~,\:
m: U...‘
‘

(1,.

,. v.

, 7'. n M
“.qv-wn"
,,.n,

”.4-

.,.. . ‘uJ-l
. c r

 

 

 

. IKHESIS

llllllllllllllllllllllllmlmllllllllllflll|l|ll|ll||ll|l||l||

31293 00885 7116

This is to certify that the

dissertation entitled

Deformation Behavior and Microstructure of TaC, and

Processing of C-TaC Composites

presented by

Chulsoo Kim

has been accepted towards fulfillment
of the requirements for

Ph.D . degree in Materials Science

 

 

 

 

 

Major professor

Date February 19, 1991

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY ‘
Michigan State
University

 

l

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

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
‘ c:\circ\datedue.pm3—p. 1

 

 

DEF ORMATION BEHAVIOR AND MICROSTRUCTURE OF TaC, AND
PROCESSING OF C-TaC COMPOSITES

By

Chulsoo Kim

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Metallurgy, Mechanics and Materials Science

1991

A A ‘+.~ a‘ir \r-sfic. «r

 

 

@«D@ P‘ Jr) (:21

ABSTRACT

DEFORMATION BEHAVIOR AND MICROSTRUCTURE OF TaC, AND
. PROCESSING OF C-TaC COMPOSITES

By

Chulsoo Kim

Deformation behavior of TaCo.99 at room temperature was studied
by means of a combination of microindentation techniques and
transmission electron microscopy. Extensive local plastic deformation
. was observed in the vicinity of microindentations. Plastic deformation of
TaC0.99 at room temperature was mainly accomplished by the motion of
edge dislocations. Long screw dislocations and the wavy character of
screw dislocations suggest the existence of a high Peierls stress for screw
dislocation motion at room temperature. Annealing of the indentation-
deformed TaCo.99 results in strongly recovered microstructures and
recrystallization, comparable to metals with high stacking fault energy.

Deformation behavior of TaCOO99 polycrystals at elevated
temperatures was investigated by compressive creep tests. From the

results on creep tests at intermediate temperatures (1400 ~1500 °C) and

 

 

 

stresses (105 - 170 MPa), the values of the stress exponent (n) and
activation energy (QC) for power-law creep of TaCo.99 were found to be
nzl.8 and Qc~150 KJ/mol. The low activation energy and the low stress
exponent suggest that the creep of TaC0.99 at intermediate temperatures
is mainly controlled by grain boundary sliding. This is further supported
by observation of void formation. The observation of dislocation
interaction in the crystal interior suggests that dislocation glide may act
as a minor contributing mechanism during creep. TEM investigation also
revealed the formation of subboundaries and dislocation networks. Most
dislocations formed during creep of TaC0.99 have no preferential
orientation, in contrast to dislocations formed at low temperatures.

The fabrication of C-fiber reinforced TaC composites was studied
using combinations of powder metallurgy techniques such as cold
pressing, slurry infiltration, sintering, and hot isostatic pressing (HIP).
The characteristics of the interface between carbon fibers and TaC were
investigated by optical microscopy, scanning electron micros00py and
auger electron spectroscopy. No chemical reaction occurred at the
interface between carbon fibers and a stoichiometric TaC matrix. Carbon
fibers suffered damage by indentation from tantalum carbide particles
during cold compaction and HIP. A tantalum carbide coating of C-fibers

prevented this indentation damage.

 

 

 

 

ACKNOWLEDGEMENTS

I would like to thank my parents and other members of my family for their love,
patience and continuous support. Also, I wish to thank Professor Giinter Gottstein under
whose direction this research was performed, for his thoughtful guidance, inspiration and
support over the entire course of my graduate studies. Additionally, special thanks to
Professor David 'Grummon whose scientific interest in my research provided a generous
amount of support and encouragement. Many thanks are extended to the other members of
my guidance committee: Dr. Iwona Jasiuk and Dr. Jack Bass for. their valuable comments
given on my thesis. Thanks are also given to Dr. Martin Crimp for helpful discussions
concerning the transmission electron microsc0pic studies.

Without the help of technical expertise, my , research would not have been
successfully completed. Therefore, I wish to express my sincere appreciation to Dr. Karen
Klomparens, Dr. Stanley Flegler and Mr. Vivion Shull of the Center for Electron Optics for
generously allowing me to use their facilities. Also, thanks go to Mr. Tom Palazzolo and
other members of the Physics Machine Shop for their help and friendship. I appreciate Dr.
Ibrahim Sekercioglu at Howmet Turbine Component Corporation for his valuable
comments and help in using the company’s facilities. Thanks are given to Mr. Richard
Schalek and Mr. Karl Tebeau for proofing drafts of my thesis. Finally, I would like to
acknowledge the financial support received from the Composite Materials and Structures

Center at Michigan State University for studying the processing of C—TaC composites.

iv

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE OF CONTENTS
Page
LIST OF TABLES vii
LIST OF FIGURES viii
CHAPTER
1. INTRODUCTION 1
2. LITERATURE REVIEW 5
2.1 PHYSICAL PROPERTIES OF TaC 5
2.2 MECHANICAL PROPERTIES OF TRANSITION METAL
MONOCARBIDES 10
2.2.1 Microhardness 10
2.2.2 Mechanical behavior of group IV carbides 14
2.2.3 Mechanical behavior of group V carbides 20
2.3 FUNDAMENTALS OF CREEP IN CRYSTALLINE SOLIDS 24
2.3.1 Creep curves 24
2.3.2 Steady—state creep 27
2.3.3 Mechanisms of creep ‘ 29
3. ROOM TEMPERATURE DEFORMATION & ANNEALING OF TaC --------- 35
3.1 INTRODUCTION , 35
3.2 EXPERIMENTAL PROCEDURE 38
3.2.1 Processing and characterization of tantalum carbide specimens -------------- 38
3.2.2 Indentation deformation and TEM specimen preparation 41
3.3 EXPERIMENTAL RESULTS 45
3.4 DISCUSSION 64
3.5 SUMMARY _- . . 73
4. INTERMEDIATE-TEMPERATURE CREEP OF TaC 74
4.1 INTRODUCTION 74
4.2 EXPERIMENTAL PROCEDURE . 79

 

 

4.2.1 Creep specimen preparation 79

 

.I'l’r

4.2.2 Creep test apparatus

 

 

 

 

 

 

 

 

 

 

 

 

 

 

80

4.2.3 Creep tests 82

4.3 EXPERIMENTAL RESULTS 86
4.4 DISCUSSION 110
4.5 SUMMARY 115
5. PROCESSING OF C-TaC COMPOSITES . 116
5.1 INTRODUCTION 116
5.2 EXPERIMENTAL PROCEDURE 123
5.2.1 Characteristics of starting materials 123
5.2.2 Fabrication of C-TaC composites 126

5.3 EXPERIMENTAL RESULTS 138
5.4 DISCUSSION 153
5.5 SUMMARY 156
6. SUMMARY AND CONCLUSIONS 157
LIST or REFERENCES 160

 

 

LIST OF TABLES -

Table Page

 

 

 

 

 

 

 

 

 

 

l. Slip planes in the transition metal carbides. (From Ref. [30]). 13
2. The chemical composition of as-received TaC specimen. 39
3. Some characteristics of TaC specimens. ' i 39
4. Dislocation contrast of single set of dislocations. 51
5. Dislocation contrast of two sets of dislocations. 53
6. Dislocation contrast of dislocations in hexagonal networks. 61
7. Quasi steady-state creep rate data of TaCOW polycrystals. 9O
8. Dislocation contrast of individual dislocations. 101
9. Dislocation contrast of dislocations in hexagonal networks. 104
10. Dislocation contrast of dislocations in networks. A 108
1]. Characteristics of as-received TaC powder and C-fiber. ' 124

 

 

ng

 

 

 

LIST OF FIGURES

Figure
1. Phase diagram of the Ta-TaC system (From Ref. [2]).

Page

 

2. Variation of the effective resolved shear stress on (001) surface with

 

{110}<1I0> and { 111 }<1I0> slip (From Ref. [30]).
3. Schematic diagram of the {111] plane in cubic carbides showing possible
slip path (From Ref. [30]).

13

15

 

4. Schematic drawing of typical creep curve observed-under a constant tensile

 

stress (From Ref. [52]).

5. Optical micrographs of as-received and HIP processed TaCogg specimen:

25

46

 

(a), (b); as-received specimen, (c), (d); HIP processed specimen.

6. Transmission electron micrographs of undeformed TaC0.99 specimen:

47

 

(a) Near grain boundary region, (b) Triple grain boundary point.

7. Transmission electron micrographs and the diffraction pattern of indentation
deformed TaCogg specimen: (a) Dislocation structures near the indentation,
(b) Selective area difiraction pattern corresponding to [111]};1 zone,

(0) Low magnification image of the indented TEM foil.

49

 

8. Burgers vector analysis of single set of dislocations: B; beam direction,

 

g; diffraction vector.

50
52

 

9. Transmission electron micrograph of two sets of dislocations.

10. Burgers vector analysis of dislocations of typel and 2: B; beam direction,

 

 

 

18. Si

 

 

11. Transmission electron micrographs of the indented and annealed Taco”:

(a), (b); High and low magnification images near the indentation area of

Page

55

 

an annealed specimen.

12. Transmission electron micrograph showing the region near the

‘57

 

recrystallized grain.

13. Transmission electron micrograph showing the subboundaries and

58

 

hexagonal networks.

14. Transmission electron micrograph showing some of sub grains and

59

 

cell structure.

 

15. Burgers vector analysis of the hexagonal networks. "

16. The Kikuchi pattern analysis of the misorientation angle between two subgrains:

(a), (d); bright field images of two subgrains labeled 1 and 2.
(c), (f) and (h); the corresponding Kikuchi patterns of subgrain 1.
(b), (e) and (g); the corresponding Kikuchi patterns of subgrain 2.

 

B; beam direction.

17. Schematic of the slip configuration near the indenter.

62
67

 

l8. Schematic of the possible formation of long screw dislocation segments

from the dislocation half-loop: (a) initial stage, (b) after glide,

69

 

e; edge dislocation, s; screw dislocation, b; Burgers vector.
19. A schematic drawing of specimen set-up for creep tests: 1.12M push rod,
2. W-pin, 3. A1203 platen, 4.12M compression block, 5. SiC platen,

6. Taco.” specimen, 7. Thermocouple, 8. A1203 extension rod,

83

 

9. Extensometer.

20. Creep curve of TaCogg obtained under the conditions of 105 MPa

87

 

and 1400 °C. -

 

 

 

21.1

22. i

23.

25.

26.

27.

28
29

 

 

21. Creep curves of Taco.” under the conditions of 105 MPa and
170 MPa at 1500 °C: 0; 105 MPa, A; 170 MPa.

Page

88

 

22. Strain rate as a function of strain: I; 105 MPa/ 1400 °C.

0; 105 MPa/1500 °C. A; 170 MPa/1500 °C.

89

 

23. TEM micrograph of TaCQ99 crept at 1300 °C and 300 MPa.

92

 

24. Optical micrographs of Taco.” crept at 1500 °C and 60 MPa:

 

(a) low magnification image, (b) high magnification image.
25. TEM micrograph of TaCogg crept at 1500 °C and 60 MPa:
B; beam direction, g; diffraction vector.

93

95

 

26. High magnification micrograph of the subboundary shown in Figure 25:

 

B; beam direction, g; diffraction vector.
27. TEM micrograph of crept Taco.” showing a subboundary and
individual dislocations: B; beam direction, g; diffraction vector.

96

97

 

28. High magnification micrograph of the subboundary shown in Figure 27. -----

29. TEM micrograph of crept TaCogg showing a twist boundary:
B; beam direction, g; diffraction vector.

---- 9s

99

 

30. High magnification micrograph of the twist boundary shown in Figure 29:

 

B; beam direction, g; diffraction vector.
31. TEM micrograph of crept TaCogg showing individual dislocations:
B; beam direction, g; difiraction vector.

 

32. Burgers vector analysis of dislocations of type 1,2 and 3:

 

B; beam direction, g; difiraction vector.
33. TEM micrographs of crept T3CQ99: (a) a subboundary and hexagonal
networks, (b) higher magnification image of the subboundary.

 

3: beam direction, g; diffraction vector.

100

102

103

105

 

 

105

 

B; beam direction, g; diffraction vector.

34. Burgers vector analysis of the hexagonal networks:

 

B; beam direction, g; diffraction vector.
35. Burgers vector analysis of the dislocation networks:

B; beam direction, g; diffraction vector.

 

36. TEM micrograph of crept TaCogg showing a subboundary

 

and a grain boundary.
37. Scanning electron micrographs of as-received TaC powders and

carbon fibers: (a) Type A powder, (b) APOLLOTM 55 fiber,

 

(c)Type B powder, ((1) AS -4 fiber.

 

38. The block diagrams of two C-TaC composite processing routes.
39. Photographs of Ta HIP containers: (a) some parts of a HIP container
(a Ta can, top lid, top and bottom spacer), (b) a vacuum-sealed container

before HIP, (c) HIP processed containers (left; a properly processed

 

container, right; a leaked container).

40. Time, temperature, and pressure profiles of HIP process.

 

41. Schematic of slurry infiltration process.

 

42. Photographs of niobium HIP containers: (a), (b) a side and front view
of a Nb container before HIP, (c), (d) a side and front view

of a Nb container after HIP respectively.

 

43. Optical and SEM micrographs of C-TaC composite:
(a), (b) optical and SEM micrographs showing the inatrix(M), interfaced)

and fiber(F) region, (c) a SEM micrograph of damaged carbon fibers. ---------

44. Auger electron spectroscopic survey of C-TaC composite.

 

45. Tantalum line scans across the interface of C-TaC composite:

xi

Page

106

107

109

125
127

129
131
133

135

139
140

 

 

 

 

Page

46. SEM micrographs of damaged C-fibers during cold compaction. 143
47. SEM micrographs of slurry infiltrated carbon fibers:

(a) perpendicular to the fibers, (b), (c) parallel to the fibers. 144
48. SEM micrographs of C-TaC composites produced by the processing route #2:

(a) transverse direction to the fibers, (b) longitudinal direction to the fibers. ------ 145
49. High magnification SEM micrographs of the specimen in Figure 12:

(a) transverse direction to the fibers, (b) longitudinal direction to the fibers. ----- 146

50. The SEM micrographs of the sintered and fractured pre-preg composite tape. ---— 148
51. The SEM micrographs of a coated carbon fiber:

(a) transverse direction to the fiber, (b) longitudinal direction to the fiber. -------- 149
52. Ta and C line scans of the coated carbon fiber:

 

(a) Ta and C line scan, (b) Ta line scan. 150
53. The SEM micrographs and a Ta line scan of sintered C-TaC composite

 

after coating the fiber. _ 152

 

Til
an
or
co
or
on

CI

 

 

CHAPTER 1
INTRODUCTION

The monocarbides of the group IV and V transition metal such as
TiC, ZrC, HfC, VC, NbC and TaC have B1 structure in which carbon
atoms are occupying the octahedral interstitial sites of the face centered
cubic (FCC) metal lattice. These materials belong to the group of Héigg
compounds and are also named as cubic carbides due to their cubic
crystal structure. Transition metal monocarbides (TMMC) exhibit a
combination of properties of three primary types of bonding in solids. For
example, their crystal structure is ionic NaCl type, their electrical and
magnetic properties are metallic, and their high elastic modulus and high
nelting point are associated with covalent bonding.

One interesting property of TMMC is that these materials exist
DVCI‘ a wide range of stoichiometry due to structural vacancies on carbon
tom sites. The chemical formula of TMMC is often written as MCx where
I represents transition metal species and x represents carbon-to-metal
atio. In the transition metal monocarbides, the ideal stoichiometry is
enerally not found and deviations from stoichiometric composition are
11' more common. It has been reported that a vacancy concentration of up
. 50% can exist on carbon lattice sites and that the vacancies tend to
:hibit ordering when their concentration is high. It has been found that

1 types of properties e.g. mechanical, thermal, electrical, magnetic etc.

 

dep

cc
of
par

fol

Hc
dil
the

re
ch
ac

av

th

 

depend significantly upon the carbon-to-metal ratio and ordering [1].

In commercial applications, these carbides are used extensively in
“cemented carbides” cutting tools, wear-resistant parts and dies because
of their great hardness. In addition to these applications, TMMC are of
particular interest for high-temperature structural applications for the
following reasons:

i) TMMC include materials with the highest melting points.

ii) TMMC are mechanically extremely strong.

iii) Above the brittle-to-ductile transition temperature, TMMC
can be plastically deformed similar to ductile fcc metals.
However, their ‘ high notch-sensitivity, poor oxidation resistance,
difficulty of fabrication and lack of reliable property data have limited
their full acceptance. A number of books and review articles on transition
metal monocarbides have been published [1-5]. The results of early

research on the mechanical properties of TMMC were proved to be less

 
   
  
   
 
   

reliable due to the fact that samples used usually lacked proper
characterization. As high-temperature technologies advance, more

accurate properties of well-characterized carbide samples are becoming

 

available.

Among transition metal monocarbides, titanium carbide has been
the most frequently selected for investigation of deformation behavior.
This is probably due to its high strength-to-density ratio and its wide
sage in cemented carbides. The other Group IV and Group V transition
etal carbides have been less frequently studied in different degrees. So
t seems to be necessary to study their mechanical behaviors in detail in

rder to fully understand and utilize these interesting materials.

 

.A
polycry
Ihcfoll

i

ii

ifi

l
tantalu
belorn
l
anneal
ricroi
(TEM)
dhloc:
Inicroi

diSlOC;

L°.at
invest
mecha
deforr

Stlldie

Was s

and '1

 

Among the various TMMC, stoichiometric tantalum carbide
polycrystals have been chosen for the current investigation because of
the following reasons:

i) TaC has the highest melting point among all solid materials.

ii) TaC is more metallic in nature than the other transition metal

carbides.
iii) TaC has good potential for low temperature ductility.

The current work consists of three related topics on stoichiometric
tantalum carbide. The characteristics of each topic can be described as
below:

First, room temperature deformation and high temperature
annealing behavior of TaC were studied by means of a combination of
microindentation techniques and transmission electron microsc0py
(TEM). The main aim of this study is to investigate details of the

islocation structure and the deformation process resulting from the

  
   
  
  
  
   
   

icroindentation at room temperature, and also the changes in
islocation structure after annealing.

Second, the creep behavior of TaC at intermediate temperatures,
.e. at temperatures below 0.5 Tm, where Tm is the melting point, was
vestigated in order to obtain information about its elevated temperature
echanical properties, in particular the mechanisms and kinetics of creep
eformation. The microstructure development during creep tests was also

udied by TEM.

Third, the processing of carbon fiber reinforced TaC composites
as studied. The characteristics of the interface between carbon fibers

d TaC were investigated by optical microscopy, scanning electron

 

mhnu
hepn
mcpn
mtuo
mmah
fundar
fine

“use

CXpen
offiu

obmh

 

microscopy (SEM) and anger electron spectroscopy (AES). In particular,
he problems encountered during processing and the possible solutions of
he problems were addressed.

The following chapter presents a literature survey of relevant
)ICVIOUS research on TMMC which include: (a) physical properties of
antalum carbide, (b) mechanical properties of TMMC and (c)
'undamentals of creep in crystalline solids. The experimental studies on
hree aforementioned tOpics are presented in separate chapters. Each of
hese chapters is composed of sections which include introduction,
:xperimental procedure, experimental results, discussion and a summary

1f the chapter. The final chapter consists of the summary and conclusions

rbtained from the current study.

 

 

21.Pi

carbon
hansi
high

struct
octah.
found

anfit)

811110
the o
TaC I
m0n1
00m
carbi
gene
iafio

flotr

 

 

CHAPTER 2
LITERATURE REVIEW

2.1 PHYSICAL PROPERTIES OF TANTALUM CARBIDE

Tantalum, a BCC transition metal, forms two compounds with
carbon: Ta2C and TaC. T32C has two allotropic phases with an a-B
transition at about 2180 °C. The crystal structure of B-TaZC, the stable
high temperature form of TazC, belongs to the hexagonal L3’ type
structure, with carbon atoms statistically distributed in one-half of the
octahedral sites. Bowman et. a1. [6], after a neutron diffraction study,
found that the crystal structure of or-Ta2C is a C6 cadmium iodide
antitype structure, due to ordering of the carbon atoms.

TaC, a monocarbide phase, crystallizes in the Bl(NaCl-type)
structure. Ta atoms occupy the FCC sublattice, while the C atoms occupy
the octahedral interstitial sites. The stacking sequence of {111} planes of
TaC can be designated as AyBocCBA'yBaCBA'y, where A B C represent Ta
atoms and or [3 7 represent the carbon atoms in the corresponding
octahedral sites. Like the other transition metal monocarbides, tantalum
carbide has a wide range of carbon deficient compositions and is
generally represented as TaCx, where x is the carbon-to-tantalum atomic
atio. Typically, the values of x for tantalum monocarbide phase range

mm 0.74 to 0.99. The melting point goes through a maximum at about

 

 

 

systc

SOC

inve
hav.
stru
seq]
beh
R01
difl
ren

0W

 

3983 °C for the TaC~o.39 composition. A phase diagram of the Ta-TaC
system [2] is shown in Figurel.

For lower carbon content the existence of an additional phase, the
so called zeta(§)-phase, has been reported [7,8]. The composition of the
C-phase is known to be close to TaCoo75. Yvon and Parthe [9] have
investigated the crystal structure of the C-phase in VC, NbC and TaC, and
have proposed that the metal atoms in the C-phase arrange in the trigonal
structure consisting of 12 close packed metal atom layers with a stacking
sequence of ABABCACABCBC. The exact arrangement of carbon atoms
between the metal layers was not determined in their X-ray work.
Rowcliffe and Thomas [10], using electron microscopy and electron
diffraction, have suggested that the C-phase of TaC is formed by the
removal of C atoms in every fourth C layer, followed by shearing of the
overlying Ta layer. The stacking sequence of the C-phase was given as
AyB’iIAyBaCBAICBAyBaCiSByCa, where arrows indicate the complete
absence of carbon atoms in the corresponding planes. Thus, the C—phase
was considered as a stacking fault with a composition of an ordered
Ta4C3 phase. Recently, Markhasev et a1. [11] have proposed that the
formation of the ordered Ta4C3 phase (C-phase) occurs by a diffusionless
slip mechanism, and is promoted by the compressive stress. However, no
microstructural evidence has been reported to support the fact that the C-
phase is an ordered phase.

In the case of the substoichiometric VC and NbC, which belong to
he Group V transition metal carbides, the existence of long range

rdering has been established and the ordered phases found were V3C7,

5C5 and Nb5C5 [12]. In contrast to those carbides, the same type of

 

 

 

- 000V

 

.Qmagom 83.5 88?? Oar—.e.“. 05 .«o Eeewfie emu—E A 0.3me

2:: 620 . o._.\u

 

 

 

 

 

 

 

 

Q. 3 ~._ 0.. mo .
2 a _ 4 A .5 00m.
OOON
on...
comma.
o+ooh .w
a
m
m.
_ w.
_ ooomq.w
__ 30m
_
_
\/ I
x .8283 m 82
\ / ,
/ ’flmo
o o o
l o \\
/ 0A \0
. I
_ _ PPILI. L..\_\_eeu _ _ oooe

 

 

ord-
difl
thn
ban

the

tha
dif
It!

18]

 

 

ordering has not been reported for tantalum carbide. The electron
diffraction pattern of single crystal TaCo.33, which was slowly cooled
through the order-disorder temperature, revealed the presence of diffuse
bands, rather than discrete superlattice spots [13]. It was suggested that
the ordering in tantalum carbide is very imperfect. Similar diffuse band
of the electron diffraction pattern have been also observed in other
compositions of tantalum monocarbide [10]. Venables and Meyerhoff
[13] have concluded that the order-disorder temperature for TaC is lower
than those of VC and NbC, and that the long range order in TaC0.33 is
difficult to develop due to the extremely low carbon diffusivity in that
temperature range. De Novion and Maurice [12] have reported that short-
range ordering is typical in TaCoc73 - TaCo.85.

Most of the physical properties of tantalum carbide have been
found to depend on the carbon-to-tantalum atomic ratio. Bowman [14]
first reported that the lattice parameter of TaC decreases linearly with
decreasing composition. Later, Storm [2] analyzed the additional data
based on well-characterized material and obtained the following
equation:

a0 = 4.2908 + 0.1673 x (1)

where a0 is the lattice parameter of tantalum carbide and x= C/Ta(i 0.01).
The reported values of lattice parameter and the density of TaC0.99 are
about 4.456 A and 14.48 g/cm3, respectively. A change in color of TaC
occurs as a function of composition. At low carbon concentrations below

about TaCo.8, the color is a metallic gray, changing to a bright golden

color near the stoichiometric composition.

 

 

defi:
rcla
the
Stci
stoi
that
ice
witl

par.

cal.
ten
she

GP

 

The electrical resistivity of TaC was found to increase with carbon
deficiency [15-17]. Cooper and Hansler [16] reported a linear
relationship between resistivity and the C/Ta ratio. This was attributed to
the fact that the carbon vacancies act as scattering centers for electrons.
Steinz and Resnick [17] measured the value of specific resistivity of
stoichiometric TaC as about 24 [IQ-cm at room temperature and reported
that the temperature coefficient of resistivity decreases rapidly with
decreasing carbon content. The magnetic susceptibility of TaC also'varies
with the composition. The magnetic property of TaC changes from
paramagnetic to diamagnetic in the range of TaCo.9 - TaCo.97 [15, 17]

The coefficient of thermal expansion and the thermal conductivity
of Taco.” at room temperature were reported as 6.3 x10'6/°C and 0.053
cal/cm-sec-°C, respectively [1]. The elastic properties of TaC at room
temperature were determined by Brown et a1. [18]. The Young’s modulus,
shear modulus and Poisson’s ratio of TaC0.99 were about 537 GPa, 216

GPa and 0.24, respectively.

 

 

r . W_ ;

2.2 M]

2.2.1

10

2.2 MECHANICAL PROPERTIES OF TRANSITION METAL CARBIDES

2.2.1 Microhardness

Westbrook [19] performed extensive measurements on the
microhardness of the polycrystalline carbides as a function of
temperature. At room temperature, TiC is the hardest carbide (Hv z 3200
kg/mmz), but it rapidly loses its hardness with increasing temperature.
The hardness of TaC is the lowest (Hv z 1800 kg/mmz) at room
temperature among‘the group IV and V transition metal carbides. An
exponential relationship between hardness and temperature was proposed
for TiC [20,21]. An Arrhenius plot showed two straight line segments,
suggesting that a process with higher activation energy dominates at high
temperature. This is in qualitative agreement with the critical resolved
shear stress of TiC single crystals in compression [22]. However,
quantitative agreement could not be found. The predicted slopes were too
low and the deflection point in the slope of the experimental curve
occurred at a rather low homologous temperature. It was suspected that
other processes or mechanisms were operative in the case of the carbides
than in the case for pure metals.

A number of researchers have shown that the hardness of TMMC is
a sensitive function of stoichiometry, as well as temperature. It was

reported that hardness values depend on stoichiometry in the opposite

 

 

way
stoic
rapi:
a at
stir]
situ.
com

can:

go
earl
pro
ban
abc
ele
bar

inc

bat

in

ha
be

00

ca

 

11

way for TiC and TaC [4, 23]. In TiC1_x, the hardness increases sharply as
stoichiometry is approached. In TaC1-x, however, the hardness increased
rapidly with deviation fromstoichiometry. Maximum hardness occurs at
a composition of TaCOO33 [15, 24]. Santoro [15] further found that
strength and hardness vary in Opposite direction with composition. This
situation casts doubt on some of the experimental results. The analysis is
complicated by the occurrence of ordered phases in the group V carbides
causing an increase of hardness.

Two possible explanations of this contrasting behavior between
group IV and V carbides were suggested, using atomic bonding in
carbides and a hard sphere model [1, 25]. Carbide formation results in a
pronounced separation of the bonding and antibonding parts of the d
band. The bonding portion of the resultant band structure is filled with
about 8-8.5 valence electrons. Deviation from stoichiometry in TiC (8
electrons) results in a depletion of electrons in the bonding portion of the
band and hence lower bond strength. Thus the hardness decreases with
increasing deviation from stoichiometry. In TaC(9 electrons), deviation
from stoichiometry deplete electrons from the antibonding portion of the
band and the average bond strength increases. Thus the hardness
increases with decreasing carbon content.

Another explanation of the composition dependence of hardness

 

has been attempted by using a hard sphere model which distinguishes
between the nearly stoichiometric carbides of group IV and V. A
comparison of lattice parameters for parent metal and carbide suggests
that group IV carbides are more ideally close-packed, whereas group V

carbides are more open structures. Hence it is argued that the greater

 

 

 

duct:
cash
for t
in la
IV b
ShOl
latti

hard

has:

is n
Sill;
0ve
ani
of
R0

011

the
be.

 

—_——7

12

ductility and preferred {111}<1T0> slip of nearly stoichiometric group V
carbides arises from easier dissociation of dislocations in this slip system
for the more open structure. With decreasing carbon content a decrease
in lattice parameter indicates a tendency to the more close-packed group
IV behavior. For the group IV carbides, any deviation from stoichiometry
should not change the arrangement of atoms significantly, since the
lattice is already close-packed. At low carbon contents, a reduction in
hardness would be expected as a result of the change in Peierls stress.
In 1971, Brookes and coworkers [26] improved existing analyses of
hardness anisotropy and showed that it can be used to determine
operative slip systems. Figure2 shows the change in effective resolved
shear stress with indenter direction for knoop impressions made on the
(100) face. The shape of the anisotrOpy curve for (100)<1T0> slip is
similar to that for (111)<1I0> slip. They further showed that the hardness
is, at a maximum when the effective resolved shear stress on the slip plane
'3 minimized. Since then, a number of KnOOp hardness measurements on
ingle crystals of carbides have been made to find out the slip systems
ver a range of temperature [25, 27-29]. By incorporating the hardness
nisotropy measurements with the results of etch-pitting and TEM, most
f the slip systems could be found and these were summarized by
owcliffe [30] (Table 1). It is seen that the choice of slip system depends
11 carbon deficiency and temperature.
Hannink et al. [27] suggested that the preference for (110) slip at
w temperatures is a result of strong, directional carbon to metal bonds
at prohibit slip on {111}. As the temperature is raised, these bonds

ecome less directional as electrons pass into metallic states, and slip

 

 

 

Fig

wit

Tal

 

 

 

 

 

 

13
r: 03 > {110}<fio>
o o
m t-
3 2
8 IL
I.” (D
m g 02. {111}<fio>
g’ r- : '
C ”’ z” 1
8 E t i
it E ' i i (001)
w to i _L l
o 45 90

ANGLE BETWEEN
INDENTER AND [100]

Figure 2. Variation of the effective resolved shear stress on (001) surface

with {110}<1T0> and {111}<1T0> slip (From Ref. [30]).

Table l. Slip planes in the transition metal carbides (From Ref. [30]).

 

Carbide Room temperature High temperature
TiC 110 111
ZrC 110 111,110,100
HfC 110 Not known
€0.34 110,111 111
bC0.83-0.96 110, 111 111
aC0.96 111 111,110

ac”3 110 111

 

 

bacon
impli
[25].
[111
stoic
mon
date
not

net:

2.2.

lan
1116
the
of

of
di.
di

 

I_———

14

becomes more similar to fcc metals, namely {111}<1T0>. Similar
implications are made in simple hard sphere models as explained earlier
[25]. However, nearly stoichiometric TaC was found to deform by
{111}<1T0> slip even at room temperature. It was also reported that
stoichiometric TaC is considerably softer than most of the other
monocarbides of group IV and V transition metals [24, 28], and that TaC
deforms plastically before it cracks, and cracks arise from dislocation
motion. Based on these findings, it. was suggested that TaC is more

metallic in nature than other carbides [28].

2.2.2 Mechanical behavior of group IV carbides

In early work on single crystal TiC, Williams [22] suggests that
lattice resistance to dislocation motion controls yielding in transition
metal carbides. It was suggested that the high Peierls stress resulted from
the occupation of the octahedral sites by carbon atoms in fee sublattices
of TiC. In 1966, Kelly and Rowcliffe [31] proposed a model for slip on
{111} planes along <1T0> in TiC that takes into account the difficulties
'of moving dislocations in this structure. Figure 3 shows a schematic
diagram of the {111} slip plane in cubic carbides and the <1T0> slip

directions such as Ble, B2B3 etc. It is unlikely that slip will proceed

 

directly in B1—>B2, B2—>B3 etc., since the {111} plane will be forced
widely apart. Instead a dislocation can dissociate as shown here, that is
BIB3—>B1C+CB3 and each of the Shockely partials Splits according to the

scheme shown in this figure. The consequence of this model is that high

 

 

 

Fig“

Sho“

 

 

15

 

 

 

 

 

0 Metal below plane

Metal above plane

 

. Carbon between metal planes

131133 —> BIC + CB3 ( a/2[1'1'01 —-> a/6[2'fi] + a/6[12l] )
BIC —> 131.0.1 + A1C( a/6[2T1‘] —9 a/6[121] + mum)

Figure 3. Schematic diagram of the {111} plane in cubic carbides

showing possible slip path (Adapted from Ref. [30]).

 

 

 

temt
dish
iron
[32-
and
disl
rest

carl

lilll

pit

001

car
she

CO]

dc:

 

16

temperature deformation will be governed by carbon diffusion near the
dislocations. Evidence in support of this mechanism has been obtained
from a number of studies in which strain rate sensitivity was measured
[32-36]. TEM observations of the dissociation of dislocations in TaC [47]
and TiC [37] were reported. The dissociation widths of partial
dislocations in TiC and TaC were about 2.5 nm and 2.7-5.2 nm,
respectively. From these values, the stacking fault energies for both
carbides were calculated to be about 170 mJ/mz.

Single crystals of TiC were deformed in bending over a temperature
range of 800 °C - 2200 °C by Williams and Schaal [38]. Dislocation etch-
pit studies on {100} cleavage surfaces revealed that slip occurred on
{111}<1T0> systems. Williams [22] subsequently conducted uniaxial
compression tests on single crystals of TiC, ZrC and NbC with various
carbon contents in the temperature range 800 °C- 1600 °C. The critical
shear stress was found to decrease exponentially with increasing carbon
content. No yield point was observed in the stress-strain curve.

Electron microscopy ofsingle crystals of TiCx (x=0.88 and 0.97)
ieformed in compression between 900 °C and 1200 °C was conducted by
Hollox and Smallman [39]. They confirmed the activation of {111}<1TO>
:lip as proposed earlier by Williams et al. [38]. Dislocation structures
ypical of FCC metals with high stacking fault energy were observed. The
>bservation of elongated dislocation loops and dipoles in the early stage
f deformation and of cell structures in more heavily deformed samples
s also consistent with the ease of cross slip. Annealing of plastically
eformed TiC is accompanied by coalescence of vacancy loops. The

titial stages appear to be associated with the formation of trails of small

 

100p
dish
iflfll

chat

and

{11
ope
to (
yie
def

cqs

Sill

30

di
dc

 

17

loops. The final stages involve the formation of a hexagonal network of
dislocations. Changes in carbon content did not appear to have any
influence on dislocation structures in TiC, but annealing kinetics were
changed.

Chatterjee et al. [40] have reaffirmed the TEM results of Hollox
and Smallman in TiC single crystals deformed in compression. They
found that around 600 °C, a gradual transition from [110}<1T0> slip to
{111}<1T0> slip occurs and around this temperature both slip systems are
operative. This was interpreted to be the governing mechanism of brittle
to ductile transition in TiC. Recently, Kurishita et a1. [32] have observed
yield drops or work softening phenomena in single crystals of TiCo.95
deformed in compression from 1007 °C to 2000 °C. A mechanical
equation of state at yielding was proposed as:

. "c Q
t = A (if) exp(-R—T- (2)

where 7' is the shear strain rate, A is a constant, tc is the critical resolved
shear stress, It is the shear modulus, m is the stress exponent, Q is the
activation energy, R is the gas constant and T is the absolute temperature.
The values of stress exponents as well as activation energies are quite
different below and above the transition temperature around 1377 °C. The
deformation rate was considered to be determined by the Peierls
mechanism below the transition temperature and by carbon diffusion
bove this temperature.

Research on deformation of polycrystalline TiC in bending and

ompression has been conducted by a number of investigators. Kelly and

 

 

\

Row:
vein
The)
half

act}
and
not
out

3111

Pit
Stl

an

 

 

18

Rowcliffe [41] investigated the plastic deformation of hot pressed
polycrystalline carbides in four point bending and under compression.
They found that a brittle to ductile transition occurs at approximately
half the melting temperature of a carbide. TiC and VC recovered
extensively on annealing for short times near 2000 °C. In a more recent
extensive study, Das et a1. [37] using four point bending and compression
found that the brittle-ductile transition temperature of polycrystalline
TiC0.97 depends also on grain size. However, they determined that the
yield stress was a function of grain size and could be described by a Hall-
Petch type relationship. A yield point drop was observed in specimens
deformed in compression at 1000 °C, but not at higher temperature where
the stress-strain curve exhibited almost parabolic behavior. The
activation energy for plastic deformation determined between 1050 °C
and 1500 °C was of the order of 3.2 to 3.6 eV. Although these values are
not close to the activation energy for C diffusion in TiC (about 5 eV), the
authors concluded that the controlling mechanism is similar to the Kelly
and Rowcliffe’s model [31]. It was also observed that a remarkable
improvement in bending strength resulted from annealing of the hot
pressed samples at 2000 °C for 4 hours in vacuum. The drastic drop in
strength between 900 °C and 1200 °C for as-hot pressed samples has been
attributed to segregation of impurities such as Fe and Co to grain
boundaries. SEM fractographs taken near the tensile surface of the
ending specimens revealed that the fracture mode is predominantly
ntergranular for both types of samples tested at temperatures ranging
rom room temperature to 1575 °C. However, evidence for transgranular

racture was also found in specimens tested at lower temperatures.

 

 

VC.

less
plat
Thi:

nod

era
of 2
TiC
inv

0111

the
has
all
abs

sta

its

str
at
pl:
lat

Rs

 

19

Boron was found to significantly increase the strength of TiC and
VC. Williams [42] reported that the critical resolved shear stress (CRSS)
of TiC crystals is increased by a factor of five at 1600 °C by addition of
less than 1% boron. He observed boride precipitates (TiBz) on {111}
planes and suggested that precipitation is the strengthening mechanism.
This was confirmed by Venables [43] who also showed that dislocation
nodes are important in acting as nucleation sites for this precipitation.
Plastic deformation in single crystals of other carbides have been
examined to a lesser extent. Williams [22] has shown that single crystals
of ZrC0.33 are stronger than TiC at all compositions between TiCogg and
TiC0.95, and this has been confirmed by Lee and Haggerty [34]. The latter
investigators also measured the strength of Zrcog as a function of crystal
orientation and induced slip on {lll}<1T0>, {110}<1T0> and
{100}<1T0> systems, when the crystal orientation was chosen such that
the Schmid factor favored slip on these systems. One observation which
as not yet been explained is that the CRSS for slip on {110}<1T0>
ppears to be slightly lower than that forslip on {111}<1TO>. The
bsence of both stacking faults and extended nodes indicates that the
tacking fault energy is high. Additional TEM studies revealed the
resence of well defined cell walls similar to earlier observations in TiC.
Darolia and Archbold [33] .found that the compressive yield
:trength of arc melted polycrystalline ZrCo.94 decreased from 77 kg/mm2
.t 1200 °C to 19 kg/mm2 at 1800 °C. Yield drops were observed with
Ilastic strain rates greater than 3 x 10 '3/sec but not with slower strain
ates. The deformation rate results were consistent with the Kelly and

owcliffe model. Hafnium carbide has been the least investigated of the

 

trans
data
illCl'l

decr

2.2

dot
the
V(
cm
at

di

 

20

transition metal carbides because of its limited availability. The tensile
data of HfC samples which contained 2 to 5% free carbon showed an
increase in strength from room temperature to 1600 °C and a rapid
decrease afterwards [23]. Microhardness of HfC0.93 single crystals at low
temperatures was measured by Rowcliffe and Hollox [28]. They also
reported that the active slip system of HfC at room temperature was

{110}<1T0>.

2.2.3 Mechanical behavior of group V carbides

Venables et al. [44] have mentioned that cubic VC is more correctly
described as a series of ordered compounds. Hollox [5, 46] has reported
the effect of ordering and composition on the CRSS in single crystals
VCX. The primary slip system is again {111}<1TO>; however, as the
carbon content is increased, the yield strength passes through a maximum
at VC0.34(V6C5) which is also the composition having the highest order-
disorder temperature. Above the brittle to ductile transition temperature,
the strength appears to be governed by one thermally activated process.
However, two thermally activated processes control the deformation
behavior of VC0.75 and so this behavior appears to be similar to that of
TiC. Williams [22] has shown that single crystals of NbC0.76 exhibit a
greater strength than either ZrC0.33 or TiC0.95. Kelly and Rowcliffe [41]
have shown that hot pressed NbC0.95 is stronger than NbCo.33 of similar
density, indicating an increase strength with carbon content over this

composition range between 1500 °C and 2000 °C. A number of creep

 

 

stus

pre

4.]

me

inv

Ta

811'

SC‘

 

 

21

studies on single crystals and polycrystals of Nbe have been conducted
previously. Some of the results of these studies are discussed in Section
4.1 along with the reported creep behaviors cf the other TMMC.

Except for some measurements of microhardness [24, 25, 28],
mechanical properties of TaC single. crystals have not been extensively
investigated. Santoro [15] studied a number of properties of carburized
TaC filaments as a function of carbon content. He has shown a maximum
in the microhardness and a minimum in the room temperature rupture
strength between TaC0.3 and TaCOO85. He correlated these changes with
several other physical changes, such as a change in color and a maximum
in diamagnetism, and suspected an electronic structural transition in this
composition range. Measurements of the bending strength of hot-pressed
TaCx (with 4% of Co binder) at high temperatures showed .similar trends
to that reported by Santoro [15], although the minimum bending strength
occurred at TaC0.9 [45]. In addition, the brittle to ductile transition
temperature increases linearly with increasing carbon content from
compositions within the TaC+Ta2C two phase field across the TaC region
to the two phase TaC+C field.

The tensile strength of polycrystalline TaCoogg was measured at
temperatures between 1800 °C and 2200 °C with a strain rate of 8.3 x
10 '4/sec [23]. The samples were made from Ta sheets, carburized in

raphite powder at 2700 °C to 3000 °C in vacuum or argon atmosphere.
he elongation became noticeable at about 2000 °C and increased rapidly
t higher temperature, while the strength decreased slowly. Below about
850 °C, no reproducible values for the elongation could be obtained.

he microstructure showed large carbide grains. At 2160 °C, the UTS was

 

0bSl

mic

abo
em]

onl

00!
Ian

del

 

22

98 MPa and an elongation of 33% was obtained. Necking was definitely
observed, but there was little, if any, separation of grains visible in the
microstructure after test.

Kelly and Rowcliffe [41] reported a ductile to brittle transition at
about 1750 °C for hot pressed polycrystalline TaC. In their study
employing four-point bend tests, they found further that even at 2000 °C
only a small permanent strain of 0.1% could be achieved prior to failure.
However, Becher [35] observed significant plastic deformation in
compression tests even at temperatures as low as 1280 °C at strain rates
ranging from 8.3 x 10’4/sec to 8.3 x 10'6/sec. He also suggested that the
deformation occurred in part by dislocation motion controlled by carbon
diffusion in contrast to the previous result by Steinitz [23].

The most extensive studies on the carburized thin sheets of TaC1_x
have been performed by Martin and coworkers [36, 47]. They performed
four-point bending experiments up to 2200 °C and strain rates ranging
from 10'7/sec to 4 x 10'4/sec. They showed that the specimens could

ndergo extensive plastic deformation above 1500 °C for a strain rate of
10’4/sec (1200 °C for strain rate of 10'7/sec). Annealing effects became
vident only at very high temperatures and low strain rates (1900 °C and
0'7/sec, 2000 °C and 10 '6/sec, T>2100 °C for higher strain rates) and
ere considered to be caused by dislocation climb. They analyzed the

esults of bending tests in terms of a strain rate law of the type:

(o-os
. _ or z -A U (3)
8 _ Aux-oz.) exp[——k—T—‘:]exp[——kT J

here, E is the strain rate, A and or are constants, C is the applied and 0;

 

 

is the
const:
The r

carbo

rcvea
and t
repor
speci
of di
Allis
sing]
carb
carb

defo

 

23

is the internal stress, V is an activation volume, k is Boltzmann’s
constant, T is the absolute temperature and AU is an activation energy.
The results of their study were interpreted on the basis of a model of
carbon diffusion assisted glide of dislocations.

TEM studies of the dislocations in deformed TaC0.7_0.8 samples
revealed stacking fault fringes bounded by Shockely partial dislocations, .
and these were interpreted as a result of shearing [10]. Martin [47] also
reported the dissociation of dislocations in the deformed TaC0.93_0.95
specimens by using weak beam dark-field microsc0py. He found the width
of dissociation to decrease with increasing carbon content [36]. Recently,
Allison et al. [48] analyzed the carbon concentration in stacking faults of
single crystals of TaCOO78. The analysis showed a significantly lower
carbon concentration in the fault. They concluded that the diffusion of
carbon away from the moving dislocation must accompany plastic

deformation.

 

2.3

2.3.

0011

16111

161!

1110

hot

ten

abs

€111
Th

C11

00

pr

ac

do

Dl

 

 

24

2.3 FUNDAMENTALS OF CREEP IN CRYSTALLINE SOLIDS
2.3.1 Creep curves

Creep is a time dependent deformation of materials under a
constant load or stress. Although creep can occur at low homologous
temperatures, creep of most materials is usually not significant at
temperatures below 0.3 Tm, where Tm is the melting point. Therefore,
most practical creep experiments have been performed at higher
homologous temperatures. The range of temperature for intermediate-
temperature creep usually lies between 0.3 Tm and 0.5 Tm. The creep
above 0.5 Tm is regarded as high-temperature creep.

Creep in a solid can be represented by a creep curve. The creep
curve is a plot of strain versus time for a given stress and temperature.

he creep rate is determined as the slope of the creep curve. Typical creep
urves of annealed crystalline materials in tension are normally
omposed of four stages (Figure 4): instantaneous strain, transient or
rimary creep, steady-state or secondary creep, and tertiary or
ccelerating creep. The creep rate continuously decreases with time
uring the transient creep and is constant during the steady-state creep.
uring the tertiary creep, the creep rate increases until creep rupture
ccurs. In compression tests, tertiary creep is absent and, following the
teady-state creep period the creep rate decreases continuously as strain

creases. The extent of each stage depends mainly on the temperature and the

 

 

Figur

°0nst

 

25

Fracture

  
  

 

Strain

 

Accelerating
creep

”Instantaneous' strain

 

Time

igure 4. Schematic drawing of typical creep curve observed under a

onstant tensile stress (From Ref. [52]).

 

 

aPP

stag

whf
pla
the
Sir
lot
the
811‘
du
01:

811

Cl]
81]

Cl]

 

26

applied stress of the creep test. Thus, a creep curve with four wells-defined
stages can be-observed under certain conditions of stress and temperature.

Phenomenologically, creep can be considered as a process during
which two competing processes of strain hardening and recovery take
place. The decrease in creep rate during primary creep is associated with
the strain hardening process, i.e. with gradual changes in the dislocation
structure. Dislocations, which are homogeneously distributed upon
loading, begin to form cells or subgrains. The constant creep rate during
the steady-state creep stage results from a dynamic balance between
strain hardening and recovery processes. The substructure development
during steady—state creep is characterized by a constant average
dislocation density and constant subgrain size despite increasing creep
Istrain.

Besides the typical creep curve, several different shapes of creep
curves have been observed for various materials and these were
summarized by Bird et al. [49]. The major difference between these creep
curve were found over the transient creep regime. For certain solid
solution alloys in which the solute drag effects on dislocations control
creep resistance, the creep rate increases with creep strain during the
ransient creep period. The creep curves of many solid solution alloys
uring the transient creep stage are often sigmoidal in nature because of
0th solute drag and strain hardening effects [50]. The shape of creep
urves also can be affected by the thermomechanical history of a

aterial. For example, a continuous increase in creep rates during

rimary creep was observed in some prestrained and recovered specimens

f pure nickel and nickel alloys [51].

 

 

 

 

2.3

0111

of
the

 

27

2.3.2 Steady-state creep

In most theoretical and experimental creep studies, the major
emphasis has been placed on the steady-state creep stage. For many
materials, it has been found that the temperature and stress dependency
of the steady-state creep rate can be best represented by an equation of
the form:

es = Ac(3)nexp(-—§ig) (4)

where és is the steady-state creep rate, Ac is the creep constant, 6 is the
applied stress, n is a constant known as stress or creep exponent, Qc'is
the activation energy for creep, k is Boltzmann’s constant, [I is the shear
modulus and T is the absolute temperature. It has been found that
different creep mechanisms can be operative depending on the range of
temperature and the applied stress for the creep test. Some important
creep mechanisms are discussed in Section 2.3.3.

The value of the activation energy for creep can depend on
emperature; for example, in Al [53]. At low or intermediate-
emperatures, the activation energy for creep is usually lower than that
or self-diffusion. For high-temperature creep, it is established that the
ctivation energy for creep is nearly equal to the activation energy for
elf-diffusion [54].

Sherby and Burke [55] discussed the structural variables that are

 

 

imp
var?
sub
dec
ela
100
0P1
or

me

ex
1111
mi
311

It

 

 

28

important in the power-law creep of polycrystalline materials. These
variables include elastic modulus, stacking fault energy, grain and
subgrain size. In general, an increase in elastic modulus yields a large
decrease in creep rate for a given stress. Thus, materials having a higher
elastic modulus can be expected to be more creep resistant than ones with
lower elastic modulus provided that other variables are the same. An
apposite trend was observed for the dependency of stacking fault energy
on the steady-state creep rate. The analysis of creep data for pure FCC
metals revealed that the creep rate increased with increasing stacking
fault energy (7). The creep rate was found proportional to 73.5. An
explanation for this behavior is due to the greater difficulty in climb for
the widely separated partial dislocations of low stacking fault energy
materials. The stacking fault energy can also influence the creep rate in
an indirect way by affecting the subgrain size of materials. It was
reported that the influence of grain size on the steady-state creep rate in
pure metals was of minor importance. On the other hand the subgrain size
seems to play an important role. In general, the creep rate of a fine
subgrain size material is much smaller than that of coarse subgrain size
material at the same stress level.

Cannon and Langdon [56] reviewed the creep behavior of ceramics
and noted that the steady-state creep behavior of ceramics was similar to
that of metals. Despite the overall similarity, they found that there were
two major differences between the creep of ceramics and the creep of
metals. First, diffusional creep with a stress exponent of 1 is a common
deformation mode in ceramics. Second, the power-law creep behavior of

ceramics yields stress exponents of about either 3 and 5, whereas pure

 

 

 

 

an

Pl'
gr

an
de

81:

 

 

 

 

29

metals usually exhibit stress exponent close to 5.

The important contribution of diffusional creep in ceramics was
attributed to a low grain boundary mobility, i.e. fine grain size, and the
preferential enhancement of diffusion of one of the ionic species along
grain boundaries in ceramics. For power-law creep behavior of ceramics,
the stress exponent of about 5 was associated with fully ductile behavior
and the stress exponent of about 3 was associated with plastic
deformation in a less ductile condition, e.g. a lack of five independent
slip systems. Cannon and Langdon [56] also concluded that the creep
mechanism for the steady—state creep behavior with the stress exponent
of about 3 was due to dislocation climb from Bardeen-Herring sources

[57] under the conditions where crystallographic slip was restricted.

2.3.3 Mechanisms of creep

Although various mechanisms for creep deformation have been
pr0posed in‘the past, there are basically two major creep mechanisms:
dislocation creep and diffusional creep. Dislocation creep occurs by
dislocation motion aided by vacancy diffusion. Depending on the level of
stress applied in the creep test, dislocation creep mechanisms can be
categorized into three creep regimes. They are known as power-law

reep, Harper-Darn creep and power-law breakdown. Diffusional creep
sually occurs at low stresses and high temperatures by diffusion of
acancies only. The diffusional creep mechanism can be categorized into

wo cases, i.e. Nabarro-Herring creep and Cable creep. In the following,

 

 

 

 

SOII

abo

Thi

010

has

dis

alc

Oil

the

Sh

po

11C

Cl'

 

30

some of characteristics of these creep mechanisms are briefly disCussed.

In the stress range of 10'4< o/pt<10'3 and at temperaturesabove
about 0.3 Tm, the stress exponent (n) in eqn.(4) varies between 3 and 8.
This regime is called power-law creep. Two different cases of power-law
creep may be distinguished. When the obstacles to dislocation motion
have dimensions on the scale of a few interatomic distances, the
dislocation may overcome these obstacles by thermally-activated glide
alone. The power-law creep for this case is known as glide-controlled
creep. Weertman [58] considered the case where dislocation glide over
the Peierls barrier was the rate-controlling process of steady-state creep.
She found that the creep rate in this glide-controlled creep followed a
power-law and the value of the stress exponent was close to 3. It was
noted that this model may be important in the case of non-metallic
crystals with high Peierls stress.

The second case of power-law creep occurs when the obstacles to
dislocation motion are too wide to be overcome by thermally-activated
glide alone. At high temperatures, gliding dislocations which are held up
by obstacles, can overcome the obstacles by climb and then continue to
glide to the next obstacles. The rate-controlling process in this case is the
diffusion of vacancies to or from the climbing dislocations. The power-
law creep by this climb-plus-glide sequence is known as climb-controlled

creep. In general, climb-controlled creep is regarded as real power-law

 

creep. From the Bailey-Orowan equation, it has been proved that the
‘natural’ stress dependency for any climb-controlled creep is a power-law
with stress exponent n=3 [59]. In the climb-controlled creep model

roposed by Weertman [60], she considered that the groups of dislocation

 

 

dip

5011

100

of:

the

glie

for

of

011

cm

 

—.

31

dipoles form from the dislocation loops which were created at dislocation
sources. She assumed that each dislocation glides the mean radius of the
loops before it climbs a half of the spacing between parallel slip planes
of neighboring dislocation sources and that the dislocation glide makes
the major contribution to the creep strain. Using a relation between the
glide velocity and the average climb velocity of dislocations, it was _
found that the creep rates varied by a power-law with a stress exponent
of 4.5. Mukherjee et al. [54] showed that many aspects of steady-state
climb-controlled creep could be correlated by the following semi-
empirical equation.
n
as = Ali—if.(g) (5)
where, A is a constant, D is the respective diffusion coefficient, and b is
the Burgers vector. It was found that for climb-controlled creep of pure
metals and solid solution alloys, the stress exponent (11) generally ranges
from 4 to 7. The equation (5) is also known as the Dam equation [54].
At low stresses, o/u S 5 x 106, the creep rate is found to be
roportional to the stress, i.e. n=1. If it is due to dislocation creep, it is
alled Harper-Darn creep. Harper and Born [61] first observed this
inear-viscous creep in Al polycrystals, which were deformed at near
elting temperature and stresses lower than 0.09 MPa. Although this
inear-viscous creep behavior is usually associated with the diffusion
reep mechanism, they found that the creep rates were about three orders
f magnitude greater than those predicted by Nabarro-Herring diffusional

reep mechanism. Harper-Darn creep was also observed in Al alloys [62]

 

 

 

and l
Harp
cond
repo
crec
slip.
area

will

that
as p
can
con
can

big

wh
an

br.

01

h:

 

 

32

and CaO [63]. Based on TEM studies, Langdon et al. [64] pr0posed that
Harper-Darn creep arises from the climb of edge dislocations under the
condition that jogs are saturated with vacancies. Mohamed et a1. [65]
reported that the dislocation substructure developed during Harper-Darn
creep in A1 showed the absence of subgrains and the evidence of cross-
slip. It is believed that Harper-Darn creep is due to iclimb controlled
creep under conditions where the dislocation density does not change
with the stress.

At high stresses, o/u 2 103, the creep rate is found to be faster
than equation (4) predicts, i.e. n is large. This creep behavior is known
as power-law breakdown or as the exponential creep region. This process
can be considered as a transition from climb-controlled to glide-
controlled flow. Luthy et a1. [66] proposed an empirical equation which
can describe the steady-state creep behavior between intermediate and
high stress regime:

n

as =Def/B(sinhat-g) ' (6)
here, Dcff is the effective diffusion coefficient. B and at are constants.
nd E is the dynamic unrelaxed Young’s modulus. The power-law
reakdown starts at a value of at -_'= (o/E)'l. I

At high temperatures and relatively low stresses, o/pt <10'4,
iffusional creep becomes the controlling mechanism of creep. For
iffusional creep, the creep rate is proportional to the stress, i.e.
ewtonian flow. Nabarro [67] and Herring [68] proposed that a non-

ydrostatic Stress field could give rise to a gradient of chemical potential

 

 

 

on .
gra

Her

wh

the

pre
the
b0

81]

 

33

on grain surfaces and thus cause a diffusional flow of matter within the
grain.This diffusion controlled creep process is referred to as Nabarro-
Herring creep which can be represented as:

D on
é = a .1..—
S N” era!2

(7)
where, “NH is a constant, Dv is the lattice diffusion coefficient, 9 is the
atomic volume and d is the grain size. Thus, for Nabarro-Herring creep,
the creep rate changes with the square of the grain size and is
proportionalitoethe applied stress. For the Nabarro-Herring mechanism,
the grains become elongated by the flow of vacancies from grain
boundaries experiencing a tensile stress to those under compressive
stress, and a simultaneous flow of the atoms in the opposite direction.
This process is important only if the flux of vacancies is large, so the
Nabarro-Herring creep is important for materials having small grain sizes
and at high temperatures where the diffusion coefficient is large.

Cable [69] proposed a model in which the diffusional mass flow
during creep occurred through grain boundaries. This diffusional creep
mechanism is called as Coble creep. Cable creep generally becomes
important at lower temperatures where ‘grain boundary diffusion

redominates over volume diffusion. The Cable creep equation can be

escribed by the following equation:

= a ng609

8‘ " de3 (8)

 

here, are is a constant, ng is the grain boundary diffusion coefficient,

 

 

 

his the e
in Coble
Nabarro
that the

the inve

 

 

34

8 is the grain boundary thickness. As can be noted from the equation (8),
in Cable creep the stress dependency of creep rate is the same as the
Nabarro-Herring creep model. The important difference lies in the fact
that the creep rate in Coble creep depends on the inverse of (13 rather than

the inverse of d2 as in Nabarro-Herring creep.

 

 

 

R0

3.1

the
def

[CS

CHAPTER w3
ROOM TEMPERATURE DEFORMATION & ANNEALING OF TaC

3.1 INTRODUCTION

Most ceramic materials are known to suffer brittle fracture below
their brittle-to-ductile transition temperature without any sign of plastic
deformation, when they are subjected to conventional uniaxial or bending
tests. The group IV and V transition metal monocarbides (TMMC) belong
to this class of materials. The main reason for the low temperature
brittleness of these materials is their high Peierls stress which comes
from the very strong covalent bond between metal and carbon atoms. At
lower temperatures the fracture strength of these materials is much lower
than the yield strength, so that brittle fracture will occur before plastic
deformation.

However, it has been shown for a number of brittle solids that
plastic flow can occur at low temperatures e.g. during indentation [70].
This has been attributed to the fact that-the stress around the indentation

as a large hydrostatic stress component and that during indentation high
hear stress can be obtained locally [71]. These promote slip while
nhibiting brittle fracture. Therefore, microhardness indentation is a

uitable method to investigate the deformation behavior of brittle solids

t low temperatures.

 

 

 

inder
etche
mien
surfa
1261
grou
pref
son
130?
NM
[29
to 1
has
Sig
del
81C
tra

me

 

36

In most previous studies on this subject, Vickers or Knoop
indentations has been made on single crystals and the surfaces were
etched to reveal the disloCations and slip traces. The anisotropy in the
microhardness of single crystals along with etch-pitting of indented
surfaces have been used to determine the slip systems of these materials
[26]. The results of KnOOp indentation experiments on single crystals of
group IV and V transition metal monocarbides indicate that slip occurs
preferentially by [110} <1T0> slip system at low temperatures except for
some particular compositions of group V transition metal monocarbides
[30]. It was reported that at room temperature plastic deformation of
NbC0.33 crystals took place by both {111} <1TO> and {110} <1T0> slip
[29]. For single crystals of TaC0.96 the preferred slip system was found
to be {111} <1T0> by etch—pitting method [24] as well as analysis of the
hardness anisotropy curve [28]. Rowcliffe and Warren [24] even observed
significant plastic flow around indentations in TaCOO95 single crystal
deformed at liquid nitrogen temperature. They concluded that
stoichiometric tantalum carbide was considerably softer than the other
transition metal carbides and thatit behaves like a relatively ductile
material;

However, in order to study details of the plastic deformation
process and the dislocation structure in a deformed material, high
magnification of transmission electron microscopy (TEM) is necessary.
TEM studies of Si single crystals deformed by indentation at room
temperature have been reported earlier [72,73]. These results showed a
high denSity of dislocations along with cracks Or dislocation loops in the

vicinity of the indentation. Annealing of the deformed crystals above 550 °C

 

 

 

 

cause
orier
inde:
resui

the r

dish
crys
anis
that
late

illde

toe
of

8111

of
ma

de

 

 

37

caused a reorientation of the dislocation loops into approximately edge
orientation or the formation of new screw-edge loops [73]. The
indentation in alumina at room temperature was studied by TEM and the
results gave evidence of plastic deformation by both slip and twinning in
the vicinity of the indentations [74]..

For TMMC, Morgan and Lewis [29] conducted a TEM study of
dislocation arrangements around Knoop indentations for Nb6C5 single
crystals in an attempt to confirm the slip process deduced from hardness
anisotr0py measurements. The transmission electron micrographs showed
that the density of dislocations near the indentation is high and that
lattice rotations occur within the small volume of deformation around the
indentation. By trace analysis, the Burgers vector of dislocations at the
periphery of the deformed region were identified to belong to the {111}
<1TO> slip system. Although the details of the study were not presented,
Rowcliffe [30] reported TEM micrographs of HfC crystals deformed at
room temperature byindentation. The result of Burgers vector analysis
of dislocations showed {110] <1TO> slip system in HfC crystals. No such
studies have been reported yet for other TMMC.

The present study was undertaken to investigate deformation behavior
of TaC0099 at room temperature by means of an indentation technique and
transmission electron microscopy. The main purpose of the study can be
described as the following: (1) to directly observe dislocation structures
in TaCOO99 crystals deformed by indentation; (2) to determine the active
slip systems of TaCoogg at room temperature; and (3) to obtain

information about the effects of annealing on plastically deformed

TaC0.99.

 

 

 

 

3.2 '

3.2.

put.
the
con
mu
Spc
cut
(Ll
mi
sch
gri
di:
tin
be

 

38

3.2 EXPERIMENTAL PROCEDURE

3.2.1 Processing and characterization of tantalum carbide specimens

A hot pressed tantalum carbide rod with a square cross section was
purchased from Cerac Inc. of Milwaukee, Wisconsin. The dimensions of
the rod were approximately 11.4 mm x 11.4 mm x '153 mm. The chemical
composition of the as-received TaC specimen is presented in Table 2. A
number of TaC0.99 specimens were cut out from the rod by using a low
speed diamond saw (ISOMETTM, Buehler Ltd., Lake Bluff, Illinois). After
cutting, these specimens were mounted with thermoplastic powder
(LUCITE®, Leco, St. Joseph, Michigan). The samples for optical
microscopy and microhardness measurement were prepared as follows. A
series of SiC papers, from 200 to 600 grit, were used for mechanical
grinding and the final polished surface was obtained by using two size
diamond compounds (MICROID® diamond compound, Leco), 6 pm and 1
pm, on nylon polishing clothes. An etchant for revealing the grain

boundaries of TaC was the solution of 50% HF and 50% HNO3 by volume.

The grain size was determined by using the linear intercept method

 

on optical micrographs. The grain size obtained was the average of ten
easurements. Microhardness was measured on the etched surface using
Vickers microhardness tester (Buehler Ltd.) with 1 kg load, 20 seconds
f loading time and 50 pm per second in loading speed. In general, the

ndentations were made in the interior of grains and the microhardness

 

 

 

Tab?

 

Ta

 

 

39

Table 2. The chemical composition of as-received TaC specimen“.

 

 

C / Ta Ratio 0.99
Impurity (%) A1 0.01 Nb 0.05
B < 0.001 Ni 0.005
Co 0.01-0.06 Si 0.01
Fe 0.01 Ti 0.005
Mo 0.02 W 0.01-0.09

 

 

 

 

* Data from CERAC Inc., Milwaukee, WI.

Table 3. Some characteristics of TaC specimens.

 

 

Specimen Grain size(ttm) Vickers hardness(kg/mm2) Density(%)

 

As-received 22 :t 3 1178 :t 89 95*

 

HIP '
Processed 57 i 6 1214 i 70 97

 

 

 

 

 

 

 

 

 

 

was

hot

pro
iso
but
am

tht

 

 

40

was determined by averaging 10 to 15 indentation measurements.

In order to improve the density, hot pressed TaC specimens were
hot isostatically pressed without encapsulation in hot isostatic pressing
(HIP) containers. The time, temperature and pressure profile of the HIP
process is presented as HIP #1 in Figure 40 of Section 5.2.2. After hot
isostatic pressing, the density of the specimens was measured by the
buoyancy method [75]. First, the specimens were cleaned with acetone
and subsequently with methanol, and dried completely before measuring
the weight in air (WA) with a balance (Mettler instrument corp.) having
an accuracy of 1 x 10'4 g. The weight of each sample used in these
measurements was about 20 grams. A 0.09 mm in diameter copper and
nickel alloy wire was tied to a sample on one end while the other end was
made into a hook for hanging on the balance. The sample was then
weighed (WT) after being. totally immersed in distilled water. It was
carefully checked that no air bubbles adhered to the sample and the wire.
The weight of the wire immersed in the distilled water up to a reference
position was measured as W0. The weight of the specimen in water (WW)
was calculated by the equation, WW: WT — W0. The density of the

Specimen (p) was computed by the following equation:

_ (WAPW- WWpA)
- (WA- WW) (9)

 

where, pW and pA are the density of water and air, respectively. The
values of pw and pA used were 0.99679 g/cm3and 1.181 x 10 ’3 g/cm3 at
a temperature of 26 °C.

Optical microscopy and hardness measurements were performed on

 

 

the I
both
pres
TaC

all!“

3.2.

WC]
and
shit
we

mu

 

41

the HIP processed specimens as mentioned before. The characteristics of
both the as-received and hot isostatically pressed TaC specimens are
presented in Table 3. In the following experiments, the HIP processed
TaC was used as the specimens for the room temperature deformation and

annealing study.

3.2.2 Indentation deformation and TEM specimen preparation.

Thin slices of TaCo.99 between 0.1 mm and 0.2 mm in thickness
were cut by a low speed diamond saw with application of a minimum load
and speed in order to prevent chipping and tapering of samples. The
slices were mechanically ground through various grades of SiC papers
and finally polished with 6 pm and 1 tun diamond compounds. Discs of 3
mm in diameter were trepanned from the above slices by using either a
electrical discharge machine (EDM) (AGIETRON, Model EMT 1.10) or
an abrasive slurry drill(South Bay Technology Inc., Model 350). In the
case of trepanning by EDM, a slice of tantalum carbide was glued to a
small steel block using a mixture of silver powder and rapid cure
adhesive. After the conductive adhesive was completely dry, the steel

lock was fastened on a magnetic holding plate and then the tank was
lled with a solution of kerosene. As an electrode, a graphite and copper
omposite material was machined into a tube having an inner diameter of

mm.The electrode was lowered to the tantalum carbide specimen and
he electrical discharge started. Typical machining speed was about 5 pm

er minute. When the cutting was completed, the discs were removed

 

 

from

moe
hea‘

COO

spe
spe
adt
of
fin
an

di

fl!

 

42

from the steel block by soaking in acetone.

For the abrasive slurry drill, a slice of tantalum carbide was
mounted onto an aluminium plate with low temperature melting wax by
heating on a hot plate to a temperature of about 120 °C followed by
cooling to room temperature. The abrasive slurry employed was a mixture
of 400 mesh boron carbide powder, glycerol and water. After applying
small amounts of slurry on the tantalum carbide specimen, a rotating
brass tube having 3 mm in inner diameter was slowly lowered onto the

specimen. A speed and a weight of the drill were adjusted to prevent

 

specimen breakage during trepanning, and occasionally fresh slurry was
added on the specimen. Typically, the time required to cut one disc out
of a 0.1 mm thick slice was about 40 minutes. After the cutting was
finished, the A1 plate was cleaned with water until free from the slurry
and reheated to remove the disc specimens. A residual of the wax on the
disc specimens was further cleaned in acetone and ethanol.

The TaCogg disc specimens were then mounted on a tantalum
carbide block of about 12 mm in thickness by using an adhesive (DUROTM
Super glue, Loctite Corp.) before applying Vickers indentations. The
microhardness indentations were made at room temperature using a
Vickers microhardness indenter with a load of 300 g applied for 5 seconds
and a loading speed of 40 um per second. On each disc, rows of 29
'ndentations were made with a spacing of about 50 pm. After indentation,

he TaC discs were carefully separated from the block by soaking in
cetone. Thin foils for TEM study were prepared in two stages. First, the
urface opposite to the indented side was dimpled to the'final thickness

f 25 um using a dimple grinder (Gatan dimple grinder Model 656). A

 

   

disc
and
grim
size

whet

side
ion:
The
whe
cur

[68'

the
the

an

IO

 

I_———

43

disc was mounted on a sample stage with low temperature melting wax
and the stage was centered with respect to a 10 mm diameter brass
grinding wheel by using an optical microscope. A small amount of 3 um
size diamond compound and distilled water was supplied to the grinding
wheel at intervals of few minutes.
In the second stage, the dimpled sample was thinned first from both
sides by ion milling (Ion Tech., Teddington, England) for about 2.5
hours, and then thinned from the dimpled side until perforation occurred.
The perforation of the TEM foil was detected by an Optical micrOSCOpe
when the light source under the foil was seen through. The voltage and
current of the ion milling process were 5 KV and 1.5 mA for each ion gun,
respectively. The pressure of Ar gas was about 1 x 10‘5 Torr. The angle
of the incident ion beam was 30 degrees. In general, the time taken until
the perforation occurred was between 15 and 20 hours. After perforation,
the samples were further thinned from both sides with an incident beam
angle of 18 degrees in order to increase the electron transparent area.
To study the annealing effect on Taco-99 after prior deformation at
room temperature, some of the indented disc samples were annealed at
1992 :1 °C (~ 0.53 rm) for 1 hour in a vacuum furnace (Centorr, Model
M-60) with a vacuum of about 2 x 10'6 Torr. In order to prevent possible
decarburization, the samples were wrapped with about 0.15 mm thick
graphite foil and placed in a graphite crucible before vacuum annealing.
fter the annealing, the samples were dimpled and thinned by ion milling
s described above. In addition to these samples, TEM foils of
ndeformed TaCo.99 were also prepared. For the undeformed specimens,

oth surfaces were dimpled for an equal amount to the final thickness of

 

 

about

elect
stage
with
rope
extr
con:
lost
The
con
con

0rd

 

I_———

44

about 25 um followed by ion milling.

The thin foils were examined in a Hitachi H-800 transmission
electron microscope (TEM) operating at 200 KV with a double tilting
stage capable of tilting i45° along both tilting axes. Usually thin foils
with a relatively large transparent area were obtained after a few
repetitions of ion milling and examination in the TEM. Because of the
extreme brittleness of tantalum carbide thin foil TEM specimens,
considerable care was needed during the preparation, transportation and
loading-unloading process of thin foil specimens for TEM investigation.
The Burgers vectors of dislocations were determined by using the
conventional gob analysis [76]. In most cases two different diffraction
conditions in which the dislocations became invisible were obtained in

order to determine the Burgers vectors unambiguously.

 

 

3.3

Spt

CO

 

45

3.3 EXPERIMENTAL RESULTS

The Optical micrographs of the as—received and HIP processed TaC
Specimens are shown in Figure 5(a), (b) and (c), (d), respectively. By
comparing these micrographs, it can be seen that considerable grain
growth has occurred during the HIP process and that the porosity has
been decreased by the HIP process. The grain size and density of both
specimens are presented in Table 3 of Section 3.2.1. The microstructure
of the HIP processed specimen in Figure 5(d) shows that many pores are
located at three-grain junctions. Some of pores located on two-grain
interfaces can be seen to separate from grain boundaries. These pores
may have resulted from the separation of pores from the grain boundaries
during grain growth. Many pores are also found inside of grains, where
they can not exhibit large shrinkage during the densification process
[77]. Some of square or triangular shapes in Figure 5(b) and (d) are due
to dislocation etch pits.

Figure 6 shows transmission electron micrographs of the TaC
specimen before the indentation experiment. Individual dislocations and
grain boundaries can be seen in Figure 6(a).The initial density of
dislocations is relatively low. The double image of some of dislocations
is probably due to three beam imaging conditions [76]. A grain boundary
triple point is shown in Figure 6(b). The grain boundaries are clean

ithout any second phase. The microstructure of TaC deformed by the

ickers indentation is shown in Figure 7(a). This micrograph was taken

 

 

 

 

.qoamoomm Bamboo; ASE £3.on £250on

“5333-3 Ref; ”:oEmoomm .36qu 33303 am can 33003-3 mo mnmnuweomfi 3230 .n 053m

46

 

47

.Emom 332509 59% 23C. 3v .=o_wou

zuquson 52w 502 A3 E0832: $605. gaseous: mo mnmnuweomfi cacao—o sommmmfimaaflt .c oSwE

 

31"

CO

 

 

48

around the area labeled as A in Figure 7(c). Figure 7(b) shows the
corresponding selective area diffraction pattern (SADP) of the bright
field image in Figure 7(a). Since the Vickers indentation was made well
inside the grain, the deformation can be considered as indentation in a
single crystal of tantalum carbide. The diffraction pattern belongs to the
[111] zone of a B1 structure, so the plane parallel to the indented foil
surface is close to a {111} plane.

In Figure 7(a), a very high density of dislocations can be seen in
the region up to the distance of few microns away from an indenter facet.
In this region dislocations are so highly tangled that the individual
dislocations can not be resolved. Outside of this region, three different
sets of dislocations can be identified. The acute angles formed by these
sets of dislocations are about 60 degrees. One set of dislocations is not
in good image contrast, but the two sets of long and straight dislocations
are thought to lie in a {111} plane parallel to the foil surface along <1T0>
directions. A set of dislocations near the bottom of the micrograph
appears to be formed by a group of end-on dislocations which lie in
another {111} plane inclined to the surface of the foil.

The Burgers vector analysis of one set of dislocations is shown in
Figure 8. The beam direction and diffraction vector for each micrograph
are denoted as B and g respectively. In Figure 8(b) and (d), the
dislocations are out of contrast implying that the condition of g-b=0 is
satisfied, where b represents the Burgers vector. Since edge dislocations
can give rise to contrast even if g-b=0, the complete invisibility condition
for a general dislocation is g-bxu=0, where u is a direction of the

islocation line. But, in practice the dislocation contrast usually becomes

 

 

49

.28 Em: 68:09: 2: he owns: couaoamcwafi Bod A3 Juan 3:2: 9 wcmucommotoo

53an cotouthc «v.3 o>to£ow 3V £33259: 2: .80: 3.533; cornea—£9 A3 ":08?on .3603.

3::on 5:859: mo 522E notofitmu 2: was winemaker: .2503... “8:38:29 .5 33E

 

.839» nouofitmu ”w £2895. 839 mm ”2330033 mo How 03:: mo 2232:. “93> muowusm .w 0.53m

_v-_uu._~__mumxe

.K

    

50

 

Hfifivmnu._o_fimumae

 

VC

d‘i

 

51

'ery weak when g-b=0 except in some special cases [76]. The resultant
[islocation contrast of this single set of dislocations is presented in Table
l. Since the dislocations are out of contrast with two diffraction vectors,
:1=[T11] and g2=[311], the direction of the Burgers vector of these
lislocations can be obtained from the cross product of diffraction
'ectors, i.e. g1 x g2 = [T11] x [311] = [0T1]. Therefore, the Burgers vector
if these dislocations is a/2 [0T1], where a is the lattice parameter of
.‘aCo.99. Since the Burgers vector and the direction of dislocations are

tarallel to each other, these dislocations are .pure screw dislocations.

Table 4. Dislocation contrast of single set of dislocations.

g I [1T1] [T11] [MT] [311]
{ Vi Inv Vi Inv

 

where, Vi and Inv denote that the dislocations are visible
and invisible respectively.

In another area of the specimen, two sets of dislocations apparently
teract with each other as shown in Figure 9. This micrograph was taken
ith a beam direction close to <112>. The intersecting angle between the
'0 sets of dislocations is approximately 60 degrees. Elongated
slocation dipoles, elongated loops and small loops are also seen in the

me micrograph. The transmission electron micrographs for Burgers

 

 

 

 

52

.mcotaoofiflu mo 33 95 mo 553328 :oboofio commmmfimaafi. .a 0:35

 

 

53

vector analysis of the two sets of dislocations are shown in Figure 10.
The two types of dislocations are denoted as 1 and 2 in Figure 10(a). The
dislocations of type 1 and 2 are approximately oriented along the [0T1]
and [1T0] directions, respectively. The dislocation contrast of the two
sets of dislocations with four different diffraction vectors is listed in
Table 5. From the result, the Burgers vectors of dislocations of type 1 and
2 were found to be a/2 [0T1] and a/2 [T10] respectively. Because some of
these dislocations obviously lie in the same slip plane, the slip plane
which contains both Burgers vectors will be (111). By considering the
direction of the dislocation lines and the Burgers vectors, both

iislocations are also pure screw type dislocations.

Table 5. Dislocation contrast of two sets of dislocations.

g [111] [T11] [11T] [1T3]

 

1 Vi Inv Vi Vi
2 Vi Vi Inv Inv

 

The microstructure of an annealed TaC0_99 specimen, which was
ndentation-deformed at room temperature, is shown in Figure 11(a).
igure 11(a) is a high magnification image of the region delineated by the
otted line in Figure 11(b). In Figure 11, a recrystallized grain marked as

can be seen in the vicinity of the indentation where the heavy plastic

eformation might be occurred during the indentation. The region near

 

 

.HOHOO>

“detonate ”w 6338:. Ewen “m ”N E; 392 we 23387.? no $9223 .832, meowezm .3 0.53m

22.7».S—2um S :oiuu.:_:um§ .___.:uu.:::ua§

54

 

 

55

‘5

| ’.
.21 X}
\'1.7.1 ‘\

 

Figure 11. Transmission electron micrographs of the indented and
annealed TaCo.99: (a), (b); High and low magnification images near the

indentation area of an annealed specimen.

 

 

 

 

 

 

 

 

 

 

56

the, recrystallized grain contains the elongated subgrains indicating
recovery during the annealing process. The high magnification image of
the region next to the recrystallized grain isishown in Figure 12. This
micrograph reveals well-developed subboundaries and hexagonal
networks which are characteristics of recovered structures of face
centered cubic (FCC) metals with high stacking fault energy, e.g.
aluminium. From the observed changes of contrast it can be concluded
that the orientation of the elongated subgrains occurred around the
recrystallized grain is not uniform. A high magnification image of some
of the subboundaries and hexagonal networks are shown in Figure 13. In
the micrograph, twist boundaries (s) are clearly shown and some of
hexagonal networks (11) are connected to the subboundaries by single
dislocation. Further away from the center of the indentation where the
dislocation density was low, the formation of a dislocation cell structure
can be observed in Figure 11(a). A higher magnification image showing
both the subgrain structure and the cell structure is presented in Figure 14.

In order to identify the Burgers vectors of dislocations forming the
hexagonal networks in the annealed specimen, a Burgers vector analysis
was performed on the hexagonal networks and is shown in Figure 15.
Three types of dislocation components of the hexagonal network are
denoted as 1, 2 and 3 in Figure 15(a). The summary of these dislocation
contrasts with five different diffraction vectors is listed in Table 6.
Calculating the cross products of diffraction vectors for which the
invisibility criterion was satisfied, the Burgers vectors of dislocation of
type 1 and 2 were found to be a/2 HOT] and a/2 [0T1]. Because only one

invisibility condition was obtained for dislocations of type 3, an

 

 

 

57

 

 

.Eaum vouafimioou 2: So: comma 2: wEBonm augment: c9502» :

ommmmamcafi. .NH can—mi

 

 

58

 

 

.8233: anwaxo: 98 32355923 2: mcgofi £35828 :ohoofio counting; .3 053m

 

 

59

 

Figure 14. Transmission electron micrograph showing some of subgrains

and cell structure.

 

 

 

[1-1-1]

(0) B=[101]. g

[0-20]

(h) B=[001]. g

[.220]

(a) B=[111], g

60

    

(f) B=[112],g=[-13.]]

[1121.z=to311]

(e) B

[ll-l]

(d) B=[1011. 8

Figure 15. Burgers vector analysis of the hexagonal networks.

 

61

Table 6. Dislocation contrast of dislocations in hexagonalnetworks.

g [ll—T] [HT] [311] [131] [020]

 

1 Vi Vi Vi Inv Inv
2 Inv Vi I nv Vi Vi
3 'Vi Inv Vi Vi Vi

 

unambiguous determination of the Burgers vector was not possible.
However, the dislocations forming the networks should satisfy the
condition that the Burgers vector is conserved at the nodes, i.e. b1 + b2
+b3 = 0, where the b1, b2 and b3 are the Burgers vector of dislocation of
type 1, 2 and 3 respectively. Thus, the Burgers vector of dislocation of
type 3 can be determined as a/2 [1T0]. The result of the directional
analysis indicates that these dislocations are mixed dislocations. Since
the plane which contains all three Burgers vector is only (111), these
dislocations might lie in the (111) plane.

In order to determine the orientation difference between subgrains
)f an annealed specimen, the Kikuchi patterns of two neighboring
subgrain were obtained and indexed (Figure 16). The Kikuchi patterns in
2igure 16(c), (f) and (h) indicate that the beam directions are [T27], [T14]
1nd [215] respectively. These Kikuchi patterns were obtained from the
torresponding subgrain labeled 1. The Kikuchi patterns for subgrain 2 are

hown in (b), (e) and (g) of Figure 16. By comparing these Kikuchi

 

 

 

 

 

 

 

 

.5503? Been ”m .N zmauwnsm we wanted Eons—Um wcmvaommotoo 05

x3 is 3 .Ae 4 Same; .8
A8 .N was _ c2092 232mg“; 95 mo woman: Sou E35

@5223 3033M wqmucommotoo 2: 25 USN Q.

:3 A3 ”mESwnsm 95 5033.“ Emma :otflcotofla 05 «0 £122; 52:3 £0:me 2:. .2 oeswmm

.m—N¢Hn—

 

62

, .32": Pain:

_ gg .,,.'

 
 

 

 

63

patterns with the corresponding Kikuchi patterns for subgrain 1, it can be
noticed that the position of Kikuchi line pairs are displaced relative to
the transmitted beam spot. The misorientation angle (0) between
subgrains 1 and 2 can be determined by the following equations: e=|a| /
L and 0=arctan e, where e is the tilt, a is the shift vector and L is the
camera length [78]. The values of Ial and L used in the present study were
10 mm and 800 mm respectively. So a is 0.0125 or the average of
misorientation angle is about 0.72 degrees. This value is comparable to
the generally reported values of the misorientation angle across low

angle grain boundaries in metals.

 

 

 

 

64

3.4 DISCUSSION

The TEM study of TaCOO99 specimens, which were deformed at
room temperature by the Vickers indentation, revealed that extensive
plastic deformation by slip occurred near the indentation area. The
results of the Burgers vector analysis of the dislocations formed during
the indentation process indicate that the slip system of TaCo_99 at room
temperature is {111}<1T0>. This result is consistent with the earlier
reported slip system of TaC0.96 at room temperature [24,28].

In order to understand the plastic deformation behavior due to the
microindentation, it is necessary to know the stress field forming around
the indentation. Despite the simple nature of the indentation process, the
stresses around an indentation are very complex. Hill [79] treated the
problem of the indentation of plastically rigid material by a smooth
wedge using the slip-line field theory. It was found that the slip-lines
meet the indenter face at 45 degrees and the indentation pressure
increased with the angle of the wedge. The pressure on the wedge was
found to be distributed uniformly and the indentation produced the pile-
up of material around the edge of the indenter. Samuel and Mulhearn [80]
experimentally investigated the deformed zone associated with the
Vickers indentation of a brass specimen. They found that the forces
immediately below the impression are largely compressive and the mode
of deformation resembles the radial flow of material away from the

indenter. The indentation produced a ‘sinking-in’ impression instead of

 

 

 

 

65

‘piling-up’ type.

Marsh [81] considered that the elasticity of the deforming material
is an important factor and pointed out that the mode of deformation in the
work of Samuel and Mulhearn [80] is analogous to the expansion of a
Spherical cavity by an internal pressure. By using the results of Hill [79],

he suggested the following expression for the indentation:

P/Y = C+KBInZ (10)

where, P is Vickers hardness number, Y is yield stress, C and K are
constants, and B and Z are functions of Y/E, where E is the elastic
modulus. It was found that for highly elastic material, i.e. for high Y/E,
the ratio of P/Y was lower than that of an ideal plastic material. For an
ideal plastic material, it is known that P/Y z 3 [79, 82].

Johnson [83] modeled the elastic-plastic indentation problem. It
was considered that the small hemispherical core formed immediately
below the indenter and within the core, a hydrostatic pressure was
assumed. Outside the core, an ideal plastic region exists in which the
stress and the displacements have radial symmetry and are the same as
the case of the expansion of cavity by internal pressure. The elastic
region lies beyond the plastic region. In case of single crystals, the
analysis of deformation behavior by indentation is more complicated due
to the marked anisotropy in plastic properties. Because polycrystalline
TaC was used in the present investigation, it was not possible to align the
edge of the indenter to a known direction on the known crystallographic

plane in a grain. So the exact quantitative analysis of stress field around

 

 

 

 

the

ob:

Ac

ha'

V3.

ca

as

 

66

the indenter was not possible.

However, a rough estimate of the stress which can cause the
observed slip under the indenter can be made by using earlier results.
According to the result of Marsh [81], the ratio of P/Y for glasses, which
have high values of Y/E, was found to be about 2. Although the exact
value of Y/E for tantalum carbide is not known, the Y/E ratio of tantalum
carbide is considered to be high due to its high Y value. So it may be
assumed that the ratio of P/Y for TaCo.99 lies between 2 and 3. Since Y
is approximately equal to 28, where S is the shear yield stress, the shear
stress is equal to the value between 0.17 — 0.25 P. Using the measured
value of Vickers hardness (P) and the reported value of the shear modulus
(u) [18] for tantalum carbide, i.e. P z 12 GPa and it z 216 GPa, the shear
stress developed during the indentation is approximately 0.01 it. At room
temperature the Peierls stress (tp) of an edge dislocation in tantalum

carbide for {111}<1T0> slip can be calculated by the equation [84,85]:

_ 2e . -flié
rp _ -l—_—v- exp( 1-vb) (11)

where, v is Poisson’s ratio, d is interplanar spacing and b is the Burgers
vector. The values of v, d and b are 0.24, al./5 and a/Ji respectively,
where a is the lattice parameter. The calculated value of the Peierls stress
is about 0.003 [1. Thus, to a rough estimation, the shear stress developed
during indentation is high enough to cause the plastic deformation of TaC
by overcoming the Peierls barrier at room temperature.

In Figure 7(a), in the vicinity of the indentation where the stresses

are very high, the highly tangled dislocation structure was formed and

 

 

 

 

this s
symn
diffe:
tang]

away

and

shoe

Fi

TC

 

67

this suggests that the slip in this region probably occurred by several
symmetric {111}<1T0> slip systems. The dislocations gliding on the
different slip planes interact with each other, and form the dislocation
tangles. This can result in high internal stresses in this region. Further
away from the indentation, where the stresses are lower, the slip occurs
on a reduced number of {111}<1T0> systems to accommodate the strain
and stress. A schematic sketch of the slip configuration for this region is

shown in Figure 17.

<111>

 

 

{1111\L \\\\ \/

\>

{111}

 

 

 

 

Figure 17. Schematic of the slip configuration near the indenter.

Two distinctive features of dislocations can be observed in this
region. First, the long and straight screw dislocations are observed along
the <1T0> slip directions. Secondly, some of the dislocations show a
wavy character as shown in Figure 8(a). The long screw dislocations

observed in the current study are very similar to the dislocations

 

 

 

 

 

observ
temper
form
of the
disloc
the th
more
segm
struc
is lit

dislo

dislc
deve
in th
able
disl
{her
be

plan
the
the:
Na

[113

Si!

 

 

68

observed in the body centered cubic (BCC) metals deformed at low
temperatures [86-88]. In BCC metals, it has been established that the
formation of long screw dislocations is mainly due to the core structure
of the screw dislocation [89]. The core structure of a/2 [111] screw
dislocations in BCC metals was described as non-planar and spread into
the three {110} planes of the [111] zone. Because of the dissociation on
more than one plane screw dislocations are not mobile and form long
segments due to the motion of the edge components. Whether the core
structure of a/2 [1T0] screw dislocation in TaCOO99 is planar or non-planar
is not known. Hence, the dislocation motion and consequently the
dislocation arrangement can not be explained at present.

However, a qualitative explanation of the formation of long screw
dislocations in TaC0.99 may be attempted. Due to high shear stress
developed during the indentation, dislocation half-loops can be formed
in the vicinity of the impression. Some of the dislocation loops, then, are
able to expand due to the action of shear stress. If the velocity of an edge
dislocation is assumed to be much higher than that of a screw dislocation,
then dislocation dipoles with long screw segments of opposite sign will
be formed as the mobile edge dislocation segments glide on the slip
plane. The schematic of this process is sketched in Figure 18. Although
the velocities of dislocations are not known for TaCo.99, previous
measurements of dislocation velocities in iron single crystals [88] and
NaCl [90] indicated that the velocity of an edge dislocation is order of
magnitude higher than that of a screw dislocation. The main reason for
the low mobility of the screw dislocation is probably its higher Peierls

stress compared to the edge dislocation. The wavy character of some of

 

 

Figur
segmt
3; ed;

SCl'C\
Thus
TaC[

dish.
be t
Wer
by 1
Btu

inv

 

69

+— 4—- /
——> -—>
s
e e5 ~ /
4J1— S
‘ b <1TO>
[111} - {111}
(a

) (b)

 

 

 

 

Figure 18. Schematic of the possible formation of long screw dislocation
segments from the dislocation half-loop: (a) initial stage, (b) after glide,
e; edge dislocation, s; screw dislocation, b; Burgers vector.

screw dislocations can also be attributed to their high Peierls stress.
Thus, it can be considered that the low temperature deformation of
Taco.” is accomplished mainly by the glide of edge dislocations.

The dislocation structure in Figure 9 shows two sets of
dislocations, dipoles and dislocation loops. A slightly darker contrast can
be noticed between the long dislocations. Because these microstructures
were not observed in the undeformed specimen, they have been formed
by the interaction of dislocations during the indentation. The result of the
Burgers vector analysis confirmed that the two sets of dislocations under

investigation were of pure screw type. Therefore, the possible

 

 

 

interactions
microstruct1
the same sli
joined by a
reaction [9?
the nodes]

by the foil.

The disloc
probably I
very short
can be fc
sePatratim
small, so:
can mm
the clong
Th.
been rep«
at low it
Was prot
TaC0.99
melting
0“lifter

 

 

70

interactions between screw dislocations, which can result in the observed
microstructure, will now be considered. When two screw dislocations on
the same slip plane are meeting at a point as in Figure 9, two triple nodes,
joined by a new dislocation segment, can be formed by a dislocation
reaction [91]. The Burgers vectors of dislocations should be conserved at
the nodes. In the present case, the new dislocation segment can be formed

by the following dislocation reaction:
a/2 [0T1] + a/2 [T10] = a/2 [T01] - (12)

The dislocation segments formed are not shown clearly in the micrograph
probably due to the fact that the length of the dislocation segments is
very short or the invisible diffracting condition is met. The screw dipoles
can be formed by the aforementioned process (Figure 18). When the
separation distance between two screw dislocations of the dipoles is
small, some part of the screw dislocation segments with opposite signs
can attract and annihilate each other. Thus, these dipoles may pinch off
the elongated or small dislocation loops (Figure 9).

The. annealing behavior of most transition metal carbides has not
been reported probably due to difficulties in obtaining large deformation
at low temperatures. In the present study, relatively large deformation
was produced during indentation at room temperature. The annealing of
TaC0.99 subsequent to cold deformation by indentation at about half
melting point resulted in an annealed dislocation structure with the
occurrence of recrystallization and strong recovery. A recrystallized

grain free of dislocations was formed in the vicinity of the impression

 

 

 

 

where the
agreement‘
formed int
Next
dislocation
networks a
during rec
Figure 12
metals wit
evidence
micrograp
distance t
carbon co
beam St]
deformati
revealed
of 2.7-5.1
the dish
contents
“€0.99:
dissociat
COliventi
diSIOCatf
used in
i

clung,“

71

where the dislocation density was very high (Figure 11). This is in
agreement with the established fact that the recrystallization nuclei are
formed in the region where the dislocation density is greatest [92].
Next to the recrystallized grain, well-defined subgrains and
dislocation networks are observed. The subgrains and the dislocation
networks are known to form by dislocation glide and climb processes
during recovery. The characteristics of these recovered structures in
Figure 12 and 13 are very similar to the recovered structures of FCC
metals with high stacking fault energy such as Al. In fact, there was no
evidence of dislocation dissociation in Taco.” from the TEM
micrographs observed. A previous study indicated that the dissociation
distance depends on the stoichiometry of tantalum carbide, i.e. as the
carbon content increases, the width of dissociation decreases [36]. Weak
beam studies of dislocations generated during high temperature
deformation of TaCx, where x was between 0.93-0.95, by Martin [47]
revealed that the dislocations were dissociated into partials with spacing
of 2.7-5.2 nm. However, it was argued that there is an uncertainty about
the dislocation dissociation because of high oxygen and impurity
contents in the specimen investigated [93]. From the current study on
TaCo.99 specimens, there is no evidence that the dislocations are widely
dissociated. The dissociation may have been too small to detect by
conventional TEM. Unfortunately, a good weak beam image of
dislocations was not successfully obtained due to the relatively thick foil
used in the current investigation.
Some of subgrains formed near the recrystallized grain were

elongated and the orientation of elongation tended to change around the

 

 

 

indentation. '
deformation

from the ind
annealing, d?
cell walls. C
the annealed
center of t
dislocation

deformed fc
the driving

dislocation:

off recrysta

72

indentation. This phenomenon is likely due to the strong anisotropy of
deformation in the TaC0.99 crystal near the indentation. Further away
from the indentation where the dislocation density was low before the
annealing, dislocations tend to form cells with low dislocation density
cell walls. Corresponding to the strain distribution owing to indentation
the annealed microstructures of TaCOO99 also vary with distance from the
center of the indentation, and far away from the indentation the
dislocation structure resembles the dislocation arrangement of the
deformed foil before the annealing. This is consistent with the fact that
the driving force of recrystallization and recovery is the strain energy of

dislocations, and that a critical driving force has to be overcome to set

off recrystallization.

 

 

 

3.5 SUMMl

For tl

microinden'

LII

. Extensis

microint

. The dish

oflong

suggest:

motion

. The laci

edge di

. Screw c'

Observe

tiimper:

. Strong]

were 01
These 1

Stackin

73

3.5 SUMMARY

For the first time the dislocation structure of TaC0.99 deformed by

microindentation at room temperature was observed by TEM.

1. Extensive local plastic deformation was observed in the vicinity of
microindentations.

2. The dislocation structure of deformed TaC0.99 consists almost entirely
of long screw dislocations. The wavy character of screw dislocations
suggests the existence of a high Peierls stress for screw dislocation
motion at room temperature.

3. The lack of edge dislocations suggests that plastic deformation of
TaC0.99 at room temperature is mainly accomplished by the motion of
edge dislocations, leaving long screw dislocation segments behind.

4. Screw dislocations were found to form nodes. Screw dipoles were
observed to pinch off, forming dislocation loops during low
temperature plastic deformation.

5. Strongly recovered microstructures and evidence of recrystallization
were observed in TaC0.99 specimens annealed after microindentation.
These phenomena strongly resemble those found in metals with high

stacking fault energy.

 

 

INTI

4.1 INTRO

As no
monocarbir
application
TMMC at
comprehen
TMMC at
to be con
P01ycrysta
presented

Keii
TiC0.915 p.
a11d 1809
bCabout:
732 mm
have stud
stress am
At stress

Was betw

 

CHAPTER 4
INTERMEDIATE-TEMPERATURE CREEP OF TaC

4.1 INTRODUCTION

As mentioned in Chapter 1, the group IV and V transition metal
monocarbides have good potential for high temperature structural
applications. Thus, it is important that the mechanical properties of
TMMC at elevated temperatures are well known and thus have been
comprehensively studied. Among these properties, the creep behavior of
TMMC at elevated temperatures is one of the most important properties
to be considered. A number of creep studies on single crystals and
polycrystals of TMMC have been reported and some of them will be
presented in the discussion below.

Keihn and Kebler [94] have conducted tensile creep tests on
TiC0.96 polycrystals within the range of temperatures between 1600 °C
and 1809 °C in vacuum. The activation energies of creep were found to
be about 544 KJ/mol below 1698 °C for the stress of 55.2 MPa, and about
732 KJ/mol above 1698 °C for the stress of 48.3 MPa. Spivak et al. [95]
have studied creep behavior of TiCo.53,o.96 polycrystals in the ranges of
stress and temperature of 4.9~29.4 MPa and 2000-2650 °C, respectively.
At stresses in excess of about 15 MPa the value of the stress exponent

was between 3 and 4. The activation energy of creep at temperatures of

74

 

 

 

 

2000-2300 ‘
carbon diffs
about 4.9-9.
KJ/mol, wh
diffusion. 'I
14.7 MPa. '
mechanism
the change:
the homoge
compressiv
of tempera
exponent o
3.5 at mod
at tempera

KJ/mol, re

deformatic

temperatur

possible 11

regiun we:

Carbon T

respective

Lei]

ZrC POIY(

in the m

MPa, The

b610w 21

 

75

 

2000-2300 °C was found to be very close to the activation energy for
carbon diffusion in TiC. At temperatures above 2300 °C and stresses of
about 4.9-9.8 MPa, the values of activation energy were about 565-598
KJ/mol, which corresponded to the activation energy for titanium self-
diffusion. The stress exponents were found to be 1 below the stresses of
14.7 MPa. They suggested that creep occurred by the Nabarro-Herring
mechanism under these experimental conditions. They also reported that
the changes in the creep rate with stoichiometry were nonmonotonic in
the homogeneity range cf TiC. Chermant et al. [96] have investigated the
compressive creep properties of polycrystalline TiCo.93 within the ranges
of temperature and stress of 1500-2000 °C and 8.8-93 MPa. The stress
exponent obtained was 2.7 at high temperatures and low stresses, and was
3.5 at moderate temperatures and high stresses. The activation energies
at temperatures below and above 1750 °C were about 380 KJ/mol and 650
KJ/mol, respectively. They concluded that the low temperature mode of
deformation is controlled by the diffusion of carbon, and at high
temperatures, deformation is controlled by the diffusion of Ti. The
possible mechanisms of deformation for the low and high temperature
region were suggested to be the thermally activated formation of kinks at
carbon vacancies and non-conservative dislocation movement,
respectively.

Leipold and Nielson [97] have studied the tensile creep behavior of
ZrC polycrystals, which contains 1-2% free carbon at grain boundaries,
in the ranges of temperature and stress of 1800-2600 °C and 2.5-34.5
MPa. They found that the activation energies for creep were 314 KJ/mol

below 2150 °C and 837 KJ/mol above 2150 °C. The higher value of

 

 

activation en
free carbon.
3. Lee and 1
Mom Sin
temperature
activation r
eontrollin g 1
of dislocatit
with the vof
specimen r
indication t
the dissoci:
Arno:
relatively
compressiv
range of 1
below 22(
“wow i
above 22!
rfispective
crystallog
below 22
diffusion
dislocatir
Nixon et 1

intervals

76

activation energy was ascribed to the effect of impurities, principally
free carbon, at grain boundaries. The stress exponent obtained was about
3. Lee and Haggerty [98] investigated the dynamic creep behavior of
ZrC0.945 single crystals at 1400 °C, 1600 °C and 2000 °C. Over the
temperature range studied, the values of stress exponent and the
activation energy were 5 and 460 KJ/mol, respectively. The rate-
controlling recovery process proposed was the diffusion-controlled climb
of dislocations, although the value of the activation energy agrees better
with the volume diffusion of C in ZrC. The TEM micrographs of a crept
specimen revealed well-defined cell walls, which may be a further
indication of dislocation climb. In addition, no evidence was found for
the dissociation of dislocations into partials in the crept specimen.
Among the group V carbides, the creep properties of NbC are
relatively well-established. Dement’yev et a1. [99] have conducted
compressive creep tests on single crystals of Nb0_89 in the temperature
range of 1800-2700 °C at stresses of 11.8-68.6 MPa. At temperatures
below 2200 °C, the values of the activation energy and the stress
exponent were about 356 KJ/mol and 3.3, respectively. At temperatures
above 2200 °C, these values were changed to 502 KJ/mol and 4.8,
respectively. Based on these values, along with optical and x-ray
crystallographic analyses, they proposed that the mechanism of creep
below 2200 °C was viscous glide of dislocations controlled by the
diffusion of carbon atoms, and that above 2200 °C the mechanism was a
dislocation-diffusion mechanism controlled by the diffusion of Nb atoms.
Nixon et al. [100] have studied the creep of HIPed Nb0.74 polycrystals in the
intervals of stress and temperature of 16.4-69.2 MPa and 1457-1827 °C,

 

 

 

 

respectively.
1657 °C and
respectively
exponents w
and TEM m
of steady-st:
and 1557-1
continued I
controlled l
Che
research or
stress used
The values
1300 °C, 4
energy inc
High volt:
arrangedi
°C, and th
above 170
concluded
and annih
0bStacles
If'gime an
in the hf
extended

 

77

 

respectively. The activation energies in the temperature ranges of 1457-
1657 °C and 1627-1827 °C were found to be 240KJ/mol and 470 KJ/mol,
respectively. In the stress ranges of 16-54 MPa and 48-70 MPa, the stress
exponents were 2.0 and 3.2, respectively. By analyzing the kinetic data
and TEM micrographs, they concluded that the controlling mechanisms
of steady-state creep were either Peierls stress or cross slip at 1457-1497 °C
and 1557-1657 °C. At 1827 °C, under low stresses, dislocation glide
continued to control creep, but under stresses above 64 MPa, creep was
controlled by climb.

Chevacharoenkul et a1. [93,101] performed extensive creep
research on NbC0.775_0.873 single crystals. The ranges of resolved shear
stress used were 6.9 to 44.4 MPa at temperatures from 1100 °C to 1800 °C.
The values of the creep stress exponent were found to be about 3 at 1100-
1300 °C, 4.6 at 1400 °C and 6.8 at 1600-1650 °C. The value of activation
energy increased from 290 KJ/mol at 1300 °C to 546 KJ/mol at 1600 °C.
High voltage electron microscopy revealed that the dislocations were
arranged in the cell structures at temperatures between 1100 °C and 1600
°C, and that the dislocations arranged into subboundaries at temperatures
above 1700 °C. From these‘kinetic data coupled with TEM studies, they
concluded that the creep mechanisms in Nbe involved dislocation glide
and annihilation. The rate controlling processes were the surmounting of
obstacles in cell walls during glide in the intermediate temperature
regime and annihilation of dislocations by Nb diffusion controlled climb
in the high temperature regime. No dissociation of dislocations or
extended nodes were found in a weak-beam dark-field TEM study of the

crept specimen.

 

 

 

Stein:
polycrystal
between 19
no reliable
was estima
found to be
the activat
addition, it
deviation ‘
creep in T.
studies on

The
informatio
elevated tr
at interme
Crhep test
TEM and
microstrua

results of

78

Steinitz [23] conducted tensile creep tests on TaCo_92_0.99
polycrystals under the stresses of 45.9 and 68 MPa, at temperatures
between 1960 °C and 2100 °C. Due to the fluctuation of carbon content,
no reliable value for the stress exponent could be obtained, but the value
was estimated to lie between 2 to 4. The activation energy of creep was
found to be about 711 KJ/mol. This value was considered to lie between
the activation energies of C diffusion and Ta diffusion in TaC. In
addition, it was found that the creep rate of TaC decreases strongly with
deviation from the stoichiometric composition. He proposed that the
creep in TaC is controlled by a diffusion of Ta in TaC. No other creep
studies on TaC have been reported yet.

The main purpose of the present investigation is to obtain
information about the mechanical behavior of TaC0.99 polycrystals at
elevated temperatures, in particular the kinetics and mechanisms of creep
at intermediate temperatures. The temperatures employed for the present
creep tests were much lower than those investigated by Steinitz [23].
TEM and optical microscopy have been conducted to study the
microstructure developed during the creep tests and to complement the

results of creep tests.

 

 

 

4.2 EXPER

4.2.1 Creep

Hot
specimens
TaC0.99 P0
used for 1
chemical l
were pres
specimen:
3 mm in
were pre;

Fir
by a sur
diamond
order to
were p01
smooth 1
181-4 mn
machine
machine
computr

deioniz‘

 

 

79

  

4.2 EXPERIMENTAL PROCEDURE
4.2.1 Creep specimen preparation

Hot pressed and HIPed TaCO.99 polycrystals were used for
specimens of the compressive creep study. The characteristics of the
TaCo.99 polycrystals were the same as those of specimens which had been
used for the room temperature deformation study in'Chapter 3. The
chemical composition and some characteristics of TaCO.99 polycrystals
were presented in Tablel and Table2 of Section 3.2.1. The creep
specimens were circular cylinders of approximately 6 mm in height and
3 mm in diameter i.e. with an aspect ratio of approximately two. They
were prepared from a TaCOO99 block as described below.

First, the top and bottom surfaces of the TaC0.99 block were ground
by a surface grinder (HARIG Model' 618 Automatic) with a 150 grit
diamond impregnated grinding wheel (Diamond Wheel Industries inc.) in
order to obtain flat and parallel surfaces. After grinding, the surfaces
were polished with a 1pm size diamond compound to obtain mirror-like
smooth surfaces. The final dimensions of the block were approximately
11.4 mm x 6 mm x 50 mm. The cylindrical creep specimens were
machined from this block by using a traveling Wire electrical discharge
machine (TWEDM, Raycon/Brother Model HS-300) equipped with a
computer control system. The TaC0.99 block was held by a small vise in

deionized water. The cylindrical specimens were trepanned by electrical

 

 

 

discharge frc
was obtained
the starting

wheel to obt

4.2.2 Creep

Com]
machine (It
temperatur
power syst
control 111
controller
MicroProf
personal t
test progr
of a moly1
KN-100 ll
and a LV]
disPlacen
Tl
consisted
in diamc.
inner shi

80

discharge from the traveling brass wire. A nearly perfectly round surface
was obtained, except for a cosmetic burr where the EDM wire cut through
the starting point. The burr was carefully ground off using the diamond

wheel to obtain the exact roundness.

4.2.2 Creep test apparatus

Compressive creep tests were performed using an MTS testing
machine (Model 810, MTS Systems Corporation), equipped with a high
temperature vacuum furnace (Centorr, Model S-60) and servohydraulic
power system. The operation of the MTS machine was controlled by a
control unit, composed of a MicroConsole (MTS 458.20), an AC
controller (MTS 458.13), two DC controller (MTS 458.11) and a
MicroProfiler (MTS 458.91). The control unit was interfaced to a
personal computer for data acquisition and external control of various
test programs. The two push rods, about 51 mm in diameter, were made
of a molybdenum alloy (TZM). The upper push rod was connected to a 10
KN-100 KN load cell and the lower push rod was connected to an actuator
and a LVDT(Linear variable differential transformer) for measurement of
displacement.

The heating elements of the high temperature vacuum furnace
consisted of a tungsten-mesh. The size of the hot zone was about 76.2 mm
in diameter and 203.2 mm high. Radiation shielding was provided by two
inner shields of W and four outer shields of M0. The furnace was capable

of operating up to 2000 °C either in vacuum or inert gas atmosphere.

 

 

 

The temper:
controller (1
The temper
thermocoup
molybdenur
amalfuncti
additional 1
of the furn:
both high-t
Two
compressiv
block was
platens, ht
between t]
apossible
temperatu
(Eagle-Pi
and 98%
creep Spe
dimensio
For

1500 °c
SPCCimer
to be {00
defoma.

cOntribu

81

The temperature inside the furnace was monitored by a temperature
controller (Honeywell, Model UDC 5000 universal digital controller).
The temperature measurement was achieved by a W-5%Re/W-26%Re
thermocouple which was insulated with BeO and protected with a
molybdenum sheath. In order to prevent possible overheating caused by
a malfunction of the thermocouple during high temperature testing, an
additional thermocouple was installed to limit the maximum temperature
of the furnace. The safety of the furnace system was further provided by
'both high-vacuum and cooling water safety interlocks.

Two TZM compression blocks were used to transmit the
compressive load to the TaC0.99 creep specimen. The upper compression
block was fixed to the upper push rod using two W-pins. The A1203
platens, having dimensions of about 6 mm x 6 mm x 1 mm, were inserted
between the TZM push rods and the TZM compression blocks to prevent
a possible diffusion bonding under the conditions of high stress and high
temperature. Because of its excellent creep resistance, hot pressed oc-SiC
(Eagle-Picher Industries, Inc.), having a purity and density of about 99%
and 98% respectively, was selected for a pair of compression platens. A
creep specimen was placed in between these compression platens. The
dimensions of the SiC platen were about 13 mm x 13 mm x 6 mm.

For the several initial creep tests at temperatures of 1300 °C to
1500 °C and stresses of 30 to 300 MPa, the displacement of the TaC0.99
specimens were measured by the LVDT. However, these data were found
to be too inaccurate because they included large contributions to the total
deformation mainly from the A1203 platens. In order to eliminate the .

contribution of the deformation of the A1203 platen, a high temperature

 

 

extensometc
displacemen
temperature
shields and
TZM compr
the A1203
about 25.4
tests is sho
For t
the creep '
temperatur
compressit
products i
data abov
reaction 1
yield stre
conseque;

the creep

4-2-3 Cre

Af

set up w

evacuate

the 10m]

82

extensometer (MTS, Model 632.51B) was employed to measure the
displacements of the specimen during the creep test. The high
temperature extensometer was equipped with water-cooled radiation
shields and two A1203 extension rods. Grooves were machined into the
TZM compression blocks in order to securely position the knife edges of
the A1203 extension rods. The gauge length of the extensometer was
about 25.4 mm. A schematic drawing of the specimen set-up for the creep
tests is shown in Figure 19.

For the present creep test apparatus, the maximum temperature of
the creep test was limited to about 1500 °C. One of main reasons for
temperature limitation was due to the reaction between the TZM
compression block and the SiC platen. The formation of the reaction
products increased as temperatures increased and could affect the creep
data above 1500 °C. In fact, local melting of the TZM block due to the
reaction had occurred at about 1700 °C. The other reason was that the
yield strength of TZM alloy significantly decreased above 1500 °C and
consequently, plastic deformation of the TZM parts could occur during

the creep test.

4.2.3 Creep tests

After the creep specimen and the extensometer had been properly
set up with a preload of about 70 N, the furnace chamber was closed and
evacuated for 3 to 4 hours. After the vacuum reached about 2 x 10‘5 Torr,

the temperature was slowly increased to the desired test temperature by

 

 

 

 

Figure 1‘.
l-TZM p.
5. SiC p1
rod, 9. E

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 19. A schematic drawing of specimen set-up for creep tests:
1.TZM push rod, 2. W-pin, 3. A1203 platen, 4.TZM compression block,

5. SiC platen, 6. TaC0.99 specimen, 7. Thermocouple, 8. A1203 extension

rod, 9. Extensometer.

 

 

manually in
tests remair
The typical
reached the
a thermal
remained cr
desired let
specimen. .
tests were
be regarde
was obtair
level was
Thrr
TaC creep
°C/105 M
the creep
MPa to 17
creep de
acquisitic
70 hours.
about 2.4
Aft
tTilnsfom
Where e i
length of

crceI) rat

84

manually increasing the set-point of the controller while the mechanical
tests remained under load control to compensate for thermal expansion.
The typical heating rate was about 20 °C/min. When the temperature
reached the test temperature, it was held for about 30 minutes to obtain
a thermal equilibrium before applying the load. The temperature
remained constant within about :1 °C. The load was applied slowly to the
desired level in order to avoid the possible fracture of the TaC0.99
specimen. A typical loading rate was about 8 N/min. Although the creep
tests were conducted under the condition of constant load, they can also
be regarded as constant stress tests since only a very small final strain
was obtained under the present creep conditions. The furnace vacuum
level was about 4 x 10'5 Torr during the creep tests.

Three test conditions were employed to obtain the kinetic data of
TaC creep. The combinations of the temperatures and stresses were 1400
°C/105 MPa, 1500 °C/105 MPa, and 1500 °C/170 MPa, respectively. For
the creep test performed at 1500 °C, the stress was increased from 105
MPa to 170 MPa during the test, in order to find the stress dependency of
creep deformation. The time interval between subsequent data
acquisition was 5 minutes and a typical creep test was run for about 60-
70 hours. The resolution of the data obtained from the computer was
about 2.44 pm.

After the creep tests, the displacement data of the specimens were
transformed to true strain using the relations: e =Al/lo and 8 = In (1+e),
where e is the engineering strain, Al is the displacement, la is the initial
length of the specimen and e is the true strain. In order to calculate the

creep rates, polynomial regression analysis was performed on each set of

 

 

 

 

about 4 data
program. Th
i.e. c = 312 i
The strain r:
correspondi
the strain it
time versus
The
about 60 is
for optical
the direct
specimen
Section 3.
between 1
along the
mm diam:
using the
Th
and an i
Side to t
ion gun
Were Sir
3.2.2.
Perpenr
examin

0f diSit

85

about 4 data points of the time versus strain curve using the PlotIt®
program. The polynomial regression model used was a quadratic equation

i.e. e = at2

+ bt + c, where a, b, c are constants and t is time in seconds.
The strain rate at a particular strain was calculated by differentiating the
corresponding regression equation and substituting the value of time for
the strain into the equation. The PlotIt® program was also used to plot
time versus strain and strain versus strain rate graphs.

The specimens crept at a temperature of 1500 °C and a stress of
about 60 MPa were used for optical and TEM microscopy. The samples
for optical microscopy were sectioned by a low speed diamond saw along
the direction parallel to the compressive direction. The procedures of
specimen mounting, polishing and etching were the same as described in
Section 3.2.1. For TEM samples, thin slices of the crept TaCo.99 specimen
between 0.1-0.2 mm in thickness were cut by a low speed diamond saw
along the direction parallel to the compressive axis. Disc samples of 3
mm diameter were trepanned from the middle portion of the slices by
using the abrasive slurry drill.

The TEM foil was obtained by a combination of dimple grinding
and an ion milling process. The disc specimens were dimpled from one
side to the final thickness of about 25 um and then ion-milled using two
ion guns. The details of the dimple grinding and ion milling procedure
were similar to the TEM foil preparation procedure described in Section
3.2.2. The direction normal to the TEM foil was approximately
perpendicular to the compressive axis of the creep test. Thin foils were

examined in a Hitachi H-800 TEM operating at 200 KV. Burgers vectors

of dislocations were determined by using the standard g~bxu analysis.

 

4.3 EXPERT

A cre
stress of ab
with an ins
decreased
instantanet
slowly thrt
about 0.64
about 105
obtained 1
20. An in

in an imn
onset of
creep in t
the exten
creep str
MP3 wet
In
Plotted
Figure 1
1400 °(
cOutinu

 

 

86

4.3 EXPERIMENTAL RESULTS

A creep curve of TaC0.99 polycrystals deformed at 1400 °C under a
stress of about 105 MPa is presented in Figure 20. The creep curve begins
with an instantaneous creep strain produced upon loading. The creep rate
decreased markedly during the initial short period of time after the
instantaneous strain. The creep rate kept decreasing, although much more
slowly through the transient creep regime. The final strain obtained was
about 0.64%. The creep curves obtained at 1500 °C under the stresses of
about 105 MPa and 170 MPa are shown in Figure 21. The creep curve
obtained under a stress of 105 MPa is very similar to the one in Figure
20. An increase in stress from 105 MPa to 170 MPa at 1500 °C resulted
in an immediate increase in the creep rate. This can be considered as the
onset of an additional transient creep. But, the extent of the transition
creep in the curve obtained from the stress increment was much less than
the extents of the transition creep in the other two creep curves. The final
creep strains obtained at 1500 °C under the stress of 105 MPa and 170
MPa were about 0.77% and 1.3%, respectively.

In order to determine the steady-state creep rate, the creep rate was
plotted versus the creep strain for three creep conditions as shown in
Figure 22. The curves obtained under the creep conditions of 105 MPa at
1400 °C and 1500 °C exhibit a similar trend that the creep rates are
continuously decreasing up to the end of creep tests. The creep rates at

the end of the creep tests were very slow for both creep conditions.

 

 

 

 

 

 

 

87

.00 ca: 23 «m2 2: mo 203280 2: Home: 35390 3.0ka mo 3.30 @030 6N 033m

AouumoE SE.
com on P on
com com so P o

P-.bi—i-trLPph..LLbi-blkt-L-pribi-i- 00°

 

U

1’6

 

N6

1 T I

4, n.

o o
(%) mums

 

 

 

 

 

 

 

5:2 o: 3 52 2: no

 

”Do 82 3 82 c: was «m2 me .3 2332350 on. 59:. 3605. no 32:0 @020 Am 9:.me

Aoommo 3 g...
0.9. can com on P om
com 00¢ com com 00 F o

L .lP-k-bbtfi.» —bl~bbrbtt —t+b len tbeblP- D} p b — p p n F. 00°

 

88

 

 

 

 

 

 

8°T .

 

89

 

do 8232 o: I do 8232 2: 8 do 8352 2: i ”£8; “8 8:23 a a 3.: £25 .2 8:5

 

3 225m
om. P 00. P omd cod 0*.0
on. P o P. p omd and 00.0 and
r L b 4 u p b P P p n p L L L n L L[ I? p o

(1-°ase_01x) mm NIVHLS

 

 

 

 

 

 

 

Although th
test conditic
to be nearly
as the quasi
and 1500 ‘
beginning :
be regardet

rates for t!

Tab1c7. Q

 

Tcmperat

~

140'

 

150

150

 

F1
the acti'

equatio:

 

90

Although the true steady-state creep stages were not reached for these
test conditions, the creep rate at the end of the creep tests were assumed
to be nearly equal to the steady-state creep rates and thus, were regarded
as the quasi steady-state creep rate. For the creep conditions of 170 MPa
and 1500 °C, the creep rate decreased with the creep strain at the
beginning and eventually attained a relatively constant value, which can
be regarded as the steady-state creep rate. The quasi steady-state creep

rates for three creep test conditions are given in Table 7.

Tab1e7. Quasi steady-state creep rate data of TaC0_99 polycrystals.

 

 

 

Temperature (°C) Stress (MPa) Quasi steady-state creep rate (x10‘9/sec)
1400 105 5 . 10
1500 105 9 .34
1500 170 20. 80

 

 

 

From the equation (4) in Section 2.3.2, the stress exponent(n) and
the activation energy for creep (QC) can be obtained using the following

equations:

log (til/(~52) (13)

 

" = log(0'1/0'2)

 

 

 

 

 

 

Where, til 3
respectivel
state creep
value of tl
energy for
A T
MPa is prc
consists c
dislocatic
interactic
dislocatit
subgrain
strain, b'
01
stress 0
Many
Specime
mainly
slightly
is indit
microg;
which

located

 

 

91

Rln (éTl/éTz)

= 14
Q, 1/T2— 1/T1 ( )

 

Where, 61 and é2 are the steady-state creep rates at stresses of 01 and 0'2,
respectively. R is the universal gas constant. $511 and (ET2 are the steady-
state creep rate at temperatures of T1 and T2, respectively. The calculated
value of the stress exponent at 1500 °C was about 1.8. The activation
energy for creep was calculated to be about 150 KJ/mol.

A TEM micrograph of a specimen crept at 1300 °C and about 300
MPa is presented in Figure 23. The dislocation structure of this specimen
consists of dislocation dipoles, numerous dislocation loops and curved
dislocation segments probably due to the dislocation-pinning and. the
interactions with other dislocations. No evidence for the formation of
dislocation cells or subgrains can be observed. The dislocation cells or
subgrains may not be formed under this creep condition with very low
strain, but also the area observed in the TEM foil was not very large.

Optical micrographs of a specimen crept at about 1500 °C under a
stress of about 60 MPa is shown in Figure 24. A low magnification
micrograph in Figure 24(a) shows the overall microstructure of the
specimen after the creep test. Many voids and cracks can be observed
mainly along the grain boundaries. The size of voids and cracks may be
slightly exaggerated due to the chemical etching. The compression axis
is indicated by two arrowheads. Figure 24(b) is a high magnification
micrograph taken in the middle of the specimen shown in Figure 24(a),
which substantiates that most of the voids and cracks are preferentially

located at those grain boundaries, which are relatively parallel to the

 

 

 

 

 

 

.3: 8m 2“ 0.. 82 a :58 .3639 no £3826 2m... .8 23E

92

 

 

 

 

 

omen: cotaoumqwaa «RE Q:

.0925 8:33:qu 32 C3 ”and; ow US“ 00 com“ 3 “mono 3605. .«o mamaumosmfi 3330 .vu 0.33m

93

 

 

94

compression axis. Some of the voids and cracks seem to have formed by
the interconnection of small pores in the grain boundaries. A few wedge
cracks were also seen at three grain junctions. The formation of the voids
and cracks at grain boundaries suggests that the grain boundary sliding
may have occurred during creep deformation.

Dislocation structures in TaC0.99 specimens crept at 1500 °C under
a stress of about 60 MPa are presented in Figure 25 to Figure 36. Figure
25 shows a subboundary, and individual dislocations. Figure 26 shows a
higher magnification micrograph of the subboundary presented in Figure
25. The subboundary is a tilt boundary, which is composed of a set of
straight parallel dislocations. Figure 27 shows another subboundary
which seems to be less perfect than the one presented in Figure 25. The
spacing between dislocations forming the subboundary is much larger
than that of the subboundary shown in Figure 26. A few dislocations of
the subboundary appear to interact with the individual dislocations inside
the subgrain. At higher magnification it becomes evident that the
dislocations which comprise the boundary have many kinks or jogs
(Figure 28).

Another TEM foil of crept Taco.” shows a well-defined twist
boundary, which is composed of two sets of long dislocations (Figure 29).
The dislocations forming the twist boundary are curved along the
boundary which implies that the boundary plane is distorted. Figure 30
shows a higher magnification micrograph of the twist boundary now
imaged with a different g vector, g=[02'2']. It can be noticed that one set
of boundary dislocations is out of contrast according to the diffraction

condition. The Burgers vector of these dislocations is probably of type

 

 

 

 

 

 

.HOuOO>

Stowaway "w £2382. anon um "5:2 co 9; Do 82 3 BPS maéofl. mo £923.83. Emana 053m

95

 

 

 

 

.882, notoSbE ”a

£3833 axon mm "mm 83E 5 =32: Emu—uncanny. 2: mo :msumoHoma eotaowmawufi swam 6N 253m

96

 

 

 

.853 :otoacmflu mu .cotoohv anon

.m ”€332:va 3533?: 93 33.52233 a wEBonm 360$. 330 we £35838 2m; .hm 3:me

97

 

98

SN owswmm 5 =39: bacaaoaasm on“ we amaumosma aoflaoamcwafi sum.“ .3 ouawwm

 

99

 

.H0u00>

:ououbfic ”w .aotoohc Egon ”m $53552 2:5 a wEBonm 360.3. 30.8 .«o £33828 Emu. .mN 033m

8 fin; u M An T a any: n. n

 

100

.882, nouoahfiv ”u

£3883 83a ”m ”mm oSmE E :32: muuvcson SE3 2: mo :mfiwosmfi ”8:35:38 swam .8 053m

. an}. .87..

 

101

a/2<01T>.

In another area of the specimen, a number of short and straight
individual dislocations can be observed (Figure 31). It can also be
noticed that some of long dislocations were pinned by these short
dislocations, and that the curved dislocation segments were formed
between the pinned-points. A Burgers vector analysis of these
dislocations is presented in Figure 32. The dislocation contrast of three
types of dislocations, denoted as 1, 2 in Figure 32(a) and 3 in Figure

32(0), is listed in Table 8.

Table 8. Dislocation contrast of individual dislocations.

 

 

g [Tm [700] [MT] [200]
1 Vi Vi Inv Vi
2 Vi ' Inv Vi Inv
3 Inv Vi Vi Vi

 

 

where, Vi and Inv denote that the dislocationsare visible
and invisible, respectively.

From the results, the Burgers vectors of dislocations of type 1, 2 and 3
were found to be a/2 [1T0], a/2[01T] and a/2[10T], respectively. Analysis

of the line directions and Burgers vectors of three types of dislocations

results that they are all mixed type dislocations.

 

s58?» :ocoatfivuu

£38.33. 839 .m umqoflaooig 1333?: wEBonm 360a. 305 we £35828 2m“. .Hm 0.53m

102

 

 

 

103

 

:ouoahfluuu £3333 Egon "m "m 93 NJ

.H0u00>

on? mo 20:30:? mo 32:“: .58: flown—7m .Nm 0.33m

 

 

104

Figure 33(a) shows a subboundary and hexagonal dislocation
networks. The high magnification micrograph of a part! of the
subboundary is shown in Figure 33(b). A set of dislocations comprising
the subboundary appears to be out of contrast (Figure 33(b)). A Burgers
vector analysis of dislocations in the hexagonal networks is given in
Figure 34. The resulting dislocation contrast (Visible or Invisible) of two
types of dislocations, denoted as 1 and 2 in Figure 34(a), is listed in Table

9.

Table 9. Dislocation contrast of dislocations in hexagonal networks.

g [1‘31] . [0:0] [702] [Th]
1 Vi Inv Vi Vi
2 Vi Vi Vi Inv

 

 

 

From the above result, the Burgers vectors of the dislocations type 1 and
2 should be a/2 HOT] and a/2 [1T0], respectively. Because of their
orientation both types of dislocations are mixed dislocations.

Figure 35 shows TEM micrographs of a Burger vector analysis
performed on another dislocation networks. Two types of dislocations
consisting the networks are denoted as 1 and 2 (Figure 35(a)). The
summary of dislocation contrast of these dislocations is presented in
Table 10. The Burgers vector of dislocation type 1 and 2 are found to be

a/2 [011] and a/2 [1T0], respectively. Again, these dislocations are mixed

105

 

 

.883 notoecmflxu £3533 Enos ”m .baunsonnpm 2: Ho 0925 :otaofawaa
HoswE SV .3333: 3:330: 95 rag—scan"; a A3 "350“,“. 308 .«o mnaauweoma Emu. .mm 053m

) 1 . u 131;. 51|\ iv. ill I‘llllllel ).\td \lliull ll 1

a H- I u.» an 3? m

 

”mm.THuu .3:an

   
  

, :
E: ,0 C

 

.Q. n.

J i. w...

 

 

. st
.HoHoo> 5522“,:va .HHoHHooHE Egon cm 3833: Haaowaxo: 2: .3 23:28 532, mHowHH—m Hum 0H .m
H

H

_

 

106

.0 .1. v.1. -,

. fly 1H Wu» H: 1.3 3.

.
1 .
A Ilifliunl’yt

HZ..HHHH.:3HHHH3H,

.n 311A-

      

    

    
 
  

H
4

EH. «6

  

 

 

107‘

.883 coHHoafihunu .

:oHHooHHv Egon Hm ”3.850: cotmooumv 2: mo $3223 .88?» .20th .2 053m

 

 

 

 

type dislocations.

Table 10. Dislocation contrast of dislocations in networks.

g [720] [111] [Off] [202]

 

1 Vi Inv Inv Vi

 

2 Vi Inv Vi Vi

Finally, Figure 36 shows a grain boundary and a subboundary
which composed of two sets of dislocations intersecting each other. A
misorientation angle across this subboundary was measured using the
Kikuchi pattern method as described in Section 3.3. The calculated value

of the misorientation angle was about 0.6 °.

 

 

 

  
  

$36559 5an a E; ravasonnnm a wEBonm gdofl. H998 mo :QmHmoHoHE 2m: 6m 053m

 

 

110

 

4.4 DISCUSSION

As discussed in Section 2.3, it is possible to elucidate the
controlling-mechanisms of creep by comparing the kinetic data of a creep
experiment with the theoretical steady-state creep models. Two important
kinetic data are required to be examined for that purpose. These are the
value of the stress exponent and the activation energy for creep. It is also
important to know the creep conditions, i.e. the applied stress and
temperature, because the operative creep mechanisms may change with
these two variables.

The temperatures employed for the present creep tests were about
1400 °C and 1500 °C. These correspond to the homologous temperatures
of about 0.39 Tm and 0.42 Tm (Tmz3983 °C), respectively. Thus, the
present investigation can be considered as intermediate-temperature
creep. Since no data exist for the temperature dependency of the shear
modulus of TaCo.99, the value of shear modulus of TaC0.99 at 1500 °C can
only be estimated roughly by using a rule of thumb, i.e. the value of shear
modulus at the melting point is one half of the value at absolute zero
temperature. Using the value of shear modulus of TaC0.99 at room
temperature (#216 GPa), the shear modulus(tt) of TaC0.99 at 1500 °C is
estimated to be about 177 GPa. The applied stresses for the present creep
experiments were about 105 MPa and 170 MPa. These correspond to
approximately 6x10'4 31 and 9.6x10'4 u, respectively. Thus, the creep

conditions applied for the present investigation are in the regime where

 

 

 

111

power-law creep behavior is usually observed for many metals and
alloys.

From the result of the creep tests of TaC0.99 polycrystals, the
calculated value of the stress exponent was close to 2. As discussed in
Section 2.3, when the creep is controlled by diffusional creep, the value
of stress exponent is 1. The stress exponents of dislocation creep
mechanisms are about 3 and 5 depending on whether the creep is
controlled by the dislocation glide or the dislocation climb. Thus, the
value of the stress exponent obtained in the present creep study lies in
the regime between diffusional creep and the dislocation glide-controlled
creep. This suggests that more than one mechanism may have contributed
to the creep of TaCo.99 under the applied conditions.

It has been known that relations exist between the activation energy
of self—diffusion and the activation energy for creep at high temperatures.
The value of activation energy for carbon diffusion in TaC was reported
as about 360 KJ/mol in the temperature range between 1800 °C and 2700
°C [102], and about 497 KJ/mol between 2360 °C and 2960 °C [103]. The
value of activation energy for Ta diffusion in TaC is not known, but it is
, considered to be higher than that for C diffusion. Within experimental
error, the value of activation energy for creep obtained in the present
study was about 150 KJ/mol. Although the value of activation energy for
diffusion generally decreases with decreasing temperature, the obtained
value of activation energy for creep is thought to be too low to match with
the values of activation energy for either C or Ta diffusion in TaC. Thus,
the low value of activation energy for creep suggests that creep of

TaCOO99 under the present creep conditions may not be controlled by

 

 

 

112

diffusion mechanisms, which is also likely since the temperatures
employed for the present creep study are lower than 0.5 Tm.

An important aspect of the creep of TaC0.99 under the present creep
conditions is the formation ‘of voids and cracks along the grain
boundaries as observed in Figure 6. Gifkins [104] and Chen et al. [105]
proposed the mechanisms for the formation of intergranular voids or
cracks. Although the details of two mechanisms are different, a necessary
prerequisite for both mechanisms is grain boundary sliding. Chang and
Grant [106] also proposed models for the formation of various shapes of
wedge cracks due to grain boundary sliding which were also observed in
this investigation. Thus, the intergranular voids and cracks formed during
creep of TaC0.99 may be due to grain boundary sliding. The formation of
intergranular voids or cracks by grain boundary sliding has also been
observed for a number of crept ceramics, e.g. MgO [107], Si3N4 [108] and
Th02 [109].

Langdon et al. [110] reviewed the experimental data on grain
boundary sliding in metals and noticed that grain boundary sliding can
make a significant contribution to the total creep deformation. They
found that the value of the stress exponent for grain boundary sliding was
lower than the stress exponent for the overall creep process and the ratio
of activation energy for grain boundary sliding and activation energy of
lattice self-diffusion ranges from about 0.5 to 1. Although the relative
contribution of grain boundary sliding to the observed creep behavior of
TaC0.99 is not known, the obtained values of stress exponent and
activation energy for creep of Taco.” at intermediate temperatures

indicate that creep is probably affected by grain boundary sliding.

 

 

 

113

Although TEM micrographs could not show the overall
substructures of crept TaCo.99 specimens, some of important features of
dislocation structure were evident. The dislocation structure of TaCO.99
crept at 1300 °C and 300 MPa reveals extensive dislocation interactions
during the creep test (Figure 23). The microstructure also shows
numerous dislocation loops which result from the pinch-off process of
dislocation dipoles. This suggests that recovery under these conditions
occurs by annihilation of dislocations of opposite sign. However, no
conclusive evidence of dynamic recovery processes such as cell or
subgrain formations were observed.

The dislocation structure of TaC0.99 specimen crept at 1500 °C and
60 MPa consists mainly of subboundaries, dislocation networks and
individual dislocations. These features indicate that the dislocation glide
activity at 1500 °C is stronger than that in the specimen crept at 1300
°C. A number of TEM micrographs (e.g. Figure 25, 31, 35, 36) reveal
interactions between dislocations occurs during creep. While the
formation of hexagonal dislocation networks and subboundaries (Figure
33) may have involved also dislocation climb, the prominent features of
the microstructure under these conditions seem to be due to dislocation
glide and dislocation interactions.

The results of the Burgers vector analysis of the dislocations
formed during the creep test substantiate that the slip system of TaCOO99
at 1500 °C is {111}<1T0>. This result is consistent with earlier reports
[25]. The characteristics of the dislocations formed during the creep test
are quite different from those of dislocations formed during the room

temperature deformation (Chapter 3). Most dislocations formed during

 

 

 

114

the creep test (e.g. Figure 31) appear to have no preferential orientation
in contrast to the long screw dislocations formed during the indentation-
deformation at room temperature. The main reason for these differences
is probably due to the fact that dislocation glide over the Peierls barriers

becomes easier at elevated temperatures due to thermal activation.

 

 

,— .

 

 

 

 

 

 

115

4.4 SUMMARY

1. From the results on creep tests at intermediate temperatures (1400-

 

1500 °C, ~ 0.4 Tm) and stresses (105-170 MPa), the values of the
stress exponent (n) and activation energy (QC) for power-law creep of

TaC0099 were found to be n==l.8 and Qc~150 KJ/mol.

2. The low activation energy and the low stress exponent suggest that ‘
the creep of Taco.” at intermediate temperatures is mainly controlled
by grain boundary sliding. This is further supported by observation of

void formation.

3. The observation of dislocation interaction in the crystal interior
suggests that dislocation glide may act as a minor contributing
mechanism during creep. TEM investigation also reveals the

formation of subboundaries and dislocation networks.

4. Most dislocations formed during creep of TaC0.99 have no preferential
orientation, in contrast to dislocations formed at low temperatures.
_ This is probably due to thermally activated dislocation motion at

elevated temperatures.

 

 

CHAPTER 5
PROCESSING OF C-TaC COMPOSITES

 

5.1 INTRODUCTION

In recent years the fiber reinforced ceramic matrix composites have
received considerable attentions primarily because of their potential
applications as a structural material at elevated temperatures. One of
main advantages of ceramics is their retention of strength at high
temperatures and low density as compared with metals. The main
disadvantage is their low fracture toughness or resistance to crack
propagation. The low toughness is a consequence of very limited
dislocation motion in ceramics. Because of their low toughness, ceramics
are susceptible to damage by thermal and mechanical shock. Also the
strength of ceramics can be drastically reduced by flaws or damages
introduced during fabrication or in service. Thus the main incentive for
developing the fiber reinforced ceramic matrix composites is to improve
their toughness as well as their strength.

In general, the fiber reinforced ceramic matrix composites can be
divided into two basic types, the system based on short fibers or whiskers
and the continuous fibers. The composites containing short random fibers
or whiskers are developed mainly for increasing the toughness of ceramic

matrices. The strength of these composites are frequently lower than

116

 

 

 

 

 

 

 

 

 

117

unreinforced matrix due to the combined effects of stress concentration
and residual thermal stresses. An improvement in strength has been
reported only for a few systems, for example SiC whisker reinforced
A1203 [111].

A greater increase in strength and toughness of ceramic matrix
composites can be obtained by reinforcing ceramics with continuous
aligned fibers. In continuous fiber system the stress concentration at the
ends of fibers is minimized and fibers have the major role as the load
bearing component while matrix has the secondary role of providing a
load transfer medium. The further advantage is that higher volume
fractions of fiber can be obtained as compared to short fiber system. Thus
the fibers are required to have high strength and high stiffness at elevated
temperatures in order to obtain the composite strengthening effect. The
work on carbon fiber and silicon carbide fiber reinforced glass and glass-
ceramic composites demonstrated the great potential for this type of
composite, both in terms of high strength and fracture toughness [112-
114].

Several important toughening mechanisms have been considered
for ceramic matrix composites. These include load transfer from a matrix
to fibers, prestressing of the matrix due to differences in the coefficients
of thermal expansion of fibers and the matrix, crack impediment by
fibers, microcracking of the matrix, crack deflection and fiber pull-out
[1.15]. It has been found that typically, in a given composite,
combinations of the aforementioned toughening mechanisms are
operative. The relative contributions of individual toughening

mechanisms generally depend on interfacial properties, microstructure of

 

 

 

 

118

composites, and the elastic constants of fibers and matrix.

The mechanical properties and behaviors of composites critically
depend on the bond strength of an interface between the fiber and the
matrix. In contrast to polymer and metal matrix composites, the
interfacial bond strength of ceramic matrix composites often need to be
low in order to improve their toughness but also need to be high enough
to improve their strength. Therefore, it is important to control the
interfacial bond strength in order to obtain the optimum combination of
toughness and strength in fiber reinforced ceramic composites.

The basic requirements of materials which can be used in fiber
reinforced ceramic composites are high melting points and high strength
often coupled with low density. The other important requirements which
must be considered in choosing a particular fiber-matrix system is the
compatibilities betWeen the two components at elevated temperatures.
Three types of compatibility between fiber and matrix are need to be
satisfied in order to form a successful composite. These are chemical
compatibility, thermal expansion mismatch and relative elastic modulus
of the fiber and matrix [116].

Among fiber reinforced ceramic composites, the glass and glass
ceramic matrices are not very attractive for high temperature applications
because the matrices loose their strength at relatively low temperature.
The potential matrix materials for higher temperature applications
include mainly the refractory ceramics such as carbides, oxides, nitrides
and borides. Although numerous potential composite systems can be
selected from the various refractory ceramics, the number of practical

systems is limited owing to the fibers that are available. At present the

 

 

119

important high performance fibers commercially available are carbon,
silicon carbide and alumina fibers. Alumina and silicon carbide fibers are
more oxidation resistant than carbon fibers and may be used without
oxidation barrier. But, in non-oxidizing atmosphere, carbon fibers
provide the highest temperature capability.

Various fabrication techniques of fiber reinforced ceramics have
been deve10ped, and a number of reviews on this subject have been
published [116-119]. Because of the refractoriness of ceramic matrices
the conventional fabrication techniques for ceramic matrix composites
employ powder processing methods. Three common problems
encountered from powder processing of ceramic matrix composites have
been reported by Rice [118]. First, the fibers can be damaged by
refractory hard matrix powders. Second, a large porosity at the fiber and
matrix interface can be formed due to insufficient formability of the
refractory matrix around the fibers. Third, the chemical interaction
between the fiber and the matrix increases due to higher processing
temperature for more refractory matrices.

In general, a fiber reinforced ceramic can be fabricated by using a
two—stage process [116]. The first stage is to incorporate the reinforcing
fibers into an unreinforced matrix and the second stage involves the
consolidation of the matrix at high temperatures. The general
microstructural requirements for high strength of the resulting fiber
reinforced composites are high volume fraction of fibers, uniform
distribution of fibers, high density and homogeneous ceramic matrix
[117].

For short fiber or whisker reinforced composites, the

 

 

 

120

reinforcements can be readily mixed with the matrix powder or slurry.
Uniform dispersion of the reinforcements becomes more difficult as the
volume fraction and length of reinforcements are increased and the
clustering of reinforcements may occur. In case of continuous fiber
composites, the most common technique of fiber incorporation is the
slurry infiltration process. In the slurry infiltration process fibers are
impregnated with the matrix by passing through a slurry of the matrix to
produce a pre-preg sheet. The pre-preg sheets are then laid into desired
configuration followed by binder burnout and consolidation processing.
The slurry usually contains the carrier liquid and an organic binder in
addition to the matrix powder. The volume fraction of fibers in a
composite can be controlled by adjusting the composition of the slurry.

The main goal of the consolidation stage is to produce a low
porosity matrix while damaging the fibers as little as possible. The
various consolidation processing techniques, such as sintering, hot
pressing and hot isostatic pressing, which have been applied for
unreinforced ceramics can also be applied to ceramic matrix composites.
Since the refractory ceramics are usually difficult to sinter to obtain full
density and furthermore the fibers tend to inhibit the densification,
ceramic matrix composites produced by sintering often exhibit poor
mechanical performance due to high porosity. The most widely used
technique of consolidating fiber reinforced ceramics is hot pressing. For
example, carbon and silicon carbide fiber reinforced glass and glass
ceramic composites has been successfully fabricated by using the slurry
infiltration process and hot pressing [112-114].

The hot isostatic pressing (HIP) is a relatively new technique in

 

 

 

 

 

121

which high isostatic gas pressure is applied to a powder part or compact
at elevated temperatures to produce a material with near theoretical
density [120-122]. An improvement in mechanical properties is reported
for hot isostatically pressed ceramic matrix composites, for example SiC
whisker reinforced alumina matrix [123] and SiC whisker reinforced
silicon nitride matrix [124] composites. The main advantage of the HIP
process is that superior quality of a product can be obtained owing to the
combined effects of low porosity and uniform and fine grained
microstructures. A further advantage is its ability to fabricate complex
shapes. The disadvantages of the HIP process are that it requires rather
complicated preparation and the processing cost is relatively high.

The other fabrication technique of fiber reinforced ceramic matrix
composites include chemical vapor deposition (CVD) infiltration and sol-
gel techniques. Silicon carbide fiber reinforced SiC matrix and Si3N4
matrix composites were fabricated by CVD infiltration technique [125].
Fabrication of A1203 and SiC fiber reinforced Alumina matrix composites
by using sol-gel technique was also reported [126]. Although these
fabrication techniques have been most recently developed, at present
these are less commonly used due to the shortcomings such as high
porosity and low yield.

As mentioned earlier in a previous chapter the transition metal
monocarbides possess very attractive properties that can be used as a
matrix of composite materials for high temperature structural
applications. However, only very few investigations on fiber reinforced
transition metal monocarbides have been reported. For example, short

graphite fiber reinforced niobium carbide composites were produced by

 

 

122

sintering of cold compacted composites, and also by the slurry
infiltration process followed by hot pressing technique [127,128]. A
carbon fiber reinforced tantalum carbide matrix composite was fabricated
by using a three-step chemical reaction process [129]. In this processing,
the yield was low and the residual porosity of the composite was 5 to 10
percent after multiple impregnations. Detailed information on these
composites, for example interfacial properties and microstructures of the
composites, cannot be obtained because of patented work.

In the present investigation, a continuous carbon fiber and
stoichiometric tantalum carbide matrix composite system was studied as
a potential material for high temperature structural applications. As
described earlier TaC possesses attractive properties for high temperature
applications including the highest melting point, high strength, high
stiffness and high temperature ductile behavior similar to f.c.c. metals.
As a reinforcement phase C-fibers exhibit high specific strength and
stiffness at high temperatures. Because the stoichiometric TaC matrix is
already saturated with carbon, the diffusion of carbon between two
components is not expected to occur at high temperatures. Therefore, the
principal advantage of C-TaC system will be its thermal and
compositional stability at high temperatures. In the present study carbon
fibers were incorporated into a TaC matrix by using either a cold
compaction process or a slurry infiltration and cold compaction process.
Hot isostatic pressing (HIP) is used as a hot consolidation technique. The
nature of the interface between carbon fibers and the tantalum carbide
matrix of the composites was investigated. The problems associated with

the processing and the possible solutions will be discussed.

 

 

 

 

 

5.2 EXPERIMENTAL PROCEDURE

5.2.1 Characteristics of starting materials

Two different types of TaC powder and carbon fibers were
employed for producing C-fiber reinforced TaC composites. The
important characteristics of these materials are summarized in Table 1. A
major difference between type A and type B of TaC powder was the initial
particle size. In both types the carbon fibers were polyacrylonitrile
(PAN)-based high modulus and high strength type carbon fibers. AS-4
fibers have a larger diameter and higher tensile strength than those of
APOLLOTM 55 fibers. But Young’s modulus of APOLLO TM 55 fibers was
higher than that of AS-4 fibers.

The morphology of TaC powders and C-fibers were observed by a
scanning electron microscope (JEOL model ISM-35C) and these are
shown in Figure 37. As can be seen in Figure 37(a) and (c), the shape of
type A and type B TaC powder is angular and rounded respectively. In
Figure 37(b) the surface of APOLLOTM 55 fibers shows many aligned
ridges which are typical of high modulus carbon fibers [130]. On the
other hand, the surface of AS-4 fibers in Figure 37(d) is very smooth.

Two kinds of C-TaC composites were fabricated using different
carbon fibers and TaC powder. APOLLOTM 55 fibers were incorporated

into type A powders, and AS-4 fibers were used with type B powders.

 

124

Table 11. Characteristics of as-received TaC powder and C-fiber.

 

 

 

Vendor

 

 

Al 0.01, Nb 0.05, B<0.001
Impurity(%) Ni 0.005, Co 0.01-0.06
Si 0.01, Fe 0.01, Ti 0.005

Cerac Inc.

 

TaC powder
Type A Type B
C/Ta ratio 0.99 0.99
Particle size(ttm) < 45 0.5 - 1.5

free C 0.08, Ca 0.003
Fe 0.007, Nb 0.13
Si 0.001, Ti 0.001

Hermann C. Starck

 

Carbon fiber

 

 

 

 

APOLLOTM 55-500 AS-4
Tensile strength(GPa) 3.45 3.79
Young’s modulus(GPa) 379 235
Ultimate elongation(%) 0.91 1.53
Filament diameter(ttm) 5 8
Vendor Fiber Materials Inc. Hercules Inc.

 

 

 

 

 

 

 

 

spa Him... H3 .5133 m 8E3 .HBHH 3 5015012 2: .533

09¢. A3 .30an Home: was mHockom OS. ~538ng mo mnaaHmoHoHE eoHHooE miccmom 41m ouswE
< .

imaamu‘ . 'H/‘o

125

 
 

 

126

5.2.2 Fabrication of C-TaC composites

 

Two different processing routes were designed to produce C-TaC
composite specimens. The block diagrams of each processing route are
shown in Figure 38. The main purpose of the processing route #1 was to
produce a composite specimen which can be used to study the
characteristics of an interface between C-fibers and TaC matrix. So, in
the processing route #1, the as-received C-fiber tows were used without
large effort in achieving exact alignments and separations of an
individual filament of fibers. Carbon fibers and TaC powder used for the
first processing route were APOLLOTM 55 fibers and type A powder
respectively.

For the processing route #1 in Figure 38, HIP containers were
fabricated from a Ta tube, Ta sheet and Ta rod. The inner diameter and
the wall thickness of a Ta tube was about 19.5 mm and 3 mm respectively.
The thickness of a Ta sheet was about 1.5 mm and the diameter of a Ta
rod was about 3.2 mm. The tube was cut into three pieces and the length
of each piece was approximately 56 mm. Ta discs with a diameter of 19.5
mm were machined out of a Ta sheet, and these were used as top lids,
bottom lids, top and bottom spacers. Each disc for a top lid was drilled
at the center to make a hole with a diameter of 3.2 mm. A Ta tube stem
of about 30 mm in length and 3.2mm in diameter was made from the Ta
rod by drilling and was inserted into a hole of a top lid. Two parts are
then tungsten inert gas (TIG)- welded in an Ar gas atmosphere. A bottom

lid was press- fitted into the Ta tube and also TIG- welded. Some of Ta

 

 

 

 

 

 

127

 

 

.moHsoH mimmoooa 380.550 05.10 93 .3 252mm? moo—a 2:1 .wm 2:3"—

 

UZ—mmmzm U—F<Em0m— #0:

t

 

 

 

 

 

mam—ZSFZOO E: ho UZ—Jafim EDDO<>

 

 

 

H

 

 

OZ-mm<OmQ Hm J<>02m- mam—z:—
AH

 

 

 

 

 

”Hus—ECU Zm—m—zc .mO ZOEZADmEHHUZH

 

 

 

:

 

 

ZOE—1045260 CHOU Hz L21><A A<DZ<E

 

 

 

2

 

 

 

qummE: 5539 .5:
H

 

 

 

 

 

mZ<U «E m0 02—15.”.5 Hm ZO—P<DU<>W

:

 

 

 

 

 

”5:28 zmmzu as @2535:
H

 

 

 

 

 

mkUfi—EOU Zm—m—y—U rmO ZO_E<ADmL<UZH

:

 

 

 

 

 

 

wan—«Q. GmEmOhm—mm .mO UZ—EHDU Hm UZ_>~_D

 

 

H

 

 

mam—E.”— ZO=m<U
21:3 awn—30m 05,—. .H—O ZO—FUxfi—EOU Dx—OU

 

 

UZ—wwm—Uozm ZOE.<K.HAEZ— 1 >mme—m

 

 

 

g mar—130m UZHmmm—UOKL

HH

 

 

mZ<U EDA<FZ<P ho ZOE—2055‘?—

 

 

HHH 52oz 025369:

 

 

 

 

 

128

discs were machined to have 8 notches around the periphery and used as
the top spacers. For the bottom spacers, some of discs are chamfered
around the edges. Some of these fabricated parts of a HIP container are
shown in Figure 39(a). The reasons for using these spacers are that the
spacers promote a uniform deformation of the HIP container and reduce
the possibility of early leakage of the container during hot isostatic
processing.

The green compacts of C-TaC composites were made as follows.
First, a small amount of TaC powder was added in a cylindrical die and
hand-pressed with a plunger after tapping the powder. The die and
plunger were made from heat treated tool steel. The dimensions of the
cylindrical body of the die were about 90 mm in length, 19.2 mm in inner
diameter and 55 mm in outside diameter. Above the TaC powder compact,
12 K tows of carbon fibers were spread and laid manually. A small amount
of TaC powder was added to the fibers and pressed manually after
tapping. By repeating the above procedure, several layers of carbon
fibers were embedded in TaC powder. Then, this sample in the die was
cold-pressed by using a MTS machine (Model 810) equipped with a
vacuum furnace. The maximum pressure applied during the compaction
was 325 MPa. After the cold compaction was completed, the green
compact was carefully removed from the die. The typical dimensions of
the green compacts are 19.2 mm in diameter and 42 mm in length.

A green compact was then carefully put in a bottom-welded Ta HIP
container with a spacer. A molybdenum foil of 0.07 mm in thickness was
inserted as a diffusion barrier between the green compact and the Ta

container. After three top spacers were dropped in, a top lid was press

 

 

 

.Cofifiaoo @322 a 3:»: .HoESHHoQ commoooa buomoa

a go: 303355 “.033on an A3 Him 283 353:8 voHHH0m1E==om> a Q: .Cooumm Saxon ES

 

m8 6: QB .58 £1 5 558:8 film a «0 2:3 088 A3 HmHoEchoQ ma «H .Ho magnumoHosm .am 3sz

0 Jul. 0

 

129

gal.

 

 

ha

130

fitted and TIG-welded with the HIP container in Ar atmOSphere. The HIP
containers were then leak tested with a helium leak tester. After the HIP
containers successfully passed leak testing, the green composite
compacts in the leak-proof HIP containers were then degassed in a
vacuum furnace (CENTORR, Model M60) at a temperature of 1000 °C for
12 hours. The final vacuum was about 3x10 ’6 Torr. After the containers
had cooled down, Ar gas was flushed into the vacuum furnace to fill the
containers. The containers were then quickly stored in a vacuum
desiccator before the vacuum sealing process. The vacuum sealing of the
HIP containers was done in a zone refiner (Materials Research
Corporation, Model EBZ-94) at a vacuum of about 2x10 '6 Torr. The final
vacuum sealed HIP containers before hot isostatic processing are shown
in Figure 39(b).

The hot isostatic pressing was performed at Howmet Turbine
Component Corporation of Whitehall, Michigan. The time, temperature
and pressure profile of the HIP process for these containers is presented
as HIP #1 in Figure 40. The specimens were heated up to about 700 °C in
Ar atmosphere and at that temperature the Ar gas pressure started to
increase. The maximum pressure of over 200 MPa was reached after 4
hours, but the pressure was lowered to about 105 MPa due to the unstable
HIP furnace. The temperature and the pressure of a steady state HIP
process were about 1900 °C and 105 MPa respectively.The steady state
was kept for 3 hours and after that the temperature and the pressure were
decreased linearly. The total time for a HIP process was about 12 hours.

The Ta containers after the HIP process are shown in Figure 39(c).

 

Although the HIP containers had been carefully prepared, one of three

 

131

2500-

2000- ,1

l500~

1000—

500-

TEMP.(0c) / PRES.(BARS)

 

 

6 a 1'0 12 14

O
n...
.p...

TIME (I-IRS.)
HIP #1: TIME. TEMPERATURE, PRESSURE PROFILE

25001

 

 

-- PRESSURE . I """ K
. A — TEMPERATURE '1 \

a . I
< 2000.. I \‘
CD I
‘2’ I ‘
Q: 1500—I I ‘
Q4 I \
\ I ‘\

a I
8 1000 I \\
a x \
OH I l
2 500~ I I
[23 l l
E4 I l

I
0 l “u-‘-I I I I \
0 2 4 6 8 10 12 14
TIME (HRS.)

HIP #2 :TIME. TEMPERATURE. PRESSURE PROFILE

 

 

2500—
' ' PRESSURE

A — WWW

m \

5 zooo~ ' ,I x

m l

‘1’ l \

m \

m 1500- I \

33 . t

I \

\ I \\

A —- I

U 1000 I ,\
(if I’ \
n. , “
E 500‘ l

ca , l
6.4 r

I
.. _’ I
0 " - ‘1 T I Fr ——I I r A ‘7
0 2 4 6 8 10 12 14 16

TIME (HRS)
HIP #3 :TIME. TEMPERATURE. PRESSURE PROFILE

Figure 40. Time, temperature, and pressure profiles of HIP process.

 

132

HIP containers was not properly consolidated probably due to a leakage.
The hot isostatically pressed containers were cut by using a low speed
diamond saw (BUEHLER, ISOMETTM). The microstructures of C-TaC
composite specimens were studied by using an optical microscope
(LECO, NEOPHOT 21) and a scanning electron microscope. The
chemical analysis of the interface between carbon fibers and tantalum
carbide was performed by using an auger electron spectroscope(AES)
(PERKIN-ELMER, Model PH1660).

For the processing route #2 in Figure 38, carbon fibers were
incorporated in a TaC matrix by the slurry-infiltration process. The
carbon fibers and TaC powder used were AS-4 fibers and type B powder.
The slurry was composed of about 87 wt% of TaC powder, 8 wt% of
ethanol and 5 wt% of polyethylenimine. A schematic of the slurry-
infiltration process is shown in Figure 41. A carbon fiber spool was held
by a motor driven pulley. A 12K tow of AS-4 fibers was stretched from
the spool and passed through a number of Teflon rollers. While the fibers
were passing through the rollers, they were spread uniformly by the
vibration of air from a loud speaker which was connected to an amplifier
and a function generator. The volume of the amplifier was adjusted to
obtain the optimum amount of spreading. The spreaded carbon fibers
were then passed through a matrix slurry tank and coated with the slurry.
The infiltrated carbon fibers were wound manually to a spool which was
wrapped with a Teflon film to prevent it from sticking. The pre-preg tapes
were dried in air for 14 hours. The typical thickness of the pre-preg tapes
was 0.5-0.7 mm. These composite tapes were cut to 55 mm length.

The green compacts of C-TaC composites were made by cold

 

 

 

 

 

133

400mm LDIQZZ}

\o/
/

 

 

 

 

.mmoooa cotubai .33: mo otqucom Av 033m

¥z<._. >m~=._._m

6.... \ n=3...

.Somm awn—n.

 

 

134

pressing in a low carbon steel die. About fifty pre-preg tapes with a thin
layer of TaC powder between each tape were stacked inside the die. The
cold pressing was performed by using a hydraulic press (TINIUS OLSEN,
Willow grove, Pennsylvania) with the maximum applied pressure of about
10.6 MPa. The dimensions of the green compacts were 70 mm x 30 mm x
30 mm. The green compacts with Mo foils and Nb spacers were then
carefully placed in Nb HIP containers which were purchased from
Advanced Fusion of Tracy, California. The photographs of a Nb HIP
container before the HIP process are shown in Figure 42(a) and (b). The
thickness of Nb sheets from which the containers and spacers were made
was about 0.36 mm. After the t0p lids of HIP containers had been TIG-
welded in Ar atmosphere, the Nb containers were heated at 500 °C for 4
hours to remove the binder and then degassed at 1000 °C for 12 hours in
a vacuum furnace. The degassed containers were filled with Ar gas and
the top stems of the containers were covered by rubber bulbs. The vacuum
sealing of the HIP containers was Carried out with an electron beam
welding machine at Howmet Corp.

The time, temperature and pressure profile of a HIP processing for
the above specimens, HIP #2, is shown in Figure 40. The Ar gas pressure
was applied after the temperature reached 1500 °C and the temperature
and pressure of the steady state processing during 3 hours were about
1950 °C and 242 MPa. The Nb container after the HIP process is shown
in Figure 42(c) and ((1). After HIPing the containers were cut by using a
SiC cutting wheel and the composite samples were optically inSpected.
After inspection, the samples were found to have low density. So, these

composite samples were again vacuum-encapsulated in Nb containers for

 

 

 

 

 

 

 

135

 

   

   

Figure 42. Photographs of niobium HIP containers: (a), (b) a side and
front view of a Nb container before HIP, (c), (d) a side and front view of

a Nb container after HIP respectively.

 

 

136

the second HIP process similar to the previous procedure. But, this time,
the outside of the containers were covered by Mo foils to prevent the
carburization from the HIP furnace, and no Nb spacers were inserted in
the containers. The profile of a HIP process for these specimens is
presented as HIP #3 in Figure 40. The temperature and the pressure at
steady state were 1960 °C and 195 MPa respectively. Steady state
conditions were kept for 4 hours. The composite specimens obtained after
the HIP process were examined by using SEM and AES.

Some individual filaments of AS-4 fibers were coated with
tantalum carbide by using a dc. magnetron sputter coating machine of the
Physics department at Michigan State University. The carbon to tantalum
atomic ratio of a target, which was purchased from Cerac Inc., was 0.99.
The impurity contents were 0.1-1% Nb, 0.01% Cu, 0.01% Ti, 0.001% A1,
0.001% Fe, and less than 0.001% of Ca, Mg, Mn and Si.The size of the
target was about 57.5 mm in diameter and 6.35 mm thickness. The target
current and voltage applied were 0.8 amperes and negative 400 volts,
respectively. The plasma voltage and current were 60 volts d.c. and 4
amperes, respectively. The emitter current was 19 amperes and the
deposition rate was about 3.3-3.6 A per second. The Ar gas pressure
during the coating was about 2.5 x 10'3 Torr. The carbon fibers were held
in a Special specimen holder machined from Al plates and were located
over the TaC target. Only one side of the fibers facing the target could be
coated each time. So, when the thickness of the coating had reached about
3 microns, the sample holder was flipped over and the other side was
coated to the same thickness.

The coated fibers were analyzed by SEM and ABS. In order to study

 

 

 

137

the effect of coating on the damage of fibers, the coated carbon fibers
were embedded in a TaC matrix slurry. The composition of the slurry was
the same as the one used previously. The dried composites were degassed
at 500 °C for 2 hours and sintered at 1980 °C for 12 hours in a vacuum

furnace. The sintered C-TaC composites were also examined by using

SEM and AES.

 

 

138

5.3 EXPERIMENTAL RESULTS

An optical micrograph of a C-TaC composite produced via
processing route #1 is shown in Figure 43(a). It can be noticed that an
interface was formed between the carbon fibers and the tantalum carbide
matrix. Figure 43(b) is a scanning electron micrograph of a fractured
composite specimen showing the interface region, and Figure 43(c)
shows the damaged carbon fibers at the interface region in the same
specimen. From these SEM micrographs it appears that the interface is
composed of damaged carbon fibers.

In order to find out the exact nature of the interface, chemical
analysis of the interface was carried out. The results of Auger electron
spectroscopy are shown in Figure 44 and Figure 45. In Figure 44, the
results of elemental spot analysis of the matrix, fiber and interface region
of the fractured C-TaC composite indicates that the interface is only
composed of carbon. Traces of oxygen were found in TaC matrix. The
tantalum elemental line scans superimposed with SEM micrographs
across the interface in Figure 45 also reveal that tantalum is not present
at the interface which is consistent with the previous result.

The damage of carbon fibers in the composite is most likely
occurred during both cold compaction and hot isostatic pressing process.
In order to investigate the damage of carbon fibers during cold
compaction process, carbon fibers were embedded in TaC powders and

cold pressed with pressure of about 325 MPa. The SEM micrographs

 

 

 

 

139

.325

E530 @0382. no £35328 Saw a CV £3on $5.”on ES Cvoofiuoui Aux/Giza”: 05 wfikofi
mamauweomfi 2mm 93 Gouge A5 A5 "0:89:00 UmHIU mo mamfiweoma 2mm and 3330 .mv 053m
0. E33 _ I. WI ,
. j .a

\

  
 

 

 

 

 

 

"VI \ .

 

Oxygen

lantalun

 

 

:1
3' Carbon Matrix Region :

1 I l l l l l I 1 I I 1 1 1 l l I l 1‘1
19.9 299.9 499.9 997.9 999.9 1995.9 1294.9 1499.9 1992.9 19
. 111119119 ENERGY. 98 91.9 2999.9

 

 

1011L_LI_11111111_111111

.21

[Wow‘nm'mmmmvmm “M'Wwwhfi

Carbon
271 Fiber Region

 

[IIIIIIIIrIUIIIITII

 

LlllllllLllllllllll

 

 

l l “r I r l l 7’ l——I f lfi T l l l l 7
18.8 288.8 488.8 687.8 886.8 1885.84 1284.8 1483.8 1682.8 1881.8 2888.8
KINETIC EHERGY, eU

104141—11I111HLQL11L1
uv'

I'WWW-18441114494919wwwlvmwvlw6.918148“I'II'I'IWIN4181189448111441w.

0 Carbon lnierphase Region ;

r l fi—I l 1 F l 7 f r l l l l I l 1 Vi]
18.8 288.8 488.8 687.8 886.8 1885.8 1284.8 1483.8 1682.8 1881.8 2888.8
KINETIC EHERGYI eU

 

 

d1
(.11
llllJllLLLllllLllL

 

Figure 44. Auger electron spectroscopic survey of C-TaC composite.

 

141

 

Figure 45. Tantalum line scans across the interface of C-TaC composite:

(a) perpendicular to the interface, (b) parallel to the interface.

 

1111

 

 

142

of this specimen in Figure 46 show that the damage of carbon fibers
occurred by indentation of the hard TaC powders during the cold
compaction process.

Since the mechanical properties of fiber reinforced ceramic matrix
composites are largely degraded by damaged fibers, it is important to use
processing techniques which will result in minimal fiber damage. The
slurry infiltration process, which was applied in the processing route #2,
was found to be successful in avoiding the damage of fibers during
incorporation of fibers into the matrix. Another purpose of using the
slurry infiltration process was to achieve uniform distribution and
alignment of carbon fibers in TaC matrix. The SEM micrographs of TaC
slurry infiltrated carbon fibers in Figure 47 indicate that no damage of
carbon fibers occurred during the slurry infiltration process.

The scanning electron micrographs of HIP processed carbon fiber
reinforced tantalum carbide matrix composite specimens are presented in
Figure 48. The image in Figure 48(a) was taken in the direction
perpendicular to the carbon fibers. It can be noticed that several
elongated voids exist between two pre-preg tapes and some pores also
exist around the fibers. This indicates that densification of the green
compact of the composites was incomplete. The micrograph taken
parallel to the fibers in Figure 48(b) shows that the alignment of carbon
fibers is relatively uniform. At higher magnification Figure 49 it is
obvious that the final density of the TaC matrix is low and that the carbon
fibers are also damaged during hot isostatic pressing.

In order to clarify the reasons for the fiber damage during the HIP

process, the pre-preg tapes were degassed at 500 °C for 3 hours and

 

143

 

Figure 46. SEM micrographs of damaged C-fibers during cold

compaction.

 

111.

 

 

 

Figure 47. SEM micrographs of slurry infiltrated carbon fibers: (a)

perpendicular to the fibers, (b), (c) parallel to the fibers.

 

145

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4-—. 4 > ‘4 fi..-__‘_ “-— — _-
V W- ”m a‘ m

fitgw —-..___.._"T'—-I__.. _ ”m—

“ w m- .. _ __

w- W ....-.

.. --_.._ __ W

_.~r m‘ M... “

-q-u— W /‘ _. w. W ‘.

. , flaunt“ - . m
....__... ...._.. 1.0.0 Pm -- ~ -.

 

Figure 48. SEM micrographs of C-TaC composites produced by the
processing route #2: (a) transverse direction to the fibers, (b)

longitudinal direction to the fibers.

 

 

 

:Z

146

 

-n 4 l} ~35 . - . -
' ., ~ , . I (L_-_ v.

‘1“ ,..\ ">4" ‘1'; ."‘v ',‘T '1‘ 1': "I; ‘8‘":‘13

\ 8‘ \i' " ‘ u. ‘3 fl" .’J ’R F '

' .‘~ - 6' ‘1. \

i ' w 5'

.. I <
m ‘ I; .
. l ‘. I _
r l :r’ ¢. k-‘ ’ . ‘._ “‘_'
I
I

‘-‘ . .
3’ VA‘J
\ l

   

A' $69.4. _‘ if

   

. .._‘J ‘5’ I ' ‘ , ‘_ 1

  

Figure 49. High magnification SEM micrographs of the specimen in

Figure 12: (a) transverse direction to the fibers, (b) longitudinal direction

to the fibers.

%

 

 

 

 

 

147

sintered at 1980 °C for 12 hours in a vacuum furnace. The sintered tapes
were fractured manually and observed in a SEM. The micrographs of the
sintered specimen are shown in Figure 50. Figure 50(a) shows that only
a low density of matrix is obtained by sintering. As can be seen in Figure
50(b), the surface of a carbon fiber is still damaged by sintering. Since
no external pressUre was applied to the sintered specimens, this
indentation damage of the fiber can only have occurred due to the high
internal pressure developed during the high temperature sintering
process. 8

From the above result, it is clear that the damage of carbon fibers
during high temperature consolidation process can not be avoided unless
a protection of carbon fibers is achieved. In order to protect the carbon
fibers from the damage during processing, the fibers were sputter-coated
with tantalum carbide. A tantalum carbide coated carbon fiber is shown
in Figure 51. The micrographs evidence that the thickness of coating is
not uniform. The result of an AES survey of a coated carbon fiber is
presented in Figure 52. The Ta and C line scans in Figure 52(a) and (b)
confirm that a tantalum carbide coating had formed on the carbon fiber.
The exact quantitative analysis of stoichiometry of the coating was not
possible. by AES. However, previous experimental studies on the
sputtering of tantalum carbide indicate preferential sputtering of carbon
occurs [131]. Thus, it is very likely that the composition of tantalum
carbide coating formed by sputter process is nonstoichiometric, with

carbon being preferentially re-sputters from the fibers.

;— m-

 

 

 

 

 

 

 

Figure 50. The SEM micrographs of the sintered and fractured pre-preg

composite tape.

 

1111

 

 

149

 

Figure 51. The SEM micrographs of a coated carbon fiber: (a) transverse

direction to the fiber, (b) longitudinal direction to the fiber.

 

 

150

 

Figure 52. Ta and C line scans of the coated carbon fiber: (a) Ta and C

line scan, (b) Ta line scan.

151

In order to see whether the coating can protect carbon fibers from
the damage during high temperature processing, tantalum carbide coated
carbon fibers were embedded in the tantalum carbide slurry and sintered
as described before. The scanning electron micrographs of this specimen
and a Ta line scan are shown in Figure 53. In Figure 53(a), it can be seen
that the coating successfully protected the carbon fiber from mechanical
damage. The interface of the coating and the fiber appears to have become
diffuse. This might be due to the fact that the interface of the coating and
the fiber became unclear by surface-polishing of the specimen. The other
possible explanation is that the interfacial diffusion may have occurred
between the nonstoichiometric tantalum carbide coating and the carbon
fiber. The Ta line scan superimposed with a SEM micrograph in Figure
53(b) shows that the tantalum carbide coating between the fiber and the

matrix exists in the specimen after high temperature sintering process.

 

152

.3

.5». .

a 8

”warn .9. k4

      

 

 

Figure 53. The SEM micrographs and a Ta line scan of sintered C-TaC

composite after coating the fiber.

 

153

 

5.4 DISCUSSION

The carbon fiber reinforced tantalum carbide composites produced
in the present study were not dense enough and free of defects to
determine the mechanical properties of this composite system. However,
some important information, which might be useful to successfully
produce similar types of composite materials, could be found from the
results of the current study. Two major problems were encountered in
processing C-TaC composites. Firstly, because of its high refractoriness,
the TaC matrix was very difficult to consolidate to full density. Secondly,
during cold compaction and hot processing invariably mechanical
damage of carbon fibers was introduced by indentation of hard tantalum
carbide powders.

Hot isostatic pressing, which was employed in the present study, is
capable of producing C-TaC composites with very high density if all the
processing variables are correctly determined. The important variables of
the HIP processing include pressure, temperature, initial density of green
compacts, material for the container and geometry of the container. The
main reason of low density of C-TaC composites obtained in the present
investigation is considered due to the brittleness of niobium container at
high temperature in the carbonaceous environment of a HIP furnace. The
brittle carburized Nb container will be cracked by high gas pressure
during the HIP process and the consolidation of the green compact will

cease as soon as the container leaks.

 

 

 

 

 

154

The M0 shield used in the third HIP process appears to reduce the
caburization of Nb. containers, but it was not sufficient to prevent
cracking of the containers. The first HIP processed C-TaC composite
specimens using tantalum containers were found to be very dense and the
containers did not show any cracks after the HIP process. Therefore,
tantalum can be considered as the best material for the HIP container
under the employed processing conditions. Although the hot pressing
technique was not applied for thepresent processing, this technique may
produce C-TaC composites with high density. The advantage of the hot
pressing technique is that it requires less complicated preparation steps,
for example the encapsulation of the specimen is not necessary.

The results of SEM and ABS studies of HIP processed C-TaC
composites indicate that no interfacial chemical reaction occurs between
carbon fibers and stoichiometric tantalum carbide. This result is
consistent with the initial theoretical prediction of compositional
stability between carbon and stoichiometric tantalum carbide. The
mechanical damage of carbon fibers was observed under various
processing conditions. The mechanical damage of carbon fibers is most
likely due to indentation of tantalum carbide powders on the surface of
the carbon fibers. At room temperature the Mohs’ hardness of tantalum
carbide is about 9. On the other hand, the Mohs’ hardness of graphite is
between 0.5 and 1. Because of this large difference in the hardness
values, the indentation damage of carbon fibers will occur during cold
compaction and hot isostatic pressing. Further, it was also observed that
the indentation damage occurred even during sintering process. Thus it

can be concluded that the mechanical damage of carbon fibers during

 

 

 

 

 

 

 

 

155

powder processing of C-TaC composites can not be avoided unless a
protection of carbon fibers is achieved.

The protection of carbon fibers can be achieved by the coating.The
materials for coating should have high hardness and high melting point,
and maintain the compositional and thermal stability with carbon fibers
and tantalum carbide matrix at elevated temperatures. The other
advantages of the coating might be that the coating can tailor the
interfacial properties of any particular composite system and help to
obtain uniform dispersion of reinforcing fibers. In the present study, it
was found that the sputter coating of tantalum carbide on carbon fibers
can prevent the mechanical damage of the carbon fiber. Other coating
techniques, for example chemical vapor deposition techniques, may be
used to produce a protective coatings with uniform thickness and

stoichiometric composition.

 

 

 

 

156

5.5 SUMMARY

The results of the present study on the fabrication of carbon fiber

reinforced tantalum carbide composites can be summarized as follows.

1. The carbon fiber reinforced tantalum carbide matrix composites can be
successfully fabricated if some of the following considerations are
taken into account.

2. No chemical reaction occurs at the interface between carbon fibers and
a stoichiometric tantalum carbide matrix.

3. The damage of carbon fibers occurs by indentation of tantalum carbide
powder during cold compaction and hot isostatic pressing.

4. The sintering process alone can cause indentation damage on
carbon fibers probably due to the high internal pressure developed
during the high temperature sintering process.

5. A tantalum carbide coating on carbon fibers prevents the indentation
damage of carbon fibers.

6. A uniform protective coating of carbon fibers is necessary to fabricate
C-TaC composites. The materials for coating should have high
hardness and high melting point, and provide compatibility with

carbon fibers and tantalum carbide matrix.

 

 

CHAPTER 6
SUMMARY AND CONCLUSIONS

The results of the current study on the deformation behavior and
microstructure of TaCo.99, and the processing of C-TaC composites are

summarized as follows.

1. Extensive local plastic deformation was observed in TaC0.99 deformed

by microindentation at room temperature.

2. The dislocation structure of deformed TaC0.99 consists almost entirely
of long screw dislocations. The wavy character of screw dislocations

suggests the existence of a high Peierls stress for screw dislocation

motion at room temperature.

3. The lack of edge dislocations suggests that plastic deformation of
Taco.” at room temperature is mainly accomplished by the motion of

edge dislocations, leaving long screw dislocation segments behind.
4. Screw dislocations were found to form nodes. Screw dipoles were

observed to pinch off, forming dislocation loops during low

temperature plastic deformation.

157

 

 

158

. Strongly recovered microstructures and evidence of recrystallization-
were observed in TaC0.99 specimens annealed after microindentation.
These phenomena strongly resemble those found in metals with high

stacking fault energy.

. From the results on creep tests at intermediate temperatures (1400-
1500 °C, ~ 0.4 Tm) and stresses (105-170 MPa), the values of the
stress exponent (n) and activation energy (QC) for power-law creep of

TaCoogg were found to be nzl.8 and Qc==150 KJ/mol.'

. The low activation energy and the low stress exponent suggest that
the creep of TaC0.99 at intermediate temperatures is mainly controlled

by grain boundary sliding. This is further supported by observation of

void formation.

. The observation of dislocation interaction in the crystal interior
suggests that dislocation glide may act as a minor contributing
mechanism during creep. TEM investigation also reveals the

formation of subboundaries and dislocation networks.

. Most dislocations formed during creep of TaCo.99 have no preferential
orientation, in contrast to dislocations formed at low temperatures.

This is probably due to thermally activated dislocation motion at

elevated temperatures.

 

 

159

10. The carbon fiber reinforced tantalum carbide matrix composites can be

successfully fabricated if some of the following considerations are

taken into account:

 

a) No chemical reaction occurs at the interface between carbon fibers

and a stoichiometric tantalum carbide matrix.

b) The damage of carbon fibers occurs by indentation of tantalum

carbide powder during cold compaction and hot isostatic pressing.

c) The sintering process alone can cause indentation damage on

carbon fibers probably due to the high internal pressure developed

during the high temperature sintering process.

d) A tantalum carbide coating on carbon fibers prevents the

indentation damage of carbon fibers.

e) A uniform protective coating of carbonfibers is necessary to

fabricate C-TaC composites. The materials for coating should have
high hardness and high melting point, and provide compatibility

with carbon fibers and tantalum carbide matrix.

 

 

 

 

 

LIST OF REFERENCES

 

 

 

 

LIST OF REFERENCES

1. L. E. Toth, Transition Metal Carbides and Nitrides, Academic Press,
New York, (1971).

2. B. K. Storm, The RefractorLQarbides. Academic Press, New York,
(1967).

3. H. H. Hausner and M. G. Bowman, Fundamentgs of Refnagory
Compounds, Plenum Press, New York, (1968).

4. W. S. Williams, “Transition-metal carbides” in Prggjess in solid state
chemistry, Vol. 6, ed. by H. Reiss and J. O. McCaldin, Pergamon
Press, New York, (1971).

5. G. E. Hollox, “ Microstructure and mechanical behavior of carbides”,
Mat. Sci. Eng., _3_, 121 (1968 / 1969).

6. A. L.Bowman, T. C. Wallace, J.L. Yarnell, R. G. Wenzel and E. K.
Storms, “The crystal structures of V2C and Ta2C”, Acta Cryst. _1_9_,

6 (1965).

7. R. Lesser and G. Braucr, “Carbide phases of tantalum (The system
tantalum-tantalum carbide)”, Z. Metallkde., 4__9_, [12], 622 (1958).

8. I. Zaplatynsky, “Observations on zeta phase in the system Ta-C”,
J. Amer. Cer. Soc., 4_9_, [2], 109 (1966).

9. K. Yvon and E. Parthe, “On the crystal chemistry of the closed packed
transition metal carbides, I. The structureof the C-V, Nb and Ta

carbides”, Acta Cryst. B26, 149 (1970).

10. D. J. Rowcliffe and G. Thomas, “Structure of non—stoichiometric
TaC”, Mat. Sci. Eng., _1_8_, 231 (1975).

160

 

 

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

4 161

B. I. Markhasev, Yu. M. Shamatov and S. E. Svintsova, “Ordering in
nonstoichiometric carbides of niobium and tantalum”, Izv. Akad.
Nauk. SSSR, Neorg. Mater., 21, 1697 (1985).

Ch. H. De Novion and V. Maurice, “Order and disorder in carbides
and nitrides”, J. Physique Colloq., fl, C7, 211 (1977).

J. D. Venables and M. H. Meyerhoff, “Ordering effects in NbC and
TaC”, Sol. Stat. Chem., Proc. of 5th Mater. Res. Symp., NBS Spec.

Pub]. 364, 583 (1972).

A. L. Bowman, “The variation of lattice parameter with carbon
content of tantalum carbide”, J. Phys. Chem. _6__5_, 1596 (1961).

G. Santoro, “Variation of some properties of tantalum carbide with
carbon content”, Trans. AIME, 227, 1361 (1963).

J. R. Cooper and R. L. Hansler, “Variation of electrical resistivity of
cubic tantalum carbide with composition”, J. Chem. Phys., _3__9_, [1],

248 (1963).

R. Steinitz and R. Resnick, “Electrical and magnetic properties of
compositions in the tantalum-carbon system”, J. Appl. Phys., _3_7_,

3463 (1966).

H. L. Brown, P. E. Armstrong and C. P. Kempter, “Elastic properties
of some polycrystalline transition-metal monocarbides”, J. Chem.

Phys., 4_5__, [2], 547 (1966).

J. H. Westbrook and E. R. Stover, “Carbides for high temperature
applications” in High temperature technology, ed. by I. E. Campbell
and Sherwood, Wiley, New York (1967).

G. V. Samsonov, M. S. Kovalchenko, V. V. Dzmelinskii and G. S.
Upadayaya, “Temperature dependence of hardness of titanium carbide
in the homogeneity range”, Phys. Stat. Sol. (a), _1_, 327 (1970).

J. H. Westbrook, “Note on the temperature dependence of hardness of
the transition metal monocarbides”, Acta Met., 3, 104 (1955).

W. S. Williams, “Influence of temperature, strain rate, surface
condition, and composition on the plasticity of transition-
metal carbide crystals”, J. Appl. Phys., _3_5_, 1329 (1964).

 

162

23. R. Steinitz, “Mechanical properties of refractory carbides at high

temperatures”, in Proc. of Conf. on Nuclear Application of
Nonfissionable Ceramics American Nuclear Society (1966).

 

24. D. J. Rowcliffe and W. J. Warren, “Structure and properties of
tantalum carbide crystals”, J. Mat. Sci., 5_, 345 (1970).

25. D. J. Rowcliffe and G. H. Hollox, “Hardness anisotropy, deformation
mechanisms and brittle—to-ductile transition in carbides”, J.
Mat. Sci., 6, 1270 (1971).

26. C. A. Brookes, J. B. O’Niel and B. A. W. Redfern, “Anisotropy in the
hardness of single crystals”, Proc. Roy. Soc. Lond. A, 322, 73
(1971).

 

27. R. H. J. Hannink, D. L. Kohlstedt and M. J. Murray, “Slip system
determination in cubic carbides by hardness anisotropy”, Proc. Roy.
Soc. Lond. A, 326, 409 (1972).

28. D. J. Rowcliffe and G. E. Hollox, “Plastic flow and fracture of
tantalum carbide and hafnium carbide at low temperatures”, J. Mat.
Sci.,_§_, 1261 (1971).

29. G. Morgan and M. H. Lewis, “Hardness anisotropy in niobium
carbide”, J. Mat. Sci., 2, 349 (1974).

30. D. J. Rowcliffe, “Plastic deformation of transition metal carbides” in
Deformgion of ceramic materials II ed. by R. E. Tressler and R. C.
Bradt, Materials Science Research, Vol. 18, 49 (1983).

 

31. A. Kelly and D. J. Rowcliffe, “Slip in titanium carbide”, Phys. Stat.
Sol., 151, K29 (1966).

32. H. Kurishita, K. Nakajima and H. Yoshinaga, “The high temperature
deformation mechanism in titanium carbide single crystals”, Mater.
Sci. Eng., :53, 177 (1982).

33. R. Darolia and T. F. Archbold, “Plastic deformation of polycrystalline
zirconium carbide”, J. Mat. Sci., l_1_, 283 (1976).

34. D. W. Lee and J. S. Haggerty, “Plasticity and creep in single crystals
of zirconium carbide”, J. Amer. Cer. Soc., §_2_, [12], 641 (1969).

35. P. F. Becher, “Mechanical behavior of polycrystalline TaC”, J. Mat.
Sci., .6, 79 (1971).

 

 

 

 

 

 

 

 

 

 

163

36. J. L. Martin, P. Lacour-Gayet and P. Costa, “Plastic deformation of
tantalum carbide up to 2200 °C” in Electron microscopy and
structure of materials” ed. by G. Thomas, R. M. Fulrath and R. M.
Fisher, University of California Press, Berkeley (1972).

37. G. Das, K. S. Mazdiyasmi and H. A. Lipsitt, “Mechanical properties
of polycrystalline TiC”, J. Amer. Cer. Soc., 6_5, 104 (1982).

38. W. S. Williams and R. D. Schaal, “Elastic deformation, plastic flow
and dislocations in single crystals of titanium carbide”, J. Appl.
Phys., 3_3_, 955 (1962).

39. G. E. Hollox and R. E. Smallman, “Plastic behavior of titanium
carbide”, J. Appl. Phys., 11,, 818 (1966).

 

40. D. K. Chatterjee, M. G. Mendiratta and H. A. Lipsitt, “Deformation
behavior of single crystals of titanium carbides”, J. Mater. Sci., 14,
2151 (1970).

41. A. Kelly and D. J. Rowcliffe, “Deformation of polycrystalline
transition metal carbides”, J. Amer. Cer. Soc., 10, 253 (1967).

42. W. S. Williams, “Dispersion hardening of titanium carbide by boron
doping”, Trans. AIME, 236, 211 (1966).

43. J. D. Venables, “The nature of precipitates in boron-doped TiC”, Phil.
Mag., _1_6, 873 (1967).

44. J. D. Venables, D. Kahn and R. G. Lye, “Structure of the ordered
compound V6C5”, Phil. Mag., _1_8_, 177 (1968)

45. H. A. Johanscn and J. G. Clearly, “The ductile-brittle transition in
tantalum carbide”, J. Electr. Chem. Soc., .1_1_3_, [4], 378 (1968).

46. G. E. Hollox, “Plastic deformation of carbides”, NBS Spec. Publ.
303, 201 (1968).

47. J. L. Martin, “Evidence of dislocation dissociation in nearly
stoichiometric tantalum carbide using the weak-beam technique”, J.
Micros., 28, 209 (1973).

48. C. Allison, M. Hoffman and W. S. Williams, “Electron energy loss
spectroscopy of carbon in dissociated dislocations in tantalum
carbide”, J. Appl. Phys. .51, 6757 (1982).

 

164

49. J. B. Bird, A. K. Mukherjee and J. E. Dorn, “Correlations between
high-temperature creep behavior and structure” in uantitative

relation between properties and microstructure, ed. by D. G.
Brandon and A. Rosen, Israel University Press, Jerusalem (1969)

 

50. W. D. Nix and B. Illschner, “Mechanisms controlling creep of single
phase metals and alloys”, in Strength of metals andfllovs vol. 3.
ed. by P. Haasen, V. Gerold and G. Kostorz, Pergamon Press Inc.,
New York (1979).

51. T. H. Hazlett and R. D. Hansen, “Influence of substructure on the
shape of the creep curve”, Trans. AIME, 41, 508 (1955).

52. J. Weertman and J. R. Weertman, “Mechanical properties, strongly

temperature-dependent” in Physical metallurgy part2, ed. by R. W.
Cahn and P. Hassen, North-Holland Physics Publishing, Amsterdam
(1983).

53. O. D. Sherby, J. L. Lytton and J. E. Dorn, “Activation energy for
creep of high-purity aluminium”, Acta Met., 5, 217 (1957).

54. A. K. Mukherjee, J. E. Bird and J. E. Dorn, “Experimental

correlations for high-temperature creep”, Trans. A.S.M., 62, 155
(1969).

55. O. D.Sherby and P. M. Burke, “Mechanical behavior of crystalline

solids at elevated temperatures”, Progress in Mat. Sci. L3, 325
(1967).

56. W. R. Cannon and T. G. Langdon, “Creep of ceramics”, (a) Part1.
Mechanical characteristics, J. Mat. Sci., L8, 1 (1983): (b) Part2. An
examination of flow mechanisms, J. Mat. Sci., _2_3_,1 (1988).

57. J. Bardeen and C. Herring, “Diffusion in alloys and kirkendall

effect”, in Imperfections in nearly perfect crystals”, ed. by W.
Schockley, J. Hollomon, R. Maurer and F. Seitz, John Wiley and Sons
Inc., New York (1952).

 

58. J. Weertman, “Steady-state creep of crystals”, J. Appl. Phys., _2__8_,
1185 (1957).

59. J. P. Poirier, Creep of Crystals, Cambridge University Press,
Cambridge (1985).

60. J. Weertman, “Dislocation climb theory of steady-state creep”, Trans.
A.S.M., _6__l_, 681 (1968).

 

 

 

 

165

61. J. Harper and J. E. Dorn, “Viscous creep of aluminium near its
melting temperature”, Acta Met., 5, 654 (1957).

62. P. Yavari, D. A. Miller and T. G. Langdon, “An investigation of
Harper-Darn creep-I. Mechanical and microstructural
characteristics”, Acta Met., 39,, 871 (1982).

63. P. J. Dixon-Stubbs and B. Wilshire, “Deformation processes during
creep of single and polycrystalline CaO”, Phil. Mag. A, 4;, 519 (1982).

64. T. G. Langdon and P. Yavari, “An investigation of Harper-Darn creep-
II. Flow process”, Acta Met., Q, 881 (1982).

65. F. A. Mohamed and T. J. Ginter, “On the nature and origin of
Harper-Dom creep”, Acta Met., _3_Q, 1869 (1982).

66. H. Luthy, A. K. Miller and O. D. Sherby, “The stress and temperature
dependence of steady-state flow at intermediate temperatures for pure
polycrystalline aluminium”, Acta Met., _2__8_, 169 (1980).

67. F. R. N. Nabarro, “Deformation of crystals by the motion of single
ions”, Report of a conference on strength of solids, The Physical
Soc., 75 (1948).

68. C. Herring, “Diffusional viscosity of a polycrystalline solid”, J.
Appl. Phys., _2_1_, 437 (1950).

69. R. L. Coble, “A model for boundary diffusion controlled creep in
polycrystalline materials”, J. Appl. Phys., 3_4_, 1679 (1963).

70. B. J. Hockey,"‘Use of the hardness test in the study of the plastic
deformation of single crystals” in The science of hardness testing .
and its research applications, ed. by J. H. Westbrook and H. Conrad,
ASM, Ohio (1973).

71. D. Tabor, “The hardness of solids”, Rev. Phys. Tech., _1_, 145 (1970).
72. V. G. Eremenko and V. I. Nikitenko, “Electron microscope
investigation of the microplastic deformation of silicon by

indentation”, Phys. Stat. Sol. (a), IA, 317 (1972).

73. M. J. Hill and D. J. Rowcliffe, “Deformation of silicon at low
temperatures”, J. Mat. Sci., 2, 1569 (1974).

 

 

 

 

 

166

74. B. J. Hockey, “Plastic deformation of aluminum oxide by indentation
and abrasion”, J. Amer. Cer. Soc., fi,[5], 223 (1971).

75. “Standard test method for density of glass by buoyancy”, ASTM
standard designation: C 693-84, Annual book of ASTM standards,
Vol. 15.02, 218 (1989).

76. P. Hirsch, A. Howie, R. B. Nicholson, D. W. Pashley and M. J.
Whealan, “Electron microscopy of thin crystals”, Robert E. Krieger
Publishing Co., Florida (1977).

77. C. H. Hsueh, A. G. Evans and R. L. Coble, “Microstructure
development during final/intermediate state sintering- I. Pore/grain
boundary separation” Acta Met., 30, 1269 (1982).

78. M. V. Heimendahl, Electron microscopy of materials, Academic
Press, New York, p. 137 (1980).

79. R. Hill, The mathematical theory of plasticity, Oxford University
Press, London (1950).

80. L. E. Samuels and T. O. Mulhearn, “An experimental investigation of
the deformed zone associated with indentation hardness
impressions”, J. Mech. Phys. Solids, 5, 125 (1957).

81. D. M. Marsh, “Plastic flow in glass", Proc. Roy. Soc. A, 279, 420
(1964).

82. D. Tabor, The hardness of metals, Oxford University Press, London
(1951).

83. K. L. Johnson, “The correlation ofindentation experiments”, J. Mech.
Phys. Solids, _1_8, 115 (1970).

84. R. Peierls, “The size of a dislocation”, Proc. Phys. Soc. London, A52,
34 (1940).

85. F. R. N. Nabarro, “Dislocations in a simple cubic lattice”, Proc. Phys.
Soc. London, A59, 256 (1947)

86. R. A. Foxall and C. D. Statham, “Dislocation arrangements in
deformed single crystals of niobium~molybdenum alloys and
niobium-9 at% rhenium”, Acta Met., l_8_, 1147 (1970).

167

 

87. F. Louchet, L. P. Kubin and D. Versely, “In situ deformation of b.c.c.
crystals at low temperatures in a high-voltage electron microscope

Dislocation mechanisms and strain-rate equation, Phil. Mag. A, 22,
[4], 433 (1979).

88. H. Saka, K. Noda and T. Imura, “Tensile test of foil specimens of iron
single crystals at room and low temperatures under observation in

high voltage electron microscope”, Crystal Lattice Defects, g, 45
(1973).

89. V. Vitek, “Effect of dislocation core structure on the plastic

properties of metallic materials” in Dislocations and properties of
real materials, ed. by M. H. Loretto, The Institute of Metals, London
(1985),p. 30.

90. E. Y. Gutmanas, E. M. Nadgornyi and A. V. Stepanov, “Dislocation

movement in sodium chloride crystals”, Soviet Physics-Solid State,
2., [4]. 743 (1963). ’

91. J. Friedel, Dislocations, Pergamon Press, Oxford (1964).

 

92. R. W. Cahn, “Recovery and recrystallization” in Physical metallurgy
part2, ed. by R. W. Cahn and P. Hassen, North—Holland Physics
Publishing, Amsterdam (1983).

93. S. Chevacharoenkul, “Kinetics and mechanisms of intermediate and
high-temperature creep in niobium carbide single crystals”, Ph. D.
thesis, North Carolina State University, Raleigh, NC, (1987).

94. F. Keihn and R. Kebler, “High-temperature ductility of large~grained
TiC”, J. Less-Common Metals, _6_, 484 (1964).

95. I. I. Spivak, R. A. Andrievskii, V. N. Rystsov, and V. V. Klimenko,
“Creep of titanium monocarbide in its homogeneity range”,
Poroshkovaya Met., 1_3_2,[7], 69 (1974)

96. J. Chermant, G. Leclerc and B. L. Mordike, “Deformation of titanium
carbide at high temperatures”, 2. Metallkde, 1;, 465 (1980).

97. M. H. Leipold and T. H. Nielson, “Mechanical properties of

hot-pressed zirconium carbide tested to 2600 °C”, J. Amer. Cer. Soc.,
51,, [9], 419 (1964).

98. D. W. Lee and J. S. Haggerty, “Plasticity and creep in single crystals
of zirconium carbide”, J. Amer. Cer. Soc., _5__2_, [12], 641 (1969).

 

 

 

168

 

99. L. N. Dement’yev, P. V. Zubarev, V. N. Kruglov, V. N. Turchin and
YB. D. Kharkhadin, “High-temperature creep of monocrystalline
niobium carbide”, Phys. Met. Metall., fl,[3], 142 (1978).

100. R. D. Nixon, S. Chevacharoenkul and R. F. Davis, “Steady-state

creep behavior of hot isostatically pressed niobium carbide”, Mat.
Res. Bull., 2_2_, 1233 (1987).

101. S. Chevacharoenkul and R. F. Davis, “Dislocation mechanisms,

diffusional processes and creep behavior in Nbe”, Acta Met., 31,
417 (1989).

102. R. Resnick, R. Steinitz and L. Seigle, “Determination of diffusivity
of carbon in tantalum and columbium carbides by layer-growth
measurements”, Trans. AIME, 233, 1915 (1965).

103. R. A. Andriyevskiy, V. V. Klimenko and Yu. F. Khromov, “Carbon
self-diffusion in carbides of IV and V transition metals”, Fiz.
Metal.Metalloved., 2_8_, [2], 293 (1969). '

104. R. D. Gifkins, “A mechanism for the formation of intergranular
cracks when boundary sliding occurs”, Acta Met., _4_, 98 (1956).

 

105. C. W. Chen and E. S. Machlin, “On the mechanism of intercrystalline
cracking”, Acta Met., 4_, 655 (1956).

106. H. C. Chang and N. J. Grant, “Mechanism of intercrystalline
fracture”, Trans. AIME, 206, 544 (1956).

107. E. M. Passmore, R. H. Duff and T. Vasilos, “creep of dense

polycrystalline magnesium oxide”, J. Amer. Cer. Soc., 42, [11],
594 (1966).

108. S. U. Din and P. S. Nicholson, “Creep of hot-pressed silicon
nitride”, J. Mat. Sci., .12, 1375 (1975).

109. L. E. Poteat and C. S. Yust, “Creep of polycrystalline Thorium
dioxide”, J. Amer. Cer. Soc., 42, [8], 410 (1966)

110. T. G. Langdon and R. B. Vastava, “An evaluation of deformation

models for grain boundary sliding” in Mechanical testing for
deformation model development, ed. by R. W. Rohde and J. C.

Swearengen, ASTM STP no. 765, 435 (1982).

111. G. C. Wei and P. F. Becher, “Development of SiC-whisker-reinforced
ceramics”, Amer. Ceram. Soc. Bull., _6_4_, 298 (1983).

 

 

169

112. R. A. J. Sambell, A. Briggs, D. C. Phillips and D. H. Bowen,

“Carbon fibre composites with ceramic and glass matrices”, J. Mat.
Sci., 1, 676 (1972).

 

 

113. J. F. Bacon, K. M. Prewo and E. R. Thompson, “Research on graphite

reinforced glass matrix composites”, NASA Contract Report 158946
(1978).

114. K. M. Prewo and J. J. Brennan, “High-strength silicon carbide
fiber-reinforced glass-matrix composites”, J. Mat. Sci., _1_e, 463
(1980).

115. R. W. Rice, “Mechanisms of toughening in ceramic matrix
composites”, Ceram. Eng. Sci. Proc., 2 [7-8], 661 (1981).

 

116. D. C. Phillips, “Fibre reinforced ceramics” in fiebrieation of

composites, ed. by A. Kelly and S. T. Mileiko, Elservier Science
Publishers, (1983) P 373.

117. J. A. Cornie, Y. M. Chiang, D.R. Uhlmann, A. Mortensen and J. M.
Collins, “Processing of metal and ceramic matrix composites,
Amer. Ceram. Soc. Bull., _6_5_, 293 (1986).

118. R. W. Rice, “Ceramic composites-Processing challenges”, Ceram.
Eng. Sci. Proc., _2_ [7-8], 493 (1981).

119. R. A. J. Sambell, “The technology of ceramic-fibre ceramic-matrix
composites, Composites, September, 276 (1970).

 

120. E. S. Hodge, “Hot isostatic pressing improves powder metallurgy
parts”, Mater. Des. Eng., _6_1_, 92 (1965).

121. P. E. Price and S. P. Kohler, “Hot isostatic pressing”, in Powder

metallurgy, Metals Handbook 9th ed. Vol 7, American Society for
Metals, (1984), P 419.

 

122. T. Fujikawa, “Hot isostatic pressing(HIP) of ceramics and HIP

equipment for ceramics”, in Fine ceramics, ed. by S. Sato,
Elservier Science Publisher, (1988).

123. L. Bjork and L. A. G. Hermannsson, “Hot isostatically pressed

alumina-silicon carbide whisker composites”, J. Am. Ceram. Soc.,
22 [8], 1439 (1989).

   

 

 

111

 

 

 

 

 

 

 

124.

125.

126.

127.

128.

129.

130.

131.

170

G. Pezzotti, I. Tanaka, T. Okamoto, M. Kozumi and Y. Miyamoto,
“Processing and mechanical properties of dense Si3N4-SiC-whisker
composites without sintering aids”, J. Am. Ceram. Soc., 12 [8],
1461 (1989).

A. J. Caputo and W. J. Lackey, “Fabrication of fiber-reinforced
ceramic composites by chemical vapor infiltration”, Ceram. Eng.
Sci. Proc., 2 [7-8], 654 (1984).

B. Capdepuy and P. Lespade, “Study of ceramic-fibre/ ceramic-
sol-gel based matrix composites”, in Technical ceramics ed. by H.
Nosbusch and I. V. Mitchell, Elservier (1986) P 233.

 

J. 0. Gibson, R. L. Schumacher and K. L. Myler, U. S. Patent 3 736
159, May 29 (1973).

J. 0. Gibson, R. L. Schumacher and K. L. Myler, U. S. Patent 3 776
000, October 16 (1973).

J. 0. Gibson and M. G. Gibson, U. 8. Patent 4 196 230, April 1
(1980).

J. B. Donnet and R. C. Bansal, Carbon fibers, Marcel Dekker, Inc.
(1984) P 27.

G. Betz and G. K. Wehner, “Sputtering of multi component

materials”, in Sputtering by particle bombardment 11, ed. by R.
Behrisch, Springer-Verlag (1983).

 

 

 

 

 

UNIV. L BR E

11111111111111.1111 9115

     

5....

 
         

 

 
     

 

   

 

“ -, » '.. . ‘>‘ .‘..‘ .'.. ... 1.- 1 . .4 -