
sulfinuz, shown in fig. 5.18, are of the classic pattern with a high initial wear
rate tapering off to a steady rate of wear after about 30 minutes running. It
can be seen from these curves that there is no simple and straightforward
relationship between load and wear rate. Fig. 5.19 shows the wear/displacement
,curve for the sulfinuzed steel at 160°C and an arm load of 0.34 kgf, for comparison
the curve at room temperature is also shown. The wear rate at 160°C has clearly
reached a steady state within the first ten or twenty minutes.
Wear/temperature curves for the mild wear of sulfinuzed En8 at different
loads are shown in figures 5.20 to 5.23. At 0.34 kgf arm load, fig. 5.20, the
wear rate follows a similar pattern to that of the tufftrided steel, however, the
peak wear rate temperature is not so clearly defined. The sharp drop in wear
rate seems to come between 130°C and 140°C, the difference in wear rate being about
0
•••••IIIIM1
77
Figure 5.18 Dependence of Wear on Time, Variable Load, Sulfinuzed En8
78;
Figure 5.19 Dependence 'of Wear on Time, Variable Temperature, Sulfinuzed En8
g x 103
em■■mt
0 40 80 120 1 6 0 200 2 4 0
79.
Figure 5.20 Dependence of Wear on Temperature, 0.34 kgf, Sulfinuzed En8
oc
80)
g x
2.0
Figure 5.21 Dependence of Wear on Temperature, 1.59 kgf, Sulfinuzed En8
1.5
-1 ff--1 I 0 40 80 120 160200240
81
Figure 5.22 Dependence of Wear on Temperature. 5.20 kgf, Sulfinuzed En8
2.0 •
0 c
0 415-* 12b 16b Z)45 24b
Figure 5.23 Dependence of Wear on Temperature, • 11.10 kgf, Sulfinuzed En8
2.0
1.5
1.0
oc
82
I L, 0 40 E3 12u 1640 200 24
83
one hundred times. The wear axis on fig. 5.21 is drawn at ten times the scale
of that on 5.20, becane of Very much lower wear rates. The pattern of the
wear behaviour is similar to that of tufftrided En8 at high loads. There is
no tendency for the wear rate to increase with temperature, and above 100°C a
gradual drop in wear rate, to about half the value at room temperature, is
observed. The wear rates overall lie between one-tenth and one-thirtieth of
the peak observed at 0.34 kgf..
At an arm load of 5.20 kgf severe wear is sometimes observed. The mild
wear curvet fig. 5.22, however, shows roughly the same pattern as before.
There is a slight, though inconclusive, rise in wear rate to a temperature of
160°C, followed by a steady decline, again to slightly under half the original
wear rate. At 11.10 kgf it is rare for mild wear to be sustained through an
entire test. Vicar rates are initially higher, the scale is, therefore, once
more that of fig. 5.17; the wear rate shows the same gradual fall to a moderate,
.but not very low value at around 200°C.
5.3.3 The ̀tear of Normalised En8
As with tufftrided and sulfinuzed En8 a rear displacement curve was
produced for normalised En8, at room temperature and an arm load of 0.34 kgf,
.fig. 5.24. The curve is not as regular in form as those of figs. 5.11 and 5.18,
but the irregularities are well within expected limits of error, and the classic
. wear/displacement pattern is still observable.
Because of the less settled behaviour of the wear displacement curve for
the normalised steel the wear/temperature curve was produced for 100 minutes
running time, fig. 5.25, rather than 60 minutes as for tufftride and sulfinuz.
The wear observed at all temperatures above 60°C was severe and was, as would be
expected, between one and two orders of magnitude higher than the observed mild
wear rates. The room temperature point marked at 25oC has been shown to give a
comparison between the different types of behaviour. It is not, of course, of
great quantitative significance to plot a mixture of mild and severe points on
such a curve.
84
Figure 5.24 Dependence of Wear on Time, 0.34 kgf, Normalised Eno
-1.5
-1.0
-0.5
0 40 80 120 160200240
85
Figure 5.25 Dependence of Wear on Temperature, 0.34 kgf, Normalised En8
g x 103
86
5.4 THE l·ITLD TO SEV'Er~ .TRANSITIOn
, ·5.4.1 The lIild to Severe Transition for Tufftrieed EnS
Ta.ble 5.2
LOADkgf
70~O
65.5
45.2
88.1
TDm
3 min.
3 min.
TErEPERATUHE
202°C
205°c
1100C
233°c
The failure loads for tufftric1ed EnS lubricated m. th cetane, and run in
a.t 100°C before loading are given in table 5.2. The first column gives the arm
·load, the se·cotid the· time of running at the f~ltlre load (in the last tvo cases
in the table, failure '\'Tas :immediate), and the final column gives the temperature
to which the bulk lubricant had risen at failure~ The tests were repeated twice
in the absence of. a lubricant. In both cases failure occurred immediately on
loading to 45.2 ker. A further pair of tests were carried out vdth the system
cooled in order to limit the temperature rise. In both these tests an arm load
of 95 kgf 'VIas exceeded at .a temperature of IlQ-1150C rn. thout a:rrJ" sign of surface
failure. This is the limit of the test l'ig.
5.4.2 The Eild to Severe TrallSition for Sulfinuzed EnS
LOAn kg! TD.!E minutes
11.31 15
9.05 30 9.05 15
11.31 15
The transition loads for sulfinuzed EnS are shown in table 5.3. The test
havi~ been run "in at lOOoC the tenperature remained steady during loading.
During weer tests severe wear was on one occasion observed at a load of 5.20 kgf,
which implies that a large amount of scatter is possible.
87
5.5 ANALYSIS
5.5.1 Identification of Debris
Information on samples of debris is given in table 5.4.
Table 5.4
Test Colour
Results of Tests
Size Magnetic X-R.D.
kgf
SULFINUZ 1 23.80 Severe Bright Yes
2 0.34 25°C Brown No ok-Fe203 L300 Je 3 0.34 200°C Black Yes Fe304 11
4 1.59 25°C Brown Slight 0.(-re203/Pe304 5 1.59 187°C Black Yes
TUITTRIDE 6 0.34 25°C Brown No c0C-re203
7 1.59 25°C Brown No
8 2.95 115°C Brown Slight 04-Pe203/Pe304
9 1.59 180°C Black Yes Fe304 tt
10 23.80 100°C Black Yes Fe304 .4300 .,1F
11 70.00 Severe Black Yes Fe304/04;Pe
With certain runs insufficient debris was available for an identification
by means of X-ray diffraction. However, on the basis of the pattern observed
with the other samples, and the information which is available, it is reasonable
to assume that they consist of the materials quoted in table 5.5.
Table 5.5 Identification
Test Material
1 -Fe
2 6; -Fe203 3 Fe304
4 0,4 --P e203/2e304 (mixture)
5 Fe304
Table 5.5 (continued) 88
Test Material
6 co.-Pe2 b_
7 ok—Fe203
8 04.-Fe203/Pe304 (mixture)
9 Fe304
10 Fe304
11 Fe /Pe304 (mixture)
5.5.2 Sub-Surface Appearance of Wear Scars
The appearance of the sectioned wear scars was as would be expected from
the literature. In all cases of mild wear there was little or no evidence of
plastic deformation and no apparent sub-surface hardening or softening. Mild
wear scars from tests with all three types of surfaces are shown both in plan
and section, plates 4, 5, 6, 10, 11, 12. Sections have been etched with 2%
Nital. The severely worn scars are similarly in accordance with expection,
plates 13, 14, 15, 16, 17, 18. There is considerable plastic deformation,
and several transferred particles can be observed in the surface, these have
hardness in the region 600-eoo TEN, a similar hardness is found in the highly strained region below the scar.
5.5.3 EPEA of Tear Scars
The analysis for sulphur gave counts of 145 for tufftrided En8 and 347 for
sulfinuzed. The 100% calibration based on a Cds standard was 700000. This
level of sulphur is very low.
89
Plate 10 Section of Mild Wear Scar on Sulfinuzed En8 (x150)
Plate 11 Section of laid Wear Scar on Tufftrided En8 (x150)
Plate 12 Section of Mild Wear Scar on Normalised En8 (x150)
90
Plate 13 Severe vlea r Sca r on Sulfinuzed :tn8 (x75)
Plate 14 Severe Wea r Scar on Tufftrided En8 (x75)
Plate 15 Severe "Hea r Sca r on Normalised En8 (x75)
cr.,■---31 r 'r
91
Plate 16 Section of Severe Wear Scar on Sulfinuzed En8 (x150)
Plate 17 section of Severe Wear Scar on Tufftrided En8 (x150)
.11
t lea' A 4ir •‘: I eleti,
ea lit lt "I"
ir• It" i *
al .#CVelitg.8.
tic iv_40 -4 e•
JPitk" AC' 1 •
4 #
.IN. •-• stle I 1,1 it 0 111 t tit 424-
gat. 41.- • .4* , 4 It-a, •Or' Aril, PAS sk - - 4
Plate 18 Section of Severe Wear Scar on Normalised En8 (x150)
92 DISCUSSION
6.1 MATERIALS
6.1.1 Microhardness Measurements
The very high hardness noted for both sulfinuz and tufftride close to the
surface, is, on its own, indicative of a high degree of protection against wear
and seizure. As has been pointed out in the introductory chapter the transition
from mild to severe wear is dependent on the hardness of the surface. It is,
therefore, not unreasonable to put dawn changes in the seizure load to the changes
in hardness.
The very even hardness profile rill indicate that there will be little
tendency to produce sharp strain gradients, which will minimise the likelihood
of a hardened surface layer flaking away. This is, of course, very different
from the more conventional case hardening methods.
6.1.2 Sulphur Concentration
The original theory for the protection offered by sulfinuzed steel was
based on the idea that a high sulphur content in the surface of the steel would
in some way minimise wear and increase seizure resistance. It was suggested
that the sulfinuz bath caused sulphur to diffuse into the steel surface.
Analyses for sulphur by J.C. Gregory (61, 64) show that for two different steels
the sulphur content of the steel surface has indeed been increased, but in prop-
ortion to the quantity of sulphur in the untreated steel. This is really rather
remarkable and leads to the question: is the sulphur going in or coming out?
On this point it is interesting that tufftrided En8 shows a significant increase
in sulphur content close to the surface. There is, nominally, no sulphur in a
tufftride bath, and so the only place this sulphur could come from is from the
steel itself. Given the similarity between the tufftride and sulfinuz baths, it
is probably worth speculating whether the sulphur in a sulfinuzed steel has, in
fact, been put in by the bath or has diffused to the surface from the bulk steel.
93
It is, of course, quite probable that a combination of both processes is
occurring.
The most important question is whether the sulphur is active as a lubricant,
and the answer must be; almost certainly not. The fact that of the two treat-
ments the tufftrided steel rith a half to a quarter the sulphur cuncerrtration has
a far higher scuffing resistance and significantly lower friction and wear makes
a theory of protection based on sulphur wholly untenable° Further, the EWA of
worn surfaces shows very little sulphur present, and indeed in almost all tests
the surface layers containing the high sulphur concentrations are worn away within
the first ten minutes. A further nail in the sulphur coffin is that these first
ten minutes exhibit the highest wear rate on the wear/time curve.
It would, of course, be foolish to reject any possibility of sulphur action.
However, there is no evidence that it occurs and the wear mechanism of these
treated steels can be understood without recourse to sulphur as an explanation.
6.2 FRICTION
The initial intention of the slow speed fricticAI tests was to obtain a
rough idea of possible areas of interest, in the behaviour of the different
surfaces. The very sharp friction drops which were observed with sulfinuzed En8,
and the more gradual drops observed with tufftrided En8 were a feature which
required some 'explanation. Excluding, for the present, the short-lived initial
drops, the major drop is similar to those obtained in boundary and E.P. studies
,(77, 78). However, the process by which the lubricant used in these tests was
purified would have removed any boundary or E.P. components from the system, and
therefore it is necessary to look at the surfaces for some possible explanation
of the change in friction.
The variation of coefficient of friction with load, for both tufftrided En8
and sulfinuzed En8 implied that below the temperature at which the lower friction
had become established the coefficient of friction did not vary significantly with
94
load. In the case of Sulfinuzed En8, above the friction drop the coefficient
of friction fell steadily with increasing load levelling off at a value of less
than 0.1 from an initial high value. The most straightforward assumption that
can be made on this information is that two different types of surface interaction
are taking place above and below the temperature associated with the drop in
friction.
A study of the time taken to run-in the various combinations of surfaces
was not very valuable as a means of understanding the mechanism of material
behaviour. In practical terms it was useful to know the running-in time to
constant friction, as friction/temperature tests were always run-in before heating
started.
As the main interest in the treatments being investigated was in their wear
resistance, no further friction experiments were carried out. The next step was
to determine whether an effect analagous to the friction drop occurred in wear
experiments at different temperatures. This meant leaving the cause of the
friction drop open.
643 EAR TESTS
6.3.1 Tufftrided En8
Taking initially the graph, fig. 5.12, for the wear/temperature relationship
of tufftrided En8 at an arm load of 0.34 kgf, it was observed that the debris at
all temperatures was in the form of a fine ponder. At temperatures below the
peak wear rate temperature (FillT) the powder was brown, but above PM the powder
was black. Samples of each type of debris collected at room temperature (25°C)
and at 180°C respectively were identified as ferric oxide, Fe203, and tri-ferric
tetroxide, Fe3 04 (magnetite). It, therefore, seems reasonable to associate the
change in wear pattern with a change in the type of oxide formed during sliding.
The very steep drop in wear rate suggests a very rapid change in mechanism. The
fact that the wear rate continues to drop, albeit more slowly until a temperature
of 240°C is reached argues against an immediate and total change from one mechanism
95
to another. It is hardly necessary to stress that a change in wear rate from
'2.5 x 10 4g/hour to 1 x110 6'dhour is of a great significance, and could in no
way be accounted for by experimental error. Similarly,the association of the
change in debris type with the change in wear rate is also significant.
The question of the initial rise in wear rate is easily explained by
reference to the literature. Quinn (52) gives for oxidational wear
Ir oc kp
where kp is the parabolic oxidation rate constant. kp is related to temperature
by an equation of the Arrhenius type
RT kp = Ape (1.39)
Ap is the Arrhenius constant and Qp the activation energy. It cannot be emphas-
ised too strongly that kp is an empirical constant, and it is not at all clear
what the physical meaning of Qp and Ap can be. The origin of kp is in the
reaction rate equation
m2 = kpt
(1.37)
where the increase in mass of oxide and t is time. kp is, therefore,
specific to a specific reaction under given conditions (49, 50, 79).
In order to determine the activation energy for the system the points on
fig. 5.9 for an arm load of 0.34 kgf, at (82°, 0.75.. 10 4g) and (105°C, 2.2 . 10 4g)
were put into the equation
_ RT V = K e
(6.1)
Allowing for the known errors in calculating flash temperatures this gave an
activation energy of between 12.0 and 15.0 kcals/gmole. This is within the same
order of magnitude as the values quoted by Kubaschewski and Hopkins (50) for a
large number of metals, including iron, oxidising in dry air. Ho-Aever, the
values quoted were for temperatures in excess of 200°C, which is well above the
temperatures attained in this system. In addition the data quoted by Kubaschewski
and Hopkins was for the oxidation of iron to magnetite and ferrous oxide (Fe0) and
96
not, as in the present case, to ferric oxide. In conclusion then, that the
-values of activation energy found for the system are of the order of magnitude
previously reported for systems of a similar nature, helps to confirm the
approach, while the differences in values between those observed here and those
reported in the literature are to be expected.
Based on these figures a curve of wear rate against total temperature was
plotted, fig. 6.1. The comparison with the experimental curves for 0.34 kgf,
1.59 kgf and 2.95 kgf arm load is shown in fig. 6.2, 6.3 and 6.4. Perhaps the
most cynical comment on the fit between the theoretical and experimental curves
is in the title of Bernard Shawls play 'Too True to be Good'.
Passing on to the curve, fig. 5.13, for wear against temperature at an arm
load of 1.59 kgf, very much the same pattern of wear is observed, with rather
higher wear rates below the transition. Identical results were obtained when
debris samples were examined, although it was not thought necessary to send a
sample of the black debris for X-ray diffraction. The nature of the black fine
magnetic powder being considered sufficient evidence to justify identification
as Fe304'magnetite.
The most interesting difference between the two curves, figs. 5.12 and 5.13,
is in the drop in URT. Looking at the curve, fig 5.14, PURT has dropped even
further and the slope of the curve immediately after the maximum has become
noticeably lower. In fig. 5.14 it is also clear that the scatter of results is
considerably increased suggesting some change in the factors influencing the
process. In addition, whereas with previous points debris could be clearly
distinguished as black or brown, the point at 130°C on fig. 5.13 was clearly a
mixture of both types of debris as were the points on fig. 5.14-between PURT and
118°C.
In order to confirm this visual impression a sample of debris was collected
at 110°C and 2.95 kgf arm load. This sample was slightly magnetic and under X-ray
diffraction proved to be a mixture of Fe 02.3 and Pe3
04.
97
g h x to-4
Figure 6.1 Dependence of Theoretical Wear Rate on Total Temperature
1.0
0.1
15.0 kc/mole
°K 'x 10-3
2.0 3.0
98
g/h x 10.4
Figure 6.2 Dependence of Wear on Total Temperature, 0.34kgf, Tufftrided En8
---d-1.0
Ferric Oxide/Magnetite Transition Line
0.I
oldx 10-3
1 1.0 2.0 3.0
99
Figure 6.3 Dependence of Wear on Total Tempe: Tufftrided En8
-0 Ferric Oxide/Magnetite Transition Line
a 0, 1.59kgf
1-10 240 30
x10 -3
100
g/h x ICY4 Figure 6.4 Dependence of Wear on Total
Temperature, 5.20kgf, Tufftrided En8
1.0
Ferric Oxide/Magnetite Transition Line
T°I
xio-31._
1.0 2.0 3.0
101
At the higher loads of 5.20 kgf and 11.10 kgf the initial rise had
vanished completely, and the uniformly black appearance of the debris suggests
very strongly that it has been pushed below the range of temperatures which
occur in this equipment. The small dip in the curves will be the subject of
discussion later in this section.
There is, therefore, undoubtedly a load effect operating on the PWRT.
Because the contact pressure in-the system is constantly changing, due to the
steadily increasing wear scar area on the rollers, the variation cannot be tied
down simply to contact pressure. A constant temperature assumption based on the
idea of the total temperature and the flash temperature occurring when asperities
penetrate the fluid film would seem a reasonable explanation.
On the basis of earlier comments concerning the exponential dependence of
oxidation on temperature, it is clear that the contacts which generate the highest
flash temperatures mill have very much the highest wear rates, until that is the
change from production of Fe203 to Fe304
takes place, when the fastest wearing
contacts will become the slowest. This process would, therefore, be expected to
result in a sharply peaked curve, such as figs. 5.12, 5.13, 5.14. It follows
from these assumptions that under conditions where the flash temperature is low
in relation to bulk temperature a rapid fall off in wear rate will occur after
MST. If, however, the flash temperature is a major part of the total temperature,
and if one assumes a distribution of flash temperatures among the individual
contacts, then a more gradual drop in wear rate would be expected.
The nature of the hydrodynamics of this system are very complicated and
constantly changing with time. It is, therefore, assumed that when the total of
bulk temperature and maximum flash temperature (based on Archara's analyses (E6) of
Blok's theory (80) ) exceed a certain critical value then the transition will occur.
This theory gives total temperature.
T =TB + 0.25/A pm)2 x 1721T. (6.2) 217
where
102
V = velocity
N'T = load
ft. = Thermal Diffusivity
Pm = Flow pressure
i° = Density of metal
C = Specific heat of metal
= Coefficient of friction
The values of coefficient of friction used in the calculation are taken
from _observations during wear tests above and below FUT. The thermal
diffusivity is taken from tables for mild steel, as are density and specific
heat. These are all considerable assumptions as there are clear differences
between the surface of the tufftrided steel and a normal mild steel. A further
question is the appropriate value of hardness. It can be seen from fig. 5.3
that the variation of hardness with distance from the surface is extremely rapid.
For the purpose of the calculations the value at one thousandth of an inch (500 VPN)
was used, a reasonable mean of the possible extremes.
The results of the calculation are shown as the horizontal line on fig. 5.17.
Calculating the expected values of PWRT for arm loads of 5.20 kgf and 11.10 kEft they
are found to be 10°C and 20°C respectively. As octane freezes at 19°C experi-
mental determination of these points is clearly not practical. ladle this is not
a rigorous proof of the theory of a constant temperature criterion for the trans-
ition, it provides very strong support. The possible reasons for the existence
of the transition are discussed later.
6.3.2 Sulfinuzed En8
For the rear of sulfinuzed En8 there is an almost identical pattern to that
for tufftrided En8. At an arm load of 0.34 kgf wear rates are found to be about
. four times those for the tufftrided steel. This may be explained by the lower
hardness, see figs. 5.2 and 5.3. Examination of sections through the surface
have sham the sulfinuzed surface to be more porous than the tufftrided surface.
The greater porosity will allow greater oxygen penetration, and would be expected
to promote oxide formation.
1:01
N.)
0
•
*.■•••.1 -,•multo• • ...to. I ir m.o.... .ta I•At/4.•1•11,114•91011.., ...tea. WIMSMA.P..111.1.■•• .14.IMP•••••11•1 ante..
Pu- V3P1= 1J11,1, oojato:: .J1J,T
uo
cooLt Jo 4:eli*u'l"fi klarg
Tur e 6.6 : • '3 ■•••, c: 1`."1 for tir..):AT of .;u1 G H T, to
i; • :ter] on a rP,:-.1'0:5:- Ion ( 1 i r
0 01
103
At an arm load of 0.34 kgf the peak wear rate temperatusAMT,was
110 - 10°C. The maximum attainable flash temperature under these conditions
is 50 ± 10°C. This gives a total transition temperature of 160 ± 20°C, which is
in agreement with the results for tufftrided En8. On this basis the high values
of wear at around 50°C on fig. 5.21 for 1.59 kgf, are the peak weir rate, fig. 5.22
fig. 5.23 for 5.20 kgf and 11.10 kgf should be well past the peak. The last two
tests are rather dubious as there were frequent instances of severe wear at these
loads, aril it is impossible to guarantee that the wear during all tests was
completely mild. It is possible that a short period of severe wear could be
followed by a recovery. Such an incident would make a very large difference to
the wear rate for the test.
The only example of a ferric oxide reaction for sulfinuzed En8 is, therefore,
at 0.34 kgf. The curve agrees with the theoretical predictions based on the
Arrhenius equation.
It is, therefore, reasonable to assume that the wear mechanism for
sulfinuzed and tufftrided En8 is identical. The reasons behind the transition
will be discussed later.
6.3.3 Normalised En8
The wear/temperature relationship for normalised En8 must be considered in
a different light from the processes involving the salt bath treated steels.
The rear in this case is not of a corrosive pattern but is the severe type of wear
described in the introduction. This type of wear is largely abrasive in character,
with detached particles forming the abrasive grains. It is clear that the wear
rate will be dependent on the size Anti number of these particles, which are in
turn dependent on the wear rate. This dependence indicates that a linear relation-
ship of wear rate with any variable is highly unlikely. It is suggested that the
increasing temperature leads to a reduction in viscosity of the cetane (85) which
in turn leads to an effective increase in the loads in the contact. Initially,
this effect is responsible for the onset of severe wear, and subsequently for the
steady increase in wear rate.
104
A further factor which may influence the wear temperature relationship is
the slight increase in 'hardness of the normalised steel close to the surface.
At very high year rates, such as those experienced in severe wear the depth of
the scar is sufficient to canse significant variations in hardness during a test,
see fig. 5.1, which will again lead to the magnification of the effect of small
changes in external conditions.
The main import of these tests is that they should demonstrate whether, or
not, the salt bath nitriding techniques are advantageous. There can be little
doubt of the superiority of performance of the treated steels under the conditions
in the tests.
6.3.4 The Variations of 7ear Rate with Load
Looking at the wear/temperature curves for tufftrided a8 it can be seen
that the increases with load from 0.34 kgf to 1.59 kgf and then falls to
.2.95 kgf. Some attempt will now be made to explain this. Firstly, if it is
assumed that all surface contact is plastic, and in the flash temperature cal- 1
culations this assumption has already been made, then the surface area in contact,
and therefore the rate of corrosion, and hence wear, will increase linearly mith
load. If the contacts are not completely plastic some increase will still occur,
'but it will not be linear. That is a sufficient explanation for the increase in
wear at PVIRT. The decrease will now be explained.
The temperature flash calculated so far has been the maximum attainable,
however, if it is assumed that it is not always attained, an explanation of the
fall in peak wear rate (PIE) might be forthcoming.
Assuming that the flash temperature at a given contact (t) is randomly
distributed between 0°C and T°C where T is the maximum attainable flash temperature,
then the number of contacts with a temperature between t and t Et °C will be given
by
N St
where N is the total number of contacts. T
This mill, therefore, give a total wear rate
105,
T 1r = S C ntt+g) dt
e
(6.3)
Where 9 is the bulk temperature. Solving this equation for the P-RT by Simpson's
formula gives a large variation in result depending on the value )f gp which is
used. Table 6.1 gives the results of the calculation.
Table 6.1
Arm Load kgf
Flash Temperature oC
Wear Rate 10 4g/hr Cp = 12.0 15 kcals/mole.
0.34 40 17.6 3.32
1.59 60 13.1 2.37
2.95 75 10.7 1.95
These results would indicate that the peak wear rate does, indeed, decrease
with increasing flash temperature. These results would also point to a value of
Cp closer to 15.0 than 12.0 keals/mole. It must be remembered that these values
were based on the simpler, and more common assumption, that all contacts were at
the maximum attainable flash temperature. However, a slight modification in the
constant term, a reduction of 2%, will place both the tests at 0.34 kgf and 2.95 kgf
into very good agreement with experimental results. The test at 1.59 kgf, however,
is a long way from agreement.
Oxidation theory produces a direct relationship between the area of the
oxidising surface (50) and the rate of oxidation. This will carry over to
oxidational wear, and Quinn's full theory (52) includes a term for the real area
of contact such that
V ck A.
The only problem is in finding the real area of contact. It may be assumed
that all the asperity interactions are plastic then the real area of contact is
directly proportional to the load. However, it has long been established (81)
that this is not the case for a smooth surface. The whole preceding analysis is
based on the assumption that there is a range of contact stress all over the
surface. Finally, introducing an area term directly proportional to load would
not bring the observed results closer to the theory.
106
A second possibility is to assume that the apparent area of contact is
-directly proportional tb the,real area of contact. This solution, although it
would make theory and practice agree, does not stand examination. As the wear
rate at a time t would be depenAnt on the amount of wear up to that time the
wear rate would be approximately exponential with time, whereas the opposite is
in fact the case. The wear rate falls with time until a steady wear rate is
set up. •
A third possibility is to assume that wear is negligible and to use the
Hertzian area as the real area of contact. This would still not make the
results observed fit the theory, and does not have the benefit of superficial
probability. The area of the wear scars observed are normally about five to ten
times the Hertzian area.
Drawing these remarks together; it can be demonstrated that a drop in
peak wear rate will occur with increasing flash temperature. It is expected that
an increase in real area of contact will act against this drop, but in order to
fully understand the variations of contact area with toad, for the system used,
much more work must be done.
6.3.5 The Ferric Oxide to IRIP;netite Transition
Tomashov (57) points out that when iron is oxidised at low partial pressures
of oxygen, magnetite will be formed. This is because ferric oxide has a high
dissociation pressure. It is also known that for high temperature oxidation of
iron the normal product is magnetite while ferrous oxide is formed above 570°C.
By high temperature is meant in excess of 2000C, and a search of the literature
has failed to ellicit any work for temperatures below 200°C.
The apparent conclusion is that at approximately 160°C something happens
which limits the amount of oxygen available to the surface. The mechanism cannot
be. a bulk lubricant phenomenon, because the bulk of the lubricant is often nowhere
near 160oC. The solution must, therefore, be on the surface, which means either
1071
a surface activated lubricant reaction, or some reaction of the surface itself.
There is no evidence for significant changes taking place on an iron surface at
160°C and, therefore, it would appear most likely that the effect is due to some
lubricant activity at the surface. It is possible that the lubricant action is
catalysed by the presence of nascent iron produced by wear. Clearly the form-
ation of any kind of lacquer or "friction polymer" on the surface will restrict
oxygen access and so encourage the formation of magnetite.
If the only evidence for this effect were the work reported here there
would be a strong temptation to conclude that the oxide transition is due to a
surface activated lubricant effect. However, it has been reported (42) that
a very similar process occurs in the dry wear of hard steels. In those exper-
iments the transition was found to be dependent on the composition of the steels.
It would, therefore, seem to be the case that for a given steel there is a critical
temperature at which the oxide type formed during sliding Changes from ferric oxide
to magnetite. The reason why this should happen is not clear.
6.3.6 The Change in 7rear Rate at the Transition
There are a number of reasons why magnetite would be associated with a
lower wear rate than ferric oxide. Ferric oxide has a hardness of about 1100 VPN
as opposed to 600 VPN for magnetite (82). This will have two effects; the
substrate has a hardness of approximately 600 VPN and, therefore, the magnetite
will be less likely to be broken up by large plastic deformations which it cannot
follow; secondly the wear debris will be much less abrasive. The fact that all
the debris is in the form of oxide does not preclude abrasion of metal by the
wear debris, as any metallic chips would be so small, of the same order as the
abrasive, that oxidation would be expected to occur very rapidly.,
Sakurai (83) has shown that the heat of adsorption of some polar molecules,
noticeably stearic acid, is about ten times higher on magnetite than it is on
ferric oxide. If oxidation of the lubricant at the transition temperatures is
producing any compounds with a tendency to act as boundary lubricants, their
108
effect will be much enhanced after the oxide transition. The results of
Clark et al.(42), however, would suggest that this can be no more than a minor
contribution to the drop in wear rate.
Observation of the experimental curves, figs. 5.12, 5.14, 5.15 and 5.16,
' shows a small rise in wear rate at temperatures above 180°C. Such a rise would
be compatible with a parabolic oxidation governed by a significantly higher
activation energy than that obderved for the formation of ferric oxide. This
is suggested by the literature (49, 50). Provided the oxidation proceeds at
a sufficient rate for the maintenance of an,adequately protective surface film
(that is, provided there is no adhesive wear), a lower reaction rate will result
in a direct reduction in wear rate as
V 04 e
A figure for Qp would have to be arawm125koals/gmole,f.6.3, there is not
.enough data to be more precise. The inflammability of cetane renders the
acquisition of data at higher temperatures inadvisable, as the consequent oxygen
starvation would be expected to distort the results.
6.4 TEE SUTELE ';IBAR TRANSITION
6.4.1 The Ybcharism of the Transition
As was explained in the introduction, the mild to severe wear transition
is dependent on contact pressure. This rig does not run at constant contact
pressure and cannot determine the pressure immediately before a transition, so
a precise value of the transition pressure is clearly not possible. The idea
behind these tests was to find if the materials would fail under any likely
running condition, and what the manner of failure would be. Published data (61,
64) indicated that sulfinuz and, to a lesser extent, tufftride were virtually
scuff proof.
The mechanism of the transition appeared to be that large particles 'peeled'
off the wear tracks, plate 19, in the manner suggested by the delamination theory.
109
Plate 19 Particle Flaking off a Surface (x200)
Plate 20 Transferred Particle on a Wear Track (x 100)
110
The particles then adhered to the surface of the track, plate 20, and caused
abrasion of the opposing surface. Plate 15 shows where two large particles on
the cone have gouged out the surface of a roller. Once this process starts,
wear becomes very rapid as the number of abrasive particles increases rapidly.
The hardness of these particles has been measured at between 600 and 700 LPN,
which is much harder than all but the surface, white layers of the coatings.
6.4.2 Sulfinazed EnB
On the basis of the wear scar areas determined on earlier tests, the
average pressure in the contact cannot have been greater than 50 kgf/mm2 at
the point where the mild/severe transition took place. Burwell and Strang (3)
had shown that this transition normally occurs when the pressure in the contact
reaches one—third the flow pressure, which is usually the yield stress of the
metal.
From microhardness measurements it is known that the flow pressure for
sulfinuzed En8 steel is between 650 and 300 kgf/mm2 dependent on depth. If the
maximum pressure in the contact is indeed the avera6, then failure is occurring
under unusually mild conditions. Assuming that no wear has taken place before
failure the absolute maximum from the Hertz theory (11) is 120 kgf/Mm2 This
. pressure would only be just sufficient to explain the failure. In any event
it is certain'from the data of figs. 5.20 to 5.23 that a significant amount of
wear had occurred, and so the lower figure will have been nearer to reality.
All that can be said is that the load distribution in a contact wearing
to conformity is not fully known. It is probable that individual joints in the
contact suffer pressures much higher than the mean. Finally, the consistency
of failure load would suggest a sharp pressure gradient in the contact in which
maximum pressure varies consistently with load.
6:4.3 Tufftrided En8
The first set of tests on Tufftrided En8, where temperature was allowed to
111
rise freely, gave a mean failure load of 70 20 kgf arm load at a temperature
of about 210°C. This is equivalent to a Hertz contact pressure of about
400 kgf/mm2, which is quite sufficient for failure to be expected. The mean
pressure based on estimates of wear scar area is not greater than 150 kgf/mm2
which is rather lower than would be expected. These results are very similar
to those for sulfinuzed En8.
The dry tests gave a failure load of about 45.4 kgf arm load. This
would suggest that a significant proportion of the load is being carried by
the lubricant film. Because of the complicated geometry of three small
conformed pads on the side of a cone, an exact hydrodynamic solution is not
possible. Using the analysis for a tilting pad thrust bearing (58), a film
carrying a load equivalent to 25 kgf arm load would have a minimum thickness of
the order of 50 4 which is of the same order as the wear debris, and would
therefore seem reasonable. The variation in viscosity of cetane with temper-
ature will also allow for the higher arm loads required when the temperature
was held down to 110°C.
To sum up, tufftrided and sulfinuzed En8 appear to exhibit a transform-
ation from oild to severe wear, under conditions which would be expected of any
metal of the same hardness.
6.5 POSSIBLE LUBRICANT EFFECTS
It is known (77, 78) that above about 120°C cetane oxidises producing,
among other things, some compounds which act as boundary lubricants. It was
observed that above 180°C the seizure resistance of sulfinuzed DO was much
enhanced, and also that in tests where debris was constantly magnetite, a
relatively small drop in wear rate could be observed above 120°C. This drop
in wear rate was at its greatest a reduction of to a half or a third the initial
value. While this is a fairly large reduction it is small compared to that
associated with the change in debris type.
112
In addition to the slight drop in wear rate there was less scatter
among data points, and a slightly lowered coefficient of friction, from 0.3 to
0.15 in the case of tufftrided En8 at 5.20 kgf arm load. These are all effects
which are consistent with the introduction of a boundary lubricant to the system.
It can be said, therefore, that chemical lubricant effects are small by
comparison with the effects associated with changes in surface oxide. rbile
a reduction in wear is noticeable under these conditions the halving of the
coefficient of friction is the most noticeable of the effects.
113 CHAPTER 7 : CONCLUSIONS
From the previous chapters it can be concluded that sulfinuzed and
tufftrided En8 steel show the same mechanisms of wear imier the conAtions
studied. Wear has been shown to increase exponentially to a critical temp-
erature of approximately 160°C, where the critical temperature is the total
of bulk and flash temperatures. This increase is in accordance with existing
theory for corrosion and corrosive wear. 'Above the critical temperature the
wear rate falls by approximately two orders of magnitude.
Below the critical temperature the wear debris is ferric oxide, Fe203,
above the critical temperature it is magnetite, Fe304. The debris in both
cases is a fine powder with a particle size of about 150 R. Analysis of
results and comparisons with the literature has shown that the change in the
type of oxide is probably due directly to an effect of temperature on the iron
— oxygen reaction. It is unlikely that the lubricant is involved.
Sulfinuzed and tufftrided En8 show a transition from mild to severe
wear under loads which would be expected for metals of the same hardness.
There is, in conclusion, nothing in the behaviour of either sulfiruzed
or tufftrided En8 steel which would indicate any special property, other than
an increase in hardness, as a result of the treatment. It must be said that
this increased hardness is very great, with a consequent improvement in
scuffing resistance over the normalised En8.
Finally, it is suggested that for maximum benefit from these steels
they should not be used where the total contact temperature is less than
160°C. It would be unwise to exceed a contact pressure of 50 kgf/am2 for
sulfinuzed En8 or 150 kgf/mm2 for tufftrided En8.
114
CHAPTM. 8 RECOMENDATIONS FOR FUTURE WORK
The first question to require an answer is why the oxidation of the
surface changes from the production of ferric oxide up to 160°C to the
production of magnetite thereafter. In addition, further experimental
work would be useful in confirming the validity of the integrated Arrhenius
equation, used to predict peak wear rates.
Immediate engineering questions which require an answer are the effect
of E.P. and boundary agents, ard how far this pattern is repeated with
different steels.
Finally, as the origin of this research was in a requirement for
cheap and reliable materials for automobile camshafts, controlled tests on
a cam and tappet tester would be useful to confirm the design criteria set
out in the conclusion section. Although the ultimate confirmation must
lie in service, a cam and tappet test would give a relatively inexpensive
confirmation of the utility of these results.
The integrated Arrhenius eeuation (6.3), can.be shown to
explain the apparent discrepancy between the curve of figure 6.4,
for the dependence of wear on total temperature for Tufftrided
En8 at a load of 5.20kgf, and the other results. It is proposed
to make this the subject of future analysis.
115,
Appendix: Program for the Calculation of Wear
A listing of the program described in Chapter 3 is given in
the two following pages.
116
G0100 PROGRAM WEAR(1NPUTODTPUTITAPE5=1NPOIITAPE6=0U1PUT,TAPE7,TAPE8) 00110 REAL LOAOI MASS 00120 04.MENSiOtv Al (6,15) ,A2 (6115) , 31(6,15) ) 62 ( 6,15) , TIME(15) , 11(15) ,S (15),11(15) 00130 REAL M 00140 1NTEUER PI LUD,A6D 00150 D...MLNS1ON TEMP(15),NAMP(15) 00160 UIMENS.&ON DM(15) 00170C READ IN DATA 00180 WRITE(611)) G019G mcAulLOAO ICONVLICON1/0,R,MHO IK 00200 WRILA6,11) 00210 READ(5,12)LUdIADO,I1,12 0022U DO luG J=11 K 00230 U0 10. 1=1,6 00240 iN LAD ( 8 1 ) AL (I ,J) g 42 (Ig J)01(I 9J) 182 (I -1 00250 100 CONTINUE 00260C PREPARE CONSTANTS 00270 IR=1,J00 00280 P=3 00290 PIE=3.14159265 00300 S2=SQRT(2.) 00310 R=R*GONVO 00320 W=(LOA0-3.)/3.*S2*CONVL 00330C READ NAME OF MATL., TIME AND TEMP.OF TEST 00640 wRITE(6,92) 00350 READ45,12)13,14,15,I6 043800 mEAu iN TLMES OF TESTS 00390 WRITE(6,90) 00400 DO 160 I=1 ,K
READ ( 51 )TIME(I),TEMP(I) 60420 160 CONTJ.10E 004300 LNTLR MA -IN OUTER LOOP 004400 START NEW PAGE 00450 DO 110 J=1,K 00460 P=P+1 66470 J.F(P.LT.3)G0 TO 170 00480 P=1 00490 WR1Ti(7,98) 0000 1/0 WRITE(7,70) 00510C WRITE NA1E OF MAIL.ITIME ANJ TEMP. 00520 limITE(7,93)I6,14,./51i6 00570 WRITE(7,62)LOAD 60580 wRITE(7,33)LUB,ADO,I1,I2 00590 WRITE(7 1 3)/TIME(J) 0060C Wm11:(7,91)TaMP(J) 00610 W;ITE(7,50) 00620 WRITE (7,51) G0630C SET VARIAJLES TO ZERO 0064u j0 130 L=1,7 00650 M(L)=0. 00660 S(L)=0. 00676 D(L)=J. 00680 130 CCNTIlUE 0C69uG GAL-GULAi:ON OF hEA.‘, LNTER "JAIN .NNiER LOOP 00700 DO 120 1=1,6 007100 GALGULAT: 4ARIA ,3LES 00720 A=A0S(A1(I,J)—m2(I,J)) 4CONVD/2.0 00730 0=AB:)(01(,:,J)-02(1,J)) 4CON40/2.0 00740 G=R—SOIRT(;-**2—(6**2)) 00750 ARLA=A4L6*PIE 60766 CALL iOLU1E(A,6,VOL,IR,CONVD) 06770 MASS=VOL+RHO 00780 1-JES=W/A -.EA 00790 WRITL(7,30)A,B,C,Ar1EA,40LIMASS,PRESS UO8COC SUM iARI43LES 00610 M(1)=M(1)+A u0620 M(2)=M(2)+ ,:i 00830 M(3)=M(3)+C 00d4u M(4)=M(4)+AREA 00850 M(5)=M(5)+40L 0066y M(6)=M(6)+MAS 00670 M(7)=M(7)+PRESS 06680C SUM SQUARES OF VARIA3LES 66890 S(1)=S(1)+A**2 vu9uU S(2)=S(2)+ 442 00910 S(3)=S(3)+0442 00920 S(4)=S(4)+ARLA."2 00930 S(5)=S(5)+/OL"2 tJu94U S(6)=S(6)+1AS**2 00950 S(7)=S(7)+PRESS442 0096JC EXIT MAI') INNER LOOP 0:97J GONTIlUE 0u966C, —Nf:R :.,dATI...)1.6C..) LOOP 00990 00 100 L=1,7 010000 PRODUCE AZAN 4ALUES
117
01610 M(L)=M(L)/o. 01020C Pm0Jt..t.:c ;JANDAm0 J14iAT1ON..) ANO SIANDA“,, Emm0,.S 01030 U(L)=(S(L)/6.-M(L)**2) 01040 IF(D(L)..0.)G0 TO 150 6105G U(L)=SURT(J(L)) 01066 DM(L)=SORT(5.0/3e.d)*D(L) 01u70 E=0(7)/M(7) 01080 150 CONTINUE u1090 IF(E.LT.J.3) GO TO 161 01100 IIRIT:_(7 1 T3) 01116 161 hmlf.:(7,70) 01120 WRITE (/,40) 01130 Wm1TE(7 JD)M(1)0(2)0(3),M(4),M(5)0(6),M(7) 0114C. WRITE(7;u:) 4)115C (I) I ( U ( y ), ( (6) 9r) (r)
01160 WRIT_Ale tl.) 01170 WNITE(71 30)0M(1)0M(2)1 0M(3).0M(4)20M(5),UM(0),0M(7) 01190 110 CONTINUE 612000 FOK0AF ,-1;ATzMENTS 01211; 10 FUfr.MAT(*!...Sf CONS(ANTSILOAO,CONVL,CON4D,M,RHOIK*) 01220 11 FORMAT (*WRITE 603m.I.CANT TYPE*) 01230 12 FORmAT(4A10) 01240 30 FORMAT(1X,7(1PE9.3,1X)) 01250 40 FOmMAT(1X,*MEAN VALUES*) 01260 50 FORMAT(5X 1 1- 4 4-,9X,*BA',9X,*G4- ) 7X,*A-:EA*,5X0VOLUME MASS P:E3SUI=.E 4.) 01270 51 FOmMAT(5Xy*MM4,2(8X,*MM4 ),7W HM Sie,4X1 4.NM CI.*1 6X,*GH4.,* N/Mm S041 01266 00 FORMAT(1X,4STANOARO UE4IATIONS*) 01290 70 FOKrAi(+ *) 01300 b0 FURMAT(1X,*TIME OF jEAi; TEST=*,F6.1,2X,*11iNUTES*) 0131u 81 FOm1-1AT(1)(,*STANOAR0 ERROR OF MEAN') 01320 32 FORMAT(1X,+LCAD = 4-,F0.2)* LOS*) 01360 ciS FOAf(1X 1 4.,_U3-c.LCAN1 4,4A1L) 01340 90 FO;MATC1X 1 +LIST T1M:.S AND TEMPErtATUr.HS OF WLAR TESTS*) 01350 91 FCH-IAT(1X,*Ti.PEI,,ATU%E-*,F6.1,ex,-OFG..EES 01360 92 FORMAT(1X,4-draTE TYPE 6F - 1ATERIAL UNO-_R 0137C 93 F0K4A1(1X / 4A16) 01420 98 FOmMAT(1H1) 01430 99 FOmMAT(1X,40U(31.0US 7tESULT,STANDAk0.0ai.AT-ON C..E.ATE-. THAN THITsr PLP.C-E- NT 01440+ OF MEAN4) 01450 STOP 01460 ENO 01470 SOaKOUTINI VOLUM'I'(A,9,VOLoIRICONVO) 01480 X=0. 01490 VOL=0. 01500 DELX=6/(2.0*FLOAT(IR)) 01510 00 1_2u M=1,iM 01520 X=X+DELX 01530 CORU=2.0 4A*SCIRT(1.0-X"2/64.4 2) 01540 R=0.675*S1J(2.6)4 CON4D 01556 THET1 01 mS1N(CCm0/(2.04M)) 01560 AmC=,:*42/2.94 (THETA-31N(THETA)) 61570 40L1=ARCELx 01580 40L=40L+40L1 4 2. 01590 1020 CONLNUE 01600 mETU.7,,N u1610 ENO
110
..REFERENCES
'Contact and Rubbing of Flat Surfaces'
J.F. Archard J. Appl. Phys. V.24 1953 pp. 981-98a
'The tear of Metals under Unlubricated Conditions'
J.F. Archard & 7. Hirst Proc. Roy. Soc., A, V.236, 1956 PP. 397-410
'Metallic rear'
J.T. Burwell & C.D. Strang
1Proc. Roy. Soc., A, 1.212, 1952 pp. 470-477
'Adhesion of Solids and The Effect of Surface Films' J.S.HcFarlane & D. Tabor Proc. Roy. Soc., A, V.202, 1950 pp. 224-243
'The Friction and rear of Metals' E. Rabinowicz
Tilley 1965
'Friction and tear'
Tragelskii
Butterworths 1965
'Friction tear and Lubrication in Vacuum' D.H. Buckley
NASA SP-277 1971
'The Delamination Theory of tear' Suh
tear V.25, 1, 1973 pp. 111-124
'Single Contacts and Multiple Encounters'
J.F. Archard
J. Appi. Phys., V.32, 1961 pp. 1420-1425
119
10 'On the Etpericial Law of Adhesive Wear'
J.T. Burwal & C.D. Strang
J. Appl. Phys., V.23, 1952 pp.18-28
11 'The Dry Wear of Steels'
N.C. Welsh
Phil. Trans. A., 257, 1964-5, pp.31-70
12
'Surface Aspects of Unlubricated Metal to Metal Wear'
T.S. Eyre & D. Maynard Wear, V.18, 1971, pp. 301-310
13 _ 'Friction and Wear of Steels in Air and Vacuum'
R. Predmore, J. Jellison, C. Sturgis
ASLE 14, 1, 1971, pp. 23-31
14 'The Effect of Combined' Stresses on the Transition from
Mild to SevereWINar,
R.D. Arnell, A.P. Herod, D.G. Teer
Wear, V.31, 1975, pp. 237-242
'Further Investigation of the Delamination Theory of Wear'
S. Jahanuir, N.P. Suh, E.P. Abrahamson, A.P.L. Turner
ABME, J. Tub. Tech. 11.96, F, 1974, pp. 631-637
• 16
'The Delamination Theory of Wear and the Wear of a
Composite Surface'
S. Jahanmir, iT.P. Suh, E.P. Abrahamson • Wear, V.32, 1975, PP.33-49
17 'Relationship Between the Coefficient of Friction and the Wear Rate of Metals'
N.P. Suh & P. Sridharan
Wear V.341 1975 pp. 291-299
18 'Principles of Abrasive Wear'
M.S. Kruschav
Wear V.23, 1974, pp. 69-88
120;
'The Influence of Solid State CoheAon of Metals and
Non4letals on the Magnitude of their Abrasive rear
Resistance'
A.K. Vijh
Wear 11.55, 1975, pp. 205-209
20 'The Adhesion of Carbon and Carbide Steels at High
Pressure and Temperature'
K. Iwata, K. Karasaka, J. Achara
Wear V.18, 1971, pp. 153-164
21 'The Appearance of the Contact Zones and the Mechanism
- of Normal Adhesion with Soft Metals'
C. Dayson & J. Lowe
Wear V.21, 1972, pp. 263-288
22 'LP.,1) Study of Adhesion of Copper and Tungsten to the
(111) Surface of Nickel'
D.H. Buckley
ABLE 13, 1970, Pp. 39-46
23 'Adhesion of Metals to a Clean Surface Studied with
LEED and Auger Spectroscopy'
D.H. Buckley
Wear V.20, 1972, pp. 89-104
24 'Adhesion of Metal Surfaces under Fretting Conditions.
I ,Like Metals in Contact'
B. Bethune & R.B. Waterhouse
Wear V.12, 1968, pp. 289-296
25 'Adhesion of Metal Surfaces under Fretting Conditions.
II "Unlike Metals in Contact'
B. Bethune & R.B.7aterhouse Wear V.12, 1968, pp. 369-374
26 'The Role of Surface Shear Strains in the Adhesion of Metals'
0.L. Anderson
Wear V.3, 1960, pp. 253-273
123.
27 'A Study of the Adhesion of Copper Using the Twist
-Compression. Technique in Vacuum'
J.M. Bradford & M.E. Sikorski Wear' V.16, 1970, pp. 413-420
• 28 'Elemental Analysis of a Friction and Wear Surface
during,Sliding using Auger Spectroscopy'
D.M. Buckley & S.V. Pepper ABLE, 7.15, 1972, pp.252-260
'Electrical Contacts'
R. Holm
Gerber, 1946
30 'Tear Particle Formation Mechanisms*
H. Koba & N.H. Cook
Ann. Report, Materials Processing Lab. Dept. Mech.
Eng. E.I.T. 1973-74
31 'The Investigation and Interpretation of the Nature
of Tear Particles'
V.C. Westcott & J.L. Middleton
Trans-Sonics Inc. Report 11.00014 73 C 0454
May 10, 1973 - March 9 1974
'Some Consierations on the Fields of Stress Connected
with Dislocations in a Regular Crystal Lattice'
J.M. Burgers
Proc. Kon. Ned. Akad. Wet. V.42, 1939, pp. 293-325
33 'Dislocations and Plastic Flow in Crystals'
A.M. Cottrell
O.U.P. 1953
34. 'UnlUbricatedBliding Between Copper and Steel Surfaces'
Kadhim & Earles
I. Mech. Eng. 5th Lub. & Wear Cony., 1967, pp. 25-30
122
- 35 'gear of Unitibricated Steel Surfaces in Sliding Contact' D,Ge Powell & S.W.E. Earles
ABLE, V.11, 1968, pp. 101-108
36 'Variations in Friction and ;Tear Between Unlubricated Steel Surfaces'
Earles & D.G. Powell
I. Mech. Eng. 5th Lub. & Tear Conv. 1967, pp. 16-24
37 'Wear Characteristics of Some Metals in Relation to Surface Temperature'
Earles & E.G. Hayler
Wear V.20, 1972, pp. 51-58' .
38 U. Simplified Theory for the Oxidative Wear of Steels'
N. Tenwick & S.W.E. Earles
Wear, V.18, 1971, pp. 381-391
39 'Observations on the 'ear of High Hardness Steels'
A. Kasak & T.A. Neumeyer
Wear, V.14, 1969, pp. 445-454
40 'Oxygen an Extreme Pressure Agent' R.O. Bjerk
ABLE, v.16, 1973, pp. 97..106
41_ 'Wear of Cast Iron in Vacuum and the Frictional
Hardened Layer'
M. Kawamoto & K. Okabayaski
Wear, V.17, 1971, pp. 123-138
42 Mild Year of Hard Steels in Air and Carbon Dioxide'
W.T. Clark, C. Pritchard, J.V. Midgley
I. Mech. Eng. Trib. Con. 1968, pp. 97-106
43 'Dry Wear of Steels as Revealed by Electron Microscopy and X-Ray Diffraction'
T.F.J. Quinn
I. Mech. Eng. Trib. Con. 1968, pp. 201-213
123
44 'The Role of Diffusion in Corrosive Wear'
P.P. Tao
ASLT:, V.11, 1968, pp. 121-130
45 'A Study of Oxidation Phenomena in Corrosive rear' P.P. Tao
ABLE, V.12, 1969, pp. 97-105
46 'Corrosive rear by Atmospheric Oxygen and Moisture'
J.K. Appeldoorn, I.B. Goldman, P.P. Tao
ASLE, V.12, 1969, pp. 140-150
47 'Scuffing as Influenced by Oxygen and Moisture' I.B. Goldman , J.K. Appeldoorn, P.P. Tao
ABLE, V.13, 1970, pp.29-35
48 'A_ Simplified Theory for the Oxidative Wear of Steels'
N. Tenwick & S.W.E. Earles
rear, V.18, 1971, pp. 381-391
49 'Metallurgical Thermochemistry'
O. Kubaschewski, E.L. Evans, CO. Alcock
Pergamon, 1967
50 'Oxidation of Metals and Alloys'
O. Kubaschewski & B.E. Hopkins
Butteriorths, 1953
51 'Theory of Corrosion and Protection of Metals' N.D. Tomashov
MacMillan, 1966
10xidational Wear'
T.P.J. Quinn
Wear, V.18, 1971, PP. 413-420
53 'Corrosive Wear as a Failure Mode in Lubricated Gears'
Goldman
• Tear, V014, 1969, PP• 431-444
124
54 'The Anti -ITear Behaviour of T.C.P. in Different
Atmospheres and Different Base Stocks'
Goldblatt & J.K. Appeldoorn
ASLE„ V.13, 1970, pp. 203-214
55 'An Interpretation of the dear Observed. with Natural
Hydrocarbons'
A. Beerbower & I.L. Goldblatt
Wear, V.18, 1971, PP. 421-425
56 'Wear Rate of Copper under Boundary Lubrication'
T. Sakurai, H. Okabe, I. Matsumura ABLE, V.14, 1971, pp. 221-225
57 'Oxygen and Sulphur Interactions with a Clean Iron
Surface and the Effect of Rubbing Contact on these
Lubrications*
D.M. Buckley
ASLE, v.17, 1974, pp. 206-212
58 'Principles of Lubrication'
A. Cameron
Longman, 1966
59 'A Simple Theory of Asperity Contact in E.H.L.' K.L. Johnson, Greenwood, S.Y. Poon
tear, V.19, 1972, pp. 91-108
, 60 'Regimes of Elastohydrodynamic Lubrication'
K.L. Johnson
J. Mech. Eng. Soc., 7.12, 1970, pp. 9-15
61 - 'Improving the Resistance of Ferrous Metals to
Scuffing, Wear, Fretting and Fatigue'
J.C. Gregory.
Metal Forming, Aug./sept./oct., 1968
62 'Sulfinuz*
I.C.I., Mond Division
125
63 Patent Specification 782, 263
'Improvements in the Production of a Thar ReSistant
'Surface on Ferrous Metal Parts'
1957
64 'JI1 Salt Bath Treatment to Improve the Resistance of
Ferrous tetals to Scuffing, 'rear, Fretting and
Fatigue'
J.C. Gregory
near, V.9, 1966, pp. 249-281
65 R.Tf. Cahn, Letter
Metal. Treatment, V.31, 1964, p.234
66
'Reducing Scuffing and rear of Ferrous Materials'
F.D. Uaterfall
Engineering, V.187, 1959, pp. 116-120
67 W.T. Chesters. Private Communication
68 'Use of Sulfur for Running in Rubbing Surfaces'
Fedyanin
Russian Engineering Journal. V.10, 1964, pp. 39 -40
69 'A Study of the Lubricating Mechanism of Sintered Iron
Impregnated with Elemental Sulfur in Sliding Contact
with liolybdenum and Tungsten'
A.A. Conte, M.J. Devine, E.R. Lamson, L. Stallings
ASLE, Lubrication Engineering, V.28, 1972, pp. 423-427
70 Patent Specification 891, 568
'Process for Carbonitriding Metals'
1960
71 Patent Specification 891, 578
'Process for Carbonitriding Metals, more especially Iron
Alloys, in Salt Baths containing Alkali Cyanide and Alkali Cyanate'
1960
126
72 'Tufftride' ,
I.C.I., Mond_ Division
73 'Thermal and Chemico-Thermal Treatments of Ferrous
Materials to Reduce Wear*
J.C. Gregory
Tribology, V.3, 1970, pp. 73-83
74 'Tufftride : Only Skin Deep?'
E. Taylor
Tecbn. Report No. C-70-36. 4, _American Society
for Metals. 1970
75 - 'Bath Nitriding for Ejector Pins and other Tool
Components'
H.R. Scbmauser
Society of Die Casting Engineers SDCE Paper 113
. '76 M. Bailey PhD Thesis. University of London, 1971
77 H. Spikes PhD Thesis. University of T,ondon. 1972
78 R. Hiley PhD Thesis. University of London, 1976
79 'High Temperature Oxidation of Metals'
P. Kofstad
Viley, 1966
80 'Theoretical Study of Temperature Rise at Surfaces
of Actual Contact under Oiliness Lubricating
Conditions*
H. Blok
I. Mech. Eng. General Discussion:Lubrication and
Lubricants. Group IV, 1937
81 'Surface Examination by Reflection Electron
Microscopy'
J.S. Halliday
Proc. I. Mech. Eng. V0169, 1955, pp. 777-781
127
82 'SurfaCe Oxide Films at Elevated Temperatures'
P. HancoCk & R. Hurst
'Advances in Corrosion Science and Technology'
Ed.' M.G. Fontana & R.T. Staehle, V.4, Plenum Press
1974
83 'Heats of Adsorption and Anti-7:ear Properties of
SoMe Surface Active Substances'
S. Hironaka, Y. 'Yelagi, T. Sakurai
J. Inst. Pet., V.17, 1975, pp. 201-205
84
'A Contribution to the Engineering Design of Machine
Elements Involving Contrashaped Contacts'
A. Horowitz
Israel Journal of Technology, V.9, 1971, pp.311-322
85 D.P. Gadd. PhD Thesis. University of London
(To be published)
86
Temperature of Rubbing Surfaces'
J.F. Archard
Wear, V.2, 1959, pp. 436-455.