vlqvn

l.p~'¢||t

 

v
o
.
r
I
v
v
.
I
u
a
-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.
-.
Z.

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

., :

 

 

 

 

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

       

5.4

 

 

 

 

 

 

 

 

 

 

 

mcmc‘sfi TWE 1u~nvfinfiu‘rv‘fl?nfmes 0'} Y 40,2 8/0 /7
lullilllfllilllllgllzllgyll ‘
LIBRARY
Michigan State
University

 

 

 

This is to certify that the
thesis entitled
Fatigue of Th02 Dispersion Strengthened
Nickel Alloy with Surface Modification by A1+ Ion

presented by Implantation

Robin Banerjee

has been accepted towards fulfillment
of the requirements for

M.S. degree in Materials Science

 

    

 

Major professor

( Dr. David S. Grummon )
Date May 17: 1989

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

PLACE N RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due. .
DATE DUE DATE DUE DATE DUE

 

 

 

 

 

1L1IL

11

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Will i ‘ «

 

 

 

 

L1

MSU Is An Aflirmdlve Action/Equal Opportunity Imam
6W”:

 

 

 

 

 

 

 

 

 

FATIGUE OF Th02 DISPERSION STRENGTHENED NICKEL ALLOY

WITH SURFACE MODIFICATION BY Al+ ION IMPLANTATION

BY

Robin Banerjee

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF SCIENCE
Department of Metallurgy, Mechanics and Materials Science

1989

rJ
an
Eff”

L-
.9 ABSTRACT
if)

FATIGUE OF Th02 DISPERSION STRENGTHENED NICKEL ALLOY

WITH SURFACE MODIFICATION BY Al+ ION IMPLANTATION

BY

Robin Banerjee

The present work was carried out to characterise the
effects of high-energy ion implantation on the fatigue
behavior of dispersion—strengthened (DS) nickel alloy.
Direct Al+ ion implantation was used to modify the surfaces
of thoria-dispersed nickel alloy specimens. The fatigue
behavior of both the unmodified and modified specimens was

- characterised in terms of cyclic stress-strain (CSS)
reSponse and by determination of low-cycle fatigue life in
stress controlled tests. The surface modified specimens
showed moderate increase in cyclic saturation stress as well
as slightly improved low-cycle fatigue life. The unmodified
specimens developed persistent slip bands (PSB) during
fatigue and cracks initiated at the PSB-matrix interface.
Surface modification by ion implantation was found to
suppress and delay the formation of persistent slip bands
and thus delayed crack initiation, resulting in higher

fatigue life.

ii

Dedicated to my parents

iii

ACKNOWLEDGEMENTS

I would like to express my sincere thanks and regards to
my advisor Dr. David S. Grummon for his continued support
and guidance throughout the length of this project. Also
thanks are due to Dr. L.T. Drzal and Mr. Michael Rich of
Composite Materials and Structures Center, Michigan State
University, and to Dr. Gary Was of Michigan Ion Beam
Laboratory, University of Michigan, for provision of various
facilities used in this work. Finally, special thanks are
due to Sherritt Gordon Mines Ltd. for providing the
material, and to the National Science Foundation for their

financial support under grant #MSM807774.

iv

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . .
CHAPTER I : INTRODUCTION . . . .

CHAPTER II : REVIEW OF THE LITERATURE

Cyclic Stress-Strain Response
of FCC Materials . . . . . . . .

Monotonic Deformation of
Dispersion Strengthened Materials .

Fatigue Behavior of
Dispersion Strengthened Materials .

The Effect of Surface Modification
on Mechanical Properties . . . . .

Ion Implantation Process
and Characteristics . . . . . . .

Effect of Ion Implantation on
Fatigue Behavior . . . . . . . . .
CHAPTER III : EXPERIMENTAL PROCEDURES

Material . . . . . . . . . . . .
Specimen Design . . . . . . . . .
Machining and Surface Preparation .
Ion Implantation . . . . . . . .
Post-Radiation Surface Analysis . .

Low Cycle Fatigue Testing . . . . .

Post-Fatigue Surface Characterisation .

vii

viii

12

18

21

23

35
38
4o
42
43
45

48

CHAPTER IV : RESULTS
Cyclic Strain Hardening Behavior
Cyclic Stress-Strain Behavior .
Low Cycle Fatigue Behavior .
Microscopic Observation of
the Specimens . . . . . . . .
CHAPTER V : DISCUSSION

Cyclic Deformation and
Plasticity Behavior . . . . .

Fatigue Crack Initiation
and Failure Mechanism . . . . .

Analysis of Fatigue Crack Growth

Low Cycle Fatigue Life Estimation

CHAPTER VI : CONCLUSIONS . . . . .
APPENDIX I . . . . . . . . . . . .

BIBLIOGRAPHY . . . . . . . . . . .

vi

Page

50
52

56

62

77

8O
84

86

96
101

104

LIST OF TABLES

Begs
Table 1 :

Chemical Composition of DS Nickel . . . . . 37

Table 2 :

Cyclic Stress-Strain Data for

Unmodified DS Nickel . . . . . . . . . . . 54
Table 3:

Cyclic Stress-Strain Data for

Modified DS Nickel . . . . . . . . . . . 55
Table 4:

Fatigue Life Data for

Unmodified DS Nickel . . . . . . . . . . . 57
Table 5:

Fatigue Life Data for

Modified DS Nickel . . . . . . . . . . . 59
Table 6:

Comparison of Fatigue Life of Pure Nickel

and DS Nickel with and without Aluminum
Ion Implantation . . . . . . . . . . . 61

vii

Figure l :

LIST OF FIGURES

Optical and Transmission Electron
Micrograph of DS Nickel . . . .

Figure 2 :

Specimen Designs for Fatigue Testing

Figure 3

Computed Atomic Concentration Profile

of Ion Implanted Specimens . . .

Figure 4

Cyclic Hardening Behavior
of DS Nickel . . . . . . .

Figure 5

Cyclic Stress-Strain Behavior
of DS Nickel . . . . . . . . .

Figure 6 :

Fatigue Life Performance
of DS Nickel . . . . . . .

Figure 7 :

Surface Features of Unmodified
DS Nickel, before and after Fatigue

Figure 8

Surface Features of Ion Implanted

DS Nickel,

before and after Fatigue

viii

Page

36

39

44

51

53

6O

66

67

Figure 9 :

Crack Tip Morphology in US Nickel

Figure 10 :
Slip Band Formation at the Edges of
Unmodified DS Nickel Specimens . . . .
Figure 11 :
Surface Morphology of a Slip Band
Formed in Unmodified DS Nickel
Figure 12 :
Fatigue Crack Initiation at the PSB-Matrix
Interface in Unmodified DS Nickel . .
Figure 13 :
Fatigue Crack Formation in Unmodified
DS Nickel, without Appreciable
Presence of Slip Bands . . .
Figure 14 :
Persistent Slip Band Formation and
Fatigue Crack Initiation at the Edges
of Modified DS Nickel Specimen .
Figure 15 :
Fatigue Crack Initiated at the Edges
of Modified DS Nickel Specimen
Figure 16 :
Fracture Surface of Unmodified

DS Nickel, Showing Microcracks
in the Matrix . . . . . . . .

ix

Page

68

69

7O

71

72

73

74

75

HAP ER
INTRODUCTION

In recent years, high-energy ion-implantation has been
succesfully used to improve surface-sensitive properties
e.g, wear resistance and corrosion resistance of various
materials. This process is now being investigated as a
viable means to improve the fatigue performance of
structural materials. Different combinations of implant and
substrate materials e.g, Al implantation in Cu [Kujore,
1980], C implantation in Ni [Sleeswyk, 1980], Ne
implantation in Cu [Bakhru, 1981], C and N implantation in
Ti-6Al-4V alloy [Vardiman, 1982], A1 implantation in Ni
[Ahmed,1985; Grummon,1987; Jones,1987] have all resulted in
improved fatigue life of the substrate materials. Various
other combinations have also been found to produce fatigue
life improvement [Mendez,1982; Jata,1983; Hohmuth,1985].
While ion implantation, in general, leads to higher fatigue
life of target materials, it has been found in some
instances e.g, C implantation in Cu and stainless steel
[Sleeswyk,l980], to decrease the fatigue life. Generally,
the increase in the fatigue life by ion implantation is
attributed to the suppression or reduction of the slip band

formation at the surface of the fatigued specimen.

It is now generally established that the fatigue failure

1

of single crystals and polycrystals of pure metals and
alloys follow a general pattern which can be summarised as:
plastic strain localization in slip bands, followed by slip
band emergence at the free surface, causing surface damage
accumulation, and finally initiating fatigue crack, which
propagates to cause catastrophic failure. With ion beam
surface modification, a hard surface layer can be formed
which can resist and delay the emergence of slip bands

through the surface and thereby increasing the fatigue life.

There has been considerable research work in recent years
on ion-implantation to improve the fatigue performance of
various materials. Studies have been done on various model
systems e.g, single crystals and polycrystals of pure metals
as well as other engineering alloys. However, so far there
has not been any study to characterise the effects of ion
beam surface modifications on the fatigue behavior of
dispersion strengthened alloys. These alloys have unique
cyclic plasticity behavior which are very different than
that of pure metals and alloys. In these alloys, the
presence of fine dispersions of hard particles in the matrix
causes extensive cross-slip in their vicinity. Consequently,
long range slip localization is prevented, resulting in a
more homogeneous slip distribution [Leverant,1968]. The
present study is intended to investigate how surface
modification by ion implantation can affect the fatigue

behavior of such materials.

For the present work, thoria-dispersed nickel (BS-Nickel)
has been chosen as the base material and aluminum ion
implantation has been used to modify its surface. The choice
of DS-Nickel as the base material has been motivated by the
fact that recently extensive work has been done on the
effect of aluminum ion implantation on the fatigue behavior
of pure Nickel [Ahmed,1985; Grummon,1987]. The cyclic
plasticity behavior of pure nickel with aluminum ion
implanted surface layer has been well characterised.
Subjecting DS-nickel to similar experimental conditions, it
would be possible to know how this matrix, having a
different cyclic plasticity behavior, interacts with the
ion-implanted layer and how that affects the fatigue
behavior of the material. Furthermore, DS-Nickel is a
representative of a whole family of dispersion strengthened
alloys and thus the knowledge gained from the study of DS-
Nickel can also be applied to other materials belonging to

this important class of materials.

QHAPTER II

REVIEW OF THE LITERATURE

Over the past few decades, considerable research work has
been done on cyclic deformation of fcc materials. While most
of these studies were done on model systems including pure
metals and alloys, there has been some studies on the
monotonic as well as cyclic deformation behavior of
dispersion strengthened alloys. In this section, some of
those studies will be reviewed which are relevant to the
present study. Also, there has been substantial research in
the application of ion-beam surface modification to improve
the cyclic deformation behavior of various materials. Later,

in this section, some of those works will be reviewed.

919119 §tress-8train Response of FCC Materials

When a material is subjected to cyclic loading above the
"fatigue limit", it fails after a finite number of load
cycles. This type of failure mode is known as fatigue
failure. Ewing and Humphrey (1902) first obtained clear
microscopic evidence of slip on the surface of the fatigued
specimens and could relate the mechanism of fatigue to
plastic deformation of the material. Today, it is fairly
well established that the stress amplitudes that cause
fatigue failure, although for most cases too small to cause

a

any "macro-yielding", are large enough to cause cyclic

4 to 10-5. These

"micro-plastic" strains of the order of 10'
small strains accumulated over a large number of cycles can
add up to a very high localized accumulated strain and can
thus initiate fatigue cracks, which then propagate to cause

failure.

The relation between the cyclic plastic strain amplitude

e and the fatigue life N was first formulated by Coffin

p1 f
and Manson (1954) by the Coffin-Manson law which states that

N = k ( constant ) ..... (1)

f'epl

For most pure fcc metals and alloys, when cycled within a

4

critical range of plastic strain of the order of 10- to

10-3, slip becomes localized into "Persistent Slip Bands" or
P883. P885 correspond to lamellae of high local glide
activity. They usually lie parallel to the primary glide
plane and can extend deep into the bulk (Helgeland, Roberts
et al,1965) and have typical thickness of about 1 p in pure
fcc metals (Mughrabi,1978). Broom and Ham (1959) and Finney
and Laird (1975) showed that at intermediate cyclic strain
amplitudes, slip is essentially confined to the P885. Slip
within the P88 is to some extent irreversible which means
that the slip during the forward and reverse cycle do not
necessarily take place in the same atomic planes (Watt,

Embury & Ham 1968, Finney & Laird 1975 ). This irreversible

nature of the slip within the P883 leads to the formation of

slip-steps at the surface and the slip traces develop a
marked relief ( notch-peak geometry ) which serves as the

preferential site for fatigue crack nucleation.

The formation of persistent slip bands is most likely to
occur where a single slip system is active and PSBs form
easily in single crystals. However PSBs have been observed
in polycrystals also, but only in grains that are oriented
for single-slip and at low strain amplitudes where single
slip is favoured over multiple slip. From the viewpoint of
fatigue crack initiation, the net extrusion of the P83 as a
whole is identified as a type I stress raiser, while the
notch-peak topography of the P83 surface as type II stress
raiser. The shear cracks are likely to initiate early at the
PSB-matrix interface due to type I stress raisers, which are
known to form early. At a later stage, they are likely to
initiate at the protruding surface of the PSB, where the
type II stress raisers cause severe stress concentration

(Mughrabi & Wang,1982).

Monotonic Deformation of Dispersion-Strengthened Materiel;

Extensive research work had been done on the monotonic
tensile behavior of dispersion-strengthened and
precipitation-strengthened alloys. Ebeling and Ashby (1966),
Jones and Kelly (1968-69) and Hirsch and Humphreys (1970)
had done extensive work on the tensile behaviour of

internally-oxidised copper alloys. Most of this work have

been directed to explain the yielding of dispersion-
strengthened alloys in terms of "Orowan mechanism".
According to Orowan’s mechanism, the yield stress of the
alloy is inversely proportional to the particle spacing by

the following relation ( Orowan, 1948),

ay=am+2T/bD ...... (2)

where ay is the alloy yield stress, am is the matrix
yield stress in shear, T is the dislocation line tension, b
is the Burger’s vector and D is the particle spacing. The
particle spacing D is defined as ( 1 / J NS) where N8 is the
number of particles intersecting unit area of the glide
plane. So, for a given volume percent of the dispersoid, the
finer the dispersion, the higher is the yield stress.
However, there is a critical size for the dispersoids, below
which the shearing or fracture of the dispersoids is
favoured over gross yielding of the matrix. In fact, Orowan
mechanism suggests that there is a critical size for the
particles, below which the particles are sheared off and
above which they are bypassed by the dislocations. Kupcis,
Ramaswami and Woo (1973), also working on internally
oxidised copper alloys, showed that when non-coherent
alumina particles were greater than 40 nm in size, the
crystal yielded at a stress suggested by Orowan mechanism,
indicating the by-passing of the particles by the
dislocations. However, when the particles were less than 15

nm in size, the material showed much lower yield strength

than required by Orowan mechanism, suggesting the shearing

of the coherent particles by dislocations.

Ansell and Lenel (1960) proposed that the yielding of
dispersion strengthened alloy occurs when the shear stress
on the particle becomes equal to either yield or fracture
stress of the particle. According to this model, the yield

stress of the alloy is given by,

0y = J ( pbp* / 2aC ) ..... (3)

where p and p* are the shear moduli of the matrix and the
particles respectively, b is the Burger’s vector, u is the
particle spacing and C is a constant. When the shear stress
due to the pile-up of dislocation loops around the particles
becomes high enough to deform or fracture the particles,
gross yielding occurs, relieving the back stress on the

dislocation sources.

The early model of yielding of a dispersion strengthened
alloy as proposed by Orowan (1948) was modified by Friedel
(1964) and Kelly and Nicholson (1964). According to them,
The critical stress required to push a dislocation between
two non-deforming particles of distance D was given by the

modified form,

a = A ( Gb / 2nD ) ln ( D / b ) ...... (4)

where, A = [ 1 + (a / l-a) Sinzo ]

Here a is the critical shear stress, G is the matrix
shear modulus, b is the Burger’s vector, a is the Poisson's
ratio and 0 is the angle between the Burger’s vector and the
dislocation line with no stress applied. ( So, a = 0 for a

screw dislocation; and 0 = 90 for an edge dislocation ).

Ashby (1966) pointed out some of the discrepancies of the
above treatment by Kelly & Nicholson. In his analysis, Ashby
took under consideration two factors which were neglected by
earlier workers e.g, a) the Orowan stress can be reduced by
the mutual interaction of two bowed-out segments and b) the
dislocation may dissociate into partials and thereby
reducing its line tension and energy. The result of his

analysis was a further modification of equation (4).

The fracture mechanisms of internally oxidised Copper
alloys containing a uniform dispersion of very fine ( 50-70
nm) oxide particles, was studied by Palmer and Smith (1966).
They investigated the fracture of single-crystals and
polycrystals under monotonic tensile deformation and
observed that the failure was initiated at the particle-
matrix interface, due to void nucleation on the oxide
particles. Fracture occurs by void growth and coalescence,
with extensive coalescence being confined to the plane of
fracture. They also observed that as far as void nucleation

is concerned, the oxide particles in the grain boundaries

10

behave in much the same way as those in the grain interior.

The formation of dislocation substructures in TD-Nickel
under monotonic tensile deformation was investigated by
Heimendahl and Thomas (1964). Upon observing the
microstructure of plastically deformed specimens, they
reported that the dislocations are pinned and tangled around
the thoria particles. The dislocation loops around the
particles were often elongated, indicating the avoidance of
the particle by cross slip. Dislocation loops were often
prismatic and not concentric as proposed by Fisher et. a1.
(1953). After large deformation (2 10%), the dislocations
tended to form a cellular structure whose size was governed
by the spacing of the thoria particles. The dislocation cell
structure in TD-Nickel was not as well developed as in pure
nickel and the size of the cells in TD-nickel was about five
times smaller. However, after very high deformation (z 90%),
the dislocation cell structure in TD-Nickel was well

developed.

Inman, Zwilsky and Boone (1964) investigated the
microstructural features of cold-rolled and recrystallised
TD-Nickel to gain insight on the strengthening effect of the
fine second phase particles. The microstructure of heavily
deformed TD-Nickel consists of elongated cells or fibres and
it shows a strong texture of (110)[001]. The
recrystallisation of TD-Nickel occurs at about the same

temperature range as that of pure nickel. The recrystallised

11

TD-Nickel shows a primary texture of (100)[001] with a
secondary texture of (110)[112]. The recrystallised TD-
Nickel showed a large number of annealing twins. The mean

particle size in TD-Nickel was about 50 nm.

Presence of a large number of narrow annealing twins was
also reported by Heimendahl and Thomas, who investigated the
microstructural feature of annealed TD-Nickel. The twin
boundaries were very irregular and thus indicated that the
thoria particles resist twin growth. Twin boundaries act as
barriers to dislocation motion and thus contribute to its
strengthening mechanism. Also, the thoria particles were
never seen to be intersected or fractured by the
dislocations. The yield stress of TD-Nickel appeared to be
much higher than that predicted, on the basis of particle

size and spacing, by the extended Orowan equation,

oy=as+(Gb¢/4x)ln[(d-2r)/2b][2/(d-2r)]
..... (5)
where G is the shear modulus of matrix, b is the Burger’s
vector, r is the mean particle radius, d is mean planar
interparticle spacing, as is the yield stress of
matrix, ¢ = 0.5( 1 + 1 /(1-a)}, and a is Poisson's ratio.

The Orowan equation however is only applicable to single
crystals. Polycrystals show much higher yield stress and
this increase in yield stress is due to the additional

contribution from grain boundaries, twin boundaries and

12

texture (Heimendahl and Thomas, 1964).

The various works discussed so far, dealt with the
monotonic deformation of dispersion strengthened alloys. The
mechanical behavior of these alloys under monotonic
deformation has been well characterised both in the macro
and micro-level. However, for the present study, it is more
important to review the studies that dealt with the cyclic
deformation and fatigue behavior of dispersion strengthened

alloys.

Fatigue Behavior of Dispersion-Strengthened Materials

The cyclic deformation behavior of dispersion
strengthened alloys differs significantly from that of pure
metals and alloys. The hard particles in the matrix are
known to cause the plastic strain homogenisation and thereby
resisting persistent slip band formation. As a result,
fatigue crack initiation is much more difficult in these
alloys. Several studies has been done on the cyclic
deformation behavior of dispersion strengthened alloys and

some of the relevant ones are reviewed here.

Wells and Sullivan (1964) characterised the low-cycle
fatigue behavior of a Nickel-base superalloy ( Udimet 700 )
whose microstructure contains random distribution of hard

second-phase particles. The initial cyclic hardening of this

13

material is followed by cyclic softening which is due to the
fracture of the precipitates. Intense slip band formation is
observed and slip-band cracking along with twin-boundary
cracking are believed to be the dominant mode of crack

initiation.

Abel and Ham (1966) studied the cyclic strain behaviour
of precipitation-strengthened aluminum alloys and pointed
out that the Bauschinger effect was extremely high in those
systems when the precipitates were not coherent. They
explained it in terms of internal stresses being built up at
the particles during forward flow. They also reported very
fast cyclic hardening, sometimes followed by softening, when
the precipitates were not coherent. Occasional cracking at
the particles were also reported. The rapid strain hardening
behavior of dispersion strengthened alloys are generally
attributed to rapid multiplication of dislocations due to
multiple slip systems. This effect is further aided by the
back stresses developed by the dislocation loops at the

particles.

Leverant (1967) investigated the fatigue behaviour of
thoria-dispersed nickel to find out whether the fine
dispersion of hard particles can induce irreversibility in
the cyclic plasticity behaviour of a wavy-slip material by
stabilizing the initial microstructure. Based on the results
obtained, he concluded that the presence of these fine, hard

particles did not affect the reversible cyclic plasticity

14

behaviour of a "wavy-slip" material at room temperature.

The presence of a fine dispersion of hard second-phase
particles in an otherwise pure matrix can significantly
alter the fatigue behaviour of fcc materials. The most
significant change may take place in the extent of cyclic
plastic strain localization during fatigue. Leverant and
Sullivan (1968) studied the effect of a fine dispersion of
Thoria particles on the fatigue behaviour of pure nickel.
They observed that the extent of surface slip was greatly
reduced by the presence of thoria particles and the slip was
more wavy and heterogeneous as compared to that in pure

nickel.

The reduction in the extent of surface slip and the
development of much finer slip can be attributed to the fact
that the fine, hard particles, uniformly dispersed over the
matrix, offer resistance to slip and cause extensive cross-
slip in their vicinity. As a result, the localization of
slip over large distances is not favoured, thus giving rise
to a fairly homogeneous distribution of slip throughout the
matrix. Furthermore, since crack initiation is greatly
influenced by the extent of strain localization, it follows
that fatigue crack initiation will be more difficult due to
the presence of these fine particles (Leverant & Sullivan,

1968).

Another important feature due to the presence of these

15

particles is the effect on the dislocation substructure at
saturation. Due to extensive cross-slip in the vicinity of
the particles, dislocation cell-walls are formed, with the
particles being included in the cell walls. The size of
these cells is independent of the plastic strain range and
is controlled by the size and distribution of the second-
phase particles (Leverant and Sullivan, 1968). The formation
of this dislocation substructure and the dependence of their
size on the distribution of second-phase particles is also
evident in the monotonic deformation behavior of Thoria
Dispersed (TD) Nickel (Heimendahl and Thomas, 1964) and
other dispersion strengthened alloys ( Kimura,1965:

Ashby,1964 ).

Further work by Mori and Mura (1976) on the slip
morphology of dispersion strengthened materials showed that
although in general the plastic deformation in these alloys
is homogeneous, heterogeneous plastic deformation sometimes
does occur in a certain range of plastic strain if the alloy
contains softening particles, as in quench-hardened aluminum

or neutron irradiated copper.

The cyclic deformation behavior of a precipitation
hardened Al-Ag alloy was investigated by Laird et. a1.
(1978). They observed that the alloys containing Guinier-
Preston (GP) zones did not cyclically soften. This supports
the "disordering" mechanism of cyclic softening of alloys

containing ordered precipitates, since GP zones are

16

inherently disordered. They also observed the formation of

cyclic strain induced stable precipitates.

Wilhelm (1980), investigating the cyclic plasticity
behaviour of age-hardened copper and aluminum alloys,
reported that below a threshold plastic strain range of the
order of 0.0001 - 0.0002, the alloys cyclically hardened to
a saturation stress. No PSBs are formed and the coherent
precipitates are not destroyed by the gliding dislocations.
However, when cycled at higher plastic strain, PSBs do form
at a critical stress and the material shows cyclic
softening. The cyclic softening is attributed to the
destruction of shearable precipitates within the P883 by the

gliding dislocations.

Vogel, Wilhelm and Gerold (1981) investigated the
development of the microstructure during the cyclic
deformation of peak aged Al-Zn-Mg single crystals. When

cycled under constant plastic strain (z 10-3

), slip bands
are formed initially which grow in volume and also in
dislocation density. At a critical dislocation density, the
slip bands collapse into a secondary slip band structure
which permits higher dislocation density. The precipitates
are sheared and destroyed within these bands and finally
narrow PSBs are formed which are devoid of any particles.

The material within the P88 remains in a supersaturated

solid solution state.

17

Steiner and Gerald (1986) investigated the fatigue
behavior of single crystals of Cu-2at.%Co alloy. The general
sequence was found to be Cyclic hardening, the formation of
P888 and cyclic softening. The cyclic softening had been
attributed to the dissolution of the precipitates within the

PSBs.

To summarize, it has been observed that the presence of
fine, hard particles or precipitates in the matrix,
substantially reduces the extent of plastic strain
localization in PSBs. This results in a more homogeneous
distribution of plastic strain over the matrix, thereby
hindering the fatigue crack initiation. The stability of the
particles or the precipitates under cyclic deformation are
critical for the improved fatigue performance of dispersion-
strengthened alloys and precipitation-hardened alloys as
well. If the second phase particles or precipitates are
fractured or dissolved in the matrix during cyclic
deformation, the alloy shows a cyclic softening with the
subsequent formation of PSBs and the overall fatigue

performance is adversely affected.

e t u fao odi a o n echanioal P o e

Substantial amount of research work has been carried out
in recent years to characterise the effects of various
surface modification techniques on the mechanical behaviour
of materials. It has been shown by various researchers that
the mechanical deformation behavior of the materials can be
influenced by modifying the state of the surface. This is
associated with the structural changes in the crystal
lattice in the surface and sub-surface regions and the
corresponding dislocation dynamics in these regions. Some

fundamental studies in this aspect are reviewed here.

Initial work by Koehler (1941) showed that an image force
is active on a screw dislocation present near an interface
between two materials having different elastic moduli. If
the dislocation lies in the material of higher elastic
modulus, it is attracted towards the interface under the
image force, whereas the dislocation is repelled if it is in
the material of lower elastic modulus. Head (1953), working
with a surface of higher elastic modulus than the bulk,
concluded that for a thin, hard surface layer, the screw
dislocations will assume an equilibrium position near the
interface where the image force acting on the dislocation is
at its minimum. Further work on the image force on the screw
as well as edge dislocations were carried out by Chou (1966)
& Conners (1967), for a more generalized combination of the

18

19

surface and the bulk elastic moduli as well as the surface
layer thickness. They showed that for a particular
combination of the surface layer and the bulk material, the
thickness of the surface layer is critically important,
because it determines the magnitude and the direction of the
image forces in the surface layer and the bulk. Greenfield
(1973) showed that the principal hardening mechanism for a
thin diffused surface layer is by means of an accomodation
dislocation network. He also concluded that this
accomodation dislocation network is greatly influenced by
the nature of the image forces in the surface layer and in

the bulk.

DeJonghe and Greenfield (1969) reported a considerable
increase in the critical resolved shear stress (CRSS) of
copper single crystal having platinum diffused surface
layers. The surface hardening was attributed to an
accomodation network of dislocations. A later work by
Greenfield (1973) was reported to have increased the CRSS of
copper single crystal by diffusing surface layer of
platinum, gold, silver and nickel as well. Again the surface
hardening was attributed to the accomodation dislocation

network.

A detailed study was undertaken by Greenfield and Purohit
(1980) to classify the effects of surface modification into
some broad categories. They worked with both discrete and

diffused surface layers and analysed the results in terms of

20

image forces, dislocation initiation and accomodation
networks. The main variables of the experiments were the
elastic modulii, yield stress and yield strain of both the
surface and the bulk for various systems. One particular
combination of interest is when the surface has lower
elastic modulus but higher yield stress than the bulk. In
this case, during cyclic loading, the stress at the surface
will initially be lower than the bulk and also the yielding
of the surface is less likely. In a situation like this, it
is possible, for a specified range of applied stress, that
the surface layer cycles elastically while the bulk sustains
cyclic micro-plastic strain. Even though the bulk material
accumulates plastic strain, the surface layer does not incur
any damage. However, when bulk plastic strain accumulates
to a critical value, the surface layer will rupture, leading
to permanent surface damage. This process would require
considerable period of time and would be manifest in an

improvement of fatigue life.

In a recent study by Meyers and Grummon (1988), finite
element method has been used to model the micromechanics of
the interaction between the ion implanted surface layer and
the underlying persistent slip band. The surface film in Al+
ion beam modified nickel substrate was found to rupture as a
result of high stress concentration at the film due to the
slip activities in the underlying subsurface slip band.
However, the rupture of the surface film does not occur

until a considerable amount of plastic strain is accumulated

21

and this results in a delay in the fatigue crack initiation

of ion beam modified specimens, causing higher fatigue life.

In recent years, ion-implantation has emerged to be a
very versatile process for surface modification.
Considerable amount of research work has been done and is
still being conducted on the beneficial effect of ion
implantation to improve the mechanical, chemical and
environmental properties of a wide range of materials. We
will discuss some of the salient features of ion
implantation process and the potential use of this process

to improve the fatigue properties of various materials.

Ten-Implantation Process and Characteristics

 

Ion-Implantation is a process where ions of an atomic
species to be implanted are generated and accelerated
through very high voltages and are made to impinge onto the
target. The process must be carried out in vacuum (~ 10"6
torr.), otherwise the ion beam will be scattered and

absorbed over a Short distance and also the surface

chemistry of the target cannot be controlled.

The main parameters that characterise the ion
implantation process are:

a) the ion species to be implanted;

b) the target species ;

c) kinetic energy of the ion beam;

22

d) the number of ions per unit area of target (fluence);
e) the beam current 3
f) the total implantation time ;

g) the target temperature during ion implantation.

Some of the salient features of ion-implantation process
that make it especially attractive as a tool for tailor-made
surface engineering are as follows ,

1) The process is extremely versatile. Almost any atomic
species can be implanted into a broad class of materials
including metals, ceramics and polymers.

2) The process can be carried out at room temperature
without causing any major changes in the bulk properties.

3) Since the process is not entirely governed by the
thermodynamic criteria, many non-equilibrium phases can be
produced. Also solid solutions beyond the thermodynamic
solubility limit can be created by this process.

However there are some disadvantages to this process of
which the most important ones are the line-of—sight

limitation and high cost.

To gain an insight into the kinetics of the ion
implantation process, let us look at the physical processes
that take place during ion implantation. As the energised
ion impinges onto the target, it is rapidly slowed down as
it continually loses its kinetic energy due to interactions
with the electrons of the target atoms. Also it transmits

quanta of energy whenever it collides with the target nuclei

23

and as a result of that, a number of target atoms are
knocked off their lattice sites in a cascading manner. Due
to a large number of incident ions, random movement of ions
occur in a cascading manner, leading to random mixing of the
ions. In fact, ion-beam mixing is one of the commonly used
method for surface modification. This mixing process can be
compared to the thermal diffusion process, however it does
not have any temperature dependence, instead the extent of
mixing is dependent on the implanted species, substrate and
beam energy. The energy transfer process between the
incident ion and the target atoms also dictates the
deposited ion concentration profile. Usually the implanted
species assume a gaussian distribution near the surface of
the target material. The depth corresponding to the peak of
the Gaussian distribution is called the "Range". The range
is again a function of the implanted species and beam
energy. Due to the surface expansion effect as well as
sputtering of target atoms, the resulting concentration
profile deviates from an ideal gaussian distribution and in
fact, the experimental data fit more closely to "Pearson IV"
distribution. Nowadays, computer codes (e.g, "Profile" from
Implant Sciences, Danvers, MA) are available to compute the
concentration profile of the implanted ions given the

process parameters.

Effect gt Ion Implantation on Fatigue Behavior

As mentioned earlier in this section, there is growing

24

evidence that the mechanical deformation behavior of the
material can be influenced by altering the surface layer. It
is now generally established that fatigue cracks usually
originate and propagate from the free surface of the
specimen. Thus the surface condition plays a major role in
the fatigue performance of the specimen. If proper surface
modification is employed, it is possible to manipulate the
deformation behavior in the surface as well as sub-surface
regions in such a way so as to prohibit fatigue crack
initiation at the free surface. Since ion implantation has
emerged in recent times to be a very flexible tool for
surface modification, considerable research has been carried
out to modify the surface of various materials using ion

implantation, in order to improve their fatigue performance.

One of the earlier studies in this area was undertaken by
Kujore, Chakraborty, Starke and Legg (1980), who studied the
effect of aluminum ion implantation on the fatigue crack
initiation of pure polycrystalline copper. The fatigue life
of copper was found to increase, after Al+ ion implantation,
by about 35% in high cycle fatigue under stress control.
Upon closer examination of the surface topography, they
observed that the extensive formation of coarse P883 in
unimplanted copper is replaced by a less severe formation of
finer PSBs in implanted specimens. They argued that the
increse in fatigue life was due to the relatively
homogeneous slip and finer slip bands in the implanted

species. It was postulated that the introduction of aluminum

25

in copper resulted in lower stacking-fault energy of the
surface layer and that caused a decrease in the extent of
cross slip, thus promoting slip reversibility. Also, since
Al+ ions are larger than Cu+ ions, a compressive residual
stress pattern may have been generated on the surface, which
would resist the formation of surface crack and thus help to

increase the fatigue life.

Similar work was carried out by Sleeswyk, Kok and Boom
(1980) to investigate the effect of C+ ion implantation on
the low-cycle fatigue life of copper, nickel and stainless
steel polycrystals. The specimen temperature during ion-
implantation rose to 100 'C. On Ion Implantation, the
fatigue life of nickel was increased by about 10%, but
surprisingly the fatigue life of copper and stainless steel
was shortened by about 20%. This was consistent with the
observation that the slip lines in implanted copper and
stainless steel were much coarser than in unimplanted
species. Although the authors could not give any
satisfactory explanation of this rather unusual decrease in
fatigue life of copper and stainless steel, they suspected
that a high concentration of dislocations underneath the
surface layer might cause micro-cracking parallel to the
surface. Nickel, on the other hand showed an increase in
fatigue life, which was associated with the formation of

finer slip bands and lower stress concentration.

Bakhru et. a1. (1981) studied the effect of Ne+ and N+ ion

26

implantation on bending fatigue performance of
polycrystalline copper and austenitic stainless steel
respectively. While Ne+ implanted copper showed a
significant increase in fatigue life, no such improvement
was evident in N+ implanted stainless steel. The surface of
the unimplanted stainless steel showed profuse slip band
formation, but after ion implantation, it showed fewer but
more sharply defined slip bands. One significant observation
was the presence of kinks in the slip bands, which indicated
the tendency of cross-slip. The occurence of cross-slip in
stainless steel is usually rare, because of its relatively
low stacking fault energy. The authors speculated that the
presence of large concentration of N atoms in implanted
layer, which acts as obstacles to dislocation glide,
necessitates higher shear stress in the slip bands, which
may then cause cross-slip. According to the authors, the
beneficial effects of lattice hardening and compressive
surface stresses due to ion implantation were offset by the
detrimental effects of radiation damage and occurrence of
cross-slip, resulting in no significant improvement in
fatigue life of stainless steel. Copper, on the other hand,
after implantation, showed almost no slip bands on the
surface and thus explains the remarkable improvement in

fatigue life.

Vardiman and Kant (1981) investigated the effect of C+
and N+ ion implantation on the fatigue life of Ti-6Al-4V

alloy. They observed a 10% and 20% increase in the endurance

27

limit by N+ and C+ implantation respectively. From the
fractographic analysis, it was found that for specimens with
a lifetime less than 105 cycles, the fatigue crack was
initiated at the surface. However, for specimens with a
lifetime greater than 105 cycles, the crack showed a
subsurface origin. The subsurface origins occured at a depth
of 25-150 u. The authors did not give any satisfactory
explanation of the observed increase in fatigue life by ion
implantation. Also, the sub-surface initiation of the

fatigue crack was not fully explained.

Another investigation was done by Mendez, Violan and
Villain (1982), to study the fatigue behavior of copper
polycrystals with ion implantation of various elements e.g,
Cu, Al, O and N ions. The most significant observation was
that the N and 0 ion implantation lead to severe inhibition
of surface slip bands, while this effect was not so severe
with Cu and Al ion implantation. Based on their findings and
some earlier reports, the authors generalised that the
implantation of gaseous elements is almost always associated
with the inhibition of slip bands. They also stated that the
implantation of gaseous elements often lead to the formation
of gas bubbles in the implanted layers, which act as
dislocation sinks, thus preventing the egress of

dislocations on the surface.

Extensive work was done by Jata and Starke (1983) to

characterize the effects of ion implantation on the fatigue

28

performance of a wide range of materials including copper,
aluminum, titanium and steel. They identified the formation
of several metastable and stable phases in each of the
substrates during implantation of various species. They
reported fatigue life improvement in Cu implanted with Al+,
Cr+ and 8+ and also in Ti-24V alloy implanted with Al+. This
improvement in fatigue life was associated with much finer
and homogeneous slip at the surface in the implanted
species. In general, transgranular cracks were observed in
unimplanted species, while implanted species showed
intergranular cracks initiating at the interface of
persistent slip bands (P88) and grain boundaries. The
result of fatigue tests on self-implanted copper did not
show any appreciable improvement, suggesting that the
lattice damage caused by ion implantation, although quite
severe at times, does not significantly contribute to
fatigue life improvement. However, ion implantation did not
always produce positive results. In fact, a/fl Ti implanted
with Al+ and 4140 steel implanted with N actually showed a
decrease in life in strain-controlled fatigue tests.
However, both of these materials showed considerable
improvement in life under stress-controlled fatigue tests.
Also they pointed out that a residual compressive stress may
be present at the surface when implanted with larger atoms.
To utilize this compressive state of stress at the surface,
one has to be careful not to overload the specimen, which
would destroy the beneficial effect of the residual

compressive stress. It must be noted, however that the

29

beneficial effect of a residual compressive stress at the
surface cannot be easily isolated from the effect of
metastable and stable phase formation during ion

implantation.

Hohmuth, Richter, Rauschenbach and Blochwitz (1985)
studied the effect of 8+ and N+ ion implantation on fatigue
of polycrystalline nickel. They found that the extent of
surface slip bands were much lower in the implanted species.
With increasing implantation dose, slip band density
decreased, however the fatigue lifetime is maximum at an

17 ions /cm2. This is

optimum implantation dose of 5 x 10
probably because at very high dose, the complete prevention
of PSB formation may lead to very high stress
incompatibility between the grains, leading to grain
boundary cracking. Another interesting point to note was the
shift of the composition profile towards the surface after

fatigue. The authors suggest that the cyclic deformation

causes boron and nitrogen atoms to diffuse to the surface.

It has long been known that ion implantation can lead to
metastable phase formation in near-surface region which
plays important role in modifying the surface sensitive
properties of the substrate. A very comprehensive study was
undertaken by Follstaedt (1985) to characterise different
metastable phases formed in various material systems. Of
particular interest to us is the fcc -> hcp transformation

of pure Ni subject to neutron irradiation or implantation

30

with Ar, He, P, Sb or even Ni. It is theorised that the
stress accompanying ion implantation is mainly responsible
for this phase transformation. Of even more interest is the
formation of amorphous phases during metalloid ion
implantation in metals. Rauschenbach et al. (1982~83) had
already applied Hagg’s embedment criterion (Hagg, 1931) to
predict the formation of amorphous or crystalline phase in a
particular material-implant system. According to the Hagg’s
rule, if the ratio (R) of the metalloid atom size to that of
the metal atom is greater than 0.59, amorphous phases appear
while R < 0.59 ensures crystalline phase formation. An
interesting model system which has been fairly well
characterised and which is of our interest too, is the Ni-Al
system. For the Al-Ni+ system, an amorphous phase nucleates
in fee Al and grows until a single amorphous phase is found
at 25 at.% Ni. At even higher Ni content, crystalline AlNi
phase forms within a matrix of the amorphous phase. For Ni-
Al+ system, we start with fcc Ni and go through an
intermediate phase NixAl, having 25-33 at.% Al and finally
get similar phases i.e, the amorphous phase ( 25 at% Ni )
and AlNi at higher Al content. It is interesting to note
that neither A13Ni and A13Ni2, which are the equilibrium

phases, are found in any of these experiments.

Another contemporary study in this area was done by Ahmed
and Potter (1985) who studied the microscopic processes

accompanying A1 ion implantation in Nickel. For specimens

17

implanted to a fluence of 6 x 10 ions/cmz, the electron

31

diffraction pattern indicated the formation of the fcc (1)
solid solution and the solubility limit of aluminum in
nickel was increased to about 30 at.% as compared to 10 at%
at equilibrium. The 1’ (Ni3Al) phase does not form during
implantation, instead an hcp phase is observed. Above 35 at%
Al, this hcp phase is replaced by the ordered p’-NiAl phase.
At even higher Al content ( z 60 at% ), an amorphous phase

is observed along with the fi’ phase.

A very detailed study was done by Wang et. al. (1985) to
characterise the surface of ion-implanted and fatigued
metals by transmission microscopy. They mainly investigated
the surface of 304 stainless steel and copper implanted with
N or Ne and proposed a model to explain the surface damage
mechanism in the implanted layer. According to their model,
a large number of uniformly distributed lattice defects,
usually Frank loops, are formed after implantation. These
defects resist dislocation glide. However, once penetrated
by the glide dislocations, these loops are annihilated. As a
result, there is little or poor resistance to dislocation
glide and eventually dislocation channels are formed. These
dislocation channels are similar to the PSBs formed in
unimplanted metals. Cyclic slip is concentrated in these
channels and eventually cause surface extrusion/intrusion

damage, leading to fatigue crack initiation.

Vardiman and Cox (1985) did another study to improve the

fatigue life of a Ni-Cr alloy using C ion implantation. They

32

reported about 17% improvement in fatigue performance which
was again attributed to the suppression of surface slip by
ion implantation. Solid solution hardening and compressive
stress at the surface were believed to be the main factors
behind reduced surface slip. The slip band cracking in
unimplanted material was replaced by, as expected, grain

boundary cracking in the implanted material.

Another recent study was done by Mendez, Violan and
Denanot (1987) who investigated the effect of nitrogen ion
implantation on the fatigue behaviour of various fcc
materials including copper, aluminum and stainless steel.
There was no significant improvement in the fatigue life of
2017 aluminum alloy or 316L stainless steel after ion
implantation, however significant improvement was reported
for N-implanted pure copper. For 2017 aluminum alloy, the
cracks were reportedly nucleated due to the cracking or
debonding of the inclusions and thus quite obviously ion
implantation has little effect on fatigue life. For 316L
stainless steel, although N-implantation drastically reduced
the extent of PSB formation, cyclic slip is so highly
localised that the implanted layer is believed to be sheared
and there was no improvement in fatigue life. In pure
copper, however, the reduction in PSB formation due to N-
implantation does result in 50% increase in fatigue life.
Also, they reported no influence of ion-implantation on the

cyclic hardening rate.

33

More recently, Grumman et. a1. (1987) have characterised
the surface morphologies of ion beam mixed surface layer in
pure nickel and correlated them with the mechanical
behaviour under cyclic loading. They evaporated alternate
layers of Ni and Al on a pure Ni substrate and ballistically
mixed them with a high-energy Kr+ beam. The surface
modification resulted in a substantial decrease in the
number and intensity of surface slip band features. The
morphology of the Persistent Slip Bands (P88) in the
modified specimens were different than those in the
unmodified specimens. The PSBs in the modified species
mostly showed a single slip step as opposed to the typical
notch-peak topography that was observed in the P888 in the
unmodied specimens. Subsurface plastic strain localization
took place in the implanted species within the P888 which
extended upto the implanted layer. However, the implanted
layer finally ruptured under the high local stress from the

impinging PSBs, but only at much higher accumulated strain.

As an extension of the latter work, the present research
was undertaken to study the effect of ion implantation on
the fatigue behaviour of DS-Nickel, which has a very fine
dispersion of inert thoria particles in an otherwise pure
nickel matrix. There have been several studies done on the
effect of ion-implantation on fatigue behaviour of pure
nickel and these have been reviewed. The obvious question
that follows is what are the effects of ion-implantation on

the fatigue behaviour of a material which is chemically

34

similar to pure nickel, but behaves differently under cyclic
plastic deformation. As we have seen earlier, the
interaction between the implanted surface layer and the
underlying cyclically deforming substrate is significant, as
it can modify various mechanical properties, particularly
the fatigue behaviour of the base material. Thus our choice
of DS-Nickel as the substrate was motivated by the fact that
it is chemically similar to pure nickel but has a very
different cyclic plasticity behaviour. The choice of the
implanted species i.e, Al+, was guided by the fact that
several studies have been done in the past to characterise
the ion-implantation dynamics of Ni-Al+ system and by
employing similar ion implantation system, we can compare
the behaviour of DS-Nickel with that of pure nickel. In the
past, there have been very few studies on the fatigue
behaviour of dispersion-strengthened alloys. So, the present
research is carried out with the hope that it will provide
us with some insight on ion implantation and its effect on

this important class of materials.

QEAPIEB_III
exps IM a R CEDU as

The experimental procedures used in this study can be
subdivided into three broad categories: specimen
preparation, low-cycle fatigue testing, and post-fatigue
surface characterisation. Low-cycle fatigue testing involved
characterising cyclic hardening behavior, cyclic stress-
strain behavior,and fatigue life. Each section will be

described below in detail.

Material

DS-Nickel was supplied by Sherritt Gordon Mines Ltd.,
Fort Saskatchewan, Canada in 0.045 inch thick sheet in hot-
rolled and pickled condition. This is a powder metallurgy
product, manufactured through hot isostatic pressing (HIP),
followed by a controlled thermo-mechanical heat treatment
schedule. Its nominal chemical composition is 98% Nickel and
2% Thoria (Thoz) with the thoria particles, 10-40 mm in
size, uniformly dispersed in a nickel matrix. The wrought
product has a characteristic microstructure comprising of
lamallae oriented along rolling direction in {100)<001> cube
texture. (fig. 1). The typical chemical analysis of DS-

Nickel is given in table 1.

35

 

36

      

   

'e‘

    

«:7

. A‘

of DS-Nickel
h of DS-Nickcl

h
icrograp

electron micrograp

Transmission

ig. 1 :

F

Optical m'

( x 10,000)(top)'

( x 50)(bottom).

( Courtesy: Sherritt Gordon Mines Ltd.)

37

 

Table 1 : Chemical composition of DS-Nickel
[ Courtesy : Sherritt Gordon Mines Ltd. ]

 

 

Element Weight (%)
Tho2 2"
C 0.005
S 0.001
Fe 0.020
Cr 0.003
Co 0.005
Ni Balance

 

 

 

 

38

Specimen Desiqn

The design of a proper specimen for fatigue testing of
such a thin sheet proved to be a serious problem. There was
no standard ASTM specimen design for such thin sheets. When
designing a good fatigue specimen, proper attention must be
paid to several critical factors, such as minimum stress
concentration in the neck region and minimum tendency of
buckling at the gauge section. For a thin sheet, there is an
acute tendency toward buckling in compression and the most
effective means of reducing this buckling tendency is to
reduce the gauge length as much as possible. However the
availability of a very small axial extensometer posed a
lower limit on the gauge length at 6 mm. An optimum fillet
was provided at the neck region to minimize stress
concentration. This design was used to prepare the
"Straight-edge specimens", which were used to characterise
the cyclic hardening and cyclic stress-strain response under

constant plastic strain (fig. 2).

The "Straight-edge specimens" required a finite straight
gauge section so that an extensometer could be mounted to be
used to measure and control the plastic strain during cyclic
deformation. However the "Stright-gauge specimens" were not
suitable for fatigue life determination experiments, because

they were prone to buckling after large accumulated plastic

39

14.28}?

 

 

 

 

‘0‘__

 

'—'>'

17.46

 

 

ril/ 1% ,
._ 1 L _.

Electropolished 1.23
region

y/l
L__L

12.5 R
+254 *O—ZS—M

 

 

 

 

 

/

l‘*‘°-l

 

 

 

 

 

All dimensioniin mm]

 

Fig. 2 : Specimen Designs tor Fatigue Testing. Straight edge
specimens ( top ) ; Hour-glass specimens ( bottom ).

 

 

 

 

40

strain. So a new specimen design with higher resistance to
buckling was needed that could be used to determine the

fatigue life of DS-Nickel.

A better choice for this purpose was the hourglass design,
which essentially had no gauge length and was thus quite
resistant to buckling. The straight edge specimens were used
to establish a load increment schedule during cyclic
hardening and cyclic stress-strain characterization. This
loading schedule was used to subject the hourglass specimens

to programmed cycling for fatigue life determination.

Machining and Surface Preparation

The machining of the specimens were done under close
tolerance i.e, within 0.001 inch of the nominal dimensions.
The straight-edge section specimens were machined in an NC
milling machine, ensuring smooth edges with minimum undercut
at the neck. The hour-glass specimens were punched out of
the sheet, using a pneumatic press. Since DS-Nickel is
mildly radioactive, special precautions were taken during
machining to minimise chip formation and to ensure operator
safety. After the initial machining, the edges were
carefully smoothed and bevelled by fine filing. The machined
specimens are then hand-polished with 600 grit abrasive
paper followed by cloth-polishing with ultra-fine diamond

paste to a mirror finish. The specimens were then rinsed and

41

cleaned in acetone and dried under air-jet.

The specimens were then dipped in lacquer to mask off the
grip regions leaving only the gauge section and the neck
region to be exposed to electro—polishing. The specimens
were given two consecutive lacquer coatings with an
intermediate air-drying for an hour. Specimens were left
overnight after the final coating to thoroughly dry out. A
small area of lacquer coating in one corner of the specimen
was wiped clean with acetone to make electrical connection

during electro-polishing.

Electro-Polishinq

The electro-polishing was done at room temperature with
an electrolyte made of 40% soln. of sulphuric acid in water.
The electro-polishing was carried out at about 5 volts dc
with a current density of 0.028-0.030 amp/mmz, with mild
stirring to ensure uniform polishing. The specimen was made
the anode while an inert nickel sheet, bent to an annular
shape, was used as cathode. The specimen was tightly secured
to the jig so that it did not wobble during polishing. The
polishing was completed in 50-60 seconds, resulting in a
highly polished, smooth surface. The electro-polished
specimens were then cleaned in water and the mask was fully
dissolved in acetone. The specimens were finally rinsed and
cleaned in acetone and methanol and dried under air-jet and

stored in a dessicator prior to ion implantation.

42

We

The ion implantation was carried out at Michigan Ion Beam
Laboratory using the 400 kV Varian Ion Implanter. Direct

implantation of Al ions was carried out at 350 keV to a

17 ions/cmz. The specimen chamber is

7

fluence of 5 x 10
maintained under a very high vacuum ( z 1.5 x 10- torr. ).
The specimens were ion-implanted in two batches with 5
specimens in each batch. They were clamped together in a
special fixture to expose only the gauge area of both sides
of the specimens to the ion beam. The beam was scanned over
the target area to ensure uniform deposition. The total dose
was monitored and controlled by a current integrator, which
was preset to a calculated value corresponding to a definite

thickness of the implanted layer. The current integrator

calculation is given below :

D = N . S / A . e ..... (6)

where

dosage ( ions/cmz)

Current integrator count

Integrator scale

a» a: 2 C
II

Aperture area

and e = electronic charge

During ion implantation, the specimens were allowed to

heat up and the Specimen temperature was measured every 30

43

minutes by means of a thermo-couple. The specimen
temperature increased slowly during ion-implantation and
reached a steady state value of 500 °C. Each batch of
specimens took about 5 hours for ion implanting on each
side. The processing parameters were maintained constant in
both batches to ensure uniform processing. After ion
implantation, the specimens were cooled down to room
temperature under high vacuum to avoid surface oxidation.
Finally, they were stored in a dessicator for fatigue

testing.

Post-Radiation Surface Analysis

The ion—implanted specimens were observed under scanning
electron microscope and phase contrast optical microscope to
detect any surface damage that may have occured during ion
implantation. The specimen surfaces were absolutely clean
and smooth and did not show any sign of surface damage due
to ion implantation. However in some specimens, the surface
did show some very fine features which resulted from
sputtering during ion implantation. The surface chemical
composition was estimated using a special-purpose software
("Profile" from Implant Sciences, Danvers, MA). The Al+ ion
concentration near the surface approximately followed a
gaussian distribution (fig. 3) with a projected range of 230
nm. This estimation is known to fit very closely to actual

measurements.

44

mcmEBmew cmEm_eE_.:o_
Lo 2:05 :ozmbcmocoo momtsm " m .mE

E: a823m aegis: 266m :38

 

 

_

08 com 2: com com 2: o
_

_ __ _ o

F

(\l

 

 

sc 1.? ”33 05.2st
.5: com ”macmm 86205

_2 0:: +_< >22 mnomeomce 2+ mm

 

01x 8 wane ‘uonenuaouog W

2:05 828888
2:53 3:38

83

 

 

 

 

 

45

Low Cycle Fatigue Testing of DS-Nickel

Ha ware

The low cycle fatigue testing of DS-Nickel was carried
out with a closed-loop servohydraulic testing machine (MTS
880) equipped with a 20,000 pounds load cell and a 5.0 mm
gauge length (MTS 632-29C-30) extensometer. The electronic
noise level of the control circuitry was less than 5
millivolts in both the load and strain channel. For our thin
sheet specimens, the proper alignment of the specimen within
the hydraulic grips was extremely important and care was
taken to minimise any misalignment. Also to ensure proper
uniaxial stress state in the specimen, special care was
taken to avoid any torsional loading of the specimen that
might result from the rotational motion of the floating
hydraulic ram. A special fixture was designed to this effect
in the form of an aluminum bar fixed to the bottom hydraulic
grip which would slide up and down along the side-posts of
the machine and would thus prevent any torsional motion of
the floating grip. To ensure proper gripping of the
specimens in the hydraulic grips, small rectangular tabs,
made of mild steel, were glued to the grip section of the
specimens. These tabs were also found to be very effective
to reduce the tendency of buckling under fully reversed

cyclic loading.

46

Quetafien

The MTS 880 system was operated by a standard input, via
an external function generator, in the form of a standard
+/- 10 V Sine wave function at a frequency of 5-10 Hz. For
our first set of experiments i.e, to characterise the
hardening behavior as well as the cyclic stress-strain
response of DS-Nickel, we chose the strain channel as the
active control mode whereas for our second set of
experiments i.e, fatigue life measurements, we chose load

control as the active mode.

The characterisation of the cyclic hardening curve
required cycling at a constant plastic strain amplitude. As
the material hardened, a continuous amplitude correction of
the total strain was needed to maintain a constant plastic
strain amplitude until the material hardened to a steady
state. For this purpose, an amplitude correction algorithm
was established. While this procedure is best suited for a
digital computer, we had to resort to manual control due to
the unavailability of a compatible system. The amplitude
correction algorithm (Grummon, 1986) can be described as

follows,

1 l
Vn+1 = Vn [ 1 - ( emp - ctp )/ (e

l] .. (7)

max {min

47

where
Vn+1 = Corrected driving signal in the active mode
Vn = Current driving signal in the active mode

6 p1 = Measured plastic strain range in current cycle

P1 = Target plastic strain range

6 = Measured max. and min. algebraic values of

6 0
max' min

total strain in the current cycle.

The total strain and the plastic strain was measured in
each cycle from the hysteresis loop and the total strain
amplitude was modified using the above equation. The process
proved very tedious and obviously required some trial and
error procedure. However, it was possible to establish a
sequential loading pattern which would result in a constant

predetermined plastic strain.

The first set of tests involved characterising the cyclic
hardening behavior of unmodified and modified DS-Nickel. The
specimens were cycled at 0.1 Hz at a constant plastic strain
range of 0.0003. The amplitude correction algorithm was used

‘to maintain constant plastic strain.

The second set of tests were designed to characterise the
cyclic stress-strain behavior of unmodified and modified DS-
Nixikel. These "Incremental step tests" were run under strain
cxnrtrol and established loading patterns were employed under
varying predefined plastic strain. The results were

pressented in the cyclic stress-strain (CSS) plot, which is

48

an analog to the monotonic stress—strain diagram. The CSS
plots for both the unmodified and modified DS-Nickel were

generated.

The third and final set of experiments involved fatigue
life determination of unmodified as well as modified
specimens. These tests were run under load control. The
steady-state stress amplitude for a predetermined plastic
strain was obtained from the CSS curves and an established
loading pattern was employed to avoid any excessive plastic
strain. These tests were run at a higher frequency of 5 Hz
to cut down the testing time. The specimens were Cycled
under fully reversed loading until a fatigue crack was
visible. Also these tests were interrupted at every 10,000
cycles to extract the surface replicas for post-fatigue
studies. The surface features of the specimen were recorded
by means of thin cellulose replicating tape which were
moistened with acetone and then pressed against the specimen
surface for about 30 seconds. The cellulose tape dissolves
in acetone and flows over the surface to create an exact

replica which was slowly peeled off and preserved.

Post-Fagigue Surface Characterisation

The surface characterisation of the fatigued specimens
involved microscopic observation of the specimen surface

under phase contrast optical microscope as well as scanning

49

electron microscope. To observe the very fine features that
may not be resolved under scanning microscope, two-stage
surface replicas were extracted at different stages of the
test and those were observed under high-voltage transmission

electron microscope.

The main objective of these surface studies was to look
for any distinct feature such as the development of
persistent slip bands and their morphologies. Since fatigue
cracks are known to initiate at persistent slip bands, their
presence and morphologies could provide informations in

regard to the fatigue crack initiation and eventual failure.

The extraction of two—stage carbon replicas involved at
first, replication of the specimen surface "in-situ" at
different stages of the test. This was done with thin
cellulose replicating tape, wetted by acetone and pressed
against the surface. As the acetone dried, the tape was
slowly peeled off after a few minutes and then it was
sputter-coated with carbon under high vacuum. To get better
resolution, we used co-deposition of carbon and platinum
using the Carbon-Platinum arc points. These carbon coated
replicas were then vapor-washed in acetone to dissolve out
the cellulose backing, leaving a thin positive carbon
replica. These were then observed under transmission

electron microscope.

CRAP ER IV
BEQQLIQ

Presentation of the results will be divided into several
areas: cyclic strain hardening behavior, cyclic stress-
strain behavior, and fatigue life performance. Quite often
the results are presented in a comparative format, where the
performance of the unmodified and modified species are
compared. Also, the results of the mechanical tests are
supplemented by microscopic observation of the surface

topography as well as fractography.

Cyclic Strain Hardening Behavior

When cycled at a constant plastic strain amplitude of

4, DS-Nickel rapidly hardens to a saturation stress

1.5 x 10’
level of 315 MPa with an accumulated plastic strain of 0.03-
0.04, which corresponds to about 50 cycles. This rapid
cyclic hardening is an important characteristic of this
material and is typical of the dispersion strengthened
alloys. No cyclic softening is observed after saturation.
With the application of A1 ion implantation, cyclic
hardening behavior is not significantly altered, although a
slight increase in the saturation stress level is observed.
This issue is discussed later. The typical cyclic hardening

behavior of DS-nickel is shown in Figure 4.

50

51

30.22-: Co 33275 9.23:6: 2.23 n

3.5 Wu .Embm "Emmi uflflsfisuum

C

md

No

To

 

_

_ :

 

 

_“ _

moood Laod ._o:coo ESE ozmmE

 

 

.............o .......................................

 

....9.1.2.3,.......................... ...................................

com

omm

'

v.5:

US$357; e
3535:: n

 

 

 

 

n
o

 

_._____

. - .
.ii‘t‘; .l‘l.\t‘\r It\§\\i(\!\z\¢ 0‘\\\n.\.((\ul .;(.l‘\l ‘7‘).\. )C \(\..r
o

w
w

w
O L

m

_~Lr________

a
.
u
u
a
u
o

.

.

.

.

. .

p n

> l.\\\\(..\\l)\(ll 4(1):) \t..3-\\ ).\(\\....)‘\.(.\I(!.O(‘)i¢
.

 

H

mm umnuu .,

m l

 

__WV__

 

oom

m;
03

com

CON
2 6226 .6 13:13.2

oor

com

Bdlll ‘apmndurg 553.115 :mnfig

omm

52

Cvclic Stress-Strain Behavior

The cyclic stress-strain behavior of DS-Nickel is
presented in the conventional Cyclic stress-strain curves or
CSS curves. The saturation stress levels at different
plastic strain ranges are recorded under plastic strain
controlled cyclic tests. The saturation stress, when plotted
against the applied plastic strain range, produces the CSS
curve. Due to the buckling tendency of the thin specimens,
the plastic strain range for tests in this study was limited
to a maximum of 0.001. The CSS curve shows a power law type
hardening behavior. The general nature of the CSS curve of
DS-Nickel remains more or less unaltered after Al ion
implantation, however there is a small increase in the
saturation stress levels over the entire range of plastic
strain amplitudes. The average increase in the saturation
stress level, in this range of plastic strain amplitudes, is
about 22 MPa, which is about a 7% increase. Figure 5 shows
the CSS curves of the unmodified as well as modified DS-
Nickel. Table 2 and Table 3 summarises the test parameters
(cyclic stress and plastic strain amplitudes) used to
characterise the CSS behavior of modified and unmodified DS

Nickel.

Cyclic Stress Amplitude, MPa

53

 

 

 

480
400 ......... a g

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

320 ,
cram-c Implanted
240
M Unmodified
160 D.S.Nickel, 0.045" sheet ------
Salutation Stiesses in
8 0 Plastic StrainControlled Tests ...........

 

 

0111i1i1i1§1

 

 

0.0 2.0 4.0 6.0 8.0 10.0

- 4
Plastic Strain Range, x 10

Fig. 5 : Cyclic Stress-Strain Behavior
of DS-Nickel

Table 2 : Cyclic Stress-Strain Data for Unmodified DS-Nickel

 

_-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I Specimen LGauge area Plastic Stress Elastic 1

Description Sq. mm. ) Strain Amplitude Modulus
BQ/US 9.2512 0.0001 264.4(MPa) 158.78 (GPa)
BQ/US 9-2512 0.0003 314.9 156.61
89/US 9.2512 0.0006 338.9 157.06
BQ/US 9.2512 0.0009 355.8 154.94
BQ/US 9.2512 0.002 401.4 155.381
B10/US 8.8228 0.001 373.3 177.97
B11/US 8.9746 0.0003 305.1 148.33
811/US 8.9746 0.0005 322.1 148.33
B11/US 8.9746 0.0006 330.6 148.33
B11/US 8.9746 0.0008 334.6 148.33
B11/US 8.9746 0.001 337.0 148.33
C3/US 9.0735 0.0003 305.6 161.79
C3/US 9.0735 0.0006 333.11 161.79
C3/US 9.0735 0.0008 347.58 161.79
C3/US 9.0735 0.0005 335.81 161.79
C3/US 9.0735 0.0009 362.28 161.79
C4/US 9.016 0.0003 320.6 —
CG/US 9.2196 0.0003 325.6 169.31
B12/US 8.742 0.0003 325.6 —

 

 

 

 

55

Table 3 : Cyclic Stress-Strain Data
for Modified DS-Nickel

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specimen Gauge area Plastic Stress Elastic
Description (sq. mm.) strain Amplitude Modulus
01/13 9.161 0.0003 320.4(MP8)164.42(GPa)
01/15 9.161 0.0005 346.4 164.42
01/13 9.161 0.0006 352.7 164.42
01/13 9.161 0.0008 363.6 154,42
C1/1S 9.161 0.001 367.0 164.42
01/13 9.161 0.0011 390.5 164.42
CZ/ls 8-7‘08 0.0003 314.1 140.25
C2/lS 8.708 0.0005 341.2 140.25
02/13 8.708 0.0006 347.3 140.25
C2/IS 8.708 0.0008 355.0 140.25
02/13 8.708 0.001 357.5 140.25
02/1s 8.708 0.0011 352.4 140.25
05/13 9.174 0.0003 334.5 169.96
05/15 9.174 0.0005 358-8 165.75
CS/IS 9.174 0.0006 373.3 165.8

 

 

56

Low-Cycle Fatique Behavior

The low-cycle fatigue behavior of DS-Nickel was
characterised by cycling under stress control at various
stress amplitudes which correspond to a plastic strain range
of 0.0001 to 0.001. The results of the fatigue life tests of
unmodified DS-Nickel are summarised in Table 4. The number
of cycles to failure (Nf) is plotted against the applied
stress amplitude (ca) on a logarithmic plot (Figure 6). The
data for the unmodified material show a linear
characteristic according to the Basquin Equation,

b

ca = af’(2Nf) ... (8)

0a = Stress amplitude

Nf = Cycles to failure

I

of = Fatigue strength coeff.

b = Fatigue strength exponent

I
For unmodified DS—nickel, of and b are 1026.0 MPa and
- 0.11 respectively. Due to the unavailability of sufficient
number of specimens, it was not possible to compute the

value of these parameters for modified DS-nickel.

The effect of Al ion implantation on the fatigue life of
DS-Nickel was evaluated by subjecting the surface-modified
DS-Nickel specimens to similar cyclic stress. However, due

to the unavailablity of large number of modified specimens,

Table 4

57

Fatigue life data for Unmodified DS-Nickel

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specimen Gauge area Stress Plastic Cycles to
Description (sq. mm.) Amplitude strain failure
D1/UH 12.963 265.9 (MPa) 0.0001 78.498
03/UH 14.426 285 0.0003 48,326
06/UH 13.037 305 0.0003 33,592
D7/UH 13.695 319 0.0003 37,665
DQ/UH 13.991 380 0.001 5,495
E1/UH 13.584 362 0.001 5,643
E2/UH 14.444 375 0-001 8.152
E3/UH 13.978 382.5 0.001 3,995
E6/UH 14.930 309.5 0.0003 32,072
E7/UH 13.815 314.8 0.0003 25,196
E8/UH 14.575 315.0 0.0003 27,395
E9/UH‘ 13.936 315.3 0-0003 35.773
E11/UH 14.762 314.8 0,0003 35,175
E12/UH 14.777 315.4 0.0003 25,445
E13/UH 14.998 314.9 0-0003 20,012

 

 

 

 

 

 

58

it was not feasible to employ various levels of stress
amplitude. Instead, a single stress amplitude ( 315 MPa )
was chosen, which corresponded to a plastic strain range of
0.0003. The results of the fatigue life tests of modified
DS-Nickel are summarised in Table 5. The average number of
cycles to failure for surface-modified DS-Nickel at this
stress amplitude was significantly higher than that of
unmodified DS—Nickel. These results are incorporated in
Figure 6 to show the comparatively higher fatigue lives of
modified DS-Nickel. Due to the inherent scatter in the
fatigue test data, it was necessary to perform a statistical
analysis on these data and based on that analysis, it can be
concluded with more than 80% confidence that the mean life
of the modified DS-Nickel specimens is significantly higher

than that of the unmodified specimens (see Appendix I).

The average life of unmodified DS-Nickel, under a
constant plastic strain range of 0.0003, is about 28,000
cycles, whereas the average life of modified DS-Nickel under
similar plastic strain is about 36,000 cycles. This
corresponds to a 28% increase in fatigue life. This result
is in line with the results of similar work by other
investigators (Jones & Morrison, 1987). Table 6 summarises
the effect of Al ion implantation on the fatigue lives of

pure Nickel as well as DS-Nickel.

59

 

:23.

 

 

 

 

 

30.3 moooé 9m; :53
35.3 33.6 «.6; 03.: 5:5
23.3 285 0.6; 35.? 538
62.3 800.0 6.30 «8.? 1:3
28mm”: cwmwsmm. grads? 3:8 .53 cozatomoo
.o 39:52 0:35 335 no; 0390 coEBmam

 

 

 

 

 

 

 

.mxw_z.mo 00:505. .2 SmdimtJIo: :mu

 

60

Stress Amplitude, MPa

 

axgzéo so oocmctotoa oz.— 03 _

 

22 .233". 2 3.93 moo

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

9m 3. 05
0mm .
62:35::
.2200 £85 2620.
com
0 86% 55.22661!
«
, 32.23. + 2 lit,
03
r
K
86

 

 

 

 

 

 

 

 

 

 

 

 

«:0.

IO
N

(D
N

edw ‘apnmdwv 958119 601

61

33 68:85 ncm mocow ...

 

 

 

 

 

 

 

mm coo on 000 mm 5m: x m .322 .8 Nofi
. . mEmB mci
x . .
Z 000 at“ 000 m; K: m .. . 6x22 836
. mEEm mmumoo
. x
we com on of. me 5 .or m ... . .322 8:0
ompz<46§_
A .\. v mm: 20. . ._< omiaozz: mwz<m
magi... z. 23me 4<_mm._.<_2
mm<mmoz_ 0_.._.w<.._n_

 

mm3.:<n_ O... mm40>0

 

 

 

 

 

coszmErE co_ E35832 305:5 ncm 5:5
_oxo_z.mo new 6x22 636 so 83 6:96“. to comtmano

m 2ng

 

 

 

62

The results presented so far contain the results of
various mechanical tests that were performed to characterise
the effects of Al+ ion implantation on the cyclic plasticity
behavior of DS-Nickel. To supplement the results of these
mechanical tests, microscopic observation of the specimens

were performed, which is detailed below.

Microscopic Observation of the Specimens

The microscopic observation of the specimens comprised
of: a) Nomarski phase contrast optical microscopy;
b) Observation of the fracture surface and the specimen
surface under the scanning electron microscope and
0) Observation of the two-stage carbon surface replica under

transmission electron microscope.

The main objective of the microscopic observation of the
specimen surface was to find if there was any formation of
persistent slip bands, which are known to initiate fatigue
cracks. Also, if there were any formation of persistent slip
bands, the ion implantation would have altered the
morphology of these bands and that would give some clue to
the micromechanism of fatigue crack initiation in DS-Nickel.
The morphology of the PSB, particularly with respect to the
formation of deep extrusions/intrusions, dictate to a large
extent whether the crack is initiated within the P885 or at

PSB-matrix interface.

63

The first set of microscopic observations consisted of
studying the electropolished surfaces of the specimens using
Nomarski phase contrast microscopy, before and after
fatigue. It was observed that there were no formation of
slip bands in unmodified DS-Nickel. No other evidence of
surface damage accumulation could either be resolved under
optical microscopy (Figure 7). The absence of any slip band
formation in fatigued DS-Nickel was unexpected, but not
quite surprising. As discussed earlier, PSB formation is
mostly favoured by a single active slip system. However, in
DS-Nickel, the presence of very fine ( z 25 nm ) thoria
particles causes the homogenisation of plastic strain by
activating multiple slip systems near their vicinity. The
effect of plastic strain homogenisation is so pronounced
that no slip bands could be observed, at least with the
optical microscope. The surface of the modified DS-Nickel
was also observed before and after fatigue, but there was
hardly any surface damage feature that could be resolved
with optical microscope ( Figure 8), except the fine
sputtering morpholgy resulting from ion implantation.
However looking at the crack tip (Figure 9),it was observed
that there were considerable surface damage near the crack

tip, but only in the form of surface rumplings.

The next step was to observe the specimen surface and the
fracture surface under the scanning electron microscope. The

specimen surfaces were observed before and after fatigue but

64

no prominent surface features were found that could be
related to fatigue crack initiation. However, an exception
to this general pattern was observed at the edges of the
thin sheet specimens, where PSB formations did take place
(Figure 10). There was no severe extrusion/intrusion
formation within the P83 (Figure 11) and crack initiation
was found to occur at the PSB-matrix interface (Figure 12).
In other cases, however, crack initiation was often found to
occur without any appreciable surface damage in the form of
slip bands (Figure 13) and it was speculated that a
mechanism of sub-surface crack initiation may be operative.
Upon observation of the modified specimens, it was found
that there was no formation of surface slip bands in the
regions that were fully exposed to the ion beam. Again
cracks were found to initiate in these specimens at the
edges, which may not have received full radiation dose

(Figure 14 & 15).

The fracture surfaces were examined under SEM to
determine the possible crack initiation site. The fracture
surfaces were found to contain numerous microcracks all over
the matrix (Figure 16). The presence of these small cracks
suggest that the failure may have resulted from a sub-
surface fatigue crack initiation. This issue is discussed at

length in later sections.

The final technique used in surface observation was to

replicate the finest surface features of the specimen and to

65

observe the replicas under very high magnification. The
procedure of creating two-stage carbon replicas has already
been discussed previously. These carbon replicas were
observed under transmission electron microscope under very
high magnification ( X 400,000 ). Observations at this high

magnification confirmed the findings mentioned earlier.

In summary, it has been observed that in fatigued,
unmodified DS-nickel, slip bands do not appear for most of
the surface area. However, slip bands have been observed at
the edges of sheet specimens. When slip bands are present,
crack initiation usually takes place at the PSB-matrix
interface. However, quite often, cracks are initiated in the
absence of any slip bands and in those cases, a sub-surface
crack initiation mechanism is speculated. In fatigued,
modified DS-nickel specimens, slip band formation was
completely suppressed in the regions which received full
radiation dose. However, slip band formation and crack
initiation was found to take place at the edges of sheet

specimens, which may not have received full radiation dose.

2666666 8922-3 62:85:: " k .9“.

x oom 05:6“. 2 0:95". 9:33 mamzmm 226m
.32. momtzw 885w uwcmzoaobomm

 

67

265666 6062-8 69:26.5. .4. ”m .m_”_

x oom 65:3 9 6:98”.
..0=< mom—ham

 

9:23.. 6:96”. 223
002.3 umcmzoaosomfi

 

68

 

Fig. 9 : Crack Tip Morphology in DS-Nickel

69

6:668me 895 8.22-8
62:88:: 6 $98 9: a. 8:668". scam 95 u S .9".

 

70

5&8QO .mxo_z.wo 82606:: E
umEtom beam 35 m to 3202822 momtzm ” I .9".

 

71

60.62.65 62:85:: E 86:25
5:62-me 2: E 5:35 x86 6:96“. N Na .9".

 

72

mucmm 95 00 00.00020 03.202004. 505:5
._0x0_z-mo 00:60:53 E 002950... x090 0:95". ”9 ..mE

 

73

. 0:950QO .050_z-mo
02:00.2 oomtzm _0 00m0m 05 “0 6:05.:
0.00.0 msmzmm 0cm. coszio... 05$ 35 6329.00 "3 .9".

 

74

2656000 5062.00 065.85.
88:5 .6 800m 85 .6 022:5 0.620 030:8 n m: .9".

 

75

02:22 05 E 8.020-222 so :20220 05
0532.0 ..22200 0.20506: to 8.250 23220 ”00.9...

- .1 wanna. .. \

 

CHAPT R V

DISCUS ION

In this section, various results of the experiments that
were presented in earlier sections will be critically
analysed and the implications of those results will be

discussed.

The results presented in the previous chapter may be

summarised as follows:

(a) DS-Nickel shows rapid cyclic hardening behavior and
ion-implantation does not significantly affect its rate of

hardening.

(b) Aluminum ion-implantation causes a 7% increase in the
saturation stress level of DS-Nickel within a range of

plastic strain amplitudes of 1.5 x 10"4 - 5.0 x 10-4.

(c) Aluminum ion—implantation causes a 28% increase in
the fatigue life of DS—Nickel at a stress amplitude of 315

MPa.

(d) For the range of plastic strain amplitudes employed
in this study, DS-nickel, for the most part, does not show
any evidence of slip band formation during fatigue. However,

76

77

slip bands were found to appear at the edges of sheet
specimens and initiated fatigue cracks at the PSB-matrix

interface.

(e) Surface modification of DS-nickel by ion-implantation
was found to completely suppress the appearance of slip
bands in the regions which received full radiation dose.
Slip bands and consequent fatigue cracks did appear at the
edges of sheet specimens, which were probably shadowed

during ion implantation process.

In light of these observations, let us now analyse and

explain each of these findings.

C clic Deformation and s Behav' r

DS-Nickel showed a very rapid cyclic hardening tendency
and it reached the saturation stress in less than 100
cycles. This behavior is typical for dispersion strengthened
alloys and this rapid rate of hardening is attributed to the
presence of very fine dispersoids which causes dislocation
"pile-up"s and activates multiple slip systems in their
vicinity due to cross-slip of the dislocations. The rate of
cyclic hardening of DS-Nickel was not significantly altered
by aluminum ion implantation and this is probably due to the
fact that the cyclic hardening rate of DS-Nickel is governed

mainly by the size and distribution of the thoria particles

78

which remains unaltered during ion implantation.

The small but noticeable increase in the saturation
stress levels of DS-Nickel after ion implantation is
apparent from the CSS curves. In fact, the saturation stress
level was increased by about 7% in the appplied plastic
strain range. It is theorised that the hard surface layer
act as a barrier to dislocation motion during cyclic
deformation and results in a pile-up of dislocations in the
near-surface region. This dislocation pile-up, in turn gives
rise to a back-stress, which results in an increase in the
stress amplitude for a prescribed plastic strain. However,
this effect may have a direct dependence on the ratio of the
modified surface area to the total cross-sectional area of

the specimen.

The increase in the low-cycle fatigue life of DS-Nickel
after aluminum ion implantation was expected and is in line
with the observations of other researchers ( Jones &
Morrison, 1987 etc.). A moderate increase in fatigue life
( 28% ) at 315 MPa stress level was observed. The increase
in the fatigue life can be explained in terms of accumulated
plastic strain during cyclic deformation. As we have noticed
before, the ion-implanted surface layer act as a barrier to
dislocation glide and result in a lower plastic strain for a
prescribed stress level. As a result, when cycled under
stress-control and at the same stress amplitude, the ion-

implanted specimen would accumulate less plastic strain than

79

the unimplanted one for equivalent number of cycles. Since
failure is initiated at a critical value of accumulated
plastic strain, the ion—implanted specimens would survive
higher number of cycles before failure. A more detailed
explanation of the higher fatigue life of ion-implanted
specimens, based on the fracture mechanics concept, will be

presented later.

The cyclic plasticity behavior of DS-Nickel is affected
due to aluminum ion-implantation. A hard, thin surface layer
which results from ion-implantation acts as an obstacle to
dislocation motions and thus tend to reduce the extent of
plastic deformation. It can be stated with reasonable
confidence that the surface layer which results from
aluminum ion implantation has higher yield strength than the
bulk. This additional hardening effect comes from various
factors e.g, introduction of numerous lattice defects due to
ion bombardment and solid solution hardening. Some other
secondary factors may contribute to the surface hardening
and they include, a) formation of accomodation dislocation
networks, and b) presence of a residual compressive stress
at the surface after ion implantation. Although the effects
of these individual factors towards surface hardening cannot
be quantified immediately, it can be logicaly assumed that
the cumulative effects of all these factors result in a

significant surface hardening.

Since the surface layer has higher yield stress, it can

80

be expected that for very small applied plastic strain, most
of the plastic strain will be present in the bulk while the
surface layer would essentially cycle elastically and thus
the surface would not experience any fatigue damage due to
plastic deformation. Another factor that might be beneficial
in this context is the elastic modulus effect. If the
elastic modulus of DS-Nickel is reduced by surface alloying
with aluminum, then the surface would experience lower
stress than the bulk. Although there is evidence that
aluminum addition does lower the elastic modulus of pure
nickel, no such experimental evidence is available for DS
nickel, whose elastic modulus is much lower than that of
pure nickel. In any case, the elastic modulus effect would
be less pronounced than the surface hardening effect. So,
for small applied plastic strain range, we will have a
situation where the bulk of the specimen undergoes plastic
strain while the surface layer essentially suffers little or
no plastic strain and cycles elastically. As a result, the
surface does not undergo any damage that might have occured

due to plastic deformation.

Fatigue Crack Initiation and Failure Meghanism

From the microscopic observation of the fatigued,
unmodified DS-nickel specimens, it was apparent that there
was hardly any formation of persistent slip bands on most

of the free surfaces of the specimens during cyclic

81

deformation of DS-nickel. The only exception to this general
observation was at the edges of sheet specimens, where slip
bands did appear and were found to initiate fatigue cracks

at the PSB-matrix interface.

It is theorised that in DS-Nickel, for the range of
plastic strain amplitudes employed in this study, the
presence of very fine thoria particles in the nickel matrix
homogenise the plastic strain resulting in a uniform
distribution of plastic strain all over the matrix. As a
consequence, there is no significant plastic strain
localization anywhere in the matrix. This explains the

absence of any persistent slip bands in fatigued DS-Nickel.

However, the situation is quite different at the edges of
sheet specimens, where the geometrical stress concentration
factor can be as high as 1.38 for hourglass specimens
(Peterson, 1974). While the hourglass specimen is cycled at
a nominal stress amplitude of 315 MPa, the maximum stress at
the edges may go upto 435 MPa. From the CSS curve of DS-
nickel (Leverant & Sullivan, 1968), it can be seen that this
maximum stress at the edges would correspond to a maximum
plastic strain amplitude of about 0.016, which is about 100
times higher than the nominal plastic strain amplitude of

1.5 x 10'4

. The direct consequence of this increased plastic
strain at the edges of the sheet specimens, is the formation
of persistent slip bands in those regions. This observation

is in agreement with the findings of Leverant and Sullivan

82

(1968), who showed evidence of slip band formation in DS-
nickel, when cycled at a range of plastic strain amplitudes

from 0.002 to 0.015.

It may be argued that for relatively low plastic strain
amplitudes (e.g, upto about 5.0 x 10-4), the plastic strain
in DS-nickel is distributed homogeneously over the matrix by
the fine thoria particles. While the plastic strain
homogenization effect of thoria particles is quite strong at
lower plastic strain amplitudes, this process is not quite
so effective at higher plastic strain amplitudes. As a
result, plastic strain localization occurs, giving rise to

the formation of persistent slip bands.

Crack initiation in DS-nickel has often been found
without the presence of any apparent slip band formation and
this observation leads to the possibility of sub-surface
crack nucleation. There have been some reports of sub-
surface crack initiation in the past ( Vardiman & Kant,
1981), but only in high-cycle fatigue regime. As far as the
sub-surface crack initiation site is concerned, the
particle-matrix interface seems to be the moSt probable
site. There have been reports of void formation at the
particle-matrix interface in monotonic tensile deformation
of internally oxidised copper alloys (Palmer & Smith, 1966)
and it may be speculated that similar formation of voids at
the interface may occur during cyclic loading and could very

well act as the crack nucleation site. However, no such hard

83

evidence is available that would decisively prove the

existence of such mechanism.

However, if sub-surface crack initiation is assumed, the
increase in the fatigue life observed in the ion-implanted
DS-Nickel can be explained, at least on a qualitative basis,
from a fracture mechanics point of view. As we know, the
stress intensity factor associated with a surface crack is
considerably higher than that associated with an interior
(sub—surface) crack of the same overall crack length. While
the exact relation depends on the loading configuration and
crack geometry, it can be stated in general that a surface
crack is associated with a higher stress intensity factor
than a sub-surface crack. Now in unmodified DS-Nickel, the
nucleated crack is a sub-surface one and with continued
cycling, it immmediately grows to a free surface and becomes
a surface crack. However, in modified DS-Nickel, the hard
surface layer resists the growth of the sub-surface crack
and forces it to grow in a non self-similar manner. In other
words, in early stages of fatigue, the presence of this hard
surface layer delays the transformation from sub-surface to
a surface (edge) crack. However, once the edge crack is
formed, the situation is identical in both the implanted and
the unimplanted specimen and the crack grows freely, leading
to catastrophic failure. Thus the increase in the fatigue
life of the implanted specimen can be solely attributed to
the time delay caused by the hard surface layer in the

transformation of a sub-surface crack to a surface (edge)

84

crack.

Analysis at Fatigue Crack grggth

The growth rate of a fatigue crack under low-cycle

fatigue condition is given by ( Tomkins, 1968 ),

(da / dn) = Acp.AD (9)
where,
A6 = the plastic strain range

D = size of the crack-tip plastic zone

AD=D -Dm.

max in

Dmax — Plastic zone Size at "max
. = Plas ' o 'ze .

Dmln tic z ne 31 atamln

From Irwin's analysis (Hellan, 1984), the size of the

crack-tip plastic zone is given by,

D

( 1 / «k2) ( K12 / ayz ) .

where, k = 1, for plane stress
k = 73, for plane strain

a = yield stress

For our case, we can assume plane stress condition, k = 1.

85

The stress intensity factor is given by,
KI = a/(na) , a = half crack length
So, the crack-tip plastic zone size is given by,

0 = a(o/ay)2

Again, AD = Dmax - Dmin
_ 2 2 _ 2
7 (a / 0y )(amax 0min )
For a fully reversed cyclic loading, a . < 0 . As far as

min

the crack growth is concerned, the compressive part (a<0)
has no significant contribution and thus we can replace it
by the lower bound, 0min = 0.

So, 0

min 0 , and AD = Dmax = a(a

For our case,

a = 315 MPa, 0y

max 350 MPa and.Aep=0.0003

-4
Thus, (da/dn) = Dmax’ASp = a (2.43 x 10 ) mm/cycle.

This gives us the fatigue crack growth rate in unmodified

DS-Nickel.

86

Now for modified DS-Nickel, we have a hard surface layer
and let us denote the yield stress of the surface layer by,
I

a = 10 where A > 1
y y ’

and 0y = yield stress of unmodified DS-nickel

So, the fatigue crack growth rate in the modified DS-

Nickel is given by,

(da/dn)’ Aep.(AD)’ = A. .a.(amaX/Aay)2

P

a (2.43 x 10-4) / 12

So, comparing, we get,

(da/dn)’ / (da/dn) = 1 / 12

We can see from this relation, that the crack growth rate
in the surface layer as compared to the bulk, is strongly
dependent on its yield stress. For instance, a 15% increase
in the surface yield stress of DS—Nickel will result in a
25% reduction in the fatigue crack growth rate in the

surface layer.

Log-gygle Fatigue Life Estimatioa

Let us recall the fatigue crack growth equation (9),

87

da/dN = Asp . AD

_ 2
For our case, AD — a(amaX/Aay)

where, A 1 , for unmodified DS-Nickel

and A > 1 , for modified DS-Nickel surface layer

So, we get,

_ 2
da/dN - Aepl.a.(amax/Aay)
or da/a = A5 .(0 /A0 )2 dN
’ p max: y
Integrating, we get,
_ 2 -
1n (ac/a0) — C / A (Nf No)

where

__ 2
C - Acp. (amax/ay)

a = initial crack length

0
aC = final (critical) crack length

N0 = Number of cycles for crack initiation
Nf = Number of cycles to failure

Rewriting, we get,
_ 2
N — NO + (A /C) 1n (ac/a0) ..... (10)

This is the generalised low-cycle fatigue life equation.

88

For unmodified DS-Nickel, this equation can be directly used
to either solve for the critical crack size (ac) or to

predict the fatigue life for a pre-defined crack size.

The number of cycles for crack initiation, No’ is typically

assumed to be about 10% of the total fatigue life, Nf.

For unmodified DS-Nickel,

Nf = 28,000 , N0 = 0.1Nf = 2,800
c = 2.43 x 10’4 , A = 1
So, from eqn. (2) , we get, aC = 456.5 ao

This relation can be later used to predict the fatigue

life of modified DS-Nickel.

Prediction of Low-Cycle Fatigue Life of Modified DS-Nigkal

 

The total fatigue life of a specimen can be subdivided
into three parts, e.g, (a) the time to initiate a flaw or
crack ; (b) the growth of the initial crack to a threshold
value where rapid growth of the crack takes over; and (c)

the rapid growth of the crack to a catastrophic failure.

Mathematically,

89

where
Nf = total fatigue life to failure (# of cycles)
N1 = time to initiate a flaw ( ,, )
NC = growth of the initial flaw to
a threshold size ( ,, )
N9 = rapid growth period ( ,7 )

Typically, N1 is assumed to be about 10% of the total
life (i.e, Ni z 0.1 Nf ). The most critical parameter is NC,
the time taken for the initial flaw to grow to a threshold
size when the rapid growth sets in. For modified DS-Nickel,
this period denotes the growth of a sub-surface sub-micron
crack through the hard surface layer to become an edge

crack.

So, for modified DS-Nickel,

NC = d / (da/dn)’

where , d = thickness of the hard surface layer

and (da/dn)’ = crack growth rate in the surface layer

Recall that, (da/dn)’ = (l/A2)(da/dn)

and also, da/dn = a (2.43 x 10-4) mm/cycle.

If we assume, a = 50 nm , d = 300 nm and A = 1.15 then,

90

we get, NC = 32,654 cycles

As mentioned before, Ni is assumed to be about 10% of
total life, Nf. For unmodified DS-Nickel, N1 = 2,800 cycles.
We can use this same value of Ni for modified DS-Nickel,
because the flaw initiation is a bulk process, which remains

unaltered during surface modification.

The final failure occurs due to the rapid growth of the
crack. This period is denoted by Ng and for modified DS-
Nickel, we get,

Ng = Nf - Nc - Ni
36,000 - 32,654 - 2,800

546 cycles
As we can see, the majority of the fatigue life of
modified DS-Nickel is spent during the slow growth of the
sub-surface crack through the ion-implanted surface layer.
Based on this observation, we can define an effective

crack growth rate for modified DS-Nickel from a simple rule

of mixture concept as follows,
(da/dn)eff = (NC/Nf)(da/dn)’ + (Ng/Nf)(da/dn)
If we assume, a = 50 nm, d = 300 nm and A = 1.15,

we get, for modified DS-Nickel,

91

9 mm/cycle

(da/dn)eff = 8.38 x 10
Furthermore, if we assume this effective crack growth
rate to be representative of the modified DS-Nickel, then,

as in unmodified DS-Nickel, we can integrate the crack
growth rate equation over the whole range of crack growth

and solving as before, we get the fatigue life eqn. (10),

N = No + (AZ/C) ln (ac/a0)

2800, c = 2.43 x 10'4

Here, N and ln (ac/a0) = 6.12,

So, for ion-implanted DS-Nickel, we get,

Nf = 2800 + 12 [6.1236 / (2.43 x 10'4)]

If we assume, A = 1.15, we get N = 36,127 cycles

f
This predicted value of Nf for modified DS-Nickel agrees
very well with the experimental value of 36,000 cycles.
Thus, based on these calculations, it can be speculated that
the surface modification of DS-Nickel by ion-implantation
creates a hard surface layer which has a yield strength
about 15% higher than the bulk and this hard surface layer
is instrumental in slower crack growth rate, therby causing

a 28% increase in the fatigue life.

92

n o t' l aw ' -N’ el

So far, it has been postulated that the increased fatigue
life in the ion-implanted DS-Nickel is due to the slower
crack growth through the hard surface layer. Now let us

recall from earlier discussion that,

Nf = N1 + NC + Ng , for unmodified DS-Nickel
I

I
and Nf = N. + N + N , for modified DS-Nickel
1 c g
As discussed before, the bulk properties of DS-Nickel
remains unaltered during surface modification and thus we

can assume Ni and Ng, which are related to the bulk

properties, to remain unaltered.
Also recall that,

NC = d / (da/dn) and Nc' = d / (da/dn)’

where,

d = the thickness of the surface layer

- N = 36,000 - 28,000 = 8,000

f

but, N ' = N [ (da/dn) / (da/dn)’] = AZNC

93

or, ( A2 - 1 ) NC = 8,000

If we assume, A = 1.15, then Nc z 24,800

Now, (da/dn) = d / Nc

here, d = 300 nm and NC 2 24,800

8

thus, (da/dn) z 1.2 x 10- mm/cycle

Also, we know for low-cycle fatigue,
(da/dn) = Acp.AD
here, Asp = 0.0003 and AD = 0.81a

So, we have,

(0.0003)(0.81a) = 1.2 x 10"8

or, a z 50 x 10.6 mm = 50 nm

Thus, the initial flaw size in DS-Nickel is calculated to

be of the order of 50 - 100 nm.

Experimental evidence of void formations at the particle-
matrix interface in internally oxidised copper alloys had
been reported by Palmer and Smith (1966). The voids, at an

early stage of formation, at the Cu - SiO2 interface, were

94

measured to be of the order of 50 - 70 nm. Thus our
theoritical prediction of initial flaw size agrees quite

well with the experimental observations.

To summarize, a detailed analysis, based on fracture
mechanics concepts, has been carried out to quantify the
effect of a thin, hard surface layer on the low—cycle
fatigue life. The analysis predicts a fatigue life of 36,127
cycles for the modified DS-nickel specimens, which is very
close to the observed fatigue life of 36,000 cycles, on the
basis of an assumed 15% increse in the yield stress of the
surface layer over the bulk. This assumption of 15% increase
in the yield stress of the surface layer seems quite
reasonable, although there is no experimental data to

support this assumption.

However, a questionable prediction of this analysis is
the initial flaw size, which according to this analysis,
comes out to be of the order of 50-100 nm. This value is
much smaller than the typically quoted initial flaw size of
25-50 p. This raises the fundamental question of the
applicability of such an analysis based on fracture
mechanics, to such a small size flaw. The situation is even
more complicated by the fact that the applied stress
amplitude is very close to the gross yield stress and
elasto-plastic fracture mechanics analysis cannot properly
describe such situations. A more fundamental analysis,

either based on dislocation theory or strain energy methods

95

may be more appropriate. However, to date there is no such
single established analysis available that can be directly

applied in such a situation.

CHA V

CONCLUSION

High energy aluminum ion implantation has been found to
be a useful technique to modify the surface of thoria
dispersion strengthened nickel by producing a thin, hard
surface layer. Using this method, it was possible to create
specific surface conditions which would otherwise be

difficult, if not impossible to produce.

The cyclic plasticity behavior of DS-Nickel with and
without surface modification by aluminum ion implantation
was characterised. It was observed that DS-Nickel undergoes
rapid cyclic hardening and aluminum ion implantation did not
significantly affect its rate of hardening. It was concluded
that ion implantation which only alters the surface does not

affect the bulk properties of the substrate.

The saturation stress levels of ion-implanted DS-Nickel
showed a small increase compared to that of unmodified DS-
Nickel, as measured in the cyclic stress-strain behavior
under constant plastic strain tests. This small increase is
attributed to the presence of the hard surface layer, which
acts as a barrier to dislocation glide, thereby causing a

hardening effect.

96

97

The low-cycle fatigue life of DS—Nickel was moderately
increased by surface modification by aluminum ion
implantation, as measured under stress controlled fatigue
tests. The presence of a hard surface layer in ion-implanted
DS-Nickel reduced the extent of plastic deformation by
acting as a barrier to dislocation motion and therefore

contributed to the increased fatigue life.

From the macroscopic and microscopic observation of the
surface of cyclically deformed specimens of DS-Nickel, it
was concluded that for the range of plastic strain
amplitudes employed in this study, DS-Nickel does not
develop persistent slip bands on most of its free surfaces
during cyclic deformation. Only exception to this pattern
was found at the edges of sheet specimens, where slip bands
did form and were found to initiate fatigue cracks at the
slip band-matrix interface. The stress concentration at the
edges of sheet specimens was high enough to produce very
large plastic strain amplitude and as a result, persistent
slip bands were formed in these regions. This observation is
in agreement with the findings of an earlier work by

Leverant and Sullivan (1968).

Often, fatigue cracks were found to initiate without
apparent surface damage in the form of slip bands and it was
concluded that a mechanism of sub-surface crack nucleation
was active in this system. It was speculated that the crack

may have initiated at the particle-matrix interface.

98

Assuming that sub-surface crack nucleation occured, the
effect of the hard surface layer in the ion-implanted DS-
Nickel towards the increase of fatigue life was explained
from a fracture mechanics point of view. The presence of the
hard surface layer reduces the crack growth rate in the
early stages of fatigue and delays the transformation of a
sub-surface crack to a surface crack. This delay is
manifested in the increased fatigue life of modified DS-

Nickel.

It must be noted, however that the fracture mechanics
analysis employed here does have some severe limitations. A
more fundamental analysis based on either dislocation theory

or strain energy methods may be more appropriate.

Scope of Future Research

The present work was directed to investigate and provide
a better understanding of the cyclic deformation behavior of
dispersion strengthened nickel alloyand the associated
effects of aluminum ion implantation, as a continuing effort
to understand the complex mechanisms associated with the
presence of an ion implanted surface layer and the

underlying cyclically deforming substrate. The present work

99

has also demonstrated the potential of ion implantation
process as a tool to modify the surface of a dispersion
strengthened alloy to improve its low-cycle fatigue
performance. A model has been proposed to characterise the
cyclic deformation behavior of DS-Nickel which is consistent
with the observations. However, it has also raised a few
interesting questions which could provide the guidelines for

further research in this area.

First of all, it has not been possible to pinpoint the
exact location of the sub-surface fatigue crack initiation.
It has been speculated that the crack might have initiated
at the particle-matrix interface, however further research
is needed, particularly involving extensive transmission

electron microscopy, to conclusively settle this issue.

Also, due to the limitation on the specimen geometry, it
has not been possible to characterise the cyclic stress-
strain behavior of DS-Nickel at higher plastic strain
ranges. Future work in this direction will provide some
useful informations which will give a more comprehensive

picture of the physical processes.

Another limitation of the present work was that the low-
cycle fatigue tests of the modified DS-Nickel were carried
out at only one stress level. It will be useful to carry out
similar fatigue tests at higher stress levels which would

give better comparison of fatigue lives between the

100

unmodified and modified DS-Nickel.

Also, to quantify the effects of ion implantation on
fatigue life, a fracture mechanics based analysis has been
presented. Although, in this case, the theoritical
prediction comes very close to the experimental
observations, the theory does suffer from severe
limitations. A more fundamental approach, based on
dislocation theory or strain energy methods would be more

appropriate.

Finally, future research on similar dispersion
strengthened alloys will be of interest. In particular,
research on aluminum and copper base dispersion strengthened
alloys and particulate-reinforced composites which are of
considerable commercial importance, will be very much in

order.

APPENDIX

APPENDIX I

Statistical Analysis of Fatigue Life Data

The low-cycle fatigue life of surface modified DS-Nickel
was found to be higher than that of unmodified DS-nickel, as
determined under stress controlled fatigue tests at an
amplitude of 315 MPa. However, due to the inherent scatter
in the data, it was necessary to perform a statistical
analysis to determine the confidence level at which the mean
fatigue life of surface modified specimens was significantly
higher than that of unmodified DS-Nickel. The "Student T-
Distribution" method was used to determine the confidence
level at which the difference between the two mean values

was statistically significant.

Experimental Data

Unmodified Specimens

Number of samples, nx = 6

Mean fatigue life, 0 = 28,166

x
Standard deviation, SX = 6,170

101

102

Surface Modified Specimens

Number of samples, ny = 2

Mean fatigue life, p 35,770

Y
Standard deviation, Sy = 3,754

Statistical Analysis

( Ref: Engineering Experiments, Lipson & Sheth, 105 )

We compute,

= + —2=6
V n my
_ _ 2 _ 2 1/2
A — [{ (nX 1)SX + (ny 1)Sy ){ nX + ny } / { v nX ny}]
= 4766
and, t2 = ( VX - py ) / A = - 1.595

Now, for 80% confidence level, 0/2 = 0.1

and from t-distribution table, t0 1 6 = 1.440
° I

Also, for 90% confidence level, 0/2 = 0.05

and from t-distribution table, t = 1.943
0.05, 6

103

since, 1.440 < ABS(t2) < 1.943

We can say with more than 80% confidence that the average
fatigue life of surface modified DS-Nickel is significantly

higher than that of unmodified DS-Nickel.

 

BIBLIOGRAPHY

1)
2)
3)
4)
5)

5)

7)

8)
9)
10)

11)

12)

13)

14)

15)

16)
17)
18)

19)

20)

21)

BIBLIOGRAPHY

Abel, A. & Ham, R.K. - Acta Met., 14, 1489 / 1495, 1966
Ahmed, M. & Potter, D.I. - Acta Met., 33, 2221, 1985.
Ansell, G.S. & Lenel, F.V. - Acta Met., 8, 612, 1960.
Ashby, M. - Z. Metalk., 55, 5, 1964.

Ashby, M.F. - Acta Met., 14, 679, 1966.

Bakhru, H., Gibson, W., Burr, C., Kumnick, A.J. &
Welsch, G.E. - Nuc. Inst. Meth., 182/183, 959, 1981.

Broom, T.T. & Ham, R.K.
5 Proc. Royal. Soc. A, 251, 186, 1959

Chou, Y.T. - Phys. Stat. Solid., 17, 509, 1966
Coffin, L.F. — Trans. ASME, 76, 923, 1954.
Conners, G.H. - Int. J. Eng. Sci., 5, 25, 1967.

De Jonghe, L.C. & Greenfield, I.G.
- Acta Met., 17, 1411, 1969.

Ebeling, R. & Ashby, M.F. - Phil. Mag., 13, 805, 1966

Ewing, J.A. & Humfrey, J.C.W.
- Phil. Trans. SOC. A, 200, 241, 1902.

Finney, J.M. & Laird, C. - Phil. Mag., 31, 339, 1975

Fisher, J.C., Hart, E.W. & Pry, R.H.
- Acta Met., 1, 336, 1953.

Follstaedt, D.M. - Nuc. Inst. Meth., B 7/8, 11, 1985.
Friedel, J. - "Dislocations", Pergamon Press, 1964.
Greenfield, I.G. - J. Sci. Indus. Res., 32, 521, 1973.

Greenfield, I.G. & Purohit, A.
- Mat. Sci. Eng., 46, 89, 1980.

Grummon, D.S. - Doctoral Thesis, U. of Mich., 1986.

Grummon, D.S., Jones, J.W., Meridon, J.M., Was, G.S. &
Rehn, L. - Nuc. Inst. Meth., B 19/20, 227, 1987.

104

 

22)

23)
24)

25)

26)
27)

28)

29)

30)

31)

32)

33)

34)

35)

36)

37)

38)

39)

40)

41)

42)

105
Grummon, D.S., Jones, J.W. & Was, G.S.
- Met. Trans., 19A, 2775, Nov., 1988.
Hagg, G. - Z. Phys. Chem., 812, 33, 1931.
Head, A.K. - Proc. Royal Soc., B, 66, 793, 1953.

Heimendahl, M.V. & Thomas, G.
- Trans. AIME, 230, 1520, 1964.

Helgeland, O. - J. Inst. Met. , 93, 570, 1965.
Hellan, K. - "Intro. Fracture Mech.", Mcgraw-Hill, 1984

Hirsch, P.B. & Humphreys, F.J.
- Proc. Royal. Soc., A, 318, 45, 1970.

Hohmuth, K., Rauschenbach, B., Kolitsch, A. &
Richter, E. — Nuc. Inst. Meth., 209 / 210, 249, 1983.

Hohmuth, K., Richter, E., Rauschenbach, B. &
Blochwitz, C. - Mat. Sci. Eng., 69, 191, 1985.

Inman, M.C., Zwilsky, K.M. & Boone, D.H.
- Trans. ASM, 57, 701, 1964.

Jata, K.V. & Starke, E.A. - J. Metals, 23, Aug., 1983.

Jones, R.L. & Kelly, A.
- "Oxide Disp. Strength.", Gordon & Breach, 229, 1968

Jones, J.W. & Morrison, D. - General abstract session,
TMS-AIME Fall Meeting, Cincinnatti, Sept. 1987.

Kelly, A. & Nicholson, R.B.
- Prog. Mat. Sci., 10, 1964.

Kimura, T. - Trans. Japan Inst. Met., 6, 217, 1965.
Koehler, J.S. - Phys. Rev., 60, 397, 1941.

Kujore, A., Chakraborty, S.B., Starke, E.A. &

Legg, K.O. - Proc. Symp. Mat. Res. Soc.; Preece &
Hirvonen, AIME, 132, 1980.

Kupcis, O.A., Ramaswami, B. & Woo, O.T.
- Acta Met., 21, 1131, 1973.

Laird, C., Langelo, V.J., Hollrah, M., Yang,N.C. &
De La Veaux, R. - Mat. Sci. Eng., 32, 137, 1978.

Lauffer, E.E. & Roberts, W.N. - Phil. Mag., 14, 65, 1966

Leverant, G.H. - Trans. AIME, 239, 1992, 1967.

43)

44)

45)

46)

47)

48)

49)

50)

51)

52)

53)

54)

55)
56)

57)

58)

59)

60)

61)

62)

63)

106

Leverant, G.R. & Sullivan, C.P.
- Trans. AIME, 242, 2347, 1968.

Manson, S.S. - Nat. Adv. Aero., Tech. note, 2933, 1954.

Mendez, J., Violan, P. & Villain, J.P.
- Scripta Met., 16, 179, 1982.

Mendez, J., Violan, P. & Denanot, M.F.
- Nuc. Inst. Meth., B 19/20, 232, 1987.

Meyers, C. & Grummon, D.S.
- General abstract session, TMS-AIME fall
meeting, Sept. 1988.
Mori, T. & Mura, T. - Mat. Sci. Eng., 26, 89, 1976.
Mughrabi, H. - Mat. Sci. Eng., 33, 207, 1978.

Mughrabi, H. & Wang, R.
- "Defects, fracture & fatigue", Eds. Nijhoff, 15, 1982

Orowan, E. - Proc. Symp. on Internal Stresses in metals
and alloys, Inst. Met., London, 451, 1948.

Palmer, I.G. & Smith, G.C.
-"Oxide Disp. Strength", Gordon & Breach, 47, 253, 1966

Peterson, R.B. - "Stress conc. factors", John Wiley,
34, 1974.

Sleeswyk, A.W., Kok, H.J.G. & Boom, G.
- Scripta Met., 14, 919, 1980.

Steiner, D. & Gerold, V. - Mat. Sci. Eng., 84, 77, 1986
Tomkins, B. - Phil. Mag., 1041, 1968.

Vardiman, R.G. & Kant, R.A.
- J. Appl. Phys., 53, 1, 690, 1982.

Vardiman, R.G. & Cox, J.E. - Acta Met., 33, 2033, 1985

Vogel, W., Wilhelm, M. & Gerold, V.
- Acta Met., 30, 21, 1982.

Wang, J.J., Welsch, G., Bakhru, H., Mashayekhi, A. &
Gibson, W. - Nuc. Inst. Meth., B 7/8, 228, 1985.

Watt, D.F., Embury, J.D. & Ham, R.K.
- Phil. Mag., 17, 1991, 1968

Wells, C.H. 8 Sullivan, C.P. - Trans. ASM, 57, 841,1964

Wilhelm, M. - Mat. Sci. Eng., 48, 91, 1981.

"tillllllllllllllllllllli