Effect of coating and exposure on the oxidation and mechanical properties of Ti-22A1-26Nb

MISSING IMAGE

Material Information

Title:
Effect of coating and exposure on the oxidation and mechanical properties of Ti-22A1-26Nb
Physical Description:
xviii, 132 leaves : ill. ; 29 cm.
Language:
English
Creator:
Dobbs, James Ross, 1962-
Publication Date:

Subjects

Subjects / Keywords:
Materials Science and Engineering thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Materials Science and Engineering -- UF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1997.
Bibliography:
Includes bibliographical references (leaves 121-131).
Statement of Responsibility:
by James Ross Dobbs.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 028626836
oclc - 39526375
System ID:
AA00014249:00001


This item is only available as the following downloads:


Full Text











EFFECT OF COATING AND EXPOSURE ON THE OXIDATION AND
MECHANICAL PROPERTIES OF Ti-22A1-26Nb















By

JAMES ROSS DOBBS


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1997

































Copyright 1997

by

James Ross Dobbs

































To my loving wife, Jane.















ACKNOWLEDGMENTS


The title of Doctor of Philosophy is bestowed on the individual, but the road to

obtaining that title is truly a team effort. My road was paved with many supportive,

capable individuals whom I would like to recognize. First and foremost I thank my wife,

Jane, for the support and love during a period in our relationship when there was

something other than her that commanded much of my time and attention. I thank my

parents and brothers and sister for their support and enthusiasm and for always being

willing to listen. I would also like to thank my thesis advisors, Prof. Mike Kaufman

whose friendship I value as much as his technical interaction and Dr. Mike Gigliotti who

never let me forget the focus of my goal. I would also like to thank Prof. Teresa Pollock

whose gentle reminders and kind inquiries gave me inspiration to finish.

I would like to thank several people who helped me conduct the experimentation

and understand the results: Mr. Mike Gilhooley for his help and guidance in LCF testing

and Mr, Chris Canestraro for his assistance in tensile testing, Mr. Keith Borst for

producing the sputter coatings, Mr. Amie Henry for teaching me the LPPS system, and

Mr. Kevin Janora for CVD coatings. I would like to thank Dr. Ravi Ravikumar and Mr.

Mike Larson for their guidance and patience while I learned and practiced TEM, Ms.

Cindy Hayden for assistance in conducting and interpreting XRD, Dr. Bob Gilmore for

acoustic microscopy and acoustic modulus measurements and Dr. John Ackerman for

guidance through many experimental decisions and his wealth of CVD knowledge and









his ability to convey that knowledge. A special note of thanks to Dr Mike Henry for

lengthy and insightful discussions on fatigue, and Prof. Harry Lipsett for many hours of

discussion on titanium metallurgy and relevant experimentation.

I would like to thank Dr. Ann Ritter for support, both financial and moral.

Thanks to Dr. Jeff Graves and Dr. Tom Cox for having enough faith in me to employee

me while I completed this work. And a special note of thanks to Ms. Karen Keever for

valuable assistance in preparation of this document.















TABLE OF CONTENTS


page


ACKNOW LEDGM ENTS ................................................................ ...................... iv

L IST O F T A B L E S ............................................................ ........................................... ix

LIST O F FIG U R E S ........................................................................... ...................... x

ABSTRA CT..................................................... .......................... xvii

CHAPTERS

1 INTRODUCTION...................... ....................... .............

2 LITERATURE REVIEW.................... ........ .....................5
Overview of Ti-Al-Nb M etallurgy.................................... ............................5
H istory..................................... ............ ... .... ......................5
Description of Phases Present in Ti-Al-Nb.................... ...... ............. 7
Deformation Behavior of C,, O and B2 Phases ...........................................14
Tensile Behavior of a2 and O Alloys................... ...........................17
Fracture Toughness and Fatigue ........................................ ......... ......22
Overview of Oxidation Behavior .............................. ....................26
Oxidation of Uncoated Titanium Alloys.................................................26
Oxidation of Coated Titanium Alloys...................... ............................28
A T i ...................................................................... ............................. 2 8
M CrA lY ........................... ............................. ...........29
Pure m etals ........................ ............................. 29
Glasses, silicates and oxides............................. .................... 30
Embrittlement of Titanium Base Alloys ................................................31
Effects of Coatings on Mechanical Properties of Ti Alloys...................33
Program A pproach................................................... .........................34

3 EXPERIMENTAL PROCEDURE .......................................................35
Material Processing ................... ........ .....................35
Specim en Blanking........................... ............................ .................... 42
Test Specim ens and Test Procedures ............................... ...........................45
Oxidation............... ..................................................................... 45









S am p les .................................................................. ............................4 5
T testing .................................................. ..................................... 45
T en sile .................................................................... .................................4 9
Sam ples .................................................. .................................... 49
T testing ...................................................................... ................. 49
F atigu e.................................................................... .................................5 0
Sam ples .................................................. .................................... 50
T testing ...................................................................... ................. 53
A analysis ............................................. ........................................ 54
Coatings.............................. ...............................56
Low Pressure Plasma Spray (LPPS) Coatings.............................................56
NiCrAlY + 40 vol% A1203............................ ................................57
FeCrAY............................................. ....................................58
Sputtered Coatings. .............................................. ......................58
Sputtered A l......................................... ......................................... 59
Sputtered Si ................... .. ........ ......................60
Sputtered Pt and Cr...................................................61
C V D coatings.............................................................. ...................... 61
S iO ..................................................... ........................................ ........6 3
T a20 .......... ......................................................................................63
MgO........................................................................................... 64

4 RESULTS AND DISCUSSION..................................... ...........................65
O xidation ................................................ ............................................... 65
Cyclic Oxidation Testing ............................................. .....................65
B baseline results................................. ......................................... 65
Al and Si Coated and Reacted Results ........................................66
Oxide Coating Results.................................................67
Metallic Coating Results ........................ ............. .....................68
Microhardness and Microprobe Evaluation............... .....................70
Baseline uncoated........................... ...........................70
FeCrAlY and NiCrAIY + A1203 ........................ .....................73
Sputtered and Reacted Al and Si.................... ...........................77
Sputtered Pt and Cr coatings ................................... ....................80
CVD Oxide Coatings.......................................................................82
Effect of Forging Strain and Cooling Rate on the Tensile and LCF
B behavior ................................................................................................ 86
Fatigue Results............................ .............. ......................86
Tensile R results ............................................................ ......................87
D iscu ssion ........................................................................... ....................8 8
Effect of Exposure at 6500C on the Tensile and LCF Behavior.......................95
Tensile Results ............. ..............................................................95
Fatigue Results ......................................................................96
Discussion ................ .........................................................98



vii









Effect of Coating and Exposure on the LCF Behavior...................................105
R esults............................................................................. ..............105
Discussion ............................................ 110

5 SUMMARY AND CONCLUSIONS ...................... .......... ......... 117
O xidation ........................................................................ ................. 117
Forging Location .................................................................. .................... 118
Environm ental Exposure .............................................................. .................. 119
Coated LCF Behavior............................. ....................120

LIST OF REFERENCES............................................121

BIOGRAPHICAL SKETCH ........................................................ ... 132









































viii















LIST OF TABLES


Table page

Table 3-1. Compositions from several locations along the length of the 15.25 cm
diameter billet, surface and center, and from the forging used in this study.............36

Table 3-2. Forging parameters for the material used in this study. The first line is step
1 and the second line is step 2....................... ..... ..................36

Table 4-1. Testing parameters and results for samples used to determine effect of
location on L C F ..................................................................................................86

Table 4-2. Tensile test data. .................................................... ................................ 95

Table 4-3. Low cycle fatigue data of environmentally exposed samples........................96

Table 4-4. Coated LCF data......................................................................................106















LIST OF FIGURES


Figure page

Figure 1.1. Specific yield strength vs. temperature comparing an a2 and orthorhombic
Ti-Al-Nb alloy with a conventional Ti alloy and IN718. After [Woo93]...................2

Figure 2.1. Titanium aluminum equilibrium phase. After [Mur87]................ ............

Figure 2.2. Comparison of the specific yield strength of a, and orthorhombic alloys
as a function of temperature. After [Row91]. ....................................................... 7

Figure 2.3. Ball model of basal plane of a showing lattice sites for Ti and Al atoms.
The a phase lattice is similar, with the Ti and Al atoms arranged randomly..............8

Figure 2.4. 900C isotherm from the Ti-Al-Nb ternary phase diagram. B2, a,, and
orthorhombic phases are in equilibrium. After [Row92]....................................... 9

Figure 2.5. Ball model drawing of the B2 phase showing the relationship between the
titanium alum inum and niobium atom s. ............................. ............................... 10

Figure 2.6. Comparison of the basal planes of the a, and orthorhombic phases ..............11

Figure 2.7. Constant 25 at% Al isopleth from the Ti-Al-Nb ternary in Figure 2.4.
A after [B an95]. .......................................... ......................... ........... ...................12

Figure 2.8. SEM Backscatter micrograph ofTi-22A1-26Nb processed in the O + B2
field and directly aged at 8150C for 4 hours. After [Woo93]..................................13

Figure 2.9. SEM Backscatter micrograph of Ti-22A1-26Nb processed in the B2 field
and directly aged at 8150C for 4 hours. After [Woo93]........................................... 14

Figure 2.10. Crystal structure and burgers vectors for the a, and O phases. After
[B an 9 5 ]...................... .... ................................................ ....................... ............ 15

Figure 2.11. Slip bands in the B2 phase in Ti-24A1- 1Nb showing the inhomogeneous
nature of slip. After [Ban90]. ............ .......................................... .................... 17

Figure 2.12. Yield strength vs. cooling rate for two a, alloys. After [Gog90] ...............17

Figure 2.13. Elongation vs. cooling rate for two a2 alloys. After [Gog90]......................18









Figure 2.14. SEM backscatter micrograph of Ti-22A1-26Nb forged in the B2 region
and directly aged at 8040C. Notice the colony of O laths that have formed off of
th e large lath ..................................................... .................................................. 19

Figure 2.15. Intergranular crack observed in tensile tested Ti-22A1-27Nb solutioned
above the P(B2) transus and aged at 8700C for 50 hours.........................................20

Figure 2.16. Fractograph of forged + direct aged material. Notice the high GAR and
the initiation of failure from an internal boundary.........................................21

Figure 2.17. SEM fractograph of material solutioned above the P(B2) transus and aged
at 870'c for 50 hours. Notice that there is mixed intergranular and transgranular
fracture across the sample and the prior P grain size is approximately 500m..........22

Figure 3.1. BSE image of a typical microstructure of the Ti-22A1-26Nb material after
B2 forging and heat treatment at 8150C for 4 hours. .......................... ........... 37

Figure 3.2. Photomacrograph of one half of the forging showing flow lines....................38

Figure 3.3. Forging half showing location of 10x macrophotographs and lines for
sectioning into metallographic samples. .......................... .........................39

Figure 3.4. Calculated iso-strain lines for the forging examined in this study ................39

Figure 3.5. Photomicrograph taken from area C. Notice the equiaxed grain structure
and low grain aspect ratio. ................................. ....... .. ...........................40

Figure 3.6. Photomicrograph taken from area P. Notice the high grain aspect ratio
resulting from the higher effective forging strain.......................................40

Figure 3.7. High resolution SEM image of area S (identified in Figure 3.3) from the
slowest cooled region in the forging...................... ............................41

Figure 3.8. High resolution SEM image of area Q (identified in Figure 3.3) from the
fastest cooled region in the forging................................... .... .................... ... 42

Figure 3.9. Fatigue and oxidation specimen sectioning diagram....................................43

Figure 3.10. Tensile and oxidation sample sectioning diagram. Shaded samples are
oxidation and white samples are tensile/creep...........................................44

Figure 3.11. Details of sectioning diagram for oxidation (shaded) and tensile (open)
samples from Figure 3.10. Calculated effective strain is also depicted....................44

Figure 3.12. Drawing of oxidation sample used in this research....................................45

Figure 3.13. Photograph of furnace used to conduct cyclic oxidation tests....................46









Figure 3.14. Oxidation pins loaded on hearth plate. The samples are the darker pins,
the thermocouples are the two white rods in the foreground. The cordierite block
is resting on the furnace platen. ........................ ..... ............................47

Figure 3.15. Typical thermal cycle recorded by strip chart recorder..............................48

Figure 3.16. Drawing of tensile sample used in this research. Dimensions are in
in ch es. .......................... .............................................................................. ....4 9

Figure 3.17. Typical stress-strain curve generated from tensile testing. Line tangent to
slope of elastic portion of curve used to determine elongation at failure is shown...50

Figure 3.18. Calculated iso-strain lines from forging showing various effective strain
from forging and respective fatigue sample location.............................................51

Figure 3.19. Calculated cooling rates from 1080C to 8700C in "C/min and
corresponding fatigue specim en location........................................................... ... 51

Figure 3.20. Drawing of fatigue sample used in this research. The dimensions are in
inches ..................................... ......................................52

Figure 3.21. Photograph of LCF samples loaded for thermal cycling. Samples are
separated by the alum ina tubes. .................................................. .................... 53

Figure 3.22. Typical LCF data taken during testing showing the hysteresis loops at
various cycles. Curve for modulus determination is also shown............................55

Figure 3.23. Illustration of sectioning and metallographic mounting of the fatigue
sam ple for evaluation. ....................................... ..................... .. .............................55

Figure 3.24. Drawing of LPPS specimen fixture.................... .......................57

Figure 3.25. SEM backscatter micrograph of Al coated and reacted sample showing
the in-situ formed A13Ti coating on the right. ..................... .. ...................59

Figure 3.26. SEM backscatter micrograph of Si coated and reacted sample showing
in-situ formed TisSi3 coating on the right. ..................... ............................60

Figure 3.27. SEM backscatter micrograph of Pt coating showing the columnar
structure and porous nature of the coating. ...............................................61

Figure 3.28. Photograph of CVD reactor........................................ ......................62

Figure 3.29. Schematic diagram of CVD reactor shown in Figure 3.28.........................62









Figure 4.1. Baseline oxidation results showing no variation between three samples
from different locations run together, and a larger variation with a sample from a
different experimental run...................... ........ ......................66

Figure 4.2. Al and Si coated and reacted oxidation results showing the decrease in
oxidation rate over the baseline. .............................................. ...................... 67

Figure 4.3. Oxide coated oxidation results showing that TaO, and MgO coated
material behaves like uncoated material and SiO, coated samples begin to gain
w eight after 400 cycles. ........................................ ............. ........................68

Figure 4.4. Results showing the large reduction in oxidation rate with the application
of the MCrAIY coatings and the effect of Cr and Pt coatings.................................69

Figure 4.5. Comparison of the oxidation rate behavior of all coatings examined in this
study. ...... ........... ............................................... .............................................. 70

Figure 4.6. Uncoated baseline oxygen content and microhardness vs. depth...................71

Figure 4.7. SEM backscatter image of baseline microhardness sample cycled for 500
cy cles...................... ............................................................................................7 2

Figure 4.8. SEM backscatter image of baseline microhardness sample cycled for 1000
cy c les...................... .............................................................................................7 2

Figure 4.9. SEM backscatter image of the NiCrAlY + A1203 coated sample after 1000
cycles at 650 C ...................................................................... .............................74

Figure 4.10. Microhardness and oxygen vs. depth for the FeCrAlY coated and 1000
cycle exposed sam ple .................................................................. ......................75

Figure 4.11. Microhardness and iron vs. depth for the FeCrAlY coated and 1000 cycle
exposed sam ple. .................... ...............................................................................75

Figure 4.12. Microhardness and oxygen vs. depth for the NiCrAlY + A1203 coated
and 1000 cycle exposed sam ple. ............................................ .................... ....76

Figure 4.13. Microhardness and Ni vs. depth for the NiCrAIY + A1203 coated and
1000 cycle exposed sam ple.................................................... .......................76

Figure 4.14. Microhardness vs. oxygen content for the Al coated and reacted sample.
Note that the distance from the interface starts at -5tm..........................................77

Figure 4.15. SEM backscatter image of Al coated and reacted sample after 1000
cycles at 6500C. Notice the crack in the AITi coating .........................................78









Figure 4.16. Microhardness vs. oxygen content for the Si coated and reacted sample.
Notice that the distance from substrate starts at -5pm..............................................79

Figure 4.17. SEM backscatter image of the Si coated and reacted sample after 1000
cy c les..................... ..........................................................................................7 9

Figure 4.18. SEM backscatter micrograph of Pt sputter coated sample cycled for 1000
cycles............. ........ ...... ....... .... .......... .. .. .... .........................81

Figure 4.19. Microhardness and oxygen content vs. depth for the Pt coated and 1000
cycle exposed sam ple .................................................................. ...................81

Figure 4.20. Plot of Cr and oxygen content vs. depth in 1000 cycle exposed oxidation
sam p le ............................................... ...................................... ............................ 82

Figure 4.21. SEM backscatter image of the SiO, coated sample after 1000 cycles at
6 5 0 C .......................................................................................... .........................8 3

Figure 4.22. Microhardness vs. oxygen content for the SiO, coated sample....................84

Figure 4.23. SEM backscatter image of Ta2,O coated sample showing area analyzed
by Auger spectroscopy. Box outlines area where compositional maps were
obtained ........................................................ ........................... ............................84

Figure 4.24. Auger compositional maps showing the concentration ofTa, Ti and
oxygen in the boxed region shown in Figure 4.23.................... ....... ............ 86

Figure 4.25. Number of cycles to failure vs. strain for various locations in the forging.
The sample number is shown beneath each symbol, and the failure location is
listed above each symbol .................... ........ ....................87

Figure 4.26. Ultimate tensile strength, yield strength and percent elongation (plastic)
vs. calculated effective strain from forging.......................... ............................88

Figure 4.27. LCF data showing the calculated a, and a, values, along with failure
location and sample number for each data point...................... ...................90

Figure 4.28. Optical fractograph of specimen taken from a highly stressed region in
the forging (N o. 52). ...................................................... ................................ 92

Figure 4.29. Secondary SEM photograph of fracture initiation site from specimen #52
show n in Figure 4.28....................... .... ...................................................... 93

Figure 4.30. Secondary SEM photograph of the fracture initiation site shown in the
small white box in Figure 4.29. ......................................... 94









Figure 4.31. Optical fractograph of sample number 6 taken from a low strain region in
the forging ................ .............. .............................................. 94

Figure 4.32. Ultimate tensile strength, yield strength and percent elongation of exposed
and unexposed samples tested in air and vacuum at room temperature and 5400C...96

Figure 4.33. Plot of number of cycles to failure vs. strain. Failure location is noted
w ith each sam ple ....................................................................... ..................... 97

Figure 4.34. Number of cycles to failure vs. strain for the unexposed and exposed
samples. This construction uses logarithmic curve fits to describe the LCF
behavior of the exposed samples. ............................................. ..................... 99

Figure 4.35. SEM backscatter image of a sample exposed for 100 cycles at 650C.
The dark region at the top of the photo is an area of high oxygen concentration.
The fracture surface is oriented to the right in this photo.....................................100

Figure 4.36. Optical micrograph of cross section from exposed tensile sample
stretched to 0.7% strain. Notice cracks run the depth of the oxygen rich zone......101

Figure 4.37. Optical micrograph of cross section from exposed tensile sample stretched
to 1.2% strain. Note the crack runs beyond the oxygen rich region .....................102

Figure 4.38. Uncoated, unexposed and exposed data, showing the failure location and
the calculated a ...................................... .................................................... 104

Figure 4.39. Oxygen concentration vs. depth from surface for the exposed fatigue
sam p les................................ ...............................................................................10 5

Figure 4.40. SiO, coated LCF data. Plot includes the polygons drawn for the baseline
uncoated unexposed and exposed material. ..................... ......................107

Figure 4.41. TaO5 coated LCF data. Plot includes the polygons drawn for the baseline
uncoated unexposed and exposed material. ..................... ....................108

Figure 4.42. MgO coated LCF data. Plot includes the polygons drawn for the baseline
uncoated unexposed and exposed material. ................... ......................109

Figure 4.43. Cr coated LCF data. Plot includes the polygons drawn for the baseline
uncoated unexposed and exposed material. ............................................110

Figure 4.44. Depth of oxygen concentration in SiO, coated LCF samples exposed in
air and argon........................ .... ............................................................... 11

Figure 4.45. Depth of oxygen concentration in TaZO, coated LCF samples after
exposure in air and argon ............................................................ ..................... 112









Figure 4.46. Depth of oxygen concentration in Cr coated LCF samples exposed to air
an d argo n ....................................................................... ... .................................. 1 13

Figure 4.47. Cr concentration depth in the Cr coated LCF samples exposed in air and
arg o n .................................................................. ............. ................................. 1 14

Figure 4.48. SEM backscatter image of Cr coated LCF sample exposed in air for 100
cycles at 6 50 C .................................................... ................... ....................... 14

Figure 4.49. Coefficient of thermal expansion comparison between coatings and Ti-
22A1-26Nb substrate........................... ....... ....................... 115















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

EFFECT OF COATING AND EXPOSURE ON THE OXIDATION AND
MECHANICAL PROPERTIES OF Ti-22A1-26Nb

By

James Ross Dobbs

December 1997
Chairman: Michael J. Kaufman
Major Department: Materials Science and Engineering

High temperature titanium base alloys are being developed to increase the thrust-

to-weight ratio in aircraft turbine engines. Conventional titanium alloys, such as Ti-6AI-

2Sn-4Zr-2Mo (wt%), can only be used to 5400C. Advanced titanium aluminide alloys

based on the a2 phase, such as Ti-24Al-11Nb (at%), and the orthorhombic phase, such as

Ti-22A1-26Nb (at%), have better high temperature tensile and creep properties than the

conventional alloys but have lower ductility. In addition to having low intrinsic ductility,

the alloys are embrittled by high temperature exposure in air.

In this study, the cyclic oxidation rate of Ti-22A1-26Nb (at%) at 6500C was

determined, and several classes of coatings, including NiCrAIY+A12O,, FeCrAIY, Pt, Cr,

Al3Ti, TisSi3, SiO,, Ta2O, and MgO, were evaluated to determine their effect on the

oxidation rate. Baseline low cycle fatigue life was established, and the effect of exposure

in air at 6500C on LCF life was determined. Based on oxidation and microhardness,









coatings ofCr, SiO,, Ta2O, and MgO were selected for a detailed study of the effect of

exposure in air at 6500C on the low cycle fatigue life of coated and uncoated material.

It is shown that exposure in air at 650C reduced the LCF life of Ti-22AI-26Nb by

as much as 2 orders of magnitude for a 100 cycle exposure. A single cyclic exposure

reduces LCF by 1.5 orders of magnitude. No coating evaluated reduced this degradation,

and in the case of SiO2 and MgO, the LCF life was worse after exposure than when no

coating was applied. It is shown that the observed reduction in LCF life is associated

with an increase in oxygen content and microhardness measured in the near surface

regions of the oxidation samples after thermal cycling. The oxide coatings were reduced

by the substrate either during coating or during exposure. The Cr coating allowed oxygen

to diffuse to the substrate.

These results indicate that further work should study those coatings that would not

be chemically reduced by titanium and that would not allow oxygen to diffuse through

the coating to the substrate.














CHAPTER 1
INTRODUCTION

Titanium base materials are used in a wide variety of applications. The medical

and chemical industries use titanium because of its resistance to corrosion and its

biocompatability, the sporting goods industry uses it because it is light weight and stiff,

and the aerospace industry uses titanium alloys for their high strength-to-weight ratio at

elevated temperatures. The use of titanium alloys in gas turbine engines is the most

demanding since the material is subjected to a corrosive environment at elevated

temperatures and, in some cases, high stresses that are frequently cyclic in nature. Thus,

titanium base alloys are attractive for use in gas turbine engines because of their low

density compared to Ni-base alloys (4.5 gm/cm3 vs 8.9 gm/cm3) and their high strength at

elevated temperatures [Pos92].

Titanium alloyed with aluminum and vanadium, Ti-6wt%Al-4wt%V, has been an

industry standard for many years [Woo72]. Below 1000C, this alloy consists of a strong

hcp a-phase precipitated in a softer bcc p-phase matrix, producing a stable two-phase

microstructure of hcp a-phase and bcc P-phase [Woo72]. The P phase is stabilized by the

addition of the vanadium whereas the addition of aluminum stabilizes the a phase. In

binary Ti-Al, single phase a is stable up to aluminum levels of approximately 11 atomic

percent. At greater aluminum levels, the disordered a phase is in equilibrium with the

ordered hcp TiAI (a2) phase [Mur87]. Single phase a, is stable from 22 to 35 atomic









percent aluminum at 5000C. The substitution of niobium (also a 3 stabilizer) for titanium

in the a, Ti3Al alloys leads to a better balance of properties, i.e. improved creep

resistance, fracture toughness and oxidation behavior. Initial work on the Ti-Al-Nb

system was carried out by McAndrew and Simcoe [Ml. ,., I] who studied alloys with up

to 20wt% Al and 30wt% Nb. More recent work conducted by Blackburn [Bla78]

indicated the promise of the TiAl class of materials as engineering alloys. The specific

tensile yield strengths of the TiAI + Nb alloys as a function of temperature are superior

to that of conventional titanium alloys, and even better than cast and wrought IN718

[Pos92] as shown in Figure 1.1. Several engine components have been produced from

Ti3A1 alloys, including afterburner nozzle seals and high pressure compressor casings for

the F100 engine and exhaust seals for the F404 engine [Lip85]

900


S800


r 700 Orthorhombic Alloy


600II 718
1 \Alloy 718

.S 500 Ti-6242


400
0 200 400 600 800 1000 1200 1400
Temperature (F)

Figure 1.1. Specific yield strength vs. temperature comparing an a( and orthorhombic Ti-
Al-Nb alloy with a conventional Ti alloy and IN718. After [W...','3 I









The discovery by Banerjee et al. [Ban88] and Rowe [Row90] that higher additions

of niobium (above approximately 12 at%) to Ti3Al leads to the formation of an ordered

Ti2AlNb orthorhombic phase (O Phase) introduced another class of materials for high

temperature applications. Alloys based on this O phase tend to have higher toughness

and tensile strength than alloys based on the a2 phase [Row91, Row91a, Row93, Smi93b,

Smi94] while maintaining similar creep resistance. Extensive evaluation of this class of

materials has been conducted in recent years [Ban95]. The deformation behavior and

dislocation processes have been evaluated [Ban91, Ban92, Ban95a, Dar94, Dou93,

Pop96a, Pop96b], as well as phase transformations and relationships [Ban93, Kau88,

Mur95, Vas96, Wey89].

A major impediment to using titanium-base materials at elevated temperatures in

oxidizing environments is the propensity for oxygen to diffuse into the titanium [Unn86].

The oxygen, being a small atom, sits in the titanium lattice interstices and causes the

titanium alloy to become very hard and thus embrittled [Cha87, Unn86]. In the case of

oa-p alloys, oxygen is a strong a stabilizing element and leads to a hard surface layer

known as "a-case". In the a2 and O alloys, there is a high solubility for oxygen in the

lattice, but these a like phases are not as readily stabilized by oxygen. Therefore, there is

no discernible change in the microstructure due to oxygen ingression during exposure.

Even so, high oxygen surface regions are embrittled and lead to a reduction in ductility,

fracture toughness and fatigue resistance [Dar95, Dar96, Lip93, Sai93, Sch95]. These

reductions in properties have been a major barrier to the implementation of this class of

materials in gas turbine engines [Pos92].









Several attempts to mitigate the environmental embrittlement have been made

with limited success. For example, 02 and O alloys have been coated with a variety of

materials, including MCrAlYs [Bri92, McK93, Sch95], glasses [Dev90, Wie89, Wie91],

silicides [Coc96], aluminides [Kun90, Smi90, Smi93b, Sub88] and Pt [Nic96]. These

coatings resulted in embrittlement, presumably due to either the formation of deleterious

reaction phases, or the coating itself was brittle and caused a reduction in low cycle

fatigue (LCF). The process of applying the coating could also contribute to the reduction

in LCF life. Subsequent exposure of these coated alloys to a high temperature oxidizing

environment typically led to negligible further embrittlement over the as-coated condition

presumably because the initial deterioration had already occurred during coating.

The objectives of this research include identifying a potential coating system and

methodology that does not lead to degradation upon application and is protective during

subsequent exposure, and to identify the mechanisms that can cause changes in the

mechanical properties of Ti-22A1-26Nb due to the presence of specific protective

coatings.















CHAPTER 2
LITERATURE REVIEW

Overview of Ti-Al-Nb Metallurgy


History

The development of supersonic jet aircraft after World War II nurtured the need

for high thrust-to-weight ratio aircraft turbine engines. This initiated the search for new

structural alloys that could withstand the high temperatures and stresses induced by a

turbine engine [Jaf80]. In aerospace applications conventional aluminum alloys are only

useable up to approximately 150C. Conventional titanium alloys like Ti-6A1-4V (wt%)

can be used at temperatures exceeding 300C. Today's commercial alloys have the

strength capability to operate at temperatures approaching 600C. Their oxidation

resistance is not yet on the same level as the mechanical properties and, consequently,

coatings have been developed that mitigate the poor oxidation behavior [Eyl84]. In the

early stages of titanium alloy development, it was recognized that the addition of

aluminum would increase the tensile and creep strength as well as the elastic modulus

[Ely84]. Much work was conducted in subsequent years to develop an accurate Ti-Al

phase diagram and in 1987 Murray [Mur87] published a version of the diagram based on

much of this work. This diagram is shown in Figure 2.1 and includes all of the phases

that form with the increasing aluminum.




















1200


noD- 0T / f N







Ti Atomic Percent Alum num A]
Ti5Aii


.0 (aoTi) I





0Atomi2 P-r"2nt Aluminum Al

Figure 2.1. Titanium aluminum equilibrium phase. After [Mur87].



Based on the phase diagram and empirical studies, it was determined that the

amount of aluminum, or its equivalent, should be kept below 9 at% to remain in the a

phase region. Higher levels of aluminum led to the formation of the intermetallic Ti3Al

and a reduction in ductility. In the 1970s it was recognized that the TiAl and TiAl class

of materials could possibly be exploited as structural engineering alloys. Ti3Al, also

known as a,, was developed by the Air Force [Lip81] and at the aircraft engine

manufacturers. It was recognized that the addition of Nb to Ti3Al improved the ductility,

creep behavior and oxidation resistance [Bla78, Sas77], and a new class of TiAl alloys

was launched. At the lead of these materials was the ternary alloy Ti-24Al-I 1Nb (at%).

The addition of vanadium and molybdenum to a, improved properties even more, and

super a(, or Ti-25Al-10Nb-3V-1Mo, was developed. The addition of Nb at levels above









about 15 at% causes the Nb atoms order on the titanium sublattice leading to a distortion

of the ordered hexagonal cell into the orthorhombic phase [Ban88]. The orthorhombic

alloys are reported to have superior tensile strengths over the a2 alloys (Figure 2.2), better

oxidation resistance and comparable creep strengths. [Row90, Row91, Row91a, Row92,

Row93, Smi93, Smi94]. It is this phase that is the focus of this study.


3.0


2.6-


2.1
U
a 1.7
E

. 1.3-


S 0.9-


0.4-


0.0I I I I I
0 100 200 300 400 500 600 700 800
Temp (*C)

Figure 2.2. Comparison of the specific yield strength of ax and orthorhombic alloys as a
function of temperature. After [Row91].


Description of Phases Present in Ti-Al-Nb

As seen in Figure 2.1, there are five distinct phases present in the Ti-Al system.

Four of these P, a, TiAl1 or a2, and TiAI or y are interesting engineering alloys. AlTi has

shown promise as a coating for titanium alloys and will be discussed in a later section.

The 0 phase is a body centered cubic (bcc) structure and is stable at higher temperatures.


Ti-22AI-27Nb
-*-Ti-25AI-1 ONb-3V-1 Mo
--Ti-24AI-17Nb-1Mo




--..\








On cooling through the 1 transus, this phase undergoes a transformation to the a phase.

The a phase is a hexagonal close packed (hcp) phase consisting of a random arrangement

of titanium and aluminum atoms. As seen in Figure 2.1 the low temperature solubility of

aluminum in titanium is approximately 10 at%; beyond this amount a 2 phase mixture of

a and TiAl, and ordered structure known as ca, exist in equilibrium. Aluminum

additions over 22 at% will create single phase a2. The basal plane of the a, lattice is

depicted by a ball model in Figure 2.3.



Il Ti Al



Ti Ti Ti Ti



Al )l Ti ; I Al Ti -. a2



STi I Ti Ti Ti



Al Ti A


al -a3
Figure 2.3. Ball model of basal plane of a, showing lattice sites for Ti and Al atoms.
The a phase lattice is similar, with the Ti and Al atoms arranged randomly.









The addition of Nb to Ti3Al ast levels beyond 10 at% results in either a 2 phase

(a + B2 or O + B2) or 3 phase (a2 + O + B2) structure as will be discussed further

below. A partial 9000C Ti-Al-Nb isotherm (Figure 2.4) indicates that the maximum

solubility of Nb in a, is 15 at%, and above 10 at% Nb the a, is in equilibrium with B2 up

to 15 at% where the a2 and B2 are in equilibrium with the O phase. The B2 phase is an

ordered bcc phase formed from the p phase. The addition of Nb causes the p phase to

order with Ti atoms occupying one site and the Al and Nb atoms randomly occupying the

other [Ban87] as shown in Figure 2.5.

Ti

















T60Tib (6 FT--
0




B2+O+u,

Compositons o f
Interest




Ti-60Nb 9000C (16500F) Ti-60AI

Ti-Al-Nb Ternary

Figure 2.4. 900C isotherm from the Ti-Al-Nb ternary phase diagram. B2, a2, and
orthorhombic phases are in equilibrium. After [Row92].









As seen in Figure 2.4, there are three important phases of interest in the Ti-Al-Nb

system at 900C, namely, ac, B2 and 0. A ball model of the basal planes of the a2 and O

phases is shown in Figure 2.6. The orthorhombic unit cell is derived by a slight distortion

of the a2 unit cell caused by the ordering of the Nb atoms on the Ti sites [Kon86]. For

complete substitution, it is clear that the O phase is an ordered ternary phase based on the

composition TiAINb.


Figure 2.5. Ball model drawing of the B2 phase showing the relationship between the
titanium, aluminum and niobium atoms.









a3 [TT20] A1I OTi *Nb

..O ) "-o-*--a[100]



[T210] O



al [210] [010]

a2 O

Figure 2.6. Comparison of the basal planes of the a, and orthorhombic phases.


Szaruga et al. [Sza92] reported that oxygen is a strong a, stabilizer in Ti-25A1-

10Nb-3V-1Mo having an effect on both the p transus temperature and the p/B2 order-

disorder transition. Oxygen is also an important alloying element in the O phase. An

alloy of composition Ti-22A1-23Nb containing above 1000 wppm oxygen is made up of

the a2, O and B2 phases. Reducing the oxygen content below 1000 wppm produces only

two phases, O and B2 [Lip93]. This indicates that oxygen stabilizes the a, phase, much

as oxygen stabilizes the a phase in conventional titanium alloys. Rhodes et al. [Rho93]

reported the same effect in the alloy Ti-22A1-27Nb, but at oxygen concentrations of

approximately 1120 wppm the equilibrium phases were still O and B2. Therefore, at

higher Nb contents it takes more oxygen to stabilize the a, phase. Ward [War93]

concluded that interstitial atoms such as oxygen can be potent strengtheners, but have an

adverse effect on ductility by limiting the number of slip systems available.










A cut through the ternary phase diagram shown in Figure 2.4 along the constant

25 at% Al line is shown in Figure 2.7. The P to B2 ordering line is shown above 1100C

for compositions around 12 at% Nb.


B2 to ri/O
Widanstatten




82 to o2
Massive


B2toO
Massive



Tempered
martensite





B2 to (type


Nb (at %)

Figure 2.7. Constant 25 at% Al isopleth from the Ti-Al-Nb ternary in Figure 2.4. After
[Ban95].



Also shown in Figure 2.7 is the range of transformations that Ti3AI + Nb alloys

experience. The transformations associated with continuous cooling of high Nb alloys

(>16 at%) from the a, + B2 region is complex. The precipitation of a2 or O laths follows


the Burgers relationship (0001)a2, (110)1; [112 0]a2 [I1 l]p and the equivalent









relationship for the O phase (001)0 1 (110)1, [110]0 1 [111]3 [Ben91]. The

transformation from ac to O phase results in a lamellar or mosaic structure [Ban88].

Microstructures obtained during thermomechanical processing of alloys with high Nb

contents depend on the processing temperature. Processing in the B2 + a, or O phase

field will yield, on cooling, a mix of primary a, or O phase surrounded by a

Widmanstatten a2 or O + B2 mixture. Processing in the B2 field results in a 100%

Widmanstatten a2 or O laths with B2 in the interlath regions. Examples of these two

microstructures are shown for the alloy used in this study, Ti-22A1-26Nb in Figure 2.8

and Figure 2.9 respectively.




















Figure 2.8. SEM Backscatter micrograph of Ti-22A1-26Nb processed in the O + B2 field
and directly aged at 815C for 4 hours. After [Woo93].




























Figure 2.9. SEM Backscatter micrograph of Ti-22A1-26Nb processed in the B2 field and
directly aged at 8150C for 4 hours. After [Woo93].


Deformation Behavior of a,, O and B2 Phases

Dislocation arrangements in deformed a, alloys [Akk91, Ker84, Kos90, Lip 80,

Lip85] and O alloys [Ban91, Ban92, Ban95a, Dou93, Pop96b, Pop96a] have been studied

in depth. This includes deformation in the B2 phase contained in both materials.

The a, phase deforms by slip [Ban90] in two distinct systems, (1120)(0001)

and (1120){1010}, and by slip in the (1126){1121} system [Kos90]. The

crystal structure and Burgers vectors for the a2 phase are shown in Figure 2.10 along with

those for the O phase [Ban95]. slip has been observed in ac alloys but are associated

with the P to a, transformation rather than tensile deformation [Ban95]. The lack of

slip and the dependence of a, alloys on
and slip create plastic

incompatibility and thus the a, alloys are considered brittle at room temperature.









The O structure is similar to the a2 phase with a slight distortion due to ternary

ordering. Figure 2.10 compares the Burgers vectors for the O and ao phases, and it can

be seen that the three 1/6(1120) vectors in a, are not equal in the O phase in either

magnitude or generation ofAPBs. Banerjee et al. [Ban91] have shown that the

deformation associated with
slip in O is very similar to that in a,, but unlike a, there

is considerable slip in the O phase alloys and these dislocations are arranged into

well defined slip bands [Kos90]. This would suggest that the O phase alloys should

possess more plasticity at room temperature and behave better under fatigue crack growth

conditions.

1/6 [1126 Al 12102]
0 T
0 Nb
S 1/4[114]




1/6 [1120] /2[100

o 0 1/4[110]


Figure 2.10. Crystal structure and burgers vectors for the a, and O phases. After
[Ban95]


Deformation in the B2 phase occurs predominantly by (111) slip on the {110},

{121} and {123}planes [Ban90]. This deformation is extremely inhomogeneous and

localized into slip bands as shown in Figure 2.11. This localization could result in the

formation of persistent slip bands and contribute to the formation of crack nucleation sites

during fatigue [Mil97].



























a FS \ pm

Figure 2.11. Slip bands in the B2 phase in Ti-24Al-11Nb showing the inhomogeneous
nature of slip. After [Ban90].


Tensile Behavior of a, and O Alloys

The discovery that the TiAl+Nb alloys have higher tensile ductility than the

binary Ti3Al alloys [Sas77] indicate that Ti3Al base alloys, including the O phase alloys,

have to be alloyed sufficiently to stabilize the P or B2 phase with Nb as the preferential 3

stabilizing element. The B2 phase delays the cleavage cracking of the a, or O laths to

higher strains by its ability to accommodate the plastic instabilities associated with the

lack of slip systems in a,. This typically occurs at the a,/B2 interfaces with its large

number of available slip systems [Ban95].

Recall that the decomposition of the B2 phase into the a( or O phase when heat

treated above the p(B2) transus results in a lath type microstructure as shown in Figure









2.9 where the a2 or O laths have the Burgers relationship with respect to the B2 phase.

The cooling rate from the p(B2) transus strongly affects the strength, as shown in Figure

2.12, and the ductility, as shown in Figure 2.13. The yield strength increases continually

with increasing cooling rate while the ductility goes through a maximum.


1400.0 --


1200.0



-1000.0



S800.0
800.0


D 24-15
0 24-11


600.0 --



400.0 I
0.01 0.1 1 10
Cooling Rate (C/sec)

Figure 2.12. Yield strength vs. cooling rate for two a, alloys. After [Gog90].










U
O 24-15 El
O 24-11 El
4.0



3.0 -



2.0



1.0 O



0.0 -E- T .
0.01 0.1 1 10 100
Cooling Rate (C/sec)

Figure 2.13. Elongation vs. cooling rate for two a, alloys. After [Gog90].


The increasing cooling rate causes a finer size lath to be formed on cooling from

the P(B2) phase field. This finer lath arrangement yields a higher yield strength due to

the Hall-Petch relationship. The low ductility at low cooling rates results from the

formation of similarly oriented lath colonies which allow cleavage to occur across many

laths with no change in crack energy [Ban90]. The reduction in ductility at high cooling

rates is associated with the reduction in scale of the microstructure and the reduction in

volume fraction of the P phase. At high cooling rates there are also a, and O laths that

have nucleated from the grain boundaries. These grain boundary nucleated laths have a

similar orientation and thus a crack that forms in one can easily propagate into the others









with little energy loss. Therefore, the optimum lath structure is a fine basketweave of the

ca2 or O phase without any grain boundary film or grain boundary initiated lath colonies.

An example of a colony of O laths is seen in Figure 2.14 where the lath colony in the

lower right hand comer has formed sympathetically from the larger O lath running from

bottom right to top left. This is a micrograph of the Ti-22AI-26Nb alloy used in this

study.


Figure 2.14. SEM backscatter micrograph of Ti-22A1-26Nb forged in the B2 region and
directly aged at 804C. Notice the colony of O laths that have formed off of the large
lath.









The thermomechanical processing history of the material can also play a role in

the macroscopic tensile behavior. Material that has been forged and directly aged has

higher tensile strength than material that has been forged, solutioned above the p(B2)

transus and then aged [Dob94]. This is again due to the failure of the a, + P at the ca

film formed along the grain boundary interface due to strain incompatibility [C I',i.I as

seen in Figure 2.15.


Figure 2.15. Intergranular crack observed in tensile tested Ti-22A1-27Nb solutioned
above the P(B2) transus and aged at 8700C for 50 hours.


The high grain aspect ratio (GAR) left over from forging in the P forged + direct

aged material forces this failure to occur on boundaries that are parallel to the stress axis

as seen in Figure 2.16, as opposed to a few boundaries that are perpendicular to the stress

axis, as seen in Figure 2.17. The 0 forged and directly aged material also did not contain









the grain boundary a, film, and the lath size was much finer [Dob94]. This is indicative

of faster cooling rates from the P region.


Figure 2.16. Fractograph of forged + direct aged material. Notice the high GAR and the
initiation of failure from an internal boundary.









Figure 2.17. SEM fractograph of material solutioned above the P(B2) transus and aged at
870'c for 50 hours. Notice that there is mixed intergranular and transgranular fracture
across the sample and the prior p grain size is approximately 500Pm.


Fracture Toughness and Fatigue

Chan determined [Cha92] that the crack growth process in a coarse basketweave

structure of Ti-24AI-11Nb (at%), one where the laths are approximately 20 jtm long and

5 tm wide, was by decohesion of slip bands, similar to the mechanism for equiaxed a,.

The tips of the microcracks terminated in the P phase, as did the tip of the main crack. In

contrast, for a fine basketweave structure where the lath size is approximately half that of

the coarse laths and where the P phase is not as continuous, the cracks propagated around

and through the p phase. Therefore, Chan concluded that the room temperature fracture

toughness of Ti-24Al- 11Nb (at%) was imparted by the P phase inhibiting microcrack

nucleation by both relaxing the incompatibility strain in the a//p interfaces and by

blunting the crack tips.

Penton et al. reported that Super a, (Ti-25A-10-Nb-3V-lMo (at%)) containing

primary ax exhibits faster crack growth rates than p solutioned and aged material

[Pen93]. This effect appears to result from early cleavage of the brittle primary a,

[Pen92, Pen93]. Ward reported that the a2 laths appeared to cleave on the basal plane

[War93b] and that the crack is then bridged by the p phase [Gog90, Luk90]. Others have

reported that the cracks also propagate along the a,/p interfaces in transformed regions

[Luk90, Tak96]. Davidson et al. [Dav91] found that, for Super a, rolled below the p

transus and then aged, small cracks were always found in the a2 phase after deformation.

These small cracks would then grow below the AK,,, for large cracks. Ravichandran and









Larsen [Rav92] concluded that growth rates of small cracks in basketweave structures of

Ti-24Al-11Nb were consistent with large crack growth rate data. Penton et al. [Pen93]

concluded that the growth rate of solutioned and aged material is in line with the fracture

of a, laths.

Miller [Mil97] has defined three distinct conditions for initiation of fatigue:

1) Fatigue limit ofpolycrystalline material which is related to some limiting

microstructural feature such as grain size, primary precipitate size or brittle

precipitate size.

2) The mechanical stress state applied to a pre-existing crack or flaw.

3) The condition for a single crystal in which a defect, such as intrusion or

extrusion, must be induced.

Miller concludes that the fatigue crack is initiated on the first cycle by one of the

above conditions. In the case of a and O alloys with limited slip systems in the lath

structures, the first condition of some limiting microstructural feature would apply. This

feature would most likely be a primary a,/ O particle or an az/ O lath, in agreement with

the observations described above.

Fatigue is typically described as consisting of the following stages: [Sur91]

1) Nucleation of permanent damage through microstructural changes;

2) Creation of microscopic cracks;

3) Coalescence of microscopic cracks to form measurable cracks;

4) Stable growth of a preferred crack; and

5) Ultimate catastrophic failure.









In classic fracture mechanics, Stage 3 indicates the end of initiation and the

beginning of propagation of the fatigue crack. Considering this model of fatigue, linear

elastic fracture mechanics (LEFM) can be utilized to help understand the behavior of

material by calculating both the initial flaw size, a, and the size of the crack at failure, af.

The equation for the stress intensity, K,, at the notch of a sharp crack is

1.12 o'ra
K, = (2.1)


3z r a 2 7r
where + In the special case when c = a, the ( term is reduced to the
8 8 c 2

equation then becomes

K, = 0.71 acr- (2.2)

This is the case for a penny shaped flaw that is very small relative to the size of the test

bar. To determine the size of the fatigue crack at failure, af, one can use the plane strain

fracture toughness, Kic for K,, and in the case where A=l, the maximum stress is equal to

Ac. Solving equation 5.2 for a, gives


a,= ( )'2 (2.3)
0.71an_ I

da
Paris showed that the fatigue crack growth increment is related to the stress intensity
dN

factor as

da
N= C(AK)" (2.4)
dN

da
where C and m are scaling constants derived from the vs. AK curve. When equation
dN

2.4 is integrated from an assumed initial flaw size a, to the critical crack size a, the









number of cycles to failure can be calculated. When the stress intensity factor for a small

crack is defined as in equation 2.2, equation 2.4 becomes


d =C(0.71Ao-r, a)" (2.5)
dN

where a has now become Ao, defined as a,,,, a,,,,. For our case where we are assuming

that the stress intensity factor is not a function of the crack depth a, equation 2.5

integrates as


CY'(AcT),7 r ,' da
CY"(Ac)mt N "2 dN = J- (2.6)
0 .,a

The resulting fatigue life is


N,= CYAr' (Im -2 1 (2.7)
_a a, 2.

for n 2.

Comparison of the fatigue crack growth data for Ti-24Al-l 1Nb (at%) generated

by Ravichandran and Larsen [Rav92] to that for Ti-21Al-25Nb (at%) generated by

Woodfield et al. [Woo94] with a similar basketweave structure indicates that the Ti-21Al-

25Nb is slightly better. This could indicate that the O + B2 alloys would behave better in

fatigue, probably due to the increase in the number of active slip systems and the increase

in Nb content in the B2 phase. Balsone et al. [Bal93] evaluated fatigue crack growth in

Ti-25A1-25Nb (at%) with a microstructure consisting of primary a2 + O laths surrounded

by the B2 phase. Comparison of dA/dN vs. AK data for Ti-24Al- 11Nb (at%) with Ti-

25A1-25Nb with a similar microstructure [Rav92], indicates that the latter has a slightly









higher crack growth rate. This is probably due to the scale and volume fraction of the

primary ax or O and the effect of the ac2/P(B2) interface.

Overview of Oxidation Behavior


Oxidation of Uncoated Titanium Alloys

The oxidation behavior of titanium base alloys has been extensively studied since

the late 1940s, starting first with unalloyed titanium. Initially there were large

disagreements between investigators about rate laws and rates of oxidation [Kof58].

More recent investigations of commercially pure titanium by Unnam et al. have shown

that the weight gain due to oxide growth and oxygen dissolution is essentially parabolic

with respect to time [Unn86]. Unnam et al. also reported that the oxygen diffusion

coefficient in Ti(O) is independent of oxygen concentration in the 1-10 at% range, and

the effective solubility limit of oxygen in pure titanium is 20 at%. They also concluded

that the diffusion coefficient of oxygen in TiO, is about 50 times that of oxygen in the

metal. Chuanxi and Bingnan reported that the addition of Nb to pure Ti improved the

oxidation resistance by improving the surface stability of the oxides [Chu92]. Chaze and

Coddet [Cha87] conducted oxidation experiments on Ti with additions of Al, Si and Cr.

They concluded that the addition of Al up to 16.5 at% and Si up to 1.5 at% improved the

oxidation resistance of Ti, while Cr additions up to 18 at% had little or no effect. In

contrast, Kahveci and Welsch evaluated the effect of Al on oxidation ofTi [Kah87] and

concluded that at least 25 at% Al is necessary for any significant improvement in the

oxidation behavior of Ti-Al alloys and that there is no appreciable effect for the addition

of Al until about 13 at%, beyond which the oxidation behavior improves with increasing









Al content. Kahveci et al. [Kah88] showed that the kinetics for oxide growth for Ti-25A1

(at%) fall between TiO2 and AIO, but are closer to the kinetics for TiO,. Qiu et al

[Qui95] reported that the addition of 11 at% Nb or 5 at% Si can improve the oxidation

resistance of TiAl and that the effect of adding both is even greater. However, no

continuous A103 scale was formed after any additions.

In a subsequent paper, Welsch and Kahveci evaluated 3 binary Ti-Al alloys with

increasing Al content and 1 Ti-Al-Nb alloy[Wel89]. They observed that the parabolic

rate constant decreased and the thickness of the oxide scale decreased with increasing Al.

They concluded that the outer oxide scale in the binary alloys is a TiO, with A120,

channels, the intermediate layer consists of TiO2, A,103 and porosity, and the inner oxide

layer is A1203 with TiO, and porosity. The alloy surface has A1,03 lamella from internal

oxidation. On the other hand, the Ti-Al-Nb alloy had a dense outer oxide scale consisting

mostly of A1203 and TiO2, an intermediate layer of TiO, and Nb0z5; and the layer next to

the metal was an Nb2Os-TiO, layer with porosity at the interface. There was no visible

internal oxidation but there was an oxygen rich zone adjacent to the oxide scale.

Chromium has been found to help improve the ability to form a protective scale in

Ti-Al-Nb alloys and is an excellent P stabilizing element [Doy95]. The Cr tends to alloy

with the oxide scale and provide a more protective scale. This indicates that

improvements in oxidation are possible by alloying, and Cr is a good alloying addition.









Oxidation of Coated Titanium Alloys

Coatings for titanium base alloys have been evaluated based predominantly on

their ability to mitigate oxidation. The coatings for Ti alloys fall into four broad classes,

Al3Ti, MCrAlY, pure metals, and glasses or oxides.

AlTi

AI3Ti has been produced as a coating by a variety of methods including (1) laser

surface alloying [Abb93] where Al powder is either fed into the molten pool or placed on

the substrate as a slurry and melted in by laser [Gal92], (2) dipping the substrate of

interest into a molten bath of Al and allowing the reaction to take place [Abd91] (3)

deposition of Al by EBPVD and subsequently reacting with the matrix during exposure in

air [Unn85], (4) deposition by sputtering and reacting in vacuum prior to exposure in

a.i [\.- .', .1 and (5) pack cementation [Kun90, Smi90, Smi93a, Sub88] where the

substrate is placed in a mixture of alumina powder, aluminum powders and halide salt

activators and reacted at 1000C where the Al will volatile and produce a coating of Al on

the substrate, which is then reacted to form AI3Ti. In all cases the coating A13Ti was

cracked on cooling from the reaction temperature. Even so, the coatings tended to impart

improved oxidation resistance despite being cracked, although oxidation did occur down

the cracks and affect the substrate [Kun90]. It was concluded that the thicker the coating,

the more susceptible it was to cracking [Smi90]. McMordie reported that the addition of

Si to the Al3Ti coating improved oxidation even more, possibly due to the formation of a

silicide at the coating/substrate interface [McM91].









MCrAlY

MCrAIY type coatings, where M is a transition metal like Ni, Fe or Co, are

typical coatings used in the gas turbine industry. The application of these coatings to Ti

has been attempted by a number of researchers, and several patents are held on the

materials and processes [Bri92, Lut90, Tob92]. The preferred method of producing this

type of coating is plasma spray [McC90, McK93, McK93a, Sch95]. For Ti3Al + Nb

alloys, the MCrAlY is sometimes mixed with an oxide powder such as AIO3 in order to

reduce the thermal expansion of the coating to better match that of the substrate [Sch95].

In all cases, the MCrAIY coatings improve the oxidation resistance of Ti substrates, such

as Ti-6A1-2Sn-4Zr-2Mo (wt%) [McC90]. There is a slight increase in hardness beneath

the coating in the as-coated condition and this is attributed to oxygen ingress during the

pre-heating in partial vacuum prior to the plasma deposition.

Pure metals

The evaluation of pure metals as oxidation resistant coatings for Ti base materials

has mostly been limited to Pt and Cr. The preferred method for deposition of Pt is ion

plating [Eyl84, Eyl85, Fuj79] and for Cr is PVD processes such as sputtering [McK90].

Eylon et al. showed that the Pt coatings reduced the surface oxidation rate of Ti-6Al-2Sn-

4Zr-2Mo (wt%) by three orders of magnitude[Eyl85] and did not degrade the high cycle

fatigue properties [Fuj80]. PtAl2, produced by multi-layer sputtering of Pt and Al and

subsequent reaction, has been evaluated as a coating for a, type alloys by Nicholls et al

[Nic96]. They reported that a 3um thick coating protected then a2 alloy MT754 (Ti-

23A1-9Nb-2Mo-0.9Si at%) from oxygen ingress at 700 and 800C, and extended the









creep life of the alloy by a factor of two under conditions of 350MPa at 6500C. They also

reported that after exposure in air for 100 hours at 7000C, there was no hardening under

the PtAl2 coating.

Glasses, silicates and oxides

Oxides such as SiO2, Y203, MgO, A1203 and ZrO, have been deposited on Ti-base

alloys by a variety of techniques. Wiedemann et al. [Wie89, Wie91] deposited SiO,,

A1203 and B,03 by a sol-gel process, and Y203, MgO, ZrO, and HfO by sputtering onto

Ti-24Al-11Nb (at%). They reported that most coatings had poor integrity and allowed

oxygen to diffuse through. Of all the coatings they tried, the MgO based coating made

from the sol-gel process behaved the best after 1 hour at 982C in air. They concluded

that coatings applied by sputtering were not effective for oxidation protection due to their

poor integrity. In contrast, Clark et al. [Cla88] deposited SiO2 by sputtering and reported

that there was a significant reduction in weight gain at 7000C after 25 hours in air.

DeVore and Osborne [Dev90] coated TiAI with borosilicates and aluminosilicates and

concluded that all coatings showed some improvement in oxidation, but the thin sol-gel

coatings were the worst while a borosilicate coating airbrushed onto the sample exhibited

the best behavior. Bedell et al. [Bed91] utilized ion beam assisted deposition to deposit a

double coating of silicon nitride and chromium on the conventional Ti alloy IMI829.

They concluded that the coating provided protection from oxidation at 7000C for 100

hours, but the SiN4 diffusion barrier is not stable with titanium. Cockram and Rapp

[Coc96] produced silicide coatings on a2 and O phase alloys by pack cementation and

reported excellent oxidation resistance up to 10000C.









Embrittlement of Titanium Base Alloys

As mentioned previously, titanium base alloys suffer from embrittlement when

exposed to air at high temperatures. This embrittlement is due to the ingress of oxygen

and the subsequent increase in hardness in the affected layer [Sha68]. The volume

fraction of the beta phase in these near surface regions in Ti-6A1-4V (wt%) is reduced at

all temperatures when exposed in air [Kah87]. The 3 transus temperature is raised by the

presence of oxygen in Ti-6A1-4V as described by the relationship T,[C] = 937 + 242.7 x

[O wt%] [Kah87]. The depth of oxygen penetration, and thus the depth of hardening due

to interstitial ingress, is a parabolic function [Kah86, She86] thus indicating that the

hardening is due to diffusion of oxygen into the substrate. Wallace conducted tensile

tests on Ti-1100 (Ti-6A1-2.75Sn-4Zr-0.4Mo-0.070-0.02Fe wt%) foils after exposure in

air at 6000C and 800C for up to 1000 hours. He concluded that the elongation was the

most sensitive parameter to minor oxidation, but tensile strength was also affected by

long exposures. He also observed a band of brittle fracture at the metal surface due to

oxygen diffusion into the metal. This band is high in oxygen and made up of the a phase

[Wal95].

Balsone studied the tensile properties of Ti-24A1-1 1Nb (at%) after exposure in air

for 10 and 100 hours [Bal89]. He reported that an annulus of embrittled material was

formed around the surface of an exposed sample that cracked during loading. These

cracks served as notches for premature failure during testing such that both the elongation

and UTS were reduced. Embrittlement of Ti-25Al-10Nb-3V-1Mo (at%) was also studied

by Saitoh and Mino [Sai93] after exposure at 7000C for 100 hours. They also reported









that the elongation and UTS was reduced, as did Meier and Pettit [Mei92] for a Nb

modified u2 alloy after a 24 hour exposure at 9000C. Other investigators have evaluated

the effects of embrittlement on fatigue behavior. Godavarti et al. [God92] reported that

the LCF life of Ti-25Al-10Nb-3V-1Mo (at%) was reduced by 2 orders of magnitude after

short exposure times at 7200C. They attributed this reduction to the formation of a brittle

layer at the surface of the test samples. They also reported that the same pre-exposure in

air had no effect on high cycle fatigue (HCF) life due to the low strains imposed during

HCF testing. Praida and Nicholas [Pra92] found that fatigue crack growth resistance in

Ti-24Al-l Nb (at%) was strongly effected by testing in air at 6500C. Hold time fatigue

crack growth rates were the most effected by elevated temperature testing, indicating an

effect of the environment on cracking. Balsone et al. [Bal93] determined that Ti-25AI-

25Nb exhibited time-dependent crack growth behavior when tested at elevated

temperature. They reported that at 6500C and 750C, there is a contribution from

environmentally assisted crack growth.

Dary et al. evaluated the effect of exposure in air and in vacuum on the tensile

properties of Ti-22A1-23Nb [Dar96, Dar94, Dar95]. They observed a decrease in UTS

and ductility in all samples exposed in air and, although there was no loss of mechanical

properties in the vacuum exposed samples, the external surfaces contained a large number

of small cracks. Although Rhodes et al. [Rho93] observed a, in Ti-22A1-23Nb at oxygen

contents greater than 0.098wt%, Dary and Pollock did not observe any a2 precipitates in

the high oxygen regions when exposed at the same temperature. They supposed that the

760C exposure was too low to cause au precipitation or that the B2 + O phases were









compositionally stable at this temperature. Chesnutt et al. [Che93] conducted tensile tests

on Ti-22A1-27Nb after either a thermal exposure (no environment) at 6500C or an

environmental exposure at 590C in air. The results indicate no by the thermal exposure

but significant embrittlement after the environmental exposure. No evaluation was

conducted on the environmentally exposed samples to establish the cause of

embrittlement.

Effects of Coatings on Mechanical Properties of Ti Alloys

Coatings to prevent the degradation of mechanical properties in Ti alloys have not

received as much attention as coatings to mitigate the poor oxidation behavior. As

described above, Ti base alloys are embrittled after exposure in air at elevated

temperatures. To prevent or at least reduce this effect, a series of evaluations have been

conducted to determine the best coatings and the effect of these coatings on the

mechanical properties in both the as-coated and exposed conditions. Fujishiro and Eylon

demonstrated that the HCF life of Ti-6A1-2Sn-4Zr-2Mo (wt%) was improved at 455C

with the application of a platinum coating [Fuj80]. For the cx alloy Ti-24.5A1-12.5Nb-

1.5Mo (at%), Schaeffer and McCarron evaluated NiCrAl + 40 vol% AlO0 deposited by

plasma spray [Sch95]. They reported that the LCF of as coated samples was

approximately the same as for the uncoated and exposed samples where the LCF life was

reduced by three orders of magnitude. Subsequent exposure of the coated sample did not

degrade the LCF life any further. McKee reported that for the same alloy, an underlayer

of chromium or tungsten prior to deposition of the NiCrAl would eliminate this

embrittlement [McK93]. He based this conclusion on microhardness profiles beneath the









coating, rather than on mechanical property data. Chesnutt et al. evaluated the LCF

response of the O alloy Ti-22A1-27Nb (at%) with and without a coating [Che93] and

concluded that the LCF life was degraded by the coating in a manner similar to that seen

by Scaeffer and McCarron [Sch95].

Program Approach

While there have been various studies addressing the oxidation of coated O phase

alloys and the effect of coatings on mechanical properties, to date there has not been a

systematic study addressing both issues. Such an study will be critical to the use of this

class of alloys in high temperature applications in air, particularly in static applications

with large thermal gradients where the stresses developed by the thermal gradients can be

large and the number of thermal cycles small, therefore subjecting a static part to low

cycle fatigue.

The approach of this program is to evaluate the oxidation, tensile and LCF

behavior of O phase alloy Ti-22A1-26Nb in both the uncoated and coated conditions. The

material will be taken from a forged disk, and the baseline properties of the material will

be determined, including the effect of location in the forging. Coatings will include

oxides and pure metals, and are intended to be barriers for oxygen diffusion and diffusion

of any topcoat that may be identified.














CHAPTER 3
EXPERIMENTAL PROCEDURE

Material Processing


A 6100 pound ingot of target composition Ti-22A1-26Nb (atom percent) was

produced at Teledyne Allvac/Vasco. The material was produced by an initial plasma arc

melting (PAM) into a 43 cm (17 inch) diameter ingot, followed by vacuum arc remelting

(VAR) into a 53 cm (21 inch) diameter ingot. This 53 cm (21 inch) diameter ingot was

sectioned into two pieces. The top section was cogged down from 53 cm (21 inch) to 41

cm (16 inch) diameter billet by heating to 12320C (2250F), and using 2.5 cm (1 inch)

drafts. This was followed by hot grinding and cooling to room temperature. This top

half was converted to 20 cm (8 inch) diameter billet and sheet bar for use in other

programs. The bottom half was cogged down from 53 cm (21 inch) to 40 cm (16 inch)

billet using a similar approach as the top half, except there was a 40% reduction initially

and fewer reheats were used. This 40 cm (16 inch) diameter billet was then GFM

converted to a 15 cm (6 inch) diameter billet. This conversion took place at 1010C

(18500F), based on beta transus measurements of a slice from the 41 cm (16 inch)

diameter billet. Four billets approximately 2.4 m (94 inches) long and 16 cm (6.3 inches)

in diameter weighing a total of approximately 2000 pounds were produced. Chemistries

were measured from several locations along the length of the ingot and from the surface

and center of these locations. The compositions are listed in Table 3-1 [Woo84].









Table 3-1. Compositions from several locations along the length of the 15.25 cm
diameter billet, surface and center, and from the forging used in this study.
Billet Forging Aim
Al (at%) 21.35 21.29 22.0+1.0
Nb (at%) 25.75 26.15 26.0+1.0
Ti Balance Balance Balance
Fe (wt%) 0.036 0.06 <0.07
C (wt%) 0.013 0.0063 <0.05
H (wt%) 0.020 0.0014 <0.0125
O (wt%) 0.071 0.075 <0.08
N (wt%) 0.012 0.012 <0.05


A 30.5 cm long section from one of these billets was forged at Wyman-Gordon in

Houston. The two step forging consisted of an O + P forging followed by a P forging.

The forging parameters are listed in Table 3-2.


Table 3-2. Forging parameters for the material used in this study. The first line is step 1
and the second line is step 2.
Initial Final Reduction Billet Temp Die Temp Strain Cooling
Height Height ratio C (F) o C (F) Rate (offdies)
cm (in) cm (in) (/min)
30.48(12) 15.88 (6.25) 1.92/1 996 (1825) 870 (1600) 0.95 Air
15.88 (6.25) 6.35 (2.50) 2.5/1 1100(2010) 870 (1600) 0.92 Fan


After the billet was forged, it was removed from the forging press and allowed to

cool to room temperature. It was then given an aging heat treatment at 8150C (15000F)

for 4 hours followed by air cooling without being subjected to a solution heat treatment.

This procedure is termed direct aging. A typical microstructure is shown in Figure 3.1. A

diametrical slice was then taken across the center of the forging for evaluation of the

material flow. Unfortunately a carbide cut-off wheel without cooling was used to section

this slice and ,as a result, the slice and both cut faces of the forging were severely









cracked. This caused some concern as to the integrity of the forging, so scanning

acoustic microscopy (SAM) was conducted to evaluate the depth of the cracking.


Figure 3.1. BSE image of a typical microstructure of the Ti-22Al-26Nb material after B2
forging and heat treatment at 815C for 4 hours.


A 1.27 cm (0.5 inch) thick section was cut off of each forging face prior to the

SAM analysis. The SAM analysis revealed no cracking on the remaining faces indicating

that all of the cracks were confined to the near surface region. The 1.27 cm (0.5 inch)

thick section was cut in half and polished and macro etched to reveal the flow lines in the

forging. The macro flow lines are shown in Figure 3.2.
























Figure 3.2. Photomacrograph of one half of the forging showing flow lines.


After the forging was pronounced sound, an extensive analysis of the variation in

grain size and grain aspect ratio versus location in the forging was conducted. The

effective strain from forging was calculated and iso-strain lines, shown in Figure 3.4,

were constructed. The calculated iso-strain lines were superimposed onto the forging

macrograph, and the location of various forging strains were identified. Figure 3.3 shows

the half diametrical slice from the forging center marked for photomacrographs at 10x

magnification. After the photographs were taken, the diametrical slice was sectioned

along the line shown into metallographic samples for microstructural analysis. The

forging was assumed to be symmetrical about the centerline and thus was sectioned to

reveal the grain structure in one quadrant of the forging. The diametrical slice was

sectioned with a carbide cut-off wheel and the samples were mounted in bakelite and

polished and etched. Microphotographs from representative locations in each samples

were taken at 50x, 500x and 1000x magnification.























*S STEL



Figure 3.3. Forging half showing location of 1Ox macrophotographs and lines for
sectioning into metallographic samples.



3.00
Eff. Strain
S. A 0.131
.' : e~%"" -" -;~, N A= 0.300
2.00 B 0600
N\ C 0.800
.2- -- -.7. N D=1.200
a aN /G F IE D E= 1.500
1.00- -- / F 1.800
G=2.100
H= 2.400
I= 2.700
.00 I I- =2.957
.00 1.00 2.00 3.00 4.00 5,00 6.00 7.00

Radius

Figure 3.4. Calculated iso-strain lines for the forging examined in this study.



Representative micrographs from the dead zone in the forging, location C, and a

more highly strained area, location P, are shown in Figure 3.5 and Figure 3.6,

respectively.



























Figure 3.5. Photomicrograph taken from area C. Notice the equiaxed grain structure and
low grain aspect ratio.


2 L-I 0 .-,

Figure 3.6. Photomicrograph taken from area P. Notice the high grain aspect ratio
resulting from the higher effective forging strain.









Notice the equiaxed grain structure from area C and the high grain aspect ratio in

area P. These two areas encompass the range of microstructures (grain sizes and aspect

ratios) that resulted from the forging process.

The rate at which the forging cooled from the P transus temperature would effect

the size and distribution of the O phase lath. Thus, the cooling rate of the forging from

the p transus temperature through 870C was calculated. There was little variation in the

cooling rate across the forging, and thus little variation in the O phase lath size would be

expected. This was confirmed by performing high resolution SEM analyses, shown in

Figure 3.7 and Figure 3.8, and little or no variation in the orthorhombic lath was

observed.


.4 t'" l .- ,. u. -. 1 : i i l r : ... : r ,:. j L" .
Figure 3.7. High resolution SEM image of area S (identified in Figure 3.3) from the
slowest cooled region in the forging.





































Figure 3.8. High resolution SEM image of area Q (identified in Figure 3.3) from the
fastest cooled region in the forging.


Specimen Blanking


After determining that the forging was sound, a plan for sectioning the oxidation

samples from the near surface dead zone and the mechanical property samples from

inside the forging was devised. This was based on the premise that the microstructure

found in this dead zone should not effect oxidation. The sectioning plan along with the

two ingot sections were submitted to the wire electro-discharge machining (EDM) shop

for sectioning as shown schematically in Figure 3.9.



































0.250" 117


Figure 3.9. Fatigue and oxidation specimen sectioning diagram


The remaining oxidation samples and the tensile samples were sectioned from the

other forging half as shown in Figure 3.10 which also shows the scrap blocks, numbered

1 through 6, that were left after sectioning. The oxidation samples were cut from near the

surface and from a block in the middle of the forging, some in low strain and some in

high strain regions, as shown in Figure 3.10 surface and the tensile samples were taken

from the more highly worked material. The details of this sectioning is shown in Figure

3.11. The open circles again represent the 5.08 cm (2 inch) long tensile/creep sample and

the shaded circles represent the oxidation samples.

































876 B70 L BI 247 260
iA
BOTTOM

B121 B105 1.20 134 135 148
B131 163 1 167








Figure 3.10. Tensile and oxidation sample sectioning diagram. Shaded samples are
oxidation and white samples are tensile/creep.


Figure 3.11. Details of sectioning diagram for oxidation (shaded) and tensile (open)
samples from Figure 3.10. Calculated effective strain is also depicted









Test Specimens and Test Procedures


Oxidation

Samples

The oxidation samples were 8.9 cm (3.5 inches) long and 0.635 cm (0.25 inch) in

diameter with a 0.3175 cm (0.125 inch) radius machined on each end (Figure 3.12). The

total surface area of each sample was 17.72 cm2. The specimen blanks were EDMed

oversize from the forging, then centerless ground to the final diameter. The radius was

machined by turning on a lathe; this made complete coverage of the sample easier and

allowed for a simpler calculation of surface area as needed for the oxidation

measurements. More importantly, the rounded ends eliminated any sharp comers which

might act as stress risers for the coating leading to premature failure of the coating during

thermal cycling.

0.250| 3.250
\ 0.250



R.125


Figure 3.12. Drawing of oxidation sample used in this research


Testing

The oxidation testing was conducted in a bottom-loading thermal cycle furnace

built by Ted Kominsky Ovens and Furnaces, Model number KS16000B. A photograph

of the furnace is shown in Figure 3.13. The temperature was controlled with a Honeywell

UDC 3000 controller using input from a type K (NiCr-NiAl) thermocouple. This









thermocouple was calibrated against a NIST standard calibrated thermocouple. The

samples were loaded vertically into a sample fixture made from a cordierite open cell

foam to facilitate rapid cooling as shown in Figure 3.14. Note the thermocouples that

protrude through the hearth plate at the same height as the oxidation pins. One

thermocouple is attached to a the controller and isthe other is attached to a strip chart

recorder that records temperature and number of cycles.


Figure 3.13. Photograph of furnace used to conduct cyclic oxidation tests.









The samples are raised into the furnace and the furnace is automatically started.

The samples remain up in the furnace for 55 minutes, then are lowered and fan cooled for

5 minutes. A typical thermal cycle taken from the strip chart recorder is shown in Figure

3.15. As seen in Figure 3.15, it takes approximately 12 minutes to reach 6500C, and 5

minutes to cool down to approximately 100C. The samples were weighed at 1, 3, 5 and

10 cycles, then every 10 cycles to 100 cycles and then every 100 cycles to 1000 cycles

where the test was terminated and the samples analyzed. The samples were handled

wearing cotton gloves during weighing to prevent contamination from the examiner and

after each weighing the samples were flipped over to ensure that one end was not always

in contact with the fixture. The samples were weighed on a Mettler 200XT digital scale

with an accuracy of 0.1 mg.











Platen




Figure 3.14. Oxidation pins loaded on hearth plate. The samples are the darker pins, the
thermocouples are the two white rods in the foreground. The cordierite block is resting
on the furnace platen.


The samples were then sectioned transverse and longitudinally and mounted in

Conductomet for analysis. Electron microprobe analysis (EMPA) was conducted at









RPI by Dr. David Wark. Microhardness measurements were conducted on a LECO DM-

400FT at a O1gfload and a 20 second load time. The tester was calibrated prior to taking

each set of measurements using the LECO calibration blocks. The data presented is from

one data point at each location; therefore, there will be some experimental error due to

variations in the microstructure or operator error in reading the microhardness

indentation. The results from the microhardness were compared to the oxygen levels

detected by the EMPA, and correlations were made concerning the effectiveness of the

various coatings to prevent oxygen ingress. The EMPA results were also used to

correlate other diffusion phenomena, such as that of the Ni and Fe into the alloy substrate,

to the microhardness results.


700 I --



600


500--


400-


S300-


200-

100 -

0 10 20 30 40 50 60
Time (mins)

Figure 3.15. Typical thermal cycle recorded by strip chart recorder.









Tensile

Samples

The tensile samples are 5.08 cm (2 inches) long and have a gauge length of 1.9 cm

(0.75 inches). The gauge diameter is 0.40 cm (0.16 inches). The sample is gripped by /4-

20 threads with a thread root radius of 0.25 mm (0.01 inches). This sample is shown

schematically in Figure 3.16.

2.000
S1.000
.375




.250-20 threads
2 places

S.160
Figure 3.16. Drawing of tensile sample used in this research. Dimensions are in inches.


Testing

The tensile tests were conducted on an Instron Model 1125 screw driven test stand

using a 5,000 pound load cell. The cross-head speed was 0.05 cm/min (0.02 in/min) and

the sample were held in threaded grips. The strain was measured from the cross-head

displacement, and load vs. time was recorded electronically and on a strip chart recorder.

Percent strain was calculated from the cross-head displacement divided by the initial

gauge length multiplied by 100. The stress was calculated by dividing the load by the

initial cross sectional area. A typical stress-strain curve is shown in Figure 3.17 for

sample B75.















S150










50




0 2 2 .2
0 2 4 6 8
Engineering Strain (%)

Figure 3.17. Typical stress-strain curve generated from tensile testing. Line tangent to
slope of elastic portion of curve used to determine elongation at failure is shown.



Fatigue

Samples

Fatigue samples were sectioned from many different locations in the as-received

forging, as discussed above. One facet of this study was to determine is the location of

the sample from the forging had any effect on the fatigue properties. To do this, the

sample location was recorded and compared to the calculated effective strain from

forging, as shown in Figure 3.18. The samples that were taken from various locations

have been divided into three groups: low (,ff <1.0), medium (1.0< 6,, < 2.0) or high (Ef

> 2.0) strain locations.


















Low Med High
0=.13160 D=1.200 G-2.100
A=.300 E= 1.500 H 2.400
B= .600 F= 1.800 1= 2.700
C= .800 2.957

Figure 3.18. Calculated iso-strain lines from forging showing various effective strain
from forging and respective fatigue sample location.



The effect of cooling rate on microstructure was also discussed above, and the

calculated cooling rate from forging was also superimposed over the diagram of fatigue

samples, as shown in Figure 3.19. The effect of location on the fatigue results was

determined and will be discussed in Chapters 4 and 5.


Low Med High
1=55 E 78 -= 115
J152 F 72 B =107
K=48 G=66 C=96
L=45 H=60 D 87
S=43

Figure 3.19. Calculated cooling rates from 1080C to 8700C in C/min and corresponding
fatigue specimen location.



The fatigue samples were 8.89 cm (3.5 inches) long with a uniform gauge length

of 2.22 cm (0.875 inches). The gauge diameter was 5.08 mm (0.200 inches) with a

surface finish of 8 microinches, and the sample was gripped by /2-20 threads. This









sample is shown schematically in Figure 3.20. In order to avoid stress concentrations at

the surface, the samples were longitudinally polished in the uniform gauge to eliminate

all circumferential grinding marks and to produce all polishing marks parallel to the

loading axis of the sample.

3.500.010
8 --- 1.750.005-






.500-20 UNF 3A
2 PLACES _- .2001.001 DIA

Figure 3.20. Drawing of fatigue sample used in this research. The dimensions are in
inches


Thermal cycling of all fatigue samples was conducted in the Kominsky thermal

cycling furnace shown in Figure 3.13. The samples were arranged horizontally on a

cordierite platen as shown in Figure 3.21. Samples were thermally cycled for 1, 10 and

100 cycles, as well as exposed in static air for 92 hours to compare with the 100 cycle

samples. Samples were also encapsulated in quartz filled with argon. These samples

were placed inside titanium tubes and inserted along with titanium sponge which was

added as an oxygen getter in quartz tubes. The quartz ampoules were evacuated prior to

backfilling and then filled to 200 torr with argon before sealing. On heating to 6500C the

pressure inside the ampoule reached approximately 760 torr.






















Figure 3.21. Photograph of LCF samples loaded for thermal cycling. Samples are
separated by the alumina tubes.


Testing

The fatigue testing was conducted on an MTS 810 servohydraulic test machine

with a 20,000 pound load cell. The testing was conducted under strain control using

resistive strain gauge extensometers (model number MTS 632.53E-14) with high

temperature alumina rods. The A ratio was 1 (,,,, = 0), calculated as

A = o/o, (3.1)




where o, is the stress amplitude defined as

O max O mn
2 (3.2)


and o,,, the mean stress defined as

O max+ O mml
n,=- 2 (3.3)




The cycling rate was 20 cycles per minute. The testing was computer controlled

using Testware SX software that incorporated data collection. Peak and valley load data









were collected electronically for every cycle. Load-strain information was recorded at

various cycles on a strip chart recorder, as shown in Figure 3.22. Hydraulic grips were

used to eliminate the need for a backing nut on the threads thus allowing for tension and

compression loading. The elastic modulus of each sample was determined by loading the

sample in the elastic region and recording the resultant strain on a strip chart recorder, as

shown in Figure 3.22. The applied load was divided by the sample cross sectional area in

the uniform gauge to determine the applied stress. The calculated stress was then divided

by the measured elastic strain in order to estimate the room temperature elastic modulus.

This room temperature modulus was multiplied by the applied strain during testing in

order to give the average, or pseudo-stress applied during the fatigue testing. This

pseudo-stress is typically reported as the alternating pseudo-stress, which is the pseudo-

stress divided by two.

Analysis

After fatigue testing, two cross-sections were cut from the sample, one containing

the fracture surface and one behind the fracture surface, as shown in Figure 3.23. The

cross section containing the fracture surface was sectioned longitudinally to observe the

crack path, the initiation site and opposite the initiation site on the longitudinal section.

These samples were mounted in Conductomet such that the initiation site in the

longitudinal section was on the same side as the cross section as shown in Figure 3.23.

This allowed the initiation site to be identified during metallography.










Specimen #52, Strain Control, 0.95%E, 20CPM, R.T.


3000 .


S 2000

1000


/ i 0.004


Strain (in/in)

Figure 3.22. Typical LCF data taken during testing showing the hysteresis loops at
various cycles. Curve for modulus determination is also shown.


Figure 3.23. Illustration of sectioning and metallographic mounting of the fatigue sample
for evaluation.


The samples were then examined in the SEM using backscatter analysis or on the


optical metallograph after etching.









Coatings


All of the samples to be coated were ultrasonically cleaned in a warm Alconox

bath for 30 minutes, followed by a de-ionized water rinse and a 2-propanol rinse. The

samples were dried with a K ,.i- .pi;' i:. 'cl and placed in sterile plastic bags, and only

handled with gloves prior to coating.

Low Pressure Plasma Spray (LPPS) Coatings

For LPPS, the oxidation pins were secured in a custom designed fixture that holds

the pin at each end and thus allows the complete pin to be coated instead of just half of

the pin. This cuts the processing time in half and eliminates any overspray in the center

of the pin. The fixture is shown in Figure 3.24. After the sample was loaded into the

fixture, the fixture was loaded into the LPPS chamber. The chamber was then pumped

down to 50 torr and backfilled with argon. The sample oscillation and rotation were

initiated and the plasma guns were turned on to pre-heat the sample to approximately

650'C. The powder feeders were then started and the sample was coated to the proper

thickness. The plasma guns and feeders were then shut off and the sample was allowed

to cool in the chamber. The sample was then removed from the single pin fixture and

placed in a fixture with several other samples where only the end radii are exposed.

These radii were also coated to insure that the weight gain versus unit area measured

during oxidation was as accurate as possible.

















































Figure 3.24. Drawing ofLPPS specimen fixture.


NiCrAlY + 40 vol% A1203.

NiCrAIY (Ni-21.7Cr-10AI-1.13Y (wt%)) powder (purchased from Praxair) was

low pressure plasma sprayed (LLPS) onto a scrap oxidation pin to assess the LPPS









process. The NiCrAIY broke away from the pin in three large sections after processing.

Since it was initially suspected that the fixture might be the cause of the lack of adherence

another trial was attempted, this time holding the pin on one half only. The result was the

same; the coating broke away from the pin after processing. The pin was sectioned in the

location of the coating failure and examined using SEM backscatter techniques. It was

determined that there was residual Ni left on the surface of the pin, indicating that the

coating was initially adherent but subsequently spalled. This spallation was probably a

result of the large thermal expansion differences between the substrate and the coating.

Consequently, a mixture of 60 vol% NiCrAlY and 40 vol% A1203 was produced by

blending, and this "cermet" was deposited on the substrate. This mixture provided a

suitable thermal expansion match between the coating and substrate and, as a

consequence, the coating was adherent.

FeCrAlY.

FeCrAIY (Fe-29.9Cr-4.9A1-0.6Y (wt%)) powder was also applied to the oxidation

pins via LPPS. Since this coating was successfully applied and did not spall, no changes

were made to the deposition process.



Sputtered Coatings.

Several elemental coatings were deposited onto oxidation pins via vacuum

sputtering. The samples were held on one end and rotated in the sputtering chamber via a

magnetic feed through. The elements were either reacted with the substrate at elevated









temperature to create a stable compound at the surface(e.g., A13Ti) or were used in their

pure elemental form.

Sputtered Al.

25 pm of Al was sputtered onto the oxidation pins in a vacuum sputtering unit.

These pins were then reacted in dry, pure argon using the following heat treatment

schedule:

Ramp to 6000C at 3000C/hour,

Hold at 6000C for 2 hours,

Ramp to 6300C at 100C/hour,

Hold at 6300C for 10 hours,

Furnace cool.

X-ray diffraction confirmed that a mixture of A13Ti and A13Nb were formed at the

surface, as shown in Figure 3.25.


Figure 3.25. SEM backscatter micrograph of Al coated and reacted sample showing the
in-situ formed A13Ti coating on the right.









Sputtered Si

Elemental Si was also sputtered onto the orthorhombic Ti-aluminide substrate in a

vacuum sputtering unit. The sputtering rate was very low and, as a consequence, only

12.7 pm of Si was deposited. The Si coated pins were reacted as follows:

Ramp to 12000C,

Hold at 12000C for 16 hours,

Furnace cool.

This cycle was designed to produce the TisSi, phase on the surface and XRD

indicated that this phase was indeed produced, as shown in Figure 3.26. The reaction

temperature was above the p transus for this alloy, and the microstructure indicated that

the prior 3 grains had indeed grown and the as forged microstructure was eliminated.




















Figure 3.26. SEM backscatter micrograph of Si coated and reacted sample showing in-
situ formed TisSi3 coating on the right.









Sputtered Pt and Cr

Pt and Cr was sputtered onto the oxidation pins in order to provide protection

based on the low oxidation rates of pure Pt, and the stability with titanium [Fuj79, Eyl85],

and because Cr has been show to provide good oxidation resistance as a coating for

titanium [McK93, McK90]. Both coatings were adherent, although the Pt coating was

porous and columnar, as shown in Figure 3.27.










P. o acting .. ...
*1; i. .... ^ .






Figure 3.27. SEM backscatter micrograph of Pt coating showing the columnar structure
and porous nature of the coating.


CVD coatings

Several oxide coatings were produced by metal organic chemical vapor deposition

(MOCVD). The idea was to evaluate the stability of the oxide in contact with the

substrate and determine if the Ti would reduce the oxide. MOCVD was selected because

of the capability of producing adherent coatings of oxides on any and every surface in the

reactor. The CVD process, unlike PVD or LPPS, is not line-of-sight, and therefore every










hot surface in the reactor will be coated. The MOCVD process consists of a metal

organic precursor that is evaporated and flowed over the substrate held in a reaction

chamber at elevated temperature, as shown in Figure 3.28 and Figure 3.29. The selection

of the precursors and the deposition temperatures are described below.



















Figure 3.28. Photograph of CVD reactor.



/ Pressure
Gauge

Furnace

Samples



Furnace -






To Vacuum Matl
Pump Liquid Mantle
Nitrnn-n Heater


Figure 3.29. Schematic diagram of CVD reactor shown in Figure 3.28.









SiO2

Oxide and silicide coatings have been reported to provide oxidation protection for

ca-p titanium [Sol85, Cla88, Bed91], r2 alloys [Wei89, Dev90] and orthorhombic alloys

[Coc96]. The mechanism is through the production of an oxide comprised mainly of

SiO,. For this reason, SiO2 was chosen as a diffusion layer for an outer coating, and

possibly as a primary oxidation coating. The reagent used to produce the SiO, was

silicon tetraethoxide [Si(OC2H,)4]. The reactor temperature was 6700C at a pressure of

500pim. The Effusion cell temperature was 900C and the flow rate of reagent was 2x10-5

gs 'cm-2. The deposition rate of the SiO, was approximately O.lnm/sec with a reagent

utilization of approximately 0.5%. The reaction taking place was

SI II i I, i SiO2+ CHsOH+C2H4

Tao,

Tantala was chosen as a barrier layer due to its higher CTE over SiO, and greater

oxidation resistance relative to TiO,. The reagent used to produce the Ta20s was tantalum

ethoxide dimer [Ta,(OC,H,) ,,]. The reactor temperature was 4250C at a pressure of

100m. The Effusion cell temperature was 1250C and the reagent flow rate was 3x10'-

gs 'cm-2. These conditions led to a deposition rate of the Ta,O, of approximately

0.lnm/sec with a reagent utilization of approximately 0.5%. A coating thickness of 0.45

tm was typically deposited on the oxidation pins and fatigue samples. The reaction

taking place was

Ta2(OC.H,),, = TaO, + 5 CZHsOH + 5 CH4









MgO

MgO has a high oxygen diffusivity in air at 6500C and therefore it probably would

not provide adequate protection as a stand alone oxidation coating. But MgO has been

evaluated as a barrier layer for fiber-reinforced titanium aluminide composites [McG95].

It was shown to dissolve oxygen into Ti-24A1-11Nb after 100 hours at 1000C, but at a

reasonably low rate. MgO has also been formed in-situ in Ti-Mg alloys produced by high

rate evaporation and quenching. After oxidation at 850C for only 10 minutes, MgO

particles were formed in the titanium matrix [War95]. For this reason and because MgO

has a higher thermal expansion coefficient than the Ti-22A1-26Nb alloy, it was chosen as

a candidate coating system.

The reagent used to form MgO was magnesium 2,4 pentane dionate and the

chemical reaction was

Mg (C,HO,), HO-, MgO + 4 CHsOH + CH4

The reaction temperature was 540C at a partial pressure of 0.5 torr. Oxygen was

the initial carrier gas and was chosen to reduce the level of carbon in the deposited film.

After deposition, the film thickness was 0.85 tim. The samples were baked out in an air

furnace for 1/2 hour at 575C, and the film thickness decreased to 0.3 pm. This is

probably due to evolution of carbon into carbon dioxide in the furnace. The source of

carbon is from the reaction of the reagent in the furnace and the incomplete formation of

the reaction products. This process is still in the lab stages and is not yet fully reduced to

practice.















CHAPTER 4
RESULTS AND DISCUSSION

Oxidation


Cyclic Oxidation Testing

Baseline results

Four baseline samples were cyclically oxidized at 6500C. Three of the samples

(136, 187 and 191) were chosen to have different levels of strain from forging, and thus

different grain size and grain aspect ratios, as discussed in Chapter 3. These samples

were chosen to determine if the strain history, which results in a slight variation in

microstructure, has any effect on the oxidation behavior. The results from these baseline

studies are shown in Figure 4.1. The difference in oxidation rates between these three

samples is small as is the weight change over time. Therefore, it can be stated that there

is little or no effect on oxidation from the strain history. The fourth sample, 174, was run

in a separate oxidation test with another series of samples as a control. The differences

observed between the samples indicates typical experimental scatter.























--&- 136
-- 187
-- 191
- 174
+-Average


0 200 400 600 800 1000
Time (hrs)

Figure 4.1. Baseline oxidation results showing no variation between three samples from
different locations run together, and a larger variation with a sample from a different
experimental run.



Al and Si Coated and Reacted Results


The results from the oxidation testing on Al and Si coated and reacted samples are

shown in Figure 4.2. The average of the baseline samples is also plotted as a reference.

As can be seen, both the Al and Si coated and reacted samples behaved well in cyclic

oxidation tests. This behavior has been observed previously for Ti coated with AIlTi

[81Str, 85Unn, 91Abd, 91McM, 92Gal, 93Abb, 93Wie] and a, coated with Al3Ti [88Sub,

90Kun, 90Smi, 93Smi]. Silicide coated Ti-22A1-27Nb has also been evaluated and


~t

.-fi


X,


- -








shown to have similar oxidation behavior as the Si coated and reacted Ti-21A1-26Nb

[96Coc].


SAl
-E- Si
--- Baseline


0 200 400 600 800
Cycles (hrs)
55 mms @ 650C, 5 mms @ R T
Figure 4.2. Al and Si coated and reacted oxidation results showing the decrease in
oxidation rate over the baseline.


Oxide Coating Results

The effect of cyclic oxidation on the oxide coated samples are shown in Figure

4.3. Both Ta20, and MgO behave similarly to the baseline alloy while the SiO, coated

sample exhibited better oxidation behavior than the baseline at short times, but began to

gain weight in a linear fashion with time after 400 cycles. This indicates that the coating

was not protective and that oxygen was diffusing through after 400 hours causing the


1000


o-
'Ir


I










sample to gain weight. These results indicate that the oxide coatings by themselves do

not provide protection from an oxidizing environment.


1.2-~~--..--------
-0- SiO2
a205
S + MgO
Baseline


S0.8 v/















0 200 400 600 800 1000
+

I 0.6'












0 200 400 600 800 1000
Cycles (hrs)
55 mins @ 650'C, 5 mins @ R.T.

Figure 4.3. Oxide coated oxidation results showing that TaO, and MgO coated material
behaves like uncoated material and SiO2 coated samples begin to gain weight after 400
cycles.



Metallic Coating Results


The results from the cyclic oxidation testing of the metallic coated samples are

shown in Figure 4.4. The results show that the MCrAlY coatings provide excellent

oxidation resistance, that Cr provides intermediate protection, and that Pt has little effect

over baseline behavior. The protection by the MCrAlY coatings has been shown in

previous work on Ct alloys [92Bri, 93McK, 95Sch] and titanium alloys [92Tob, 90Lut,









90McC, 93McK]. On the other hand, platinum, in the form of PtAl2, has been shown to

reduce the cyclic oxidation rates in conventional titanium alloys and in Ti3AI [93Nic].

Pure platinum coatings have also been shown to improve creep resistance [85Eyl] and

decrease oxidation rate [79Fuj] of Ti-6242. Finally, pure Cr coatings have also been

shown to reduce the weight gain over time of Ti-64 and Ti-6242 [90McK].

1.2 -
-NiCrAIY+AI203
-A- FeCrAIY
1- -Pt
--C--- Cr Bs
---- Baseline

le0.8 -V

& \

en~
0.6
U


0.4


0.2-- -- -


0 200 400 600 800 1000
Cycles (hrs)
55 mins @ 650aC, 5 mins @ R T

Figure 4.4. Results showing the large reduction in oxidation rate with the application of
the MCrAlY coatings and the effect of Cr and Pt coatings.


The results from all of the cyclic oxidation tests are compared in Figure 4.5. As

can be seen, the metallic coatings, excluding Pt, provide excellent oxidation resistance










after 1000 cycles, whereas the Ta20O and MgO coatings behave similarly to the baseline,

and the SiO, coated samples begin to degrade after 400 hours.


1.2



1



- 0.8



S
. 0.6



0.4


- Al
-- Si
-o S10
X- Ta20
+ MgO
N-- NiCrAIY+AlO0
-A- FeCrAIY
-- Pt
- Cr
-- Baseline


0 200 400 600 800 1000
Cycles (hrs)
55 mms 650C, 5 mins @ R.T

Figure 4.5. Comparison of the oxidation rate behavior of all coatings examined in this
study.


Microhardness and Microprobe Evaluation


Baseline uncoated


As shown in Figure 4.1, the three samples with varying forging strains behaved

similarly, so after 500 cycles samples 191 was removed for analysis. Both microhardness

and oxygen content vs. depth are plotted in Figure 4.6. Notice that in both curves, the

oxygen content and microhardness follow the same trend. Further work is needed to









determine if the high value of oxygen in the 1000 cycle sample at 25gm is scatter or an

indication of localized oxidation, possibly an oxide.

25 --- 1100

SOxygen (at%) 1000 cycles
SOxygen (at%) 500 cycles 1000
20-A- O p hardness (KHN) 1000 cycles
'- hardness (KHN) 500 cycles 900


800
15 800

nt -700

O ,10 \
St 600

500
5-
400

0 300
0 10 20 30 40 50 60
Distance from Coating/Substrate Interface (gim)

Figure 4.6. Uncoated baseline oxygen content and microhardness vs. depth.



The backscatter SEM images of the two microhardness samples (Figure 4.7 and

Figure 4.8) indicate that the dark zone near the surface of the sample is high in oxygen.

The microhardness indentations visible in these figures clearly show that these regions

are much harder that the base material. The depths of oxygen rich zone in the 500 and

1000 cycle samples are approximately 20pm and 35pm respectively. Dividing the square

of the depth (in cm) by the time in seconds, it is possible to estimate the diffusivity of

oxygen into the O + B2 lattice at 6500C as:

x = (cm) [4.1]































Figure 4.7. SEM backscatter image of baseline microhardness sample cycled for 500
cycles.
























Figure 4.8. SEM backscatter image of baseline microhardness sample cycled for 1000
cycles.
cycles.









or

X2
D =- (cm2/sec) [4.2]
t

This yields a D of approximately 3.4 x 10-12 cm2/sec. This estimate ignores the oxide

present on the surface of the sample as well as any interaction that oxide has with the

substrate, and the fact that these are no isothermal experiments. These are a valid

assumptions given the fact that the oxide has a diffusivity of oxygen that is 50x that of the

metal [Unn86]. The diffusion coefficient of oxygen for CP cL-Ti at 6500C is

approximately 2.5 x 10-12 cm2/sec [Kub83] and for Ti-24Al-15 Nb at 10270C is

approximately 2.6 x 10-12 cm2/sec [Roy96]. These values correlate well to the 3.4 x 10-

12 cm2/sec value calculated from the baseline samples.

FeCrAIY and NiCrAlY + A1203

Brindley et al. [Bri92] have evaluated the oxidation behavior of both of these

coatings, and found that while they provide excellent resistance to oxygen ingress into

titanium aluminides, the elements in the coatings themselves diffuse into the substrate

and cause embrittlement. Figure 4.9 is an SEM backscatter image of the NiCrAlY +

A1l03 coated specimen after 1000 cycles. The NiCrAlY + A1203 plasma sprayed coating

can be seen in the lower right hand comer. Notice that the microstructure of the matrix

next to the coating has changed from the lath morphology to a fine, equiaxed structure.

Work conducted by Schaeffer et al. [Sch95] on U2 coated with NiCrAIY + A1203

demonstrated that the coating alone, without exposure, degraded the LCF life by two


orders of magnitude.

























Figure 4.9. SEM backscatter image of the NiCrAIY + A1203 coated sample after 1000
cycles at 6500C.


The results from this study support Schaeffer's results, as can be seen from the

microhardness and oxygen profiles in Figure 4.10 through Figure 4.12. Figure 4.10 and

Figure 4.12 compare the oxygen level in the substrate to the measured microhardness.

Notice that there is simply one point under the coating that contains oxygen, and beyond

that point the level is below the detectability limit of the EMPA. Figure 4.11 and Figure

4.13 show the Fe and Ni concentration vs. microhardness. The Fe and Ni content seems

to correlate with microhardness, and the Ni and Fe levels both taper off with distance

from the interface, as does the microhardness. This could contribute to a loss in ductility

and thus fatigue life. Recall that LPPS is conducted in a partial pressure of argon and air,

and the sample is preheated to 650C for several minutes prior to coating during which

oxygen could ingress. A diffusion barrier between the substrate and coating might

possibly control the ingress of Ni or Fe from the coating into the substrate, but the pre-

heat might still cause embrittlement.














10 Oxygen (at%) 650
hardness (KHN)
600
8

o :550
S6
4500
4
0 450 ,

2
400

0 350


-2 -' 300
0 10 20 30 40 50 60 70
Distance from Coating/Substrate Interface (pm)

Figure 4.10. Microhardness and oxygen vs. depth for the FeCrAlY coated and 1000
cycle exposed sample.


Fe (at%)
hardness (KHN)


S02


S015


550


500


450


0 10 20 30 40 50 60 70
Distance from C (,.,Iri Su.,hiraji Interface (pm)

Figure 4.11. Microhardness and iron vs. depth for the FeCrAlY coated and 1000 cycle
exposed sample.













Oxygen (at%)

hardness (KHN)


2.5


2


S 1.5


0


500

C-
450


400


0 10 20 30 40 50 60 70
Distance from Coating/Substrate Interface (pm)

Figure 4.12. Microhardness and oxygen vs. depth for the NiCrAIY + Al203 coated and
1000 cycle exposed sample.


NI (at%)

hardness (KHN)


02


500
C-

450


400


-0,1 .1 .. 300
0 10 20 30 40 50 60 70
Distance from Coating/Substrate Interface (pm)

Figure 4.13. Microhardness and Ni vs. depth for the NiCrAIY + A1203 coated and 1000
cycle exposed sample.










Sputtered and Reacted Al and Si


Elemental Al and Si were sputtered onto the orthorhombic under vacuum. The Al

coated samples were then ramped to 6000C in two hours and held for two hours followed

by ramping to 6300C in 3 hours and holding for 10 hours. This produced A13(Ti,Nb) as a

surface reacted coating, as verified by XRD. The Si coated samples were ramped to

12000C as quickly as possible and held for 16 hours. The intent was to produce the TiSi3

phase and miss the TiSi phase. This was accomplished as can be seen in Figure 4.17,

and also verified by XRD. The microhardness profiles vs. the oxygen content are shown

in Figure 4.14 for the Al coated sample and in Figure 4.16 for the Si coated sample.

Figure 4.15 and Figure 4.17 are SEM backscatter images of the Al and Si coated and

reacted samples, respectively, obtained after cycling at 650'C for 1000 cycles.

025 -- 800

Oxygen (at%)
0.2
hardness (KHN) 700

0.15


0 5

O 500
0.05


0 400



-0.05 -- -- 300
-5 5 15 25 35 45 55 65
Distance from Coating/Substrate Interface (pm)

Figure 4.14. Microhardness vs. oxygen content for the Al coated and reacted sample.
Note that the distance from the interface starts at -5tm.





















i I E




Figure 4.15. SEM backscatter image of Al coated and reacted sample after 1000 cycles at
6500C. Notice the crack in the AlTi coating.


Notice that the A13(Ti,Nb) coating has cracks that run to the substrate, yet the

coating behaved very well in oxidation. The SEM backscatter images do not show any

indications that oxygen diffused preferentially down the cracks and into the substrate.

This could be due to the fact that on heating, the coating expanded more than the

substrate and thus the cracks were not open during exposure. Microhardness

measurements were taken in the reacted layer because in optical microscopy the layer

looked like the substrate. Therefore, the oxygen analysis and the microhardness start at

minus 5tm in both the Al and Si coated cases. The zero point is the location of the

coating/substrate interface after the reaction. This indicates that the coating was the

major source of oxidation and hardness. Directly beneath the reacted coating, the oxygen

content was negligible and the microhardness had returned to that of the substrate. But

since the coating was atomistically attached to the substrate, any cracks in the coating

would propagate into the substrate and cause limited LCF life.











0.14-


0.12-


0.1 -


0.08-
--

0.06-
S0.04
0 0.04-


Oxygen (at%)

- hardness (KHN)


900


800


700
(5
600


500


002 400


0 300


-0.02 -. 200
-5 5 15 25 35 45 55 65
Distance from Coating/Substrate Interface (im)

Figure 4.16. Microhardness vs. oxygen content for the Si coated and reacted sample.
Notice that the distance from substrate starts at -5pm.


, i '- "-'r*''i.




1 1- 1 f '.

*


Figure 4.17. SEM backscatter image of the Si coated and reacted sample after 1000
cycles.









Sputtered Pt and Cr coatings

Pt was sputtered onto the orthorhombic substrate to a thickness of about 1 pm.

The Pt coating (Figure 4.18) was columnar, porous and discontinuous. As a result, it

allowed oxygen to diffuse to a similar depth as in the uncoated material. The

microhardness and oxygen profiles (Figure 4.19) are consistent with the poor behavior of

the coating in oxidation. Had the coating been dense and adherent, it should not have

allowed oxygen to pass at 6500C according to previous work conducted on Pt coated Ti-

6242 [Fuj79] where the rate of weight gain was reduced from 3x10-1 to 2x104 at 5930C.

Their coating, deposited by ion plating, was approximately 1 Im thick and was dense and

continuous. Sputtered Cr exhibited better oxidation behavior than Pt during cyclic

oxidation, as shown in Figure 4.4. The level of oxygen beneath the coating after

exposure is shown in Figure 4.20, along with the level of Cr. Oxygen content for the

unexposed sample was zero as determined by EMPA. There is essentially no Cr diffused

into the matrix after 1000 cycles at 6500C, and very little oxygen, compared to the

baseline of 20 at%.





















Pt coating

*hA.^ fc:--`--; .-<, ..j '


Figure 4.18. SEM backscatter micrograph of Pt sputter coated sample cycled for 1000
cycles.


25


Oxygen (at%)
hardness (KHN)


700 _

0-
600


500


0 10 20 30 40 50 60
Distance from Coating/Substrate Interface (I'm)

Figure 4.19. Microhardness and oxygen content vs. depth for the Pt coated and 1000
cycle exposed sample.










2-
S-- Oxygen
S- Cr


1.5-





1-





0.5-



0 Im n m 1


0 5 10 15 20 25
Depth (pim)
Figure 4.20. Plot of Cr and oxygen content vs. depth in 1000 cycle exposed oxidation
sample.



CVD Oxide Coatings


The CVD SiO2 coating was dense, adherent, and appeared iridescent green in

color. The Ta20, and MgO coatings were also dense and adherent. All coatings were

approximately 0.5 im thick.

As can be seen from Figure 4.3, the initial oxidation behavior of the SiO2 coating

was excellent. This data is an average of three separate tests conducted with the SiO,

coating. However, after 400 cycles, the oxidation behavior of the SiO2 coated samples

began to behave in a more linear relationship with time instead of parabolic. This is

typically an indication that there is no protective oxide present and that oxygen is simply




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EPLB20GAP_09T53Y INGEST_TIME 2013-11-16T01:44:02Z PACKAGE AA00014249_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES