Group Title: microstructural evolution of aluminum during the course of high temperature creep
Title: The Microstructural evolution of aluminum during the course of high temperature creep
Full Citation
Permanent Link:
 Material Information
Title: The Microstructural evolution of aluminum during the course of high temperature creep
Physical Description: xxii, 529 leaves. : illus. ; 28 cm.
Language: English
Creator: Connell, Richard Grant, 1938-
Publication Date: 1973
Copyright Date: 1973
Subject: Aluminum -- Creep   ( lcsh )
Metals -- Effect of high temperatures on   ( lcsh )
Metallurgical and Materials Engineering thesis Ph. D
Dissertations, Academic -- Metallurgical and Materials Engineering -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 517-527.
General Note: Typescript.
General Note: Vita.
Statement of Responsibility: by Richard G. Connell.
 Record Information
Bibliographic ID: UF00098184
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000585115
oclc - 14183894
notis - ADB3747


This item has the following downloads:

PDF ( 11 MBs ) ( PDF )

Full Text




A Dissertation Presented to the Graduate Council
of the University of Florida
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy


CQNNBLL, Richard Grant, Jr., 1938.

The University of Florida, Ph.D., 1973
Engineering, metallurgy

SUniverlsty Microfilms. A XERD Company, Ann Arbor, Michigan



To my wife, Linda, and children,
Deborah and Richard


The author wishes to express his gratitude to the

members of .his supervisory committee for their service

on the committee and for their guidance during his course

of study. In particular, the author is indebted to Dr.
Frederick N. Rhines for his continuous interest in the

research and for the many hours of discussion related to

hot deformation. Discussions with Professors Robert T.

DeHoff, Ellis D. Verink, Jr., and Robert E. Reed-Hill were

of significant benefit and are hereby acknowledged.

It is with gratitude that the author acknowledges the

financial support of the Engineering and Industrial Experi-

ment Station at the University of Florida, under whose
sponsorship a majority of this research was conducted.


ACKNOWLEDGEMENTS . . . . . . ... iv

LIST OF TABLES . . . . . . ... viii

LIST OF FIGURES . . . . . . . x

ABSTRACT . . . . . . . .... . xx


I INTRODUCTION . . . . . ... .1

1.1. Scope of the Research . . . 2
1.2. The Scientific Background
for the Research . . . . 2


2.1. The Material . . . . .. 58
2.2. Fabrication of Creep Specimens . 60
2.3. Creep Testing . . . . .. 67
2.4. Metallographic Techniques .. . 73


3.1. Observations of Surface Features 90
3.2. The Development of Substructure 99
3.3. Grain Boundary Serrations
Formed During Creep . . . .. 117
3.4. Relations Between Grain Boundary
Serrations and the Subgrain
Morphology . . . . . .. .120
3.5. The Nature of Subgrain Rows . . 124
3.6. The Effect of Grain Boundaries
on Deformation . . . . .. 166
3.7. The Microstructural Evolution
During Creep I: Results from
Constant Load and Constant
Temperature Creep Tests . . .. .167



III (continued)

3.8. The Microstructural Evolution
During Creep II: Results from
Variable Load and Temperature
Creep Tests . . . .... 287
3.9. The Evidence for Shear Along
Subgrain Boundaries . . . .. 382
3.10. A Summary of the Observations
and Results . . . . ... 400
IV DISCUSSION . . . . . . ... .408

4.1. Grain Boundaries and Deformation 412
4.2. A Schematic Representation of
High Temperature Creep . . .. .428
4.3. The Microstructural Evolution
with Respect to the Schematic
Representation of Hot Deformation 434
V CONCLUSIONS . . . . . . ... .453

VI FUTURE WORK . . . . . . ... .456

6.1. The Kinetics of Shearing
Along Subgrain Boundaries . .. 456
6.2. A Correlation Between the Onset
of Tertiary Creep and Micro-
structure . . . . . . 457
6.3. The Effects of Minor Alloying
Additions to Aluminum Upon the
Kinetics of Sub-boundary Shearing 458
6.4. The Effects of Subgrain Struc-
tures Produced by Creep Upon the
Mechanical Properties at Ordinary
Temperatures . . . . .. 460

ANODIZING . . . . . . . 463



CONDITIONS . . . . . . . 465

THE MEAN OF NL .. . . . . . .470

TOLERANCE LIMITS .. . . . ... 471

FOR SPECIMEN S3-12 . . . . .. 473

BOUNDARIES . . . . . . . 478

BOUNDARIES . . . . . . . . 513

BIBLIOGRAPHY . . . . . . . . 517

BIOGRAPHICAL SKETCH . . . . . . ... 528


Table Page

I Chemical Analysis of Aluminum . . .. 59

II Planes Corresponding to Surface Bands
and Boundaries Between Subgrain Rows . 147

III Etch Pit Measurements: Grain #1 .... .149

IV Etch Pit Measurements: Grain #2 .... 150

V Etch Pit Measurements: Grain #3 .... .151

VI Etch Pit Measurements: Grain #10 . .. 152

VII Etch Pit Measurements: Grain #5 .... .153

VIII The Creep Testing Conditions for the
Evaluation of the Microstructural
Evolution During Creep . . . ... 169

IX The Conditions of Creep Testing and
Secondary Creep Rates . . . . .. 171

X Assumed Activation Energies, AH;
Stress Sensitivity Constants, n;
Intercepts, log A; and Sums of the
Squares of the Deviations,
E(log ZHi-log'ZHi)2 . . . . . 173

XI Specimens, Creep Conditions, and
Microstructural Parameters . . . . 178

XII The Rates of Subgrain Boundary
Formation During the Creep of
Aluminum . . . . . . . . 207

XIII Assumed Activation Energies, AH;
Stress Sensitivity Constants, m;
Intercepts, log B; and Sums of the
Suares of the Deviations,
E(log Ci-logC)2 . . . . . . 209


LIST OF TABLES (continued)

Table Page

XIV Specimens, Creep Conditions, Strain,
and fSGR/fSG-SGR. . . . . 235

XV Creep Testing Conditions for the
Preliminary Experiments . . . .. 290

XVI Creep Testing Conditions, Strain Rates
and Microstructural Parameters for
Specimens Tested Under Conditions of
Stepped Variations in Stress and
Temperature . . . . . . .. .319

XVII Creep Tests of Specimen S5-30 . . .. .393

IV-1 Specimens Creep Tested Under
Conditions of Constant Load and
Constant Temperature . . . . ... 465

IV-2 Specimens Tested at Constant Strain
Rate . . . . . . . . ... 467

IV-3 Specimens Creep Tested Under Conditions
of Variable Load and Variable Temperature. 468

VII-1 Lineal Intercept Counts . . . ... 473

VII-2 Counts of Grains Containing Subgrains
and Subgrain Rows, and the Angles
Between Subgrain Rows and the Tensile
Axis . . . . . . . . . 476

_ ---


Figure Page

1 Shear stress versus shear strain for
f.c.c. single crystals . . . . . 53

2 Flow sheet for the fabrication of
Series 1 creep specimens . . . ... 61

3 Flow sheet for the fabrication of
Series 3 creep specimens . . . ... 62

4 Flow sheet for the fabrication of
Series 4 specimens . . . . . . 63

5 Flow sheet for the fabrication of
Series 5 creep specimens . . . ... 64

6 Creep specimen dimensions. . . . ... 65

7 A photomicrograph of the square grid
photoengraved on the sides of some
creep specimens . . . . . .. 66

8 Schematic diagram of creep testing
apparatus . . . . . . . .. 68

9 Creep specimen, grips and hanger rods . 69

10 Mounted metallographic specimens .... 74

11 Automatic mechanical polishing
apparatus . . . . . . . . 76

12 Anodizing cell . . . . . .... 77

13 Schematic microstructure of a creep
strained specimen . . . . ... 81

14 Schematic diagram of smooth and
serrated grain boundary serrations,
based upon the period and amplitude
of serrations . . . . . .... 83

LIST OF FIGURES (continued)

Figure Page

15 Intercepts of a test line with grain
boundaries . . . . . . . 84

16 Surface bands in specimen 4 tested at
2000C under an initial stress of 1,000
psi for 18.98 hours . .. . .... . 92

17 Surface bands in specimen S1-2 tested
at 2000C under an initial stress of
1,000 psi for 24.05 hours to a strain
of 0.049 . . . . . . ..... 94

18 Surface bands in specimen S1-6 tested
at 2000C under an initial stress of
1,000 psi for 12.03 hours to a strain
of 0.032 . . . . . . . . 95

19 Surface bands in specimen S3-6 tested
at 425*C under an initial stress of 150
psi for 4 hours to a strain of 0.016 . 96

20 Surface structure of specimen S3-18
tested at 2750C under an initial stress
of 800 psi for 5.38 hours to a strain
of 0.176 . . . . . . . ... 98

21 Microstructures of creep specimens . .. 102

22 Microstructures of creep specimens
tested at 4250C . . . . . .. 103

23 Microstructures of creep specimens
tested at 3500C . . . . . .. 105

24 Microstructures of creep specimens
tested at 2750C . . . . . . 107
25 Microstructures of creep specimens
tested at 200*C under an initial stress
of 800 psi . . . . . . .... .109
26 Microstructures of fine-grained creep
specimens tested at 200"C under an
initial stress of 1,000 psi . . .. 110

LIST OF FIGURES (continued)

Figure Page

27 Microstructures of coarse-grained creep
specimens tested at 2000C under an
initial stress of 1,000 psi .. . . 111

28 Microstructures of creep specimens
tested at 200C under an initial stress
of 1,200 psi . . . . . .... .112

29 Microstructures of creep specimens
tested at a nominal temperature of
1300C . . . . . . . ... 113

30 Microstructures of specimens tested
at room temperature at constant strain
rates . . . . . . . . . 114

31 Microstructures illustrating wavy and
dentate grain boundary serrations . . 119

32 The microstructure of specimen SS-7
illustrating well-defined subgrain rows
in adjacent grains and very sharp,
dentate grain boundary serrations . 122

33 The microstructure of specimen S1-4
illustrating that the boundaries around
a single grain, "A," may be smooth, "S,"
wavy, "W," or dentate, "D." ... . ... 123

34 Position of metallographic specimen
with respect to the geometry of creep
specimens . . . . . . . . 126

35 The microstructure of specimen Sl-8
illustrating the alternating shading
of subgrain rows as observed under
polarized light . . . . . ... 128

36 Photomicrographs showing the subgrain
row pattern in specimen 5-7 ..... . 134

37 The orientation of subgrain rows in
specimen S5-7 . . . . . ... .135

38 Stereographic plots of planes (AB and
AC) corresponding to boundaries
separating subgrain rows . . . . 136

LIST OF FIGURES (continued)

Figure Page

39 Strip-wood model illustrating the three-
dimensional nature of subgrain rows . 138

40 The microstructure of specimen S5-7
showing subgrain rows and etch pits
which indicate crystallographic
orientation . . . . . . . 140

41 Traces of surface bands on specimen S5-8.. 144

42 Subgrain row traces in specimen S5-8 . 145

43 Stereographic projection of the orien-
tation of grain #1 and the planes
determined from the traces of surface
bands and from the traces of subgrain
rows . . . . . . . . . 154
44 Stereographic projection of the orien-
tation of grain #2 and the planes deter-
mined from the traces of surface bands
and from the traces of subgrain rows . 155

45 Stereographic projection of the orien-
tation of grain #3 and the planes deter-
mined from the traces of surface bands
and from the traces of subgrain rows . 156

46 Stereographic projection of the orien-
tation of grain #10 and the planes deter-
mined from the traces of surface bands
and from the traces of subgrain rows . 157
47 Stereographic projection of the orien-
tation of grain #5 and the planes deter-
mined from the traces of surface bands 158

48 Photomicrographs showing the alignment
between surface bands and subgrain rows .164
49 E(log ZHi-log~ZHi)2 versus AH . . . 174

50 Log ZH (ZH = e H/RT) versus log a
(a stress (psi)) . . . . . .. 175

51 Grain boundary area, Sv(GB), versus
time, t. . . . . . . . . 184


LIST OF FIGURES (continued)

Figure Page
52 Grain boundary area, Sv(GB), versus
strain, . . . .(..G . . ..... 192
53 Subgrain boundary area, Sv(SGB), versus
time, t . . . . . . ..... 200
54 E(log Ci-log~Ci)2 versus AH . . .. 210

55 Log C (C, ev(TSGB) eAH/RT)versus
log a (a stress (psi)) . . . .. 211
56 Subgrain boundary area, Sv(SGB), versus
strain, e . . . . . . ... 213

57 Fraction of grains containing subgrains,
fSg, and fraction of grains containing
subgrain rows, fSGR, versus time, t . 221
58 Fraction of grains containing subgrains,
fSG, and fraction of grains containing
subgrain rows, fSGR, versus strain, e . 227

59 Histograms showing the distribution
of the angles between subgrain rows
and the tensile axis . . . . . 236
60 Fraction of the grain boundary area
which is serrated, Sv/Sv(GB), versus
time, t . . . . . . . .... 247
61 Fraction of the grain boundary area
which is serrated, Sv /Sv(GB), versus
strain, e . . .. . . . . ..... 253

62 Fraction of grain boundary area which
is serrated, Sv/Sv(GB4, versus
fractions of grains which contain
subgrains, fSG, subgrain rows, fSGR,
and only subgrains of random
morphology, fSG-fSGR .. ....... . 261
63 The change in grain boundary aniso-
tropy, e (f)-e (i), versus time, t . . 273

64 The change in grain boundary aniso-
tropy, e (f)-e (i), versus strain, e . 281

LIST OF FIGURES (continued)

Figure Page

65 Strain, e,'and temperature, T, versus
time, t, for specimen SS-5 . . . . 291

66 Strain, e, and temperature, T, versus
time, t, for specimen SS5-6 . . .. 292

67 Strain, e, and temperature, T, versus
time, t, for specimen S5-7 . . ... 293

68 Strain, e, and temperature, T, versus
time, t, for specimen S5-9 . . ... 294

69 Strain, e, and temperature, T, versus
time, t, for specimen S5-11 . . ... .295

70 Strain, e, and temperature, T, versus
time, t, for specimen 5-12 ...... .296

71 Strain, e, and temperature, T, versus
time, t, for specimen S5-13 . . ... 297

72 Microstructure of specimen S5-5
illustrating subgrain rows and
dentate grain boundary serrations . . 300

73 Microstructure of specimen S5-6
illustrating wavy grain boundary
serrations and the random morphology
of subgrains . . . . . .... 304

74 Microstructure of specimen S5-7
illustrating subgrain rows and dentate
grain boundary serrations .. ... .307

75 Microstructure of specimen S5-9
illustrating subgrain rows and
grain boundary serrations . . ... 310

76 Microstructure of specimen 5-11
illustrating the subgrain structure
and grain boundary serrations devel-
oped during 24 hours of creep at 300C
under an initial stress of 450 psi . 313

LIST OF FIGURES (continued)

Figure Page
77 Microstructure of specimen S5-12
showing that no observable subgrain
structure was developed during 24
hours of creep at 200C under an
initial stress of 450 psi . . . . 314

78 Microstructure of specimen SS-13
illustrating the subgrain structure
and grain boundary serrations devel-
oped in a stepped temperature creep
test under an initial stress of
450 psi . . . .. .. . . 315

79 Strain, c, and temperature T, versus
time, t, for specimen S5-38 . . .. 326

80 Strain, e, and temperature, T, versus
time, t, for specimen S5-31 . . 327

81 Strain, a, and temperature, T, versus
time, t, for specimen S5-43 . . .. 328

82 Strain, a, and temperature, T, versus
time, t, for specimen 5-45 . . . 329

83 Strain, a, and temperature, T, versus
time, t, for specimen S5-40 . . .. 330

84 Strain, e, and temperature, T, versus
time, t, for specimen S5-41 . . .. 331

85 Strain, e, and temperature, T, versus
time, t, for specimen SS-42.. . . 332

86 Strain, a, and temperature, T, versus
time, t, for specimen S5-44 . . . 333

87 Strain, e, and temperature, T, versus
time, t, for specimen SS-39 . . . 334

88 Grain boundary area, Sv(GB), versus
time, t . . . . . . .... 336

89 Subgrain boundary area, Sv(SGB), versus
time, t . . . . . . . . 340

LIST OF FIGURES (continued)

Figure Page

90 Fraction of grains containing sub-
grains, fSG, versus time, t . . .. 343
91 Microstructures illustrating the
effects of precreep upon subgrain
structure . . . . . . . . 346
92 Fraction of grains containing subgrain
rows, fSGR, versus time, t, for speci-
mens of the first experiment ...... . 350
93 Fraction of the grain boundary aroa
which is serrated, Sv/Sv(GB), versus
time, t . . ... . . . ...... 354

94 Fraction of the grain boundary area
which is serrated, Sv/Sv(GB), versus
fractions of grains containing sub-
grains, fSG, subgrain rows, fSGR, and
the random morphology of subgrains,
fSG-fSGR * . . . . . . 357
95 Histograms showing the distribution
of angles between subgrain rows and
the tensile axis . . . . .... 361

96 Change in grain boundary anisotropy,
e (f)-e (i), versus time, t . . .. 365
97 Laue x-ray diffraction photograph
from specimen SS-38 . . . . .. .368
98 Laue x-ray diffraction photograph
from specimen S5-31 . . . . .. 369
99 Laue x-ray diffraction photograph
from specimen S5-43 . . . . .. 370
100 Laue x-ray diffraction photograph
from specimen SS-45 . . . . ... 371
101 Laue x-ray diffraction photographs
from specimen SS-40 . . . . .. 372
102 Laue x-ray diffraction photograph
from specimen SS-41 . . . . .. 374


LIST OF FIGURES (continued)

Figure Page
103 Laue x-ray diffraction photograph
from specimen S5-42 . . . . ... 375

104 Laue x-ray diffraction photograph
from specimen S5-44 . . . . .. 376

105 Laue x-ray diffraction photograph
6f specimen S5-39 . . . . . .. 377

106 Photomicrographs showing the alignment
of subgrain rows and ledges in the
edge of a creep specimen . . . . 384

107 Photomicrograph showing the ledge
in the edge of a creep specimen
produced by grain boundary shear .... 387

108 Photomicrograph showing surface bands
aligned with shear steps in a photo-
engraved grid . . . . . .. 388

109 Photographs of a three-dimensional
model of a serrated grain boundary . 389

110 Displacement of surface reference
scratches in specimen S5-30 . . .. 394

111 Microstructure of specimen S5-30
after test . . . . . . . . 401

112 Grain boundary shearing as an
adjustment required by a violation
of the condition of contiguity . . .. 417

113 The development of subgrain rows .... .420

114 Three-dimensional lattice bending
and the formation of sub-boundaries . 422

115 Shearing along the boundaries
separating subgrain rows . . . .. 425

116 Three-dimensional view of bi-
directional shearing to produce
dentate grain boundary serrations . 427

117 Block diagram of the schematic repre-
sentation of high temperature creep . .429


LIST OF FIGURES (continued)

Figure Page

VIII-1 Twist boundary formed by rotating
one-half of a crystal with respect
to the other through an angle, p . .. 479
VIII-2 Two-dimensional grain sheared to
form a tilt orientation with respect
to the plane of the boundary interface 481
VIII-3 The formation of a simple tilt
boundary by joining two grains which
have been sheared oppositely . . .. .482
VIII-4 Single tilts (a) and (b), and a
skew (c) of a crystal with respect
to the plane of the boundary
interface . . . . . . .. 483
VIII-5 Geometric difference between a twist
boundary (a) and a skew boundary (b) .485
VIII-6 Combinattons of tilts and a skew
of a crystal with respect to the
plane of the boundary interface . . 486
VIII-7 Classification of tilt boundaries . 487
VIII-8 Combinations of symmetrical and non-
symmetrical tilts in forming a
grain boundary . . . . .... 490
VIII-9 Stereographic projections showing
the differences in the orientations
of the conjoint crystals with
respect to the plane of the boundary
interface . . . . . . .. 494
VII-li0 Stereographic plots of grain bound- -
aries corresponding to those shown
in Figure VIII-8 . . . . .... 498
VIII-11 Stereographic plot of the boundary
formed between two simple cubic
crystals by double symmetrical tilting 504
VIII-12 Stereographic plots of the orienta-
tions of conjoint grains relative to
the plane of the boundary interface .506

Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


Richard G. Connell, Jr.
March, 1973

Chairman: Dr. Frederick N. Rhines
Major Department: Materials Science and Engineering

In polycrystalline metals it is well recognized that
the microstructures which develop during the course of hot
deformation are more inhomogeneous than those produced by
equivalent amounts of deformation at ordinary temperatures.
As the microstructure evolves during the course of hot defor-
mation from that of an initially annealed state to that of
a typical hot deformed structure, the grain boundaries, as
observed in metallographically prepared surfaces, assume
paths which alternate sharply back and forth across their
originally smooth contours. With large amounts of strain,
the grain boundaries attain a distinctively dentate form.
A detailed investigation of the microstructural evolu-
tion during the hot creep of aluminum was made to determine
the relations that exist between the formation of dentate
grain boundary serrations and the mechanisms involved in

high temperature creep. Quantitative metallographic tech-
niques were employed to evaluate the microstructures pro-

duced by increasing amounts of creep strain at several

temperatures between 1000C and 425C under various initial
applied stresses from 150 to 4,000 psi. The microstructures
produced during creep were retained for metallographic eval-
uation by quenching the creep specimens from the temperature
of deformation while the loading was maintained.
Concurrent with the evolution of the microstructure in
aluminum during high temperature creep, subgrain rows grad-
ually form within grains until all of the grains contain
subgrain rows. When the rows are first formed, the crystal-
lographic misorientation across the boundaries separating
the rows of subgrains is small; with the progress of defor-

mation it increases until the sub-boundaries represent mis-
orientations equivalent to those of ordinary grain boundaries.
When the misorientation across the sub-boundaries becomes
large, and subgrain rows exist in grains on opposite sides
of common grain boundaries, the grain boundaries take on an
appearance which is dentate in form. Late in the progress

of creep, shearing occurs along the boundaries separating
subgrain rows.
Based upon the results of this research, and upon con-
siderations of the constraints to deformation imposed by the
presence of grain boundaries, it is proposed that the dentate
grain boundary serrations formed during the hot creep of

aluminum are the result of macroscopic shearing which occurs

along the boundaries of subgrain rows. This shearing is

cooperative across grain boundaries.

Shear along the boundaries separating subgrain rows

and formation of dentate grain boundary serrations commences

concurrently with the onset of tertiary creep. Further,

sub-boundary shearing is viewed as a recovery process.

Therefore, it is proposed that an additional amount of

recovery associated with sub-boundary shearing contributes
to an accelerating creep rate and the beginning of third-

stage creep.




The way in which a polycrystalline metal responds by

plastic deformation to an enduring stress at an elevated

temperature depends upon the way in which the microstructure

is altered during the course of deformation. The tempera-

ture and the stress control the microstructural evolution;

and, in turn, the microstructural evolution controls the

plastic response of the metal.
Initiation of the present research was prompted because

little is known of the mechanisms by which metals deform at

high temperature except that there exist important differ-

ences between the plastic response at high temperature and

that at ordinary temperature. It is generally agreed upon

that metals at high temperature deform rather more inhomo-

geneously than at ordinary temperatures. Further, the

great rate-sensitivity of hot working metals may be ex-
plained partially by the competition that develops between
work hardening and thermal recovery. The effects of inhomo-

geneous deformation and thermal recovery are both manifested,
at least in part, by changes in the microstructure during

deformation. Therefore, a logical approach to investigate

the mechanisms by which metals deform is to monitor the

microstructural evolution during the course of hot deforma-
tion and to study its interrelations amongst the plastic

response of the metal and the parameters of the deformation:
stress, temperature, strain and strain rate.

1.1. Scope of the Research

A major portion of this research is confined to a
study of the microstructural evolution which takes place
concurrently with the high temperature creep of high purity

aluminum. A special emphasis has been placed upon discover-
ing the mechanism by which dentate grain boundary serrations
form during hot working. From the outset it was believed
that such an approach would lead to a significantly more
complete understanding of hot deformation in terms of the
shear which results in microscopic shape changes and the
nature of the internal disorder left behind by the various
hot deformation processes. As a result of the present
research, these objectives have been realized.

1.2. The Scientific Background
for the Research

The current research, confined primarily to a study of
the microstructural evolution during the course of high
temperature creep, has required a background of scientific
knowledge in several metallurgical subject areas. In addi-
tion to the obvious topics dealing with the historical

background and the phenomenological nature of creep, six
other subjects are reviewed in the ensuing sections. These
six subjects -- the microstructural features of slip, the
microstructural features of deformation bands and kink
bands, the formation of subgrains, the formation of grain
boundary serrations, grain boundary sliding* and dynamic
recovery -- are all important to the research because each

in some way is connected to the interrelation between micro-
structural evolution during creep and the mechanical response
of a metal to the various parameters of deformation: stress,
temperature, strain and strain rate.

1.2.1. A Brief Historical Review

The beginning of a scientific approach to the subject
of hot deformation dates back to just before the turn of the

century and the discovery of the crystallinity of metals.
Among notable contributors to an understanding of the crys-
talline nature of metals were Stead [1], Desch [2], Humfrey

[3], Rosenhain and Humfrey [4] and Osmond and Cartaud [5].
Since this early work, the concepts of deformation have
progressed through three overlapping periods: 1898 to the
present, crystallographic slip; 1910 to the early 1920's,
the amorphous metal hypothesis; late 1940's to the present,

*The term "grain boundary sliding" is used in this disser-
tation in order to maintain consistency with the literature;
however, the author believes the term to be misleading in
that it implies that shear occurs along the grain boundary
interface rather than in a layer of material adjacent to
the interface.

microstructural and atomistic views of deformation. The
concepts of crystallographic slip began development with

MUgge's [6,7,8] study of slip lines in relation to crystal-
lographic planes. In the face-centered-cubic metals --

copper, gold, silver and lead -- he found that slip followed
(111) planes. Simultaneously, Ewing and Rosenhain [9,10]

discovered slip in lead and proposed block movements which
accompany deformation.
The amorphous metal hypothesis had its beginning in the

work of Beilby [11], who, upon studying the earlier work of
Lord Rayleigh on polished metal surfaces, first ascribed a

hardening effect to a highly distorted structure. The
amorphous metal hypothesis was extended by Rosenhain along

with several other investigators [4,12-17] to the structure
of grain boundaries. The observation that grain boundaries
resist deformation was taken as evidence for the presence
of a hard, deformation resistant structure cementing
together individual grains in an aggregate. Rosenhain and
Ewen [14] found that intergranular fractures could be pro-

duced in lead, tin, gold and bismuth under very small
stresses at temperatures from 30 to 200C below the melting
point. The authors believed that such an observation could
be accounted for only by a layer of amorphous material along
the grain boundary surfaces. At very high temperatures such
an amorphous film would behave in a viscous manner when sub-
jected to small stresses. Further support of this concept

was obtained when "intercrystalline fractures" were

observed in low carbon steels which had been ruptured at

elevated temperatures [4].

Chappell [18] in his study of transcrystalline frac-

tures in low carbon steels postulated that what often

appeared as intergranular fractures at high temperatures

may in fact be the result of fracture along paths separating

grain fragments. Thus, the suggestion was that grains be-

came fragmented during hot deformation and that these frag-

ments were cemented together by an amorphous cement. The

rupture was presumed to result from fracture through the

amorphous cement.

In 1916, Howe [19] summarized the state of knowledge

concerning the types of motion which occur within poly-

crystalline metals during deformation. Howe's classifica-

tion of the types of motion is given as follows:

Types of Motion


Noncrystalline Movement of irregular
grain fragments
Crystalline Annealing (broad twins)
Twinning Mechanical (Neumann
lamellae or narrow

In making the distinction between fluid motion and block

motion, Howe made the following statement in connection

with fluid motion:


So with steam expanding in a cylinder, with
glass on the pontil, clay on the potter's wheel,
and putty and dough when kneaded, ubiquitous
stress causing ubiquitous and almost necessarily
irregular shear. [19, p. 293]
Block movement, on the other hand, was believed by Howe to

consist of the motion of whole blocks of structure which in

themselves remained intact throughout relative movement with
respect to one another. Jefferies [20] proposed the concept
of the equicohesive temperature in relation to the viscous

behavior of grain boundaries at high temperature. The equi-

cohesive temperature was defined as that temperature above
which the cohesion in amorphous regions adjacent to grain

boundaries was less than that of crystalline metal. Hence,
above the equicohesive temperature viscous flow along the
grain boundaries would occur. In relation to the behavior
of grain boundaries during deformation at elevated tempera-
tures, the viscous nature of an amorphous grain boundary
cement persisted to the early 1920's and the beginnings of
x-ray diffraction.

The advent of dislocation theory, the early application
of the electron microscope to the study of metals [16], the
confirmation of the existence of subgrains and Mott's [21]
ideas of polygonization ushered in the current period of

research in which deformation is investigated primarily in
terms of the microstructure and dislocation mechanisms.

1.2.2. The Phenomenolojical Nature of Creep
The first creep tests at ordinary temperature for long
duration (33months) were made by Vicat [22]. Mechanical
properties of metals at high temperature and under varying
rates of loading became of considerable interest at the
turn of the century. Hanson [23] contributed important
information in showing that at elevated temperatures coarse-
grained aluminum was more creep resistant than fine-grained
metal. That creep occurs in three stages was demonstrated
by McVetty [24], who also proposed a mathematical formula-
tion of creep curves. Primary creep
In the first stage of creep, called primary creep, the

creep rate diminishes with time, indicating that hardening
processes are active. Andrade [25] found for primary creep
that the creep strain is proportional to the one-third root
of the time according to the equation:

1 = l (1 + Otl/3) ek (1)

where 1 is the measured length of the specimen at time t.
1 is the initial length of the specimen, t is the time and
B and k are constants. Such an expression has been found to
be generally applicable in expressing primary creep behavior.
Other equations have been proposed to express the primary
creep strain in terms of time and/or the instantaneous
strain rate. Some of the important contributors in this

area are Cottrell and Aytekin [26], Garofalo [27] and
de Lacombe [28]. Their equations are given in Appendix I.
It is now generally held that the hardening developed
during primary creep is related to substructural changes
that take place [29,30]. The development of a cell struc-
ture and the "entanglement of dislocations" presumably con-
tribute significantly to hardening. As primary creep pro-
gresses the original dislocation "entanglements" disperse
and a well-defined subgrain structure begins to form. The
subgrain boundaries formed by the in situ precipitation of
dislocations (polygonization) into walls result in a more or
less stable substructure. Polygonization, at least in metals
with high stacking fault energies, aluminum for example,
requires that dislocations must climb to surmount che stress
fields of dislocation pile-ups and glide to stable positions
in the subgrain boundary as proposed by Mott [21]. Secondary creep
Second stage, or secondary creep,is defined as that
creep deformation which proceeds at a substantially constant
rate. This implies that the rate of thermal softening
(recovery) is just great enough to balance the rate of work
Several empirical mathematical expressions are of
common use in expressing the dependence of the secondary
creep rate upon stress and temperature. McQueen [31], in a
review paper on deformation mechanisms in hot working, has

summarized these expressions. For low stresses, the power

S A N e(-AH/RT) (2)

where e is the strain rate, a is the stress, AH is the
apparent activation energy, R is the gas constant, T is the
absolute temperature and A and N are constants, is of gen-
eral applicability in fitting creep data, but is not suit-
able for the results of hot working. A modified power law,

S- A' an (3)

where e is the strain rate, a is the stress, n is a function
of temperature and A' is a constant has been used to fit
secondary creep data and has often been used for the results
of hot working experiments. The reciprocal of n increases
linearly from about 0.04 to 0.2 as the temperature increases
from 0.55 Tm to 0.9 Tm, respectively.
An exponential equation of the form

S= A" e(Ba) e(H/RT) (4)

where e, a, AH, R and T have their usual meanings, and A"
and 8 are constants, is of use in analyzing the results of
creep and hot working data at high stresses. Deviations
from linearity in plots log e versus a at high stresses;
however, create difficulties in the determination of appar-
ent activation energies.

A more general expression, the hyperbolic sine equation,

i A"'[sinh(cc)]n' e(-AH/RT) (5)

where a, s, AH, R and T have their usual meanings and A"',
a and n' constants, can be applied to creep and hot working
data over wide ranges of strain rate and stress. At high
stress, equation (5) is a good approximation of the exponen-
tial equation, equation (4), and at low stress, it is a good
approximation to the power equation, equation (2).
In comparing creep data at different stresses and tem-

peratures, the Zener-Hollomon parameter,

Z i eH/R (6)

where e, AH, R and T have their usual meanings, has been
widely employed. By plotting the Zener-Hollomon parameter,
Z, versus the stress, a, on logarithmic scales, a straight
line is obtained, the slope of which is the exponent, N, in
the power law equation, equation (3).
Methods for analyzing combined primary and secondary
creep data have been considered by Conway and Mullikin [32].
They proposed an equation of the form

e = a + bt1/3 + ct2/3 + dt (7)

where e is the strain rate, t is the time, and a, b, c and
d are constants. This equation, applied to the creep data

from arc-cast tungsten tested at 24000C under a stress of
800 psi, proved to better express the creep strain as a
function of time than did the equations of Andrade [25],
de Lacombe [28], Cottrell and Aytekin [26] or Garofalo [27].
Secondary creep (constant strain rate) demands that
there exist a balance between work hardening and recovery.
The behavior of the subgrain structure during secondary
creep has been taken as an indicator of the rate of recovery.
There is evidence that the subgrain size developed during
steady state creep is strongly stress dependent [33-35].
Barrett et al. [36] have indicated that the subgrain size
is constant during the course of steady state creep. Sub-
grain size is not importantly related to temperature as long
as the temperature is above that required for recovery [33,
34]. That the subgrain size remains constant implies that
lattice dislocations must be continuously precipitated in
subgrain boundaries.
Metals with high stacking fault energies polygonize
readily during creep; metals with low stacking fault ener-
gies reduce their energy and soften through recrystallization
[37]. It seems reasonable to expect recrystallization to
occur when the driving force for recrystallization cannot
be lowered at a sufficiently rapid rate by other recovery
processes. Lead, for example, under certain creep condi-
tions apparently recrystallizes almost continuously during
secondary creep [38,39].

Under extremely small stresses, it has been proposed

by Nabarro [40] and Herring [41] that, when the temperature
is sufficiently high, creep proceeds primarily by the stress
directed diffusion of vacancies. According to the model pro-

posed, plastic flow results when material is transported by
self-diffusion from grain boundaries under compressive

stresses to boundaries under tensile stresses. Harper et al.

[42] indicate that low stress creep arises from a process

involving the motion of dislocations. These investigators
found that at low stresses the creep rates were about three
orders of magnitude greater than could be accounted for by
Nabarro-Herring diffusional creep. Further, similar creep

behavior was observed in single crystals, which, according
to the diffusional creep model, would not occur because of
the lack of grain boundaries. The authors believed that the
motion of jogged screw dislocations could well account for
the creep behavior of aluminum under very small stresses.
At stresses below 13.5 psi the creep rate in aluminum was
found to increase linearly with the applied stress. Tertiary creep
Third stage, or tertiary, creep is characterized by a

period of accelerating elongation prior to rupture. The
accelerating strain rate has been most commonly related to
a reduction in the cross-section of the specimen, either by
necking or by the presence of internal cavities developed
first at grain boundary quadruple points or triple lines.

Third stage creep and its accelerating creep rate was corre-
lated to a reduction in cross-section by Andrade [43].
Early investigations by Howe [44] of the strength of creep
tested copper and silver showed that low temperature creep
produced irreparable damage to the metal. Even annealing
subsequent to creep testing did not repair the damage sus-
tained during creep. Therefore, structural changes in the
metal itself during secondary creep are suspect in leading
to a change in the response of the metal to its loading, and,
hence, to the onset of tertiary creep.

Nemy and Rhines [45]j in studying the creep behavior
of aluminum alloy 52S-0 (5052-0), found that creep at 4000C
under a stress of 1,500 psi produced no internal cavities
even when creep had progressed well into the third stage.
Some specimens were subjected to creep testing long enough
to produce a well-defined neck, remachined to remove the
neck and again creep tested under identical conditions of
temperature and stress. When these specimens were retested,
the creep rate proceeded at an accelerating rate much as if
the initial test had been continued. This experiment pro-
duced conclusive evidence that neither necking of the speci-
men nor internal cavities were responsible for the onset of
tertiary creep. Tensile tests of similar specimens after
creep showed a progressive loss in tensile strength. The
authors pointed out that these results were reminiscent of
Howe's observations, and support the concept of "creep

damage." At the present, the nature of the microstructural

features which are connected with "creep damage" are not

1.2.3. The Microstructural Features of Slip

Since the early work of MUgge [6-8] and Bwing and Rosen-

hain [9,10] on the relationships between slip and deformation,

the concepts of slip have evolved through the amorphous metal

hypothesis to the current concepts of dislocation theory.

Slip is manifested in two forms as observed on the

polished surfaces of deformed metal crystals:
1. Slip bands, observable at low magnifications,
have spacings and displacements of the order
of one to ten microns. In appearance they
may be straight, wavy, forked, continuous or
2. Slip lines, whose observation usually re-
quires the magnification and resolution
capabilities of an electron microscope,
represent spacings and displacements on the
order of several angstroms to one micron.

Heidenreich and Shockley [46] were the first to estab-

lish that slip bands consist of lamallae of fine slip lines.

It has been observed that, with progressive deformation,

coarse slip bands develop from a fine slip structure [47-49].
In aluminum deformed at slow strain rates in the tem-

perature range 200 to 3500C, Wood and Rachinger [50] have
demonstrated that as the temperature of deformation is
raised, the slip bands observed on polished surfaces become

less sharply defined, more wavy in appearance and fewer in

number. Coincident with these changes, x-ray diffraction

studies showed that at the lowest temperature (200C) the
structure developed was like that of a cold worked metal;
at an intermediate temperature (2000C), a subgrain structure
developed; and at the highest temperature (3500C), a sub-
structure approaching the size of individual grains developed.
Servi and Grant [51], investigating the slip band spacing
on the surfaces of creep deformed aluminum specimens, found
that the average spacing between slip bands decreased linearly
with increasing stress and was independent of the temperature.
In an earlier research, Hanson and Wheeler [52] had
observed that aluminum deformed in creep for 252 hours at
250*C under a stress of 2,800 pounds per square inch elon-
gated without the formation of coarse slip bands. On the
basis of this observation, they postulated that deformation
occurred by uniform slip too fine to be observed with the

optical microscope. McLean [48,53], some twenty years later,
followed up this postulate and studied in detail the surface
markings on aluminum deformed in creep. Both interference
and phase contrast optical microscopy were employed. The
major findings of this investigation were
1. The existence of fine slip lines between
coarse slip bands was confirmed.

2. The extension due to prominent slip con-
tributed less than one-half to the total
extension. Most of the remaining extension
was ascribed to fine slip.
3. The number of prominent slip bands and the
amount of slip within each band increases
with increasing grain size and/or stress.

4. The fraction of the total extension con-
tributed by prominent slip bands increases
with increasing grain size and with in-
creasing stress.

5. A connection between prominent slip bands
and fine slip bands was emphasized; the
former were simply those positions where
fine slip occurred in an intensified

In the face-centered-cubic metals, slip deviates from
the ideal, the major deviation being cross-slip [54]. Slip
lines corresponding to slip on a {111) plane are ideally
straight; however, they sometimes follow zig-zag steps when
the slip is locally altered to a conjugate {111) slip plane.
The two slip planes share a common slip direction. Cahn [54],
Ogilvie and Boas [55], Gifkins [31] and McLean [48] have
observed cross-slip in aluminum.
Maddin et al. [56,57] have identified prominent and
intimate cross-slip. Prominent cross-slip usually appears
as a zig-zag pattern, while intimate cross-slip is distin-
guished by the visibility, at high magnifications, of fine
slip elements connecting the main slip bands. At low mag-
nifications intimate cross-slip often appears as overlapping
discontinuous slip bands.
Cahn [54] has shown that in aluminum low temperatures of
deformation favor intimate cross-slip, whereas high tempera-
tures (500C) favor prominent cross-slip. At room temperature
increasing amounts of extension greatly increases the amount
of intimate cross-slip. The electron microscopic studies of
Trotter [58] on the slip of aluminum during creep confirmed the

existence of cross-slip as previously observed by optical

techniques. Using replica techniques, Trotter found that

slip zones do not consist of parallel slip lamellae as dis-

covered by Heidenreich and Shockley [46], rather they con-

sist of intimate segments of primary and cross-slip. The

directions of such shear zones do not coincide with the

trace of the slip plane, in fact, they may be rather far

removed from it. Slip bands formed at high temperatures

often appear as broad, wavy lines.

A number of researchers [54,59-62] have investigated

the continuity of slip at grain boundaries, and it is appar-

ent that slip cannot proceed across grain boundaries without

a change in direction. The continuity of slip lines at a

boundary observed on a two-dimensional surface is coinci-

dental. In three dimensions, for the continuity of slip

lines to exist at more than a point on a grain boundary sur-

face requires that slip planes in conjoint crystals meet at

the boundary in a line. Such would be the case for simple
tilt boundaries.

The inhomogeneity of deformation as observed by the

inhomogeneity of slip has been pointed out by Urie and Wain
[63] for the plastic deformation of coarse-grained aluminum,

and by Gifkins [39] for the creep of lead. The evidence
presented by Urie and Wain suggests that grain to grain con-

straints play an important role in the way in which individual

grains deform. The strength of grain to grain constraints in

I __

retarding deformation is apparently related to the misorien-
tation between neighboring grains. Gifkins [39] proposed
that the inhomogeneous deformation within grains assists
relative grain movements in lead. High creep rates tend to
favor inhomogeneous deformation within grains. Slip lines

were observed in two or three directions within one grain,
and inhomogeneity was observed as clusters of slip lines
within isolated portions of individual grains.

1.2.4. The Formation of Subgrains
Conclusive proof for the existence of subgrains was
provided by the work of Lacombe and Beaujard [59], in which
etch-figures were used to study deformed polycrystalline
aluminum. The etch-figure study was supported by x-ray
back-reflection patterns, and it was found that subelements
within a single grain were misoriented by small angles,
usually of the order of 15 to 30 minutes.
The formation of subgrain structures during the course
of creep was first observed in iron by Jenkins and Mellor

[64], and since then they have been observed in many other
metals. Evidence for the formation of subgrains during the
course of creep or hot deformation has been found in alumi-
num [65-67], a-iron [64], nickel [37], iron-3.1 percent
silicon [68], zinc [69,70], cadmium [71], magnesium [72,73],
tin [74], lead [75], a tin-antimony alloy [76], noibium [77]
and copper [78,79]. The effect of stacking-fault energy
Metals which form subgrains during the course of creep
or hot deformation are usually those which have high

stacking-fault energies. Hardwick and Tegart [37], for
instance, in comparing the hot deformation microstructures

of aluminum, nickel and lead, found that both aluminum and
nickel formed subgrains and that lead showed no evidence of
a subgrain structure. Metals which rapidly polygonize do
not, in general, recrystallize during creep. Therefore, it
seems that polygonization and recrystallization are competi-
tive processes, and that the relative strengths of each of

these processes is dependent upon the stacking-fault energy.
If the stacking-fault energy is high, polygonization is

favored. The size of subgrains
Recovery by polygonization, a process which requires
prior lattice bending, is regarded as the rate-controlling
mechanism of the high temperature creep of aluminum. The
size of the subgrains has been assumed to be a measure of

the rate of recovery. During primary creep, subgrains form
and attain an equilibrium size which is dependent upon the
temperature and the strain rate [80-83]. The stable sub-
grain size increases with increasing temperature and de-
creases with increasing strain rate, or stress, in the case
of creep. In creep testing, the temperature and the stress
are independent variables; however, they are related in a

unique manner to the strain rate. Jonas et al. [80] claim

that this is true only after a stable subgrain size is

attained, as is the case for secondary creep. Sherby and
Dorn [84] suggested that the size of the subgrains is the

same for a given stress, regardless of the temperature. The
strain rate of secondary creep, then, must assume a value

related to the subgrain size which is related strongly to
the stress and weakly to the temperature. It is the balance

between work hardening and recovery (by polygonization) which

results in a given stable subgrain size. The greater the

rate of working, the greater is the rate of recovery required
to balance the work hardening. The greater the rate of
recovery, the smaller is the stable subgrain size. There-

fore, a stable subgrain size is responsible for the relation-
ship among the strain rate, the stress and the temperature.

Raising the temperature or decreasing the stress, subsequent
to achieving secondary creep, causes the subgrain size to

increase to a new equilibrium value [82,84]. This new
equilibrium subgrain size dictates a new secondary creep
rate. On the other hand, if the temperature is decreased or
the stress is increased, the equilibrium subgrain size becomes
smaller and the steady state creep rate is dependent upon the
smaller subgrain size.
For creep deformation to proceed at a constant rate, a

balance between the rates of work hardening and recovery
must be maintained. If polygonization is the rate-controlling

process, then the presence of a stable subgrain size during
secondary creep implies that dislocations are removed from
the lattice by continuous precipitation in pre-existing sub-

grain boundaries. Such a concept may seem untenable,
because as dislocations are added to subgrain boundaries, the
angular misorientations across the boundaries would increase
and intuitively result in a hardening effect. Thus, the
creep rate would diminish. Two alternatives to this problem
exist: (1) the subgrain size remains constant, but their
boundaries continuously disperse and reform at an equilibrium
spacing [85]; or (2) the subgrain size decreases concurrent
with increases in the angular misorientations of the sub-

McQueen et al. [85], after investigating the hot extru-
sion of aluminum, proposed that the subgrain structure
developed during steady state hot deformation may not be
stable, but only the average size of the subgrains remains
constant. These authors contend that subgrain boundaries
must continuously reform at equilibrium spacing following
their disorganization by a high flux of dislocations. In
addition to finding that the subgrain size remains constant
during deformation, it was proposed that the misorientation
between subgrains is small and remains constant. In this
way a constant dislocation density is maintained during
steady state deformation.

Voznesensky and Rosenberg [86] monitored the size of
subgrains in nickel throughout the course of creep, and
found that as rupture is approached, there is a continuous
refinement of the subgrains. They also discovered that the
degree of crystallographic misorientation between neighbor-
ing subgrains increases with.increasing deformation. Subgrain boundaries
In the last two decades there has been considerable

disagreement over the amount of crystallographic misorienta-
tion which may exist across a subgrain boundary between con-
jugate subgrains. McLean [87-89], Servi et al. [90], Wood
and Rachinger [50] and Ramsey [69] have all favored the
development of large angles of misorientations between adja-
cent subgrains. Misorientations as large as 16.6 have been

measured by McLean [87] in creep deformed aluminum. Small
angles of misorientation have been favored by Garofalo et al.

[91], Hammad and Nix [92] and by McQueen et al. [85]. Green
et al. [93], in tantalum creep tested at 19000C under a
stress of 500 psi up to a strain of 16 percent, found no
misorientation angles between adjacent subgrains greater
than 0.5 degrees.
Walls of edge dislocations have been shown by Cahn [94]
to form upon heating bent aluminum single crystals. The
observation coincides with the views of Mott [21], and leads
to the conclusion that, in the most elementary case, sub-
grain boundaries represent simple tilt misorientations between

subgrains. Cottrell [95] summarizes the characteristics of
subgrain boundaries formed during creep, noting that the
similarities between subgrain structures formed during creep
and those formed upon annealing cold worked metals suggest
that the subgrain boundaries form in simple tilt walls by
climb and glide of edge dislocations. Objections to this
view have been raised by Wood and his associates [50,66,
96,97] and are restated by Cottrell [95] in the following
1. The sub-structures produced during creep form
at a lower temperature than in the usual
polygonization experiments, and appear to
form directly during the deformation without
an intermediate stage during which the lattice
is continuously bent.
2. The sub-boundaries form an irregular, poly-
hedral network instead of parallel planes
with a definite crystallographic orientation.
These objections, based upon the studies of subgrains
formed during creep, appear to point out two conclusions
about the nature of the substructure formed during creep.
First, it seems that the formation of subgrain boundaries
in creep is accelerated by the action of the enduring applied
stress. Second, since the boundaries between subgrains do
not follow definite crystallographic orientations, the
boundaries must in themselves be rather complex in terms
of their dislocation content. Such complexity may result
from the operation of several slip systems within each
grain, and when dislocations from these various slip systems
come together in planes, it is not difficult to understand

that the resultant boundary may be rather complex. A recent
investigation by Orlovi and Cadek [78] on the origin of the

dislocation substructure during the high temperature creep

of copper, shows that under certain creep conditions sub-
grain boundaries are formed not only perpendicular to the
slip direction, but also in the primary slip planes. The

latter boundaries were found to represent twist misorienta-

tion angles of up to 20 minutes. Subgrain rows
Bands or rows of subgrains have been observed by
several investigators [26,48,68,98,99] to form within grains

during the course of creep. McLean [48] has shown quali-
tatively that slip lines observed on the surface of aluminum
creep specimens are related to the internal subgrain struc-
ture. Rows or bands of subgrains correspond to rather
irregularly curved slip lines which lie between very promi-
nent slip bands. The subgrains, separated by prominent slip
bands, have significantly different crystallographic orien-

tations, as observed by polarized light microscopy, on
opposite dides of the prominent slip bands. Bands of sub-

grains were observed in aluminum specimens crept at 200C
under stresses in the range of 666 to 1,500 psi to total
extensions as large as 50 percent [87]. The bands of sub-
grains were described as polygonization bands and were
thought to coincide with previously formed deformation bands
with rather poorly defined boundaries. McLean [26] proposed

that the boundaries of deformation bands representing local-
ized bending provided the necessary sites for subgrain-like
boundaries to form through the climb and glide of disloca-
tions. Ramsey [98] observed in creep deformed aluminum that
subgrains formed within kink bands resulting in a banded
structure of subgrains. When no prominent slip bands were
observed on the surface of specimens (slow rates of deforma-
tion) subgrains did not appear preferentially in bands.
Lytton et al. [68], in studying the creep behavior of (100)
[001]-oriented polycrystalline iron-3.1 percent silicon,
found that when the orientation of the grains with respect
to the tensile axis was favorable for the predominate oper-
ation of a single slip system, dislocation pile-ups formed
in rows and with concurrent recovery during. creep. The
development of polygonized boundaries in parallel rows re-
sulted. Initially dislocation pile-ups and polygonized rows
formed adjacent to grain boundaries. The parallel bands of
dislocation walls were aligned with primary <111> slip
directions, whereas the polygonized rows were formed perpen-
dicular to these slip directions. Commercial purity aluminum
deformed at 410C at a strain rate of 2.7x10" sec1 and
subsequently annealed at the same temperature for nine
seconds developed elongated grains with bands of subgrains.
The bands of subgrains were oriented at about 45 degrees to
the tensile axis [99]. Serrations in the grain boundaries
were observed at positions where the subgrain bands'

intersected the grain boundaries. Further, Evans and
Dunston [99] found that the subgrain bands were extremely

stable during further annealing after deformation. This,
no doubt, was due to the stability of the boundaries which
separated the subgrain bands. These boundaries were found

to represent misorientations of 7 or 8 degrees, while the
boundaries separating subgrains along the length of the

bands represented misorientations of only about one-half

1.2.5. The Microstructural Features of
Deformation Bands and Kink Bands

The appearance of deformation bands is associated with

turbulent flow. Intuitively, it seems that easy glide should
give way to turbulent flow, when a crystal attains an orien-
tation such that two or more slip systems are stressed
approximately to the same level [100]. Results on brass
confirm this behavior; however, in aluminum easy glide ceases
long before symmetrical orientations are reached. Deforma-
tion at elevated temperatures should tend to further restrict
the occurrence of easy glide in aluminum because of the
thermal activation of secondary slip systems; thus, turbu-
lent flow should predominate. Barrett and Levenson [101],
in studying the structure of polycrystalline aluminum after
compression, discovered parallel bands extending across
entire grains. When the bands first became visible they
were straight, and the orientation difference across the
boundaries of a band was on the order of only a few degrees.

As the amount of deformation increased, the bands became
curved and the orientation differences across their bounda-
ries increased. Such bands are not to be confused with
twins, because the band boundaries tend to increase misori-
entation with increasing deformation. The material on each

side of a band boundary has a cube plane in common, and the
bands follow (100) planes or possibly some other planes in
the [001] zone. Further, these authors observed that the
bands apparently have different widths when observed under
different conditions of illumination, a phenomenum which
they ascribe to a more or less continuous range of orienta-
tions across narrow regions at the boundaries of bands.
This feature of deformation bands was further supported by
Gay and Honeycombe [102] when they observed asterisms in
Laue patterns in cases where the incident x-ray beam
straddled a deformation band. Honeycombe [103,104] demon-
strated that the lattice is rotated within deformation bands
formed in aluminum.
The primary features of deformation bands as discovered
by Honeycombe [103,104], Cahn [54] and Chen and Mathewson
[105] are as follows:
1. After only a few percent of strain, bands
appear on the surfaces of crystals as waves
with sharp crests. As the amount of
deformation increases, the bands become
forked and less regular.
2. The lattice within bands is rotated with
respect to the lattice outside the bands.

3. In slightly deformed crystals, the plane of
the deformation band is invariably perpen-
dicular to a <110> slip direction to within
a few degrees. As the deformation is increased,
the plane of the band becomes rotated away from
its original position by an angle, 8. The
angle, 0, is approximately equal to the rota-
tion of the crystal lattice of the band. The
axis of rotation of the band lies in the slip
plane normal to the slip direction.
4. Most of the deformation bands form at the
start of deformation and persist throughout
the process, increasing in rotation as the
amount of deformation increases.

5. The initially formed slip lines cross deforma-
tion bands, altering direction only to the
extent of the lattice curvature across the
band. Slip lines formed later in deformation
do not cross deformation bands. Short seg-
ments of slip bands on intersecting slip
systems are formed within the bands.
6. Deformation bands represent inhomogeneous
deformation. Material on either side of a
deformation band may differ in strain by
as much as 5 percent.

7. Crystals which deform (in tension) in double
slip from the beginning of deformation do
not form observable deformation bands.
Barrett [106] found that when iron crystals
were compressed with their <100> or <111>
axes parallel to the compression axis, no
bands resulted.

8. Specimens deformed by pure bending or shear
do not form deformation bands.

9. Grain boundaries have no apparent effect
upon the formation of bands.

A double fiber texture, <111>+<001>, has been produced

in extruded single crystals of aluminum by Reed and McHargue
[107]. The resultant texture had a strong <111> component

and a weak <001> component; their relative strengths depended

upon the temperature of deformation and the initial

orientation of the single crystal. The authors attributed

the observed effects of temperature and initial orientation

upon the resultant texture to the formation and interaction
of deformation bands.
Recently Mitchell et al. [108] and Ahearn et al. [109]

have described a new kind of deformation band in copper-
aluminum solid solution alloys. The experimental observa-

tions establish that polyslip-oriented single crystals form
deformation bands when they undergo avalanches of single

slip during an abrupt relaxation process.
Cottrell [100] cites the tendency to use the term "kink

band" in place of "deformation band," because of their
crystallographic similarities. A kink band represents oppos-
ing localized lattice bending at the parallel boundaries of

the band which, according to Hess and Barrett [110], results

from the gradual and progressive rotation of the lattice.
Kink bands are usually associated with the hexagonal-close-

packed metals, and are viewed as being produced by organized

slip on many parallel planes. The plane about which the
kink forms is called the "kink plane." It is normal to the

slip direction and symmetrically oriented with respect to
the slip planes on either side of the sharp bend.
Cahn [54] has produced evidence of subgrains within

deformation bands, and Mott [21] proposed that the bounda-
ries of deformation bands are preferred sites for polygoni-

1.2.6. The Formation of Grain Boundary Serrations
Grain boundary serrations, the dentate or wavy offsets
produced in grain boundaries during the course of hot defor-
mation, have been observed in a wide variety of metals and
alloys deformed at elevated temperatures over a range of
strain rates encompassing those commonly encountered in
creep and hot deformation. The metals which reportedly
form grain boundary serrations are nickel [111-114],
nickel-copper alloys [112], nickel-aluminum alloys [112],
nichrome [114], austenitic stainless steel [115], magnesium
[73,116], uranium [117], lead [39], aluminum [53,87,118-121],
aluminum-magnesium alloys [120,122], an aluminum-20 percent
zinc alloy [123] and niobium [77].
The main characteristics of grain boundary serrations
and the hot working conditions under which they form are
generally agreed upon. There exists a temperature range of
hot working in which serrations form. Below a certain cri-
tical temperature, usually associated with that required for
recovery, serrations either do not form, or they are of such
a small size that they are unresolvable by the ordinary
techniques of optical microscopy. Above a rather higher
temperature they become destroyed by recrystallization during
the course of hot working [111]. On the other hand, Gifkins
[39] believes that grain boundary serrations formed in lead
may be the results of recrystallized grains formed along
grain boundaries. Serrations become more sharply defined
with increasing temperatures and increasing amounts of

deformation [111,113,115,124] within the temperature inter-
val of their formation. The appearance of grain boundary
serrations vary depending upon the temperature and the rate
of deformation [113]. Low temperatures and fast rates of
working favor sharp serrations with more or less straight
sides, while higher temperatures and slow rates of working
favor a less sharp form of serrations which impart a wavy
appearance to the grain boundary. Smeal [114] has reported
that serrations become less pronounced as the grain size is
The size and/or spacing of serrations along a grain
boundary have been associated with the slip spacing [120],
or with the subgrain structure [51,119,123,124]. Photo-
micrographs in the papers by McLean [53,87,118] show an
association between grain boundary serrations and subgrains
in creep deformed aluminum. This observation was also made
by Presland and Hutchinson [73], Brumner and Grant [120],
Evans and Dunston [99] and Brinson and Argent [77]. That
there exists an association between grain boundary serra-
tions and subgrains has led some researchers [39,120] to
attribute the formation of serrations to the migration of
grain boundaries. When a newly formed subgrain boundary
impinges upon a grain boundary, the angles between the grain
boundary and the subgrain boundary are nearly 90 degrees.
The surface tension of the subgrain boundary causes the
migration of grain boundary segments from their original
positions towards the line of the subgrain boundary, resulting

in more nearly equilibrium angles at the juncture between

the subgrain boundary and the two grain boundary segments.
This mechanism accounts for the observation that the apex

of a serration is often joined to a subgrain boundary.
Walter and Cline [121] attribute the formation of serra-
tions to grain boundary migration into the shear zone adja-
cent to a shearing grain boundary, but they do not connect

the migration with the existence of subgrain boundaries.

Mullendore and Grant [124] subscribe to the concept that

grain boundary migration results in serrations. They have
been led to the conclusion, by the studies of grain boundary

sliding in polycrystals and bicrystals of an aluminum-2

percent magnesium alloy, that grain boundary displacements
can be the result of slip crossing the grain boundary.
Sliding along the boundary between two conjugate grains

represents an unresolved component of shear when slip in
one grain changes direction in crossing the grain boundary.
When such a process occurs in the vicinity of a serration

peak, a dislocation accumulation is built up on one side of
the boundary. The result of the dislocation accumulation

produces complex strain fields around the serration peak
and causes the grain boundary to migrate, thus altering the
form of the serration peak. Even a slightly irregular
boundary represents an unstable condition in the presence

of grain boundary sliding, and growth, or at least altera-
tions, of the serrations should occur as long as appreciable

grain boundary sliding occurs. By this model, the authors
have avoided the necessity of requiring the surface tension

of subgrain boundaries to provide the driving force for

grain boundary migration.
Rows or bands of subgrains were observed in aluminum

by Evans and Dunston [99] and in niobium by Brinson and

Argent [77]. The original grain boundaries were serrated

at intersections of subgrain rows or bands with the bounda-

ries, the size of the serrations being equal to the width

of the subgrain bands. It is of particular interest that

the bands of subgrains observed by Evans and Dunston were

oriented at nearly 45 degrees to the tensile axis.
A model of crystal deformation during creep has been

proposed by McLean [87] and extended [118] to show that

crystal deformation controls grain boundary shear. Rows of

subgrains can be accounted for by this model. The operation
of the model is best described by the following sequence of

1. Initially the lattice planes are flat.

2. During primary creep, some dislocations
approach the boundary and, in becoming
immobile, produce deformation bands.
Others form deformation bands elsewhere
in the crystal.

3. During the course of secondary creep,
polygonization occurs first near the
grain boundaries, then in whole crystals.
Sub-boundaries form along the boundaries
of deformation bands, resulting in what
McLean calls "polygonization bands."


4. Late in secondary creep, subgrain boundaries
form within the "polygonization bands,"
segmenting the bands into subgrains; hence,
subgrain bands.
5. Further deformation causes the subgrains
within the "polygonization bands" to rotate
with respect to each other, resulting in
shear along subgrain and grain boundaries.
The model does not account for the spasmodic behavior of

grain boundary sliding, nor for the formation of grain

boundary serrations. Had the author considered the possible

occurrence of long range shear along the boundaries between
adjacent polygonization bands, then, at least in principle,
the formation of grain boundary serrations would have been

1.2.7. Grain Boundary Sliding
Sliding upon mutual interfaces of metal crystals was

reported as early as 1913 by Rosenhain and Humfrey [4] for
a-iron. Observations of grain boundary sliding led Andrade
[43] to believe that grain boundaries possessed a capacity
for viscous sliding which could be at least partly responsi-
ble for the constant strain rate behavior of secondary creep.
Some two decades after its demise in the mid-1920's, K8 [125]
revived the amorphous metal hypothesis in studying the stress
relaxation across grain boundaries, an effect which he
ascribed to a viscous behavior of the grain boundaries. He
concluded from his investigations that the viscosity of
grain boundaries corresponds to that of a supercooled liquid
metal. However, McLean and Farmer [74] pointed out that the

viscosity of liquid metals under supercooled conditions
would result in grain boundary sliding rates many times
greater than those observed experimentally.
Wood et al. [67] were the first to conclude that

adjacent subgrains may move relative to one another by
sliding along their mutual boundaries. Their conclusion

was based upon the following three observations:

1. Subgrain boundaries become observable as
surface.relief between neighboring
2. X-ray reflections from subgrains show dis-
placements after further deformation.
3. Scribe marks across subgrain boundaries
become displaced as the result of shear
along subgrain boundaries.
The authors also concluded that the shape of grains may

change during hot deformation by doformation within sub-
grains as well as by internal block movements of the sub-
grains relative to one another. McLean [87] also presented

evidence that subgrain boundary sliding does occur.
Grain boundary sliding has been linked with grain

boundary migration and crystal deformation. In high purity
aluminum, Chang and Grant [119] observed spasmodic sliding

of grain boundaries and connected the rest periods between
sliding periods with grain boundary migration. Sliding
along grain boundaries was found to occur through a thick-
ness of material adjacent to, and on both sides of, the
grain boundary. It was concluded that grain boundary slid-
ing causes localized deformation within grains as evidenced

by "folds" at grain boundary triple junctions opposite the
shearing boundary, and by the appearance of subgrains whose

boundaries intercept the boundaries which have sheared.

The strain produced in primary creep was attributed primarily
to initial grain boundary sliding.

Gifkins [39,126], investigating the creep of lead, also

associated grain boundary sliding with grain boundary migra-

tion. He presumed that, in addition to grain boundary migra-
tion, inhomogeneous deformation within grains also aided

grain boundary sliding. Because sheared grain boundaries
in lead often exhibited steps, Gifkins believed that alter-
nate shearing and migration of the boundary occurred. The contribution of sliding
to total deformation

The first attempt to discover the contribution of grain
boundary sliding to the total creep extension was made by

Rachinger [127]. In creep tested aluminum, internal grains

elongated very little while the surface grains elongated to
a marked degree. Rachinger attributed the effect to grain

boundary sliding. That the internal grains remained equi-
axed suggested that grain boundary sliding was the predomi-
nant mode of creep deformation within the bulk of the speci-
men. The exterior surfaces of the specimen apparently

suffered little deformation from grain boundary sliding.
The contribution of grain boundary sliding to total extension
was found to be greater the lower the stress and the higher

the temperature, up to a temperature above which extensive

grain growth occurred (350*C). Grain boundary shear may

account for as little as 6 percent or as much as 95 percent

of the total deformation, depending upon whether measure-

ments were made at the surface or within the bulk of the

creep specimen, respectively.
Davies et al. [128] measured the contribution of grain

boundary sliding to the total creep strain in high purity

(99.97 percent) polycrystalline gold. Measurements made on
the surface of creep specimens tested at 395*C showed that

the contribution was as high as 67 percent at a stress of
2,225 psi. The contribution decreased linearly with increas-
ing stress at this temperature to about 8 percent at a stress

of 5,650 psi.
The work of Rachinger [127] has been reconfirmed by.

Davies et al. [129] who found, in nickel, that the contribu-
tion of grain boundary sliding to the total strain was

smaller at the surface than in the interior of creep speci-

mens. Additional experiments with gold, nickel, copper
and aluminum showed that the ratio of the contribution of

grain boundary sliding at the surface to that measured in
the interior of creep specimens was always less than unity.
The ratios obtained were as follows: aluminum, 0.67;
nickel, 0.82; copper, 0.89; and gold, 0.93. The smaller
the ratio, the greater is the difference between the surface
and interior measurements. All specimens were tested at the

same homologous temperature and within the same order of

magnitude of the strain rate. The authors suggest that the

variations in the contribution of grain boundary sliding to
the total strain reported in the literature may be due, in

part, to surface effects in large-grained specimens.

Typical values of the contributions of grain boundary slid-
ing to the total creep strain are: aluminum, 0.15 [74];

nickel, 0.35 [65]; copper, 0.50 [130]; and gold, 0.67 [128].
Measurements of the contribution of grain boundary
sliding to the total deformation in aluminum creep tested
between 3370C and 474C, made by Fazan et al. [131], indi-
cated that the contribution may be between 65 and 95 percent
of the total creep strain. The value of grain boundary
sliding obtained for the same total strain was found to be

independent of temperature. The apparent activation energy

for grain boundary sliding was found to be between 35 and
38.5 Kcal/mole, a value which agrees with earlier determina-
tion [34,132,133]. Since the apparent activation energy for
grain boundary sliding is nearly equal to that for the creep
process as a whole, it was concluded that the rate control-

ling processes for grain boundary sliding were identical with
those for creep.
In several papers [48,53,87,118] McLean had shown that
the relative contribution of grain boundary sliding to the
total strain was independent of the temperature at a constant
stress, but decreased with increasing stress at a constant

temperature. In this regard, the results of Fazan et al.
[131] were in total agreement with those of McLean. Harper
et al. [42], in studying grain boundary sliding during the
creep of aluminum, found that the ratio of grain boundary
sliding to the total strain was not only independent of the
temperature, but also independent of the strain. Contrary
to these results, Brunner and Grant (120] discovered that
with increasing temperature, the contribution of grain
boundary sliding to the total elongation begins near the
equicohesive temperature, reaches a maximum, then decreases
with further increase in temperature or decrease in strain.
They also found that the contribution of grain boundary
sliding to the total strain is insensitive to grain size
over a rather wide range of grain sizes. In contradiction
to the work of Harper et al. [42], the ratio of the strain
due to grain boundary sliding to the total strain was mark-
edly affected by the amount of strain, the ratio being
greater in the early stages of creep and diminishing with
increasing strain. Moreover, Brunner and Grant [120] con-
nected grain boundary sliding with grain boundary migration,
and suggested that the formation of grain boundary serra-
tions impedes the progress of grain boundary sliding as the
strain increases. Barrett et al. [134] suggest that, in the
creep of copper, an increase in the secondary creep rate as
the grain size is decreased below 0.1 mm in average diameter
may be attributable to an increased contribution of grain
boundary sliding to the total creep strain.

..:::.'.~..: f::~:~.

McLean [48,53,87,118] and Brunner and Grant [135] have
derived equations for the contribution of grain boundary
sliding to the total extension. McLean's derivation is
based upon a statistical average of the amount of shear on
many grain boundaries and is applicable to fine grain mate-
rials, whereas the equation derived by Brunner and Grant
considers contributions for single grain boundaries.
Stevens [136] has reviewed the equations used by various
researchers [48,127,128,131,137-141] to express the contri-
bution of grain boundary sliding to the total creep exten-
sion. The author indicates that the previously derived
equations usually involve doubtful assumptions, are not
rigorously derived or are often in serious error. However,
Rachinger's equation and the way in which he applied it
[127] seems to be the most suitable and easily applied for
the measurement of grain boundary contributions at the sur-
faces of creep specimens. Mechanisms of grain boundary sliding
In the last two decades, the studies of grain boundary
sliding have produced a great variation in the observed
characteristics of the sliding process. Consequently, the
mechanisms proposed for grain boundary sliding also are
varied. The following paragraphs review briefly some of the
more prominent studies and the proposed mechanisms.
It has been postulated by Feltham [142] that grain
boundary sliding is dependent upon crystal deformation; and

that this deformation, in turn, is dependent upon recovery.
Recovery was viewed by the author to consist of the "melting"
of regions of good crystallographic fit between neighboring
grains. The regions of good fit were considered as barriers
to both dislocation motion and grain boundary sliding as
first proposed by Mott [143]. The concept of "melting"
coincides with viscous behavior during grain boundary sliding.
Crussard and Friedel [144] suggested that grain boundary
sliding takes place along the boundary by the movement of
partial dislocations in planes parallel to the boundary.
Dislocations approaching a grain boundary which acts as a
barrier to dislocation motion dissociate into partial and
move parallel to the boundary. Such a concept supports the
observation that sliding occurs within a thickness of mate-
rial adjacent to the boundary. On the other hand, it appears
generally untenable because in the high stacking fault metals,
such as aluminum, partial dislocations do not readily form.
McLean and Farmer [74] have proposed that in grain
boundary sliding there exists a viscous flow such that as
the temperature decreases,the viscosity of the grain boundary
increases. However, in considering the viscosity of a super-
cooled liquid metal layer at the grain boundary, they
pointed out that the layer of supercooled liquid would pre-
dict a sliding rate many times faster than those observed
experimentally. To overcome this difficulty, it was con-
sidered that sliding may be inhibited at low temperatures by

the formation of ledges in the grain boundaries produced by
plastic deformation of the grains. Such ledges were viewed
to be large on an atomic scale. In order for sliding to
proceed, these ledges must be sheared through. At high tem-
perature, crystal slip was presumed to aid grain boundary
sliding. At low temperatures, where the deforming stresses
are high, the number of active slip sources increases sharply;
but the proportion of these sources which assist sliding pre-
sumably decreases. Thus, at low temperatures grain boundary
sliding was not believed to be controlled by crystal deforma-
tion, because most of the slip goes into forming ledges in

the grain boundary which has a retarding effect upon grain
boundary sliding.
Macroscopic grain boundary sliding is the result of
many small-scale shears, each of approximately the same mag-
nitude, according to the ideas of Couling and Roberts [137].
Macroscopic shear is at least partially blocked by the
elastic behavior at grain boundary triple junctions. Elas-
tic recovery of the macroscopic shear in the reverse sense
was thought to be prevented by grain boundary migration
which was regarded as a recovery mechanism. Once such a
recovery process takes place, the boundary is again free to
shear. Hence, the concept that grain boundary sliding occurs
intermittently and alternately with boundary migration was
A part of the graip boundary sliding, that which results
from activated sites near the grain boundary, was attributed

by Tung and Maddin [145] to a component of displacement
parallel to the boundary surface. The activated sites

correspond to regions properly oriented for crystal shear
in a direction parallel to the boundary. Regions of poor
crystallographic fit between the grains were believed to
constitute barriers to boundary shear. These authors pro-
posed that the regions of poor fit are removed by diffu-
sional processes which they identified as grain boundary

migration. The idea of poor fit described by Tung and
Maddin appears to be counter to the proposals of Mott [143].
Martin et al. [146], from a study of grain boundary
sliding in 0-brass, concluded that grain deformation was the
controlling factor in the rate of grain boundary sliding.
In addition, they demonstrated that the creep rate at high
temperatures was controlled by the resistance of grains to

deformation, rather than by the resistance of grain bounda-
ries to shear.
Grain boundary sliding in oriented tricrystals was
investigated by Weinberg [147]. It was confirmed that the
rate of sliding is dependent upon the grain boundary mis-
orientation angle. As the angle of misorientation increased,
so did the rate of sliding. Boundaries with less than five
degrees of misorientation did not shear. Grain boundary
shear was observed to be confined to one side of the bound-
ary and it occurred in successive layers adjacent to the
boundary, each layer being further removed from the boundary.

The author attributed this observed behavior to grain bound-
ary migration. It was also observed that sliding occurred
without overall deformation of the grains; however, as
sliding progressed, the regions adjacent to the grain
boundaries became rather severely deformed.
Gifkins [148], incorporating various features of several

of the aforementioned mechanisms for grain boundary sliding,
suggested a mechanism based upon the existence of subgrain
boundaries adjacent to shearing boundaries and upon the
movement of dislocations by climb and glide in close prox-
imity to the grain boundary. Dislocations from within the
subgrains adjacent to the grain boundary either precipitate
in the subgrain boundaries or pile up against the grain
boundary. The dislocations in the pile-ups move toward the
boundary by climb and glide and in the process create a
deficiency of vacancies. To replenish the vacancy deficiency,
vacancies diffuse from the boundary toward the climbing dis-
locations, thus establishing a vacancy flow along the bound-
ary toward the climbing dislocations. Vacancy flow in the
grain boundary is suggestive of a high degree of disorder
or possibly as "melting" as proposed by Mott [143]. A
vacancy flow in the grain boundary is regarded as causing
relaxation of the boundary which allows grain boundary slid-
ing to progress. After climb, dislocations are more uni-
formly distributed along the grain boundary, a condition
which Gifkins believed was favorable for grain boundary

migration. Migration was thought to occur once sliding had

By applying a direct shear stress to the boundaries in
aluminum bicrystals, Biscondi and Goux [149] investigated
the characteristics of grain boundary shear. They found

evidence which suggests that two regions of behavior exist

with respect to the initiation of sliding, depending upon a

critical temperature or stress. For a given stress (or tem-

perature) there exists a critical temperature (or stress)
below which sliding begins only after an incubation period.

The critical conditions of stress and/or temperature which

separate the two regions of behavior depend upon the purity
of the metal and the angular misorientation between the con-
jugate grains. The higher the purity and the smaller the
misorientation angle, the greater is the critical tempera-

ture (or stress) for.a given stress (or temperature). Grain
boundary sliding was observed to occur in alternate periods
of rapid displacement and rest. Grain boundary migration
was associated with the rest periods and was found to vary

in degree from point to point along the boundary. For a con-

stant stress, the logarithm of the initial sliding rate

decreases linearly with the inverse of the absolute tempera-
ture, within the temperature region where an incubation
period exists. The heat of activation for sliding was found
to be dependent upon the misorientation angle; the smaller
the misorientation, the greater the heat of activation.

Horton et al. [150] observed that grain boundary slid-
ing occurs in zinc bicrystals in the temperature range 3200

to 3900C within the existence of slip in both of the con-

jugate grains. Sliding apparently did not occur within a

layer of material adjacent to the grain boundary, but was
rather sharply defined at the grain boundary interface.

Because a band of subgrains was observed next to sheared

grain boundaries, the authors suggested that, during the
course of sliding, dislocations are ejected from the bound-

ary into the conjugate grains. Grain boundary migration was
observed and its direction and degree appeared to be associ-

ated with the subgrain structure. It was proposed that the
rate of grain boundary sliding depends upon the ease with
which dislocations can be absorbed by the conjugate grains.
As the substructure develops adjacent to the boundary during
creep, the rate of dislocation absorption becomes increas-
ingly slower; thus, the rate of grain boundary sliding

becomes diminished.
Walter and Cline [121] investigated grain boundary slid-

ing in aluminum at 3150C using constant strain rate (10-4
sec-1) tensile tests. Sliding occurred continuously, and
when the amount of deformation became great enough for defor-
mation bands and subgrains to be clearly observed, sliding
ceased. Migration of the grain boundaries occurred spas-
modically, and in such directions as to accommodate the
changing stresses created by the grain boundary shear.

Migration also ceased about the same time as did grain
boundary sliding. The authors believed that even slight
curvatures of grain boundaries might result in shear zones
adjacent to grain boundaries and provide a driving force for
grain boundary migration.
When grain boundary sliding occurs along boundaries
which are not planar -- and, indeed, grain boundaries are
seldom perfectly planar -- it is reasonable to expect that
some other process might be occurring concurrently with
sliding. Raj and Ashby [151] have considered some of these
other processes, which they call accommodation processes, in
detail. Underlying their work is the assumption that it is
the accommodation process which controls the rate of slid-
ing, rather than any intrinsic property of the boundary it-
self. The accommodation may be purely elastic, diffusional,
or it may involve plastic flow by dislocation processes.
Regardless of the accommodation process, there exists some
upper limit to which it can operate. When this limit is
attained, sliding along irregular boundaries ceases. The
particular nature of the accomnadation and its limit of
operation then control the rate and extent of sliding,
Slide hardening in aluminum has been linked to the
occurrence and extent of crystal deformation by Beevers
et al. [152]. Small additions of copper to aluminum de-
creased the slide hardening effect and promoted increased
rates of grain boundary sliding.

Horton [153], as the result of grain boundary sliding

experiments with bicrystal specimens of zinc, aluminum and

aluminum containing small additions of copper in solid

solution, has reported several important features of slide
hardening. It was found that:

1. The rates of sliding and crystal creep defor-
mation have the same time and temperature
dependencies, but behave differently with
respect to the stress and to alloying copper
with aluminum.

2. If a power law stress dependence of the creep
rate is assumed ( =. on), then the exponent n
for sliding is greater than unity but less
than that for crystal deformation.
3. At low stresses, where sliding occurs without
macroscopically observable crystal deforma-
tion, sliding is not affected by additions of
copper to aluminum.
4. At intermediate shear stresses, copper addi-
tions to aluminum decrease the rate of
crystal deformation and increase the rate
of sliding.
5. Tests made at constant shear rates indicate
that the rate of sliding is linearly propor-
tional to the shear stress developed, and
the ratio of the amount of sliding to the
shear stress increases with increasing tem-
From these observations it was concluded that grain boundary
sliding is not controlled by crystal deformation in any
unique way. Horton proposes that grain boundary sliding is
controlled by slide hardening, a process which is apparently
independent of the rate of crystal deformation.
McLean [154] has developed a two-dimensional model for
grain boundary sliding which considers the rate of sliding

in terms of dislocations meeting a grain boundary and then

moving along the boundary by a series of climb and glide
processes. Coincident with these processes, vacancies are
either emitted from or absorbed by the boundary during the

course of sliding. The model also takes into account the

orientations of conjugate grains and the orientation of the

boundary surface with respect to the tensile axis. Appar-
ently, the assumption of continuous sliding was made. This
is contrary to the findings of a majority of other investi-

Another recent model for high temperature low stress

creep has been proposed by Stevens [155]. The creep strain
is attributed to grain boundary sliding and diffusional
transport. None of the strain results from dislocation
mechanisms. Grain boundary sliding is viewed as an accommo-
dation mechanism which maintains continuity of the metal at

grain boundaries in addition to contributing large amounts
of strain to the total creep strain. Calculations based

upon the model predicts that 60 percent of the total
strain results from grain boundary sliding.
Much of the variation in the observed characteristics

of grain boundary sliding and in the mechanisms based upon
these characteristics, no doubt, arises because of differ-
ences in investigative techniques. In spite of the wide
variation, there exists some majority agreement on several
points listed below:

1. Grain boundary sliding proceeds spasmodically
in alternate periods of rapid displacement
and periods of rest.
2. Grain boundary sliding is controlled in degree
and overall rate by the rate and extent of
crystal deformation.

3. The misorientation between conjugate crystals
influences the average rate of grain boundary
shear. The greater the misorientation, the
greater is the rate of sliding.
4. Grain boundary sliding occurs within a layer
of materials adjacent to the boundary inter-

5. Grain boundary migration accompanies grain
boundary sliding.
6. The contribution of grain boundary sliding to
total creep strain is significantly large.

That grain boundary migration is associated with slid-
ing as part of the same phenomenon seems most doubtful.
Grain boundary migration has been observed in polycrystalline

specimens, in tricrystals and in bicrystals which have been
stressed so as to apply bending moments or stress concentra-

tions to the grain boundary. Therefore, it is likely that
grain boundary sliding has not been effectively isolated
from the operation of other processes by such experiments.

The classic paper of Rhines et al. [156] on the subject
of grain boundary sliding in aluminum bicrystals describes
the essential characteristics of the phenomenon as it occurs
essentially free of other processes. A summary of these
characteristics is given in the following statements:

1. There exists an incubation period prior to
the onset of sliding. Short induction periods
are associated with large orientation differ-
ences between the conjugate grains, high

temperatures and large stresses. The onset
of sliding is abrupt regardless of the
length of the induction period.

2. Grain boundary sliding occurs spasmodically
in alternating periods of rapid shear and
periods of rest. Such behavior is not
influenced by temperature surges or mechani-
cal shock.

3. As sliding progresses, the displacement
surges become smaller and the rest periods
become longer.
4. The direction of sliding coincides with the
direction of the maximum resolved shear
stress in the plane of the grain boundary.
5. Sliding occurs in a layer of material adja-
cent to and only on one side of the grain
boundary. Subgrains formed within this zone
are rotated relative to each other through
greater angles than elsewhere.

6. Sliding is not uniform along the full length
of the grain. It may occur not only to
different extent, but also at different
times along the boundary.

7. The average grain boundary displacement is
proportional to the cube root of time.
8. Grain boundary sliding follows an Arrhenius
behavior as does the creep of individual
9. The overall rate of sliding increases with
the angular misorientation between grains.

10. Grain boundary migration was not observed.

11. Crystal deformation precedes the onset of
grain boundary sliding and continues through-
out the course of creep testing.

Based upon these observed characteristics of grain

boundary sliding, the authors proposed that during the incu-
bation period, slip, impinging on the boundary surface,

causes localized bending all along the boundary. At some

point, enough energy has been stored for a recovery process
to occur. When the recovery takes place, the layer of mate-
rial adjacent to the grain boundary becomes softer. When
this layer of material coincides with a direction of high

resolved shear stress, yielding parallel to the grain bound-
ary occurs until the material has again become hardened.
This process continues throughout creep in a repetitive


1.2.8. Dynamic Recovery
Dynamic recovery is that portion of recovery operative
during deformation which is strongly stress dependent. At

low temperatures it is the predominate form of recovery;
while at higher temperatures, usually above T/Tm = 0.5, it

is superimposed upon the thermally activated recovery pro-
cesses. The thermally activated recovery processes are those
which are responsible for the softening obtained prior to
recrystallization in the heating of prestrained metals.
The concepts of dynamic recovery in terms of disloca-
tion mechanisms have been described thoroughly in three

papers: Seeger [157], Seeger et al. [158], and Haasen [159].
Based upon these papers, a brief description of dynamic
recovery as it applies to aluminum follows.
Dynamic recovery at ambient temperatures is associated
with the onset of Stage III deformation as shown in the
shear stress versus shear strain curve for face-centered-
cubic single crystals, Figure 1. The high stresses attained



8 I i



I i


Sheat Strain

Figure 1. Shear stress versus shear strain for F.C.C.
single crystals.

at the end of Stage II, which represents work hardening,

cause slip dislocations to surmount obstacles, thus reducing

the rate of work hardening. At room temperature, aluminum

exhibits no Stage I and very little Stage II behavior, indi-

cating that dynamic recovery occurs almost at the beginning

of deformation. That aluminum exhibits recovery at temper-

atures well below those required for thermally activated

recovery processes is attributable to its very high stacking-

fault energy. In terms of dislocation behavior, a high

stacking-fault energy means that dislocations do not readily

dissociate into partial; therefore they may cross-slip with


Two processes have been postulated for dynamic recovery

in face-centered-cubic metals: cross-slip of screw dislo-

cations and the collapse of Lomer-Cottrell dislocations
which act as barriers to screw dislocation movement. The

former process, that of cross-slip, is favored because
theoretical calculations show that the activation energy for

cross-slip is always smaller than that for the collapse of
Lomer-Cottrell dislocations. Further supporting evidence

for the cross-slip process based upon surface observations

of face-centered-cubic metals deformed into Stage III has
been provided. From the onset of Stage III, the amount of

cross-slip observed at the surface continuously increases
with increasing deformation. Such a theory explains why

aluminum at room temperature begins Stage III at very low

stresses. These low stresses are too small to account for
a significant number of dislocation pile-ups, hence work
hardening (Stage II) does not occur.
In contrast to single crystals, dislocation pile-ups
and attendant work hardening do form in polycrystalline

aluminum deformed at room temperature. Evidence that these

pile-ups do exist after deformation is observed by poly-
gonization during heating subsequent to deformation. This
implies that grain boundaries offer barriers to dislocation
movement, and that the barriers must be of such a nature as
to impart bending to the lattice. A similar situation, as
previously described in Section 1.2.4, occurs in the pure
bending of single crystals which polygonize upon heating

With reference to Seeger's paper [157], it has been
proposed that cold working into Stage III may result in near
tilt boundaries of edge dislocations when edge dislocation
segments are left behind in the cross-slip plane by the
annihilation of opposing screw dislocation segments.
Parallel sets of such edge dislocation segments represent
bending which, with addition deformation, may form the
parallel boundaries of deformation bands. Elastic inter-
actions amongst the edge dislocations would then tend to
align them into well-formed tilt boundaries. Geometrically,
these tilt boundaries would be much like boundaries formed
through thermally activated polygonization. In summary, if

a metal recovers easily at the temperature of deformation
by the cross-slip process, it is then possible for rather
sharp sub-boundaries to be formed without the operation of

thermally activated recovery processes.
During the course of high temperature creep, dynamic

recovery is superimposed upon the more commonly thought of
thermally activated recovery process of polygonization.
Because of the ease with which aluminum polygonizes during
high temperature deformation, it has been the recent trend

of some researchers [29,31,82,85,94,160-163] to attribute
dynamic recovery to polygonization as it arises from the
thermally activated processes. Mukherjee et al. [29] have
considered cross-slip to be included in dynamic recovery
processes in that, along with dislocation climb, it may
play a role in the development of subgrain boundaries through

The proposals of McQueen [31] and McQueen et al. [85,
161] that the subgrain boundaries developed during the hot
working of aluminum are not stable, but go through repeated
processes of dispersion and reformation, "repolygonization"

[85], may be yet another process of dynamic recovery.
It has been shown that recovery during hot working [80]
or under the action of an applied stress during the heating
of prestrained specimens [84,164] produces greater softening
than can be accounted for by static recovery.

Embury et al. [165], in studying the substructures of
various cold worked ferrous alloys and copper, have found
that the face-centered-cubic metals do not harden as readily

as do the body-centered-cubic ones, a fact they attribute'to

dynamic recovery during working.

Langford and Cohen [166,167] have attributed a portion
of the dynamic recovery in the drawing of iron wires to dis-
location annihilation and a loss of cell wall area to cell
coalescence. The motion of the cell walls under the applied

stresses accounts for the annihilation of dislocations.
In commercial purity aluminum (1100) deformed by rapid
compression at temperatures between 4000 and 5000C, McQueen
and Hockett [168] have observed that the subgrain structure
developed during deformation resists subsequent recrystal-
lization. This observation suggests that the degree of
recovery during deformation was large; thus, there is little
remaining driving force for recrystallization.



The experimental techniques employed in this research

have been many and varied. Those techniques which were

used most widely are described in detail in this chapter,

while those used only for special experiments are described

in appropriate sections of Chapter III, Observations and

2.1. The Material

High purity aluminum was chosen as the material for

this research, because it is easily fabricated into test

specimens, it is easily prepared for metallography, it

requires no protective atmosphere at high temperatures, it

readily develops grain boundary serrations during hot

deformation and it deforms without forming twins. The

metal was obtained in the form of twenty-five-pound pigs,

intended as remelt stock, from Consolidated Aluminum Com-

pany, Jackson, Tennessee. An analysis furnished by the
vendor is shown in Table I.

The level of the impurities was believed to be low

enough not to greatly affect the microstructure developed

during hot deformation.


Table I

Chemical Analysis of Aluminum

Element Weight Percent

Al 99.9980

Si 0.0003

Fe 0.0009

Cu 0.0001

Mg 0.0005

Ca 0.0002


2.2. Fabrication of Creep Specimens

Creep specimens were fabricated from the as-cast

aluminum by milling slabs from the pigs, rolling in three

stages with intermediate anneals, machining the specimens,

grinding the surfaces, and final annealing. Several series

of specimens were prepared for creep or tensile testing in

this way with some variations as outlined in the flow sheets
given in Figures 2 through 5.

After the final anneal each specimen was electropolished

and anodized so that lineal intercept counts could he made

for the determination of the initial grain boundary area

and initial grain boundary anisotropy. Formulae of solu-

tions for electropolishing and anodizing are given in Appen-

dix II. Anodizing rendered the microstructure sensitive to

polarized light which facilitated the lineal intercept count-

ing. A drawing showing the dimensions of the creep speci-

mens is given in Figure 6.

Following the quantitative metallographic counting, the

specimens were again electropolished to remove the anodized

film. A photoengraved grid was etched on the sides of some
specimens as reference marks for observations of surface
deformation. A spacing of 1.01x10-2 inches separated the
lines of the square grid. Appendix III gives the details
of the photoengraving process using commercially available

materials. A photomicrograph of the photoengraved grid is

shown in Figure 7.

0.650" Slab milled from ingot.
Cold roll.
0.440" I
Anneal 1 hour at 4000C, air cool.
Cold roll.
Anneal 1 hour at 4000C, air cool.

Cold roll.


Machine creep specimens and grind
surfaces on 180, 320 and 600 grit
SiC papers.

I .

S1-1 through Sl-5

Anneal 1 hour at 400C,
furnace cool to 2000C,
hold for 6-1/2 hours,
air cool.

S1-6 through S1-9

Anneal 1 hour at 5000C,
furnace cool to 2000C,
hold for 9-1/2 hours,
air cool.

Electropolish (Electropolishing Solution #1)
Quantitative Metallography (Grain Boundary Area)
Electropolish (Electropolishing Solution #1)

Figure 2. Flow sheet for the fabrication of Series 1
creep specimens.

0.650" Slab milled from ingot.
Cold roll.
0.440" I
Anneal 1 hour at 400C, air cool.
Cold roll.
0.210" .-
Anneal 1 hour at 4000C, air cool.
Cold roll.
0.135" I
Machine creep specimens and grind
surfaces on 180, 320 and 600 grit
SiC papers.

Anneal 1 hour at 4000C, air cool.

-Specimens Anneal 12 hours at 4250C, air cool.
S3-1 through S3-6
-Specimens Anneal 12 hours at 3500C, air cool.
S3-7 through S3-12
-Specimens Anneal 12 hours at 275"C, air cool.
S3-13 through S3-18
-Specimens Anneal 12 hours at 200*C, air cool.
S3-19 through S3-24
-Specimens Anneal 12 hours at 125C, air cool.
S3-25 through S3-30

Electropolish (Electropolishing Solution #2)
Quantitative Metallography (Grain Boundary Area)
Electropolish (Electropolishing Solution #2)

Figure 3. Flow sheet for the fabrication of Series 3
creep specimens.


0.650" Slab milled from ingot.
Cold roll.
0.440" I
Anneal 1 hour at 4000C, air cool.
Cold roll.
Anneal 1 hour at 4000C, air cool.
Cold roll.
0.135" I
Machine creep specimens and grind
surfaces on 180, 320 and 600 grit
SiC papers.

Anneal 2 hours at 4000C, air cool.

Electropolish (Electropolishing Solution #2)

Figure 4. Flow sheet for the fabrication of Series 4

0.625" Slab milled from ingot.
Anneal 3 hours at 4000C, air cool.
Cold roll.
Anneal 1 hour at 400*C, air cool.
Cold roll.
0.210" I
Anneal 1 hour at 400C, air cool.
Cold roll.
0.135" I
Machine creep specimens and grind
surfaces on 180, 320 and 600 grit
SiC papers.

Anneal 2 hours at 4000C, air cool.

Electropolish (Blectropolishing Solution #2)
Quantitative Metallography (Grain Boundary Area)
Electropolish (Electropolishing Solution #2)

Figure 5. Flow sheet for the fabrication of Series 5
creep specimens.

0 Oso


Figure 6. Creep specimen dimensions. Dimensions
are in inches.

r-. -I-- --.- ~ ur~mnrcr-~-r-.~*~,;/ -rr~ nrrurr.l+r;~r*l.tI~ nm~

Figure 7. A photomicrograph of the square grid photo-
engraved on the sides of some creep specimens.

2.3. Creep Testing

2.3.1. Creep Testing Apparatus
Air furnaces for creep testing, constructed in the
laboratory, feature two-layer non-inductive windings of
18 gage Kanthal-A resistance wire wound on impervious
mullite tubes. The inside dimensions of the furnace were
2 inches in diameter by 24 inches in length. A side viewing

port at midlength of the furnace tube permitted observation
of the specimen during creep testing. Both ends of the
furnace and the viewing port were open to the air. A sche-
matic diagram in Figure 8 shows the methods employed for
supporting and loading the specimen and the provisions made
for quenching the specimen at the termination of a creep
test. Stainless steel all-thread rod was used for upper and
lower hanger rods. Specimen grips were also fabricated from
stainless steel. Hanger rods, grips and a specimen are shown
in the photograph of Figure 9. The specimens were loaded
into the furnace from the bottom by pulling the assembly of
both hanger rods, grips and specimens up into the furnace
with an overhead chain. The upper hanger was supported by a
clevis block resting on two steel angles above the furnace.
Loading was accomplished by hanging a three-pound coffee can
containing the desired weight of lead shot on the lower
hanger rod. After loading the specimen a quench tank was
placed beneath the weight and filled with tap water.


Upper Hanger Support

L and N Millitemp
Nodel 7120
Upper Hanger

r** Resistance Windings

SViewing Port
110 l Thermocouple

Control Thermocouple

Specimen Grips

Lower Hanger eight L and N Speedomax G

--a-c o -c Quench Tank

Figure 8. Schematic diagram of creep testing apparatus.

.' ~ ~ ` ',FVmfl r. aj". 'f:~ll4': "C' '~;R-~)-r..rrrs ; lv,.rf/..t4



Figure 9. Creep specimen, grips and hanger rods.

The temperature of the furnace was controlled by a

Leeds and Northrup Model 7120 Millitemp controller connected

to a chromel alumel thermocouple. Measurement of the speci-

men temperature was made using a chromel-alumel thermocouple

in contact with the specimen surface. The thermocouple was

connected either to a Leeds and Northrup Model 8690 Poten-

tiometer or a Leeds and Northrup Speedomax G recording poten-

tiometer. The specimen temperature was maintained within

5SC of the desired temperature. It was found that the tem-

perature along the length of the specimen varied no more

than 5C.

2.3.2. Creep Testing Sequence

Two types of creep tests were made using the apparatus

described in the preceding section; one in which the tem-

perature and load remained constant throughout the duration

of the test, and one in which either or both the temperature

and the load was altered during the test. The sequence of

steps in performing tests under constant temperature and

S constant temperature was as follows:

1. The furnace was heated to the desired temper-

2. The specimen was loaded into the furnace and
allowed to heat to the testing temperature.
3. The load was applied to the specimen.

4. At the end of the test, the upper support was
released and the specimen was lowered with
the load still applied into the quench tank
by means of the overhead chain.


The sequence of steps in performing tests under conditions

of variable temperature and variable load was as follows:
1. The specimen was loaded into a cold furnace.

2. The load was applied to the specimen.

3. Power was applied to the furnace and the
temperature allowed to rise to and equili-
brate at the set point.

4. To increase or decrease the temperature, the
set point on the controller was adjusted
accordingly and the temperature of the fur-
nace was allowed to increase or decrease to
the new set point.

5. In cases where the load was to be increased
or decreased simultaneously with a tempera-
ture change, the load was increased or
decreased simultaneously with changing the
set point of the controller.

6. At the end of the test, the upper support
was released and the specimen was lowered
with load still applied into the quench tank
by means of the overhead chain.

Quenching specimens under load at the end of a creep test
was accomplished within a time period of one second after
the specimen passed out of the furnace. In cases when rup-

ture of the creep specimen terminated the creep test, the
lower portion of the specimen was immediately quenched. The

time between rupture and immersion of the specimen in water
was calculated to be 0.35 seconds.

2.3.3. Measurement of Strain
Throughout this research the use of the term strain

was taken as the engineering strain, i.e., the change in
length of the specimen divided by the original length. For

creep specimens tested under conditions of constant tempera-
ture and constant load, the strain was calculated from the
change in length between two reference scratches on the
specimen surface. When creep specimens were tested under

conditions of variable temperature and variable load, the

strain was periodically measured using a time-lapse photo-
graphic technique. A motor-driven Nikon Photomic FTN
camera fitted with an extension bellows and a 135 mm lens
was focused on the surface of the specimen in order to

periodically photograph reference scratches on the surface
of the specimen. The camera was activated by a clock-motor

intervalometer set for one exposure per hour. More frequent
exposures could be made by manually activating the camera.

The distances between the images of the reference scratches

on negatives obtained in this manner were measured and the
strain calculated.

2.3.4. The Specimens and Their Testing Conditions

Specimens from Series 1, 3, 4 and 5 were tested under
various conditions of loading and temperature. A complete
list of these specimens and their conditions of deformation

is given in Appendix IV.

2.4. Metallographic Techniques

Metallographic techniques employed in this research were
especially developed to facilitate rapid preparation of
polished surfaces with a high degree of reproducibility and
a minimum of manual effort. Quantitative metallographic
measurements were standardized and statistical methods for
computing tolerance limits were employed. The following
several sections describe the procedures employed.

2.4.1. Metallographic Preparation
Sections about 1-1/4 inches long were cut with a
jeweler's saw from the center of the gage length of the
creep specimens and mounted so that a surface parallel to
one side of the specimen could be ground and polished.
Specially designed specimen mounts, 2 inches in diameter,
were cast from Caulk NuWeld dental plastic. Electrical con-
tact was made with copper alligator clips fastened to the
specimens and cast into the mounting material. A photograph
of a mounted specimen in Figure 10 shows the physical con-
figuration of the mount which was designed to be used with
an automatic mechanical polishing device. Each specimen
was ground on 180, 320 and 600 grit SiC abrasive papers
using a solution of four parts of water and one part of a
commercial liquid hand soap (Franklin's 40 % Liquid Hand
Soap or Spence Liquid Hand Soap No. 40). The specimens were
polished using a slurry containing one level tablespoon of

Figure 10. Mounted metallographic specimens.

AB 1200 mesh emery, 300 ml commercial liquid hand soap
(Franklin's 40% Liquid Hand Soap or Spence Liquid Hand Soap
No. 40) and 900 ml of water on an AB microcloth. In both
grinding and polishing, a specially constructed automatic
polishing device mounted on a Geoscience Unipol polisher
was used. The polishing device utilized a 20 rpm synchron-

ous motor and friction drive to rotate the specimen about
an axis normal to the surface of the polishing wheel. A
1,400 gm weight resting on the specimen mount applied a
dead load to the specimen. The speed of the polishing wheel
was adjusted to about 150 rpm. The abrasive slurry was
dripped onto the polishing wheel just ahead of the specimen
position. The essential features of the polishing apparatus
are shown in the photograph in Figure 11. About twenty
minutes polishing time after grinding was required to pro-
duce a highly polished abrasive-free surface. Following
mechanical polishing, the specimens were electrolytically
polished and anodized in a polycarbonate cell using the solu-
tion for anodizing listed in Appendix II. The specimen to
cathode distance was 1 inch. A photograph of the anodizing
cell is shown in Figure 12. In addition to leaving behind
an anodized film which renders the microstructure sensitive
to polarized light, the anodizing procedure also lightly
electropolished the surface, removing any residual fine
scratches left behind by mechanical polishing. A major
advantage of this metallographic preparation is that the

Figure 11. .Automatic mechanical polishing apparatus.

Figure 12. Anodizing cell.

prepared surfaces remain suitable for metallographic exami-
nation for several days 'without special storage precautions.
If for any reason a specimen required repolishing, the hard,
abrasion-resistant anodized film was first removed by immers-
ing the specimen for 1 minute in an aqueous solution contain-
ing 10 volume percent of hydrofluoric acid.

2.4.2. Quantitative Metallography
For all of the metallographic work in this research,
both photographic and quantitative metallography, a Bausch
and Lomb Research II Metallograph was used. This instrument
is superior to other available instruments for polarized
light examination because of the high contrast afforded by
the employment of a Foster calcite prism which maintains a
precisely fixed relation between polarizing and analyzing
filters. With this sort of design, the specimen must be
rotated in order to affect the various polarized extinctions
due to the variations in surface orientations of the speci-
men. A single drawback to the design, however, is that very
often not all of the microstructural features may be observed
at any one rotational setting of the specimen. For this
reason, it was necessary, during quantitative metallographic
counting, to rotate the specimen to observe all the features,
then reposition the specimen in the proper reference to the
test line or grid to make the necessary counts. Such rota-
tions of the specimen during counting were extremely tedious
and time-consuming because of the care that was necessary to
reindex the specimen after each rotation.

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs