Citation
The Impact of Carbon on Single Crystal Nickel-Base Superalloys

Material Information

Title:
The Impact of Carbon on Single Crystal Nickel-Base Superalloys Carbide Behavior and Alloy Performance
Creator:
Wasson, Andrew
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (260 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Materials Science and Engineering
Committee Chair:
Fuchs, Gerhard E.
Committee Members:
Dempere, Luisa A.
Dehoff, Robert T.
Holloway, Paul H.
Subhash, Ghatu
Graduation Date:
8/7/2010

Subjects

Subjects / Keywords:
Alloys ( jstor )
Carbides ( jstor )
Crack initiation ( jstor )
Fatigue ( jstor )
Heat resistant alloys ( jstor )
Heat treatment ( jstor )
Hypertension ( jstor )
Materials science ( jstor )
Porosity ( jstor )
Specimens ( jstor )
Materials Science and Engineering -- Dissertations, Academic -- UF
carbide, carbon, crack, creep, crystal, fatigue, initiation, mechanical, modification, morphology, single, superalloy
Genre:
Electronic Thesis or Dissertation
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
Materials Science and Engineering thesis, Ph.D.

Notes

Abstract:
Advanced single crystal nickel-base superalloys are prone to the formation of casting grain defects, which hinders their practical implementation in large gas turbine components. Additions of carbon (C) have recently been identified as a means of reducing grain defects, but the full impact of C on single crystal superalloy behavior is not entirely understood. A study was conducted to determine the effects of C and other minor elemental additions on the behavior of CMSX-4, a commercially relevant 2nd generation single crystal superalloy. Baseline CMSX-4 and three alloy modifications (CMSX-4 + 0.05 wt. % C, CMSX-4 + 0.05 wt. % C and 68 ppm boron (B), and CMSX-4 + 0.05 wt. % C and 23 ppm nitrogen (N)) were heat treated before being tested in high temperature creep and high cycle fatigue (HCF). Select samples were subjected to long term thermal exposure (1000 degrees C/1000 hrs) to assess microstructural stability. The C modifications resulted in significant differences in microstructure and alloy performance as compared to the baseline. These variations were generally attributed to the behavior of carbide phases in the alloy modifications. The C modification and the C+B modification, which both exhibited script carbide networks, were 25% more effective than the C+N modification (small blocky carbides) and 10% more effective than the baseline at preventing grain defects in cast bars. All C-modified alloys exhibited reduced as-cast gamma/gamma prime eutectic and increased casting porosity as compared to baseline CMSX-4. The higher levels of porosity (volume fractions 0.002 - 0.005 greater than the baseline) were attributed to carbides blocking molten fluid flow during the final stages of solidification. Although the minor additions resulted in reduced solidus temperature by up to 16 degrees C, all alloys were successfully heat treated without incipient melting by modifying commercial heat treatment schedules. In the B-containing alloy, heat treatment resulted in the transformation of script MC (M ? metal, C ? carbon) carbide networks into clusters of small, spherical MC carbides without a significant change in composition. Formation of topologically close packed phases during thermal exposure was suppressed in the B-containing alloy due the decomposition of primary MC carbides and the preferential formation of secondary M23C6 carbides. All of the modified alloys exhibited shorter creep rupture lifetimes than the baseline at all creep conditions (950 degrees C/300 MPa, 850 degrees C/550 MPa, 750 degrees C/800 MPa). The most significant decrease in lifetime occurred at the 750 degrees C condition due to large primary creep strains of up to nearly 10% in the C-containing alloys. In HCF testing at 850 degrees C, the presence of carbides and increased porosity led to reduced lifetimes in the modified alloys. HIP (hot isostatic pressing) processing significantly improved fatigue performance, accounting for average lifetime increases ranging from 77% to 4490% over Un-HIPed material. HIP also isolated the effect of carbide morphology on fatigue behavior by changing active crack initiation sites from pores to carbides. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2010.
Local:
Adviser: Fuchs, Gerhard E.
Statement of Responsibility:
by Andrew Wasson.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
10/9/2010
Resource Identifier:
004979661 ( ALEPH )
769019262 ( OCLC )
Classification:
LD1780 2010 ( lcc )

Downloads

This item has the following downloads:


Full Text






A 002-1: OnU Rtl M~ :HT ~ ...... I'I V :.-
II ml -


r .,r, rl ,l1 r o. r. i (.i *- STl I M I'., l" 4,-' C0 1 L
, tri pirt *_\i t.a l<: I 'i if I r1 'norei r r ,t'- J .. .,.1 ,0


C 043-1: I..A .C.. HTsH.. T. 110.54.
1n


B 022-2: t.. HIm.e..S T.


Intiation atlpore wrth carid* Stage ll I Iack c rgrowt*
ertload ,,, .


IsxU


D 061-2: sm..P...cM. NHTHT. IS M.t 7A.s.fel,


Initiation aporewtl cartid@* Stage Icrackgrowth* .... !". Initiation atpom wil card* Stage IIcrackgrowth* "
carload .... .. .. overload
Figure H-2. Summaries of HCF fracture surfaces from "Round 2" HCF tests, Un-HIPed,
HT SHT specimens tested at 15 Hz. A) Baseline, B) C modification, C) C+B
modification, D) C+N modification.


240


LI
%^^^^^^!t









(Figure 5-24B). The deformation appeared to "flow" around the carbides, and some of

the highly deformed regions separated from carbides. The absence of carbide cracking

in the longitudinal sections very near the fracture surface suggests that the cracking of

carbides visible on the fracture surfaces occurred during final fracture.

5.4.4 Summary

Creep tests at 750 C/800 MPa resulted in the largest differences in lifetime

between baseline CMSX-4 and the C-modified alloys among the three creep testing

conditions. The C+B modification, in particular, showed significantly shorter rupture

lifetimes and exhibited much larger primary creep strains than the other variants.

Microstructural examination revealed less cracking at pores and carbides and more

deformation of the y/y' matrix than at higher test temperatures. The results here

highlight the importance of testing at multiple conditions. The B-containing alloy

exhibited very similar creep deformation to the baseline at 950 OC, but a decrease in

temperature of 200 OC, coupled with an increase in applied stress, led to dramatically

dissimilar behavior between the alloys.

5.5 Analysis

The following sections cover several studies designed to compliment creep results

and evaluate possible explanations for the observed behavior. These studies sought to

connect creep deformation processes to sample observations. The efforts included

measurements from tested specimens and compositional data from metallographic

sections.

5.5.1 Sizes of Pores and Cavities

Casting pores have been reported to grow during both heat treatment [55,127] and

creep deformation [52]. Direct, quantitative size comparisons were made between


128









This creates discontinuities in the microstructure that can be detrimental to mechanical

performance [102]. Slow ramp rates and hold times at intermediate temperatures allow

for diffusion and the reduction of segregation during the SHT to reduce the likelihood of

incipient melting. The heat treatments used in this study were modifications of a

commonly used, eight-stage commercial SHT for CMSX-4, seen in Figure 3-2

[103,104]. Two separate treatments, shown in Table 3-3, were developed: a "low"

temperature (LT) SHT with a maximum temperature of 1310 C and a "high"

temperature (HT) SHT with a maximum temperature of 1325 C. The LT treatment was

designed to obtain good homogenization of the alloy without strongly affecting carbide

morphologies, and the HT treatment aimed to break down large, script carbide networks

into smaller features. Work with the polycrystalline superalloy M963 found that

increasing the SHT temperature led to increased decomposition of primary MC carbides

[61]. The majority of mechanical test specimens underwent the high temperature SHT.

The high temperatures and precise temperature control required to conduct a SHT

necessitated the use of an Elatec vacuum furnace system that is capable of

temperatures up to 1400 C and vacuum levels lower than 10-4 Torr (0.133 Pa). The

temperature of the furnace was controlled by a Honeywell controller to within

approximately 1.5 C of the setpoint by two thermocouples lowered very close to the

surface of the test bars. A third thermocouple was used as a survey that monitors the

front end of the furnace hot zone. All three thermocouples were type C and consisted of

tungsten and rhenium sheathed in individual molybdenum jackets. Cooling was done

through the injection of ultra high purity helium and the forcing of the gas over the bar

samples with a fan to produce cooling rates greater than 250 OC/minute. The high


















J W W
Ti TJ T.8 W W
Ti W TJ T

A|1o Energy (keV) ,
Figure 4-21. Small, rounded carbide particles in fully heat treated condition of the
carbon and boron (C+B) modification. A) SEM micrograph of deep etched
sample, B) EDS spectrum from one of the small carbides indicating carbide
composition did not change despite morphology transformation.




B








1 pm2 4 1 12 14

Figure 4-22. Small, tantalum (Ta) rich carbide in the fully heat treated condition. A) TEM
micrograph (STEM mode), B) EDX spectrum from indicated point. The Cu
peak is from the TEM specimen grid.

Table 4-5. Average spot wavelength dispersive spectroscopy (WDS) compositional (wt.
%) measurements of carbides in the fully heat treated (HT SHT + two-step
age) condition.
C modification C+B modification C+N modification
Ta 76.12 75.96 65.43
C 10.41 9.16
Al 0.003 0.003 0.61
Co 0.26 0.29 2.45
Cr 0.28 0.47 1.17
W 8.90 10.55 7.16
Ti 0.31 0.14 0.36
Ni 2.84 3.11 6.49
Hf 0.87 0.32 16.30


110









7.2.1 Effect of Carbon and Minor Additions on Heat Treatability

The reduction of alloy solidus temperature is a primary concern when adding

"grain boundary" elements such as C, B, N, Hf, and Zr to SX superalloys. Even small

additions of these elements can significantly depress the melting point and increase the

risk of incipient melting during heat treatment. Homogenizing temperatures are often

lowered to avoid the possibility of melting, and this can result in incomplete solutioning

of the y' phase. The reduction of solidus temperatures removes flexibility from heat

treatment schedules by narrowing the heat treatment window. Liu and colleagues found

that additions of 0.05 wt. % and 0.1 wt. % C to a 1st generation SX superalloy lowered

the incipient melting point to a level that prevented complete solutioning of the y' phase

[46]. The inability to fully homogenize these modified alloys led to reduced creep

strength due to non-uniform y' precipitates [81]. Difficulties in homogenizing without

melting have caused some alloy developers to remove the SHT step entirely from heat

treatment schedules for SX alloys containing grain boundary elements. The standard

heat treatment for CMSX-486, a commercial SX superalloy with 0.07 wt. % C, 0.017 wt.

% B, 1.2 wt. % Hf, and 0.005 wt. % Zr, involves applying a two-step age heat treatment

to the alloy in the as-cast condition [12,29]. While this may be a relatively simple and

low cost heat treatment, the ability to homogenize before forming the strengthening

precipitates would likely lead to improved properties.

Using modified variations of the standard commercial heat treatment for CMSX-4,

the modified CMSX-4 alloys were successfully solutioned and aged to produce the

desired y/y' microstructures for optimum mechanical performance. Although these

alloys contained fewer grain boundary elemental additions than CMSX-486, this

represents a significant step in the heat treatment of C-modified SX superalloys. The


190









over Un-HIPed material. HIP also isolated the effect of carbide morphology on fatigue

behavior by changing active crack initiation sites from pores to carbides.









Table 6-1. Predicted minimum cyclic lifetimes from fracture mechanics approach and
corresponding actual fatigue lifetimes for all HCF test specimens. Actual
results that are lower than estimations are shown in red italics.
Material Condition Alloy Modification Estimated Minimum N Actual N
(cycles) (cycles)
Baseline 137625 520425
Baseline 181413 295457
Baseline 146663 327785
C modification 587976 138712
C modification 131037 78729
Un-HIPd HT C modification 100755 61825
Un-HIPed HT SHT
C+B modification 358278 674213
C+B modification 34887 110549
C+B modification 33013 52880
C+N modification 407058 233287
C+N modification 46094 75426
C+N modification 124713 56574
Baseline 858624 6993205
Baseline 543713 742208
Baseline 1726761 8797800
C modification 92861 166089
C modification 112008 176144
C modification 109539 151054
HIPed HT SHT
C+B modification 5021227 1685075
C+B modification 111854 872343
C+B modification 339802 1338727
C+N modification 1575937 4504625
C+N modification 1342884 5947002
C+N modification 1062095 6315833
Baseline 341765 291575
Baseline 153999 378911
C modification 54289 91324
HIPed LT SHT C modification 70091 110432
C+B modification 102664 160733
C+B modification 100936 168957
C+N modification 96160 82946


181









LIST OF ABBREVIATIONS


Terms and Phrases

AC Air cooling

AES Auger electron spectroscopy

AFRL Air Force Research Laboratory

APB Anti-phase boundary

ASM American Society of Materials

ASTM American Society for Testing and Materials

BSE Backscattered electron

DS Directionally solidified

DTA Differential thermal analysis

EDS Energy dispersive spectroscopy

EDX Energy dispersive X-ray

FCC Face centered cubic

FIB Focused ion beam

GB Grain boundary

HCF High cycle fatigue

HIP Hot isostatic pressing

HTAL High Temperature Alloys Laboratory

HT SHT High temperature solution heat treatment

LAB Low angle boundary

LCF Low cycle fatigue

LMP Larson Miller parameter

LT SHT Low temperature solution heat treatment

LVDT Linear variable differential transducer














































Figure 5-3. SEM secondary electron-backscattered electron pairs of carbide and y/y'
matrix cracking in longitudinal sections of tensile specimens tested at 850 OC.
A) SE of C modification, B) BSE of C modification, C) SE of C+B modification,
D) BSE of C+B modification, E) SE of C+N modification, F) BSE of C+N
modification. The applied stress direction is vertical.


134









minimum applied loads. In some cases, localized deformation resulted in de-cohesion

between carbide and matrix (Figure 6-10D). Metallography of several thread sections,

which are not subjected to significant stresses during testing, did not reveal any local

deformation at carbides. This confirmed that the plastic deformation occurred during

HCF loading and was not due to the HIP process.

Several longitudinal sections were deep etched in order to more thoroughly

examine changes to carbides from fatigue loading. This effort verified the observations

made in the light etched condition, namely cracked carbide plates in the C modification

(Figure 6-11A) and intact small carbides in the C+B modification (Figure 6-11B).

Interesting markings were also observed on the surfaces of some carbides, as shown in

Figure 6-12. These short, closely spaced features have been identified in the literature

[70] as evidence of slip bands in the y/y' matrix impinging on carbides, and these

features are thought to be an effect of localized plastic deformation.

The edge surfaces of several longitudinal sections were examined, motivated by

reports of recrystallization of nano-sized grains at the surface of fatigue specimens

caused by surface preparation [129]. Although no signs of recrystallization were found,

appreciable oxide formation was observed. Oxide layers, rich in Cr and Al, formed on

the surfaces of fatigue specimens at test temperature (Figure 6-13A). The oxide

formation caused depletion of Al and formation of a y phase layer below. These layers

were thicker on specimens that had experienced longer fatigue lifetimes. Cracking was

observed to form at the oxide layer and extend into the y layer (Figure 6-13B). Although

no surface crack initiations were found to cause any of the fatigue failures in this study,


159









Several stray grains, similar to the one observed after creep at 850 C (Figure 5-

18B), were identified in tested HCF specimens. The grains had carbides at the

boundaries, as shown in the Figure 7-5 image captured from a HT SHT HIPed N-

containing sample. The possibility of dynamic re-crystallization is greater under cyclic

loading conditions (HCF) as compared to a monotonic load (creep). The localized cyclic

deformation that develops at carbides provides more potential sites for grain nucleation

[141]. Even so, the grains observed are unlikely the result of dynamic re-crystallization

because the temperature of fatigue testing (850 C) is probably too low for this type of

process to occur. It remains unclear why these stray grains were observed only after

mechanical testing.

Carbides act as stress concentrators that can lead to development of dislocation

structures at the interface between carbides and matrix during cyclic HCF loading. In

addition, carbides can act as barriers to slip band and dislocation movement, which

leads to accumulation of plastic deformation near carbides as more load cycles are

applied to the material. The localized deformation serves as a predecessor to crack

initiation at carbides.

7.4.5 Fatigue Mechanisms and Effects on Cyclic Lifetimes

Observed crack initiation mechanisms had a significant impact on HCF lifetimes of

CMSX-4 and the modified alloys tested in this study. Recent work has identified

multiple fatigue lifetime distributions and S-N curve behaviors within the same alloy due

to different active failure mechanisms, as was shown in Figure 2-9. The report

concluded that alloys will have at least two competing fatigue failure modes, namely

internal crack initiation versus surface crack initiation [100]. Fatigue results from the

current investigation of C-modified CMSX-4 alloys suggest more than two competing


209
























S.' (D


(1)
(0
(D




















Figure 7-6. Characteristic HF crack initiation sites and associated relative lifetimes.
Ith














Figure 7-6. Characteristic HCF crack initiation sites and associated relative lifetimes.


215
























30 35 40 45 50 55 60 65 70 75 80
29 (o)
Figure 4-30. XRD spectrum of C+N modification in the fully heat treated and exposed
condition. Peak profile indicates decomposition of MC carbide and formation
of M23C6 carbides during long-term exposure.


Figure 4-31. SEM micrographs of topologically close packed (TCP) phases in deep
etched, thermally exposed samples. A) Baseline CMSX-4, B) C modification,
C) C+N modification.


115









[14] K.A. AI-Jarba and G.E. Fuchs, "Effect of carbon additions on the as-cast
microstructure and defect formation of a single crystal Ni-based superalloy,"
Materials Science and Engineering A, vol. 373, pp. 255-267, May 2004.

[15] S. Tin and T.M. Pollock, "Phase instabilities and carbon additions in single-crystal
nickel-base superalloys," Materials Science and Engineering A, vol. 348, pp. 111-
121, May 2003.

[16] J. Mihalisin, J. Corrigan, M. Launsbach, E. Leonard, R. Baker, and B. Griffin,
"Some Effects of Carbon in the Production of Single Crystal Superalloy Castings,"
Superalloys 2004, Champion, Pennsylvania: TMS, 2004, pp. 795-800.

[17] Y. Zhou and A. Volek, "Effect of carbon additions on hot tearing of a second
generation nickel-base superalloy," Materials Science and Engineering: A, vol.
479, pp. 324-332, Apr. 2008.

[18] L.R. Liu, T. Jin, N.R. Zhao, X.F. Sun, H.R. Guan, and Z.Q. Hu, "Formation of
carbides and their effects on stress rupture of a Ni-base single crystal superalloy,"
Materials Science and Engineering A, vol. 361, pp. 191-197, Nov. 2003.

[19] Y.H. Kong and Q.Z. Chen, "Effect of minor additions on the formation of TCP
phases in modified RR2086 SX superalloys," Materials Science and Engineering
A, vol. 366, pp. 135-143, Feb. 2004.

[20] D. Porter and K. Easterling, Phase Transformations in Metals and Alloys, Boca
Raton, FL: Taylor & Francis Group, 2004.

[21] P. Auburtin, T. Wang, S. Cockcroft, and A. Mitchell, "Freckle Formation and
Freckle Criterion in Superalloy Castings," Metallurgical and Materials Transactions
B, vol. 31B, pp. 801-811, Aug. 2000.

[22] S. Tin, "Carbon Additions and Grain Defect Formation in Directionally Solidified
Nickel-Base Superalloys," PhD Thesis, University of Michigan, 2001.

[23] D. Ness and P.D. Lee, "Unpublished Research," 2007.

[24] N. D'Souza, M. Newell, K. Devendra, P. Jennings, M. Ardakani, and B. Shollock,
"Formation of low angle boundaries in Ni-based superalloys," Materials Science
and Engineering: A, vol. 413-414, pp. 567-570, Dec. 2005.

[25] X. Yang, D. Ness, P.D. Lee, and N. D'Souza, "Simulation of stray grain formation
during single crystal seed melt-back and initial withdrawal in the Ni-base
superalloy CMSX4," Materials Science and Engineering: A, vol. 413-414, pp. 571-
577, Dec. 2005.


248









6.2.1 Test Results

Fatigue lifetimes (Figure 6-7) were significantly greater than those for the Un-

HIPed, HT SHT material. Note the order of magnitude difference between the y-axes in

Figure 6-7 and Figure 6-1. The baseline, aside from the shortest lifetime data point to

be explained below, once again outperformed the modified alloys. The longest fatigue

lifetime among all the samples tested was the baseline specimen that failed after

8,797,800 cycles. The C+N modified alloy exhibited the second longest fatigue lifetimes

ranging between approximately 4 and 6.5 million cycles, followed by the C+B

modification (800,000 1.5 million cycles). The C modification clearly had the shortest

fatigue lifetimes and experienced the smallest lifetime improvement due to HIP

processing. The lifetimes were, in fact, very near those for Un-HIPed specimens. Note

that the lack of scatter in the results gives the appearance of one point for the three

closely plotted C modification data points.

6.2.2 Fractographgy

As in the Un-HIPed samples, all fatigue crack initiations occurred at internal

features. Representative initiation sites are shown in Figure 6-8. On the fracture

surfaces of the two longer lifetime baseline samples, primary crack initiation was

identified at small pores. These pores were much smaller than those responsible for

crack initiation in the Un-HIPed condition. The initiation site for the lowest lifetime

(742,208 cycles) baseline test contained a pore or y/y' eutectic region that was

significantly larger than any others observed at this condition. This was likely a rare

defect that was not removed by heat treatment and HIP processing, causing the shorter

lifetime. Crack initiation occurred at carbides of various morphologies in the modified

alloys. Blocky carbides approximately 20 40 pm in size were observed at initiation


157









performance than the Un-HIPed specimens. The C modification HIPed test was

interrupted by a heating element failure at 67.8 hrs and 1.16 % strain. The test was

stopped as an interrupted test, and results will be discussed when relevant.

5.3.1 Test Results

The overall time vs. strain curves for the seven full tests are shown in Figure 5-11.

The general shapes of the curves are similar as all specimens at this condition

underwent different amounts of primary, steady-state, and tertiary creep. All three of

the HIPed specimens (shown as solid lines) had creep lifetimes longer than any of the

other specimens, with HIPed, baseline CMSX-4 having the longest lifetime (653.8 hrs).

The HIP treatment resulted in a 38% increase in rupture lifetime for the baseline, 35%

increase for C+N modification, and a 15% increase for C+B modification. Among the

Un-HIPed CMSX-4 samples, the baseline once again had the best creep lifetime (473. 3

hrs). The onset of tertiary creep occurred in a similar manner for all alloys, as indicated

by the positive slopes of the strain rate vs. strain curves in Figure 5-12. Differences in

lifetimes, therefore, were influenced by relative minimum creep rates and strain rate

increases associated with tertiary creep. A comparison of minimum creep rates (Table

5-4) further highlights the effect of HIP on creep behavior at 850 C. The minimum

creep rates for the HIPed specimens were significantly lower than for Un-HIPed

samples.

The first 1% and 100 hrs of the time vs. creep strain curves are shown in Figure 5-

13. The non-smooth regions of the curves are due to slight extensometer slips. The

C+B modification samples, both HIPed and Un-HIPed, experienced more primary creep

strain (approximately 0.5%) than the other CMSX-4 alloys. HIP processing did not

appear to have a strong effect on primary creep behavior at this condition. Primary


123






















Figure 2-5. Decomposition of MC carbide after long term exposure (Adapted from [59]).
















400 *


-cc son 4oe ma o soy 'eve
T*vuipesalhie. C
Figure 2-6. Tensile yield strength and ultimate tensile strength vs. temperature for 2nd
generation CMSX-4 (points) and 3rd generation CMSX-10 (lines) SX
superalloys (Adapted from [9]).









THE IMPACT OF CARBON ON SINGLE CRYSTAL NICKEL-BASE SUPERALLOYS:
CARBIDE BEHAVIOR AND ALLOY PERFORMANCE


















By

ANDREW JAY WASSON


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

2010










A 014-1t IPd: LTS.T. 15... 1.575tyt



I *^ 4'1 .1. 1
,/'o 1 .




10OX-xa iitination siae (Qxde present)
f.1 r, fi l, [. i* i l i I Irh( l r( r. r1.. .M
jr<,,iti airog 4 angf i irygil iv lm i irare r ,,,. .. ..n


B 033-2: HIP.. C ,tr.LT ST;: 15 Ki;~14%eki
0 I


averloati


Id


"30i |- '. 500Y$exacl initiation site (ode preseni 3 500KB E-eacl inilallon site(e redei p san
Initiationatporewthcartiadl*StageIIcrackgrowth* .... Initiationatporewih carbide Stage II crakgrwth* "
isOy load ,J ... .. .. werloaf ... -..., ..
Figure H-7. Summaries of HCF fracture surfaces from "Round 7" HCF tests, HIPed, LT
SHT specimens tested at 15 Hz. A) Baseline, B) C modification, C) C+B
modification, D) C+N modification.


245


C 044-1: I...... C..; LT S& ..... e.33


D 071-1: .I;tl. ...I sa. .. i.....M


I J









blocking mechanism was confirmed by the observation of pores in direct contact with

carbides (Figure 4-5). The pores identified as crack initiation sites in all Un-HIPed HCF

samples of modified alloys were also associated with carbides; as were the pores at the

center of the square features on creep fracture surfaces. This suggests that the most

detrimental pores are those that are near carbides. An assessment of as-cast porosity

in a C-containing alloy is incomplete without careful observation of regions with

carbides. Examination of interdendritic regions away from carbides may give the false

impression that C additions reduce pore formation. The reports of decreased porosity in

C-containing superalloys may not fully account for pores adjacent to carbides.

While the large lattice parameters of interdendritic carbides may alleviate some

microshrinkage during solidification, the blocking of molten alloy by carbides is the more

dominant effect of C additions on casting porosity in SX superalloys. This results in

increased porosity due to the formation of pores near carbides. This negative side

effect of including C in SX superalloy chemistry is an important consideration that could

warrant changes in alloy processing for modified alloys. For example, the addition of C

may require incorporating a HIP cycle into the heat treatment schedule to control

porosity levels.

7.2 Heat Treatment and Exposure

The evolution of microstructures during heat treatment and thermal exposure

revealed further differences in alloy behavior between baseline CMSX-4 and the three

modified alloys. The following sections relate observed behavior to established

mechanisms of microstructural evolution in order to explain these differences.


189












1.0E+07


QU
m
U)


'1.OE+06 o- Un-HIPed HT SHT
"O 0 HIPed HT SHT
o 0
A HIPed LT SHT
.4-0
0o

S1.OE+05 -----------------
5o 0
4 0 0
0




1.OE+04
Baseline C C+B C+N

CMSX-4 Variant

Figure 6-22. HCF lifetime results for all tests. The y-axis is presented in the log scale.


182


U S
0


S









5-10 SEM micrographs of longitudinal sections of specimens tested in creep at 950
oC/300 M Pa.. ............... ...... .. ........ ........................ .......... 140

5-11 Time vs. strain curves for 850 C/550 MPa creep tests. All curves exhibit all
three regions of creep strain ....................................................................... 141

5-12 Creep strain vs. creep strain rate curves for 850 C/550 MPa creep tests.......... 141

5-13 Time vs. 1% strain curves for 850 C/550 MPa creep tests ............................. 142

5-14 SEM micrographs of fracture surfaces from creep tests of Un-HIPed (did not
undergo hot isostatic pressing) specimens at 850 C/550 MPa....... ................. 143

5-15 SEM micrographs of fracture surfaces from creep tests of HIPed specimens at
850 C/550 M Pa. ........... .......... ........... .... ............ ...... ......... 144

5-16 SEM micrographs of longitudinal sections of Un-HIPed specimens tested in
creep at 850 C/550 M Pa.. ...................... .. .. .. ....... ............... ... ............ 145

5-17 SEM micrographs of longitudinal sections of HIPed specimens tested in creep
at 850 C/550 M Pa.................. ..................... ............................ 146

5-18 SEM micrographs from longitudinal section from HIPed C modification creep
specimen tested at 850 C/550 MPa interrupted at 1.16% strain.. ................... 146

5-19 Time vs. strain curves for 750 C/800 MPa creep tests ................................. 147

5-20 Creep strain vs. creep strain rate curves for 950 C/800 MPa creep tests......... 147

5-21 Time vs. 3% strain curves for 750 C/800 MPa creep tests ............................. 148

5-22. SEM micrographs of fracture surfaces from creep tests at 750 C/800 MPa.. ... 149

5-23 SEM micrograph of ductile dimples near a pore on the fracture surface of a
C+B modified CMSX-4 specimen crept at 750 C/800 MPa........................... 149

5-24 SEM micrographs of longitudinal sections of specimens tested in creep at 750
C/800 M Pa.. .................... .......... ............... 1.......... .......... 150

5-25 SEM micrographs of hole-like features on the fracture surfaces of baseline
CMSX-4 specimens tested in creep at 850 C/550 MPa ............................... 151

5-26 Representative EDS line scan across carbide phases in a longitudinal section
of C+B modified CMSX-4 tested in creep at 750 C/800 MPa....................... 151

5-27. Creep data summarized in the Larson-Miller parameter (LMP) curve format..... 152

6-1 HCF lifetime results for alloys in the Un-HIPed, HT SHT condition.................. 167









LIST OF REFERENCES


[1] A.K. Sehra and J. Whitlow, "Propulsion and power for 21st century aviation,"
Progress in Aerospace Sciences, vol. 40, pp. 199-235, May 2004.

[2] R. Bowman, "Superalloys: A Primer and History," Superalloys 2000, Champion,
Pennsylvania: TMS, 2000.

[3] R. Decker, "Strengthening Mechanisms in Nickel-Base Superalloys," Zurich,
Switzerland: Climax Molybdenum Company, 1969.

[4] G. Dieter, Mechanical Metallurgy, New York: McGraw-Hill, Inc., 1986.

[5] R. Reed, The Superalloys: Fundamentals and Applications, Cambridge:
Cambridge University Press, 2006.

[6] F. Schubert, T. Rieck, and P. Ennis, "The growth of small cracks in the single
crystal superalloy CMSX-4 at 750 and 1000C," Superalloys 2000, Champion,
Pennsylvania: TMS, 2000, pp. 341-346.

[7] B. Wilson, "The Primary Creep Behavior of Single Crystal, Nickel Base
Superalloys PWA 1480 and PWA 1484," PhD Thesis, University of Florida, 2008.

[8] J. Yu, X. Sun, T. Jin, N. Zhao, H. Guan, and Z. Hu, "Effect of Re on deformation
and slip systems of a Ni base single-crystal superalloy," Materials Science and
Engineering: A, vol. 458, pp. 39-43, Jun. 2007.

[9] G. Erickson, "The Development and Application of CMSX-10," Superalloys 1996,
Champion, Pennsylvania: TMS, 1996, pp. 35-44.

[10] A. Yeh, A. Sato, T. Kobayashi, and H. Harada, "On the creep and phase stability
of advanced Ni-base single crystal superalloys," Materials Science and
Engineering: A, vol. 490, pp. 445-451, Aug. 2008.

[11] R. MacKay, T. Gabb, J. Smialek, and M. Nathal, "A New Approach of Designing
Superalloys for Low Density," Journal of Materials, vol. 62, pp. 48-54, 2010.

[12] I. Wright and T. Gibbons, "Recent developments in gas turbine materials and
technology and their implications for syngas firing," International Journal of
Hydrogen Energy, vol. 32, pp. 3610-3621, Nov. 2007.

[13] S. Tin, T. Pollock, and W. King, "Carbon additions and grain defect formation in
high refractory nickel-base single crystal superalloys," Superalloys 2000,
Champion, Pennsylvania: TMS, 2000, pp. 201-210.


247









group will undoubtedly serve me well in my professional career. All of the current and

former members of the High Temperature Alloys Laboratory (HTAL) deserve praise for

providing helpful insight and helping me to keep everything in perspective. Finally, I

would like to acknowledge the financial support of the National Science Foundation

(Grant Numbers 0072671 and 0353952) that contributed to these efforts.









temperatures. The APB energy has a direct effect on the y' shearing process. The

established mechanism for shearing of y' involves pairs of a/2 <110> {111} dislocations

[5]. The first dislocation enters the y' phase and creates the disordered APB region,

and the second dislocation follows along the same slip plane and restores order. These

dislocation pairs are often referred to as superdislocations. A larger APB energy

represents a greater barrier that must be overcome if shearing is to occur. Alloying

adjustments to improve strength must be made with care because they can also affect

heat treatment capabilities and additional phase formations [3].

2.3 Advancements in Superalloy Development

Advances in superalloy design have been driven largely by the need for

improvements in high temperature strength required for better engine performance.

The first Ni-base superalloys were equiaxed polycrystalline in nature, but innovations in

processing technology enabled the development of directionally solidified (DS) alloys.

These DS alloys have a columnar grain structure that eliminates any grain boundaries

that are transverse to the primary stress direction. These transverse boundaries tend to

reduce creep strength as higher diffusion rates along the boundaries enable more creep

deformation. Even more recently, a shift has been made from DS to SX alloys, which

contain no grain boundaries. The removal of grain boundaries has allowed for an

increase in temperature capability. Turbine engine components that can withstand

higher temperatures enable greater operating efficiencies. The progression in

processing over the years can be seen in Figure 2-1 [6].

SX superalloys have been developed over the years in several different

"generations" that represent some general alloying trends. The 1st generation SX

superalloys unveiled in the early 1980s did not contain grain boundary strengthening









1.1 times greater than the diffusivity at the maximum temperature of the LT SHT (1310

C). Increasing the temperature, and hence diffusivity, beyond 1325 C would likely

result in incipient melting in the alloy regardless of ramp rate and the number of steps

taken to reach maximum temperature. Extended hold times at maximum temperature

would be the only manner to promote significant carbide breakdown, but the heat

treatments would then become much more expensive and impractical from a production

standpoint. Longer times at elevated temperatures may also promote the formation of

unwanted TCP phases. It was determined that carbide morphologies in the studied

alloys could not be transformed during the SHT process. Observed carbide

decomposition during SHT of polycrystalline alloys can be attributed to differences in

carbide stability arising from variations in MC carbide compositions between

polycrystalline and SX superalloys.

7.2.3 Carbide Morphology Change in Boron-Containing Alloy

During the two-step aging heat treatment, some script carbide networks

transformed into groups of smaller, rounder carbide particles (Figure 4-21A) without a

significant change in carbide composition. This morphology change occurred most

frequently in the C+B modification, where the majority of script networks broke down

into these smaller carbide features. Most of the carbide networks in the C modification

retained their morphology after full heat treatment. The morphology change from

needle and plate-like features to more spherical shapes indicates a transformation

designed to minimize surface energy. While a sphere has a larger relative strain energy

than a plate or a needle [122], it has a lower surface energy. Surface energy has a

greater impact on morphology for coherent or semi-coherent phases while strain energy

dominates for incoherent phases. Work with MC carbides in SX superalloy systems has


192









Ti

W

Zr

A1203

CO

HCI

HNO3

H20

H202

MC

M6C

M23C6

MoO3

NaCI

Symbols

a

C

d

D

Do

da/dN

K

n

Nv


Titanium

Tungsten

Zirconium

Aluminum oxide

Carbon monoxide

Hydrochloric acid

Nitric acid

Water

Hydrogen peroxide

Metal carbide ('M' represents a variety of possible atomic species)

Metal carbide ('M' represents a variety of possible atomic species)

Metal carbide ('M' represents a variety of possible atomic species)

Molybdenum trioxide

Sodium chloride (rock salt)



Crack radius

Crack growth material constant

Precipitate size (length of a side)

Diffusivity

Arrhenius pre-exponential factor

Crack growth rate (distance per cycle)

Coarsening rate

Crack growth material exponential constant

Number of unpaired electrons for an element









[83] D.W. MacLachlan and D.M. Knowles, "Modelling and prediction of the stress
rupture behaviour of single crystal superalloys," Materials Science and
Engineering A, vol. 302, pp. 275-285, Apr. 2001.

[84] L. Zhao, N. O'Dowd, and E. Busso, "A coupled kinetic-constitutive approach to the
study of high temperature crack initiation in single crystal nickel-base superalloys,"
Journal of the Mechanics and Physics of Solids, vol. 54, pp. 288-309, Feb. 2006.

[85] K. Cheng, C. Jo, T. Jin, and Z. Hu, "Influence of applied stress on the [gamma]'
directional coarsening in a single crystal superalloy," Materials & Design, vol. 31,
pp. 968-971, Feb. 2010.

[86] R. Reed, D. Cox, and C. Rae, "Damage accumulation during creep deformation of
a single crystal superalloy at 11500C," Materials Science and Engineering: A, vol.
448, pp. 88-96, Mar. 2007.

[87] T. Tinga, W. Brekelmans, and M. Geers, "Directional coarsening in nickel-base
superalloys and its effect on the mechanical properties," Computational Materials
Science, vol. 47, pp. 471-481, Dec. 2009.

[88] L. Shui, T. Jin, S. Tian, and Z. Hu, "Influence of precipitate morphology on tensile
creep of a single crystal nickel-base superalloy," Materials Science and
Engineering: A, vol. 454-455, pp. 461-466, Apr. 2007.

[89] A. Pineau and S.D. Antolovich, "High temperature fatigue of nickel-base
superalloys A review with special emphasis on deformation modes and
oxidation," Engineering Failure Analysis, vol. 16, pp. 2668-2697, Dec. 2009.

[90] E. Silveira, G. Atxaga, and A. Irisarri, "Influence of the level of damage on the high
temperature fatigue life of an aircraft turbine disc," Engineering Failure Analysis,
vol. 16, pp. 578-584, Mar. 2009.

[91] M. Kim, D. Kim, and O. Oh, "Effect of [gamma]' precipitation during hot isostatic
pressing on the mechanical property of a nickel-based superalloy," Materials
Science and Engineering: A, vol. 480, pp. 218-225, May 2008.

[92] K. Harris, G. Erickson, S. Sikkenga, W. Brentnall, J. Aurrecoechee, and K.
Kubarych, "Development of Two Rhenium-Containing Superalloys for Single-
Crystal Blade and Directionally Solidified Vane Applications in Advanced Turbine
Engines," Journal of Materials Engineering and Performance, vol. 2 (4), pp. 481-
487, Aug. 1993.

[93] M. Lamm and R. Singer, "The Effect of Casting Conditions on the High-Cycle
Fatigue Properties of the Single-Crystal Nickel-Base Superalloy PWA 1483,"
Metallurgical and Materials Transactions A, vol. 38A, pp. 1177-1183, Jun. 2007.


254









generation SX superalloy creep tested at 980 C [138]. In Un-HIPed alloys, pores act

as pre-existing cavities that readily grow and form cracks as they incorporate vacancies

and small voids during creep. This process leads to more rapid failure than in HIPed

alloys, where cavities must nucleate from coalescing vacancies and voids before

growing and leading to fracture. The clear difference in size of cavities at the center of

square crack features for HIPed and Un-HIPed specimens was shown in Figure 5-25.

The growth of the pore in the Un-HIPed specimen causes an earlier onset of cracking

and thus failure than the more slowly forming creep cavity in the HIPed specimen. The

reduction of pores through HIPing does not improve creep behavior at very high

temperatures because cavities can rapidly form at TCP phases, aided by increased

vacancy diffusion rates.

HIP processing can clearly improve the intermediate temperature creep

performance of SX superalloys. Despite this fact, C-containing SX superalloys such as

CMSX-486 do not undergo HIP processing as part of the standard heat treatment

schedule, which employs only a two-step aging treatment [29]. The work here

demonstrates that a successful HIP cycle can be incorporated into the processing of C-

modified SX superalloys to improve creep performance.

7.3.3 Re-Crystallization Near Carbides?

Observation of an isolated grain in the longitudinal section of a HIPed C

modification interrupted creep specimen (Figure 5-18B) warranted further investigation

to determine if it was evidence of dynamic re-crystallization near carbides during creep

deformation. Examination of thread sections of HIPed creep specimens revealed no

evidence of grains, indicating that re-crystallization did not occur during HIP processing.

A thorough review of re-crystallization behavior in superalloys indicated that the


201


































Figure 6-10. SEM micrographs of longitudinal sections of HIPed, HT SHT specimens
tested in HCF. A) Baseline, B) C modification (BSE), C) C+B modification, D)
C+N modification. The applied stress direction is vertical.


- r 2OyCm


Figure 6-11. SEM micrographs of features observed in deep etched longitudinal
sections of HIPed, HT SHT HCF specimens. A) cracked carbide plate in the C
modification, B) intact small, rounded carbides in the C+B modification. Stress
direction is vertical.


172









A study by Xie and colleagues used in-situ SEM observation to capture the steps

involved in fatigue crack initiation at A1203 particles in Rene'95, a polycrystalline

superalloy disc alloy produced using powder metallurgy methods. At applied stresses

below the yield stress, dislocations built up at the alumina particles and locally

increased the stress concentration. As material damage increased, the particles

cracked or separated from the surrounding microstructure and caused crack

propagation into the matrix [36]. Shenoy and others recognized the role of hard ceramic

phases such as oxides and carbides in limiting fatigue lifetime. They incorporated

effects of these phases into a cyclic plasticity model for predicting LCF lifetime in a DS

superalloy. Fatigue cracks that formed due to carbide cracking or de-cohesion events

were modeled as micro-notches [35]. All of these described mechanisms of crack

initiation are consistent with the observations from the modified CMSX-4 fatigue

specimens. Figure 6-10D shows localized deformation at a blocky carbide and

associated de-cohesion.

Localized deformation has also been attributed to increased elastic modulus in

regions containing defects or secondary phases. Ravi Chandran pointed to localized

modulus increases and associated increases in stress as the cause of fatigue crack

formation at secondary phases in a number of alloys [100]. While local differences in

elastic modulus may play a role, experimental observations suggest that the

accumulation of localized deformation at interfaces between secondary phases

carbidess in the current study) and the matrix is the key factor leading to fatigue crack

initiation.


208









of the beam interaction volume in the material (up to several microns deep). An

example of one of these maps with the corresponding EDS spectra is shown in Figure

4-29. The area of the carbide labeled with a '1' is remaining MC carbide. Region '2' is

a M23C6 carbide that has formed on the decomposing MC carbide, and carbides '3' and

'4' are M23C6 carbides that formed nearby. Additional carbide maps are presented in

Appendix C. Some of the maps contain TCP phases, which will be discussed in the

next section.

XRD results confirmed decomposition of MC carbide and formation of secondary

M23C6 carbides due to long-term exposure. The MC carbide peaks were less intense

and the M23C6 peaks more intense than at the post-SHT and post-age conditions,

particularly for the C+B modification. A characteristic spectrum is shown in Figure 4-30

(additional spectra in Appendix B).

4.4.3 Topologically Close Packed (TCP) Phase Formation

Formation of detrimental TCP phases over long term exposures is a concern in

advanced SX superalloys such as CMSX-4 [63], and determining the role of minor

additions on TCP phases was an important aspect of this study. TCP phases were

observed after thermal exposure in all alloys except for the C+B modification, but the

location and quantities of TCP phase varied amongst the modifications. Baseline

CMSX-4 had short, needle-like TCP phases that formed sporadically in dendrite core

regions (Figure 4-31A). Thin TCP needles (Figure 4-31 B) were also observed at

dendrite cores and near decomposed carbides in the C modification. TCP phases

formed the most frequently in the C+N modification, in which groups of connected plates

and needles were observed in the vicinity of decomposed carbides (Figure 4-31 C).































30 35 40 45 50 55 60 65 70 75 80





















2e (o)
B -MC
















30 35 40 45 50 55 60 65 70 75 80
20 ()


C -MC

A V/V'





C
S





30 35 40 45 50 55 60 65 70 75 80
20 (o)

Figure B-1. XRD spectra for deep etched samples in the as-cast condition. A) C
modification, B) C+B modification, C) C+N modification.


224









to 800. Signal intensities were plotted against 29 angles and diffraction peaks were

indexed. Spectra were compared with data in the Personal Computer Powder

Diffraction File (PCPDF) database and with literature data to confirm results.

3.3.5 Transmission Electron Microscopy (TEM)

Dislocation structures of mechanically tested material and the finest

microstructural features were examined using transmission electron microscopy (TEM).

TEM samples were prepared by using the focused ion beam (FIB) to extract thin foils

from polished and light etched cross sections of HCF test specimens. The FIB uses a

gallium ion beam to mill thin rectangular samples from the bulk. Superalloy samples

require more thinning, to approximately 100 pm thickness, than other materials due to a

relatively high average atomic number that makes electron transparency more difficult.

FIB lift-out samples were positioned onto copper grids covered with carbon film. All

TEM samples were oriented parallel to the stress axis and therefore near the [001] zone

axis.

A JEOL 200CX at an accelerating voltage of 200 kV was used for the majority

of the TEM analysis. Images were captured to qualitatively observe dislocations,

carbide-matrix interfaces, and very fine secondary y' precipitates. A JEOL 2010F high

resolution TEM equipped with scanning transmission electron microscopy (STEM) and

energy dispersive x-ray (EDX) capabilities was utilized for quantitative compositional

data of phases observed in TEM.

3.4 Quantification

In addition to qualitative observations of microstructure at various conditions,

some quantitative measurements were conducted to compare the alloy modifications.

































Figure 5-2. SEM micrographs of fracture surfaces from tensile tests at 850 C. A)
Baseline, B) C modification (BSE), C) C+B modification, D) C+N modification
(BSE).


133











A MC
S- M236
A y/y'


,A4Ai4J


30 35 40 45 50 55 60 65 70 75 80
26 ()


30 35 40 45 50 55 60 65 70 75 80
20 ()


30 35 40 45 50 55 60 65 70 75 80
2e ()

Figure B-3. XRD spectra for deep etched samples after aging heat treatment. A) C
modification, B) C+B modification, C) C+N modification.


226


h6WAZA- 71"-~'l LLL









Table A-3. Orientation data for the CMSX-4 C+B
Bar Number y (0) 6 (0)
41 -1.4 2.3
42 -3.2 3.1
43 1.1 -2.7
44 4.4 9
45 0.5 -1.4
46 5.8 3.4
47 8 -22.4
48 5.7 1.9
49 0.6 -2.7
50 2.3 1.5
51 4.4 2.9
52 -1.1 -0.9
53 0.9 1.4
54 2.6 2.3
55 4 2.6
56 2.3 1.5
57 4.3 -1.7
58 1.2 -5.4
59 1.4 -1.6
60 -0.3 5.5


Table A-4. Orientation
Bar Number
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80


data for
y (0)
-10.7
5.1
37.2
-0.4
0
0.6
32
-8.3
4.4
4.4
2.6
2.1
1.8
2.9
-1.9
3.5
0.6
-18.1
-0.2
2.1


modification SX cast bars.
a (0)
2.7
4.4
2.9
10
1.5
6.7
23.7
6
2.8
2.7
5.3
1.4
1.6
3.5
4.8
2.7
4.6
5.5
2.1
5.5


the CMSX-4 C+N modification SX cast bars.
S(o) a (0)
-26.3 28.3
5.3 7.4
5.9 37.6
3.3 3.4
5.6 5.6
1.4 1.5
23.2 38.8
-2.5 8.7
-0.5 4.4
2.9 5.3
4.1 4.8
5.3 5.7
1.3 2.2
1.9 3.4
2.3 3
2.4 4.2
0.2 0.6
-37.2 40.8
-0.9 0.9
4.1 4.6


p (0)
-15.8
18.6
16.7
35.6
-45
32.3
-26.4
-15
26.7
-24.5
37.5
-18.8
-1.6
41.2
-39.6
-18
-20.7
1.9
3.5
37.5


p (0)
-44.9
-23.7
-43.1
18.4
-45
-44.6
29.2
-15.5
15.4
32.2
-19.7
-31.7
-42.3
-43
-31.2
12.1
1.6
-16.9
-30.6
-14.2


222









3.2.4 Hot Isostatic Pressing (HIP)

Selected alloy bars underwent a HIP treatment at PCC Airfoils in Minerva, OH.

The HIP process is commonly used in the superalloy industry and involves applying

isostatic pressure at elevated temperatures to effectively reduce or eliminate casting

porosity and improve mechanical properties [105-107]. In this study, reducing casting

porosity helped to isolate the effect of carbides on mechanical performance. The HIP

cycle conducted at PCC is shown in Table 3-3. Due to the risk of incipient melting at

the highest temperature of the HIP cycle, 1313 OC, the bars underwent SHT before

being sent for HIP processing. The cooling rate at the end of the HIP cycle was slow

enough that some y' coarsening may result. To eliminate this coarsening, bars that

were HIPed underwent a partial solution heat treatment (Table 3-3) before aging. The

maximum temperature of the partial solution heat treatment was the same as the

maximum temperature of the SHT for a given test bar, but slow heat up rates and lower

temperature holds were eliminated.

3.3 Sample Preparation and Characterization Instrumentation

A key component of this study was the examination of alloy microstructure at

different stages of processing or testing in order to correlate structure with observed

material behavior. Metallographic samples were prepared from alloy bars and

mechanical test specimens at every condition before being analyzed with a variety of

characterization tools at the University of Florida Major Analytical Instrumentation

Center (MAIC).

3.3.1 Metallography

Metallographic samples were prepared from as-cast, post-SHT, fully heat

treated, and thermally exposed samples cut from alloy bars and from all tested










t LbJ
4.


A

1 mm 1 mm











Figure 4-4. SEM backscattered electron (BSE) micrographs of transverse cross
sections used to estimate porosity volume fraction. A) baseline, B) C
modification, C) C+B modification, D) C+N modification.
s" -, ,, ",- ,
h- ,,
I IIP-8 i..






modification, C) C+B modification, D) C+N modification.


Figure 4-5. SEM micrograph showing casting porosity in immediate vicinity of carbides.


102


~f~:-









diameter decorated the periphery of a region where a MC carbide underwent complete

dissolution (Figure 4-26B). These decomposition zones were almost entirely y' phase

(dark contrast in micrograph). These y' layers are commonly reported as a product of

carbide decomposition processes [56,59]. The decomposing MC carbide provides C

(forms M23C6 carbides) and Ti (forms Ni3(AI,Ti) y') and the y matrix provides Ni and Al

(forms Ni3(AI,Ti) y') and Cr (forms M23C6 carbides). Local enrichments of Ni, Al, and Ti

occur with precipitation of M23C6 carbides, and y' forms and grows around the M23C6

carbides. The process can be summarized by the reaction MC + y M23C6 + y'. The

size of these circular regions indicates that the decomposing MC carbides were the

small, spherical carbide particles that were observed after aging heat treatment in the

C+B modification (Figure 4-21A).

Examination of exposed samples in the deep etched condition presented a clearer

picture of carbide decomposition mechanisms. Secondary M23C6 carbides formed near

partially decomposed script networks in the C modification, as shown in Figure 4-27A.

The plates on the left side of the image represent the remaining MC carbide and the

small carbides on the right are the secondary carbides that have formed. The EDS

spectrum for the carbide circled in the micrograph indicated high Cr content (Figure 4-

27B).

Secondary M23C6 carbides were observed on the surfaces of some decomposed

carbides in the C+B and the C+N modifications. These types of carbides were

observed most frequently in the B-containing alloy, in which they formed both near and

on the MC phase (Figure 4-28A). Long term service exposures of polycrystalline

superalloy samples exhibited continuous layers of M23C6 film that formed around









achieve the input conditions. Fatigue behavior of alloys in this condition was controlled

by porosity.

6.1.1 Test Results

Fatigue lifetime results for this condition are presented in Figure 6-1. Results were

plotted as lifetime vs. alloy modification, as opposed to more common stress vs. lifetime

plots, because all testing was conducted at the same condition. As indicated in the

figure, the tests at 20 Hz showed longer lifetimes than the 15 Hz tests. This was due to

the decreased stress amplitudes in the 20 Hz tests. Overall, baseline CMSX-4

outperformed the C-containing modifications, which had similar lifetimes amongst them.

All fatigue lifetimes were significantly below 1 million cycles, and 7 of the 9 C-containing

specimens had lifetimes below 200,000 cycles.

6.1.2 Fractography

All fracture surfaces were examined using SEM to characterize crack initiation

sites, crack growth behavior, and overall appearance. Characteristic crack initiation

sites for the Un-HIPed specimens are shown in Figure 6-2. In all Un-HIPed tests, failure

causing fatigue crack initiations occurred at internal pores. The porosity at initiation

sites in the modified alloys was associated with carbides, and cracked carbides were

frequently observed inside of pores.

The fracture surfaces of baseline alloy specimens contained smooth planes at

angles roughly 450 from the stress direction that were continuous across most of the

surface. This indicates large scale cleavage along {111} type crystallographic planes

that occurred after crack initiation and initial crack growth. Crack growth regions were

small on the baseline fracture surfaces and consisted of areas of stage-I crack growth

(along crystallographic planes of high stress) and stage-Il crack growth (in a direction


154























Si 1: Ta rich MC carbide





3,i--, Gr- ric cabid


,.,. 3 Cr rich M2._C, carbide



liz l


5 Cr rich M..C;. carbide


.1~


2: Cr rich M:, carbide




E*Ohf~~


4 Ta rich MC carbide



P O



,6 Ni rich matrix
5 -,1 .'* '. '


*uI- I r,
II' -E A :I ......


Figure C-2. Carbide map of thermally exposed sample in the deep etched condition in
the C+B modification.


230









of the small carbides to one another leads to a greater amount of plastic deformation

than if the particles were widely spaced apart. Cracks initiate from cracking of the

strained matrix or de-cohesion between matrix and carbide.

Crack initiation from de-cohesion of discrete, blocky carbides from the matrix took

the longest to occur among the carbide initiation mechanisms. The HIPed C+N

modification specimens exhibiting these types of initiations had the longest fatigue

lifetimes of any of the C-modified alloys. Blocky carbides are not as prone to cracking

as the thin carbides in script networks. In addition, most blocky carbides were spaced

far enough apart that the local deformation fields of each did not interact with one

another. The levels of plastic deformation and interfacial strain necessary to cause de-

cohesion and crack initiation at individual blocky carbides require a large number of

fatigue cycles.

The longest fatigue lifetimes, and therefore the most delayed crack initiation,

occurred in the HIPed baseline CMSX-4 material. The crack initiations leading to failure

in these samples originated at very small pores or voids. These were either residual

features remaining after HIP processing or microstructural discontinuities that

developed during fatigue loading at 850 OC. The mechanism is likely similar to that for

initiation at carbides, with local plastic deformation accumulating cyclically at the small

void until a crack forms.

Testing at a variety of HCF conditions would be needed to determine the stresses

at which specific fatigue crack initiation mechanisms operate and to to fully develop S-N

curves for the alloys studied. The work completed, however, demonstrates that minor


211










1
0.9
0.8

0.7
0.6
CI-
c,, 0.5
C. 0.4
D -- Baseline
0.3
T0. -C modification
0.2 ---- -C+B modification
0.1 -- --- -- -C+N modification
0
0 50 100 150

Time (hrs)

Figure 5-7. Time vs. 1% strain curves for 950 C/300 MPa creep tests.

Table 5-3. Time to various % creep strains for tests at 950 C/300 MPa. All times are in
hrs.
0.01% 0.1% 0.2% 0.5% 1% 2% 5% 10% Rupture
Baseline 0.06 11.1 30.3 91.7 138 173 218 254 300
C modification 2.09 15.5 38.7 89.7 125 156 195 225 251
C+B modification 0.2 5.86 16.2 69.5 128 171 219 255 290
C+N modification 0.56 10.7 25.4 75.4 120 151 193 227 261


137









fractions. The measurements here (Table 4-3) can be presented only as carbide area

fractions in the transverse cross sections. This data can be interpreted as a

representation of carbide "blockiness". Carbides in the N-containing alloy were the

blockiest and appeared larger and less frequently in cross section than carbides in the

other two modifications.

The deep etching process removes both the y and y' phases and yields a much

better representation of actual carbide morphology. When viewed only in cross section,

carbide appearances can be deceiving. Regions that seem to have groupings of small,

discrete carbide particles may actually contain large carbide networks. A comparison of

how carbide networks appear in the two different etched conditions is shown in Figure

4-7.

A very detailed examination of carbides in the deep etched condition for both

transverse and longitudinal samples was conducted to compare carbide morphologies

between the modifications. Characteristic carbide morphologies for the three C-

containing alloys are shown in Figure 4-8. The C modification as well as the C+B

modification resulted in carbides that formed in script-like networks consisting of

connected rods and plates. Some regions consisted of arrays of parallel rods which

were partially "filled-in" from incomplete plate formation during solidification, as seen in

Figure 4-9. Cores of carbide networks resided between primary dendrite arms and

carbide rods branched off of the cores in between secondary dendrite arms. The

manner in which the carbides conform to the edges of the dendrite arms can be seen in

Figure 4-10. The C+N modification exhibited blocky carbides that showed less spatial

orientation in relation to the dendrites than the script networks found in the other two









Table A-1. Orientation data for baseline CMSX-4
Bar Number y (0) 6 (0)
1 11.8 -3.3
2 1.7 1.5
3 -3.2 3
4 4.2 0
5 2.3 3.1
6 -0.3 0.4
7 3.8 -1.1


-3.8
-0.8
8.2
0.6
0.4
0.1
0.4
3
1
1
0.3
0.9
1.2


Table A-2. Orientation data for
Bar Number y (0)
21 2.6
22 -0.6
23 2
24 -0.3
25 1.9
26 -1.6
27 -0.1
28 -0.9
29 -0.9
30 0.9
31 2.2
32 0.2
33 1.1
34 2.4
35 1.5
36 2.2
37 1.1
38 1.2
39 3
40 2.5


-4
-1.6
7.5
1.4
3.5
-3.7
4.2
-1.6
1.2
2.7
1.6
1.4
4.1


SX cast bars.
a (0)
12.2
2.3
4.4
4.2
3.9
0.5
4
5.6
1.8
11.1
1.5
3.5
3.7
4.3
3.4
1.6
2.9
1.6
1.6
4.2


the CMSX-4 C modification SX cast bars.
S(o) a (0)
3.4 4.3
1.2 1.3
-1 2.2
5.4 5.4
0 1.9
7.7 7.8
-2.2 2.2
6.5 6.5
3.2 3.3
5.8 5.8
0.7 2.3
1.1 1.1
-2 2.3
1.5 2.8
-0.8 1.7
-1.8 2.8
1.6 1.9
1.2 1.7
-0.5 3.1
-0.4 2.5


p (0)
-13
34.1
-14.5
-4.5
25.3
-21.5
-20.1
-15.4
-26.7
40.8
-19.8
-41.4
42.2
-5.2
-14.1
24.7
-24.4
35.9
32
-11.6


S(0)
11.1
36.5
-32.6
34.9
-31.5
41
16
2.9
-45
10.6
-31.3
6
-18.6
33.7
29.1
-33.7
-20.7
42.4
-28.5
-7.2


221






























A1C M-MF.j c lm'j, D
Figure 5-8. SEM micrographs of fracture surfaces from creep tests at 950 C/300 MPa.
A) Baseline, B) C modification, C) C+B modification, D) C+N modification.


138










A o-il ,


B 038-1: .IP.Cr I*, lC,- .,H


IU


I NO


Crack Propagat


Initation atpore Stage I crack growth on multiple plaes t .1._...
Soverload U .,. .. ....


B


Iniation at carbide ckster Stage I crack growth In ,
mitiple directrrns Stage II crack growtti overload ... ..


D 065-1: 4..... ... ........ 4

^^^^^jgtpg. 1BRx


Inititon at cartde Stage I crack growth overload .! .! Iniitiotnon alocky carbide mrxed Stage 4 and Hi crack "'
Figure H-4. Summaries of HCF fracture surfaces from "Round 4" HCF tests, HIPed, HT
SHT specimens tested at 15 Hz. A) Baseline, B) C modification, C) C+B
modification, D) C+N modification.


242


re,
~. ,


r
:: ~i

10DDX


C 056-1: I......T..ST. IS.. E.. flSt. I
--





























.- 1: Ti, Ta rich MC carbide







Sa U*
*LUJnsV


-II


2: Ti, Ta ic h MC carbide






---- A-- -. *i


4


.- 3.-




*w ',. I


Ti, Ta rich MC carbide


I I i,


-- ... ......


Figure C-4. Carbide map of thermally exposed sample in the deep etched condition in
the C+N modification.


232





















~ x .-- ;-

Figure 5-25. SEM micrographs of hole-like features on the fracture surfaces of baseline
CMSX-4 specimens tested in creep at 850 C/550 MPa. A) Un-HIPed, B)
HIPed.


0 10 20 30 40
distance along scan (pm)


50 60


Figure 5-26. Representative EDS line scan across carbide phases in a longitudinal
section of C+B modified CMSX-4 tested in creep at 750 C/800 MPa. A) SEM
micrograph with location of line scan, B) EDS line scan result showing relative
concentrations of elements along the length of the scan.


151









initiation sites, however, were similar to those in the Un-HIPed condition, indicating that

the particular HIP cycle applied to this group of bars did not reduce porosity as much as

the cycle applied to the HIPed, HT SHT bars.

6.3.1 Test Results

Fatigue lifetimes (Figure 6-16) were significantly shorter than those for the HIPed,

HT SHT condition. The plot of the lifetimes, in fact, resembles the plot for the Un-

HIPed, HT SHT HCF tests. The two baseline alloy specimens exhibited the longest

lifetimes (378,911 cycles and 291,575 cycles), and all the modified alloy samples had

lifetimes between 75,000 and 200,000 cycles. Only one data point was collected for the

C+N modification because a power outage during setup of the second test severely

bent the specimen in compression, rendering it non-usable.

6.3.2 Fractography

Fracture surfaces from the HIPed, LT SHT tests were also very similar to those in

the Un-HIPed, HT SHT condition. All failure-causing fatigue cracks initiated at internal

pores. Carbides were observed at the initiation pores in the modified CMSX-4

specimens as in the Un-HIPed material. Characteristic initiation sites are presented in

Figure 6-17. The light colored oxide layer at the initiation site in Figure 6-17A is a result

of exposure of the fracture surface to high temperature after failure. In this case, the

clamshell furnace was not removed from around the specimen until several hours after

failure.

The overall appearance of fracture surfaces was consistent with observations from

Un-HIPed samples, with smooth crystallographic planes 450 to the stress axis for the

baseline and large, circular Mode-I crack growth planes for the modified alloys.


161









Table 3-3. All heat treatments cycles used in this study. GFQ is Gas Furnace Quench


and AC
Cycle



HT SHT





HT SHT
HIPed




LT SHT
HIPed


is Air Cool.
SHT
12800C/2hr -
12960C/2hr -
13100C/2hr -
13180C/2hr -
13220C/2hr -
13250C/4hr/GFQ
12800C/2hr -
12960C/2hr -
13100C/2hr -
13180C/2hr -
13220C/2hr -
13250C/4hr/GFQ
12800C/2 hr -
12900C/2 hr -
12950C/2 hr -
13000C/2hr -
13050C/4hr -
13100C/4hr/GFQ


HIP


Partial SHT


11400C/6hr/AC
+
871 C/20hr/AC


ramp to
13130C/103
MPa AC



ramp to
13130C/103
MPa AC


12000C/10min-
12800C/1hr-
13250C/1 hr/GFQ



12000C/10min-
12800C/1hr-
13100C/1 hr/GFQ


11400C/6hr/AC
+
8710C/20hr/AC



11400C/6hr/AC
+
871 C/20hr/AC


Figure 3-3. Scanning electron microscope (SEM) micrographs of carbides. A) light
etched condition and B) deep etched condition.


Age











1.0E+07

*


C,,

o

0) 1.0E+0 ------ ------


W 1 .OE+05 ---------- L-------- L-------- ------ L --- -- ^- -- -

C modification
o= Script carbide
networks

1.0E+04
0 2000 4000 6000 8000 10000 12000

Elliptical Area of Initiating Feature (um2)

Figure 6-20. Crack initiating feature areas vs. fatigue lifetimes for HIPed, HT SHT HCF
specimens. The red dashed line represents a power curve trendline for the
data. The y-axis is presented in the log scale.


179









cooling rate minimizes precipitation and coarsening of the y' phase. The outer walls of

the furnace and the copper electrical connections for the graphite heating elements

were H20-cooled.

3.2.2 Aging Heat Treatments

After homogenization, a two-step aging heat treatment was performed on all

alloys to produce the desired microstructure. A primary age at 1140 C for 6 hours

formed a high volume fraction of strengthening y' precipitates, and the secondary age at

871 OC for 20 hours produced fine, "secondary" y' precipitates in the y channels.

All aging heat treatments were performed in air in Carbolite box furnaces. The

temperatures and times of heat treatment were low enough that oxidation was not

considered to be a problem. Temperature control in the box furnaces was maintained

using two K-type thermocouples directly in contact with the alloy bar surface, which led

to control of temperature to within 3 C. Quenching was performed by rapid removal of

bars from the furnace for air cooling (AC) on alumina racks. This method produced

cooling rates in excess of 100 C/min fast enough to avoid undesired y' precipitate

coarsening.

3.2.3 Long Term Exposure

The long term microstructural stability of baseline CMSX-4 and the modified

compositions was examined by thermally exposing fully heat treated alloys in air at

1000 C for 1000 hours. Structural changes due to exposure that were evaluated

included y' coarsening, carbide decomposition, and formation of TCP phases. All

thermal exposures were done in the Carbolite box furnaces using the same technique

as the aging treatments.









[105] J. Chang, Y. Yun, C. Choi, and J. Kim, "Development of Microstructure and
Mechanical Properties of a Ni-Base Single-Crystal Superalloy by Hot-lsostatic
Pressing," Journal of Materials Engineering and Performance, vol. 12 (4), pp. 420-
425, Aug. 2003.

[106] R. Stevens and P. Flewitt, "Hot isostatic pressing to remove porosity & creep
damage," Materials & Design, vol. 3, pp. 461-469, Jun. 1982.

[107] H.Y. Bor, C. Hsu, and C.N. Wei, "Influence of hot isostatic pressing on the fracture
transitions in the fine grain MAR-M247 superalloys," Materials Chemistry and
Physics, vol. 84, pp. 284-290, Apr. 2004.

[108] "Standard Test Method for Determining Volume Fraction by Systematic Manual
Point Count," ASTM Standards 2008, West Conshohoken, PA: ASTM
International, 2008.

[109] S. Das, J. Seol, Y. Kim, and C. Park, "Structure and mechanical properties of Ni-
Cr alloy produced by single roll strip casting," Materials & Design, vol. 31, pp. 570-
573, Jan. 2010.

[110] Y. Liu, R. Hu, J. Li, H. Kou, H. Li, H. Chang, and H. Fu, "Hot working characteristic
of as-cast and homogenized Ni-Cr-W superalloy," Materials Science and
Engineering: A, vol. 508, pp. 141-147, May 2009.

[111] Z. Dong, X. Li, G. Yuan, Y. Cong, N. Li, Z. Hu, Z. Jiang, and A. Westwood,
"Fabrication of protective tantalum carbide coatings on carbon fibers using a
molten salt method," Applied Surface Science, vol. 254, pp. 5936-5940, Jul. 2008.

[112] J. Yang, Q. Zheng, X. Sun, H. Guan, and Z. Hu, "Relative stability of carbides and
their effects on the properties of K465 superalloy," Materials Science and
Engineering: A, vol. 429, pp. 341-347, Aug. 2006.

[113] J. Yang, Q. Zheng, X. Sun, H. Guan, and Z. Hu, "Topologically close-packed
phase precipitation in a nickel-base superalloy during thermal exposure,"
Materials Science and Engineering: A, vol. 465, pp. 100-108, Sep. 2007.

[114] B. Yan, J. Zhang, and L. Lou, "Effect of boron additions on the microstructure and
transverse properties of a directionally solidified superalloy," Materials Science
and Engineering: A, vol. 474, pp. 39-47, Feb. 2008.

[115] L. Zheng, C.Q. Gu, and Y.R. Zheng, "Investigation of the solidification behavior of
a new Ru-containing cast Ni-base superalloy with high W content," Scripta
Materialia, vol. 50, pp. 435-439, Feb. 2004.


256









3.4.3 Fracture Surfaces

Noteworthy features observed on fracture surfaces were measured using the

public software program Image Tool from the University of Texas Health Science Center

(UTHSC). The program allows the user to import micrographs and calibrate measuring

tools to determine the actual lengths of features. The sizes of the features at the crack

origin for all HCF tests were found by measuring the major and minor axes to estimate

the elliptical area. The shortest distance from the fatigue crack origin to the surface of

the specimen was also measured for every HCF test.

The same UTHSC software was used to measure and compare micropores on the

fracture surfaces of creep and tensile tests to quantify how pores may change during

creep testing. The pores were assumed to be elliptical, and a minimum of 10 pores

were measured for each sample.

3.5 Mechanical Testing

All tensile, creep, and HCF testing was conducted in the High Temperature

Alloys Lab (HTAL) at UF. Test conditions were determined through a detailed literature

review of previous work involving SX Ni-base superalloys. All testing was conducted on

fully aged material that had undergone either the LT or HT SHT and was either in the

HIPed or Un-HIPed condition. The majority of testing was on the HT SHT material.

3.5.1 Specimen Machining

Heat treated bars were sent to Joliet Metallurgical Laboratories for machining into

cylindrical test specimens. Samples for all types of testing were machined to the same

geometry, as shown in the drawing in Figure 3-4. Each test bar yielded 2 test

specimens. Final machining steps were done using low stress grinding techniques to

prevent residual stresses at the sample surface. Specimens were carefully measured









to exposure and the average precipitate sizes increased significantly. The graph in

Figure 4-25 illustrates that a greater degree of coarsening occurred in the C-modified

alloys. The B-containing alloy exhibited the largest post-exposure y' size (0.91 pm).

Assuming that volume diffusion is the controlling factor, coarsening of y' can be

represented by a cube law and Eqn. 4-1 [119,120]. The value of 'K' represents a

coarsening rate constant for a given alloy and temperature that is controlled by the

equilibrium concentration of y' solute in y, the diffusion coefficient of y' solute in y, and

the y/y' interfacial free energy. The values of 'K' (Table 4-6) were estimated using

average precipitate sizes before (do) and after (d) exposure and the time of exposure.

All of the modified alloys had larger rate constants than baseline CMSX-4, with the C+B

modification having the largest 'K' value.

d3 do3 = Kt (Eqn. 4-1)

4.4.2 Carbide Decomposition

All carbides underwent significant changes due to high temperature exposure.

Partial carbide dissolution and formation of M23C6 secondary carbides occurred in all

three modified alloys. Carbides were first observed in the light etched condition. Script

networks in the C modification, which had remained almost fully intact and 100 200

pm in size during heat treatment, were much smaller (10 50 pm) due to

decomposition during exposure. Figure 4-26A shows a small, Ta rich carbide in the C

modification that is likely a remnant of a decomposed carbide network. Small, Cr rich

secondary carbides were observed in the vicinity of decomposed primary carbides in all

C-containing alloys. The compositions, high Cr and significant presence of W, indicated

these secondary carbides to be M23C6 carbides [18,39,57,80,119]. In some instances

in the C+B modification, small, rounded M23C6 carbides approximately 1 3 pm in






































1: Cr rich M GC, carbide



IL -

"" F| +b'
c" __ .Jk ,, ,, r
t t rt.A


,,

:... -
hill -*


3: Ta rich MC carbide 4 Ta rich MC carbide,
1 some Cr present

:*. ,. q ,.\
"- ", L.. I' -

Figure C-1. Carbide map of thermally exposed sample in the deep etched condition in
the C modification.


229


2 Cr rich M,,C, carbide


- c- -'


.% N M W .
^ f?~ vraw










0.25
Baseline
-C modification
0.2 --C+B modification
-C+N modification
0.1 ----------------------------------
a 0.15
C
I 0.1

(D
0.05


0
0 5 10 15 20 25 30

Creep Strain (%)

Figure 5-6. Creep strain vs. creep strain rate curves for 950 C/300 MPa creep tests.
The increase of strain rate with additional strain is indicative of tertiary creep
and material damage accumulation.

Table 5-2. Minimum creep rates for tests conducted at 950 C/300 MPa.
Minimum Creep Rate (%/hr)
Baseline 0.0050
C modification 0.0041
C+B modification 0.0049
C+N modification 0.0057


136









and helped offset volume shrinkage through lattice expansion by interstitial C atoms.

Work at the University of Florida has shown an increase in microporosity due to minor

carbon additions [14,27,42]. The increase was attributed to interdendritic carbides

blocking fluid flow during the last stages of solidification. This theory is supported by the

observation of pores adjacent to carbides. This kind of pore-carbide spatial relationship

has also been seen in Mar-M247, a DS superalloy [51]. It appears that the fluid

blocking ability of carbides that is helpful in preventing thermosolutal convection and

defects may also lead to increased porosity.

It should be noted that some variability exists in the quantification of porosity, as

volume fractions are estimated from micrographs of two-dimensional cross sections.

Synchrotron tomography is a relatively new technique to characterize pore distribution

and shape in three dimensions [52], but it is not yet widely used in superalloy research.

Several other microstructural features have been examined with respect to C

content. Quantitative studies of as-cast microstructures revealed that C significantly

reduced the volume fraction of y/y' eutectic by changing eutectic formation temperature

[14,53]. This is a favorable effect because it improves the ability to achieve full

homogenization during heat treatment. Heat treatability can also be improved with a

fine dendrite structure and a small primary dendrite arm spacing (PDAS). PDAS

measurements indicate the average distance between solute rich dendrite cores, and

smaller values represent shorter diffusion lengths to achieve homogenization during

solution heat treatment [26]. The addition of C has been found to have no significant

impact on the overall dendritic structure or PDAS [45].










S- MC
M2306
A y/y'





E





30 35 40 45 50 55 60 65 70 75 80
29 (o)

Figure 4-23. XRD spectrum of the carbon and nitrogen (C+N) modification in the post-
aged, deep etched condition.


Figure 4-24. SEM micrographs of y' precipitates in the C modification. A) fully heat
treated condition and B) thermally exposed condition. Significant y' phase
coarsening occurs due to the 1000 C/1000 hr exposure.

Table 4-6. Quantitative measurements for exposed samples. Standard deviations are
shown in parentheses.
Baseline C C+B C+N

y' Volume Fraction (%) 70.2 (2.78) 74.0 (2.50) 70.9 (2.28) 70.8 (2.90)

y' Size (pm) 0.79 (0.05) 0.86 (0.05) 0.91 (0.07) 0.86 (0.04)
Coarsening Rate 11 107 17 107 107
Constant: K (pm3/sec) 1.48 x 10- 1.77 x 10 1.48 10


111









Primary creep is typified by an initially high creep rate followed by a rapid

decrease in strain rate due to work hardening as deformation is introduced in the

material [74]. Significant primary creep strains have been observed in advanced 2nd

and 3rd generation alloys at intermediate temperatures (650 C to 850 C) and high

stresses (above 500 MPa) [75,76]. At these conditions, shearing of y' precipitates and

development of significant creep anisotropy can occur [7]. The shearing of precipitates

by <112> {111} dislocations creates slip heterogeneity that leads to elliptical cross

sections of tested specimens [77,78]. A consequence of this mechanism is the

sensitivity of primary creep to minor changes in crystallographic orientation.

Misorientations greater than 100 from the [001] direction can lead to drastically higher

levels of primary creep and shorter overall creep lifetimes [79]. The likelihood of y'

shearing and large primary creep strain has been tied to y/y' lattice misfit, which impacts

dislocation behavior at the y/y' interface [75]. Smaller misfit values lead to fewer misfit

dislocations at the interface and allow for easier shearing of precipitates. Modifications

to alloy compositions result in changes to lattice misfit values, and this is believed to be

the cause of increased primary creep in later generations of SX superalloys. The role of

C in primary creep has not been thoroughly explored, but possible effects prudent to

examine include shear dislocation-carbide interactions and local differences in

composition caused by carbide formation.

While significant work hardening occurs in the primary creep region, a balance

between work hardening and recovery or damage processes is characteristic of the

secondary creep regime. The creep in this region is often referred to as steady-state

because the creep strain reaches a minimum, constant (or nearly constant) rate. This









APPENDIX A
CAST BAR MISORIENTATION DATA

Orientation data for each SX bar was provided by PCC Airfoils (Tables A-1

through A-4). Primary angles were measured using Laue diffraction patterns. Gamma

(y) is defined as a rotation about a <001> axis lying within reference plane and

perpendicular to the reference direction. Delta (5) represents rotation about the normal

to the plane with both the <001> nearest to the reference direction and the <001>

direction within the reference plane perpendicular to the reference direction. Beta (13) is

the clockwise rotation about the reference direction measured from the reference plane

to the nearest {001} plane passing through the <001> direction nearest to the reference

direction. Alpha (a) represents the angle between the reference direction and the

<001> direction, and it is derived from the other three angle values [101]. All

mechanical tests were conducted using specimens from bars with a values less than

100.


220









modified SX components. Several tests should be interrupted after a small amount of

primary creep to compare deformation mechanisms in C-containing and C-free alloys.

Using FIB techniques, TEM specimens can be selectively prepared near carbides to

identify any local differences in deformation mechanisms, such as y' precipitate

shearing, that may explain increased primary creep levels.

The most significant impact of carbide phases on SX superalloy performance

identified in this study was the reduction of HCF lifetimes due to crack initiation at

carbides and pores associated with carbides. HCF testing at a variety of temperatures

and stresses would help identify the conditions at which certain crack initiation sites and

crack formation mechanisms are active. Fatigue crack growth rate and LCF testing

should also be conducted on SX superalloys modified with C to further isolate the effect

of carbides on fatigue crack propagation.


219









heat treatments (Table 3-3) involved very slow ramp rates (1 C/sec), multiple steps,

and precise temperature control to slowly homogenize the alloy. The slow ramps

reduced segregation of regions rich in low melting point elements and thereby

minimized the risk of incipient melting. Even the B-containing alloy, which had a solidus

temperature of 1327 OC, did not experience any observed incipient melting during the

HT SHT, which had a maximum temperature of 1325 OC.

Although heat treatments of large SX gas turbine components are more difficult

than for small test bars, this exercise demonstrated that the implementation of careful

heat treatment design and tight temperature control can produce traditional superalloy

microstructures in SX alloys modified with C and other minor additions.

7.2.2 SHT Temperature and Stability of As-Cast Carbides

As described above, the SHTs used in this study satisfactorily homogenized the

alloys. Another goal of this study was to determine if the temperatures of SHT could

effectively "breakdown" carbides from large, script networks to smaller features that are

presumed less detrimental to mechanical properties. This attempt was in part motivated

by the work of He and others with a polycrystalline superalloy that exhibited increased

carbide decomposition as the SHT temperature was increased [61]. Two heat

treatments, HT SHT and LT SHT, were designed to study the effect of heat treatment

temperature on carbide morphology.

As described in section 4.5.3, neither the LT SHT nor the HT SHT resulted in

significant carbide breakdown. Both heat treatments caused slight changes on carbide

surfaces that indicated the early stages of decomposition, but no changes in carbide

morphology were observed. Estimates of relative C diffusivities indicated that the

maximum temperature of the HT SHT (1325 OC) resulted in a diffusivity that was only


191









There are reports in the literature that are somewhat contradictory to one another

regarding the creep conditions at which carbides improve properties and those at which

carbides hurt properties. Kong reported that additions of C, B, and Hf improved creep

behavior of a 2nd generation SX superalloy at a "low" temperature condition (850 C/430

MPa) and negatively effected behavior at two "high" temperature conditions (950 C/210

MPa and 1050 C/165 MPa). Improvements were attributed to a reduction of porosity

that was beneficial to the tertiary creep behavior, and the reductions in lifetime were tied

to higher creep rates during the primary and secondary stages due to irregular y'

precipitates and carbide cracking [38]. Liu and colleagues correlated the benefits and

detriments of carbides to the same microstructural features, but they observed these

effects at opposite creep conditions compared to the Kong report. Carbides were

beneficial to creep lifetimes at the "high" temperature condition (1038 OC/172 MPa) and

detrimental at the "low" temperature condition (871 0C/552 MPa) [81]. In both reports,

creep strain curves showing lifetime "improvements" due to C additions indicate only

minor differences in behavior and rupture lifetime that could be attributed to scatter.

While reduction of porosity and pinning of dislocations by carbides could improve creep

behavior, the limited data presented may not provide enough evidence to confirm

beneficial effects of C on creep behavior.

In the present study, carbides reduced creep rupture lifetimes compared to

baseline CMSX-4 at all three creep conditions. Any potential benefits of carbides were

outweighed by the negative effects, which included increased porosity and carbide

cracking. Both of these effects lead to greater damage in the alloy and would be

expected to have the greatest impact on tertiary creep behavior (rapid damage


198











1.0E+07


o |

o 1.0E+06 -



O *


1.0E+05 -.. -
o --,
II-
1.0E+04 -------------
0



1.0E+04
0 5000 10000 15000 20000 25000 30000 35000 40000

Elliptical Area of Initiating Feature (pm2)

Figure 6-19. Crack initiating feature areas vs. fatigue lifetimes for all HCF specimens.
The red dashed line represents a power curve trendline for the data. The y-
axis is presented in the log scale.


178


.1
b









ACKNOWLEDGMENTS

I would like to take a moment to thank all of those who have helped me during this

long and sometimes trying process. I could not have succeeded in my efforts without

the support of my parents and my sister. They have been an invaluable source of

advice regarding important life decisions that I have made over the past several years.

I would also like to give a special thanks to my fiance, Lauren Parigi, for her steady

love and support throughout my graduate school journey. She has been there for me

throughout all of the ups and downs, listening to my ramblings about research issues or

simply helping me relax and take a break from my work. Even though we have been

living several hours apart for the last couple of years, she has been my primary source

of strength in difficult times.

Siemens Energy in Orlando, Florida is also deserving of a special thanks. They

gave me with an opportunity to work in an industrial setting for several summers during

my graduate work. This provided me with invaluable experience that helped further my

growth as a developing engineer. I would like to especially recognize Kevin Sheehan,

Allister James, Sachin Shinde, and Cynthia Klein for their guidance.

A significant portion of this work was conducted at the Major Analytical

Instrumentation Center (MAIC) at the University of Florida (UF), and I owe a special

thanks to the staff for their assistance, particularly Dr. Amelia Dempere, Rosabel Ruiz,

and Wayne Acree. I would like to thank my research advisor, Dr. Gerhard Fuchs, for his

advice and support over the years. When I finished my undergraduate degree, I was

faced with a difficult decision regarding where to attend graduate school. I decided to

remain at UF because of Dr. Fuchs and his ability to balance directing students and

challenging them to think independently. The privilege of conducting research in his









4.6 Summary

The effects of alloy modifications included increased porosity and reduced

eutectic, but the y/y' microstructures developed through heat treatment were largely

unchanged. Minor additions of B and N, however, had a significant impact on carbide

phases in C-modified CMSX-4. The addition of N changed the as-cast carbide

morphology from large script to smaller and blockier. Carbides in the B-containing alloy

transformed from script MC networks to smaller, rounder MC carbides during heat

treatment. In addition, the presence of B led to suppression of TCP phase formation

during long term exposure.









CHAPTER 3
EXPERIMENTAL PROCEDURE

The experimental methods, techniques, and instrumentation used to carry out this

study are described below. The aim of this chapter is to provide not only the details of

the procedures, but insight into the motivation behind them. A brief discussion of how

the procedures were developed is included.

3.1 Materials

The alloys examined in this study were cast as part of a larger research effort to

study the role of C in Ni-base superalloys. One phase of the study involved the addition

of varying levels of C to a model 3rd generation SX superalloy [45,27]. The work

represented here involves another aspect of the study examining the additions of B and

N along with C to CMSX-4, a common commercial Ni-base superalloy. As discussed

previously, CMSX-4 is a 2nd generation SX superalloy with approximately 3 wt. % Re

content included to improve high temperature creep strength over 1st generation alloys.

It was chosen for study because it is widely used throughout the high temperature alloys

industry.

The baseline alloy and the C-containing modifications were cast at PCC Airfoils in

Minerva, Ohio from the same master heat of CMSX-4, with composition as shown in

Table 3-1. All of the minor additions were made just prior to pouring for optimal

composition control. Additions were made by wrapping the appropriate powder

(graphite powder C, boron powder B, CrN powder N) in Ni-foil and adding it to the

melt just prior to pouring. The desired and actual elemental contents of the

modifications are shown in Table 3-2. Cylindrical bars, each with a length of 12.5 cm

and a diameter of 1.25 cm, were cast in multi-bar cluster molds using the Bridgman









fatigue mechanisms that impact cyclic lifetimes. These mechanisms include, (i) crack

initiation at large pores followed by significant Mode-I crack growth, and (ii) local plastic

deformation at carbides followed by carbide cracking or de-cohesion between carbide

and y/y' matrix. Figure 7-6 contains representative SEM images of observed crack

initiation sites and the associated, relative fatigue lifetimes. Explanations for the

observed behavior will now be presented.

The shortest lifetimes were associated with crack initiations at large casting pores

in specimens that had not been HIPed or had been incompletely HIPed. Increased

porosity in the C-containing alloys was attributed to the presence of carbides and led to

shorter lifetimes than in baseline CMSX-4. Casting pores result in the earliest crack

initiation because they act as pre-existing circular or elliptical cracks that can cause

propagation (normal to the stress axis) of cracks from the edge of a pore into the matrix.

Crack initiation at large carbide networks resulted in fatigue lifetimes for HIPed, C

modification specimens that were not much greater than Un-HIPed lifetimes. Initiations

at networks were triggered by the cracking of thin rod and plate-like carbide features.

The small cross-sectional areas and brittle nature of these features make them

susceptible to cracking at a relatively low number of cycles. The cyclic development of

small amounts of plastic deformation at carbide-matrix interfaces further contributes to

cracking of the networks.

Crack initiations at clusters of rounded carbide particles, observed in HIPed C+B

modification specimens, resulted in longer lifetimes than initiation at script carbides.

The rounded particles are less likely to crack than the thin network features, and the

build up of more localized deformation is required before crack initiation. The proximity


210









[49] J. Cormier, P. Villechaise, and X. Milhet, "[gamma]'-phase morphology of Ni-
based single crystal superalloys as an indicator of the stress concentration in the
vicinity of pores," Materials Science and Engineering: A, vol. 501, pp. 61-69, Feb.
2009.

[50] Q.Z. Chen, Y.H. Kong, C.N. Jones, and D.M. Knowles, "Porosity reduction by
minor additions in RR2086 superalloy," Scripta Materialia, vol. 51, pp. 155-160,
Jul. 2004.

[51] H.S. Whitesell and R.A. Overfelt, "Influence of solidification variables on the
microstructure, macrosegregation, and porosity of directionally solidified Mar-
M247," Materials Science and Engineering A, vol. 318, pp. 264-276, Nov. 2001.

[52] T. Link, S. Zabler, A. Epishin, A. Haibel, M. Bansal, and X. Thibault, "Synchrotron
tomography of porosity in single-crystal nickel-base superalloys," Materials
Science and Engineering: A, vol. 425, pp. 47-54, Jun. 2006.

[53] L.R. Liu, T. Jin, N.R. Zhao, Z.H. Wang, X.F. Sun, H.R. Guan, and Z.Q. Hu, "Effect
of carbon additions on the microstructure in a Ni-base single crystal superalloy,"
Materials Letters, vol. 58, pp. 2290-2294, Jul. 2004.

[54] A. Szczotok, J. Richter, and J. Cwajna, "Stereological characterization of
[gamma]' phase precipitation in CMSX-6 monocrystalline nickel-base superalloy,"
Materials Characterization, vol. 60, pp. 1114-1119, Oct. 2009.

[55] D. Anton and A. Giamei, "Porosity distribution and growth during homogenization
in single crystals of a nickel-base superalloy," Materials Science and Engineering,
vol. 76, pp. 173-180, Dec. 1985.

[56] G. Lvov, V. Levit, and M. Kaufman, "Mechanism of primary MC carbide
decomposition in Ni-base superalloys," Metallurgical and Materials Transactions
A, vol. 35A, pp. 1669-1679, Jun. 2004.

[57] X. Qin, J. Guo, C. Yuan, J. Hou, and H. Ye, "Precipitation and thermal instability of
M23C6 carbide in cast Ni-base superalloy K452," Materials Letters, vol. 62, pp.
258-261, Jan. 2008.

[58] X. Qin, J. Guo, C. Yuan, J. Hou, and H. Ye, "Thermal stability of primary carbides
and carbonitrides in two cast Ni-base superalloys," Materials Letters, vol. 62, pp.
2275-2278, May 2008.

[59] X. Qin, J. Guo, C. Yuan, C. Chen, J. Hou, and H. Ye, "Decomposition of primary
MC carbide and its effects on the fracture behaviors of a cast Ni-base superalloy,"
Materials Science and Engineering: A, vol. 485, pp. 74-79, Jun. 2008.


251









circular crack [94]. Carbides represent another possible crack initiation site in C-

containing alloys. Their brittle nature makes them prone to cracking, which can cause

crack formation in the surrounding matrix and limit fatigue lifetime [35]. A failure

analysis conducted by Silveira demonstrated that even when they do not act as the

primary initiation site, cracked carbides can accelerate the onset of failure, as shown in

Figure 2-8 [40].

Localized plastic deformation near carbides has been recently identified as an

important aspect of the fatigue process. The stress concentrations in the vicinity of

carbides can lead to the formation and motion of dislocations that locally deform the

microstructure at macroscopic stresses below the yield stress [36]. These deformed

regions represent a localization of strain that can accelerate the crack initiation process

[39]. Continued load cycling can increase the deformation near carbides and develop

strains at the interface of the carbide and the y/y' matrix. De-cohesion of carbide from

the matrix can occur if the interfacial strain becomes large enough. This separation

event creates a site for a fatigue crack to form, propagate, and lead to failure [95,96].

Additional study is needed to explain the role of localized deformation at carbides on the

overall fatigue behavior.

Small amounts of B, along with C, have been shown to impact fatigue behavior of

IN 718, a polycrystalline superalloy. The addition of 29 ppm B resulted in longer

thermo-mechanical fatigue (TMF fatigue involving both strain and thermal cycling)

lifetimes, which was attributed to reduced y' phase coarsening [97,98]. Fatigue crack

growth (FCG) rates significantly decreased in IN 718 with 29 ppm B as compared to 12

ppm B. Clusters of B atoms were credited with retarding dislocation slip, which created









The presence of minor elemental additions is known to affect carbide precipitation

behavior. The presence of N has been connected to higher formation temperatures and

blockier carbides. The proposed theory involves carbides preferentially nucleating on

preexisting TiN particles high in the mushy zone [44]. Minor concentrations of B have

also been identified to change carbide precipitation behavior in SX superalloys [47], but

the specific effects have not been established. A report by Chen and others on a 2nd

generation alloy suggests that the effect of minor elements such as B and Hf is

determined by how they change carbide lattice parameter [37,48]. The theory proposes

that an increased carbide lattice parameter leads to larger lattice misfit between carbide

and matrix, and therefore carbides will assume a blockier structure to minimize the

surface area/volume ratio. This type of relationship between changes in carbide lattice

parameter and carbide morphology has not been established in other alloy systems.

Despite recent findings, continued efforts are required to better understand the role that

minor additions play in carbide formation.

Casting pores are generally accepted to negatively affect mechanical properties

because they act as internal flaws that generate significant local stress concentrations

[49]. There is disagreement in the field as to whether casting microporosity increases or

decreases with the presence of C. The differing opinions revolve around the role of

interdendritic carbides in the final stages of solidification, when pore formation occurs. It

was reported by Liu that additions of C to a 1st generation superalloy decreased both

the size and frequency of micropores [46]. Chen found that modifying a 2nd generation

superalloy with C, B, and Hf significantly lowered the microporosity [50]. Both of these

positive results were attributed to primary carbides that formed in interdendritic pools









or r phase, Ni3(Ti,Ta), can form between shrinking MC carbides and secondary

carbides on the periphery [56,58,59]. The formation of these transition zones and

interfaces is believed to have a negative effect on mechanical properties. An example

of such decomposition is shown in Figure 2-5 [59].

Long term exposure in superalloys can result in the unwanted formation of

topologically close packed (TCP) phases, named as such because they have planes of

close packed atoms that are separated by relatively large spacings. These brittle,

needle or plate-like phases remove strengthening elements from the matrix and are

susceptible to cracking. Later generation alloys with high refractory content are

particularly prone to the formation of these phases [63]. The effect of C on TCP phase

formation seems to be directly related to the carbide decomposition process. TCP

phases were observed to form directly on decomposing MC carbides, which was

attributed to local enrichment of TCP phase forming elements due to the decomposition

process [59]. Chen, however, reported a mechanism by which carbide decomposition

retarded TCP formation [37]. A SX alloy modified with small amounts of C and B

formed less TCP phase during exposure than an identical alloy without modifications.

The formation of secondary carbides consumed elements such as Cr that would

otherwise form TCP phases. The possibility of carbides improving long term alloy

stability is a promising one that demands further attention.

2.5.3 Tensile Properties

Tensile properties of superalloys are largely controlled by interactions of

dislocations with one another and with y' precipitates. Yield strengths tend to initially

increase with temperature as multiple octahedral {111} slip systems are activated and

then decrease rapidly as thermally activated dislocation climb or cross-slip of









APPENDIX D
COMPOSITIONAL RESULTS FOR TCP PHASES

This section contains semi-quantitative EDS results for the probed TCP phases in

exposed specimens. Note that no TCP phases were observed in the C+B modification.

Table D-1. Semi-quantitative compositions (wt. %) for TCP phase in baseline CMSX-4
exposed at 1000 OC for 1000 hrs.
Phase Co Cr Ni Re Ti W
1 9.1 44.3 0.7 46.0
2 6.9 39.5 -1.1 52.4
3 9.7 9.5 13.4 67.4
4 10.1 10.2 14.1 24.4 0.26 40.9


Table D-2. Semi-quantitative compositions (wt. %) for TCP phase in the
exposed at 1000 OC for 1000 hrs.
Phase Al Co Cr Re Ta W Ti
1 14.4 11.1 55.2
2 3.7 9.8 7.7 31.3


7.0 11.7
10.8
5.9
11.3
12.6
2.3
9.7


6.02
8.7
48.0
24.6
10.9
40.1
9.9


24.6
11.1

17.4
25.6
10.1


10.9
14.1

17.7


34.6


40.1


51.0


C modification


Ni
19.3
47.4


- 75.3
- 17.8
- 20.4
- 50.1
- 19.0
1.4 4.5
- 19.4


3.53
3.72


8.5


Table D-3. Semi-quantitative compositions (wt. %) for TCP
modification exposed at 1000 OC for 1000 hrs.


Re W
46.7
16.3
14.6
36.6
37.8
46.2
57.4
12.9 43.6
30.3 34.3
13.5 38.6
32.6
19.6 41.5


phase in the C+N


Ni
19.8
26.2
43.5
32.6
36.6

15.7
17.3
10.5
13.5
9.9
14.6


233


Phase
1
2
3
4
5
6
7
8
9
10
11
12


Co
12.0
4.4
5.5
11.6
12.3
28.6
9.0
12.3
9.5
9.8
8.7
8.5


Cr
12.1
51.1
31.9
11.2
10.9
22.5
10.7
10.1
10.3
9.0
9.4
8.6


C
9.4




2.6
7.3
3.83
2.6
7.4
3.1
4.5









observed directly in SEM and WDS analysis. A small amount of secondary M23C6

carbide formation occurred in addition to the MC carbide morphology transformations

that did not involve changes in composition. A characteristic spectrum is shown in

Figure 4-23 (additional spectra in Appendix B).

4.4 Exposed Microstructure

Long-term alloy stability has been identified as a significant issue in SX

superalloys with high refractory content. To evaluate microstructural stability, bar

sections that had undergone HT SHT and the two-step aging heat treatment were

thermally exposed in air for 1000 hrs at 1000 C. This exposure resulted in y' phase

coarsening, carbide decomposition with associated secondary carbide formation, and

TCP phase formation. Each topic will now be presented separately for clarity.

4.4.1 yly' Structure and Coarsening

Coarsening of the y' phase occurs during high temperature exposure and can

significantly reduce the strength of the alloy as precipitates grow and lose coherency

with the matrix. The primary driving force for coarsening is reduction of y/y' interfacial

area as the average size of precipitates increases and the number of precipitates

decreases [63].

Significant y' coarsening occurred in all alloys during exposure. The precipitates

remained cuboidal with some slight rounding of the corners. A comparison between the

fully heat treated and the exposed microstructure for the C modification is shown in

Figure 4-24. The precipitate size distribution clearly increases during exposure. Some

precipitates in the exposed condition were similar in size to the heat treated precipitates

and others were significantly larger. Average measurements for y' volume fraction and

precipitate size are presented in Table 4-6. The volume fractions increased slightly due









In addition to C content and carbide formation temperature, the composition of MC

carbides has an impact on their morphology. Unique alloy systems have varying

concentrations of MC carbide forming elements and will preferentially form carbides of a

certain composition. This appears to determine the prevalent carbide morphology.

7.1.3 Carbon and Casting Porosity

The effect of C additions on casting porosity in SX superalloys is not entirely clear.

There have been reports published describing both decreases [46,50] and increases

[14,27,42] in porosity due to the presence of C. As is the case with most C effects, the

observed behavior has been tied to interdendritic carbides.

The reduction of porosity observed by Liu [46] and Chen [50] has been attributed

to MC carbides "offsetting" volume shrinkage that occurs during the final stages of

solidification. The carbides have larger lattice parameters than the surrounding y and y'

phases, and this is believed to alleviate microshrinkage and reduce pore formation.

Chen has also reported a qualitative decrease in porosity as carbide volume fraction

increases [37].

Decreased porosity would be a significant and beneficial effect of C additions, but

it has not been consistently observed. Porosity has been found to increase with the

addition of C in SX superalloys with chemistries very similar to the alloys described

above in which porosity decreases were seen. The increases in porosity have been

attributed to blockage, by MC carbides, of molten alloy flow during the last stages of

solidification. This leads to pores that are associated with interdendritic carbides [51].

The current investigation provides further evidence of increased casting porosity

due to C modifications. As was shown in Table 4-2, all of the modified CMSX-4 alloys

exhibited higher as-cast porosity volume fractions than the baseline. The carbide


188









CHAPTER 6
RESULTS: FATIGUE BEHAVIOR

Fatigue testing was conducted to study the effects of alloy modifications on crack

initiation and fatigue lifetimes. Gas turbine components are subjected to a variety of

cyclic loading conditions, including LCF and TMF, due to engine startup and shutdown.

HCF conditions develop from high frequency vibrations present during engine operation.

HCF failures are controlled by crack initiation, which is often quickly followed by

catastrophic failure. HCF cracks can form at a variety of material defects or secondary

phases, and creating potential sites for crack formation is a primary concern when

introducing a phase carbidess in this case) to an alloy.

All HCF tests were conducted at 850 C, with an input stress R-ratio of 0.1 and

input stress range of 690 MPa. Under this condition, the maximum applied tensile

stress is below the alloy yield stress. Three rounds (round = 1 test of each alloy variant)

of testing were conducted on HT SHT material in both the Un-HIPed and HIPed

conditions, and 2 rounds of testing were conducted on the LT SHT material in the HIPed

condition. The gage sections of all fatigue specimens were carefully polished to reduce

the likelihood of crack formation at surface machining marks. This facilitated the study

of microstructure-controlled fatigue behavior.

6.1 Un-HIPed, HT SHT

The first 3 rounds of fatigue testing were performed on alloys that had undergone

the HT SHT and no HIP processing. A frequency of 20 Hz was used for the first round,

but the servo-hydraulic load levels did not reach the desired maximum and minimum

values. Actual stress ranges were near 620 MPa and actual R values were

approximately 0.15. The frequency was reduced to 15 Hz for all subsequent testing to


153

































Figure 5-17. SEM micrographs of longitudinal sections of HIPed specimens tested in
creep at 850 C/550 MPa. A) Baseline, B) C+B modification, C) C+N
modification. The applied stress direction is vertical.


Figure 5-18. SEM micrographs from longitudinal section from HIPed C modification
creep specimen tested at 850 C/550 MPa interrupted at 1.16% strain. A)
intact carbide rods, B) stray grain. The applied stress direction is vertical.


146









CHAPTER 4
RESULTS: MICROSTRUCTURAL EVOLUTION

The following sections describe the microstructures of the four alloys in the as-cast

condition and at various stages of heat treatment and thermal exposure. The y/y'

structure and secondary phases are covered with particular focus on the carbides in the

C-modified CMSX-4 alloys.

4.1 As-Cast Characteristics

In order to characterize the as-cast microstructures, alloys were first analyzed in

both the polished and light etched conditions on the optical microscope and the SEM.

Further details of carbide phase formation were obtained after deep etching and

thorough SEM analysis.

4.1.1 Differential Thermal Analysis (DTA)

Phase formation temperatures were determined using DTA curves for all alloys in

the as-cast and SHT conditions. The temperature values are shown in Table 4-1. In

general, the C additions reduced the solidus and liquidus temperatures as expected

[41,42,46]. The B-containing alloy exhibited the lowest solidus temperature and

therefore the greatest risk of incipient melting during heat treatment. The solidus and

liquidus both increased after SHT due to the reduction in segregation that occurred

during solidification. The carbide formation temperatures were all within 7 OC of each

other and between 13 OC and 20 OC below the liquidus temperature. This indicates that

MC carbide formation occurred at the same general height in the mushy zone for all

three modifications.









additions are relatively low cost, can be controlled to within several parts per million

(ppm), and offer the flexibility to add unique combinations elements at different

concentrations.

The purpose of this study was to examine the effects of C modifications to a

commercially relevant, 2nd generation SX superalloy (CMSX-4). Additions of C, C+B,

and C+N were made, and the alloy behaviors were compared to one another and to

baseline CMSX-4. The goal was to present a complete picture of how C impacts

multiple aspects of alloy behavior. The impact of heat treatment on carbide morphology

was assessed and, throughout the study, efforts were made to draw connections

between carbide morphology and alloy performance. Due to the large number of

micrographs, graphs, and tables presented in the following chapters, all figures have

been inserted at the end of the chapters to improve reading flow.









It is important to note that gas turbine engines are generally designed and

operated to avoid conditions that cause primary creep in critical components.

Therefore, increased primary creep strains associated with C-modified superalloys may

not be as significant in the assessment of alloy performance as more critical aspects

such as steady state creep and fatigue behavior.

7.3.2 Improvements in Creep Performance Due to HIP Processing

Although HIP processing of superalloys is performed primarily to improve fatigue

behavior, it can also benefit creep performance. The work of Chang involving HIP

processed CMSX-4 found that HIPed material experienced a 185% increase in rupture

lifetime at 950 C/355 MPa as compared to the Un-HIPed condition [105]. Results from

the current study also showed increases in rupture life, ranging from 15% to 38%, due

to HIPing in all alloys tested at 850 C/550 MPa. Lifetime improvements from HIPing,

however, are not observed at all creep conditions. Reed and others tested CMSX-4 at

a very high temperature condition (1150 C/100 MPa), and found that HIP processing

did not yield any lifetime improvement. Rapid formation of TCP phases and subsequent

creep cavitation at these phases resulted in similar rupture lifetimes for both HIPed and

Un-HIPed CMSX-4 [86].

The varying effects of HIP processing at different conditions can be explained by

examining the creep cavitation process. Vacancy condensation during creep

deformation can result in the formation of small voids that can grow into cavities [86].

Nucleation of these cavities occurs during the steady-state (secondary) stage of creep

and precedes cavity growth and the onset of tertiary creep. The last stages of

deformation before failure are controlled by cavity growth [137]. Yu identified cavitation-

controlled damage as the dominant mechanism inducing creep fracture in a 2nd


200









LIST OF TABLES


Table page

2-1 Compositions of some specific single crystal nickel-base (Ni-base) superalloys
in w t% [9] .............. ............. .... ............ .. ............53

3-1 Chemical compositions of master heat of CMSX-4 ............................................... 74

3-2 Amount of minor elements in baseline CMSX-4 and modifications......................... 74

3-3 All heat treatments cycles used in this study. GFQ is Gas Furnace Quench and
A C is A ir Cool. ................................................................. ... ........... 75

3-4 High cycle fatigue (HCF) test matrix with sample identification ............................ 76

3-5 Creep test m atrix.................... ............................. ................. 77

4-1 Differential thermal analysis (DTA) results showing phase transformation
temperatures. ................................... ............................... ........ 99

4-2 Estimated porosity volume fractions in the as-cast condition.............................. 101

4-3 Estimated carbide area fractions in the as-cast condition. .................................... 103

4-4 Quantitative measurements for fully heat treated samples................................. 109

4-5 Average spot wavelength dispersive spectroscopy (WDS) compositional (wt. %)
measurements of carbides in the fully heat treated (HT SHT + two-step age)
condition. ............ ................................ .............................. 110

4-6 Quantitative measurements for exposed samples. .................. .. ............ 111

4-7 Energy dispersive spectroscopy (EDS) semi-quantitative concentration (wt. %)
measurements of the topologically close packed (TCP) phases shown in
Figure 4-31 .................................... ............................... ......... 116

5-1 Im portant values from tensile testing at 850 C.................................................... 132

5-2 Minimum creep rates for tests conducted at 950 oC/300 MPa. ........................... 136

5-3 Time to various % creep strains for tests at 950 oC/300 MPa................ ............ 137

5-4 Minimum creep rates for tests conducted at 850 C/550 MPa. ................................ 142

5-5 Time to various % creep strains for tests at 850 C/550 MPa................ ............ 142

5-6 Maximum and minimum creep rates for tests conducted at 750 C/800 MPa....... 148









alloying additions in CMSX-4 have a direct effect on carbide morphologies, which

determine crack initiation mechanisms and fatigue lifetimes.


212









particularly in the rod and plate carbides comprising the networks in the C modification

(Figure 6-5B). The cracking did not extend significantly into the surrounding y/y' matrix

as it did in tested creep specimens.

Some cracks observed within 500 pm of the fracture surface had zigzagged crack

paths, as shown in Figure 6-6. This indicates the formation and growth of cracks along

octahedral {111} slip planes. These cracks are similar to those that have been reported

elsewhere in SX superalloys tested in HCF [39]. The cracks form angles with the stress

axis that are similar to those formed by the crystallographic facets observed on the

fracture surfaces.

6.1.4 Summary

Baseline CMSX-4 outperformed the C-modified alloys in HCF testing of Un-HIPed

material. All crack initiations occurred at internal pores, and initiating pores in the

modified alloys were associated with carbides. Fracture surfaces and microstructures

revealed cracking along crystallographic planes and Mode-I cracking perpendicular to

the stress direction. Final fracture regions in the baseline alloy exhibited uninterrupted

crystallographic cleavage, while overloaded areas in the modified alloys were much

rougher in nature.

6.2 HIPed, HT SHT

Three rounds of HCF tests were conducted on material that had undergone the HT

SHT cycle and HIP treatment. HIP processing improved fatigue lifetimes and resulted

in different fatigue initiation behavior as compared to the Un-HIPed alloys. Fatigue

lifetimes were impacted by crack initiation sites.


156









BIOGRAPHICAL SKETCH

Andrew Wasson was born on Clark Air Force Base in the Philippines and grew up

in places across the United States from Alaska to Florida and many others in between.

Andrew graduated with honors from Niceville High School in Niceville, Florida in 2002.

He went on to attend the University of Florida in Gainesville, Florida and began his

studies majoring in chemistry. He soon felt a draw towards engineering and changed

his major to materials science and engineering with a specialty in metals, and he

graduated cum laude in 2006 with his bachelor's degree.

Andrew chose to remain at the University of Florida for graduate school to

continue working with his senior research advisor, Dr. Gerhard Fuchs, on high

temperature alloys. During his graduate program, he completed two summer

internships at Siemens Energy in Orlando, FL working on materials used in gas turbine

power generating engines. He received his Ph.D. from the University of Florida in the

summer of 2010 and began employment in Houston, Texas working as a materials

engineer in the energy industry.


260









elements (C, B, Zr, and Hf) that were present in their polycrystalline and DS

predecessors. The justification at the time was that an alloy devoid of grain boundaries

did not need any elements that were originally included to strengthen such boundaries.

It should be noted that small amounts of Hf are included in most modern SX superalloys

to improve oxidation resistance. The 2nd generation in the early 1990s introduced

approximately 3 wt. % Re, which is an excellent solid solution strengthener that slows

down diffusional processes such as y' coarsening. The reasons that Re provides such

significant improvements in strength are not entirely understood, but changes to y/y'

misfit and the formation of clusters of Re atoms are thought to be responsible [7,8].

Negative aspects of adding Re include increased density and cost and reduction of

long-term alloy stability. The improvements associated with Re were clear, however,

and further increases in Re up to about 6 wt. % characterized the 3rd generation alloys

commercialized in the mid 1990s [9]. Examples of some common commercial

compositions from the leading companies in SX alloy development can be seen in Table

2-1 [9]. Alloy development in the 2000s consisted of efforts to improve stability of Re-

bearing alloys through ruthenium (Ru) additions [10] and reductions of Re content to

produce cheaper, lower density superalloys while maintaining high temperature creep

resistance [11].

It should be noted here that development of coating systems for both oxidation

and thermal protection has become very important in the superalloys industry over the

past 20 years [12]. Advanced coatings can improve component performance without

any changes to the substrate superalloy. This study, however, will focus on SX

substrate alloys and will address coatings only when directly relevant.










I 7.* ,. -.


* ,iM


Figure 3-4. Drawing for cylindrical specimen used for all mechanical tests.

Table 3-4. High cycle fatigue (HCF) test matrix with sample identification. Tests shown


in italics were conducted at 20 Hz.


Heat
Treatment
Cycle
HT SHT
HT SHT HIPed
LT SHT HIPed
HT SHT
HT SHT HIPed
LT SHT HIPed
HT SHT
HT SHT HIPed
LT SHT HIPed
HT SHT
HT SHT HIPed
LT SHT HIPed


Alloy
Modification


Baseline


C only


C+B


C+N


880

TQ03

in 4
S 30

100.
0 *


DoM


All others were conducted at 15 Hz.

Specimens


006-2
011-1
014-1
021-2
038-1
033-2
042-2
056-1
044-1
062-2
065-1
071-1


0.O97


002-1
015-1
004-2
022-2
037-1
033-1
043-1
057-2
055-1
061-2
076-2
079-2


002-2
015-2

022-1
037-2

043-2
057-1

061-1
065-2


0.133


time (seconds)

Figure 3-5. Triangular waveform used in high cycle fatigue (HCF) testing.


-~~~~ ~ ~ ~ -n
-~ -T -


.V1.
.,_. ,-, -.


































Figure 6-2. SEM micrographs of crack initiation sites on fracture surfaces from HCF
tests specimens in the Un-HIPed condition. A) Baseline, B) C modification, C)
C+B modification, D) C+N modification. All initiations for Un-HIPed specimens
occurred at pores.


Figure 6-3. SEM micrographs of features observed on fracture surfaces of Un-HIPed
HCF specimens. A) large crack growth plane perpendicular to the stress axis
and centered at crack initiation site (red arrow indicates initiation site and
black arrows indicate general crack growth directions), B) crack propagation
outward from crack initiation and growth region.


168









[26] B. Wilson, E. Cutler, and G. Fuchs, "Effect of solidification parameters on the
microstructures and properties of CMSX-10," Materials Science and Engineering: A, vol.
479, pp. 356-364, Apr. 2008.

[27] E. Cutler, "Effect of Carbon Additions and Carbide Morphology on the
Microstructure and Mechanical Properties of Ni-Base Superalloys," PhD Thesis,
University of Florida, 2006.

[28] Q. Chen, C. Jones, and D. Knowles, "The grain boundary microstructures of the
base and modified RR 2072 bicrystal superalloys and their effects on the creep
properties," Materials Science and Engineering A, vol. 385, pp. 402-418, Nov.
2004.

[29] J. Wahl and K. Harris, "CMSX-486 Alloy Update," Proceedings of ASME Turbo
Expo 2009: Power for Land, Sea and Air, Orlando, FL: ASME, 2009.

[30] K. Harris and J. Wahl, "Improved Single Crystal Superalloys, CMSX-4
(SLS)[La+Y] and CMSX-486," Superalloys 2004, Champion, Pennsylvania: TMS,
2004, pp. 45-52.

[31] R. Barrie, T. Gabb, J. Telesman, P. Kantzos, A. Prescenzi, T. Biles, and P.
Bonacuse, "Effectiveness of shot peening in suppressing fatigue cracking at non-
metallic inclusions in Udimet 720," Materials Science and Engineering: A, vol.
474, pp. 71-81, Feb. 2008.

[32] K.S. Chan, "Roles of Microstructure in Fatigue Crack Initiation," International
Journal of Fatigue, In Press, Accepted Manuscript.

[33] P. Kantzos, P. Bonacuse, J. Telesman, T. Gabb, R. Barrie, and A. Banik, "Effect
of powder cleanliness on the fatigue behavior of powder metallurgy Ni-disk alloy
Udimet 720," Superalloys 2004, Champion, Pennsylvania: TMS, 2004, pp. 409-
417.

[34] R.J. Morrissey and P.J. Golden, "Fatigue strength of a single crystal in the
gigacycle regime," International Journal of Fatigue, vol. 29, pp. 2079-2084, Sep.
2007.

[35] M.M. Shenoy, R.S. Kumar, and D.L. McDowell, "Modeling effects of nonmetallic
inclusions on LCF in DS nickel-base superalloys," International Journal of Fatigue,
vol. 27, pp. 113-127, Feb. 2005.

[36] X. Xie, L. Zhang, M. Zhang, J. Dong, and K. Bain, "Micro-mechanical behavior
study of non-metallic inclusions in P/M disk superalloy Rene'95," Superalloys
2004, Champion, Pennsylvania: TMS, 2004, pp. 451-458.


249




















wi*wI -~I~~LFl -.OII
,,,- U4


S2 Cr rich M.,C, carbide
1: Ta rich MC carbide



HI.I












exposure (1000 OC/1000 hrs). The EDS spectra shown are from the
3: Cr rich M. C,- carbide 4: Cr rich M: ,C carbide









corresponding numbers in the micrographs.










114









recorded when the specimen fractured. Samples were removed from the heat as

quickly as possible to minimize oxidation, which can hinder imaging of fracture surfaces.

3.5.4 Creep Testing

Creep tests were performed in air at 750 C/800 MPa, 850 C/550 MPa, and 950

C/300 MPa. The conditions were chosen to study the effect of carbides on creep

behavior at different regimes with varying deformation mechanisms. Most creep testing

was done on Un-HIPed HT SHT material, but several tests at the intermediate condition

were also run on HIPed HT SHT alloys. All tests ran until rupture except for one test

that was interrupted at about 1% strain. The creep test matrix is shown in Table 3-5.

Testing was conducted with Satec M-3 creep frames equipped with NuVision Mentor

software. Loading occurs by adding weights to a pan that is connected to a lever arm to

apply a load to the sample that is 16 times greater than the pan load. All threads were

lubricated as described previously to facilitate sample removal after testing. An

extensometer was attached with screws that fit into small indentations machined near

the threads of the sample. This extensometer was connected to a Linear Variable

Differential Transducer (LVDT) to measure displacements. Three K-type

thermocouples were tied to the gauge section, each connected to one of three zones of

the clamshell furnace. The sample was heated under a small pre-load until all

thermocouples read within 5 C of the test temperature. After a one hour soak, the test

load was applied in a step fashion by adding weights to the pan. An elastic modulus

value at the test temperature was determined from strain measurements taken after

each loading step.

Creep strain data was collected every 12 seconds during the first hour and every

minute for the remainder of the test. This data was used to produce creep strain vs.











I /
g i



Figure 2-7. Typical creep strain vs. time curve showing the three creep regions
(Adapted from [73]).










Figure 2-8. SEM micrograph of the growth of a fatigue crack across broken carbides
(Adapted from [40]).























Figure 4-10. SEM micrograph showing carbide network and secondary dendrite arms.


0 2 4 6 8 10 12 14 16
keV
Figure 4-11. Characteristic energy dispersive spectroscopy (EDS) spectrum for a
carbide in the as-cast condition. Carbides in all three modified alloys had
similar spectra.


MC













30 35 40 45 50 55 60 65 70 75 80
29 (o)
Figure 4-12. X-ray diffraction (XRD) spectrum of carbon (C) modification in the as-cast
and deep etched condition. Peak profile is consistent with MC-type carbide.


105









2.5.5 Fatigue Properties

Fatigue is the manner by which alloys fail from cyclic loading at stresses below the

ultimate tensile stress. Fatigue failures are particularly problematic because they often

occur suddenly without any clear warning signs [4]. The process begins with crack

nucleation at regions of localized strain. Crack growth follows until overload ductile

failure occurs in the remaining cross-section. Low cycle fatigue (LCF) generally refers

to fatigue in which plastic deformation occurs, and failure is generally crack growth

controlled. Most LCF testing is strain-controlled. In gas turbines, LCF conditions

develop due to start-up and shutdown cycles and from thermal strains. High cycle

fatigue (HCF) is stress-controlled and involves repeated loading below the yield stress,

and failure is crack initiation controlled. High frequency vibratory stresses in gas

turbines can lead to HCF conditions in blades [89].

Observed fatigue crack initiation sites in SX superalloys include pores, inclusions,

and the material surface [31,32,36,89]. Surface initiations are more likely to occur in

test specimens or components with rough surface finishes or scratches [90]. The

primary strategy for improving fatigue performance of an alloy is to reduce the size and

number of potential crack formation sites. Porosity can be greatly reduced through the

hot isostatic pressing (HIP) process, which combines high pressure and temperature to

collapse and "heal" casting pores [91]. During the development of CMSX-4, HIP

processing was shown to improve the 750 C HCF strength (to 107 cycles) of the alloy

by approximately 50% [92]. Attempts have been made to correlate the size of crack

initiating features to cyclic lifetimes, with some limited success [93]. Work at the Air

Force Research Laboratory involves minimum fatigue lifetime predictions using a

fracture mechanics approach in which the crack forming feature is assumed to be a









4.2.2 Porosity

Casting porosity has been shown to increase during homogenization in SX

superalloys due to coalescence of vacancies at high temperature [55], and the alloys

studied in this investigation demonstrated this effect. Pores observed were clearly

larger (Figure 4-15) in the post-SHT condition as compared to the as-cast state. The

dozens of pores examined in both conditions revealed an increase in maximum pore

size from ~ 5 10 pm to ~ 10 20 pm due to SHT. This size increase was observed

for both the baseline and modified alloys. The growth of pores is a side effect of

homogenization that may have a negative impact on mechanical properties.

4.2.3 Carbides

The HT SHT was designed to promote decomposition of large carbide networks to

smaller features that may be less detrimental to mechanical properties. SEM

examination revealed only minor carbide changes, and these changes were similar for

both the LT SHT (Figure 4-16) and HT SHT (Figure 4-17). Some of the carbides did not

undergo any changes during SHT and their surfaces remained smooth. A large number

of them, however, exhibited surface roughness that indicated early stages of carbide

decomposition. Small nodes were evident on the surfaces of carbide plates in the C

and C+B modifications and blocky carbides in the C+N modification, as shown in Figure

4-16A and Figure 4-17C. Decomposition reached a more advanced stage for some

carbides in the C+B modification as discrete, small carbides were observed near larger

carbide networks (Figure 4-16B). No change in carbide composition accompanied any

of the morphology changes due to SHT. Although most carbide networks remained

intact after SHT, the results indicate that partial carbide dissolution through diffusion of

C from primary MC carbides did occur in all alloys. The LT SHT (maximum temperature

































Figure 4-16. SEM micrographs of carbides after low temperature (LT) SHT. A) C
modification showing small nodes on carbide plate surface, B) C+B
modification (BSE) showing breakdown of network into smaller carbide
particles, and C) C+N modification (BSE) showing carbide surface roughness.


107









can be seen as well as the onset of tertiary creep and strain rate increase. Note that

just below 0.5% and after approximately 80 hrs the C modification rate begins to rapidly

increase and crosses over the baseline curve. The times to various % strains are

presented in Table 5-3. The results help to further illustrate the trends reported above.

5.2.2 Fractography

The fracture surfaces were similar for all alloys crept at 950 OC (Figure 5-8), with

groups of connected, square features linked by torn ridges. These square regions

indicate microcrack formation and subsequent octahedral slip in directions

perpendicular to the applied stress axis. The outward growth of the square features is

in the <011> type direction as intersections of active {111} slip planes occur ahead of

the crack tip [82]. Microcracks began at pores or creep cavities in the baseline alloy

and at cracked carbides or carbide/pore combinations in the modified alloys.

Secondary cracking on square crack planes was observed more in the C-containing

alloys than the baseline.

EDS analysis of carbides on fracture surfaces revealed that carbide decomposition

occurred during creep at 950 OC. Figure 5-9 shows carbides at the center of a square

crack plane in the C+N modification and the corresponding EDS spectrum. This type of

composition was not observed for any carbides in the fully heat treated condition, but it

also does not indicate as much Cr enrichment as the M23C6 carbides observed after

1000 OC/1000 hr exposure. The observed carbides represent an intermediate stage of

decomposition of MC to M23C6 carbides. The length of time (250 300 hours) was not

long enough and test temperature (950 OC) was not high enough to get the full

transformation that was observed after long-term exposure.


121




Full Text

PAGE 1

THE IMPACT OF CARBON ON SINGLE CRYST AL NICKEL-BASE SUPERALLOYS: CARBIDE BEHAVIOR AND ALLOY PERFORMANCE By ANDREW JAY WASSON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010 1

PAGE 2

2010 Andrew Jay Wasson 2

PAGE 3

To my family and soon-to-be wife for supporti ng me every step of the way 3

PAGE 4

ACK NOWLEDGMENTS I would like to take a moment to thank all of those who have helped me during this long and sometimes trying process. I could not have succeeded in my efforts without the support of my parents and my sister. They have been an invaluable source of advice regarding important life decisions that I have made over the past several years. I would also like to give a special thanks to my fianc, Lauren Parigi, for her steady love and support throughout my graduate school journey. She has been there for me throughout all of the ups and downs, listening to my ramblings about research issues or simply helping me relax and take a break from my work. Even though we have been living several hours apart for t he last couple of years, she has been my primary source of strength in difficult times. Siemens Energy in Orlando, Florida is also deserving of a special thanks. They gave me with an opportunity to work in an i ndustrial setting for several summers during my graduate work. This provided me with inva luable experience that helped further my growth as a developing engineer. I would like to especially recognize Kevin Sheehan, Allister James, Sachin Shinde, and Cynthia Klein for their guidance. A significant portion of this work wa s conducted at the Major Analytical Instrumentation Center (MAIC) at the University of Flor ida (UF), and I owe a special thanks to the staff for their assistance, particularly Dr. Am elia Dempere, Rosabel Ruiz, and Wayne Acree. I would like to thank my re search advisor, Dr. Gerhard Fuchs, for his advice and support over the years. When I finished my undergr aduate degree, I was faced with a difficult decision regarding where to attend graduate school. I decided to remain at UF because of Dr. Fuchs and his ability to balance directing students and challenging them to think i ndependently. The privilege of conducting research in his 4

PAGE 5

group will undoubtedly serv e me well in my professional career. All of the current and former members of the High Te mperature Alloys Laboratory (HTAL) deserve praise for providing helpful insight and helping me to keep everything in perspective. Finally, I would like to acknowledge the financial support of the National Science Foundation (Grant Numbers 0072671 and 0353952) that contributed to these efforts. 5

PAGE 6

TABL E OF CONTENTS page ACKNOWLEDGMENTS ..................................................................................................4 LIST OF TABLES ..........................................................................................................10 LIST OF FIGURES ........................................................................................................12 LIST OF ABBREVIATIONS ...........................................................................................19 ABSTRACT ...................................................................................................................25 CHA PTER 1 INTRODUC TION ....................................................................................................28 2 BACKGRO UND ...................................................................................................... 31 2.1 Overview ...........................................................................................................31 2.2 Strengthening ...................................................................................................32 2.3 Advancements in S uperalloy Development ......................................................33 2.4 Carbon Additions in Singl e Crystal Ni-Base Superalloys ..................................35 2.5 Effects of Carbon ..............................................................................................38 2.5.1 As-Cast Characteristics ...........................................................................38 2.5.2 Microstructural Evolution .........................................................................43 2.5.3 Tensile Properties ....................................................................................45 2.5.4 Creep Properties .....................................................................................46 2.5.5 Fatigue Properties ...................................................................................49 2.6 Summary ..........................................................................................................51 3 EXPERIMENTAL PROCEDURE ............................................................................58 3.1 Materials ...........................................................................................................58 3.2 Heat Treatments ...............................................................................................59 3.2.1 Solution Heat Treatment (SHT) ...............................................................60 3.2.2 Aging Heat Treatments ............................................................................62 3.2.3 Long Term Exposure ...............................................................................62 3.2.4 Hot Isostatic Pressing (HIP) ....................................................................63 3.3 Sample Preparation and Charac terization Instrumentation ...............................63 3.3.1 Metallography ..........................................................................................63 3.3.2 Optical Microscopy ..................................................................................65 3.3.3 Scanning Electron Microscopy (SEM) .....................................................65 3.3.4 X-Ray Diffraction (XRD) ..........................................................................66 3.3.5 Transmission Electron Microscopy (TEM) ...............................................67 3.4 Quantification ....................................................................................................67 3.4.1 Carbide and Porosity Volume Fraction ....................................................68 6

PAGE 7

3.4.2 Volume Fraction and Size ....................................................................68 3.4.3 Fracture Surfaces ....................................................................................69 3.5 Mechanical Testing ...........................................................................................69 3.5.1 Specimen Machining ...............................................................................69 3.5.2 Tensile Testing ........................................................................................70 3.5.3 Fatigue Testing ........................................................................................71 3.5.4 Creep Testing ..........................................................................................72 4 RESULTS: MICROSTRUCTURAL EVOLUTION ....................................................78 4.1 As-Cast Characteristics ....................................................................................78 4.1.1 Differential Thermal Analysis (DTA) ........................................................78 4.1.2 Dendrite Structure ...................................................................................79 4.1.3 / Eutectic .............................................................................................79 4.1.4 Casting Porosity ......................................................................................80 4.1.5 Primary Carbides .....................................................................................80 4.2 Post Solution Heat Treatment Microstructure ...................................................83 4.2.1 Homogenization ......................................................................................83 4.2.2 Porosity ...................................................................................................84 4.2.3 Carbides ..................................................................................................84 4.3 Post Aging Heat Tr eatment Microstructure .......................................................85 4.3.1 / Structure ...........................................................................................86 4.3.2 Carbides ..................................................................................................87 4.4 Exposed Microstructure ....................................................................................89 4.4.1 / Structure and Coarsening .................................................................89 4.4.2 Carbide Decomposition ...........................................................................90 4.4.3 Topologically Close Packed (TCP) Phase Formation ..............................93 4.5 Analysis ............................................................................................................94 4.5.1 Carbon Diffusion ......................................................................................94 4.5.1.1 Post-SHT vs. post exposed carbides .............................................96 4.5.1.2 Low temperature (LT) SH T vs. high temperature (HT) SHT ...........96 4.5.2 Alloy Stability Predictions ........................................................................97 4.6 Summary ..........................................................................................................98 5 RESULTS: CREEP DEFORMAT ION ...................................................................117 5.1 Elevated Temperat ure Tensile Testing ...........................................................117 5.1.1 Test Results ..........................................................................................118 5.1.2 Fractography .........................................................................................118 5.1.3 Longitudinal Sections ............................................................................119 5.2 Creep Testing at 950 C/300 MPao..................................................................119 5.2.1 Test Results ..........................................................................................119 5.2.2 Fractography .........................................................................................121 5.2.3 Longitudinal Sections ............................................................................122 5.2.4 Summary ...............................................................................................122 5.3 Creep Testing at 850 C/550 MPao..................................................................122 5.3.1 Test Results ..........................................................................................123 7

PAGE 8

5.3.2 Fractography .........................................................................................124 5.3.3 Longitudinal Sections ............................................................................124 5.3.4 Summary ...............................................................................................125 5.4 Creep Testing at 750 C/800 MPao..................................................................125 5.4.1 Test Results ..........................................................................................126 5.4.2 Fractography .........................................................................................127 5.4.3 Longitudinal Sections ............................................................................127 5.4.4 Summary ...............................................................................................128 5.5 Analysis ..........................................................................................................128 5.5.1 Sizes of Pores and Cavities ...................................................................128 5.5.2 Creep Specimen Elli pticity and Primary Creep ......................................129 5.5.3 Local Compositional Differences and Primary Creep ............................130 5.6 Summary ........................................................................................................131 6 RESULTS: FATIGUE BEHAVIOR ........................................................................153 6.1 Un-HIPed, HT SHT .........................................................................................153 6.1.1 Test Results ..........................................................................................154 6.1.2 Fractography .........................................................................................154 6.1.3 Longitudinal Sections ............................................................................155 6.1.4 Summary ...............................................................................................156 6.2 HIPed, HT SHT ...............................................................................................156 6.2.1 Test Results ..........................................................................................157 6.2.2 Fractographgy .......................................................................................157 6.2.3 Longitudinal Sections ............................................................................158 6.2.4 Transmission Electron Microscopy (TEM) .............................................160 6.2.5 Summary ...............................................................................................160 6.3 HIPed, LT SHT ...............................................................................................160 6.3.1 Test Results ..........................................................................................161 6.3.2 Fractography .........................................................................................161 6.3.3 Longitudinal Sections ............................................................................162 6.3.4 Summary ...............................................................................................162 6.4 Analysis ..........................................................................................................162 6.4.1 Crack Initiating Feature Size vs. Cyclic Lifetime ....................................163 6.4.2 Fracture Mechanics Approach ...............................................................164 6.5 Summary ........................................................................................................165 7 DISCUSSION .......................................................................................................183 7.1 As-Cast Characteristics ..................................................................................183 7.1.1 Carbides and Reduction of Grain Defects .............................................183 7.1.2 Minor Additions and Carbide Morphology ..............................................186 7.1.3 Carbon and Casting Porosity .................................................................188 7.2 Heat Treatment and Exposure ........................................................................189 7.2.1 Effect of Carbon and Minor Additions on Heat Treatability ....................190 7.2.2 SHT Temperature and Stabi lity of As-Cast Carbides ............................191 7.2.3 Carbide Morphology Chang e in Boron-Containing Alloy .......................192 8

PAGE 9

7.2.4 Coarsening of Phase .........................................................................194 7.2.5 Alloy Modifications and TCP Phase Formation .....................................195 7.3 High Temperature Mec hanical Properties Creep .........................................197 7.3.1 Competing Roles of Ca rbides in Creep Deformati on .............................197 7.3.2 Improvements in Creep Perf ormance Due to HIP Processing ...............200 7.3.3 Re-Crystallization Near Carbides? ........................................................201 7.4 High Temperature Me chanical Properties High Cy cle Fatigue (HCF) ..........202 7.4.1 Role of C in Changing Fracture Appearance .........................................202 7.4.2 Isolating Effects of Carbi de Morphology Through HIP Processing ........203 7.4.3 Possible Effects of Minor Additions in the Atomic Form ........................205 7.4.4 Local Plastic Deformation Near Carbides ..............................................206 7.4.5 Fatigue Mechanisms and Effects on Cyclic Lifetimes ............................209 8 CLOSING RE MARKS ...........................................................................................216 8.1 Research Conclusions ....................................................................................216 8.2 Future Work ....................................................................................................218 APPENDIX A CAST BAR MISORIEN TATION DA TA ..................................................................220 B XRD SPEC TRA ....................................................................................................223 C COMPOSITIONAL MAPS FROM D EEP ETCHED EXPOSED SAMPLES ...........228 D COMPOSITIONAL RESULT S FOR TCP PH ASES ..............................................233 E PHACOMP APPR OACH .......................................................................................234 F CREEP RESULTS FO R CMSX-48 6 .....................................................................235 G ELLIPTICAL MEASUREMENT S OF CREPT SAMPLES ...................................... 238 H HCF FRACTURE SURF ACE SUMMARI ES .........................................................239 LIST OF REFERENCES .............................................................................................247 BIOGRAPHICAL SKETCH ..........................................................................................260 9

PAGE 10

LIST OF TABLES Table page 2-1 Compositions of some specific singl e crystal nickelbase (Ni-base) superalloys in wt% [9]. ...........................................................................................................53 3-1 Chemic al compositions of master heat of CMSX-4. ................................................74 3-2 Amount of minor elements in bas eline CMSX-4 and modifications. ........................74 3-3 All heat treatments cycles used in this study. GFQ is Gas Furnace Quench and AC is Air Cool. ....................................................................................................75 3-4 High cycle fatigue (HCF) test matrix with sample ide ntification. ..............................76 3-5 Creep test matrix. ....................................................................................................77 4-1 Differential thermal analysis (DTA) results showing phase transformation temperatures. .....................................................................................................99 4-2 Estimated porosity volume frac tions in the as-cast condition. ...............................101 4-3 Estimated carbide area fractions in the as-cast condition. ....................................103 4-4 Quantitative meas urements for fully heat treated samples.. .................................109 4-5 Average spot wavelength dispersive sp ectroscopy (WDS) compositional (wt. %) measurements of carbides in the fu lly heat treated (HT SHT + two-step age) condition. ..........................................................................................................110 4-6 Quantitative measurements for exposed samples. ...............................................111 4-7 Energy dispersive spectroscopy (EDS ) semi-quantitative concentration (wt. %) measurements of the topol ogically close packed (TCP) phases shown in Figure 4-31. ......................................................................................................116 5-1 Important values from tensile testing at 850 C..o...................................................132 5-2 Minimum creep rates for tests conducted at 950 C/300 MPa.o.............................136 5-3 Time to various % cr eep strains for tests at 950 C/300 MPa..o.............................137 5-4 Minimum creep rates for tests conducted at 850 C/550 MPa.o.............................142 5-5 Time to various % cr eep strains for tests at 850 C/550 MPa..o.............................142 5-6 Maximum and minimum creep rates for tests conducted at 750 C/800 MPa.o......148 10

PAGE 11

5-7 Time to various % cr eep strains for tests at 750 C/800 MPa..o.............................148 5-8 Average elliptical areas of pores/cavities observed on fracture surfaces of failed specimens tested at 850 C.o.............................................................................150 6-1 Predicted minimum cyclic lifetimes fr om fracture mechanics ap proach and corresponding actual fatigue lifetimes for all HCF test specimens.. ..................181 7-1 Number of defect free bar s cast from baseline CMSX -4 and the 3 C-containing modifications studied.. ......................................................................................213 A-1 Orientation data for bas eline CMSX-4 SX cast bars. ............................................221 A-2 Orientation data for the CMSX-4 C modification SX cast bars. .............................221 A-3 Orientation data for the CMSX-4 C+B modification SX cast bars. ........................222 A-4 Orientation data for the CMSX-4 C+N modification SX cast bars. ........................222 D-1 Semi-quantitative compositions (wt. %) for TCP phase in baseline CMSX-4 exposed at 1000 C for 1000 hrs.o.....................................................................233 D-2 Semi-quantitative composit ions (wt. %) for TCP phase in the C modification exposed at 1000 C for 1000 hrs.o.....................................................................233 D-3 Semi-quantitative composit ions (wt. %) for TCP phase in the C+N modification exposed at 1000 C for 1000 hrs.o.....................................................................233 F-1 Minimum creep rate for CMSX-486 tested in creep at 850 C/550 MPa.o..............236 F-2 Time to various % creep stra ins for CMS X-486 tested at 850 C/550 MPa. All times are i n hrs.o................................................................................................237 G-1 Ellipticity values (ratio of major diam eter to minor di ameter) for all tested creep specimens. .......................................................................................................238 11

PAGE 12

LIST OF FIGURES Figure page 2-1 Schematic showing the progressi on in pr ocessing of superalloys.. .........................53 2-2 Representations of grain defects in single crystal (S X) superalloys. .......................54 2-3 Rupture life vs. misorientation for CMS X-3 and CMSX-486 tested at 982 C/248 MPa (Adapted from [29,30]).o..............................................................................54 2-4 The two most commonly observed MC (M metal, C carbon) carbide morphologies in SX superalloys.. .......................................................................54 2-5 Decomposition of MC carbide afte r long term expos ure (Adapted from [59]). .........55 2-6 Tensile yield strength and ultimate tensile strength vs. temperature for 2 generation CMSX-4 (points) and 3 generation CMSX-10 (lines) SX superalloys (Adapted from [9]).nd rd...........................................................................55 2-7 Typical creep strain vs. time curv e showing the three creep regions (Adapt ed from [73]). ...........................................................................................................56 2-8 SEM micrograph of the growth of a fatigue crack across broken carbides (Adapted from [40]). ............................................................................................56 2-9 Failure map showing different curve behavior for different fatigue mechanisms (Adapted from [100]). ..........................................................................................57 3-1 SEM micrograph of CMSX-4 after soluti on heat treatment (SH T) and two-step aging heat treatment. ..........................................................................................74 3-2 Temperature profile for comm on commercial SHT for CMSX-4: 1277 C/2 hr 1288 C/2 hr 1296 C/3 hr 1304 C/3 hr 1313 C/2 hr 1316 C/2 hr 1318 C/2 hr 1321 C/2 hr Gas Furnace Quench (GFQ).o o o o o o o o............................74 3-3 Scanning electron microscope (SEM) mi crographs of carbides.. ............................75 3-4 Drawing for cylindrical specimen used for all mechanica l tests. ..............................76 3-5 Triangular waveform used in high cycle fatigue (HCF) testing. ...............................76 4-1 Optical micrographs of transverse, as-cast CMSX-4 alloy microstructures.. ...........99 4-2 Optical micrographs of longitudinal as-cast CMSX-4 alloy microstructures.. ........100 4-3 / eutectic in as-cast microstructure. ..................................................................101 12

PAGE 13

4-4 SEM backscattered electron (BSE) micr ographs of transverse cross sections used to estimate porosity volume fraction.. ......................................................102 4-5 SEM micrograph showing cas ting porosity in immediate vicinity of carbides. .......102 4-6 SEM BSE micrographs of as-cast carb ides in transverse cross sections. ............103 4-7 SEM micrographs showing di f ferent appearance of carbides.. .............................103 4-8 Representative SEM micrographs of carbide morphologies in the deep etched condition.. .........................................................................................................104 4-9 SEM micrograph of partial carbide plat e formation in the C modification in the as-cast condition. ..............................................................................................104 4-10. SEM micrograph showing carbide net work and secondary dendrite arms. ........105 4-11. Characteristic energy dispersive s pectroscopy (E DS) spectrum for a carbide in the as-cast condition.. ...................................................................................105 4-12. X-ray diffraction (XRD) spectrum of carbon (C) modification in the as-cast and deep etched condition. ....................................................................................105 4-13 SEM micrographs showing similar / structures i n the post-SHT condition. .....106 4-14 SEM micrographs of post-SHT residual eutectic.. ...............................................106 4-15 SEM micrograph showing a por e in the post-SHT condition.. .............................106 4-16 SEM micrographs of carbides after low temperature (LT) SHT.. .........................107 4-17 SEM micrographs of carbides after high temperature (HT) SHT. ........................108 4-18 XRD spectrum of the C modification in the pos t-H T SHT, deep etched condition.. .........................................................................................................108 4-19 Representative SEM microgr aph of the fully heat treated / microstructure observed in all alloys. .......................................................................................109 4-20 Secondary, fine particles in channels of fully heat treated alloys.. ................109 4-21 Small, rounded carbide particles in fu lly heat treated condition of the carbon and boron (C+B) modification.. .........................................................................110 4-22 Small, tantalum (Ta) rich carbide in the fully heat treated condition.. ..................110 4-23 XRD spectrum of the carbon and nitr ogen (C+N) modification in the post-aged, deep etched condition. .....................................................................................111 13

PAGE 14

4-24 SEM micrographs of precipitates in the C modification.. .................................111 4-25 Average size ( m) of precipitates in the heat treated condition and after exposure at 1000 C for 1000 hours..o...............................................................112 4-26 SEM micrographs of light etched spec imens after high temperature exposure (1000 C/1000 hrs)..o.........................................................................................112 4-27 Deep etched C modification sample after high temperature exposure (1000 C/1000 hrs)..o...................................................................................................113 4-28 SEM micrographs of deep etched sample s after high temperature exposure (1000 C/1000 hrs)..o.........................................................................................113 4-29 Map of carbides in the deep etched c ondition after high temp erature exposure (1000 C/1000 hrs)..o.........................................................................................114 4-30 XRD spectrum of C+N modification in the fully heat treated and expos ed condition.. .........................................................................................................115 4-31 SEM micrographs of topol ogically clos e packed (TCP) phases in deep etched, thermally exposed samples.. ............................................................................115 4-32 SEM micrograph of deep etched C+B modification sample after high temperature expos ure (1000 C/1000 hrs)..o.....................................................116 5-1 Stress strain curves for tensile tests conducted at 850 C on fully heat treated (HT SHT) samples..o..........................................................................................132 5-2 SEM micrographs of fracture surf aces from tensile tests at 850 C..o....................133 5-3 SEM secondary electron-backscatter ed electron pairs of carbi de and / matrix cracking in longitudinal sections of tensile specimens tested at 850 C..o.........134 5-4 SEM micrographs of a longitudinal se ction from the basel ine CMSX-4 tensile specimen tested at 850 C..o..............................................................................135 5-5 Time vs. strain curves for 950 C/300 MPa creep tests..o......................................135 5-6 Creep strain vs. creep strain rate curves for 950 C/300 MPa creep tests..o..........136 5-7 Time vs. 1% strain curves for 950 C/300 MPa creep tests.o.................................137 5-8 SEM micrographs of fracture su rfaces from creep tests at 950 C/300 MPa..o......138 5-9 Chromium (Cr) rich carbides on the fracture surface of a C+N modification specimen crept at 950 C/300 MPa..o................................................................139 14

PAGE 15

5-10 SEM micrographs of longitudinal sections of specimens tested in creep at 950 C/300 MPa..o....................................................................................................140 5-11 Time vs. strain curves for 850 C/550 MPa creep tests. All curves exhibit all three regions of creep strain.o............................................................................141 5-12 Creep strain vs. creep st rain rate curves for 850 C/550 MPa creep tests.o.........141 5-13 Time vs. 1% strain curves for 850 C/550 MPa creep tests.o...............................142 5-14 SEM micrographs of fracture surfaces from creep tests of Un-HIPed (did not undergo hot isostatic pressing) specimens at 850 C/550 MPa..o......................143 5-15 SEM micrographs of fracture surfaces from creep tests of HIPed specimens at 850 C/550 MPa..o.............................................................................................144 5-16 SEM micrographs of longitudinal sections of Un-HIPed specimens tested in creep at 850 C/550 MPa..o...............................................................................145 5-17 SEM micrographs of longitudinal sections of HIPed specimens tested in creep at 850 C/550 MPa..o.........................................................................................146 5-18 SEM micrographs from longitudinal se c tion from HIPed C modification creep specimen tested at 850 C/550 MPa interrupted at 1.16% strain..o...................146 5-19 Time vs. strain curves for 750 C/800 MPa creep tests.o.....................................147 5-20 Creep strain vs. creep st rain rate curves for 950 C/800 MPa creep tests.o.........147 5-21 Time vs. 3% strain curves for 750 C/800 MPa creep tests.o...............................148 5-22. SEM micrographs of fracture surfaces from creep tests at 750 C/800 MPa..o...149 5-23 SEM micrograph of ductile dimples nea r a pore on the fracture surface of a C+B modified CMSX-4 s pecimen crept at 750 C/800 MPa.o............................149 5-24 SEM micrographs of longitudinal sections of specimens tested in creep at 750 C/800 MPa..o....................................................................................................150 5-25 SEM micrographs of holelike features on the fractu re surfaces of baseline CMSX-4 specimens tested in creep at 850 C/550 MPa.o.................................151 5-26 Representative EDS line scan across carb ide phas es in a longitudinal section of C+B modified CMSX-4 tested in creep at 750 C/800 MPa..o........................151 5-27. Creep data summarized in the Larson-Miller para meter (LMP) curve format. ....152 6-1 HCF lifetime results for alloys in the Un-HIPed, HT SHT condition. ......................167 15

PAGE 16

6-2 SEM micrographs of crack initiation site s on fracture surfaces from HCF tests specimens in the Un-HIPed condition. ..............................................................168 6-3 SEM micrographs of featur es observed on fracture surfaces of Un-HIPed HCF specimens.. ......................................................................................................168 6-4 Example catalog entry for Un-HIP ed, C modification HCF specimen.. ..................169 6-5 SEM micrographs of longi tudinal sections of Un-HIPed specimens tested in HCF.. ................................................................................................................169 6-6 SEM micrograph of zigzagged crack path indicating cracking along {111} octahedral slip planes in C+B modified Un-HIPed HCF specimen.. .................170 6-7 HCF lifetime results for alloys in the HIPed, HT SHT condition. ............................170 6-8 SEM micrographs of crack initiation site s on fracture surfaces from HCF tests specimens in the HIPed, HT SHT condition.. ....................................................171 6-9 SEM micrograph of crystallographic cra cking observ ed on the fracture surface of a C+B modification HCF sample in the HIPed, HT SHT condition. ...............171 6-10 SEM micrographs of longit udinal sections of HIPed, HT SHT specimens tested in HCF.. ............................................................................................................172 6-11 SEM micrographs of feat ures observed in deep etched longitudinal sections of HIPed, HT SHT HCF specimens.. ....................................................................172 6-12 SEM micrographs of shortly spaced f eatures observed on carbides in deep etched longitudinal sections of HIPed, HT SHT HCF specimens.. ...................173 6-13 SEM micrographs of surface oxide obs erved on carbides in longitudinal sections of HIPed, HT SHT HCF specimens.. ..................................................173 6-14 TEM micrographs of carbides and disloc ations in foils from H IPed, HT SHT HCF specimens.. ..............................................................................................174 6-15 TEM micrograph of fo il prepared from a baseline HIPed, HT SHT HCF specimen.. ........................................................................................................174 6-16 HCF lifetime results for alloys in the HIPed, LT SHT condition. ..........................175 6-17. SEM micrographs of crac k initiation sit es on fracture surfaces from HCF tests specimens in the HIPed, LT SHT condition.. ....................................................176 6-18 SEM micrographs of longitudinal sections of HIPed, LT SHT specimens tested in HCF.. ............................................................................................................177 6-19 Crack initiating featur e areas vs. fatigue lifetimes for all HCF specimens.. .........178 16

PAGE 17

6-20 Crack initiating featur e areas vs. fatigue lifetimes for HIPed, HT SHT HCF specimens.. ......................................................................................................179 6-21 Ratios of distance from sample surfac e to elliptical area of crack initiating feature vs. fatigue lifetimes for all HCF specimens.. .........................................180 6-22 HCF lifetime results for all tests.. .........................................................................182 7-1 Schematics of phase changes during heat treatment and long term thermal exposure.. .........................................................................................................213 7-2 Schematics of typical HCF fracture su rfaces, presented as side profile views.. ....213 7-3 SEM micrograph of transverse section from C+N modification HCF test specimen with a fatigue lif etime of 6,315,833 cycles. .......................................214 7-4 SEM micrograph of longitudinal sect ion from C modification HCF test specimen.. ........................................................................................................214 7-5 SEM micrograph of longitudinal secti on from C+N modification HCF test specimen.. ........................................................................................................214 7-6 Characteristic HCF crack initiation sites and associated relative lifetimes. ...........215 B-1 XRD spectra for deep etched samples in the as-cast condition.. ..........................224 B-2 XRD spectra for deep etched samples after HT SHT.. .........................................225 B-3 XRD spectra for deep etched samples after aging heat treatment.. .....................226 B-4 XRD spectra for deep etched samp les after thermal exposure at 1000 C for 1000 hrs..o.........................................................................................................227 C-1 Carbide map of thermally exposed samp le in the deep etched condition in the C modification. .....................................................................................................229 C-2. Carbide map of thermally exposed sa mple in the deep etched condition in the C+B modification. .............................................................................................230 C-3 Map of TCP phases and carbides in thermally exposed sample in the deep etched condition in the C+N modification. ........................................................231 C-4 Carbide map of thermally exposed sa mple in the deep etched condition in the C+N modification. .............................................................................................232 E-1 Example of PHACOMP calculat ion to predict alloy stability. .................................234 F-1 Full creep curve for CMSX-486 s pecimen tested in creep at 850 C/550 MPa.o....235 17

PAGE 18

F-2 Creep strain rate vs. creep strain for CMSX-486 specimen tested in creep at 850 C/550 MPa.o..............................................................................................236 F-3. Time to 2% creep strain vs. time for CMSX-486 specimen tested in creep at 850 oC/550 MPa. ..............................................................................................236 F-4. SEM BSE micrograph of fracture surf ace from creep test of CMSX-486 at 850 C/550 MPa.o.....................................................................................................237 F-5 SEM micrograph of longitudinal secti on of CMSX-486 specimen tested in creep at 850 C/550 MPa.o..........................................................................................237 H-1 Summaries of HCF fracture surfaces from Round 1 HCF tests, Un-HIPed, HT SHT specimens tested at 20 Hz.. .....................................................................239 H-2 Summaries of HCF fracture surfaces from Round 2 HCF tests, Un-HIPed, HT SHT specimens tested at 15 Hz.. .....................................................................240 H-3 Summaries of HCF fracture surfaces from Round 3 HCF tests, Un-HIPed, HT SHT specimens tested at 15 Hz.. .....................................................................241 H-4 Summaries of HCF fracture surfaces from Round 4 HCF tests, HIPed, HT SHT specimens tested at 15 Hz.. .....................................................................242 H-5 Summaries of HCF fracture surfaces from Round 5 HCF tests, HIPed, HT SHT specimens tested at 15 Hz.. .....................................................................243 H-6 Summaries of HCF fracture surfaces from Round 6 HCF tests, HIPed, HT SHT specimens tested at 15 Hz.. .....................................................................244 H-7 Summaries of HCF fracture surfaces from Round 7 HCF tests, HIPed, LT SHT specimens tested at 15 Hz.. .............................................................................245 H-8. Summaries of HCF fracture surfaces from Round 8 HCF tests, HIPed, LT SHT specimens tested at 15 Hz.. .....................................................................246 18

PAGE 19

LIST OF ABBREVIAT IONS Terms and Phrases AC Air cooling AES Auger electron spectroscopy AFRL Air Force Research Laboratory APB Anti-phase boundary ASM American Society of Materials ASTM American Society for Testing and Materials BSE Backscattered electron DS Directionally solidified DTA Differential thermal analysis EDS Energy dispersive spectroscopy EDX Energy dispersive X-ray FCC Face centered cubic FIB Focused ion beam GB Grain boundary HCF High cycle fatigue HIP Hot isostatic pressing HTAL High Temperat ure Alloys Laboratory HT SHT High temperature solution heat treatment LAB Low angle boundary LCF Low cycle fatigue LMP Larson Miller parameter LT SHT Low temperature solution heat treatment LVDT Linear variable differential transducer 19

PAGE 20

MAIC Major Analytical Instrumentation Center PCPDF Personal Computer Powder Diffraction File PDAS Primary dendrite arm spacing PHACOMP Phase computation S-N Stress versus cycle (common pl ot for fatigue lifetime data) SE Secondary electron SEM Scanning electron microscopy SFE Stacking fault energy SHT Solution heat treatment SIMS Secondary ion mass spectrometry STEM Scanning transmission electron microscopy SX Single crystal TCP Topologically close-packed TEM Transmission electron microscopy TMF Thermo-mechanical fatigue UTHSC University of Texas Health Science Center WDS Wavelength dispersive spectroscopy XRD X-ray diffraction Alloys CMSX-2 1st generation single crystal alloy CMSX-3 1st generation single crystal alloy CMSX-4 2nd generation single crystal alloy CMSX-10 3rd generation single crystal alloy CMSX-486 Single crystal alloy with added grain boundary elements IN 718 Polycrystalline alloy 20

PAGE 21

LMSX-1 Experimental si ngle crystal alloy M 963 Polycrystalline alloy Mar-M247 Polycrystalline alloy PWA 1480 1st generation single crystal alloy PWA 1483 1st generation single crystal alloy PWA 1484 2nd generation single crystal alloy Rene N4 1st generation single crystal alloy Rene N5 2nd generation single crystal alloy Rene N6 3rd generation single crystal alloy Elements and Compounds Al Aluminum B Boron C Carbon Co Cobalt Cr Chromium Fe Iron Hf Hafnium Mo Molybdenum N Nitrogen Nb Niobium Ni Nickel O Oxygen Re Rhenium Ru Ruthenium Ta Tantalum 21

PAGE 22

Ti Titanium W Tungsten Zr Zirconium Al2O3 Aluminum oxide CO Carbon monoxide HCl Hydrochloric acid HNO3 Nitric acid H2O Water H2O2 Hydrogen peroxide MC Metal carbide (M represents a va riety of possible atomic species) M6C Metal carbide (M represents a va riety of possible atomic species) M23C6 Metal carbide (M represents a variety of possible atomic species) MoO3 Molybdenum trioxide NaCl Sodium chloride (rock salt) Symbols a Crack radius C Crack growth material constant d Precipitate size (length of a side) D Diffusivity Do Arrhenius pre-exponential factor da/dN Crack growth rate (distance per cycle) K Coarsening rate n Crack growth material exponential constant Nv Number of unpaired el ectrons for an element 22

PAGE 23

Q Activation energy for diffusion R Universal gas constant (oft en written as 8.314 J/mol*K) Sv Area per unit volume t Time T Temperature Vv Volume fraction Average gamma prime precipitate size Units Angstroms nm nanometers m micrometers mm millimeters cm centimeters mL milliliters g gram kV kilovolt mA milliampere Hz hertz kHz kilohertz sec seconds min minutes hrs hours ppm parts per million wt. % weight percent at. % atomic percent 23

PAGE 24

kJ/mol kilojoules per mole oC degrees Celsius K degrees Kelvin 24

PAGE 25

Abstract of Dissertation Pr esented to the Graduate School of the University of Fl orida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy THE IMPACT OF CARBON ON SINGLE CRYSTAL NICKEL-BASE SUPERALLOYS: CARBIDE BEHAVIOR AND ALLOY PERFORMANCE By Andrew Jay Wasson August 2010 Chair: Gerhard Fuchs Major: Materials Science and Engineering Advanced single crystal nickel-base s uperalloys are prone to the formation of casting grain defects, which hi nders their practical implement ation in large gas turbine components. Additions of carbon (C) have recently been identif ied as a means of reducing grain defects, but the full impact of C on single crystal superalloy behavior is not entirely understood. A study was conducted to determine t he effects of C and other minor elemental additions on the behavior of CMSX-4, a commercially relevant 2nd generation single crystal super alloy. Baseline CMSX-4 and three alloy modifications (CMSX-4 + 0.05 wt. % C, CMSX-4 + 0.05 wt. % C and 68 ppm boron (B), and CMSX-4 + 0.05 wt. % C and 23 ppm nitrogen (N)) were heat treated before being tested in high temperature creep and high cycle fatigue (HCF). Select samples were subjected to long term thermal exposure (1000 oC/1000 hrs) to assess microstructural stability. The C modifications resulted in significant differences in microstructure and alloy performance as compared to the baseline. These variations were generally attributed to the behavior of carbide phases in the alloy modifications. 25

PAGE 26

The C modification and the C+B modificati on, which both exhibited script carbi de networks, were 25% more effective than t he C+N modification (small blocky carbides) and 10% more effective than the baseline at pr eventing grain defects in cast bars. All C-modified alloys exhibited reduced as-cast / eutectic and increased casting porosity as compared to baseline CMSX-4. The higher levels of porosity (volume fractions 0.002 0.005 greater than the baseline) were attrib uted to carbides blocking molten fluid flow during the final stages of solid ification. Although the minor additions resulted in reduced solidus temperature by up to 16 oC, all alloys were successf ully heat treated without incipient melting by modifying commerc ial heat treatment schedules. In the Bcontaining alloy, heat treatment resulted in the transformation of script MC (M metal, C carbon) carbide networks into clusters of small, spherical MC carbides without a significant change in composition. Formati on of topologically close packed phases during thermal exposure was suppressed in the B-containing alloy due the decomposition of primary MC carbides and the preferential formation of secondary M23C6 carbides. All of the modified alloys ex hibited shorter creep rupture lifetimes than the baseline at all creep conditions (950 oC/300 MPa, 850 oC/550 MPa, 750 oC/800 MPa). The most significant decrease in lif etime occurred at the 750 oC condition due to large primary creep strains of up to nearly 10% in the C-containing alloys. In HCF testing at 850 oC, the presence of carbides and increased porosity led to reduced lifetimes in the modified alloys. HIP (hot isostatic pressing) processing significantly improved fatigue performance, accounting for average lifetim e increases ranging from 77% to 4490% 26

PAGE 27

27 over Un-HIPed material. HIP also isolated the effect of carbide morphology on fatigue behavior by changing active crack initia tion sites from por es to carbides.

PAGE 28

CHA PTER 1 INTRODUCTION Progress in the gas turbine industry is driven largely by technological advances in component design and alloy development. The business of selling gas turbine engines is highly competitive, with ai rcraft propulsion engine companies vying for military contracts with gover nments and civilian contracts with commercial aircraft manufacturers. Companies producing larger industrial gas turbines for power generation contend with each other for business fr om corporations that produce and sell electricity. The highly competitive nature of the industry leads to rapid implementation of new technology as companies try to gain ad vantages over their competitors. In some cases, new processes or materials are in corporated into products without a complete understanding of how they improve performance. For exam ple, single crystal (SX) nickel-base (Ni-base) superalloys with additi ons of rhenium (Re) have been used in engine components because they provide better mechanical strength at high temperatures, but the mechanisms responsib le for the increased strength are not entirely understood. The recent incorporation of carbon (C) in SX Ni-base superalloy chemistries is another case of implementat ion without complete comprehension of all associated effects. Elements such as C and other grain boundary strengtheners are included in polycrystalline superalloys, but they are generally absent fr om SX alloys. Recently, however, these elements have been reintroduced to SX superalloys because they have been shown to reduce casting defects Although there are several proposed theories that attempt to ex plain the benefits of C, there is no es tablished mechanism describing exactly how C is so helpful in casting. Even greater uncertainty surrounds 28

PAGE 29

the effects of these elem ents on mechanical performance. Despite these gaps in understanding, the constant drive for savings has led to the incorporation of C in SX superalloy chemistries. The inclus ion of C in SX superalloys is becoming more common as gas turbine components become larger and more complicated, making them more difficult to cast without defects. Alloys containing C are par ticularly prevalent in large industrial gas turbine (IGT) engines. It is vital to improv e the knowledge of how these minor additions affect alloy performance. The presence of C leads to the formation of brittle carbide phases, which introduce a potent ial negative effect on high temperature mechanical properties. Creep rupture str ength is often the limiting des ign factor for gas turbine components that employ SX superalloys, but fatigue performance is also an important consideration that can im pact component lifetime. Se veral new technologies being developed for future aircraft propulsion, such as the pulsed detonation engine, may rely more heavily on fatigue performance due to the more prominent role of cyclic (as opposed to monotonic) loading [1 ]. It is pragmatic, t herefore, to understand how changes to SX superalloy microstructure impact both high te mperature creep and fatigue properties. The morphology of carbides is likely a significant factor in determining how C affects performance. Discovering c onnections between morphology types and associated alloy behavior would help clarify t he picture of C in SX superalloys. Minor additions of elements such as hafnium (Hf), boron (B), niobium (Nb) nitrogen (N), and zirconium (Zr), along with C, may provide the key to engineering carbide morphologies to provide the best balance of alloy castability and mechanical performance. These 29

PAGE 30

30 additions are relatively low cost, can be cont rolled to within severa l parts per million (ppm), and offer the flexibility to add uni que combinations elements at different concentrations. The purpose of this study was to examin e the effects of C modifications to a commercially relevant, 2nd generation SX superalloy (CMSX-4). Additions of C, C+B, and C+N were made, and the alloy behaviors were compared to one another and to baseline CMSX-4. The goal was to present a complete picture of how C impacts multiple aspects of alloy behavior. The im pact of heat treatment on carbide morphology was assessed and, throughout the study, effo rts were made to draw connections between carbide morphology and alloy perfo rmance. Due to the large number of micrographs, graphs, and tables presented in the following chapters, all figures have been inserted at the end of the chapters to improve reading flow.

PAGE 31

CHA PTER 2 BACKGROUND 2.1 Overview The term superalloys refers to a class of Ni, cobalt (Co), and iron (Fe) alloys first developed in the middle of the 20th century to perform at higher temperatures than traditional steels. Ni-base super alloys have been specifically tailored as the material of choice for the turbine section of gas turbine engines used for aviation propulsion and power generation. Turbine blade and vane components operate in the hottest areas of the engine in the post-combustion gas path. These alloys ar e some of the few in the world that can maintain their structural in tegrity at temperatures up to 80% of their melting point. The combinati on of oxidation and hot corrosi on resistance, resistance to creep deformation, and fatigue resistance makes t hem the ideal material choice for the high stress, high temperature envir onments in which they operate [2]. There have been several advancements in superalloy development that have been the most impactful on the industry. In addition to cooling passages and coating systems, removal of grain boundaries a llowing the production of single crystalline components and the addition of refractory elements such as Re were key innovations propelling the industry. Both changes resu lted in tremendous improvements in high temperature creep strength t hat paved the way for higher performance engines. With this improved performance, however, came new obstacles in casting. SX alloys with high refractory contents have proven to be more difficult to cast than their predecessors. In order to take advantage of newer alloys without sacrificing prof its, alloy developers have recently explored ways to improv e castability. Newly improved casting 31

PAGE 32

technologies have helped, and t he addition of minor levels of C has been identified as a cheap and effective way to reduce cast ing defects and rejection rates. 2.2 Strengthening Ni-base superalloys are precip itation harden ed alloys that consist of two phases: a matrix (FCC Ni) and high volume fraction, coherent precipitates (ordered L12 Ni3Al). Service microstructures are attained via a so lution heat treatment followed by a multistep aging heat treatment to form the precipitates. This micros tructure leads to multiple strengthening mechanisms in the alloy. Dislocation motion in the matrix is impeded by solid solution strengthening elements such as tungsten (W), Re, chromium (Cr), and molybdenum (Mo) [3]. In additi on, the various alloying elem ents, especially Co, impact the stacking fault energy (SFE) of the system. A lower SFE is desired because it produces greater separation dist ances between partial dislocat ions. Partials that are further apart are less likely to recombi ne and cross-slip [4]. The SFE becomes particularly important at el evated temperatures when a num ber of FCC slip systems can be active. Dislocations that come into contact wit h precipitates will either shear the precipitates, bypass them to continue deforming the material, or form high concentration dislocation structures at the / interface. The st rength of the interfac e is controlled by / misfit (a measure of t he difference in lattice param eters of the precipitate and matrix) and anti-phase boundary (APB) energy, which is affected by additions of aluminum (Al), Nb, tantalum (T a), and titanium (Ti). Larger / misfits lead to higher coherency strains and formation of geometrically necessary dislocations at the interface. Increased coherency strain can im prove strength, but it also reduces long term creep capabilities by increasing the driving force for phase coarsening at high 32

PAGE 33

temperatures. The A PB energy has a direct effect on the shearing process. The established mechanism for shearing of involves pairs of a/2 <110> {111} dislocations [5]. The first dislocation enters the phase and creates the disordered APB region, and the second dislocation follows along the same slip plane and restores order. These dislocation pairs are often referred to as superdislocations. A larger APB energy represents a greater barri er that must be overcome if s hearing is to occur. Alloying adjustments to improve strengt h must be made with care because they can also affect heat treatment capabilities and additional phase formations [3]. 2.3 Advancements in Superalloy Development Advances in superalloy design have been driven largely by the need for improvements in high temper ature strength required for better engine performance. The first Ni-base superalloys were equiaxed polycr ystalline in nature, but innovations in processing technology enabled the development of directionally solidified (DS) alloys. These DS alloys have a columnar grain stru cture that eliminates any grain boundaries that are transverse to the pr imary stress direction. These transverse boundaries tend to reduce creep strength as higher diffusion rate s along the boundaries enable more creep deformation. Even more recently, a shift has been made from DS to SX alloys, which contain no grain boundaries. The remova l of grain boundaries has allowed for an increase in temperature c apability. Turbine engine co mponents that can withstand higher temperatures enable greater operating efficiencies. The progression in processing over the years c an be seen in Figure 2-1 [6]. SX superalloys have been developed over the years in several different generations that represent some general alloying trends. The 1st generation SX superalloys unveiled in the early 1980s did not contain grain boundary strengthening 33

PAGE 34

elements (C, B, Zr, and Hf) that were present in their polycrystalline and DS predecessors. The justificati on at the time was that an a lloy devoid of grain boundaries did not need any elements that were original ly included to strengthen such boundaries. It should be noted that small amounts of Hf are included in most modern SX superalloys to improve oxidation resistance. The 2nd generation in the early 1990s introduced approximately 3 wt. % Re, which is an excell ent solid solution strengthener that slows down diffusional processes such as coarsening. The reasons that Re provides such significant improvements in strength are not entirely underst ood, but changes to / misfit and the formation of clusters of Re atoms are thought to be responsible [7,8]. Negative aspects of adding Re include incr eased density and cost and reduction of long-term alloy stability. The improvements associated with Re were clear, however, and further increases in Re up to about 6 wt. % characterized the 3rd generation alloys commercialized in the mid 1990s [9]. Examples of some common commercial compositions from the leading companies in SX alloy development can be seen in Table 2-1 [9]. Alloy development in the 2000s consisted of efforts to improve stability of Rebearing alloys through ruthenium (Ru) additions [10] and reductions of Re content to produce cheaper, lower density superalloys wh ile maintaining high temperature creep resistance [11]. It should be noted here that development of coating systems for both oxidation and thermal protection has become very import ant in the superalloys industry over the past 20 years [12]. Advanced coatings can improve component performance without any changes to the substrate superalloy. This study, however, will focus on SX substrate alloys and will address coatings only when directly relevant. 34

PAGE 35

2.4 Carbon Additions in Single Crystal Ni-Base Superalloys As described above, C, B, and other g rain boundary elements were removed from the first SX superalloys. Their re moval eliminated the formation of brittle secondary phases such as carbides and borides. C and B can also depress solidus temperatures, and their absence allowed for increased heat treatment temperatures to more easily achieve homogenization. With in the last decade, however, the reintroduction of C into SX superalloys has prov en useful in improving castability in the following ways [13-19]: Reducing formation of grain defects such as stray grains and low angle boundaries (LABs), which are generally defined as misorientations of less than 15o [20]. Increasing tolerance of LABs by st rengthening any boundaries that may form. Decreasing oxide and surface scale formation and hot tearing during solidification. Before the role of C in reducing casting defects is explained, an understanding of how grain defects form in SX castings wil l be developed. Formation of boundaries and unwanted grains represents a signi ficant obstacle to the so lidification of SX alloys. These harmful defects can form when a dens ity inversion occurs between the denser bulk liquid and the less dense interdendritic liquid in the mushy zone (area between 100% solid and 100% liquid). This inversion and associated turbulence in the liquid can lead to plumes of solute atoms and convecti ve instabilities [13]. These instabilities represent a disruption in t he SX solidification and can lead to boundary formation at the region impacted. The flow of liquid may c ause fragmentation of solidified dendrite arms in the liquid that become nucleation point s for new, unwanted grains. A grain that nucleates in the liquid provides a surface for another grain to form and can lead to formation of freckle chains of small, equiaxed grains [21,22 ]. A schematic [23] of this 35

PAGE 36

process and a photograph of a SX bar with a fre ckle chain are shown in Figur e 2-2. There has also been work that indicates sm all misorientation LABs can result from plastic deformation of dendrit e stems during growth [24 ]. Efforts to reduce the occurrence of these phenomena hav e included adjusting solidification velocities and thermal gradients [25]. These defects have a higher propensity to form in 2nd and 3rd generation alloys containing more heavy solute elements, such as Re, which promote density inversions [26]. Methods of improv ing casting efficiency, therefore, have demanded greater attention as SX a lloy development has progressed. Additions of C to SX alloys have been found to reduce the occurrence of solidification grain defects [ 14,13] through the forma tion of interdendritic, primary, MCtype (M-metal, C-carbon) carb ides in the mushy zone. Some reports have proposed that these carbides reduce the driving force for density inversion by consuming heavy, carbide-forming elements such as W and Ta [ 13,15]. This is t hought to favorably reduce solidification segregation and the likelihood of thermosolutal convection. Recent work, however, has shown that the addition of small amounts of C to 2nd and 3rd generation alloys did not produce any signific ant changes in segregation [14,27]. These findings suggest that reduced segregation is un likely the primary c ause of grain defect reduction, and it is more likely that the ph ysical presence of the MC carbides in the interdendritic region interferes with fluid flow and prevents harmful thermosolutal convection [14]. Carbide morphology and size therefore, can be expected to impact the effectiveness of C in reducing grain defects. Additions of C also improve defect tolerance by strengthening LABs that may unintentionally form during casting. Chen and others showed that the presence of C 36

PAGE 37

and B improved the creep ruptur e strength of a SX superallo y cast with an intentional boundary. Carbides formed at the boundary and inhibited slidi ng [28]. Scrap criterion for SX components inv olve numbers and types of grain defects as well as critical LAB angles. Misorientation angles above critical values result in decreased mechanical properties and rejection of parts. The presence of carbides improves the grain defect tolerance and casting yields. Cannon-Mu skegon, a prominent superalloy developer, has incorporated the benef its of C into one of their newe st alloys, CMSX-486. It is an alloy with chemistry very similar to CMSX4 with additions of 0.07 wt. % C, 0.015 wt. % B, and 0.005 wt. % Zr. CMSX486 has been shown to have improved creep rupture life at larger misorientation angles (Figure 2-3) than CMSX-3, a 1st generation SX alloy with no grain boundary strengthening elements [29,30]. The marketing of CMSX-486 as an alloy with improved grain defect toleranc e and good overall properties is a strong indication of the importance of C additions to current and future SX superalloys. Improved alloy cleanliness is another added benefit of minor levels of C. Through a process known as a carbon boi l, C reacts with oxygen (O) in the alloy melt to form gaseous CO, which removes O and reduces the formation of oxide scale and inclusions [16]. Oxide inclusions are highly undesirable because they can act as primary sites for fatigue crack formation and limit fatigue lifetimes [31-36]. A small am ount of C was also found to reduce the tendency for hot tearing (fracture of the alloy during solidification) in CMSX-4 [17]. This was attri buted to reduced formation of / eutectic film on the surface of the casting. Improved castability has become increas ingly important re cently as engine companies have begun to use advanced SX a lloys for larger, more complex, and 37

PAGE 38

therefore more costly components Land-bas ed gas turbines are now produced larger to keep up with growing power demands, and new blade designs involving complicated cooling passages are utilized to incr ease engine operating temperatures and performance. Any grain defects that form in the thin cross-sections between cooling holes can lead to mechanical failure. Allo y castability improvements are necessary for successful scaling-up of SX component manuf acturing, and the use of minor amounts of C has proven to be an effective way to si gnificantly reduce casting problems. As mentioned above, Cannon-Musk egon has identified the need for C in their SX alloy CMSX-486, and General Electric includes C in their 2nd generation Rene N5 alloy. Additions of C and other grain boundary elements will be necessary as engine improvements and alloy development moves forw ard, and it is important to gain an understanding of their impac t on SX superalloy systems. 2.5 Effects of Carbon Some potential negative effects come al ong with the casting benefits of C in SX superalloys. The interdendritic carbides cr edited with reducing casting defects can also be detrimental to mechanical properties [35, 37-40]. Large carbide networks can create significant stress concentrations and their brittle nature makes them prone to cracking. It is therefore critical to understand the various effects of C on superalloy systems so the potential negative effects c an be minimized. The sectio ns that follow highlight relevant work on the effects of C in superalloys and focus primarily on SX systems. Polycrystalline and DS alloys are only discussed when relevant. 2.5.1 As-Cast Characteristics As mentioned above, the majori ty of the work in the lit erature reports a direct correlation between the presence of C and the reduction of casting defects in SX 38

PAGE 39

superalloys The optimal beneficial amount of C, however, is not as clear. Al-Jarbas work with LMSX-1 (a 3rd generation alloy) revealed that additions of C up to 0.15 wt. % decreased the tendency to form solidification defects [14]. Tin studied a wide range of 2nd and 3rd generation alloys and f ound that 0.125 wt. % C was the ideal level for reducing defects, and additional levels beyond that did not lead to further improvement [13,15]. In 2004, Howmet Castings develop ed a mathematical relationship based on alloy chemistry and predicted carbide phas e formation to determine the C levels required for the best alloy cleanliness [16]. Actual C levels in newly developed SX superalloys are lower than the ideal levels fo r castability as presented in the literature. CMSX-486 contains approximately 0.07 wt. % C and Rene N5 has approximately 0.05 wt. % C [29]. Alloy development is often quite proprietary in nature, but it is likely that these reduced C levels are designed to form lower carbide volume fractions and maintain heat treatment windows. Minor additions of grain boundary elements such as C, B, and N have a tendency to reduce alloy liquidus and solidus tem peratures [41,42]. A reduced solidus temperature leads to a greater risk of meltin g during heat treatment at localized areas rich in low melting point el ements [43]. Significant decr eases in solidus temperature cause difficulties with homogeni zation as the gap between the solvus and solidus narrows. Alloy solidus temperatures incr ease during heat treatment [14], however, and heat treatments can be conducted with multiple steps and slow ramp rates to minimize the risk of melting. A variety of primary (formed during initial solidification) MC-type carbides have been reported in literature. Regardle ss of morphology, MC carbides have been 39

PAGE 40

observed only in interdendritic regions and ar e rich in refractory metal elements, particularly Ta and W. They have a NaCl-type cr ystal structure, which is very similar to the FCC and L12 structures of the surrounding / The most common two types observed are blocky, faceted carbides and larger script carbide networks of interconnected rods and plates, as shown in Figure 2-4. The script networks can be viewed as dendritic in shape because t hey form between primary dendrites with protrusions extending between se condary dendrite arms [27]. Researchers have offered several explanat ions for the formati on of different carbide morphologies, and most revolve around the temperatur e or location of formation. Formation of carb ides at temperatures near the liquidus, or high in the mushy zone, is believed to allow more time fo r growth into blocky, equilibrium structures without interference from dendrite arms [44]. A more c oupled growth of dendrites and carbides is thought to occur at lower carbid e formation temperatures and is likely to produce carbide network structures [45]. Despite these observations, a consistent connection between carbide formation temper ature and carbide morphology cannot be made. Tin reported carbide precipitation that occurred just below the liquidus temperature resulted in the fewest casting defects, but carbide morphologies were the same regardless of formation temperature [15]. Al-Jarba discove red an increase in carbide formation temperature with increas ed C content but no change to the script carbide morphology [14]. Liu, however, did observe a morphology change from blocky to script carbides when C content was increased in a 1st generation superalloy [46]. It is clear that as-cast carbide formation is not entirely understood a nd further study is needed. 40

PAGE 41

The presence of minor elemental additions is known to affect carbide precipitation behavior. The presence of N has been connected to higher formation temperatures and blockier carbides. The proposed theory invo lves carbides preferenti ally nucleating on preexisting TiN particles high in the mushy z one [44]. Minor concentrations of B have also been identified to change carbide precipit ation behavior in SX superalloys [47], but the specific effects have not been establis hed. A report by Chen and others on a 2nd generation alloy suggests that the effect of minor elements such as B and Hf is determined by how they change carbide lattice parameter [37,48]. The theory proposes that an increased carbide lattice parameter leads to larger lattice misfit between carbide and matrix, and therefore carbides will assume a blockier structure to minimize the surface area/volume ratio. This type of re lationship between changes in carbide lattice parameter and carbide morphol ogy has not been established in other alloy systems. Despite recent findings, continued efforts are required to better understand the role that minor additions play in carbide formation. Casting pores are generally accepted to negatively affect mechanical properties because they act as internal flaws that gener ate significant local stress concentrations [49]. There is disagreement in the field as to whether casting microporosity increases or decreases with the presence of C. The differing opinions revolve around the role of interdendritic carbides in the final stages of solidification, when pore formation occurs. It was reported by Liu that additions of C to a 1st generation superalloy decreased both the size and frequency of micropores [ 46]. Chen found t hat modifying a 2nd generation superalloy with C, B, and Hf significantly lowe red the microporosity [ 50]. Both of these positive results were attributed to primary carbides that formed in interdendritic pools 41

PAGE 42

and helped offset volume shrinkage through latti ce expansion by int erstitial C atoms. Work at the University of Florida has show n an increase in microporosity due to minor carbon additions [14,27,42]. The increase was attributed to interdendritic carbides blocking fluid flow during the last stages of solidification. Th is theory is supported by the observation of pores adjacent to carbides. This kind of pore-carbide spatial relationship has also been seen in Mar-M247, a DS superalloy [51]. It appears that the fluid blocking ability of carbides t hat is helpful in preventing thermosolutal convection and defects may also lead to increased porosity. It should be noted that some variability exis ts in the quantification of porosity, as volume fractions are estimated from microg raphs of two-dimensional cross sections. Synchrotron tomography is a relatively new technique to characterize pore distribution and shape in three dimensions [52], but it is not yet widely used in superalloy research. Several other microstructural features have been examined with respect to C content. Quantitative studies of as-cast mi crostructures revealed that C significantly reduced the volume fraction of / eutectic by changing eutectic formation temperature [14,53]. This is a favorable effect becaus e it improves the ability to achieve full homogenization during heat treatment. Heat treatability can also be improved with a fine dendrite structure and a small primary dendrite a rm spacing (PDAS). PDAS measurements indicate the average distance between solute rich dendrite cores, and smaller values represent shorter diffusi on lengths to achieve homogenization during solution heat treatment [26]. The addition of C has been f ound to have no significant impact on the overall dendritic structure or PDAS [45]. 42

PAGE 43

2.5.2 Microstructural Evolution All superalloys undergo carefully tailored heat treatment steps to create particular microstructures before they are put into serv ice. It is important to understand these microstructural changes and the role of C in these thermally driven processes. Lattice misfit between the and phases, which is a product of alloy composition, determines precipitate morphology and size. Precipitat ion hardening treatments generally result in cuboidal phases for alloys with high refractory content such as CMSX-4 [54]. There does not seem to be any signific ant impact of C on heat treated size and volume fraction [27,45]. Other than slight adjustm ents to homogenizing temperatures to avoid inic ipient melting, SX superalloy heat treatment schedules are not changed when C is present. Casting porosity in SX superalloys has been shown to increase during homogenization. Anton and others reported that new pores formed and existing pores grew by a diffusion-controlled mechani sm at several different homogenizing temperatures [55]. The high diffusion rates required for homogenization also allow vacancies to coalesce and new vacancies to form, which leads to the increased porosity. Although most measur ements of porosity are done in the as-cast state, this finding indicates that porosity levels in the heat treated condition may be higher, which could have an unexpected adverse impact on alloy performance. A significant amount of work and a wi de variety of resu lts have been reported regarding the stability of MC carbides during heat treatment. The established mechanism of carbide transforma tion involves the interstitial diffusion of C out of MC carbides and the formation of smaller, secondary carbides with carbide forming 43

PAGE 44

elements in the surrounding matrix [56]. Reactions involving the most common secondary carbides are shown here: MC + M6C + MC + M23C6 + Signs of carbide decomposition include MC carb ides that are smaller than in the as-cast condition and the appearance of new, small plat e-like or spherical carbides. These types of transformations have been observe d for years in polycrystalline superalloys [57-61], where the secondary carbides tend to form preferentially on grain boundaries. The products of MC decomposition seem to be tied to the relative amounts of certain elements in the matrix. The M6C carbides are rich in W and Mo, and M23C6 carbides have high levels of Cr. These same types of transformations have been observed in SX superalloys, but reports are quite varied dep ending on the alloy system and the specific heat treatment applied. Decomposition of MC carbide and precipitation of M6C carbides [18] or M23C6 carbides [39] due to a standard heat treatment of C-modified SX superalloys have been published, but MC carbides have also been found to remain stable and unchanged during hea t treatment in other SX systems [62]. Carbide changes are often reported as side effects of superalloy precipitati on heat treatments. There is a need for better understanding of ca rbide stability during heat treatment and whether connections can be made between certain heat treatments and specific carbide transformations. Long term service exposure at high temperat ures can also result in significant carbide decomposition. In this type of transformation, secondary carbides often form on or near the edges of decomposed MC carb ides. Diffusion transition zones of phase 44

PAGE 45

or phase, Ni3(Ti,Ta), can form between shrinking MC carbides and secondary carbides on the periphery [56,58,59]. T he formation of these transition zones and interfaces is believed to have a negative effe ct on mechanical properties. An example of such decomposition is shown in Figure 2-5 [59]. Long term exposure in superalloys can re sult in the unwanted formation of topologically close packed (TCP) phases, named as such because they have planes of close packed atoms that are separated by rela tively large spacings. These brittle, needle or plate-like phases remove strengt hening elements from the matrix and are susceptible to cracking. Later generation alloys with high refractory content are particularly prone to the formati on of these phases [63]. The effect of C on TCP phase formation seems to be directly related to the carbide deco mposition process. TCP phases were observed to form directly on decomposing MC carbides, which was attributed to local enrichment of TCP phase forming element s due to the decomposition process [59]. Chen, however, reported a me chanism by which carbide decomposition retarded TCP formation [37]. A SX alloy modified with small amounts of C and B formed less TCP phase during exposure than an identical alloy without modifications. The formation of secondary carbides consumed elements such as Cr that would otherwise form TCP phases. The possibility of carbides improving long term alloy stability is a promising one that demands further attention. 2.5.3 Tensile Properties Tensile properties of superalloys are la rgely controlled by interactions of dislocations with one another and with precipitates. Yield st rengths tend to initially increase with temperature as multiple octahedral {111} slip systems are activated and then decrease rapidly as thermally activat ed dislocation climb or cross-slip of 45

PAGE 46

disloc ations from the octahedral to the cubic plane reduces the effectiveness of the strengthening precipitates [64, 65]. The ultimate tensile strength follows a similar temperature trend, as seen in the Figure 2-6 plot [9]. Several other, more complicated dislocation-precipitate interaction mechanisms have also been identified as possible causes for the strength peaks [66]. In CM SX-4, the maximum yield stress occurs at around 800 oC [64]. A different set of deforma tion mechanisms are believed to operate above this temperature as compared to below it. The presence of carbides seems to only have minor effects on tensile behavior. Stepanova reported that discrete carbide particles acted to slightly increase the r oom temperature yield and ultimate tensile strengths of a 1st generation alloy [67]. Other work has indicated that script carbide network cracking led to a decrease in yi eld strength at room temperature and 950 oC [45]. The competing roles of carbides (dislo cation obstacles vs. crack formation sites) in tensile behavior seem to be connected to carbide morphology. 2.5.4 Creep Properties Creep strength is a primary controlling factor in the design of superalloys, and creep is often identified as the main cause of failures in gas turbine components [68,69]. Polycrystalline superalloys rely on the forma tion of secondary carbides to slow down diffusion and prevent sliding at grain boundaries [70-72], but effects of carbides on SX creep deformation are not as well established. Creep deformation is generally divided into three stages: primary, secondary or ste ady-state, and tertiary. A creep strain vs. time plot with all three regions is shown in Fi gure 2-7 [73]. The initial elastic strain is very minor in superalloys. Observed effects of C on each creep region will now be discussed. 46

PAGE 47

Primary creep is typified by an initia lly high creep rate followed by a rapid decrease in strain rate due to work hardening as deformation is introduced in the material [74]. Signific ant primary cr eep strains have been observed in advanced 2nd and 3rd generation alloys at inte rmediate temperatures (650 oC to 850 oC) and high stresses (above 500 MPa) [75,76]. At these conditions, shearing of precipitates and development of significant creep anisotropy c an occur [7]. The shearing of precipitates by <112> {111} dislocations creates slip heterogeneity that leads to elliptical cross sections of tested specimens [77,78]. A consequence of this mechanism is the sensitivity of primary creep to minor c hanges in crystallographic orientation. Misorientations greater than 10o from the [001] direction can lead to drastically higher levels of primary creep and shorter overa ll creep lifetimes [79] The likelihood of shearing and large primary cr eep strain has been tied to / lattice misfit, which impacts dislocation behavior at the / interface [75]. Smaller misfit values lead to fewer misfit dislocations at the interface and allow for easier shearing of precipitates. Modifications to alloy compositions result in changes to latti ce misfit values, and this is believed to be the cause of increased primary creep in later generations of SX superalloys. The role of C in primary creep has not been thoroughly explored, but possible effects prudent to examine include shear disloc ation-carbide interactions and local differences in composition caused by carbide formation. While significant work hardening occurs in the primary creep region, a balance between work hardening and recovery or damage processes is characteristic of the secondary creep regime. The creep in this region is often referred to as steady-state because the creep strain reaches a minimum, c onstant (or nearly constant) rate. This 47

PAGE 48

rate often fits to an Arrhenius-type equat ion with an activation ener gy very near the activation energy for diffusion within the system. Carbides can contribute to strengthening by acting as barriers to dislocat ion motion. Smaller MC carbides [37], secondary M23C6 and M6C carbides [8,18], and nano-scaled M23C6 carbides confined to channels [80] have been identified as the mo st effective at pinning dislocations. De-cohesion of carbides from the matrix and carbide cracking, however, may lead material damage that counteracts any carbide strengthening effects. Script MC carbide networks with thin features are more prone to cracking than blockier features [37,81]. to Tertiary creep involves the rapid incr ease of strain and a ccumulated damage that leads to necking and final failure. Cracks grow from pores or creep cavities until the cracks link up and failure occurs by void coalescence [82-84]. Carbide cracking accelerates this process by introducing mo re damage and providing additional sites for crack formation and growth [69]. Rafting is a phenomenon in which di rectional coarsening of the phase occurs under applied stress at high temp eratures (greater than 900 oC). For alloys with negative (a < a) misfits, such as CMSX-4 and mo st commercial superalloys, the channels are in compression and the is in tension near the interface. The channels parallel to the applied tensile stress direction become elastica lly strained, and rafting of perpendicular to the tensile stress direction a lleviates some of this strain and reduces the free energy [85-88]. T he high temperatures are r equired to enable the diffusion necessary to form the rafts. Small C additi ons do not significantly alter the rafting process [18] except for local interf erence of rafts at carbides [38]. 48

PAGE 49

2.5.5 Fatigue Properties Fatigue is the manner by which alloys fail from cyclic loading at stresses below the ultimate tensile stress. Fatigue failures are particularly problematic because they often occur suddenly without any clear warning sign s [4]. The process begins with crack nucleation at regions of localized strain. Cr ack growth follows until overload ductile failure occurs in the re maining cross-section. Low cycle fatigue (LCF) generally refers to fatigue in which plastic deformation occurs, and failure is generally crack growth controlled. Most LCF testing is strain-c ontrolled. In gas turbines, LCF conditions develop due to start-up and shutdown cycles and from thermal strains. High cycle fatigue (HCF) is stress-controlled and involves repeated loading below the yield stress, and failure is crack initiation controlled. High frequency vibratory stresses in gas turbines can lead to HCF conditions in blades [89]. Observed fatigue crack initiation sites in SX superalloys include pores, inclusions, and the material surface [31,32,36,89]. Surface initiations are more likely to occur in test specimens or components with rough surf ace finishes or scratches [90]. The primary strategy for improving fatigue performance of an alloy is to reduce the size and number of potential crack formation sites. Porosity can be greatly reduced through the hot isostatic pressing (HIP) process, which combines high pressure and temperature to collapse and heal casting pores [91]. During the dev elopment of CMSX-4, HIP processing was shown to improve the 750 oC HCF strength (to 107 cycles) of the alloy by approximately 50% [92]. Attempts have been made to correlate the size of crack initiating features to cyclic lifetimes, with some limited su ccess [93]. Work at the Air Force Research Laboratory involves mini mum fatigue lifetime predictions using a fracture mechanics approach in which the cr ack forming feature is assumed to be a 49

PAGE 50

circular crack [94]. Carbides represent another possible crack initiation site in Ccontaining alloys. Their brittle nature make s them prone to cracking, which can cause crack formation in the surrounding matrix and limit fatigue lifetim e [35]. A failure analys is conducted by Silveira demonstrated that even when they do not act as the primary initiation site, cracked carbides can a ccelerate the onset of failure, as shown in Figure 2-8 [40]. Localized plastic deformation near carbides has been recently identified as an important aspect of the fatigue process. The stress concentrations in the vicinity of carbides can lead to the formation and motion of dislocations that locally deform the microstructure at macroscopic stresses bel ow the yield stress [36]. These deformed regions represent a localization of strain that can accelerate the crack initiation process [39]. Continued load cycling can increase the deformation near carbides and develop strains at the interface of the carbide and the / matrix. De-cohesion of carbide from the matrix can occur if the interfacial strain becomes large enough. This separation event creates a site for a fa tigue crack to form, propagate, and lead to failure [95,96]. Additional study is needed to explain the role of localized deformation at carbides on the overall fatigue behavior. Small amounts of B, along with C, have been shown to impact fatigue behavior of IN 718, a polycrystalline superalloy. T he addition of 29 ppm B resulted in longer thermo-mechanical fatigue (TMF fatigue invo lving both strain and thermal cycling) lifetimes, which was attributed to reduced phase coarsening [97,98]. Fatigue crack growth (FCG) rates significantly decreased in IN 718 with 29 ppm B as compared to 12 ppm B. Clusters of B atoms were credited with retarding dislocation slip, which created 50

PAGE 51

a tortuous crack path and improv ed fatigue lifetimes [99]. This report is unique from others involving minor additions and mechanical properties becaus e it connects changes in behavior to atoms in the lattice inst ead of carbides. It is unclear if B has these effects on fatigue in SX systems. The variety of observed fatigue mechani sms operating in superalloys and other metallic systems has led some to reconsider traditional methods of representing fatigue behavior. Fatigue lifetimes are frequently pl otted versus stress amplitude or stress range to generate S-N (S stress, N cycles to failure) curves. Generally one curve is generated from test data for a given alloy, and this curve is used for design. Ravi Chandran recently published a paper that explored the possibility of multiple S-N curves to represent different failure mechanisms o ccurring in the same alloy. Significantly shorter lifetimes were observed for samples with surface crack init iations than for those with internal initiations. The failure map s hown in Figure 2-9 repr esents the competition amongst crack initiators and how certain condi tions can activate par ticular mechanisms [100]. Microstructural features control fatigue initiation and ultimately the overall fatigue behavior of the alloy. Carbides have an import ant role in these initiation processes that must be clarified to fully understand the ramifi cations of including C in SX superalloys. 2.6 Summary The improvements in castability are enough to justify additions of C to SX superalloys, but the potential negative effects on mechanical performance should not be overlooked. In an effort to investigate this topic, a t horough study of C-modified CMSX4 has been conducted. High temperat ure mechanical testing and detailed characterization were used to understand the ways in which C impacts alloy behavior. This investigation also sheds some light on the viability of employing minor elemental 51

PAGE 52

52 additions and modified heat treatments to change carbide morphology which in turn can result in improved properties.

PAGE 53

Figure 2-1. Schematic show ing the progression in processing of superalloys. A) polycrystalline to B) directionally solidified grains to C) single crystal alloy (Adapted from [6]). Table 2-1. Compositions of some spec ific single crystal nickel-bas e (Ni-base) superalloys in wt% [9]. Alloy Generation Co Cr Mo W Ta Re Al Ti Nb Hf Ni Pratt & Whitney PWA 1480 1st 5.0 10.0 4.0 12.0 5.0 1.5 Bal. PWA 1483 1st 9.0 12.8 1.9 3.8 4.0 3.6 4.0 Bal. PWA 1484 2nd 10.0 5.0 2.0 6.0 9.0 3.0 5.6 0.1 Bal. General Electric Rene N4 1st 8.0 9.0 6.0 6.0 4.0 3.7 4.2 0.5 Bal. Rene N5 2nd 8.0 7.0 2.0 5.0 7.0 3.0 6.2 0.2 Bal. Rene N6 3rd 12.5 4.2 1.4 6.0 7.2 5.4 5.8 0.2 Bal. Cannon-Muskegon CMSX-2 1st 5.0 8.0 0.6 8.0 6.0 5.6 1.0 Bal. CMSX-3 1st 5.0 8.0 0.6 8.0 6.0 5.6 1.0 0.1 Bal. CMSX-4 2nd 9.0 6.5 0.6 6.0 6.5 3.0 5.6 1.0 0.1 Bal. CMSX-10 3rd 3.0 2.0 0.4 5.0 8.0 6.0 5.7 0.2 0.1 0.0 Bal. 53

PAGE 54

Figure 2-2 Represent ations of grain defects in single crystal (SX) superalloys. A) A schematic showing how density inve rsion between bulk and interdendritic liquid can lead to convective currents and stray grain formation [23] and B) example of a freckle chain. Figure 2-3. Rupture life vs. misorientation for CMSX3 and CMSX-486 tested at 982 oC/248 MPa (Adapted from [29,30]). Figure 2-4. The two most commonly observe d MC (M metal, C carbon) carbide morphologies in SX superalloys. A) sma ll and blocky and B) script network. 54

PAGE 55

Figure 2-5. Decomposition of MC carbide a fter long term exposure (Adapted from [59]). Figure 2-6. Tensile yield st rength and ultimate tensile st rength vs. temperature for 2nd generation CMSX-4 (points) and 3rd generation CMSX-10 (lines) SX superalloys (Adapted from [9]). 55

PAGE 56

Figure 2-7. Typical creep strain vs. ti me curve showin g the three creep regions (Adapted from [73]). Figure 2-8. SEM micrograph of the growth of a fatigue crack across broken carbides (Adapted from [40]). 56

PAGE 57

Figure 2-9. Failure map showing differ ent curve behavior for different fatigue mechanism s (Adapted from [100]). 57

PAGE 58

CHA PTER 3 EXPERIMENTAL PROCEDURE The experimental methods, techniques, and instrumentation used to carry out this study are described below. The aim of this c hapter is to provide not only the details of the procedures, but insight into the motiva tion behind them. A brief discussion of how the procedures were developed is included. 3.1 Materials The alloys examined in this study were cast as part of a larger research effort to study the role of C in Ni-base superalloys. One phase of the study involved the addition of varying levels of C to a model 3rd generation SX superalloy [45,27]. The work represented here involves anothe r aspect of the study examin ing the additions of B and N along with C to CMSX-4, a common commercial Ni-base superalloy. As discussed previously, CMSX-4 is a 2nd generation SX superalloy with approximately 3 wt. % Re content included to improve high te mperature creep strength over 1st generation alloys. It was chosen for study because it is widely used throughout the high temperature alloys industry. The baseline alloy and the C-containing modi fications were cast at PCC Airfoils in Minerva, Ohio from the same master heat of CMSX-4, wit h composition as shown in Table 3-1. All of the minor additions we re made just prior to pouring for optimal composition control. Additions were made by wrapping t he appropriate powder (graphite powder C, boron powder B, CrN powder N) in Ni-foil and adding it to the melt just prior to pouring. The desir ed and actual elemental contents of the modifications are shown in Table 3-2. Cylin drical bars, each with a length of 12.5 cm and a diameter of 1.25 cm, were cast in multi-bar cluster mold s using the Bridgman 58

PAGE 59

technique in the [001] orient ation. The Bridgman method for DS and SX casting is a commonly used technique that involves care fully controlled withdrawal rates of the casting from the furnace to c ontrol solidification [5]. A wit hdrawal rate of 20 cm/hr and a thermal gradient of 30 40 oC/cm were used for the test bars. A total of 80 single crystal bars (20 for each alloy variation) we re cast and each was subject to the standard quality control practices at PCC Airfoils. Each bar was macroetched to reveal any surface casting defects, and Laue back reflection was used to confirm that the crystallographic orientati on of each bar was within 10o of the [001] di rection. Small deviations from the desired casting ori entation have been shown to have minimal effects on mechanical properties of SX superalloys [79], and minor misorientations are permitted in order to improve casting yields. A table with the misorientation angles for all alloy bars used in this study c an be found in Appendix A. The angle is determined from the three other angles and is defined as the deviation between the reference direction (along the longitudinal axis of the bar) and the [001] direction [101]. The other angles are defined in Appendix A. 3.2 Heat Treatments The unique high temperature strength of Ni-base superalloys is attributed to the two phase / microstructure. This structure, c onsisting of a high volume fraction of coherent, cuboidal precipitates in a matrix and very fine secondary particles in the channels, is produced using a series of care fully controlled heat tr eatments. The alloy is first homogenized with a high temperature solution heat treatment (SHT) designed to solution precipitates that form during soli dification and to reduce the chemical segregation that arises duri ng the casting process. After homogenization, a two-step aging heat treatment is applied to form the precipitates. An example of the fully heat 59

PAGE 60

treated, two phase microstructure ( darker phase, lighter phase) is shown in Figure 3-1. It should be noted that the high rates of cooling ensured that the observed microstructural changes occurred at high temperature and not dur ing the cool down process. 3.2.1 Solution Heat Treatment (SHT) The goal of the SHT is to reduce chem ical segregation present after dendritic solidification and to dissolve the phase, including the / eutectic. Later generation SX superalloys have high contents of heavy refr actory elements such as Re, which can lead to increased elemental segregation and slower atomic diffusion. High temperatures and longer heat treatment times are required to induce atomic diffusion sufficient enough to homogenize these newer high performance alloys [26]. The maximum temperature for a particular SH T is selected to be greater than the solvus temperature and below the solidus of the alloy. This temperature r ange is often referred to as the heat treatment wi ndow, and it was determined in this study from the results of differential thermal analysis (DTA) [41]. The DTA technique involves the application of identical thermal cycling to a sample and a reference material. The temperature differentials are plotted and abrupt changes ar e identified at temperatures where phase changes occur. DTA curves were used to identify the solvus temperature range (near 1295 oC) and alloy solidus temperature (1343 oC). The carbon modifications slightly lowered the solidus temper ature, and this was account ed for when determining the maximum SHT temperature. Ramping to the maximum SHT temperatur es was carried out using a series of steps at progressively higher temperatures to prevent incipient melting. Incipient melting is localized melting at areas enriched in elements with lower melting points [43]. 60

PAGE 61

This creates discontinuities in the microstr ucture that can be detr imental to mechanical performance [102]. Slow ramp rates and hold times at i ntermediate temperatures allow for diffusion and the reduction of segregation dur ing the SHT to reduce the likelihood of incipient melting. The heat treatments used in this study were modifications of a commonly used, eight-stage commercial SHT for CMSX-4, seen in Figure 3-2 [103,104]. Two separate treatm ents, shown in Table 3-3, were developed: a low temperature (LT) SHT with a maximum temperature of 1310 oC and a high temperature (HT) SHT with a maximum temperature of 1325 oC. The LT treatment was designed to obtain good homogenization of the alloy without strongly affecting carbide morphologies, and the HT treatment aimed to break down large, script carbide networks into smaller features. Work with the polycrystalline superalloy M963 found that increasing the SHT temperature led to incr eased decomposition of primary MC carbides [61]. The majority of mec hanical test specimens underwent the high temperature SHT. The high temperatures and precise temper ature control required to conduct a SHT necessitated the use of an Elatec vac uum furnace system that is capable of temperatures up to 1400 oC and vacuum levels lower than 10-4 Torr (0.133 Pa). The temperature of the furnace was controll ed by a Honeywell controller to within approximately .5 oC of the setpoint by two thermoc ouples lowered very close to the surface of the test bars. A third thermocouple was used as a survey that monitors the front end of the furnace hot zone. All thr ee thermocouples were type C and consisted of tungsten and rhenium sheathed in individual molybdenum jackets. Cooling was done through the injection of ultra high purity heliu m and the forcing of the gas over the bar samples with a fan to produce cooling rates greater than 250 oC/minute. The high 61

PAGE 62

cooling rate minimizes precip itation and coarsening of the phase. The outer walls of the furnace and the copper electrical c onnect ions for the graphite heating elements were H20-cooled. 3.2.2 Aging Heat Treatments After homogenization, a two-step aging heat treatment wa s performed on all alloys to produce the desired micros tructure. A primary age at 1140 oC for 6 hours formed a high volume fr action of strengthening precipitates, and t he secondary age at 871 oC for 20 hours produced fine, secondary precipitates in the channels. All aging heat treatments we re performed in air in Carbolite box furnaces. The temperatures and times of heat treatment were low enough that oxidation was not considered to be a problem. Temperature control in the box furnaces was maintained using two K-type thermocouples directly in c ontact with the alloy ba r surface, which led to control of temperature to within oC. Quenching was performe d by rapid removal of bars from the furnace for air cooling (AC) on alumina racks. This method produced cooling rates in excess of 100 oC/min fast enough to avoid undesired precipitate coarsening. 3.2.3 Long Term Exposure The long term microstructural stability of baseline CMSX-4 and the modified compositions was examined by thermally ex posing fully heat treated alloys in air at 1000 oC for 1000 hours. Structural changes due to exposure that were evaluated included coarsening, carbide dec omposition, and formation of TCP phases. All thermal exposures were done in the Carbolit e box furnaces using the same technique as the aging treatments. 62

PAGE 63

3.2.4 Hot Isostatic Pressing (HIP) Selected alloy bars underwent a HIP treatment at PCC Airfoils in Minerva, OH. The HIP process is commonly used in the superalloy industry and involv es applying isostatic pressure at elevated temperatures to effectively reduce or eliminate casting porosity and improve mechanical properties [105-107]. In this study, reducing casting porosity helped to isolate the effect of ca rbides on mechanical performance. The HIP cycle conducted at PCC is shown in Table 3-3. Due to the risk of incipient melting at the highest temperature of the HIP cycle, 1313 oC, the bars underwent SHT before being sent for HIP processing. The cooling rate at the end of the HIP cycle was slow enough that some coarsening may result. To elimi nate this coarsening, bars that were HIPed underwent a partial solution heat treatment (Table 3-3) before aging. The maximum temperature of the partial soluti on heat treatment was the same as the maximum temperature of the SHT for a given test bar, but slow heat up rates and lower temperature holds were eliminated. 3.3 Sample Preparation and Ch aracterization Instrumentation A key component of this study was the ex amination of alloy microstructure at different stages of processing or testing in order to correlate structure with observed material behavior. Metallographic samp les were prepared from alloy bars and mechanical test specimens at every conditi on before being analyze d with a variety of characterization tools at the University of Florida Major Analytical Instrumentation Center (MAIC). 3.3.1 Metallography Metallographic samples were prepared fr om as-cast, post-SHT, fully heat treated, and thermally exposed samples cut from alloy bars and from all tested 63

PAGE 64

specimens. All sectioning was per formed using liquid cooling to minimize the increase in sample temperature. Bars were cut into 1 cm thick sections using a LECO abrasive cutoff wheel, and test specimens were se ctioned using an Allied diamond saw. One half of each failed test specimen was used to section off the fractu re surface, and the other half was used to generat e longitudinal cross secti ons of the gage section. Observation of longitudinal cross sections was conducted on regions less than 5 mm from the fracture surface. All fracture surfaces were cleaned in an ultrasonic methanol bath and hot glued onto aluminum stubs for exam ination. All metallographic specimens were mounted in diallyl phthalate and polished using silicon ca rbide paper with successively finer grits from 240 to 1200. Final polishing was done using alumina powder (5 m, 3 m, 1 m, and 0.3 m) slurry suspensions on polishing cloths. Samples were rinsed with water between each polishing step and cleaned in an ultrasonic bath of methanol after the final step. Most polished samples were light etched with Pratt & Whitney #17 etchant (100 mL H20 + 100 mL HCl + 100 mL HNO3 + 3 g MoO3). This etchant preferentially attacks the phase and reveals the / structure (Figure 3-1) and features such as eutectics and carbides. These flat cross-sections, however, can lead to misunderstandings of carbide morphology. For example, a carbide network consisting of cylindrical rods may appear as a group of discrete r ound particles when viewed in cr oss-section. Therefore, a deep etch was also used to further rev eal carbide morphologies. The deep etch procedure involved submerging a polished sample in a mixture of 30 mL H20 and 80 mL HCl and then adding 60 mL of H2O2. The procedure must be performed in a fume hood 64

PAGE 65

due to the violent reaction that occu rs on the sample surface as both and are removed to expose the carbides. A com pari son of carbide appe arance using each of the etchants is shown in Figure 3-3. 3.3.2 Optical Microscopy A LECO light optical microscope was used for inspection of dendrite structures and carbides in the as-cast condition and during the heat treat ment development process. It was particularly useful for obs ervation of residual eutectics that indicated incomplete homogenization during SHT. The optical microscope was also used occasionally during metallographic preparation to confirm the effectiveness of the final polishing steps. 3.3.3 Scanning Electron Microscopy (SEM) Light etched and deep etched metallograph ic samples as well as fracture surfaces were characterized extensively using a JEOL 6400 scanning electron microscope (SEM). Exposed mounting materi al was coated with graphite paint to allow a path for electrons to travel and to avoid extensive charging. An accelerating voltage of 15 kV and working distances 25 mm (fra cture surfaces) and 15 mm (metallographic samples) were used. Larger working distance s allow for larger depth of field to capture features at various heights on the fracture surfaces. The majority of the SEM images were c aptured in secondary electron (SE) mode, which uses secondary electrons emitted from the sample by a variety of inelastic scattering events to produce good topographical contrast. SE mode was particularly useful in producing images that were thr ee-dimensional in appearan ce of the various carbide morphologies observed in deep etc hed samples. Backscattered electron (BSE) mode was also used to detect compositiona l differences between various phases or 65

PAGE 66

regions of a sample. BSE mode detects el ectrons emitted from the sample due to elastic interactions. Areas with higher conc entrations of heavier elements elastically scatter more electrons to produce a greater BSE sig nal and appear brighter in contrast than areas with lig hter elements. Qualitative and semi-quantit ative chemical information was obtained using the energy dispersive spectroscopy (EDS) det ector on the SEM. El ements present at a particular point or along a line scan can be detected and their relative compositions can be determined. EDS analysis was used prim arily to determine elements present in various carbide and TCP phases. More prec ise compositional data was collected for certain phases using wavelength dispersi ve spectroscopy (WDS) on a JEOL 733 Superprobe with a beam voltage of 12 kV and a beam size of 0.5-1.0 m. All WDS measurements were taken from polished and un-etched samples. 3.3.4 X-Ray Diffraction (XRD) The types of carbide phases present in the alloys at various heat treated conditions were identified using X-ray diffraction (XRD). Thin disk samples approximately 10 mm thick were cut from as-cas t alloy bars. One slice of each of the carbon-containing modifications was analyzed at each stage of heat treatment and after long term exposure. The thin slices we re polished and deep etched as described above in order to reveal as much carbide surface as possible for x-ray interaction. Samples were attached to glass slides with double stick tape before insertion into the XRD chamber. A Phillips APD 3720 automated diffractometer system using Cu K radiation was used to examine all samples at 40 kV a nd 20 mA. The step size of the scans was 0.02o with a dwell time of 1 second. Seve ral initial scans were performed from 2o to 135o to identify peak locations of interest, and all subsequent scans were run from 20o 66

PAGE 67

to 80o. Signal intensities were plotted against 2 angles and diffraction peaks were indexed. Spectra were compared with data in the Personal Computer Powder Diffraction File (PCPDF) database and with li terature data to confirm results. 3.3.5 Transmission Elect ron Microscopy (TEM) Dislocation structures of mech anically tested material and the finest microstructural features were examined using transmission electron microscopy (TEM). TEM samples were prepared by using the focu sed ion beam (FIB) to extract thin foils from polished and light etched cross sections of HCF test specimens. The FIB uses a gallium ion beam to mill thin rectangular sa mples from the bulk. Superalloy samples require more thinning, to approximately 100 m thickness, than other materials due to a relatively high average atomic number that makes electron transparency more difficult. FIB lift-out samples were pos itioned onto copper grids cover ed with carbon film. All TEM samples were oriented parallel to the stress axis and therefore near the [001] zone axis. A JEOL 200CX at an accelerating voltage of 200 kV was used for the majority of the TEM analysis. Images were captur ed to qualitatively observe dislocations, carbide-matrix interfaces, and very fine secondary precipitates. A JEOL 2010F high resolution TEM equipped with scanning trans mission electron microscopy (STEM) and energy dispersive x-ray (EDX) capabilities was utilized for quantitative compositional data of phases observed in TEM. 3.4 Quantification In addition to qualitative observations of microstructure at various conditions, some quantitative measurements were conduct ed to compare the alloy modifications. 67

PAGE 68

Quantification methods were determined from literature research and from past experience with superalloys. 3.4.1 Carbide and Po rosity Volume Fraction As cast carbide and pore volume frac tions were determined using a manual point counting method as described in ASTM standar d E562-05[108]. A fine rectangular grid with 1225 line intersections was overlaid on SEM BSE micrographs taken at 2000X magnification for carbides and 50X for pores and the number of intersections at carbides or pores was counted to estimate the volume fraction. Ten fields of view were used for each alloy. 3.4.2 Volume Fraction and Size The average sizes of precipitates were measured before and after exposure to determine the effect of the minor additions on coarsening behavior. Volume fraction was determined using the ASTM method described above and size was evaluated using the mean linear intercept method. Ten lines were drawn on each of the 10 fields of view (10,000X magnification) for a tota l of 100 lines for each sample. The following equation was used to determine average precipitate size: = 4*Vv/Sv (Eqn. 3-1) where is the average size, Vv is volume fraction, and Sv is area per unit volume found by multiplying the number of intersections per line length by 2. Coarsening rates (K) were determined from the values for the fully heat treated condition and the heat treated and exposed condition using the following equation: K = ( exposed 3 heat treated 3)/t (Eqn. 3-2) 68

PAGE 69

3.4.3 Fracture Surfaces Noteworthy features observed on fractu re surfaces were measured using the public software program Image T ool from the University of Texas Health Science Center (UTHSC). The program allows the user to import micrographs and calibrate measuring tools to determine the actual l engths of features. The sizes of the features at the crack origin for all HCF tests were found by meas uring the major and minor axes to estimate the elliptical area. The shortest distance from the fatigue crack origin to the surface of the specimen was also measured for every HCF test. The same UTHSC software was used to measure and compare micropores on the fracture surfaces of creep and tensile te sts to quantify how pores may change during creep testing. The pores were assumed to be elliptical, and a minimum of 10 pores were measured for each sample. 3.5 Mechanical Testing All tensile, creep, and HCF testing wa s conducted in the High Temperature Alloys Lab (HTAL) at UF. Te st conditions were determined through a detailed literature review of previous work involving SX Ni-base superalloys. All testing was conducted on fully aged material that had undergone either t he LT or HT SHT and was either in the HIPed or Un-HIPed condition. The majority of testing was on the HT SHT material. 3.5.1 Specimen Machining Heat treated bars were sent to Joliet Me tallurgical Laboratorie s for machining into cylindrical test specimens. Samples for all ty pes of testing were machined to the same geometry, as shown in the drawing in Figur e 3-4. Each test bar yielded 2 test specimens. Final machining steps were done using low stress grinding techniques to prevent residual stresses at the sample su rface. Specimens were carefully measured 69

PAGE 70

with digital calipers after machining to veri fy dimensions. The tensile and creep tests were run with specimens in the as-machi ned condition. All HCF specimens underwent additio nal polishing of the gauge section using progressively finer silicon carbide paper up to 1200 grit. This was done to improve surface finish and promote fatigue crack initiations at interior features as opposed to the specimen surface. 3.5.2 Tensile Testing Tensile testing was performed at 850 oC in air on baseline CMSX-4 and all 3 modifications in the Un-HIPed, fully heat tr eated condition (HT SHT). Results of tensile tests were used to help develop conditions for creep testing. A servo-hydraulic Instron load frame with a Satec 8800 series controlle r and a Merlin data collection software program was used. The Merlin program acquired load data fr om the load cell and strain data from the knife edge extensom eter secured to the gauge sect ion of the test sample. Stress-strain curves were generated from the data and yield strengths, ultimate tensile strengths, and elongations we re compared between samples. A constant cross-head speed of 0.25 cm/min. was employed for all tests. Before loading into the test frame, a ll coupling threads on the test specimen and pull rods were sprayed with boron nitride high temperature lubricant. After the sample was secured into the test frame, two K-type thermocoup les were tied onto the gauge section with Ni-Cr wire. Heat was applied to the setup with a clamshell furnace with temperature capability up to 1000 oC that was wrapped around the test specimen. Fiberfrax insulation was draped over the outside of the fur nace for insulation. The furnace controlled the temperature during testing to within .7 oC. One of the thermocouples was connected to the furn ace for the control and the other was connected to a handheld thermocouple reader to verify specimen temperature. 70

PAGE 71

Samples were allowed to soak for thirty minutes once the test temperature was reached to ensure uniform temperature befor e the tensile test was run. 3.5.3 Fatigue Testing Stress-controlled HCF tests were per formed on Un-HIPed and HIPed HT SHT material and HIPed LT SHT material. HCF lo ading conditions develop in turbine blades due to variations in mean stresses caused by flame impingement or due to vibratory stresses [89]. HCF failures ar e characterized by crack initiation at small defects and rapid crack propagation [96]. These types of fatigue failures are a primary concern with a microstructure that contains brittle se condary phases, such as carbides, that are present in the modified CMSX-4 alloys. The HCF test matrix is shown in Table 3-4. Tests were conducted using the same Instro n servo-hydraulic frame and controller as the tensile tests. All tests were conducted at 850 oC using the clamshell furnace as in tensile testing. After the sample was se cured into the frame and the thermocouples attached while the specimen was under load c ontrol, a small posit ive load was applied to the sample to prevent compression during heating. Once the specimen soaked at 850 oC for thirty minutes, a positive mean load was applied and triangular waveform was activated to produce the HCF conditions. The mean load and wave amplitude were determined for each test by measuring t he specimen diameter and calculating the values needed to produce a stress range of 690 MPa and an R-ratio of 0.1. The cyclic applied stress conditions are represented in Figure 3-5. The first round of HCF tests (one test for each alloy vari ation) was conducted at 20 Hz. At this frequency, how ever, the maximum and minimum load endpoints were not reached. A frequency of 15 Hz was used for a ll remaining HCF tests to reach stresses nearer to the desired input test conditions. The number of cycl es to failure was 71

PAGE 72

recorded when the specimen fractured. Sa mples were removed from the heat as quickly as possible to minimize oxidation, which can hinder im aging of fracture surfaces. 3.5.4 Creep Testing Creep tests were performed in air at 750 oC/800 MPa, 850 oC/550 MPa, and 950 oC/300 MPa. The conditions were chosen to study the effect of carbides on creep behavior at different regimes with varying def ormation mechanisms. Most creep testing was done on Un-HIPed HT SHT material, but several tests at the intermediate condition were also run on HIPed HT SHT alloys. All tests ran until rupture except for one test that was interrupted at about 1% strain. The creep test matrix is shown in Table 3-5. Testing was conducted with Satec M-3 creep frames equipped with NuVision Mentor software. Loading occurs by adding weights to a pan that is connected to a lever arm to apply a load to the sample that is 16 times greater than the pan load. All threads were lubricated as described previously to fac ilitate sample removal after testing. An extensometer was attached with screws that fit into small indentations machined near the threads of the sample. This extens ometer was connected to a Linear Variable Differential Transducer (LVDT) to m easure displacements. Three K-type thermocouples were tied to the gauge sectio n, each connected to one of three zones of the clamshell furnace. The sample was heated under a small pre-load until all thermocouples read within oC of the test temper ature. After a one hour soak, the test load was applied in a step fashion by adding weights to the pan. An elastic modulus value at the test temperat ure was determined from strain measurements taken after each loading step. Creep strain data was collected every 12 seconds during the first hour and every minute for the remainder of t he test. This data was used to produce creep strain vs. 72

PAGE 73

73 time and creep strain rate vs. creep strain pl ots. Other data reco rded included elastic modulus during loading and time to specific creep strain levels.

PAGE 74

Table 3-1. Chemical compositions of master heat of C MSX-4. Element Ni W Mo Re Al Ti Ta Co Cr Hf wt. % 61.4 6.40 0.60 2.90 5. 64 1.03 6.60 9.60 6.40 0.10 Table 3-2. Amount of minor elements in bas eline CMSX-4 and modifications Carbon (wt.%) Boron (wt.%) Nitrogen (wt.%) CMSX-4 Variation aim actual aim actualaim actual Baseline 0.000 0.002 trace 0.0004trace 0.0002 C 0.050 0.066 trace 0.0011 trace 0.0003 C + B 0.050 0.063 0.005 0.0068 trace 0.0003 C + N 0.050 0.063 trace 0.0011 0.0025 0.0023 Figure 3-1 SEM micrograph of CMSX-4 after solution heat treatment (SHT) and twostep aging heat treatment. 1270 1280 1290 1300 1310 1320 1330 020040060080010001200time (minutes)T (oC)1321oC Figure 3-2. Temperature profile for common commercial SHT for CMSX-4: 1277oC/2 hr 1288oC/2 hr 1296oC/3 hr 1304oC/3 hr 1313oC/2 hr 1316oC/2 hr 1318oC/2 hr 1321oC/2 hr Gas Furnace Quench (GFQ). 74

PAGE 75

Table 3-3 All heat treatments cycles used in this study. GFQ is Gas Furnace Quench and AC is Air Cool. Cycle SHT HIP Partial SHT Age HT SHT 1280oC/2hr 1296oC/2hr 1310oC/2hr 1318oC/2hr 1322oC/2hr 1325oC/4hr/GFQ 1140oC/6hr/AC + 871oC/20hr/AC HT SHT HIPed 1280oC/2hr 1296oC/2hr 1310oC/2hr 1318oC/2hr 1322oC/2hr 1325oC/4hr/GFQ ramp to 1313oC/103 MPa AC 1200oC/10min 1280oC/1hr 1325oC/1hr/GFQ 1140oC/6hr/AC + 871oC/20hr/AC LT SHT HIPed 1280oC/2 hr 1290oC/2 hr 1295oC/2 hr 1300oC/2hr 1305oC/4hr 1310oC/4hr/GFQ ramp to 1313oC/103 MPa AC 1200oC/10min 1280oC/1hr 1310oC/1hr/GFQ 1140oC/6hr/AC + 871oC/20hr/AC Figure 3-3. Scanning electron microscope (SE M) micrographs of carbides. A) light etched condition and B) deep etched condition. 75

PAGE 76

Figure 3-4. Drawing for cylindrical spec imen used for all mechanical tests. Table 3-4 High cycle fatigue (HCF) test matr ix with sample identification. Tests shown in italics were conducted at 20 Hz. All others were conducted at 15 Hz. Alloy Modification Heat Treatment Cycle Specimens HT SHT 006-2 002-1 002-2 HT SHT HIPed 011-1 015-1 015-2 Baseline LT SHT HIPed 014-1 004-2 HT SHT 021-2 022-2 022-1 HT SHT HIPed 038-1 037-1 037-2 C only LT SHT HIPed 033-2 033-1 HT SHT 042-2 043-1 043-2 HT SHT HIPed 056-1 057-2 057-1 C+B LT SHT HIPed 044-1 055-1 HT SHT 062-2 061-2 061-1 HT SHT HIPed 065-1 076-2 065-2 C+N LT SHT HIPed 071-1 079-2 Figure 3-5 Triangular waveform used in hi gh cycle fatigue (HCF) testing. 76

PAGE 77

77 Table 3-5. Creep test matrix. Specimens Alloy Modification Heat Treatment Cycle 750 oC/800 MPa 850 oC/550 MPa 950 oC/300 MPa HT SHT 019-1, 019-2 012-2 012-1 Baseline HT SHT HIPed 011-2 HT SHT 040-1 039-2 039-1 C only HT SHT HIPed 038-2 HT SHT 051-1, 051-2 048-2 048-1 C+B HT SHT HIPed 056-2 HT SHT 073-1 069-2 069-1 C+N HT SHT HIPed 076-1

PAGE 78

CHA PTER 4 RESULTS: MICROSTRUCTURAL EVOLUTION The following sections describe the microstruc tures of the four alloys in the as-cast condition and at various stages of heat treatment and thermal exposure. The / structure and secondary phases are covered wit h particular focus on the carbides in the C-modified CMSX-4 alloys. 4.1 As-Cast Characteristics In order to characterize the as-cast micr ostructures, alloys were first analyzed in both the polished and light etched conditions on the optical microscope and the SEM. Further details of carbide phase formati on were obtained after deep etching and thorough SEM analysis. 4.1.1 Differential Thermal Analysis (DTA) Phase formation temperatures were determi ned using DTA curves for all alloys in the as-cast and SHT conditions. The temperat ure values are shown in Table 4-1. In general, the C additions reduced the solidus and liquidus temperatures as expected [41,42,46]. The B-containing alloy exhibited t he lowest solidus temperature and therefore the greatest risk of incipient melting during heat treatment. The solidus and liquidus both increased after SHT due to t he reduction in segregat ion that occurred during solidification. T he carbide formation temperatures were all within 7 oC of each other and between 13 oC and 20 oC below the liquidus temperature. This indicates that MC carbide formation occurred at the same general height in the mushy zone for all three modifications. 78

PAGE 79

4.1.2 Dendrite Structure All alloys e xhibited dendritic solidificati on microstructures that are typical for Nibase superalloys, whose complicated chemical compositions lead to solidification instabilities. The dendrites grow vertically in the [001] directi on and solidify before the interdendritic regions. This process leads to significant segregation, with dendrite cores rich in heavier elements with higher melting points. Optical micrographs of transverse cross sections for all variants indicate similar structures, as shown in Figure 4-1. The dendrites have a lighter contrast than the inderdendritic regions. Secondary and tertiary dendrite arms formed in all alloys, but carbid es in the modifications impeded arm growth in some regions. Carbides, / eutectic, and pores (clearly visible as black circles) are all found in the interdendritic regions. The micrographs in Figure 4-2 were captured from longitudinal cross sections and show carbides (dark phases) between the primary dendrites and between secondary arms. Eu tectic regions (light phases) in the interdendritic area represent t he last liquid to solidify. 4.1.3 / Eutectic Reduced / eutectic has been reported as one of the beneficial results of adding C to SX superalloys [14,46]. Optical micr oscopy, where eutectic appears light (Figure 4-3A), indicated overall eutectic content, and SEM analysis, where eutectic appears dark (Figure 4-3B), provided a closer look at individual eutecti c pools. The modified alloys contained fewer and smaller regions of / eutectic than baseline CMSX-4, and eutectics were often observed in direct proximit y to carbides, as shown in Figure 4-3C. Semi-quantitative EDS analysis revealed eutectic areas to be rich in relatively low melting point elements such as Ni, Co, and Al. 79

PAGE 80

4.1.4 Casting Porosity Additions of C have been reported to either decrease [46,50] or increase [14,27,42] casting por osity in SX superalloys Determining the connection between C and porosity is central to understanding the impact on mechanical properties. Detailed metallographic examination of multiple samples for each alloy variant enabled thorough characterization of pores. The porosity volu me fractions were slightly higher in the modified alloys than the baseline, as shown in Table 4-2. Volume fraction estimates were calculated from SEM micrographs in BSE mode of polished and un-etched transverse cross sections, as shown in Fi gure 4-4. All observed pores were in interdendritic regions and were circular or elliptical in shape. This pore morphology indicated that the pores likely formed due to microshrinkage effects and not from gas formation during solidification. Pores in the modified alloys were commonly in the immediate vicinity of carbides (Figure 4-5) The pore-carbide spatial relationship is likely due to the carbides blocking the molten alloy during the final stages of solidification [14,41,51]. This blocking effect leads to the increased porosity levels in the C-containing alloys. 4.1.5 Primary Carbides The carbide phases that decorated the inte rdendritic regions of all modified alloys were first characterized by examining light etched, flat cross sections with the SEM. BSE mode provided better contrast than SE mode between carbides (light) and / matrix (dark) because it is more sensitive to compositional differences. Images such as those in shown in Figure 4-6 were used in an attempt to determine carbide volume fractions in a similar manner to porosity volume fraction. The unique carbide shapes seen in these alloys significantly hamper the ability to accurately determine volume 80

PAGE 81

fractions. The measurements here (Table 4-3) can be presented only as carbide area fractions in the transverse cross sections. This data can be interpreted as a representation of carbide blockiness. Carb i des in the N-containing alloy were the blockiest and appeared larger and less frequently in cross section than carbides in the other two modifications. The deep etching process removes both the and phases and yields a much better representation of actual carbide mor phology. When viewed only in cross section, carbide appearances can be deceiving. Regions that seem to have groupings of small, discrete carbide particles may actually contai n large carbide networks. A comparison of how carbide networks appear in the two differ ent etched conditions is shown in Figure 4-7. A very detailed examination of carbi des in the deep etched condition for both transverse and longitudinal samples was c onducted to compare carbide morphologies between the modifications. Characteristic carbide morphologies for the three Ccontaining alloys are shown in Figure 4-8. The C modification as well as the C+B modification resulted in carbides that formed in script-like net works consisting of connected rods and plates. Some regions c onsisted of arrays of parallel rods which were partially filled-in from incomplete plat e formation during solidi fication, as seen in Figure 4-9. Cores of carbide networks resided between primary dendrite arms and carbide rods branched off of the cores in between secondary dendrite arms. The manner in which the carbides conform to the edges of the dendrite arms can be seen in Figure 4-10. The C+N modification exhibited blocky carbides that showed less spatial orientation in relation to t he dendrites than the script networks found in the other two 81

PAGE 82

modifications. These blocky carbides were also much smaller (20 50 m) than the script networks (100 200 m). As-cast carbides in SX superalloy s are generally MC-type, but carbide compositions can vary significantly between different alloys depending on content of carbide forming elements such as W, Ta, Ti, and Hf [19,37,38,48 ]. EDS measurements on a large number of carbides were c onducted to determine relative amounts of elements present in carbides. Results indicated that as-cast carbides in all modified alloys were rich in Ta and contained sma ller amounts of W and Ti A few of the measurements detected small am ounts of Hf and Cr in the ca rbides. An example of a characteristic EDS spectrum for a carbi de is presented in Figure 4-11. Precise concentrations cannot be obtained with EDS, es pecially when a very low atomic weight element such as C is present, but semiquantitative analysis indicated MC-type carbide chemistry ranges and high levels of Ta (~ 60 80 wt.% or ~ 40 55 at.%). Even though Ta (6.5 wt.%) and W (6.4 wt.%) are present in similar concentrations in the bulk alloy, Ta is clearly the preferred ca rbide former in these modified alloys. In order to verify as-cast carbide type, XRD was conducted on deep etched samples. Deep etched specimens have mo re exposed carbide surface and therefore produce stronger carbide signals. All of the analyzed samples produced peak profiles characteristic of MC-type carbides [109117]. The identified peaks corresponded very closely to those for pure TaC, with the slight differences in peak locations due to the minor presence of Ti and W. The confirm ed MC carbides have the NaCl-type crystal structure, with C occupying the cation intersti tial sites and Ta occupying the majority of the anion sites. A characteristic XRD spectr um is shown in Figure 4-12. Additional 82

PAGE 83

XRD spectra can be found in Appendix B. Differences in peak intensities and the presence of some / peaks are due to varying amount s of material removed during deep etching. 4.2 Post Solution Heat Treatment Microstructure SHTs are normally designed to eliminate or greatly reduce chemical segregation without causing incipient melting. The treatm ents in this study, particularly the HT SHT, were designed to induce carbide transformations as well as homogenize the alloy. The results were assessed with particula r focus on changes to the carbides. 4.2.1 Homogenization Both SHTs achieved the desired level of homogenization by significantly reducing segregation between dendrit ic and interdendritic regions. The / structures ( precipitates form during cooling from heat treatment temperature) appeared similar for both homogenization treatm ents (LT SHT and HT SHT). A dditionally, there were no clear differences in homogenization or / structure between baseline CMSX-4 and the modifications, as shown in Figure 4-13. Fo r both SHTs, a very sma ll amount of residual eutectic was observed in all al loys, but the eutectic pools were larger after LT SHT than after HT SHT, which dissolved more of the eutectic regions (Figure 4-14 ) No incipient melting was observed in any of the post-SHT microstructures. It should be noted here that further increases to maximum SHT may have fully dissolved eutectic regions [37,81], but the higher temperatures would have significantly increased the risk for incipient melting. The possibility of usi ng a specifically designed SHT for each alloy variation was considered, but it was determi ned that using the same treatments for all alloys was the most effective way to make direct comparisons. 83

PAGE 84

4.2.2 Porosit y Casting porosity has been shown to in crease during homogenization in SX superalloys due to coalescence of vacancie s at high temperature [55], and the alloys studied in this investigation demonstrated th is effect. Pores observed were clearly larger (Figure 4-15) in the post-SHT condition as compared to the as-cast state. The dozens of pores examined in both condition s revealed an increase in maximum pore size from ~ 5 10 m to ~ 10 20 m due to SHT. This size increase was observed for both the baseline and modified alloys. T he growth of pores is a side effect of homogenization that may have a negativ e impact on mechanical properties. 4.2.3 Carbides The HT SHT was designed to promote decomposition of large carbide networks to smaller features that may be less detri mental to mechanical properties. SEM examination revealed only minor carbide ch anges, and these changes were similar for both the LT SHT (Figure 4-16) and HT SHT (Fi gure 4-17). Some of the carbides did not undergo any changes during SHT and their surfac es remained smooth. A large number of them, however, exhibited surface roughness that indicat ed early stages of carbide decomposition. Small nodes were evident on t he surfaces of carbide plates in the C and C+B modifications and blocky carbides in the C+N modification, as shown in Figure 4-16A and Figure 4-17C. Decompositi on reached a more advanced stage for some carbides in the C+B modification as discrete, small carbides were observed near larger carbide networks (Figure 4-16B). No change in carbide composition accompanied any of the morphology changes due to SHT. Although most carbide networks remained intact after SHT, the results indicate that partial carbide dissolution through diffusion of C from primary MC carbides did occur in a ll alloys. The LT SHT (maximum temperature 84

PAGE 85

of 1310 oC) had similar effects on carbides as the HT SHT (maximum temperature of 1325 oC). Both treatments involv e temperatures and times that cause some amount of carbide decomposition, but neither one results in complete transformation of morphology. The difference in maximum temperatures was not great enough to cause noticeable changes in carb ide behavior during SHT. EDS analysis of carbide surface featur es and new, discrete carbides confirmed that all carbides remained Ta rich and MC-type. None of the analyzed carbides had chemistries indicative of traditional secondary M23C6-type or M6C-type carbides. XRD was once again enlisted to verify carbide pha ses present after SHT. A characteristic spectrum confirming the presence of MC-t ype carbides is shown in Figure 4-18 (additional spectra in Appen dix B). Additionally, t he peak at approximately 43o identifies the presence of some secondary, likely M23C6-type, carbides [ 110]. No such peaks were observed in the as-cast condition. This indicates that some secondary carbide formation did occur during SHT even though no su ch carbides were observed directly in the SEM analysis. Note that HT SHT specim ens were used because the majority of mechanical testing was performed with test s pecimens that had undergone the HT SHT. All subsequent XRD spectra are from material that underwent the HT SHT as part of its heat treatment schedule. 4.3 Post Aging Heat Tr eatment Microstructure The two-step aging heat treatment is designed to form uniform, cuboidal primary strengthening precipitates (from 1st aging step) and sm aller, secondary particles in channels (from 2nd aging step) to provide the required mechanical properties for service. The post-age examin ation focused not only on the / structure, but on the changes in carbide morphology that occurred. 85

PAGE 86

4.3.1 / S tructure Aging heat treatment s produced the desired size, vo lume fraction, and cuboidal morphology of phase in all alloys. A representative SEM micr ograph of the fully heat treated structure is presented in Figure 4-19. This structure, with primary precipitates measuring ~ 0.5 m edge dimension and separated by channels measuring ~ 0.05 0.2 m in width, was observed in all CM SX-4 variants. Secondary, spherical particles measuring ~ 20 60 nm in diameter decorated the channels. The microstructures were consistent whether the sample had undergone the LT SHT or the HT SHT. Average quantitative measurem ents from samples that underwent the HT SHT and aging heat treatment are show n in Table 4-4. The variations between the alloy modifications were remarkably small, as all of the volume fractions were between 66 % and 67 %. The size values were slightly larger for the C-containing alloys, but all sizes were within statistical scatter. Although primary precipitate size and morphology are the clearest indicators of a satisfactory heat treatment, secondary formation also warrants attention. Fine particles can change deformation processes and mechanical strength by effectively hindering dislocation motion in channels [118]. High magnifications (> 20,000 X) are required to image these very small features as shown in Figure 4-20. In the SEM micrograph in Figure 4-20A, the secondary particles are visible as dark circles in the light channels. This image, however, capt ures only the particles that were preferentially etched during the light etch procedure. In TEM micrographs, the particles appear as small, light colored spheres, as shown in Figure 4-20B. This method captures the high density of fine in the channels. Overall, the minor additions of C, B, 86

PAGE 87

and N did not seem to impact the heat treatment response of CMSX-4. The / microstructures were very similar for t he baseline and the modified CMSX-4 alloys. 4.3.2 Carbides The MC carbides underwent significant changes during aging heat treatments, particularly the alloy with C and B modifica tions. The breakdown of carbide networks seen after SHT was much more advanced after the aging steps. Small, rounded carbides were observed in some regions of the C modification and to a greater extent in the C+B modification. The SEM micrograph in Figure 4-21A shows a group of carbides with this rounded morphology. The frequency of these smaller carbides, coupled with the small number of retained script carbide networks in the C+B modification, suggests that the majority of the networks in this alloy broke down during heat treatment. The majority of the script carbide networks in the C modification and blocky carbides in the C+N modification were retained with small surface changes similar to those observed after SHT. The drastic morphology change for carbides in the C+B modification was further investigated. These types of transformati ons during heat treatm ent are generally the result of decomposition accompanied by changes in carbide composition and type [18,39]. The morphology changes studied her e, however, occurred without significant changes in composition. ED S measurements of small carb ides indicated that they remained Ta rich. The representat ive EDS spectrum in Figure 4-21B is remarkably similar to EDS spectra for as-cast carbides (Figure 4-11 ), with minor amounts of Ti and W in addition to high Ta concentration This finding is significant because it represents carbide transformation during heat treatm ent without formation of traditional M23C6 or M6C secondary carbides. 87

PAGE 88

TEM observation revealed carbides less than 5 m in size that formed during heat treatment, as shown in Figure 4-22A. Th is image of a small, round carbide was captured using the ST EM mode in the high resolution TEM. Figure 4-22B presents EDX measurements (probe size of 3 nm) taken from the spot indicated. Compositional data from a carbide this small cannot be accurately obtained using EDS in the SEM because the probe size is larger than the feature. Semi-quant itative EDX results indicated that this small carbide remained Ta rich (~79.01 wt. %). This serves as further evidence that some transformation during heat treatment of carbide networks into small carbide particles occurred without a significant change in composition. Post aged carbide compositions were com pared using WDS, which provides more accurate quantitative data than EDS. Aver age results are shown in Table 4-5. Although there were slight variations between the alloys, all carbides had high concentrations of Ta. The low atomic weight of C makes it difficult to accurately measure with WDS, as evidenced by the lack of detected C in the carbides in the C+N modification. For both the C and C+B modifications, W had the second highest concentration behind Ta amongst the carbide forming elements. In the C+N modification, however, Hf was the element with the second highest concentration in fully heat treated carbides. Differences in compos ition from carbides in as cast condition, along with observations of surface changes to carbides, suggest that appreciable atomic diffusion occurred between the carb ides and the surrounding matrix during heat treatment. XRD results were similar to those in the post-SHT condition, indicating the presence of MC carbides and M23C6 carbides even though only MC carbides were 88

PAGE 89

observed directly in SEM and WDS analys is. A small amount of secondary M23C6 carbide formation occurred in addition to t he MC carbide morphology transformations that did not involve changes in composition. A characteristic spectrum is shown in Figure 4-23 (additional spectra in Appendix B). 4.4 Exposed Microstructure Long-term alloy stability has been identified as a significant issue in SX superalloys with high refractory content. To evaluate microstructural stability, bar sections that had undergone HT SHT and the two-step aging heat treatment were thermally exposed in air for 1000 hrs at 1000 oC. This exposure resulted in phase coarsening, carbide decomposition with a ssociated secondary carbide formation, and TCP phase formation. Each topic will now be presented separately for clarity. 4.4.1 / Structure and Coarsening Coarsening of the phase occurs during high te mperature exposure and can significantly reduce the strengt h of the alloy as precipitat es grow and lose coherency with the matrix. The primary driving force for coarsening is reduction of / interfacial area as the average size of precipitates increases and the number of precipitates decreases [63]. Significant coarsening occurred in all alloys dur ing exposure. The precipitates remained cuboidal with some slight rounding of the corners. A co mparison between the fully heat treated and the exposed microstructu re for the C modification is shown in Figure 4-24. The precipitate size distribution clearly incr eases during exposure. Some precipitates in the exposed condition were sim ilar in size to the heat treated precipitates and others were significantly la rger. Average measurements for volume fraction and precipitate size are presented in Table 4-6. The volume fractions increased slightly due 89

PAGE 90

to exposur e and the average precipitate size s increased significantly. The graph in Figure 4-25 illustrates that a greater degree of coarsening occurred in the C-modified alloys. The B-containing alloy ex hibited the largest post-exposure size (0.91 m). Assuming that volume diffusion is the controlling factor, coarsening of can be represented by a cube law and Eqn. 4-1 [119,120]. The value of K represents a coarsening rate constant for a given alloy and temperature that is controlled by the equilibrium concentration of solute in the diffusion coefficient of solute in and the / interfacial free energy. The values of K (Table 4-6) were estimated using average precipitat e sizes before (do) and after (d) exposure and the time of exposure. All of the modified alloys had larger rate co nstants than baseline CMSX-4, with the C+B modification having the largest K value. d3 do 3 = Kt (Eqn. 4-1) 4.4.2 Carbide Decomposition All carbides underwent significant changes due to high temperature exposure. Partial carbide dissolution and formation of M23C6 secondary carbides occurred in all three modified alloys. Carbides were first obs erved in the light etched condition. Script networks in the C modification, which had remained almost fully intact and 100 200 m in size during heat treatment, we re much smaller (10 50 m) due to decomposition during exposure. Figure 4-26A shows a small, Ta rich carbide in the C modification that is likely a remnant of a decomposed carbide network. Small, Cr rich secondary carbides were observed in the vicinity of decomposed primary carbides in all C-containing alloys. The compositions, high Cr and significant pres ence of W, indicated these secondary carbides to be M23C6 carbides [18,39,57,80,119]. In some instances in the C+B modification, small, rounded M23C6 carbides approximately 1 3 m in 90

PAGE 91

diameter decorated the peripher y of a region where a MC carbide underwent complete dissolution (Figure 4-26B). These decom position zones were almost entirely phase (dark contrast in micrograph). These layers are commonly reported as a product of carbide decomposition proce sses [56,59]. The decomposing MC carbide provides C (forms M23C6 carbides) and Ti (forms Ni3(Al,Ti) ) and the matrix provides Ni and Al (forms Ni3(Al,Ti) ) and Cr (forms M23C6 carbides). Local enrichments of Ni, Al, and Ti occur with precipitation of M23C6 carbides, and forms and grows around the M23C6 carbides. The process can be su mmarized by the reaction MC + M23C6 + The size of these circular regions indicates that the decomposing MC carbides were the small, spherical carbide particles that were observed after aging heat treatment in the C+B modification (Figure 4-21A). Examination of exposed samples in the deep etched condition presented a clearer picture of carbide decomposit ion mechanisms. Secondary M23C6 carbides formed near partially decomposed script netwo rks in the C modification, as shown in Figure 4-27A. The plates on the left side of the image represent the rema ining MC carbide and the small carbides on the right are the secondar y carbides that have formed. The EDS spectrum for the carbide circled in the mi crograph indicated high Cr content (Figure 427B). Secondary M23C6 carbides were observed on t he surfaces of some decomposed carbides in the C+B and the C+N modifications. These types of carbides were observed most frequently in the B-containing alloy, in which they formed both near and on the MC phase (Figure 4-28A). Long term service exposures of polycrystalline superalloy samples exhibited continuous layers of M23C6 film that formed around 91

PAGE 92

decomposing MC carbides [ 56]. In contras t, the M23C6 carbides observed in this study appeared as discrete growths. The growth s represent an intermediate stage in the transformation from MC to M23C6 carbides. Longer exposure times would be expected to lead to further decomposition and covera ge of MC carbide by a continuous M23C6 layer. Figure 4-28B shows secondary carb ide formation on a decomposing MC in the C+N modification. Semi-quantitative EDS measurements of the partially decomposed MC carbide in the N-containing alloy showed higher Ti content (41.37 at. % Ti, 29.48 at. % Ta) than the MC carbide in the B-containing alloy, which was richer in Ta (24.75 at. % Ti, 39.88 at. % Ta). These trends were cons istent for multiple carbides analyzed in each alloy. The MC carbide in the C+B m odification (Figure 4-28A) is lighter in BSE contrast than the M23C6 carbides because Ta is a denser element than Cr, and the MC carbide in the C+N modification (Figure 428B) is darker in BSE contrast than the M23C6 carbides because Ti is less dense than Cr. The Ti rich decomposed carbide core in the N-containing alloy may support previous findi ngs [44] that TiN particles can precipitate during solidification and provide heter ogeneous nucleation sites for carbide or carbonitride formation. The DTA results pres ented above indicate that the presence of N does not cause carbide formation higher in the mushy zone. The Ti rich cores represent only a difference in carbide composition between alloys, not evidence of formation of carbides or ca rbonitrides on TiN particles. Maps showing relative locations of remaining MC carbides and newly formed M23C6 carbides were created to gain a better understanding of carbide decomposition during exposure. Efforts were made to take EDS measurement s at the thickest regions of small carbides because the compositiona l signals can come from the deepest regions 92

PAGE 93

of the beam interacti on volume in the ma terial (up to several microns deep). An example of one of these maps with the corr esponding EDS spectra is shown in Figure 4-29. The area of the carbide labeled with a is remaining MC carbide. Region is a M23C6 carbide that has formed on the decomposi ng MC carbide, and carbides and are M23C6 carbides that formed nearby. Additi onal carbide maps are presented in Appendix C. Some of the maps contain TCP phases, whic h will be discussed in the next section. XRD results confirmed decomposition of MC carbide and formation of secondary M23C6 carbides due to long-term exposure. The MC carbide peaks were less intense and the M23C6 peaks more intense than at the post-SHT and post-age conditions, particularly for the C+B modification. A charac teristic spectrum is shown in Figure 4-30 (additional spectra in Appendix B). 4.4.3 Topologically Close Packed (TCP) Phase Formation Formation of detrimental TCP phases over long term exposures is a concern in advanced SX superalloys such as CMSX-4 [63], and determining the role of minor additions on TCP phases was an important as pect of this study. TCP phases were observed after thermal exposure in all alloys except for the C+B modification, but the location and quantities of TCP phase varied amongst the modifications. Baseline CMSX-4 had short, needle-like TCP phases that formed sporadically in dendrite core regions (Figure 4-31A). Thin TCP needles (Figure 4-31B) were also observed at dendrite cores and near decomposed carbides in the C modification. TCP phases formed the most frequently in t he C+N modification, in which groups of connected plates and needles were observed in the vicinity of decomposed carbides (Figure 4-31C). 93

PAGE 94

The TCP phases shown were rich in Cr, Co, Re, and W (Table 4-7). The EDS results of other analyzed TCP phases (Appe ndix D) showed enrich ment of the same elements in varying amounts. B ased on composition ranges and morphologies, the TCP phases observed were likely phase [37,121]. Varying stoichiometric ratios have been reported in phase, which has a tetragonal crystal structure. Similarities between atomic species lead to some interchangeability in which atoms fill lattice positions. TCP formation near MC carbides is believed to be caused by short range segregation of TCP-forming elements in these regions [19, 37]. Longer exposure times would likely result in further TCP needle and plate forma tion and removal of more solid solution strengthening elements from the matrix. The lack of TCP in the B-containing alloy is of particular interest. The only phases observed in the post-exposure microstructure (other than and ) were partially decomposed MC carbides and secondary M23C6 carbides on or near the MC phases (Figure 4-32). The B-containing alloy ex hibited the most carbide decomposition, including complete dissolution in some cases, and secondary M23C6 carbide formation during heat treatment and exposure. 4.5 Analysis The data presented in the pr eceding sections was carefully examined to improve the depth of understanding of microstructural evolution in the studied alloys. The following sections detail these efforts. De scriptions of the techniques and results of analysis are included. 4.5.1 Carbon Diffusion There were two different ty pes of carbide decomposit ion processes observed. The first involved morphology changes without corresponding changes in composition. 94

PAGE 95

MC carbides broke down into smaller MC carbides or small MC carbide nodes formed on the surface of larger MC carbides (Fi gur e 4-17C). These types of changes were observed after SHT and after the aging heat treatment. The second process involved the decomposition of MC carbides into secondary M23C6 carbides that formed near the decomposed MC carbides or directly on the surf aces of the MC carb ides (Figure 4-32). These changes were observed via SEM only after long-term exposure, but M23C6 carbides were detected with XRD after SHT and aging. Both of these processes rely on the interstitial diffusion of C out of MC carbide and into or through the surrounding matrix [56]. Estimates for relative diffusi on coefficients were made usi ng ratios of the Arrhenius equation for a diffusion coefficient at a given temperature [122]. The formula is shown in Eqn. 4-2 where D is the diffusion coefficient, Do is a pre-exponential factor, Q is the activation energy for diffusion, R is the universal gas constant (8.314 J/mol*K), and T is temperature. To compare coefficients at tw o temperatures, the equations were divided and several of the variables cancelled out. The ratio between the coefficients can then be found with the temperatures and the activation energy. Precise activation energies for diffusion are difficult to obtain for s uperalloy systems due to t he large number of atomic species present. Therefore a simp lified system of C diffusing out of TaC and then through bulk Ni (the pr imary element in the / matrix) served as the basis for the comparisons. Estimated values of Q (161. 1 kJ/mol for diffusion of C in Ta [123] and 138 kJ/mol for diffusion of C in Ni [124]) were used for the coefficient comparisons. It should be noted here that only minor differ ences have been estimated in activation energies for diffusion of C in Ta and C in TaC [125]. 95

PAGE 96

D = Doe-Q/RT (Eqn. 4-2) 4.5.1.1 Post-SHT vs. post exposed carbides Using the method described above, the diff usion coefficient for C in Ta at the maximum SHT temperature (1310 oC or 1325 oC) was found to be about 20 times greater than the coefficient at the exposure temperature (1000 oC). The coefficient for C in Ni was an estimated 14 times greater at the SHT temperature t han at the exposure temperature. During SHT, the diffusion of C is great enough to provide enough C near the surface of large MC carbides to fo rm MC carbide nodes. During exposure, however, the diffusion of C is considerabl y slower and may not be present in large enough concentrations to re-fo rm MC carbide. The 1000 hrs of exposure, however, may provide enough time for Cr and W in the matrix to diffuse and form M23C6 carbides with the C that is present. 4.5.1.2 Low temperatur e (LT) SHT vs. high temperature (HT) SHT Carbides underwent similar degrees of partial decomposition due to both SHTs. The diffusion coefficient for C in Ta and C in Ni at maximum temperature for the HT SHT was estimated to be only 1.1 times grea ter than at the maximum temperature for the LT SHT. The activation energy for diffusion of C out of the MC ca rbides is such that decomposition begins to occur as a result of both heat treatm ents. The maximum temperature of the HT SHT is near enough to the incipient melting point that increasing the diffusivity by increasing temperature is not a viable option. The only way to drive more C out of the MC carbides to promot e morphology change is to increase the length of time at the maximum heat treatment temperatures. 96

PAGE 97

4.5.2 Alloy Stability Predictions The growing importance of phase stabilit y and prevention of TCP phase formation in newly developed alloys has given rise to several computer model methods designed to predict TCP phase formation based on allo y chemistries. One of the most recognized is the PHACOMP (Phase Computati on) approach, which attempts to predict phase stability based on solute distribution among phases. Mathematical calculations are made to estimate the am ount of each element in the matrix, which is then multiplied by the number of unpaired electrons (Nv) for the given element. These values are then summed to get an overall Nv number for a given alloy. An alloy with N an n. v value greater than 2.45 is said to be susceptible to phase formatio Alloy Nv values for the four CMSX-4 variati ons were calculated using elemental Nv numbers from Donachies ASM Superalloys Technical Guide [126]. An example calculation is shown in Appendix E. All of the calculated Nv values were considerably lower than the 2.45 threshold: baseline 2.24, C modification 2.17, C+B modification 2.17, C+N modification 2.18. Three of the four alloys contained TCP phases after thermal exposure even though a ll were predicted to be stable. Also, the one alloy that did not contain TCP phase, the C+B modification, had virtually the same Nv value as the other two modified alloys. The reason for the discrepancy is likely the simplicity of the model. The PHACOMP appr oach does not take into account residual segregation at dendrite core regi ons (location of TCP phase in the baseline) or localized compositional differences near decomposing carbides (location of TCP phase in C modification and C+N modification). 97

PAGE 98

98 4.6 Summary The effects of alloy modifications included increased porosity and reduced eutectic, but the / microstructures developed thr ough heat treatment were largely unchanged. Minor additions of B and N, however, had a significant impact on carbide phases in C-modified CMSX4. The addition of N changed the as-cast carbide morphology from large script to smaller and blo ckier. Carbides in the B-containing alloy transformed from script MC networks to sm aller, rounder MC carbides during heat treatment. In addition, the presence of B led to suppression of TCP phase formation during long term exposure.

PAGE 99

Table 4-1 Differential thermal analysis (DTA) results showing phas e transformation temperatures. Baseline C C+B C+N As Cast solidus (oC) 1319 1318 1316 1322 Heat Treated solidus (oC) 1343 1339 1327 1340 As Cast liquidus (oC) 1386 1382 1382 1383 Heat Treated liquidus (oC) 1389 1386 1384 1384 As Cast carbide (oC) n/a 1369 1362 1367 Heat Treated carbide (oC) n/a 1369 1366 1369 Figure 4-1. Optical micrographs of transverse, as-cast CMSX-4 alloy microstructures. A) Baseline, B) C modification, C) C+B modification, D) C+N modification. 99

PAGE 100

100 Figure 4-2. Optical micrographs of longitudi nal as-cast CMSX-4 alloy microstructures. A) Baseline, B) C modification, C) C+B modification, D) C+N modification.

PAGE 101

Figure 4-3. / eutectic in as-cast microstructure. A) light contrast in optical micrograph, B) dark contrast in SEM microgr aph, C) eutectic near carbide. Table 4-2. Estimated porosity volume fractions in the as-cast condition. Baseline C C+B C+N Porosity Volume Fraction 0.011 0.014 0.013 0.016 90% Confidence Interval .3 x 10-3.4 x 10-3.9 x 10-4 .6 x 10-3 101

PAGE 102

Figure 4-4. SEM backscattered electron (BSE) micrographs of tra nsverse cross sections used to estimate porosity vo lume fraction. A) baseline, B) C modification, C) C+B modifica tion, D) C+N modification. Figure 4-5. SEM micrograph showing casting porosity in immediate vicinity of carbides. 102

PAGE 103

Figure 4-6. SEM BSE micrographs of as-cast carbides in transverse cross sections. A) C modification, B) C+B modifi cation, C) C+N modification. Table 4-3. Estimated carbide area fractions in the as-cast condition. C C+B C+N Carbide Area Fraction 0.032 0.042 0.059 90% Confidence Interval .3 x 10-3.0 x 10-2 .2 x 10-2 Figure 4-7. SEM micrographs showing differ ent appearance of carbides. A) light etched and B) deep etched conditions. 103

PAGE 104

Figure 4-8. Represent ative SEM micrographs of carbide morphologies in the deep etched condition. A) C modification, B) C+B modification, C) C+N modification. Figure 4-9. SEM micrograph of par tial carbide plate formation in the C modification in the as-cast condition. 104

PAGE 105

Figure 4-10. SEM micrograph showing carbid e network and secondary dendrite arms. Figure 4-11. Characteristic energy dispersive spectroscopy (EDS) spectrum for a carbide in the as-cast condition. Carb ides in all three modified a lloys had similar spectra. 30354045505560657075802 (o)Intensity (A.U) -MC Figure 4-12. X-ray diffraction (XRD) spectrum of carbon (C) modification in the as-cast and deep etched condition. Peak profile is consistent with MC-type carbide. 105

PAGE 106

Figure 4-13. SEM micrographs showing similar / structures in the post-SHT condition. A) Baseline and B) C modification. Figure 4-14. SEM micrographs of post-SHT re sidual eutectic. A) LT SHT and B) HT SHT. The smaller eutectic pools after HT SH T are due to more dissolving of eutectic at higher temperature. Figure 4-15. SEM micrograph showing a por e in the post-SHT condition. Pores increased in size due to SHT. 106

PAGE 107

107 Figure 4-16. SEM micrographs of carbides after low temperature (LT) SHT. A) C modification showing small nodes on carbide plate surface, B) C+B modificatio n (BSE) showing breakdown of network into smaller carbide particles, and C) C+N modification ( BSE) showing carbide surface roughness.

PAGE 108

Figure 4-17. SEM micrographs of carbides after high temperatur e (HT) SHT. A) C modification showing remaining r od and plat e-like features, B) C+B modification showing signs of carbide decomposition, and C) C+N modification showing nodes on surface of blocky carbide. 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / Figure 4-18. XRD spectrum of the C modification in the post-HT SHT, deep etched condition. Presence of MC carbide and M23C6 carbide is confirmed. 108

PAGE 109

Figure 4-19. Representative SEM mi crograph of the fu lly heat treated / micro structure observed in all alloys. Table 4-4. Quantitative measurements for fully heat treated samples. The size represents the length of an edge of the cuboidal precipitate. Standard deviations are shown in parentheses. Baseline C C+B C+N Volume Fraction (%) 66.2 (1.25) 66.5 (1.59) 66.3 (2.57) 66.9 (2.27) Size (m) 0.45 (0.02) 0.47 (0.02) 0.49 (0.03) 0.46 (0.03) Figure 4-20. Secondary, fine particles in channels of fully heat treated alloys. A) SEM micrograph of C modification, B) TEM micrograph of C+N modification. The dark phase on the right side of the TEM image is a MC carbide. 109

PAGE 110

Figure 4-21. Small, rounded carbide particles in fully heat treated condition of the carbon and boron (C+B) modification. A) SEM micrograph of deep etched sample, B) EDS spectrum from one of the small carbides indicating carbide composition did not change despite morphology transformation. Figure 4-22. Small, tantalum (T a) rich carbide in the fully heat treated condition. A) TEM micrograph (STEM mode), B) EDX spectr um from indicated point. The Cu peak is from the TEM specimen grid. Table 4-5. Average spot wavelength dispersi ve spectroscopy (WDS) compositional (wt. %) measurements of carbides in the fully heat treated (HT SHT + two-step age) condition. C modification C+B modifi cation C+N modification Ta 76.12 75.96 65.43 C 10.41 9.16 Al 0.003 0.003 0.61 Co 0.26 0.29 2.45 Cr 0.28 0.47 1.17 W 8.90 10.55 7.16 Ti 0.31 0.14 0.36 Ni 2.84 3.11 6.49 Hf 0.87 0.32 16.30 110

PAGE 111

30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / Figure 4-23. XRD spectrum of the carbon and nitrogen (C+N) modification in the postaged, deep etched condition. Figure 4-24. SEM micrographs of precipitates in the C m odification. A) fully heat treated condition and B) thermally exposed condition. Significant phase coarsening occurs due to the 1000 oC/1000 hr exposure. Table 4-6. Quantitative m easurements for exposed sample s. Standard deviations are shown in parentheses. Baseline C C+B C+N Volume Fraction (%) 70.2 (2.78) 74.0 (2.50) 70.9 (2.28) 70.8 (2.90) Size (m) 0.79 (0.05) 0.86 (0.05) 0.91 (0.07) 0.86 (0.04) Coarsening Rate Constant: K (m3/sec) 1.10 x 10-7 1.48 x 10-7 1.77 x 10-7 1.48 x 10-7 111

PAGE 112

Figure 4-25. Average size ( m) of precipitates in the heat treated condition and after exposure at 1000 oC for 1000 hours. Error bars r epresent the 90% confidence interval. Coarsening occurs to a greater degree in the C-modified alloys than baseline CMSX-4. Figure 4-26. SEM micrographs of light et ched specimens after high temperature exposure (1000 oC/1000 hrs). A) small, Ta rich carbide in C modification, B) group of small, Cr rich secondary carbides in C+B modification. 112

PAGE 113

Figure 4-27. Deep etched C modification samp le after high temperature expos ure (1000 oC/1000 hrs). SEM micrograph and correspondi ng EDS spectrum of Cr rich M23C6 carbide. Figure 4-28. SEM micrographs of deep etched samples after high temperature exposure (1000 oC/1000 hrs). A) C+B modification, BSE image of Cr rich carbide (dark phase) on Ta rich MC carbide (light phase), B) C+N modification, BSE image of Cr rich carbide (light phase) on Ti rich MC carbide (dark phase). 113

PAGE 114

Figure 4-29. Map of carbides in the deep etched condition after high temperature exposure (1000 oC/1000 hrs). The EDS spectra shown are from the corresponding numbers in the micrographs. 114

PAGE 115

30354045505560657075802 (o)Intensity (A.U.) -MCM23C6 / Figure 4-30. XRD spectrum of C+N modification in the fully heat treated and exposed condition. Peak profile indicates dec omposition of MC carbide and formation of M23C6 carbides during long-term exposure. Figure 4-31. SEM micrographs of topologically close packed (TCP) phases in deep etched, thermally exposed samples. A) Baseline CMSX-4, B) C modification, C) C+N modification. 115

PAGE 116

116 Table 4-7. Energy dispersive spectroscopy (EDS) semi-quantitative concentration (wt. %) measurements of the topologically close packed (TCP) phases shown in Figure 4-31. Baseline C modification C+N modification Cr 10.2 10.9 10.8 Co 10.1 12.6 13.0 Ni 14.1 19.0 18.1 Re 24.4 17.4 13.7 W 40.9 40.13 44.37 Figure 4-32. SEM micrograph of deep etched C+B modification sample after high temperature expos ure (1000 oC/1000 hrs). Secondary M23C6 carbides formed on decomposing MC carbides. No TCP phase was observed in this alloy.

PAGE 117

CHA PTER 5 RESULTS: CREEP DEFORMATION The following sections describe the results of all creep testing performed in this investigation. The effects of carbides on creep and stress rupture are unclear for SX superalloys, with reports varying regarding w hether carbides improve creep strength by acting as barriers to dislocation motion [8,18, 37,80] or accelerate creep failure through carbide cracking [68,81]. It is important to conduct testing at multiple conditions because different areas of service components experience different temperatures and stresses. Testing was in itially planned only at 950 oC/300 MPa and 850 oC/550 MPa, but observed primary cr eep effects during the 850 oC testing warranted testing at a 750 oC/800 MPa condition. All creep testing was initially planned for HT SHT specimens that had not undergone the HIP treatment. Concurrent HCF testing, however, revealed a significant effect of HIP, and some HT SHT, HIPed material was crept at 850 oC for comparative purposes. A limited number of samples, along with the need to use multiple repeats for the more highly variable fatigue testing, resulted in fewer repeats of creep tests. Still, careful post-test analysis a llowed for identification of trends regarding minor additions and creep deformation. 5.1 Elevated Temperature Tensile Testing A tensile test was conducted for each alloy at 850 oC to help determine creep conditions and to compare tensile properties. Results confirmed sim ilar yield strengths for the baseline and the C-modified alloys. Creep testing, therefore, occurred at comparable fractions of t he yield strength for all alloys without changing stress conditions between tests. All tensile test s were conducted on fully heat treated (HT SHT) material in the Un-HIPed condition. 117

PAGE 118

5.1.1 Test Results Figure 5-1 shows tensile stress-strain curves for the four tests. All of the curves have the same general shape, but the baseline experienced significantly more elongation than the modi fied alloys, as presented in Table 5-1. Other important tensile values are also presented in the table. It should be noted that yield stresses were estimated at the point wher e the curve began to deviate from linearity. The standard 0.02% offset was not used because the elongation curves were generated from the tester head movement and not directly from t he sample due to extensometer slippage. All curves exhibited a significant increase in stress after yielding and then gradual decrease in stress until failure. Baseli ne CMSX-4 had the highest ultimate tensile strength due to a greater amount of hardening after yielding than in the modified alloys. 5.1.2 Fractography Representative micrographs of tensile fractu re surfaces are presented in Figure 52. Fracture surfaces of the modified alloys contained rough features and signs of ductile fracture, while the features seen in the baseline were smoother in nature. Areas that appear to be cleavage facets represent deformation along slip planes. Baseline ductile features had pores in the center (Figure 5-2A). The cracked carbides observed were often in direct contact with porosity and appeared to be inside of pores. A large number of carbides were visible on the frac ture surfaces of the modified specimens, indicating that the crack paths likely fo llowed regions containing carbides. The presence of carbides created more crack formation sites and may account for the reduced ultimate tensile strengths and tensile elongations (Table 5-1) in the modified alloys. Cracks were observed in carbides of all morphologies and in all three Ccontaining alloys. 118

PAGE 119

5.1.3 Longitudinal Se ctions Light etched longitudinal sections were pr epared from each tested tensile sample for observation in the SEM. These sect ions revealed cracks perpendicular to the loading axis, which were accompanied by carb ide cracking in the three modified alloys. Most carbides, of both script and blocky mor phologies, had cracks that extended into the surrounding / matrix. These cracks were the most clearly observed when using BSE mode, as shown in the SE BSE pair images in Figure 5-3. This confirms the fracture surface observations that cracking of carbides created an increased number of crack formation sites. Carbide cracking, as well as increased porosity reduced the strength and ductility of the modifications as compared to baseline CMSX-4. Cracks in the baseline alloy were observed at sites of porosity (Figure 5-4A). There was also evidence of more plastic deformation in t he baseline specimen (F igure 5-4B). The triangular appearance of the phase indicates shearing of pr ecipitates along {111} slip planes [7]. Plastic deformation processes ad vanced further in baseline CMSX-4 before significant cracking and failure, leading to increased ductility. 5.2 Creep Testing at 950 oC/300 MPa Creep tests at 950 oC and 300 MPa were conducted on heat treated, Un-HIPed material that had undergone the HT SHT. Tertiary creep played a dominant role at this condition. Differences in observed behavior we re attributed to variations in material damage accumulation. 5.2.1 Test Results All tests at this condition resulted in the same general shape of the time vs. strain creep curves (Figure 5-5). The creep rates stayed relatively constant for the first 50 100 hours (less than 1 % strain) before the creep rates began to increase for the 119

PAGE 120

remainder of the test. This is more clearly illustrated in Figure 56, in whic h the creep data is plotted as creep strain vs. creep strain rate. T he creep rate increased more rapidly for the C modification and the C+N modi fication than rates for the baseline and C+B modification. The faster rate increase indicates a gr eater level of accumulated damage in the form of di slocation formation in channels, creep cavitation, crack formation and growth, and void coalescence [82-84]. The baseline specimen showed the greatest creep elongation (29.14%) and r upture lifetime (300 hrs) while the C modification had the smallest elongation (2 1.65%) as well as the shortest rupture lifetime (251 hrs). Minimum creep rates were used as estima tes for steady-state creep rates for all creep tests. The rates for the 950 oC condition, where steady-state deformation only occurred for about 1% strain befor e the onset of tertiary creep, are shown in Table 5-2. Note that the minimum creep rate at this condition does not corre late to the overall creep behavior due to the strong influence of tertiary creep. The C modification had the lowest minimum creep rate (0.0041 %/hr), but also had the shortest lifetime due to a greater increase in creep rate as the test progressed. Time to the first several % creep strain can be very important data for SX superalloys used in engine components with very tight tolerances. While stress rupture of a turbine blade may not occur until greater than 20% strain, creep strain of 2% may cause the blade to contact the engine case, resulting in failure. Ther efore, the time vs. creep strain curves were re-plotted to get a cl earer picture of these important regions. The zoomed-in plot for the 950 oC testing is shown in Figure 5-7. The curves are very similar for all four alloys in th is region. The slight differ ences in steady-state creep rates 120

PAGE 121

can be seen as well as the onset of tertiary creep and strain rate i ncrease. Note that just below 0.5% and after approximately 80 hrs the C modification ra te begins to rapidly increase and crosses over the baseline curve. The times to various % strains are presented in Table 5-3. The results help to further illustrate the trends reported above. 5.2.2 Fractography The fracture surfaces were si milar for all a lloys crept at 950 oC (Figure 5-8), with groups of connected, square f eatures linked by torn ridges. These square regions indicate microcrack formation and subsequ ent octahedral slip in directions perpendicular to the applied stress axis. The outward growth of the square features is in the <011> type direction as intersections of active {111} slip planes occur ahead of the crack tip [82]. Microcracks began at pores or creep cavities in the baseline alloy and at cracked carbides or carbide/pore co mbinations in the modified alloys. Secondary cracking on square crack planes was observed more in the C-containing alloys than the baseline. EDS analysis of carbides on fracture surf aces revealed that carbide decomposition occurred during creep at 950 oC. Figure 5-9 shows carbides at the center of a square crack plane in the C+N modification and the corresponding EDS spectrum. This type of composition was not observed for any carbides in the fully heat treated condition, but it also does not indicate as mu ch Cr enrichment as the M23C6 carbides observed after 1000 oC/1000 hr exposure. The observed carbi des represent an intermediate stage of decomposition of MC to M23C6 carbides. The length of time (250 300 hours) was not long enough and test temperature (950 oC) was not high enough to get the full transformation that was observed after long-term exposure. 121

PAGE 122

5.2.3 Longitudinal Se ctions Light etched longitudinal sections observed in SEM (Figure 5-10) confirmed the presence of cracks that grew perpendicular to the applied stress direction and revealed that rafting occurred in all alloys at 950 oC. The rafted microstructure in the baseline alloy is represented in Figure 5-10A, in wh ich rafts approximately 5 10 m long formed uniformly throughout the specimen. Similar ra ft structures formed in the C-containing alloys, but rafting was interrupted at carbides as is generally observed [38]. Rods and plates that made up carbide networks in the C modification (Figure 5-10B) and blocky carbides in the C+N modification (Figure 5-10D) experienced extensive cracking. The majority of the carbide cracks extended into the surrounding matrix, confirming observations on the fracture surfaces. The small, spherical MC carbides in the C+B modification remained intact during creep (F igure 5-10C). Rare script networks that were retained during heat treatment in the B-containing alloy, however, did undergo cracking. 5.2.4 Summary Tertiary creep caused by damage in the form of microcracking at pores, cavities and carbides dominated defo rmation behavior at the 950 oC/300 MPa condition. The C modification and the C+N modification expe rienced more cracking than the other two alloys. This resulted in faster increas es in creep rates and shorter creep lifetimes. 5.3 Creep Testing at 850 oC/550 MPa A total of nine creep tests (4 HT SHT Un-HIPed, 4 HT SHT HIPed, and 1 CMSX486 specimen) were conducted at the 850 oC/550 MPa condition. Results of the CMSX-486 test are shown in Appendix F. All creep curves exhibited varying degrees of the three major stages of creep deformation. The HIPed samples exhibited better creep 122

PAGE 123

performance than the Un-HIPed specimens. The C modification HIPed test w as interrupted by a heating element failure at 67.8 hrs and 1.16 % strain. The test was stopped as an interrupted test, and results will be discussed when relevant. 5.3.1 Test Results The overall time vs. strain curves for the se ven full tests are shown in Figure 5-11. The general shapes of the curves are similar as all specimens at this condition underwent different amount s of primary, steady-state, and te rtiary creep. All three of the HIPed specimens (shown as solid lines ) had creep lifetimes longer than any of the other specimens, with HIPed, baseline CMSX-4 having the l ongest lifetime (653.8 hrs). The HIP treatment resulted in a 38% increase in rupture lifetime fo r the baseline, 35% increase for C+N modification, and a 15% in crease for C+B modification. Among the Un-HIPed CMSX-4 samples, the baseline once again had the best creep lifetime (473. 3 hrs). The onset of tertiary creep occurred in a similar manner for all alloys, as indicated by the positive slopes of the strain rate vs. st rain curves in Figure 5-12. Differences in lifetimes, therefore, were influenced by re lative minimum creep ra tes and strain rate increases associated with tertiary creep. A comparison of minimum creep rates (Table 5-4) further highlights the effe ct of HIP on creep behavior at 850 oC. The minimum creep rates for the HIPed specimens were significantly lower than for Un-HIPed samples. The first 1% and 100 hrs of the time vs. cre ep strain curves are shown in Figure 513. The non-smooth regions of the curves ar e due to slight extensometer slips. The C+B modification samples, bot h HIPed and Un-HIPed, experi enced more primary creep strain (approximately 0.5%) than the other CMSX-4 alloys. HIP processing did not appear to have a strong effect on primary cr eep behavior at this condition. Primary 123

PAGE 124

creep strains develop from sheari ng of precipitates [75, 77,78] and are likely independent of the diffe rences in porosity levels betwe en HIPed and Un-HIPed material. Table 5-5 contains the times to reach specific strain levels in the 850 oC creep tests. These specific values further demonstrate the trends described above. The HIPed specimens exhibited slower accumulation of creep strain, which became more apparent as strain increased. 5.3.2 Fractography Creep fracture surfaces for Un-HIPed specim ens (Figure 5-14) were very similar to those tested at 950 oC/300 MPa. Linked up square featur es connected by ductile tear ridges were visible over the ma jority of the surface. The square features were centered around pores and creep cavities (with associated ca rbides in the C-containing alloys) in the Un-HIPed condition. Similar to the 950 oC fracture surfaces, square features containing carbides exhibited more secondary cracking. In the HI Ped specimens, the features were smaller and centered around cr acked carbides in the modifications (Figure 5-15). Many of the square crack growth planes in the baseline HIPed specimen did not have pores or cavities at the cent er (Figure 5-15A). EDS analysis revealed all carbides on fracture surfaces to be Ta rich. No Cr rich carbides were observed on the fracture surfaces, revealing that decomposition from MC to M23C6 carbides seen during creep testing at 950 oC did not occur at 850 oC. 5.3.3 Longitudinal Sections SEM imaging of light etched longitudinal se ctions revealed cracking normal to the stress direction as in the 950 oC samples, but no rafting was observed. In the UnHIPed condition, cracks preferentially fo rmed at pores in baseline CMSX-4 (Figure 516A), and large carbide networks in the C m odification underwent significant cracking 124

PAGE 125

that extended into the surrounding matrix (Fi gure 5-16B). In the C+B modification, the small, spherical carbides remained intac t, but de-cohesion and crack formation occurred at the interface between carbides and / matrix (Figure 5-16C). D e-cohesion and cracking was observed in the C+N modification carbides, but some of the blocky carbides remained intact, as shown in Figure 5-16D. Longitudinal sections from the HIPed specimens also showed signs of carb ide cracking and de-cohes ion (Figure 5-17). A longitudinal section was prepared from the center of the gage section of the HIPed, C modification interrupted (at 1.16% st rain) test specimen. All carbide networks remained intact (Figure 5-18A), confirming that carbide cracking did not occur during primary creep or the b eginning of steady-state creep. A stray grain with carbides at the boundaries was also observed in this sample, as shown in Figure 5-18B. Note that no such grains were detected in the as -cast or any heat treated condition. 5.3.4 Summary Baseline CMSX-4 performed better in creep at 850 oC/550 MPa than the Cmodified alloys. HIP processing improved creep lifetimes by reducing minimum creep rate. The B-containing alloy, both in the HIPed and Un-HIPed conditions, experienced slightly more primary creep st rain than the other alloys. Post-tested microstructures revealed carbide cracking and de-cohesion of carbides from the surrounding matrix. 5.4 Creep Testing at 750 oC/800 MPa Creep tests were conducted at 750 oC/800 MPa in an attempt to observe the impact of minor additions on primary creep. Te sts at this condition involved only UnHIPed HT SHT material. Significant prim ary creep strains developed in all alloys and impacted overall cr eep lifetimes. 125

PAGE 126

5.4.1 Test Results All three stages of creep are apparent in t he full time vs. strain creep curves in Figure 5-19. Each sample under went signifi cant primary creep strain followed by an extended period of steady-state creep and a ve ry small amount of tertiary creep near failure. The initial C+B modification test resulted in a rapid and unexpected creep strain of approximately 10%. A second test of t he C+B modification was conducted to confirm the large primary creep strain observed in the first test. T he flat regions of the strain rate vs. strain curves (Figure 5-20) r epresent steady-state creep near the minimum creep rate. This region accounted for a larger portion of the overall creep strain at 750 oC than at the other two conditi ons. The relative heights of the maximum strain rate levels in Figure 5-20 clearly illustrate the differences in magnit ude of primary creep. The C modification and the C+N modification underwent faster primary creep rate than baseline CMSX-4, and the C+B modification exper ienced significantly fa ster strain rates than all of the alloys. In addition to identifying minimum creep rates as done for the other conditions, maximum creep rates were also determined for the 750 oC tests (Table 5-6). The values in the table repres ent the maximum and minimum st rain rate values of the curves in Figure 5-20. The maximum rate s are in the same ranked order as the minimum creep rates. The B-containing te sts had significantly higher maximum and minimum rates than the other alloys, and the ra tes for the baseline te st were the lowest. The zoomed-in versions of time vs. strain curves for the 750 oC creep tests are presented to 3% strain in Figure 5-21 instead of 1%, as in the other conditions, to capture the extent of the di fference between the C+B modification and the other alloys. Both B-containing samples reached 3% creep strain after only one hour of testing, and 126

PAGE 127

primary creep continued at bot h a faster rate and for a longer time than in the other alloys. Times to reach specific strain levels in the 750 oC creep tests are presented in Table 5-7. This test condition (lowest temperature and highest stress) produced the largest differences in creep rupture lifetimes between base line CMSX-4 and the Cmodifications, particularl y the B-containing alloy. 5.4.2 Fractography The 750 oC/800 MPa fracture surfaces (Figure 5-22) featured fewer of the square crack features which were so prevalent at t he two higher test temperatures. The square features that were observed, however, were similar in that they had either pores or cracked carbides at their centers. The ma jority of the surfaces were comprised of stepped tear ridges larger than those observed at 850 oC and 950 oC. Closer examination of tear ridges revealed small dimples, characteristic of ductile failure, as shown in Figure 5-23. As in the other conditions, carbides of all morphologies experienced some cracking. A ll analyzed fracture surface carbides were Ta rich, giving no indications of chemical carbide decomposition. 5.4.3 Longitudinal Sections Post-tested 750 oC/800 MPa microstructures (Figure 5-24) were characterized by deformed regions of / and fewer cracked carbides than at higher temperature testing. The baseline microstructure contained / stretched in the stress direction and regions in which precipitates were sheared (Figure 524A). At this lower temperature, dislocations do not experience as much th ermally activated climb and are hence more likely to shear through the strengthening precip itates at large enough stresses. In the C+B modification, highly deformed / was observed near sma ll, spherical carbides 127

PAGE 128

(Figure 5-24B). The deformation appeared to flow around the carbides, and some of the highly deformed regions sepa rated from carbides. The absence of carbide cracking in the longit udinal sections very near the frac ture surface suggests that the cracking of carbides visible on the fr acture surfaces occurr ed during final fracture. 5.4.4 Summary Creep tests at 750 oC/800 MPa resulted in the lar gest differences in lifetime between baseline CMSX-4 and the C-modified alloys among the three creep testing conditions. The C+B modification, in particu lar, showed significantly shorter rupture lifetimes and exhibited much larger primary creep strains than the other variants. Microstructural examination revealed less cracking at pores and carbides and more deformation of the / matrix than at higher test te mperatures. The results here highlight the importance of te sting at multiple condition s. The B-containing alloy exhibited very similar creep defo rmation to the baseline at 950 oC, but a decrease in temperature of 200 oC, coupled with an increase in applied stress, led to dramatically dissimilar behavior between the alloys. 5.5 Analysis The following sections cover several studi es designed to compliment creep results and evaluate possible explanatio ns for the observed behavior. These studies sought to connect creep deformation processes to samp le observations. The efforts included measurements from tested specimens and compositional data from metallographic sections. 5.5.1 Sizes of Pores and Cavities Casting pores have been reported to grow during both heat treat ment [55,127] and creep deformation [52]. Direct, quantitativ e size comparisons were made between 128

PAGE 129

pore-like features on tensile fracture surfaces and those on creep fracture surfaces for tests conducted at 850 oC. The average elliptical areas calculated from measurements of the major and minor axes, are shown in Table 5-8. The pores in the crept samples were larger than those in the tensile test specimens for all alloys, suggesting growth of pores did occur during creep deformation. The majority of the pores observed in the Cmodified alloys contained carbides inside of t hem, and it is difficult to determine whether these represent original casting pores or ca vities that formed during deformation. An interesting correlation exists between the measured pore sizes and the sizes of carbide features for each modified alloy. For example, the measured elliptical cavity sizes for the C modification are similar to the size of th e script carbide networks in this alloy. This suggests that the size of cavities which form at carbides during high temperature deformation may be impacted by the carbide size. As expected, hole-like featur es observed at the center of square crack growth planes were much smaller in HIPed samples than in Un-HIPed material creep tested at 850 oC (Figure 5-25). The presence of very small holes (< 5 m) in the HIPed material indicates the following possibilities: HIP pr ocessing did not entirely eliminate casting porosity, small pores formed during heat treatment steps after HIP, small cavities formed during creep deformation, or any combination of the three. 5.5.2 Creep Specimen Ellipticity and Primary Creep Round (diameter1 = diameter2) creep specimens that undergo large primary creep strains often exhibit elliptical (diameter1 > diameter2) cross sections after testing due to heterogeneity of slip [77,78,128 ]. The major and minor diamet ers of gage sections for all crept specimens were measured directly adjacent to the fractu re surface. The results are shown in Appendix G. The only two samples that exhibited significant 129

PAGE 130

elliptical ratios, 1.15 and 1.13, were the two C+B modification samples tested at 750 oC that experienced nearly 10% pr imary creep strain. The ot her samples tested at this condition experienced smaller primary creep strains between 2% and 3%, and the gage sections remained nearly circular. Misorientation of greater than 10o from the preferred [001] direction has been identified as a cause of la rge primary creep strains and a ssociated elliptical cross sections in SX superalloys [79]. This does not appear to be the caus e of the high levels of primary creep in the C+ B modification specimens, however, as the cast bar orientation data (Appendi x A) confirms an angle (misorientati on from the [001] direction) of only 5.3o in the bar from which the samples were machined. 5.5.3 Local Compositional Di fferences and Primary Creep Local difference in composition was also explored as a potential cause of the large primary creep strains at 750 oC in the B-containing alloy. A hypothesis was developed after the observation of highly deformed microstr ucture in the direct vicinity of small, spherical carbides (Figur e 5-24C). Shearing of precipitates, which can lead to large primary creep strains, is largely de termined by compositionally sensitive / lattice misfit [75]. Carbides change local chemistr ies and misfit values by consuming certain carbide forming elements such as Ta. If the lo cal composition varies significantly, misfit can change enough to produce regions near carb ides that are more susceptible to shearing than the rest of the microstructure. To evaluate the possibility of such an effect, EDS line scans were produced in pos t-creep microstructures to check for any clear signs of composition changes near carb ides. Three line scans were conducted in the C modification specimen, 5 in the C+N modi fication specimen, and a total of 7 in the two C+B modification samples. One su ch scan, of a B-containing sample, is 130

PAGE 131

131 represented in Figure 5-26. In all of the scans, the carbides were clearly enriched in Ta and Ti, but there were no indications of Ta or Ti depletion in the vicinity. Based on this result, the primary creep behav ior observed could not be conclusively tied to local differences in composition near carbides. 5.6 Summary All creep rupture data was plotted in t he form of Larson-Miller Parameter (LMP) curves (Figure 5-27) to present an overall vi ew of the results. Plotting creep rupture data in this manner is a common practice used to compare creep behavior of different alloys. The parameter value is determined from the time to rupture and the test temperature, and then it is plotted agains t the applied stress [5]. Curves and data points that are higher and further to the right represent be tter creep rupture strength. Generation of material LMP curves requires a large number of tests at a wide variety of creep conditions. The data presented here aims to demonstrate trends and not to provide representative material curves. Among the Un-HIPed materials, the baseline alloy clearly outperformed the Cmodified variants. The low parameter values for the Bcontaining alloy (in blue) at 800 MPa applied stress demonstrate th e strong effect of primary creep on r upture life at 750 oC. HIP processing im proved creep performance, as indicated by the solid point s to the right of the curves.

PAGE 132

0 200 400 600 800 1000 1200 1400 05101520253035Strain (%)Tensile Stress (MPa)A 0 200 400 600 800 1000 1200 1400 05101520Strain (%)Tensile Stress (MPa)B C 0 200 400 600 800 1000 1200 1400 0510152025Strain (%)Tensile Stress (MPa) 0 200 400 600 800 1000 1200 1400 05101520Strain (%)Tensile Stress (MPa)D Figure 5-1. Stress strain curves for tensile tests conducted at 850 oC on fully heat treated (HT SHT) samples. A) Base line, B) C modification, C) C+B modification, D) C+N modification. Table 5-1. Important values from tensile testing at 850 oC. Estimated yield stresses are similar, but the baseline exhibits greater UTS and elongation values than the C-modified alloys. Estimated Yield Strength (MPa) Ultimate Tensile Strength (MPa) Elongation (%) Baseline 854.4 1201.6 30.05 C modification 841.0 1151.2 16.90 C+B modification 843.7 1142.9 20.12 C+N modification 836.0 1110.2 15.23 132

PAGE 133

Figure 5-2. SEM micrographs of fracture surfaces from tensile tests at 850 oC. A) Baseline, B) C modification (BSE), C) C+ B modification, D) C+N modification (BSE). 133

PAGE 134

t C. n, Figure 5-3. SEM secondary electron-backsca matrix cracking in longitudinal sec A) SE of C modificatio n, B) BSE of C m D) BSE of C+B modificati modification. The applied stress direction is vertical. ttered electron pairs of carbide and / ions of tensile specimens tested at 850 oodification, C) SE of C+B modificatio on, E) SE of C+N modifi catio n, F) BSE of C+N 134

PAGE 135

Figure 5-4. SEM micrographs of a longitudinal section from the baseline CMSX-4 tensile specimen tested at 850 oC. A) crack formation, B) shearing of precipitates. The applied stress direction is vertical. 0 5 10 15 20 25 30 35 050100150200250300350Time (hrs)Creep Strain (%) Baseline C modification C+B modification C+N modification Figure 5-5. Time vs. strain curves for 950 oC/300 MPa creep tests. All samples underwent significant tertiary creep. 135

PAGE 136

0 0.05 0.1 0.15 0. 0.25 2051015202530Creep Strain (%)Creep Strain Rate (%/hr) Ba seline C modification C+B modification C+N modification Figure 5-6. Creep strain vs. creep strain rate curves for 950 oC/300 MPa creep tests. The increase of strain rate with additional strain is indicative of tertiary creep and material damage accumulation. Table 5-2. Minimum creep rates for tests conducted at 950 oC/300 MPa. Minimum Creep Rate (%/hr) Baseline 0.0050 C modification 0.0041 C+N modification 0.0057 C+B modification 0.0049 136

PAGE 137

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 050100150Time (hrs)Creep Strain (%) Baseline C modification C+B modification C+N modification oFigure 5-7. Time vs. 1% strain curves for 950 C/300 MPa creep tests. Table 5-3. Time to various % creep strains for tests at 950 oC/300 MPa. All times are in hrs. 0.01% 0.1% 0.2% 0.5% 1% 2% 5% 10% Rupture Baseline 0.06 11.1 30.3 91.7 138 173 218 254 300 C modifica tion 2.09 15.5 38.7 89.7 125 156 195 225 251 C+B modification 0.2 5.86 16.2 69.5 128 171 219 255 290 C+N modification 0.56 10.7 25.4 75.4 120 151 193 227 261 137

PAGE 138

Figure 5-8. SEM micrographs of fracture surfaces from creep tests at 950 oC/300 MPa. A) Baselin e, B) C modifi cation, C) C+B modification, D) C+N modification. 138

PAGE 139

Figure 5-9. Chromium (Cr) rich carbides on the fracture su rface of a C+N modification specimen crept at 950 oC/300 MPa. A) SEM SE micrograph, B) SEM BSE micrograph, C) EDS spectrum. 139

PAGE 140

140 Figure 5-10. SEM micrographs of longitudinal sections mens tested in creep at 950 oC/300 MPa. A) Baseline, B) C modification, C) C+B modification, D) C+N modification. The applied st ress direction is vertical. of speci

PAGE 141

0 5 10 15 20 25 0100200300400500600700Time (hrs)Creep Strain (%) Baseline_HIP Baseline C+B modification_HIP C+B modification C+N modification_HIP C+N modification C modification oFigure 5-11. Time vs. strain curves for 850 C/550 MPa creep tests. All curves exhibit all three regions of creep strain. 0 0.05 0.1 0.15 0.2 0.25 0.3a0.35te0.4 (%/h0.45 0.5 0 510152025Creep Strain (%)Creep Strain R r) Baseli ne_HIP Baselin e C+B mod_H ification IP C+B modification C+N mod_H ification IP C+N modification C modific Figure 5-12. Creep strain vs. creep strain rate curves for 850 oC/550 MPa creep tests. ation 141

PAGE 142

Table 5-4. Minimum creep rates for tests conducted at 850 oC/550 MPa. Minimum Creep Rate (%/hr) Baseline 0.0125 Baseline_HIP 0.0079 C modification 0.0164 C+B modification 0.0143 C+B modification_HIP 0.0070 C+N modification 0.0106 C+N modification_HIP 0.0085 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0204060801Time (hrs)Creep Strain (%) 00 Baseline_HIP Baseline C+B modification_HIP C+B modification C+N modification_HIP C+N modification C modification oFigure 5-13. Time vs. 1% strain c urves for 850 C/550 MPa creep tests. able 5-5. Time to various % creep strains for tests at 850 oC/550 MPa. All times are in hrs. 0.01% 0.1% 0.2% 0.5% 1% 2% 5% 10% Rupture T Baseline 0.08 0.72 1.58 15.6 57.2 134 283 394 473.3 Baseline_HIP 0.05 1.61 5.28 35.9 97.4 215 401 541 653.8 C modification 0.01 0.57 5.18 29 63.1 119 244 341 386.8 C+B modification 0.02 0.02 0.47 7.7 38.6 100 233 335 410 C+B modification_HIP 1.28 1.29 1.69 3.9 49.9 132 302 421 472.1 C+N modification 0.16 0.8 2.84 26.9 73.9 163 295 391 432 C+N modification_HIP 0.26 0.76 1.28 16.4 64.5 162 368 503 583.4 142

PAGE 143

i u F gure 5-14. SEM micrographs of fracture s not undergo hot isostatic pressing) specimens at 850 Baseline, B) C modification, C) C+B rfaces from creep tests of Un-HIPed (did oC/550 MPa. A) modification, D) C+N modification. 143

PAGE 144

Figure 5-15. SEM micrographs of fracture surfaces from cr eep tests of HIPed specimens at 850 oC/550 MPa. A) Baseline, B) C+B modification (BSE C+N modification. Note that no fractu re surface is sho ), C) wn for the HIPed C modification sample because that te st was interrupted after 1.16 % creep strain. 144

PAGE 145

145 Figure 5-16. SEM micrographs of longitudinal sections of Un-HIPed specimens tested in creep at 850 oC/550 MPa. A) Baseline, B) C modification (BSE), C) C+B modification, D) C+N modi fication. The applied stress direction is vertical.

PAGE 146

Figure 5-17. SEM micrographs of longitudinal sections of HIPed specimens tested in creep at 850 oC/550 MPa. A) Baseline, B) C+B modification, C) C+N modification. The applied stress direction is vertical. Figure 5-18. SEM micrographs from longitudi nal section from HIPed C modification creep specimen tested at 850 oC/550 MPa interrupted at 1.16% strain. A) intact carbide rods, B) stray grain. The applied stress direction is vertical. 146

PAGE 147

0 5 10 15 20 25 0100200300400500600 Time (hrs)Creep Strain (%) Baseline C modification C+B modification_Test 1 C+B modiification_Test 2 C+N modification oFigure 5-19. Time vs. strain curves for 750 C/800 MPa creep tests. All curves exhibit substantial primary creep st rain followed by extended periods of steady-state creep and a small amount of tert iary creep near rupture. 0 0.5 1Cr1.5 2 2.5 3 3.5 4 4.5 5 0 5 10152025 Creep Strain (%)eep Strain Rate (%/hr) Baseline C modification C+B modification_Test 1 C+B modiification_Test 2 C+N modification Figure 5-20. Creep strain vs. creep strain rate curves for 950 oC/800 MPa creep tests. 147

PAGE 148

Table 5-6. Maximum and minim um cr eep rates for tests conducted at 750 oC/800 MPa. Maximum Creep Rate (%/hr) Minimum Creep Rate (%/hr) Baseline 0.795 0.0147 C modification 1.44 0.0202 C+B modification_Test 1 4.94 0.0913 C+B modification_Test 2 3.87 0.0674 C+N modification 1.37 0.0178 0 0.5 1 1.5 2 2.5 3 e Creep Strain (%) Baseline C modification C+B modification_Test 1 C+B modiification_Test 2 C+N modification01 2345 im T (hrs) Figure 5-21. Time vs. 3% strain for 750 oC/800 MPa creep tests. % creains for tests atoC/800 MPa. All times are in hrs. 0.01% 0.1% 0.2% 0.5% 1% 2% 5% 10% Rupture c urves Table 5-7. Time to various ep str 750 Baseline 0.12 0.52 0.68 1.04 1.49 2.57 69 365 501.5 C modification 0.03 0.24 0.34 0.52 0.74 1.36 61.7 262 281.4 C+B modification_1 0.09 0.31 0.41 0.54 0. 67 0.84 1.17 4.08 92.2 C+B modification_2 0 0.4 0.52 0.69 0.86 1.07 1.48 5.21 135.3 C+N modification 0.12 0.27 0.35 0.53 0.75 1.47 85 312 370.1 148

PAGE 149

Figure 5-22. SEM micrographs of fracture surfaces from creep tests at 750 oC/800 MPa. A) Baseline, B) C modifi cation, C) C+B modification, D) C+N modification. Figure 5-23. SEM micrograph of ductile dimple s near a pore on the fracture surface of a C+B modified CMSX-4 s pecimen crept at 750 oC/800 MPa. 149

PAGE 150

150 Figure 5-24. SEM micrographs of longitudinal sections of specimens tested in creep at 750 oC/800 MPa. A) Baseline, B) C modification, C) C+B modification, D) C+N modification. The applied st ress direction is vertical. Table 5-8. Average elliptical areas of pores /cavities observed on fracture surfaces of failed specimens tested at 850 oC. Baseline (m2) C modification (m2) C+B modification (m2) C+N modification (m2) Tensile 47.7 141.5 52.2 106.3 Creep 76.0 314.8 65.2 195.8

PAGE 151

Figure 5-25. SEM micrographs of hole-like feat ures on the fracture surfaces of baseline CMSX-4 specimens tested in creep at 850 oC/550 MPa. A) Un-HIPed, B) HIPed. Intensity (A.U.) Ni Ta Ti Al B0102030405060distance along scan ( m) section of C+B modified CM SX-4 tested in creep at 750 Figure 5-26. Representative EDS line scan across carbide phases in a longitudinal a. A) SEM micrograph with location of line scan, B) EDS line scan result showing relativ e oC/800 MP concentrations of elements al ong the length of the scan. 151

PAGE 152

100 1000LMP, P = T[log(t) + 20]*(10 )20 22242628303234-2Applied Stress (MPa) Baseline C modification C+B modification C+N modification Baseline_HIP C+B modification_HIP C+N modification_HIPFigure 5-27. Creep data summari zed in the Lars on-Miller parameter (LMP) curve determined from test temperature and ime and plotted against the applied stress. format. The x-axis parameter value is rupture lifet 152

PAGE 153

CHA PTER 6 RESULTS: FATIGUE BEHAVIOR Fatigue testing was conducted to study the effects of alloy modifications on crack initiation and fatigue lifetimes Gas turbine components are subjected to a variety of cyclic loading conditions, including LCF and TMF, due to engine startup and shutdown. HCF conditions develop from high frequency vi brations present during engine operation. HCF failures are controlled by crack initia tion, which is often quickly followed by catastrophic failure. HCF cracks can form at a variety of material defects or secondary phases, and creating potential sites for cra ck formation is a primary concern when introducing a phase (carbides in this case) to an alloy. All HCF tests were conducted at 850 C, with an input stress R-ratio of 0.1 and input stress range of 690 MPa. U nder th is condition, the maximum applied tensile stress is below the alloy yield stress. Thr ee rounds (round = 1 test of each alloy variant) of testing were conducted on HT SHT material in both the Un-HIPed and HIPed conditions, and 2 rounds of testing were conduc ted on the LT SHT material in the HIPed condition. The gage sections of all fatigue sp ecimens were carefully polished to reduce the likelihood of crack formati on at surface machining marks. This facilitated the study of microstructure-controlled fatigue behavior. 6.1 Un-HIPed, HT SHT The first 3 rounds of fatigue testing we re performed on alloys that had undergone the HT SHT and no HIP processing. A frequency of 20 Hz was used for the first round, but the servo-hydraulic load levels did not reach the desired maximum and minimum values. Actual stress ranges were near 620 MPa and actual R values were approximately 0.15. The frequency was reduced to 15 Hz for all subsequent testing to o 153

PAGE 154

achieve the input conditions. Fatigue behav ior of alloys in this condition was controlled by porosity. 6.1.1 Test Results Fatigue lifetime results for this condition ar e presented in Figure 6-1. Results were plotted as lifetime vs. alloy modification, as opposed to more common stress vs. lifetime plots, because all testing was conducted at the same condition. As indicated in the figure, the tests at 20 Hz show ed longer lifetimes than the 15 Hz tests. This was due to the decreased stress amplitudes in the 20 Hz tests. Overall, baseline CMSX-4 outperformed the C -containing modifications, which had similar lifetimes amongst them. s were significantly below 1 million cycles, and 7 of the 9 C-containing spec alloys was associated with carbides, and cracked carbides were frequently observed inside of pore s growth. Crack growth regions were small on the baseline fracture surfaces and c onsisted of areas of stage-I crack growth (along crystallographic planes of high stress) and stage-II crack growth (in a direction All fatigue lifetime imens had lifetimes below 200,000 cycles. 6.1.2 Fractograph y All fracture surfaces were examined usi ng SEM to characterize crack initiation sites, crack growth behavior, and overall appearance. Characteristic crack initiation sites for the Un-HIPed specimens are shown in Figure 6-2. In all Un-HIPed tests, failure causing fatigue crack initiations occurred at internal pores. The porosity at initiation sites in the modified s. The fracture surfaces of baseline alloy specimens contained smooth planes at angles roughly 45o from the stress direction that were continuous across most of the surface. This indicates large scale cl eavage along {111} type crystallographic plane that occurred after crack initiation and initial crack 154

PAGE 155

normal to the stress a xis) [4]. Fracture su rfaces of the modified alloys were overall rough very Areas ies for ite path from crack formation to failure. An gure 64. This fracture surface contains two circular crack grow rew en. p the he er in appearance than the baseline and featur ed broken carbides, steps, and few smooth crystallographic facets. Circular Mode-I (stage-II) crack planes (Figure 63A) were observed around the primary initia tion sites, representing crack growth perpendicular to the stress direction. These planes were highly reflective in light and were much larger than the crack growth regi ons seen in the baseline specimens. of the overloaded, final frac ture regions showed signs of rapid crack propagation in a direction outward from the crack init iation and growth areas (Figure 6-3B). Features of each fracture surface were cataloged and summarized on individual Microsoft PowerPoint slides for ease of refe rence. The fracture surface summar all HCF tests can be found in Appendix H. Each slide contains an overall low magnification image of the fractu re surface, SEM micrographs of the crack initiation s and other features, and a description of the example is shown in Fi th regions around two distinct crack init iation sites. Both crack planes likely g simultaneously, which contributed to the relati vely low cyclic lifetime in this specim 6.1.3 Longitudinal Sections Light etched longitudinal sections we re prepared from each Un-HIPed HCF specimen to observe the post-tested microstr uctures in a similar manner as in cree testing. The / structure was largely intact except for some slight stretching in stress direction, indicating that no signific ant general plastic deformation occurred. T majority of the pores observed in the baseline were free of cracks (Figure 6-5A). The proximity of carbides and pores was conf irmed here, as shown in Figure 6-5C and Figure 6-5D. Cracks in carbide phases fo rmed perpendicular to t he stress direction, 155

PAGE 156

particularly in the rod and plate carbides comprising the networks in the C modification (Figure 6-5B). The cracking did not extend signific antly into the surrounding as it did in tested creep specimens. Some cracks obs / matrix erved within 500 m of the fracture surface had zigzagged crack paths reported ress IPed es re regions in the baseline alloy exhibited uninterrupted e overloaded areas in the modified alloys were much rough e HT sulted as shown in Figure 6-6. This indi cates the formation and growth of cracks along octahedral {111} slip planes. These cracks are similar to those that have been elsewhere in SX superalloys tested in HCF [39 ]. The cracks form angles with the st axis that are similar to those formed by the crystallographic facets observed on the fracture surfaces. 6.1.4 Summary Baseline CMSX-4 outperformed the C-modifi ed alloys in HCF testing of Un-H material. All crack initiations occurred at internal pores, and initiating pores in the modified alloys were associated with carbides Fracture surfaces and microstructur revealed cracking along crystallographic pl anes and Mode-I cracking perpendicular to the stress direction. Final fractu crystallographic cleavage, whil er in nature. 6.2 HIPed, HT SHT Three rounds of HCF tests were conducted on material that had undergone th SHT cycle and HIP treatment. HIP processing improved fati gue lifetimes and re in different fatigue initiation behavior as co mpared to the Un-HIPed alloys. Fatigue lifetimes were impacted by crack initiation sites. 156

PAGE 157

6.2.1 Test Results Fatigue lifetimes (Figure 6-7) were signi ficantly greater than those for the UnHIPed, HT SHT material. Note the order of magnitude difference between the y-axes Figure 6-7 and Figure 6-1. The baseline, aside from the shortest lifet ime data point to be explained below, once again outperformed the m in odified allo ys. The longest fatigue lifetim rtest e 6.2.2 for lifetime. Crack initiation o ccurred at carbides of various morphologies in the modified alloys. Blocky carbides approximately 20 40 m in size were observed at initiation e among all the samples tested was the baseline specimen that failed after 8,797,800 cycles. The C+N modified alloy exhi bited the second longest fatigue lifetimes ranging between approximately 4 and 6.5 million cycles, followed by the C+B modification (800,000 1.5 million cycles). The C modification clearly had the sho fatigue lifetimes and experienc ed the smallest lifetime improvement due to HIP processing. The lifetimes were, in fact, ve ry near those for Un-HIPed specimens. Note that the lack of scatter in the results gives the appearance of one point for the thre closely plotted C modification data points. Fractographgy As in the Un-HIPed samples, all fatigue crack initiations occurred at internal features. Representative init iation sites are shown in Figur e 6-8. On the fracture surfaces of the two longer lifetime baseline samples, primary crack initiation was identified at small pores. These pores were much smaller than those responsible crack initiation in the Un-HIPed condition. T he initiation site for the lowest lifetime (742,208 cycles) baseline test contained a pore or / eutectic region that was significantly larger than any others observed at this condition. This was likely a rare defect that was not removed by heat treatm ent and HIP processing, causing the shorter 157

PAGE 158

sites in the C+N modification samples. T he C+B modification had crack initiatio consisting of groups of small, rounded ca rbides, an n sites d crack initiations in the C s, were at cracked carbide networks 100 e of s verload fracture regions were not as ified alloys as in the Un-HIP ed condition. The final fracture areas conta f rved in the baseline samples. In the C modification, cr nd r od-like features was appa reased modification, which had the shortest fatigue lifetime 200 m in size. The flat, circular crack growth planes seen in Un-HIPed samples were absent in the HIPed fracture surfaces. Growth of cracks was primarily along high stress crystallographic planes (stageI cracking). An example of this type of crack growth originating at the primary initiation site is shown in Figure 6-9. Three distinct, ridged features indicate the crack growth directions. Some initiation sites did show evidenc mixed stage-I and stage-II cracking in the immediate vicinity of the initiation feature. Overall, the crack growth regions were much smaller in the HIPed specimens a compared to the Un-HIPed crack planes. The o rough in the mod ined long, smooth crystall ographic cleavage planes similar to those observed on the baseline fracture surfaces. 6.2.3 Longitudinal Sections Similar to the Un-HIPed condition, overall / structures showed little evidence o uniform plastic deformation. Characteristic f eatures present in the longitudinal sections of each alloy are shown in Figure 6-10. No cracking was obse acking of thin plate a rent, as shown in Figure 6-10B. Small, rounded carbides in the C+B modification remained intact (Figure 6-10C). Signs of localized plastic deformation were present near carbides, particularly in the C+B modifi cation and the C+N modification. Inc cyclic lifetimes resulted in larger localized stresses from repeated maximum and 158

PAGE 159

minimum applied loads. In some cases, localized deformation resulted in decohesion between carbide and matrix (Fi gure 6-10D). Metallography of several thread sections, which are not subjected to significant stresse s during testing, did not reveal any local des. This confirmed th at the plastic deformation occurred during HCF ns in identified in the literature ip bands in the / matrix impinging on carbides, and these featu ns cker on specimens that had experienc ed longer fatigue lifetimes. Cracking was obse deformation at carbi loading and was not due to the HIP process. Several longitudinal sections were deep etched in order to more thoroughly examine changes to carbides from fatigue loadi ng. This effort verified the observatio made in the light etched condition, namely cra cked carbide plates in the C modification (Figure 6-11A) and intact small carbides in the C+B modification (Figure 6-11B). Interesting markings were also observed on t he surfaces of some carbides, as shown Figure 6-12. These short, closely spaced features have been [70] as evidence of sl res are thought to be an effect of localized plastic deformation. The edge surfaces of several longitudinal sections were exam ined, motivated by reports of recrystallization of nano-sized grai ns at the surface of fatigue specime caused by surface preparation [129]. Although no signs of recrystallization were found, appreciable oxide formation was observed. Ox ide layers, rich in Cr and Al, formed on the surfaces of fatigue spec imens at test temperature (Figure 6-13A). The oxide formation caused depletion of Al and formation of a phase layer below. These layers were thi rved to form at the ox ide layer and extend into the layer (Figure 6-13B). Although no surface crack initiations were found to caus e any of the fatigue failures in this study 159

PAGE 160

these cracks represent another possible cr ac k initiation mechanism that could poten g ng. C s nt t 6.3 HIPed, LT SHT Two rounds of HCF testing were th alloys that had undergone HIP proce ent tially lead to failure at longer cyclic lifetimes. 6.2.4 Transmission Elect ron Microscopy (TEM) Localized plastic deformation was studi ed using TEM and a FIB technique to selectively acquire foils at carbide phases in tested HCF samples. Evidence of heavy dislocation formation was observed at carbide matrix interfaces, as shown in Figure 6 14A. Dislocations in high concentrations were present in the immediate vicinity of carbide phases, and some cracks formed at ca rbides and extended into the surroundin matrix (Figure 6-14B). These dislocation concentrations contrasted with the limited dislocation structures in channels observed away from carbides and in the baseline alloy (Figure 6-15). These studies confi rmed the formation of localized plastic deformation at carbides during fatigue loadi 6.2.5 Summary HIP processing significantly improved fatigue lifetimes in all alloys except for the modification. All crack initiations occurred at pores in the baseline alloy and at carbide in the modified alloys. Baseline alloy specimens had longer fatigue lifetimes than the modifications, and the lifetimes of the m odified samples corresponded to the differe carbide morphologies and their roles in crack in itiation. Localized plastic deformation a carbides was observed with SEM and confirmed using TEM. c onducted wi ssing and the LT SHT schedule. The goal was to isolate potential heat treatm effects on carbide crack initiation sites by comparing the fatigue behavior at this condition to the results for the HIPed, HT SHT alloys. Fatigue lifetimes and crack 160

PAGE 161

initiation sites, however, were simi lar to thos e in the Un-HIPed condition, indicating the particular HIP cycle applied to th that is group of bars did not reduce porosity as much as 6.3.1 ly l in a result ase, the clam the cycle applied to the HIPed, HT SHT bars. Test Results Fatigue lifetimes (Figure 6-16) were significa ntly shorter than those for the HIPed, HT SHT condition. The plot of the lifetimes in fact, resembles the plot for the UnHIPed, HT SHT HCF tests. The two basel ine alloy specimens exhibited the longest lifetimes (378,911 cycles and 291 ,575 cycles), and all the modified alloy samples had lifetimes between 75,000 and 200, 000 cycles. Only one data point was collected for the C+N modification because a power outage duri ng setup of the second test severe bent the specimen in compression, rendering it non-usable. 6.3.2 Fractography Fracture surfaces from the HIPed, LT SHT tests were also very similar to those in the Un-HIPed, HT SHT condition. All failure-c ausing fatigue cracks initiated at interna pores. Carbides were observed at the initiation pores in the modified CMSX-4 specimens as in the Un-HIPed material. Char acteristic initiation si tes are presented Figure 6-17. The light colored ox ide layer at the initiation site in Figure 6-17A is of exposure of the fracture surface to high te mperature after failure. In this c shell furnace was not removed from ar ound the specimen until several hours after failure. The overall appearance of fracture surfaces was consistent with observations from Un-HIPed samples, with smooth crystallographic planes 45o to the stress axis for the baseline and large, circular Mode-I cra ck growth planes for the modified alloys. 161

PAGE 162

6.3.3 Longitudinal Se ctions Features of the examined longitudinal sections are shown in Figure 6-18. Residual porosity retained a fter the HIP process was apparent in all of the variants. Deformed structure at a number of pores suggested that i ncomplete pore closure occurred, as shown in Figure 6-18A. The HIP process relies on isostatic pressure and high temperature to collapse pores in on themselves and join collapsed edges through diffusion bonding. Significant deformation was visible at pores, but many of them did had tact carbides (Figure 6-18C and Figure 6-18D) were s s st corresponding initiation site sizes, and the second not undergo complete closure. More in seen than in the fatigue tested HIPed, HT SHT samples. Shorter lifetimes and fewer tensile load cycles developed smaller stresses at carbides, making them les prone to cracking. Crack initiation occu rred at pores before localized stresses were large enough to crack the carbide phases. 6.3.4 Summary HCF specimens in the HIPed, LT SHT condition behaved in a similar manner to the Un-HIPed, HT SHT material. As was the case in all HCF testing, fatigue lifetimes for baseline CMSX-4 were greater than the modified alloys. The low cyclic lifetimes, a compared to the HIPed, HT SHT alloys, we re caused by fatigue crack initiation at internal pores. It was determi ned that the HIP cycl e applied to these sample bars did not reduce porosity enough to significantly im prove fatigue lifetime or change the crack initiation mechanisms. 6.4 Analysis The following sections describe several techniques used to correlate fatigue lifetimes to observed fatigue crack initiation features. The first approach involved plotting fatigue lifetimes again 162

PAGE 163

applied a estim r the ,131]. HT of initiating feature and its size. The three shorte wn in Figure 6-20 belong to C modi not ot re method developed at the Air Force Research Laboratory (AFRL) [94] to ate mi nimum fatigue lifetimes. 6.4.1 Crack Initiating Feature Size vs. Cyclic Lifetime Sizes of crack initiation sites are often plotted against fatigue lifetimes to determine possible trends [130,131]. Generally, larger fa tigue crack initiation sites are associated with shorter fatigue lifetimes. Estimated area s for crack initiation features (pores o carbides) from all HCF specimens were det ermined from SEM micrographs. Due to variety of shapes observed, f eature areas were estimated as ellipses with major and minor diameters, as has been performed in recent superalloy fatigue studies [130 The plot showing the initiating crack feature areas and the corresponding fatigue lifetimes is shown in Figure 619. Although there is consi derable scatter in the data, a general trend correlating the longer fatigue lifetim es with smaller crack initiation areas is apparent. Figure 6-20 contains data only for the 12 HCF tests conducted on HIPed SHT alloys. The smallest initiation sites re sulted in the longest fatigue lifetimes. It is important here to note that connections can also be made between the type st lifetimes sho fication samples that experienced cr ack initiation at cracked script carbide networks greater than 8000 m2 in estimated elliptical area. Fatigue lifetimes were also plotted against se veral other size characteristics of the initiation sites. Plots of length to width ra tios (ellipticities) vs. fatigue lifetimes did exhibit any clear trends. It was anticipated that sharper f eatures with high er length to width ratios may have been associated with s horter fatigue lifetimes, but this was n the case. Lifetime data was also plotted agai nst the distance from the initiating featu to the outer surface of the specimen. It would be expect ed that initiations occurring 163

PAGE 164

nearer to the surface would result in shorte r fatigue lifetimes. The plots, however, did not show any obvious trends. A plot of the ratio of the distance from the edge to the area of the initiating feature, however, did exhibit a lif etime trend (Figur e 6-21). Lo fatigue lifetimes were associated with larger ratio values corresponding to longer distances from the edge and/or smaller ellipt ical areas. The only two data points with ratio values greater than unity showed the tw o longest fatigue lifetimes. Overall, fatigu behavior was controlled by bot nger e h the type of initiating feature and its corresponding size. 6.4.2 Fracture Mechanics Approach Recent work at the AFRL wit h HCF behavior of PWA 1484 (a 2nd generation SX superalloy) involved a fracture mechanics approach to estimate minimum fatigue lifetimes [94]. This approach was applied to all HCF data from the current investigation to compare actual cyclic lifetimes to t hose predicted by the method. The approach involves a series of assumptions that likely oversimplify the actual fatigue process. The assumptions are as follows: a fatigue crack init iates on the first cycle, the crack initiation area is circular with crack radius a, the growth of the cra ck is controlled by small crack growth behavior with a corresponding C and n value, and the crack grows through Mode-I crack growth governed by Eqn. 6-1 The da/dN value repr esents the growth of the crack during each cycle, and C and n are ma terial constants determined by crack growth data for CMSX-4 published by Schubert [6]. The values for K were determined for each test by measuring the crack initia ting feature and assuming circular crack geometry. Predicted minimum lif etimes were determined from the number of cycles to cause crack growth large enough that the remaining cr oss-sectional area of the specimen was such that the maximum applied load caused overload stress greater than the ultimate tensile strength. 164

PAGE 165

da/dN = C( K)n (Eqn. 6-1) The predicted cyclic lifetimes and correspondi ng actual lifetimes for all HCF tests are shown in Table 6-1. Actual lifetimes that were lower than the minimum predicted lifetimes are shown in red italics. It can be seen that only 7 of the 12 tests of Un-HIPed HT SHT alloys resulted in cyclic lifetimes gr eater than the minimum predicted value, but 11 of the 12 tests of HIPed HT SHT material had actual lifetimes greater than the predictions. For the HIPed LT SHT tests, 5 of the 7 samples lasted longer than the prediction. It is important here to stress the limitations of this approach. Mode-I crack growth is assumed even though both Mode-I (normal to the stress direction) and crystallographic crack growth along high stress planes were observed in the tested specimens. It is also unlikely that a fa tigue crack forms during the first cycle of each h to the data confirms that, in addition to the size, the type of initiating feature may impact fatigue lifetime. Tests that ex hibited crack initiation at carbides or very small pores generally had fatigue lifetimes longer than the predicted minimum lifetimes. 6.5 Summary ll HCF data is presented in Figure 6-22 in log form for easier comparison over a wide range of cyclic lifetimes. At each material c ondition, baseline CMSX-4 outperformed the modified alloys in terms of av erage cyclic lifetime. The application of HIP processing to the HT SHT alloys, as represented by solid circles in the plot, significantly improved fatigue lifetimes in all alloys except for the C modification. HIP processing also changed the active initiation sites from pores to carbides in the Ccontaining alloys. This served to isolate the effect of carbide morphology on fatigue lifetime for material in the HIPed, HT SHT co ndition. Crack initiation at script carbide HCF test. Even with these shortcomings, the exercise of applying this approac A 165

PAGE 166

166 networks in the C modification corresponded to s horter lifetimes than crack formation at groups of small, rounded carbides (C+B modifica tion) or at small, blocky carbides (C+N modification). The HIP processing applied to the HIPed, LT SHT material did not significantly improve fatigue lifetimes because porosity was not reduced enough to change the mechanism of fatigue crack initiati on. Fatigue lifetimes were impacted by the type of crack initiating feat ure (pore, script carbide network etc.) in addition to the size of the crack initiation site.

PAGE 167

0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 1,000,000# of cycles to failureBaselineCC+BC+NCMSX-4 Variant 20 Hz Figure 6-1. HCF lifetime results for allo ys in the Un-HIPed, HT SHT condition. 167

PAGE 168

Figure 6-2. SEM micrographs of crack initiation sites on fracture surfaces from HCF tests specimens in the Un-HIPed condition. A) Baseline, B) C modification, C) C+B modification, D) C+N modification. All initiations for Un-HIPed specimens occurred at pores. Figure 6-3. SEM micrographs of features obs erved on fracture surfaces of Un-HIPed HCF specimens. A) large crack growth plane perpendicular to the stress axis and centered at crack initiation site (r ed arrow indicates initiation site and black arrows indicate general crack gr owth directions), B) crack propagation outward from crack initiation and growth region. 168

PAGE 169

-HIPed, C modification HCF specim en. The ix H. Figure 6-4. Example c atalog entry for Un complete catalog can be found in Append Figure 6-5 IPed specimens tested in HCF. A) Baseline, B) C modification (BSE), C) C+B modification, D) C+N modification. The applied stress direction is vertical. SEM micrographs of longitudinal sections of Un -H 169

PAGE 170

igure 6-6. SEM micrograph of zigzagged cr a ck path indicating cracking along {111} octahedral slip planes in C+B modifi ed Un-HIPed HCF spec imen. The applied stress direction is vertical. F 0 1,000,000 2,000,000 3,000,000 4,000,000 5,000,000to f6,0 00,000 7,000,000 8,000,000 9,000,000 10,000,000# of cycles ailureBaselineCC+BC+NCMSX-4 Variant Figure 6-7. HCF lifetime results for allo ys in the HIPed, HT SHT condition. 170

PAGE 171

Figure 6-9. SEM micrograph of crystallographic cracking observed on the fracture surface of a C+B modification HCF samp le in the HIPed, HT SHT condition. The red area indicates the cluster of sma ll carbides at the initiation site and the black arrows indicate crack growth directions. Figure 6-8. SEM micrographs of tests specimens in the HIPed, HT crack initiation sites on fracture surfaces from HCF dition. A) Baseline, B) C SHT con modification, C) C+B modi fication, D) C+N modifica tion. Crack initiating features are identifi ed by red circles. 171

PAGE 172

Figure 6-10. SEM micrographs of longitudinal sections of HIPed, HT SHT specimens tested in HCF. A) Baseline, B) C modifi cation (BSE), C) C+B modification, D) C+N modification. The applied st ress direction is vertical. Figure 6-11. SEM micrographs of features observed in deep etched longitudinal sections of HIPed, HT SHT HCF specim ens. A) cracked carbide plate in the C modification, B) intact small, rounded ca rbides in the C+B modification. Stress direction is vertical. 172

PAGE 173

173 Figure 6-12. SEM micrographs of shortly spaced features observed on carbides in deep etched longitudinal sections of HI Ped, HT SHT HCF specimens. A) C modification, B) C+N modification. Figure 6-13. SEM micrographs of surface oxi de observed on carbides in longitudinal sections of HIPed, HT SHT HCF specim ens. A) Al and Cr rich oxide layer and Al depleted phase layer, B) cracks forming at the surface and extending into phase layer. The applied stress direction is vertical.

PAGE 174

Figure 6-14. TEM micrographs of carbides and dislocations in foils from HIPed, HT SHT HCF specimens. A) high concentration of dislocations near carbides, B) carbide cracking due to local strain. Figure 6-15. TEM micrograph of foil prepared from a base line HIPed, HT SHT HCF specimen. Limited dislocation structures can be seen in the channels. 174

PAGE 175

0 50,000 100,000 150,000# of cy200,000 250,000 300,000 350,000 400,000 450,000 500,000cles to failureBaselineCC+BC+NCMSX-4 Variant Figure 6-16. HCF lifetime re sults for alloys in the HIPed, LT SHT condition. 175

PAGE 176

176 Figure 6-17. SEM micrographs of crack initiati on sites on fracture surfaces from HCF tests specimens in the HIPed, LT SHT condition. A) Baseline, B) C modification (BSE), C) C+B modifica tion, D) C+N modification (BSE).

PAGE 177

177 Figure 6-18. SEM micrographs of longitudinal sections of HIPed, LT SHT specimens tested in HCF. A) Baseline, B) C modifi cation (BSE), C) C+B modification, D) C+N modification. The applied st ress direction is vertical.

PAGE 178

1.0E+04 1.0E+07 1.0E+05ycles1.0 E+06 0 020000 00035rea of Initiatin m2)# of c to failure (log sca Figure 6-19. Crack initiati vs. fatigue for all HCens. The red dashed line represents a power curve trendline for the data. The yaxis is presented in the log scale. 5000 1500 10000 2500030 00040000Elliptical A g Feature (le)ng feature areas lifet imes F specim 178

PAGE 179

1.0E+04 1.0E+05 1.0E+06 1.0E+07 020004000600080001000012000# cycles to failure (log scale) C modification script carbide networks Elliptical Area of Initiating Feature (um2) Figure 6-20. Crack initiating feature areas vs. fatigue lifetimes for HIPed, HT SHT HCF specimens. The red dashed line represent s a power curve trendline for the data. The y-axis is presented in the log scale. 179

PAGE 180

1.0E+04 1.0E+05# of cycleso fare (g scale)1.0 E+06 1.0E+07 00.2 1.21.41.6 tilulo in the 0.40.60.81Distance from edge/Area of feature ( m-1) Figure 6-21. Ratios of distance from sample surface to ellipt ical area of crack initiating feature vs. fatigue lifetim es for all HCF specimens. The red dashed line represents a power curve trend line for the data. The y-ax is is presented log scale. 180

PAGE 181

Table 6-1. Predicted minimum cyclic lifetim es from fracture mechanics appr oach results that are lower than estimations are shown in red italics. (cycles) (cycles) and corresponding actual fatigue lifetimes fo r all HCF test specimens. Actual Material Condition Alloy Modification Estimated Minimum N Actual N Baseline 137625 520425 Baseline 181413 295457 C modification 587976 138712 C modification 100755 61825 Baseline 146663 327785 C modification 131037 78729 Un-HIPed HT SHT C+B modification 358278 674213 C+B modification 34887 110549 C+B modification 33013 52880 C+N modification 407058 233287 C+N modification 46094 75426 C+N modification 124713 56574 Baseline 858624 6993205 Baseline 543713 742208 Baseline 1726761 8797800 C modification 92861 166089 C modification 112008 176144 C modification 109539 151054 C+B modification 5021227 1685075 C+B modification 111854 872343 C+B modification 339802 1338727 C+N modification 1575937 4504625 C+N modification 1342884 5947002 C+N modification 1062095 6315833 HIPed HT SHT Baseline 341765 291575 Baseline 153999 378911 HI C modification 54289 91324 C modification 70091 110432 C+B modification 102664 160733 C+B modification 100936 168957 Ped LT SHT C+N modification 96160 82946 181

PAGE 182

182 1.0E+04 1.0E+05 1.0E+06 1.0E+07# of cycailure (log scale) -Un-HIPedHT SHT -HIPedHT SHT -HIPedLT SHT les to fBaselineCC+BC+NCMSX-4 VariantFigure 6-22. HCF lifetime result s for all tests. The y-axis is presented in the log scale.

PAGE 183

CHA PTER 7 DISCUSSION The purpose of this chapter is to expl ain the observed alloy behavior within the framework of current literature and previously proposed theories, thus developing an understanding of the effects of C and other minor additions on microstructure and high temperature mechanical properties of SX Ni-base superalloys. Carbides, particularly carbide morphologies, are identified as t he controlling factor for behavior, and the viability of reducing harmful effects of carbides through minor additions and heat treatment is assessed. 7.1 As-Cast Characteristics Several clear differences we re identified between the as-cast microstructures of baseline CMSX-4 and the modified alloys. A dditions of C are made to SX superalloys for the express purpose of improving castability, and their impact on solidification microstructures is well covered in the fiel d. Several inconsis tencies amongst reported results and explanations, however, have some what clouded the picture of C in SX superalloys. The following sections address these topics with discussion of results from this study coupled with an analytical l ook at theories in the literature. 7.1.1 Carbides and Reduction of Grain Defects of C reduce formation of grain defects dur ing SX superalloy solidification. Both Two explanations have been presented regar ding the manner by which additions mechanisms involve the reduction of thermosolutal convection currents that are responsible for defect formation. The first suggests that interdendritic MC carbides consume enough heavy carbide forming elements, such as Ta and W, to reduce chemical segregation in the liquid. The t heory follows that a le ss segregated liquid will 183

PAGE 184

be less likely to experience densit y invers ions of heavy elements. The density inversions could generate convective cu rrents and lead to grain defect nucleation events [13,15]. The second explanation proposes that the presence of MC carbides in the in 7]. lly t o s that n. terdendritic regions may interfere with convective fluid fl ow in these areas [14,2 This mechanism requires carbide phase forma tion to sizes large enough to physica block fluid flow at the time when grai n defect formation would otherwise occur. The question regarding which of the tw o proposed mechanisms contributes mos to the reduction of grain defects is an impor tant consideration for alloy design. Later generation SX superalloys are more prone to grain defect formation because of their high refractory content. Verifying that ca rbide formation reduces segregation enough t prevent grain defects could give alloy devel opers increased flexibility. Refractory contents could be maintained as long as t he proper level of C was present to form carbides that tied up heavy elements. If it is only the physical presence of carbide leads to reduction of grain defects, ideal C levels could be determined based on the proper amount to form carbi des of the proper size and sh ape to interrupt convective fluid flow. Evidence from the current investigati on and other recent work suggests that reduction of segregation in interdendritic r egions is not likely the primary cause of decreased grain defect formation. No signifi cant changes in segregation were found to result from the minor addition of C to 2nd and 3rd generation SX superalloys [14,27]. At C levels generally used in SX alloys, the amount of elements such as Ta and W consumed by carbides is not great enough to cause noteworthy changes in segregatio MC carbides do not seem to significantly impact surrounding segregation of carbide 184

PAGE 185

formi bides both Re at 2.9 wt. %) are contributors to any segre alloys varies with changes to ca ere modification and the C+B m odification were free of grain defects. Casti betw ng elements, as was dem onstrated by the EDS line scans (Figure 5-26). Concentrations of Ta, the richest element in MC carbides, were not lower near car as compared to regions of / matrix far from the carbides. It should also be noted that carbide formation does not tie up a significant amount of W or any Re. In CMSX-4, of these elements (W at 6. 4 wt. % and gation-driven instabilities that occur. Prevention of de fects through reduced segregation may be a more realistic possibili ty in alloys in which Ta is the only significant refractory element that causes convective instabilities. The effectiveness of C in reducing grain defects in SX rbide morphology, offering evidence of the fluid blocking mechanism. Casting defect data from the alloy bars used in this st udy is presented in Table 7-1 [41]. Th were 20 bars cast for each alloy modificati on, and a bar was considered defect free only if it did not contain visi ble casting grain defects after macroetching and visual examination. While 15 of t he 20 bars for the baseline were defect free, 17 of the 20 bars in both the C ng of C+N modification bar s, however, resulted in only 12 of 20 defect free bars. Fewer grain defects were observed in the modifications containing script MC carbide networks than in the C+N modification featur ing smaller, blocky MC carbides. The carbide networks of large, connected plates and closely spaced rods are likely more efficient at controlling fluid flow in the in terdendritic regions than the blocky, faceted carbides. The script networks exhibit a tree-lik e structure, with trunks or cores forming een primary alloy dendrites and branc hes of carbide rods extending between secondary dendrite arms. The network cores are more effective at disrupting fluid flow 185

PAGE 186

than the more dispersed, blocky carbides. The increase in grain defects in the C+N modification may indicate t hat blocky carbides dispersed in the interdendritic regions may facilitate nucleatio n of grain defects. The lack of evidence supporting reduction of segregation due to MC carbide formation, coupl ed with fewer observed grain defects in alloys with large script carbide networks, indicates the reduction of grain defects attributed to C additions occurs due to physica l blocking of fluid flow by carbides 7.1.2 Minor Additions and Carbide Morphology Minor additions, often less than 100 ppm, of certain elements have been found to impact as-cast MC carbide morphologies. In the current study, the addition of 6 8 ppm B did n g the ence The blocky carbides obser ved in this study, however, were still confined to the interdendr owth without the influe ride ally decomposed blocky carbides in the C+N modifi cation did indicate that the cores were richer in Ti than carbides in the other alloys. This may point to formation of carbides on ot significantly change the script carb ide morphology observed when only C was added. The addition of 23 ppm N, however, did change the morphology from script networks to smaller, blocky carbides with face ted sides (Figure 4-8C). Work by Huan and others attributed the formation of blocky ca rbides in N-containing superalloys to heterogeneous nucleation of carbides on existing TiN particles in the alloy melt [44]. It was reported the presence of these particles at positions high in the mushy zone allowed for carbides to form into faceted, equilibrium structures wit hout the interfer of dendrite arms itic regions and showed no signs of gr nce of dendrite arms. In addition, DT A results (Table 4-1) indicated that MC carbides did not form higher in the mushy z one in the N-containing alloy. The blocky morphology is not likely caused by early nuc leation of carbides on preexisting nit particles high in the mushy zone. EDS anal ysis of thermally exposed and parti 186

PAGE 187

TiN particles, but not at a higher temper ature than standard carbide formation. The partic y rbide composition. It should be noted of d gy. les could provide a larger number of nu cleation sites that results in the smaller more dispersed carbides observ ed in the N-containing alloy. Lattice misfit between MC carbides and the surrounding matrix has also been identified as a cause of differences in carbide morphology. A report by Chen and colleagues examined several variat ions of a C-containing 2nd generation SX superalloy (developed by Rolls-Royce) with varying addi tions of B and Hf [37]. They estimated lattice parameters of carbides based on m easured carbide compositions. The main finding was that blockier carbides had larger lattice parameters, and therefore greater lattice misfits with the / matrix. Carbides with larger lattice misfits are presumed to form blockier shapes to reduce the surface ar ea to volume ratio. Further work with these modified alloys confirmed through latti ce parameter calculations from TEM diffraction patterns that carbides with lar ger lattice parameters exhibited blockier morphologies. This correlation was even appar ent from various carbides within the same sample. As a general trend, carbides with higher Hf concentrations had larger lattice parameters and exhibited blocki er morphologies [48]. Detailed WDS measurements of carbides in the heat treated condition (Table 4-5) indicated higher concentrations of Hf in the blocky carbides found in the C+N modification. The block morphology, therefore, is likely due to differ ence in ca that the concentration of Hf was 0.1 wt. % in all alloys. The increase in Hf in the carbides was not due to an increase of Hf in t he alloy, but rather by a minor addition 23 ppm N. Perhaps the presence of N c aused changes in the carbide nucleation an growth processes that resulted in the increased Hf content and blockier morpholo 187

PAGE 188

In addition to C content and carbide formation temperature, the composition of M carbides has an impact on their morpholog y. Unique alloy systems have varying concentrations of MC carbide forming elem ent s and will preferentially form carb certain composition. This appears to det ermine the prevalent carbide morphology. 7.1.3 Carbon and Casting Porosity The effect of C additions on casting porosit y in SX su C ides of a peralloys is not entirely clear. There en attributed g the final stages of solidi the re seen. The increases in porosity have been attrib f oys have been reports published describing both decreases [46,50] and increases [14,27,42] in porosity due to the presence of C. As is the ca se with most C effects, the observed behavior has been tied to interdendritic carbides. The reduction of porosity observed by Li u [46] and Chen [50] has be to MC carbides offsetting volume shrinkag e that occurs durin fication. The carbides have larger lattice param eters than the surrounding and phases, and this is believed to alleviate mi croshrinkage and reduce pore formation. Chen has also reported a qualitative decrease in porosity as carbide volume fraction increases [37]. Decreased porosity would be a significant and beneficial effect of C additions, but it has not been consistently observed. Porosity has been found to increase with addition of C in SX superalloys with chemistr ies very similar to the alloys described above in which porosity decreases we uted to blockage, by MC carbides, of molten alloy flow during the last stages o solidification. This leads to pores that are associated with interdendritic carbides [51]. The current investigation provides furt her evidence of increased casting porosity due to C modifications. As was shown in Ta ble 4-2, all of the modified CMSX-4 all exhibited higher as-cast porosity volume fr actions than the baseline. The carbide 188

PAGE 189

blocking mechanism was confirmed by the obs ervation of pores i n direct contact with carbides (Figure 4-5). The por es identified as crack initiation sites in all Un-HIPed HC samples of modified alloys were also associated with carbides; as were the pores at th center of the square f eatures on creep fracture surfaces. This suggests that the most detrimental pores are those that are near ca rbides. An assessment of as-cast poro in a C-containing alloy is incomplete without careful observation of regions with carbides. Examination of interdendritic regi ons away from carbides may give the false impression that C additions reduc e pore formation. The reports of decreased porosit C-containing superalloys may not fully account for pores adjacent to carbides While the large lattice parameters of inte rdendritic carbides may alleviate some microshrinkage during solid ification, the blocking of molten alloy by carbides i dominant effect of C additions on ca F e sity y in s the more sting porosity in SX superalloys. This results in is negative side effec hat could ee increased porosity due to the formation of pores near carbides. Th t of including C in SX superalloy chemistr y is an important consideration t warrant changes in alloy processing for modifi ed alloys. For example, the addition of C may require incorporating a HIP cycle into the heat treatment schedule to control porosity levels. 7.2 Heat Treatment and Exposure The evolution of microstructures duri ng heat treatment and thermal exposure revealed further differences in alloy behav ior between baseline CMSX-4 and the thr modified alloys. The following sections relate observed behavior to established mechanisms of microstructural evolution in order to explain these differences. 189

PAGE 190

7.2.1 Effec t of Carbon and Minor Additions on Heat Treatability The reduction of alloy solidus temperat ure is a primary concern when adding grain boundary elements such as C, B, N, Hf, and Zr to SX superalloys. Even additions of these elements can significantly depress the melting point and increase the small risk o n gues found e rd to the he represents a significant step in the heat treatment of C-modi fied SX superalloys. The f incipient melting during heat treatm ent. Homogenizing tem peratures are ofte lowered to avoid the possibility of melting, and this can result in incomplete solutioning of the phase. The reduction of solidus tem peratures removes flexibility from heat treatment schedules by narrowing the heat tr eatment window. Li u and collea that additions of 0.05 wt. % and 0.1 wt. % C to a 1st generation SX superalloy lowered the incipient melting point to a level t hat prevented complete solutioning of the phas [46]. The inability to fully homogenize these modified alloys led to reduced creep strength due to non-uniform precipitates [81]. Difficu lties in homogenizing without melting have caused some alloy developers to remove the SHT step entirely from heat treatment schedules for SX alloys containing grain boundary elem ents. The standa heat treatment for CMSX-486, a commercial SX superalloy with 0.07 wt. % C, 0.017 wt. % B, 1.2 wt. % Hf, and 0.005 wt. % Zr, invo lves applying a two-step age heat treatment alloy in the as-cast condition [12,29]. While this may be a relatively simple and low cost heat treatment, the ability to homogenize before forming the strengthening precipitates would likely lead to improved properties. Using modified variations of the standar d commercial heat treatment for CMSX-4, the modified CMSX-4 alloys were successfu lly solutioned and aged to produce t desired / microstructures for optimum mechanical performance. Although these alloys contained fewer grain boundary elem ental additions than CMSX-486, this 190

PAGE 191

heat treatments (Table 3-3) involved very slow ramp rates (1 oC/sec), multiple steps, and precise temperature control to slowly homogenize the alloy. The slow ramps reduced segregation of regions rich in low melting point elements and thereby minimized the risk of incipient melting. Ev en the B-containing alloy, which had a s temperature of 1327 olidus maximum temperature of 1325 oC. f large SX gas turbine components are more difficult than reful t are ed d ts, HT SHT and LT SHT, were desig ned to study the effect of heat treatment temp e oC, did not experience any observ ed incipient melting during the HT SHT, which had a Although heat treatments o for small test bars, this exercise dem onstrated that the implem entation of ca heat treatment design and tight temperature control can produce traditional superalloy microstructures in SX alloys modifi ed with C and other minor additions. 7.2.2 SHT Temperature and Stability of As-Cast Carbides As described above, the SHTs used in this study satisfactorily homogenized the alloys. Another goal of this study was to determine if the temperatures of SHT could effectively breakdown carbides from large, script networks to smaller features tha presumed less detrimental to mechanical properti es. This attempt was in part motivat by the work of He and others with a polycrystalline superalloy that exhibited increase carbide decomposition as the SHT tem perature was increased [61]. Two heat treatmen erature on carbide morphology. As described in section 4.5. 3, neither the LT SHT nor the HT SHT resulted in significant carbide breakdown. Both heat treatments caused slight changes on carbid surfaces that indicated the early stages of decomposition, but no changes in carbide morphology were observed. Estimates of re lative C diffusivities indicated that the maximum temperature of the HT SHT (1325 oC) resulted in a diffusivity that was only 191

PAGE 192

1.1 times greater than the diffusivity at t he maximum temperature of the LT S HT (1310 re heat e formation of unwa out a ost scrip t networks broke down into t y y oC). Increasing the temperatur e, and hence diffusivity, beyond 1325 oC would likely result in incipient melting in the alloy regar dless of ramp rate a nd the number of steps taken to reach maximum temperature. Extended hold times at maximum temperatu would be the only manner to promote significant carbide breakdown, but the treatments would t hen become much more expensive a nd impractical from a production standpoint. Longer times at elevated temperat ures may also promote th nted TCP phases. It was determined that carbide morphologies in the studied alloys could not be transformed during the SHT process. Observed carbide decomposition during SHT of pol ycrystalline alloys can be attributed to differences in carbide stability arising from variati ons in MC carbide compositions between polycrystalline and SX superalloys. 7.2.3 Carbide Morphology Change in Boron-Containing Alloy During the two-step aging heat treatm ent, some script carbide networks transformed into groups of smaller, rounder carbide particles (Fi gure 4-21A) with significant change in carbide composition. This morphology change occurred m frequently in the C+B modification, where th e majority of hese smaller carbide features. Most of the carbide networks in the C modification retained their morphology after full heat treatment. The morphology change from needle and plate-like features to more spherical shapes indicates a transformation designed to minimize surface energy. While a sphere has a larger relative strain energ than a plate or a needle [122], it has a lowe r surface energy. Surface energy has a greater impact on morphology for coherent or semi-coherent phases while strain energ dominates for incoherent phases. Work with MC carbides in SX superalloy systems has 192

PAGE 193

confirmed them to be semi-coherent wit h an orientation relationship with the / matrix [14,48]. The semi-coherency of the MC carb ides increases the influence of surface energy and causes the morphology changes observed. The question remains, however, why this transformati on is observed predominantly in the B-containing alloy. As discuss ed in section 7.1.2, lattice misfit between M C carbides and the / matri to ttice s f ly cause g B. o nge. Both W and Mo have been ident rbides x can affect carbide morphologies. Carb ides with larger misfits can be expected have morphologies which minimize carbide matrix interfacial area. Estimated la parameter values were calculated from WDS carbide compositions for the C modification and the C+B modification to dete rmine if lattice misfit acted as a driving force for morphology change in the B-containing alloy. No significant difference was found between the carbides in both alloys, as MC carbide lattice parameters were estimated at approximately 4.34 for each case. It should be noted that the estimate used an atomic radius of 0.77 for C because the specific val ue for the atomic radius o C in a metallic bond is not readily availabl e in the literature [48]. These estimates indicate that increased lattice misfit betw een carbide and matrix was not a like of increased carbide morphology transformation in the alloy modification containin Although estimated lattice parameters were very similar for carbides in the tw alloy modifications in question, differences in concentrations of particular elements may explain the varying degrees of morphol ogy cha ified to weaken binding forces in MC ca rbides, increasing the likelihood of carbide transformation at elevated temperatures [132]. A report by Qin and colleagues correlated thermal stability of Ti -rich MC carbides with W/Ti at omic ratios in the ca [58]. Carbides with larger W/ Ti ratios exhibited less stabili ty. Calculations of W/Ta 193

PAGE 194

ratios (Ta is the primary MC carbi de forme r in the studied CMSX-4 alloys) from average WDS compositions of carbides revealed a higher ratio in the B-containing alloy (0.14) than in the B-free alloy (0.11). The increas ed W content may expl ain the transforma from script carbide networ ks to spherical carbides dur ing heat treatment. The Ta content is high enough, however, to maintain the MC-type composition even after the morphology change. tion 7.2.4 s ler precipitates shrink and terminate [63]. Significant coarsening occ e majority of the ng (Table 4-6) than base by Coarsening of Phase Coarsening of phase occurs during high te mperature exposure and can signific antly reduce strength as precipitates grow larger and lose coherency with the matrix. The primary driving force for coarsening is reduction of / interfacial area a larger precipitates grow and smal urred in all four alloys during ex posure (Figure 4-25), but th precipitates remained cuboidal. A small amount of directional coarsening was observed near carbides. Directional coar sening in SX superalloys during thermal exposure is an indication of local chemic al segregation [133]. Directional coarsening near carbides, along with TCP phase forma tion at decomposing carbides, indicates compositional differences in these regions as compared to areas free of carbide phases. All alloy modifications exhibited greater rates of coarseni line CMSX-4. The C+B modification had the highest coarsening rate (1.77 x 10-7 m3/sec) and the largest size after exposure (0.91 m). The increased precipitate coarsening in the modified alloys may be ex plained by the presence of MC carbides. Carbides tie up W, Ti, and Ta, which can al ter the diffusional coarsening process changing / lattice misfit and interfacial energy. In particular, the MC carbides in the 194

PAGE 195

modi ermal r temperature exposure varied significantly amon e, n here 4-7), and they tended to form in regions enriched in t hese elements. In the baseline, this was fied CMSX-4 alloys occupied significant amounts of Ta. Carbides have been shown in some alloy systems to consume enough Ta to measurably reduce stress rupture lifetimes due to the loss of solid solu tion strengthening [134]. A relatively slow diffusing element compared to other elemen ts in superalloy systems, Ta has been shown to partition to the phase and retard coarsening [135]. Reduced Ta concentrations in the / matrix due to carbide formation may explain why the modified alloys experienced more precipitat e coarsening than baseline CMSX-4. While minor additions slightly increased coarsening rates during long-term th exposure, the precipitates remained cuboidal. No significant breakdown or widespread, abnormal c oarsening of the / microstructure occu rred. Maintaining precipitate morphology while avoi ding spherical or irregular has proven critical fo retaining creep strength at high temperatures [63]. The mi nor increase in coarsening rate induced by the minor additions does not act as a significant deterrent to the implementation of C-containi ng chemistries in SX superalloys. It should be noted, however, that changes to coarsening for a given alloy will be determined by the formation of carbides and their impact on the chemistry of the and phases. 7.2.5 Alloy Modifications and TCP Phase Formation Formation of TCP phases during high gst the CMSX-4 variants. The C+N m odification formed the most TCP phas while the baseline and C modification both formed similar amounts of TCP phase. The most noteworthy result was the lack of any visible TCP phase in the C+B modificatio after exposure. The first step in describing these unique behaviors is to identify w the TCP phases formed. TCP phases were ri ch in Cr, Co, Ni, Re, and W (Table 195

PAGE 196

at dendrite cores with residual segregation of heavy elem ents such as Re and C modification and the C+N modification, the TCP phase formation occurred near decomposing MC carbides. The decomposition proces W. In the s resulted in local enrichment of W, w areas te matrix C and for TCP s C dification (Figure 7-1A), the carbide network remains after heat treatment and then partially decomposes duri ng exposure, releasing some C and W into hich is a key element for TCP phase fo rmation. The TCP phases formed in with the ideal quantities of TCP phase forming element s. The lack of TCP phase formation in the Bcontaining alloy can be attributed to carbide decomposition and secondary carbi de formation. The C+B modification exhibited the most carbide decomposition among the alloys. In some cases, comple MC carbide dissolution occurred. This provided t he greatest amount of C in the for Cr-rich M23C6 carbide formation, and more sec ondary carbides were in fact observed in the C+B modification than any ot her alloy. Therefore, while partial M carbide decomposition in the C modificati on and the C+N modification released W led to TCP phase formation near carbides, more advanced MC carbide decomposition in the C+B modification led to M23C6 carbide formation and less Cr available phase formation. Several reports on long te rm stability of superalloys have tied the suppression of TCP to secondary carbide formation [18,19,37]. These carbide consume elements, particularly Cr and W, which are known TCP phase formers. Preferential formation of secondary carbides as opposed to detrimental TCP phases is therefore achieved with a minor addition of B. Simplified schematics showing the mechanisms for TCP phase formation in the modification and TCP phase suppression in the C+B modification are shown in Figure 7-1. In the C mo 196

PAGE 197

the s m TCP carbide of es with Cr in the matrix. The rea. king C modifications to SX superalloys. Creep results are discu s acking dition. urrounding matrix. The W combines with Cr Co, and Re in the matrix to for phase near the original carbide. In the C+B modification (Figure 71B), the MC network breaks down into small, spherical MC carbides during heat treatment. The spherical MC carbides completely dissolve during exposure, releasing a large amount C into the matrix. This C forms a group of M23C6 carbid local depletion of Cr in the matrix prevents TCP phase formation in the a The suppression of TCP phase formation through MC carbide decomposition and M23C6 secondary carbide formation is an intriguing aspect of C additions to SX superalloys. It represents another possible b enefit of C in SX superalloys. This also creates an opportunity for alloy developers to optimize minor elemental additions to improve long term alloy stability. 7.3 High Temperature Mechan ical Properties Creep Adequate creep rupture strengt h is a necessity for any alloy considered for gas turbine components subjected to high temper atures and stresses. Any new alloy or alloy modification that does not improve upon or at least maintain the creep capabilities of current alloys may not be viable for applicat ion. The effect on creep strength is a primary concern when ma ssed here within the cont ext of observed behavior in the current literature. 7.3.1 Competing Roles of Carbides in Creep Deformation Most studies on the effects of C on creep properties focus on the roles of carbide in the deformation process. Both primar y and secondary carbides have been credited with improving creep properties by pinning dislocations [8,18,37], but carbide cr has been shown to negatively impact creep st rength [37,81]. The net impact is generally determined by which of the competi ng effects dominates at a given con 197

PAGE 198

There are reports in the lit erature that are somewhat contradictory to one another regarding the creep conditions at which carbi des improv e properties and those at which carbides hurt properties. Kong reported that additions of C, B, and Hf improved creep behavior of a 2 C/430 d and conditions compar ed to the K ong report. Carbides were bene Pa) and ly tter. ugh evidence to confirm to nd generation SX superalloy at a l ow temperature condition (850 oMPa) and negatively effected behavior at two high temperature conditions (950 oC/210 MPa and 1050 oC/165 MPa). Improvements were attr ibuted to a reduction of porosity that was beneficial to the tertiary creep behavior, and the reducti ons in lifetime were tie to higher creep rates during the primary and secondary stages due to irregular precipitates and carbide cracking [38]. Liu and colleagues correlated the benefits detriments of carbides to the same microstr uctural features, but they observed these effects at opposite creep ficial to creep lifetimes at the high tem perature condition (1038 oC/172 M detrimental at the low temperature condition (871 oC/552 MPa) [81]. In both reports, creep strain curves showing lifetime improve ments due to C additions indicate on minor differences in behavior and rupture lifetime t hat could be attributed to sca While reduction of porosity and pinning of dislocations by carbides could improve creep behavior, the limited data presented may not provide eno beneficial effects of C on creep behavior. In the present study, carbides reduc ed creep rupture lifet imes compared baseline CMSX-4 at all three creep conditions. Any potential benefits of carbides were outweighed by the negative effects, wh ich included increased porosity and carbide cracking. Both of these effects lead to greater damage in the alloy and would be expected to have the greatest impact on tertiary creep behavior (rapid damage 198

PAGE 199

accumulation and crack growth) and st eady-state creep behavior (balance of strengthening and damage accumu lation). Pores and carbides both facilitate creep cavitation and the advancement of creep deformation. Interestingly, the largest reduction in creep lifetimes of the modified alloys occurred at 750 oC/800 MPa (Figure 5-19) and was ca used by increased primary creep strains. Primary creep behavior is controlled by deformation of precipitates and is generally not associated with da mage from cracking at pores or carbides, which occurs during steady state and tertiary creep. T he B-containing alloy exhibited particularly large primary creep strains and shor t rupture lifetimes at the 750 re near clusters of spherical MC carbides (Figu differ to y once again revolve around Ta. In addition to retarding coarsening as discussed above, Ta has been found to suppress the oC condition. Observations of highly deformed microstructu re 5-24C) prompted a study of possibl e local segregation and lattice misfit ences near carbides. As presented in section 5.5.3, no clear differences in composition were found near carbides. In creased primary creep rates due to irregular precipitates were unlikely because heat treatments produced regular / microstructures in the m odified alloys. Regular precipitate morphology has been identified as a key aspect of maximizing creep performa nce [62]. The key understanding primary creep in the modified alloys ma the complicated set of disloc ation structure formations th at lead to primary creep [76,136]. Ta content in the / matrix is reduced due to the presence of Ta rich MC carbides, which may cause increased susceptibil ity to large primary creep strains in modified alloys. 199

PAGE 200

It is important to note that gas tu rbine engines are generally designed and operated to avoid conditions that cause primary creep in crit ical components. Therefore, increased primary creep strains associated with C-modified superalloys not be as significant in the assessment of alloy performance as more critical aspects such as steady state creep and fatigue behavior. 7.3.2 Improvements in Creep Pe rformance Due to HIP Processing Although HIP processing of superalloys is performed primarily to improve fatigu behavior, it can also benefit creep performanc e. The work of Chang involving HIP processed CMSX-4 found that HIPed material experienced a 185% increase in rupture lifetime at 950 may e e s. Reed and others tested CMSX-4 at a ver sing t and can be explained by defor -oC/355 MPa as compared to the Un-HIPed condition [105]. Results from the current study also showed increases in rupture life, ranging from 15% to 38%, du to HIPing in all alloys tested at 850 oC/550 MPa. Lifetime improvements from HIPing however, are not observed at all creep condi tion y high temperature condition (1150 oC/100 MPa), and found that HIP proces did not yield any lifetime im provement. Rapid formation of TCP phases and subsequen creep cavitation at these phases resulted in si milar rupture lifetimes for both HIPed Un-HIPed CMSX-4 [86]. The varying effects of HIP processing at different conditions examining the creep cavitation process. Vacancy condensation during creep mation can result in the fo rmation of small voids that c an grow into cavities [86]. Nucleation of these cavities occurs during the steady-state (secondary) stage of creep and precedes cavity growth and the onset of tertiary creep. The last stages of deformation before failure are controlled by cavity growth [137]. Yu identified cavitation controlled damage as the dominant mechani sm inducing creep fracture in a 2nd 200

PAGE 201

generation SX superalloy creep tested at 980 oC [138]. In Un-HIPed alloys, pore as pre-existing cavities that readily grow and form cracks as they incorporate vacancie and small voids during creep. This process l eads to more rapid failure than in HIPed alloys, where cavities must nucleate fr om coalescing vacancies and voids before growing and leading to fracture. The clear difference in size of cavities at the center o square crack features for HIPed and Un-HIP ed specimens was shown in Figu The growth of the pore in the Un-HIPed spec imen causes an earlier onset of crackin and thus failure than the more slowly forming creep cavity in the HIPed specimen. Th reduction of pores through HIPing does not improve creep behavior at very high temperatures because cavities can rapidl y form at TCP phases, aided by increased s act s f re 5-25. g e vaca oys such as tion g creep sing. A thorough review of re-crystallization behavior in superalloys indicated that the ncy diffusion rates. HIP processing can clearly improve the intermediate temperature creep performance of SX superalloys. Despite this fa ct, C-containing SX superall CMSX-486 do not undergo HIP processing as part of the standard heat treatment schedule, which employs only a two-step agi ng treatment [29]. The work here demonstrates that a successful HIP cycle can be incorporated into the processing of Cmodified SX superalloys to improve creep performance. 7.3.3 Re-Crystallization Near Carbides? Observation of an isolated grain in t he longitudinal section of a HIPed C modification interrupted creep specimen (Figur e 5-18B) warranted further investiga to determine if it was evidence of dynamic re-crystallization near carbides durin deformation. Examination of thread sections of HIPed creep specimens revealed no evidence of grains, indicating t hat re-crystallization did not o ccur during HIP proces 201

PAGE 202

observed g rain was not likely a result of dy namic re-crystallizati on. The most common case of re-crystallization in superalloys o ccurs when the alloy is plastically deformed before undergoing SHT [139]. Ex treme care is taken in handling SX components before SHT, and cleaning processes such as grit blasting are avoided [ 140]. Generally, th temperature must be above the e solvus for re-crystallization to occur. The creep test temp be t formed rature Mechanical Prope rties High Cycle Fatigue (HCF) when making C modi fatigue ce following ine e cleav erature (850 oC) is not likely high enough to provide enough thermal energy for re crystallization to occur. Furthermore, dy namic re-crystallization has been found to less likely at monotonic creep loading conditions than at cyclic fatigue loading conditions [141]. The observed feature can only be des cribed as an anomalous grain tha during initial solidificat ion or during the SHT. 7.4 High Tempe Negative effects on fatigue properties are a primary concern fications to SX superalloys. The formati on of carbides introduces possible crack initiation sites. The HCF studies in the current study indicate d that the presen of carbides did reduce fatigue lifetimes in the C-modified CMSX-4 al loys. The sections serve to further examine the causes of these lif etime reductions. 7.4.1 Role of C in Ch anging Fracture Appearance Clear differences were apparent between the HCF fracture surfaces of basel and modified samples. Schematics representing side profile views of fracture surfaces for typical baseline and C-modified samples ar e shown in Figure 7-2. Baseline fractur surfaces, in both the HIPed and Un-HIPed conditions, consisted of smooth, highly reflective planes approximately 45o from the stress axis. These planes contained age characteristics such as river patterns that could be traced back to the crack origin. Crack initiation occurred at pores and grew by stage-I crack propagation along 202

PAGE 203

{111} crystallographic planes. The overload failure involved cleavage that continued along {111} slip planes, resulting in a uni form planar appearance across the entire fracture surface. The C-containing fr acture surfaces were much rougher and had multiple steps, some smooth crystallographic facets, and fractured carbides. In the UnHIPed specimens, the initial crack growth regi ons were circular and perpendicular to the nitial crack growth occurred by stage racks azaki the eferential crack initiation sites and crack growth stress axis, indicating stage-II crack propagation. I -I crack propagation for HIPed specimens in which crack initiation occurred at carbides. The roughness and variety of planes seen in the overload fracture regions can be attributed to the presence of carbides. Carbides disrupt single planar slip c and cause crack deviation to other planes. This type of mixed plane fracture has also been observed on the fracture surfaces from TMF tests of SX s uperalloys. Ok observed a combination of {100} and {111} planes for TMF samples oriented in the [001] direction parallel to the stress axis [103]. The change in fracture surface appear ance in the modified alloys is due to interference of planar cleavage by carbides during overload fracture. Cleavage along crystallographic planes occurr ed across the majority of t he fracture surface in the baseline samples. This likely has a very minor impact on fatigue lifetime as compared to changes to crack initiation behav ior in the C-containing alloys. 7.4.2 Isolating Effects of Carb ide Morphology Through HIP Processing Improved fatigue lifetime is perhaps the most well known benefit gained from HIP processing of superalloys. A report by Harri s detailing the initial development of CMSX 4 noted a 50% increase in 750 oC HCF lifetime due to HIP pr ocessing to eliminate micropores [92]. In addition to significantly improving lifetimes, HIP processing of modified CMSX-4 alloys changed pr 203

PAGE 204

mech e h tion ity. re 00 e des in the modi anism s. In Un-HIPed specimens, crac ks initiated at pores associated with carbides and grew in the form of circular Mode-I crack planes. T he surfaces of these crack planes were highly reflective in li ght when observed with t he naked eye. Th polished appearance developed as opposite surfac es of the crack plane contacted eac other during each minimum stress half cycle afte r crack formation. This type of flat crack growth has been observed elsewher e in low R-ratio HCF testing of SX superalloys at high temperatures [142]. Specimens that underwent HIP processing featured cracks initiating at carbides and stage-I crack growth along crystallographic planes. It should be noted that the HIPed LT SHT specimens exhibited crack initia behavior characteristic of the Un-HIPed sample s due to incomplete removal of poros Fatigue cracks likely form after a fewer num ber of cycles when initiating at pores as compared to carbides. Pores are pre-exis ting discontinuities in the microstructu that act as circular cracks. Cracks can readily form at areas of high stress concentration at pore edges. Work by Lamm in volving HCF testing of PWA 1483 at 8oC reported that fatigue cracks pref erentially form at pores if the pores are similar in siz to carbides [93]. Applying a HIP cycle eliminat es or greatly reduces the size of casting pores, effectively changing the fatigue crack init iation sites from pores to carbi fied CMSX-4 alloys and im proving fatigue lifetimes. Observed crack initiation mechanisms at carbides included carbide cracking and de-cohesion between carbide and matrix, bot h of which occur only after localized deformation develops from repeated cyclic loading. These mechanisms were associated with particular carbide sizes and morphologies that had strong correlations to fatigue lifetime. Cracking of large scr ipt carbide networks in the C modification 204

PAGE 205

resulted in signific antly lower lifetimes than de-cohesion of smaller, blocky carbide the C+N modification. The relationship between size of crack initiating feature an fatigue lifetime for the HIPed HT SHT specim ens was shown in Figure 6-20. The script networks retained during heat treatment resu lted in the lowest fatigue lifetimes CMSX-4 modified only with C. The change in preferred fatigue crack initiation sites due to HIP process critical in isolating the effects of ca rbide morphology on HCF behavior. Although differences were observed in crack growth behavior for cracks initiating at pores (M I growth perpendicular to the stress axis) and carbides (stage-I growth along crystallographic planes), the lifetime differences are likel y due to delayed crack initiation in HIPed specimens. 7.4.3 Possible Effects of Minor Additions in the Atomic Form The majority of the discussion on the ro le of minor elemental additions in mechanical behavior has s in d for ing was oderevolved around carb ide phases. Possible effects of the adde creased B ous slip or made cross-slip more difficu lt in the alloy with hi gher B concentration. Although most of the minor additions to CMSX-4 are incorporated into carbides, interstitial C, B, and N atoms are present in solution. While no obvious differences in d elements in the atomic form (not as part of metal carbides) will now be addressed. A series of publications [ 97-99] by Xiao and co lleagues on wrought, polycrystalline IN 718 superalloys focused on va rying levels of B and their impacts on mechanical behavior. The presence of only 29 ppm B, as opposed to 12 ppm B, resulted in slower fatigue crack growth rate s and longer LCF lifetimes. The in content was credited with chang ing crack paths from smooth and straight to tortu and rough. This effect was attributed to clus ters of interstitial B atoms that retarded dislocation 205

PAGE 206

crack as tions of tested HCF specimens. The / structure directly adjacent to ca wo e n faces of carbides and the / matrix. As shown in Figure 6-12, ar to those identified by Zhao cation ased o growth behavior between the alloys were observed on HCF fracture surfaces or longitudinal sections, intersti tial additions may impact fatigue crack growth behavior to some extent. These minor effects on cra ck growth rate, however, would not have great of an influence on fatigue lif etime as the effects of carbides on crack initiation (the controlling factor for HCF behavior). 7.4.4 Local Plastic Deformation Near Carbides The localized deformation that developed in the vicinity of carbides during HCF loading was identified as an im portant contributing factor in crack initiation mechanisms at carbides. Evidence was first obser ved during SEM analysis of longitudinal and transverse cross sec rbides was deformed and appeared str etched to varying degrees. The deformation was greater in specimens that exhibited longer fa tigue lifetimes and a greater number of loading cycles. Figure 7-3 shows plastic deformation in between t carbides in C+N modification HCF sample that underwent over 6 million cycles befor failure. Deep etching of tested specimens re vealed further signs of active deformatio processes at the inter edges of carbides featured closely spaced, linear features simil as evidence of slip band impingement on carbides [70]. TEM analysis confirmed the increased levels of plastic deformation due to the presence of carbides. Dislo densities were significantly greater near carbi des (Figure 6-14) than away from carbides or in baseline CMSX-4 specimen s (Figure 6-15). It is wo rth noting that the incre plastic deformation in the carbide-containi ng specimen is apparent even though the tw tests had similar fatigue lifetimes (6,315, 833 cycles for the C+N modification and 6,993,205 cycles for the baseline). 206

PAGE 207

The literature contains multiple reports of plastic deformation at secondary pha se developing during fatigue loadi ng. In addition to carbi des in SX superalloys [39,95], oxide inclusions in polycrystalline superalloys [35,36] and steel [143] have been identified as sites of strain localization. Yi and fellow researchers correlated localize strain near carbides with crystallographic {111} cracks that initiated along shear b in these regions during ultrasonic (20 kHz) s d ands fatigue testing of PWA 1484 at 1000 oC. They ce -4 temp mation d in several ways. The most commonly described process involv le reported that t he unique stress states a nd locali zed cyclic deformation required for these types of cracks to form were comm on at low temperatur es, and the occurren of cracking along octahedral planes at 1000 oC was due to the ultrasonic loading frequency [39]. Similar behavior involving crystallographic cracking associated with localized deformation at carbides, however, was observed in m odified, HIPed CMSX specimens tested in HCF at a much lowe r frequency (15 Hz) and a relatively high erature (850 oC). An example of such an observa tion is presented in Figure 7-4. Also recall that crystallographic, stage-I cr acks were observed originating at primary crack initiation sites on HT SHT HIPed fatigue fracture surfaces of C-containing alloys (Figure 6-9). This type of cyclic strain localization and crystallographic crack for at high temperatures does not appear to be unique to ultrasonic frequencies. The cyclic development of localized, plas tically deformed areas near secondary phases has been explaine es the impingement of persistent slip bands at the particle, followed by shear localization and the increase of strain until cracking or de-cohesion between the partic and the surrounding matrix [35,36,95,143]. A similar de-cohesion and crack initiation process has even been reported at / eutectic in a 1st generation SX superalloy [144]. 207

PAGE 208

A study by Xie and colleagues used in-sit u SEM observation to capture the steps involv ed in fatigue crack initiation at Al2O3 particles in Rene, a polycrystalline superalloy disc alloy produced using powder metallurgy methods. At applied stresses below c DS sistent with the observ ations from the modified CMSX-4 fatigue spec alized a role, exper imental observations suggest that the accu the yield stress, dislocations built up at the alumina particles and locally increased the stress concentration. As ma terial damage increased, the particles cracked or separated from the surrounding microstructure and caused crack propagation into the matrix [ 36]. Shenoy and others recognized the role of hard cerami phases such as oxides and carbides in lim iting fatigue lifetime. They incorporated effects of these phases into a cyclic plasticity model for predicting LCF lifetime in a superalloy. Fatigue cracks that formed due to carbide cracking or de-cohesion events were modeled as micro-notches [35]. All of these described mechanisms of crack initiation are con imens. Figure 6-10D shows localized deformation at a blocky carbide and associated de-cohesion. Localized deformation has also been attri buted to increased elastic modulus in regions containing defects or secondary phases. Ravi Chandran pointed to loc modulus increases and associated increases in stress as the cause of fatigue crack formation at secondary phases in a number of alloys [100]. While local differences in elastic modulus may play mulation of localized deformation at interfaces between secondary phases (carbides in the current study) and the matrix is the key factor leading to fatigue crack initiation. 208

PAGE 209

Several stray grains, similar to the one observed after creep at 850 oC (Figure 18B), were identified in tested HCF spec imens. The grains h 5ad carbides at the boundaries, as shown in the Figure 7-5 image captured from a HT SHT HIPed Ncontaining sample. The possibility of dynamic re-crystallization is greater under cyclic loading conditions (HCF) as compared to a monotonic load (creep). The localized cyclic deformation that develops at carbides provides more potential sites for grain nucleation [141]. Even so, the grains observed are unlikel y the result of dynamic re-crystallization because the temperature of fatigue testing (850 oC) is probably too low for this type of process to occur. It remain s unclear why these stray grains were observed only after mechanical testing. Carbides act as stress concentrators that can lead to development of dislocation structures at the interface between carbides and matrix during cyclic HCF loading. In addition, carbides can act as barriers to slip band and dislocation movement, which leads to accumulation of pl astic deformation near carbides as more load cycles are applied to the material. The localized deformation serves as a predecessor to crack initiation at carbides. 7.4.5 Fatigue Mechanisms and Effects on Cyclic Lifetimes Observed crack initiation mechanisms had a significant impact on HCF lifetimes of CMSX-4 and the modified alloys tested in this study. Recent work has identified multiple fatigue lifetime distributions and S-N curve behaviors within the same alloy due to different active failure mechanisms, as was shown in Figure 2-9. The report concluded that alloys will have at least two competing fatigue failure modes, namely internal crack initiation versus surface cra ck initiation [100]. Fatigue results from the current investigation of C-modified CMSX4 alloys suggest more than two competing 209

PAGE 210

fatigue mechanisms that impact cyclic lifetim es. These mechanis ms include, (i) t large pores followed by significant Mode-I cr crack initiation a ack growth, and (ii) local plastic deformatio at carbides car or e and igure 7-ins repres SEM im f observe itiation sites and the associated, relative fatigue lifetimes. Explanations for the observed behavior will now be presented. The shortest lifetimes were associated with crack initiations at large casting pores in specimens that had not been HIPed or had been incompletely HIPed. Increased porosity in the C-containing al loys was attributed to the presence of carbides and led to shorter lifetimes than in basel ine CMSX-4. Casting pores re sult in the earliest crack initiation because they act as pre-existing circular or elliptical cracks that can cause propagation (normal to the stress axis) of cra cks from the edge of a pore into the matrix. modificati s at networks were triggered by the cracking of thin rod and plate-like carbide features. The small cross-sectional areas and brittl e nature of these f eatures make them susceptible to cracking at a relatively low nu mber of cycles. The cyclic development of small amounts of plastic deformation at carbide-matrix interfaces further contributes to cracking of the networks. Crack initiations at clusters of rounded carbide particles, observed in HIPed C+B The rounded particles are less likely to crack than the thin network features, and the build up of more localized deformation is requi red before crack initiation. The proximity n followed by bi de cracking de-cohesion b tween carbide / matrix. F 6 conta entative ages o d crack in Crack initiation at large carbide networks re sulted in fatigue lifetimes for HIPed, C on specimens that we re not much greater than Un-H IPed lifetimes. Initiation modification specimens, resulted in longer lif etimes than initiation at script carbides 210

PAGE 211

of the small carbides to one another leads to a greater amount of plastic deformation than if the particles were widely spaced apar t. Cracks initiate from cracking of the strained matrix or de-cohesio n between matrix and carbide. Crack initiation from d e-cohesion of discrete, blocky carbides from the matrix took the longest to occur among the carbide in itiation mechanisms. The HIPed C+N modification specimens exhibiting these types of initiations had the longest fatigue ng as the thin carbides in scrip far enough apart that the local deformation fields of each did not interact with one another. The levels of plastic deformation and interfacial strain necessary to cause decohesion and crack initiation at individual blocky carbides require a large number of fatigue cycles. The longest fatigue lifetimes and therefore the most delayed crack initiation, in these s al features remaining after HIP processing or microstructural discontinuities that developed during fatigue loading at 850 oC. The mechanism is lik ely similar to that for initiation at carbides, with local plastic def ormation accumulating cyclically at the small void until a crack forms. Testing at a variety of HCF conditions would be needed to determine the stresses at which specific fatigue crack initiation me chanisms operate and to to fully develop S-N r lifetimes of any of the C-m odified alloys. Blocky carbides are not a s prone to cracki t networks. In addition, most blocky carbides were spaced occurred in the HIPed baseline CMSX-4 material. The crack initiations leading to failure amples originated at very small por es or voids. These were either residu curves for the alloys studied. The work co mpleted, however, demon strates that mino 211

PAGE 212

212 alloying additions in CMSX-4 have a direct effect on carbide morphologies, which determine crack initiation mechanisms and fatigue lifetimes.

PAGE 213

Table 7-1. Number of def ect free bars cast from bas eline CMSX-4 and the 3 Ccontaining modification no observed grain defects after casting are considered defect free [41]. modification modification modification s studied. Bars that had Baseline C C+B C+N # of defect free bars 15/20 17/20 17/20 12/20 HT Exposure C W HT Exposure B C C C C MC carbide M23C6carbide TCP phaseACr Cr igure 7-1. Schematics of phase changes du ring heat treatment and long F term thermal exposure. A) C modification, TCP phas e forms near partially decomposed MC carbide, B) C+B modification, M23C6 carbides form preferentially from fully decomposed MC carbide and suppress formation of TCP phase. Figure 7-2. Schematics of ty pical HCF fracture surfaces, presented as side profile views. A) Baseline specimen, B) C-containing specimen. 213

PAGE 214

Figure 7-3. SEM micrograph of transverse se ction from C+N modification HCF test ion from C modification HCF test specimen with a fatigue lifetime of 6,315,833 cycles. Localized plastic deformation can be seen in between the two carbides. g Fi ure 7-4. SEM micrograph of longitudinal sect specimen. Localized plastic deformation near carbides and associated crystallographic cracking can be seen. Figure 7-5. SEM micrograph of longitudinal section from C+N modification HCF test specimen. A stray grain with carbi des at the boundaries is shown. 214

PAGE 215

50 m small void or pore 50 m small, blocky carbideDecreasing Fatigue Lifetime 100 mcluster of rounded carbide particles50 m casting pore 100 m script carbide network Figure 7-6. Characteristic HCF crack initiati on sites and associated relative lifetimes. 215

PAGE 216

CHA PTER 8 CLOSING REMARKS This investigation highlights the vari ous effects on alloy performance of C additions in SX superalloys. Minor element al modifications designed to improve castability should be made carefully with an eye towards carbide phase formation and its impact on mechanical properties. The pr esence of carbides changes SX superalloy behavior and should be taken into account in gas turbine engine design. 8.1 Research Conclusions containing variants have led to the following conclusions: Differences in as-cast carbide morphologies are likely due to variations in MC ppm B resulted in script carbide networks, while the addition of 0.05 wt. % C along cky carbides. While all MC carbides were enriched in Ta, those in the N-containing alloy had higher concentrations of Hf, which may result in larger lattice misfit between carbide and matrix. The presence of C increases casting por osity in modified CMSX-4 superalloys due to carbide blocking of molten alloy during t he final stages of solidification. This mechanism was confirmed by observation of pores associated with carbides. The effectiveness of C in reducing ca sting grain defects changes with carbide morphology. Interdendritic networks are more effective than dispersed, blocky carbides at physically blocking convective fluid flow to prevent defect formation. Although additions of C, B, and N reduce alloy solidus te mperatures by up to 16 oC, modified alloys can still be succe ssfully homogenized and heat treated by making minor adjustments to established superalloy heat treatment schedules. The desired / microstructures were achieved in both the baseline and modified alloys. Increased SHT temperature, by itself, does not result in full breakdown of carbide networks during homogenization. Minor changes to carbide surfaces were observed after both the LT SH T (maximum temperature of 1310 oC) and the HT SHT (maximum temperature of 1325 oC). The experiments and analys is comprising this study of CMSX-4 and three Ccarbide composition. The addition of 0.05 wt. % C and 0.05 wt. % C along with 68 with 23 ppm N produced smaller, blo 216

PAGE 217

The B-containing alloy exhibited rbide morphology without a significant change i Script carbide networks transformed to groups of small, rounded Ta rich MC carbides. The B is likely due to small variations in carbide co mposition that impact carbide interfacial ys caused by carbides consuming elements such as Ta, Ti, and W. o o o led to decreased creep ruptur e lifetimes. Although further testing and analys is is needed, the larger primary cr eep strains are likely due to local differences in composition and / misfit near carbides. HCF testing of Un-HIPed material indi cated that the modified alloys had lower fatigue lifetimes than the baseline. Decreas ed lifetimes were attributed to crack initiation at pores associated with carbides These initiations resulted in formation of large, circular Mode-I cracks pe rpendicular to the stress direction. HIP processing significantly improved fatigue performance in all CMSX-4 variants, resulting in average lifetime increases ranging from 77% to 4490%. In the Cmodified alloys, HIPing changed the preferre d fatigue crack initiation sites from pores to carbides and isolated the effect of carbide mor phology on crack initiation behavior. Crack initiation mechanisms at carbides involve the development of localized plastic deformation followed by carbide cr acking or carbide de-cohesion from the matrix. A greater amount of deformati on, and therefore a gr eater number of fatigue cycles, is required to form failure-c ausing cracks at small blocky carbides (C+N modification) and groups of small, ro und carbides (C+B mo dification) than at large script carbide networks (C modification). It is cl ear that both the size and type of crack initiating feature have an effect on fatigue behavior. a change in ca n carbide compositi on during aging HT. greater degree of carbide network breakdow n compared to the alloy without energy and stability. During high temperature exposure (1000 oC/1000 hrs), the C-modified allo exhibited phase coarsening rates that were 35 61% higher than baseline CMSX-4. The increased coarsening is attributed to differences in / lattice misfit Formation of TCP phase during high temper ature exposure was suppressed in the B-containing alloy due to the decomposition of MC carbides and preferential formation of secondary M23C6 carbides. Improved long term alloy stability is a promising potential benefit of C in SX superalloys. Additions of C to CMSX-4 reduced creep performance at all conditions (950 C/300 MPa, 850 C/550 MPa, 750 C/800 MPa) due to increased porosity and carbide cracking. HIP processi ng improved creep per formance at 850 oC/550 MPa, although baseline CMSX-4 still outperformed the modified alloys. Increased primary creep strains were observed in the C-modified alloys as compared to the baseline, and these large initial strains (up to approximately 10%) 217

PAGE 218

8.2 Future Work Conued study of ces in SX supe ys is importan the progression of the industry. As a result of this w ork several ure research phave beco apparentfurther undershow C and ot inor addition behavior.hese opportun It appears that carbide morphologies are stro ngly affected by carbide composition, and additional efforts are needed to draw connections be tweentive eleme concentrations and morphology. Improved under standing of these relationships will allow for the prediction of carbide morphol ogy based on alloy composition (particularly concentrs of carbide fng elements) C content. The specific roles of minor additionsh as B, N, Hf Nb in alteringide morphologies also need continued and Auger ele copy (Aould be used tect minor elements present at levels less than 100 ppm. It is assu would hdentify specifi ions, such asbide cores, wiher atomic concentrations. Different combinations of minor elements, such as ppm additions of both B and N, could be investigated to i dent carbide morp gy producing best balance of improved castability and minimiz egative mechl property effects. Addition, alloys could be studied wit h mindditions and to isolate a possible effects of t he elemental additions. Further creep testing at low temperat und high stress needed to explain ased primaryp strains in C ified alloys. t from the gas turbine industry should be solicited to determine the tin arbid ral lo t for fut aths me to tand her m s impact SX superalloy T ities are briefly described below. rela ntal ation ormi and suc and carb attention. Techniques such as secondar y ion mass spectrometry (SIMS) ctronctros spe ES) c to de med that B and N reside in carbides, but their detection elp i c reg car th hig ify a hol o the ed n anica ally or a no C ny res a ses i fully incre e cre -m od Inpu importance of primar y creep behavior in C218

PAGE 219

219 mount of primary p to comparermation mec Using FIB techniques, TEM specimens can be selectively prepared near carbides to identify deformation mechanisms, such as precipitate shearingat may explain The most significant impact of carb idees on SX suploy perform identifiethis study was reduction of Hlifetimes due ack initiatio carbidesd pores assoc with carbide CF testing at i ety of temperatures and stres would help idtify the conditio t which certaick initiationand crack fotion mechanisre active. Fa crack growth and LCF t should abe conductedX superalloydified with C to further isolate the effect f carbides on fatigue crack propagation. modified SX components. Several tests s hould be interrupted after a small a cree defo hani sms in C-containing and C-free alloys. any local differences in th in creased primary creep levels. phas eral ance d in the CF to cr n at an iated s. H a var sse en ns a n cra sites rma ms a tigue rate esting lso on S s mo o

PAGE 220

APPENDIX A CAST BAR MISORIENTATION DATA amma eference direction. Delta ( ) represents rotation about the normal to the plane with both the < 001> nearest to the reference direction and the <001> direction within the referenc e plane perpendicular to the re ference direction. Beta ( ) is the clockwise rotation about the reference di rection measured from the reference plane to the nearest {001} plane passing through the <001> direction nearest to the reference direction. Alpha ( ) represents the angle between the reference direction and the <001> direction, and it is derived from t he other three angle va lues [101]. All mechanical tests were conducted using specimens from bars with values less than 10o. Orientation data for each SX bar was pr ovid ed by PCC Airfoils (Tables A-1 through A-4). Primary angles were measured using Laue diffracti on patterns. G ( ) is defined as a rotation about a <001> axis lying within reference plane and perpendicular to the r 220

PAGE 221

Table A-1. Orientation data for baseline CMSX-4 SX cast bars. Bar Numb er (o) (o) (o) (o) 1 11.8 -3.3 12.2 -13 2 1.7 1.5 2.3 34.1 3 -3.2 3 4.4 -14.5 4 4.2 0 4.2 -4.5 5 2.3 3.1 3.9 25.3 6 -0.3 0.4 -21.5 7 3.8 -1.1 -20.1 8 -3.8 -4 5.6 -15.4 9 -0.8 -1.6 1.8 -26.7 10 8.2 7.5 11.1 40.8 11 0.6 1.4 1.5 -19.8 12 0.4 3.5 3.5 -41.4 13 0.1 -3.7 3.7 42.2 14 0.4 4.2 4.3 -5.2 15 3 -1.6 3.4 -14.1 16 1 1.2 1.6 24.7 17 1 2.7 2.9 -24.4 18 0.3 1.6 1.6 35.9 19 0.9 1.4 1.6 32 20 1.2 4.1 4.2 -11.6 0.5 4 Table A-2. Orientation data for the CMSX-4 C modification SX cast bars. Bar Number (o) (o) (o) (o) 21 2.6 3.4 4.3 11.1 22 -0.6 1.2 1.3 36.5 23 2 -1 2.2 -32.6 24 -0.3 5.4 5.4 34.9 25 1.9 0 1.9 -31.5 26 -1.6 7.7 7.8 41 27 -0.1 -2.2 2.2 16 28 -0.9 6.5 6.5 2.9 29 -0.9 3.2 3.3 -45 30 0.9 5.8 5.8 10.6 31 2.2 0.7 2.3 -31.3 32 0.2 1.1 1.1 6 33 1.1 -2 2.3 -18.6 34 2.4 1.5 2.8 33.7 35 1.5 -0.8 1.7 29.1 .7 37 -20.7 38 1.2 1.2 1.7 42.4 39 3 -0.5 3.1 -28.5 40 2.5 -0.4 2.5 -7.2 36 2.2 -1.8 2.8 -33 1.1 1.6 1.9 221

PAGE 222

222 Table A-3. Orientation data for the CMSX-4 C+B modification SX cast bars. Bar Number (o) (o) (o) (o) 41 -1.4 2.3 2.7 -15.8 42 -3.2 3.1 4.4 18.6 43 1.1 -2.7 2.9 16.7 44 4.4 9 10 35.6 45 0.5 -1.4 1.5 -45 46 5.8 3.4 6.7 32.3 47 8 -22.4 23.7 -26.4 48 5.7 1.9 6 -15 49 0.6 -2.7 2.8 26.7 50 2.3 1.5 2.7 -24.5 51 4.4 2.9 5.3 37.5 52 -1.1 -0.9 1.4 -18.8 53 0.9 1.4 1.6 -1.6 54 2.6 2.3 3.5 41.2 55 4 2.6 4.8 -39.6 56 2.3 1.5 2.7 -18 57 4.3 -1.7 4.6 -20.7 58 1.2 -5.4 5.5 1.9 59 1.4 -1.6 .1 3.5 60 -0.3 5.5 5.5 37.5 2 Table A-4. Orientation data for the CMSX-4 C+N modification SX cast bars. Bar Number (o) (o) (o) (o) 61 -10.7 -26.3 28.3 -44.9 62 5.1 5.3 7.4 -23.7 63 37.2 5.9 37.6 -43.1 64 -0.4 3.3 3.4 18.4 65 0 5.6 5.6 -45 66 0.6 1.4 1.5 -44.6 67 32 23.2 38.8 29.2 68 -8.3 -2.5 8.7 -15.5 69 4.4 -0.5 4.4 15.4 70 4.4 2.9 5.3 32.2 71 2.6 4.1 4.8 -19.7 72 2.1 5.3 5.7 -31.7 73 1.8 1.3 2.2 -42.3 74 2.9 1.9 3.4 -43 75 -1.9 2.3 3 -31.2 78 40.8 -16.9 79 -0.2 -0.9 0.9 -30.6 80 2.1 4.1 4.6 -14.2 76 3.5 2.4 4.2 12.1 77 0.6 0.2 0.6 1.6 -18.1 -37.2

PAGE 223

APPENDIX B XRD SPECTRA XRD data was collected from deep et ched samples at each stage of heat treatment and after long term exposure. This data wa s collected to confirm the presence of specific carbide phases. T he resulting XRD spectra and identifie d peaks are presented here. 223

PAGE 224

30354045505560657075802 (o)Intensity (A.U) A -MC 30354045505560657075802 (o)Intensity (A.U) B-MC 30354045505560657075802 (o)Intensity (A.U) -MC / C Figure B-1. XRD spectra for deep etched samples in the as-cast condition. A) C modification, B) C+B modifica tion, C) C+N modification. 224

PAGE 225

30354045505560657075802 (o) Inteity (A..) nsU -MCA M C 23 6 / 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / B 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / C Figure B-2. XRD spectra for deep etched sample s after HT SHT. A) C modification, B) C+B modification, C) C+N modification. 225

PAGE 226

30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / A 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / B 3035 4045505560657075802 (o)Intensity (A.U.) -MC M23C6 / C Figure B-3. XRD spectra for deep etched sa mples after aging heat treatment. A) C modification, B) C+B modifica tion, C) C+N modification. 226

PAGE 227

227 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / A 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / B 30354045505560657075802 (o)Intensity (A.U.) -MC M23C6 / C Figure B-4. XRD spectra for deep etched sa mples after thermal exposure at 1000 oC for 1000 hrs. A) C modification, B) C+B modification, C) C+N modification.

PAGE 228

APP ENDIX C COMPOSITIONAL MAPS FROM DEEP ETCHED EXPOSED SAMPLES Long term thermal exposure of the Ccontaining CMSX-4 alloys resulted in formation of secondary M23C6 carbides and TCP phases near partially decomposed MC carbides. Maps of SEM micrographs and EDS spots were generated to document these regions. Several of the maps are presented here. 228

PAGE 229

Figure C-1. Carbide map of thermally expos ed sample in the deep etched condition in the C modification. 229

PAGE 230

Figure C-2. Carbide map of thermally expos samthe de etchedition the C+B motion. ed ple in ep don c in da ific 230

PAGE 231

Figure C-3. Map of TCP phases and carbides in thermally exposed sample in the deep etched condition in the C+N modification. 231

PAGE 232

232 Figure C-4. Carbide map of thermally expos ed sample in the deep etched condition in the C+N modification.

PAGE 233

APP ENDIX D COMPOSITIONAL RESULTS FOR TCP PHASES This section contains semi -quantitative EDS results fo r the probed TCP phases in exposed specimens. Note that no TCP phases were observed in the C+B modification. Table D-1. Semi-quantitative compositions (wt. %) for TCP phase in baseline CMSX-4 exposed at 1000 oC for 1000 hrs. Phase Co Cr Ni Re Ti W 1 9.1 44.3 0.7 46.0 2 6.9 39.5 1.1 52.4 3 9.7 9.5 13.4 67.4 4 10.1 10.2 14.1 24.4 0.26 40.9 Table D-2. Semi-quantitative compositions (wt. %) for TCP phase in the C modification exposed at 1000 oC for 1000 hrs. Phase Al Co Cr Re Ta W Ti Ni C 1 14.4 11.1 55.2 19.3 2 3.7 9.8 7.7 31.3 47.4 7.0 11.7 3 6.02 75.3 3.53 6 11.3 .6 0.1 7 9 17.4 40.1 19.0 8 2.3 40.1 25.6 17.7 1.4 4.5 8.5 9 9.7 9.9 10.1 51.0 19.4 4 10.8 8.7 24.6 34.6 17.8 5 5.9 48.0 11.1 10.9 20.4 3.72 24 12.6 10. 14.1 5 Table D-3. Semi-quantitative compositi ons (wt. %) for TCP phase in the C+N modification exposed at 1000 oC for 1000 hrs. Phase Al Co Cr Re W Ti Ni C 1 12.0 12.1 46.7 19.8 9.4 2 2.1 4.4 51.1 16.3 26.2 3 3.4 5.5 31.9 14.6 1.1 43.5 4 11.6 11.2 36.6 7.9 32.6 5 2.4 12.3 10.9 37.8 36.6 6 28.6 22.5 46.2 2.6 7 9.0 10.7 57.4 15.7 7.3 8 12.3 10.1 12.9 43.6 17.3 3.83 9 9.5 10.3 30.3 34.3 10.5 2.6 10 9.8 9.0 13.5 38.6 5.3 13.5 7.4 11 8.7 9.4 32.6 9.9 3.1 12 8.5 8.6 19.6 41.5 14.6 4.5 233

PAGE 234

APPENDIX E PHACOMP APPROACH COppr at todicbise so istribution among phases. It involves mat hematical calculations to estimate the amount of each elem ent in the matrix, which is then multiplied by the number of unpaired electrons (Nv) for the given element. An alloy with an overall Nv value greater than 2.45 is predicted to be prone to phase formation. Calculations were made for all four alloy variations. An example showing one of the calculations is presented here. The PHA MP a oach tempts pre t phase sta lity ba d on lute d C modification STEP 1AlCoCrHfReTaWTiNiCSum wt. %69.66.50.12.96.56.4160.90.07100 at. % STEP 2 Subtracting TaC000000.170000.17 at. % remaining13.384545379.87.520.030.9422.11.2662.50.17 STEP 3 M23C6 secondary carbides Subtracting for Cr21(W)2C6000.580000.06000.17 at. % remaining13.384545379.86.950.030.9422.041.2662.50 STEP 4No borides STEP 5Gamma Prime Ni3(Al+Ti+Nb+Ta+Zr+0.03*Cr) STEP 6 Ni3(16.863616) Subtracting Gamma Prime13.3845453700.2300201.2650.60 at. % in Gamma09.86.720.030.9402.04011.9031.4 Rebalanced at. Fraction00.310.2100.0300.0600.380 STEP 7 Multiply by Nv numbers 45 atom ic wt. (g/mol) 26.982 58.95217818618118447.958.712 mol 0.222370469 0.160.1300.020.040.030.021.040.011.66 13.38454537 9.87.520.030.942.162.11.2662.50.33100 TaC primary carbides El ectron Hole Numbers 1.714.66N/A3.664.660.61 at. Fraction E Hole Number0 0.53100.1100.300.2302.17<2. Figure E-1. Example of PHACOMP calc ulation to predict alloy stability. 234

PAGE 235

APPENDIX F CREEP RESULT S FOR CMSX-486 The CMSX-486 specimen showed more primary creep at 850 oC/550 MPa than any of the CMSX-4 alloys tested in this study The C+B modification of CMSX-4 demonstrated increased primary cr eep strains than the other alloys in creep tes 850 also ting at the C+B modification has n the presence of B and primary creep. oC and 750 oC. CMSX-486 contains 0.015 wt.% B and 0.007 wt.% B. More testing is required to further study a possible connection betwee 0 5 10 15 20 25 050 200250Creep Strain (%)30 100150 300Time (hrs) CMSX-486 igure F-1. Full creep curve for CMSX-486 specimen tested in creep at 850 oC/550 MPa. F 235

PAGE 236

0 0.4 0.6 0.8 1.2 1.4Creep Strain Rate (/hr)0.2 1 0 51015202530Creep Strain (%)% CMSX-486 Figure F-2. Creep strain rate vs. creep stra in for CMSX-486 specimen tested in creep at 850 oC/550 MPa. Table F-1. Minimum creep rate for CMSX-486 tested in creep at 850 oC/550 MPa. Minimum Creep Rate (%/hr) CMSX-486 0.0226 0 051015202530Tim0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2e (hrs)Creep Strain (%) CMSX-486 Figure F-3. Time to 2% creep strain vs. time for CMSX-486 specimen tested in creep at 850 oC/550 MPa. 236

PAGE 237

237 Table F-2. Time to various % creep strains for CMSX-486 tested at 850 oC/550 MPa. All times are in hrs. 0.01% 0.1% 0.2% 0.5%1% 2% 5% 10% Rupture CMSX-486 0.04 0.2 0.28 0.5 1.1 29.1 123 199 268.8 Figure F-4. SEM BSE micrograph of fracture surface from creep test of CMSX-486 at 850 oC/550 MPa. EDS chemical analysis of carbides identified them as Ta rich, MC-type carbides. Figure F-5. SEM micrograph of longitudinal section of CMSX-486 specimen tested in creep at 850 oC/550 MPa. De-cohesion has occu rred at the carbide/matrix interface.

PAGE 238

APPENDIX G ELLIPTICAL MEASUREMENTS OF CREPT SAMPLES Gage section ellipticity has been identified as an indicator of slip heterogeneity and large primary creep strains. Diameter measurements were taken from the gage sections (near the fracture surface) of all tested creep specimens to determine ellipticities (diameter ratios). The results are shown below. Table G-1. Ellipticity values (ratio of major dmeter to minor diameter) for all tested creep specimens. 750 oC/800 MPa Alloy Speci men Diameter Ratio i a Baseline 1.00 C modification 1.00 C+B modification_1 1.15 C+B modification_2 1.13 C+N modification 1.02 850 oC/550 MPa Alloy Speci men Diameter Ratio Baseline 1.01 C modification 1.01 C+B modification 1.01 C+N modification 1.01 tion_HIP 1.01 CMSX-486 1.02 950 oC/300 MPa Alloy Speci men Diameter Ratio Baseline_HIP 1.01 C+B modification_HIP 1.00 C+N modifica Baseline 1.00 C modification 1.01 C+B modification 1.03 C+N modification 1.01 238

PAGE 239

APP Fracture surfaces of all HCF test s scanning electron microscope (SEM) to det fracture features. In o rder to catalog compiled. Each summary contains an over surface, SEM micrographs of the crack initia ENDIX H HCF FRACTURE SURFACE SUMMARIES pecimens were characterized using the ermine crack initiation sites and other the results, a summary of each specimen was all low magnification image of the fracture on site and other noteworthy features, a ti nd Figure H-1. Summaries of HCF fracture surf aces from Round 1 HCF tests, Un-HIPed, HT SHT specimens tested at 20 Hz. A) Baseline, B) C modification, C) C+B modification, D) C+N modification. a description of the path from crack forma here represents one round of HCF tests, consis tion to failure. Ea ch of the figures presented ting of one test of each alloy variation. 239

PAGE 240

240 B Figure H-2. Summaries of HCF fracture surf aces from Round 2 HCF tests, U n-HIPed, HT SHT specimens tested at 15 Hz. A) Baseline, B) C modification, C) C+ modification, D) C+N modification.

PAGE 241

Figure H-3. Summaries of HCF fracture surf aces from Round 3 HCF tests, U n-HIPed, HT SHT specimens tested at 15 Hz. A) Baseline, B) C modification, C) C+ modification, D) C+N modification. B 241

PAGE 242

Figure H-4. Summaries of HCF fracture surf aces from Round 4 HCF tests, H IPed, HT SHT specimens tested at 15 Hz. A) Ba seline, B) C modification, C) C+B modification, D) C+N modification. 242

PAGE 243

Figure H-5. Summaries of HCF fracture surf aces from Round 5 HCF tests, HIPed, HT SHT specimens tested at 15 Hz. A) Ba seline, B) C modification, C) C+B modification, D) C+N modification. 243

PAGE 244

Figure T e, B) C modification, C) C+B H-6. Summaries of HCF fracture surf aces from Round 6 HCF tests, HIPed, H SHT specimens tested at 15 Hz. A) Ba selin modification, D) C+N modification. 244

PAGE 245

seline, B) C modification, C) C+B Figure H-7. Summaries of HCF fracture surf aces from Round 7 HCF tests, HIPed, LT SHT specimens tested at 15 Hz. A) Ba modification, D) C+N modification. 245

PAGE 246

igure H-8. Summaries of HCF fracture surf aces from Round 8 HCF tests, HIPed, LT rred during loading of the F SHT specimens tested at 15 Hz. A) Ba seline, B) C modification, C) C+B modification. Note that a power inte rruption occu C+N modification specimen, accounting for the lack of a fracture surface for this round. 246

PAGE 247

LIST OF REFERENCES A.K. Sehra and J. Whitlow, Propulsi on and power for 21st century a [1] viation, Progress in Aerospace Sciences vol. 40, pp. 199-235, May 2004. [2] ] G. Dieter, Mechanical Metallurgy New York: McGraw-Hill, Inc., 1986. [5] 6. 41-346. ] J. Yu, X. Sun, T. Jin, N. Zhao, H. Guan, and Z. Hu, Effect of Re on deformation 6 0] A. Yeh, A. Sato, T. Kobayashi, and H. Harada, On the creep and phase stability 1] R. MacKay, T. Gabb, J. Smialek, and M. Nathal, A New Approach of Designing [12] as turbine materials and technology and their implicat ions for syngas firing, International Journal of 3] S. Tin, T. Pollock, and W. King, C arbon additions and grain defect formation in R. Bowman, Superalloys: A Primer and History, Superalloys 2000 Champion, Pennsylvania: TMS, 2000. [3] R. Decker, Strengthening Mechanisms in Nickel-Base Superalloys, Zurich, Switzerland: Climax Molybdenum Company, 1969. [4 R. Reed, The Superalloys: Fundamentals and Applications Cambridge: Cambridge University Press, 200 [6] F. Schubert, T. Rieck, and P. Ennis, The growth of small cracks in the single crystal superalloy CMSX-4 at 750 and 1000C, Superalloys 2000 Champion, Pennsylvania: TMS, 2000, pp. 3 [7] B. Wilson, The Primary Creep Behavior of Single Crystal, Nickel Base Superalloys PWA 1480 and PWA 1484, PhD Thes is, University of Florida, 2008. [8 and slip systems of a Ni base single-crystal superalloy, Materials Science and Engineering: A vol. 458, pp. 39-43, Jun. 2007. [9] G. Erickson, The Development and Application of CMSX-10, Superalloys 199 Champion, Pennsylvania: TMS, 1996, pp. 35-44. [1 of advanced Ni-base single crystal superalloys, Materials Science and Engineering: A vol. 490, pp. 445-451, Aug. 2008. [1 Superalloys for Low Density, Journal of Materials vol. 62, pp. 48-54, 2010. I. Wright and T. Gibbons, Recent dev elopments in g Hydrogen Energy vol. 32, pp. 3610-3621, Nov. 2007. [1 high refractory nickel-base single crystal superalloys, Superalloys 2000 Champion, Pennsylvania: TMS, 2000, pp. 201-210. 247

PAGE 248

[14] K.A. Al-Jarba and G.E. Fuchs, Eff ect of carbon additions on the as-cast microstructure and defect formation of a single crystal Ni-based superalloy, Materials Science and Engineering A vol. 373, pp. 255-267, May 2004. 5] S. Tin and T.M. Pollock, Phase instabili ties and carbon additions in single-crystal 16] J. Mihalisin, J. Corrigan, M. Launsbach, E. Leonard, R. Bake r, and B. Griffin, ings, 7] Y. Zhou and A. Volek, Effect of ca rbon additions on hot tearing of a second 8] L.R. Liu, T. Jin, N.R. Zhao, X.F. Sun, H.R. Guan, and Z.Q. Hu, Formation of 9] Y.H. Kong and Q.Z. Chen, Effect of minor additions on t he formation of TCP eering 0] D. Porter a nd K. Easterling, Phase Transformations in Metals and Alloys Boca [21] ft, and A. Mitchell, Freckle Formation and Freckle Criterion in Superalloy Castings, Metallurgical and Materials Transactions [22] Nickel-Base Superalloys, PhD Thesis, University of Michigan, 2001. [23] 4] N. D'Souza, M. Newell, K. Devendra, P. Jennings, M. Ardakani, and B. Shollock, 5] X. Yang, D. Ness, P.D. Lee, and N. D'Souza, Simulation of stray grain formation [1 nickel-base superalloys, Materials Science and Engineering A vol. 348, pp. 11 121, May 2003. [1 Some Effects of Carbon in the Production of Single Crystal Superalloy Cast Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 795-800. [1 generation nickel-base superalloy, Materials Science and Engineering: A vol. 479, pp. 324-332, Apr. 2008. [1 carbides and their effects on stress ruptur e of a Ni-base single crystal superalloy, Materials Science and Engineering A vol. 361, pp. 191-197, Nov. 2003. [1 phases in modified RR2086 SX superalloys, Materials Science and Engin A vol. 366, pp. 135-143, Feb. 2004. [2 Raton, FL: Taylor & Francis Group, 2004. P. Auburtin, T. Wang, S. Cockcro B vol. 31B, pp. 801-811, Aug. 2000. S. Tin, Carbon Additions and Grain De fect Formation in Directionally Solidified D. Ness and P.D. Lee, Unpublished Research, 2007. [2 Formation of low angle boundarie s in Ni-based superalloys, Materials Science and Engineering: A vol. 413-414, pp. 567-570, Dec. 2005. [2 during single crystal seed melt-back and initial withdrawal in the Ni-base superalloy CMSX4, Materials Science and Engineering: A vol. 413-414, pp. 571577, Dec. 2005. 248

PAGE 249

[26] B. Wilson, E. Cutler, and G. Fuchs, Effect of solidificat ion parameters on the structures and pr operties of CMSX-10, Materials Science and Engine micro ering: A vol. 79, pp. 356-364, Apr. 2008. [27] University of Florida, 2006. [28] p properties, Materials Science and Engineering A vol. 385, pp. 402-418, Nov. [29] urbo Expo 2009: Power for Land, Sea and Air Orlando, FL: ASME, 2009. [30] nd P. at none and Engineering: A vol. 474, pp. 71-81, Feb. 2008. [32] Press, Accepted Manuscript. t 84, Sep. 2007. [35] ase superalloys, International Journal of Fatigue vol. 27, pp. 113-127, Feb. 2005. [36] 5, Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 451-458. 4 E. Cutler, Effect of Carbon A dditions and Carbide Morphology on the Microstructure and Mechanical Properties of Ni-Base Superalloys, PhD Thesis, Q. Chen, C. Jones, and D. Knowles, The grain boundary microstructures of the base and modified RR 2072 bicrystal superalloys and their effects on the cree 2004. J. Wahl and K. Harris, CMSX-486 Alloy Update, Proceedings of ASME T K. Harris and J. Wahl, Improved Single Crystal Superalloys, CMSX-4 (SLS)[La+Y] and CMSX-486, Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 45-52. [31] R. Barrie, T. Gabb, J. Telesman, P. Kantzos, A. Pr escenzi, T. Biles, a Bonacuse, Effectiveness of shot peening in suppressing fatigue cracking metallic inclusions in Udimet 720, Materials Scienc K.S. Chan, Roles of Microstr ucture in Fatigue Crack Initiation, International Journal of Fatigue In [33] P. Kantzos, P. Bonacuse, J. Telesman, T. Gabb, R. Barrie, and A. Banik, Effec of powder cleanliness on the fatigue behavior of powder metallurgy Ni-disk alloy Udimet 720, Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 409417. [34] R.J. Morrissey and P.J. Golden, Fatigue strength of a single crystal in the gigacycle regime, International Journal of Fatigue vol. 29, pp. 2079-20 M.M. Shenoy, R.S. Kumar, and D.L. Mc Dowell, Modeling effe cts of nonmetallic inclusions on LCF in DS nickel-b X. Xie, L. Zhang, M. Zhang, J. Dong, and K. Bain, Micro-mechanical behavior study of non-metallic inclusions in P/M disk superalloy Rene'9 249

PAGE 250

[37] Q.Z. Chen, N. Jones, and D.M. Knowles, The microstructures of base/modified RR2072 SX superalloys and their effects on creep properties at elevated temperatures, Acta Materialia vol. 50, pp. 1095-1112, Mar. 20 02. igue of a single crystal Ni-base superalloy at 1000 C, Materials Science and Engineering: A vol. 0] E. Silveira, G. Atxaga, E. Erauzkin, and A. Irisarri, S tudy on the root causes for 1] E. Cutler, A. Wasson, and G. Fuchs, Effect of minor allo ying additions on the vol. 2] E. Cutler, A. Wasson, and G. Fuchs, Effect of minor allo ying additions on the 3] S. Hegde, R. Kearsey, and J. Beddoes, Resolidification behavio r of single-crystal [44] Hu Effect of small am ount of nitrogen on carbide characteristics in unidi rectional Ni-base superalloy, Metallurgical and [45] ostructure and the Mechanical Properties of Model Single Crystal Ni-Base Superalloy, PhD Thesis, [46] ng, X.F. Sun, H.R. Guan, and Z.Q. Hu, Effect of carbon additions on the microstructure in a Ni-base single crystal superalloy, [47] Superalloy, Materials Science Forum vol. 486-487, pp. 460-463, 2005. [48] y on MC 2086 SX superalloys, Scripta Materialia vol. 47, pp. 669-675, Nov. 2002. [38] Y. Kong, Q. Chen, and D. Knowles, Eff ects of minor additi ons on microstructure and creep performance of RR2086 SX superalloys, Journal of Materials Science vol. 39, pp. 6693-7001, 2004. [39] J. Yi, C. Torbet, Q. F eng, T. Pollock, and J. Jones, Ultrasonic fato443, pp. 142-149, Jan. 2007. [4 the premature failure of an aircraft turbine blade, Engineering Failure Analysis vol. 16, pp. 639-647, Mar. 2009. [4 solidification of single-cr ystal Ni-base superalloys, Journal of Crystal Growth 311, pp. 3753-3760, Jul. 2009. [4 carbide morphology in a single crystal Ni-base superalloy, Scripta Materialia vol. 58, pp. 146-149, Jan. 2008. [4 superalloys, Scripta Materialia vol. 57, pp. 837-840, Nov. 2007. X. Huang, Y. Zhang, Y. Liu, and Z. Materials Transactions A vol. 28A, pp. 2143-2147, Oct. 1997. K.A. Al-Jarba, Effect of Carbon Addition on t he Micr University of Florida, 2003. L.R. Liu, T. Jin, N.R. Zhao, Z.H. Wa Materials Letters vol. 58, pp. 2290-2294, Jul. 2004. C. Jo, Effect of Boron on Solidifi cation Microstructure in a Single Crystal Q.Z. Chen, C.N. Jones, and D.M. Knowle s, Effect of alloying chemistr carbide morphology in modified RR2072 and RR 250

PAGE 251

[49] J. Cormier, P. Villechaise, and X. Milhet, [gamma]'-pha se morphology of Nibased single crystal superalloys as an indica tor of the stress concentrat vicinity of pores, Materials Science and Engineer ion in the ing: A vol. 501, pp. 61-69, Feb. 2009. [50] Jul. 2004. [51] regation, and poros ity of directionally solidified MarM247, Materials Science and Engineering A vol. 318, pp. 264-276, Nov. 2001. [52] glecrystal nickel-base superalloys, Materials Science and Engineering: A vol. 425, pp. 47-54, Jun. 2006. [53] a Ni-base single crystal superalloy, Materials Letters vol. 58, pp. 2290-2294, Jul. 2004. [54] e nickel-base superalloy, Materials Characterization vol. 60, pp. 1114-1119, Oct. 2009. [55] ization ring vol. 76, pp. 173-180, Dec. 1985. [56] loys, Metallurgical and Ma terials Transactions A vol. 35A, pp. 1669-1679, Jun. 2004. [57] ity of ers vol. 62, pp. 258-261, Jan. 2008. [58] des 9] X. Qin, J. Guo, C. Yu an, C. Chen, J. Hou, and H. Ye, Decomposition of primary oy, Q.Z. Chen, Y.H. Kong, C.N. Jones, and D.M. Knowles, Porosity reduction by minor additions in RR2086 superalloy, Scripta Materialia vol. 51, pp. 155-160, H.S. Whitesell and R.A. Overfelt, I nfluence of solidification variables on the microstructure, macroseg T. Link, S. Zabler, A. Epishin, A. Haibel, M. Bansal and X. Thibault, Synchrotron tomography of porosity in sin L.R. Liu, T. Jin, N.R. Zhao, Z.H. Wang, X.F. Sun, H.R. Guan, and Z.Q. Hu, Effect of carbon additions on the microstructure in A. Szczotok, J. Richter, and J. Cwa jna, Stereological characterization of [gamma]' phase precipitation in CMSX-6 monocrystallin D. Anton and A. Giamei, Porosity dist ribution and growth during homogen in single crystals of a nickel-base superalloy, Materials Science and Enginee G. Lvov, V. Levit, and M. Kaufman, Mechanism of primary MC carbide decomposition in Ni-base superal X. Qin, J. Guo, C. Yu an, J. Hou, and H. Ye, Precipi tation and thermal instabil M23C6 carbide in cast Ni-base superalloy K452, Materials Lett X. Qin, J. Guo, C. Yu an, J. Hou, and H. Ye, Thermal stability of primary carbi and carbonitrides in two cast Ni-base superalloys, Materials Letters vol. 62, pp. 2275-2278, May 2008. [5 MC carbide and its effects on the fracture behaviors of a cast Ni-base superall Materials Science and Engineering: A, vol. 485, pp. 74-79, Jun. 2008. 251

PAGE 252

[60] A. Shulga, Boron and carbon behavior in the cast Ni-base superalloy EP962, Journal of Alloys and Compounds vol. 436, pp. 155-160, Jun. 2007. 1] L. He, Q. Zheng, X. Sun, H. Guan, Z. Hu, A. Tieu, C. Lu, and H. Zhu, Effect of 128-136, May 2005. y, 7. ures on the eralloys, 4] A. Sengupta, S. Putatunda, L. Bartosie wicz, J. Hangas, P. Nailos, M. Peputapeck, superalloy g ture N -92, May ine of 9] W.D. Pridemore, Stress-Rupture Characterization in Nickel-Based Superalloy 0] K. Zhao, Y. Ma, and L. Lou, Improve ment of creep rupture strength of a liquid [6 heat treatment on microstructures and t ensile properties of Ni-base superalloy M963, Materials Science and Engineering A vol. 398, pp. [62] J. Yu, X. Sun, N. Zhao, T. Jin, H. Guan, and Z. Hu, Effec t of heat treatment on microstructure and stress rupture life of DD32 single crystal Ni-base superallo Materials Science and Engineering: A, vol. 460-461, pp. 420-427, Jul. 200 [63] M.V. Acharya and G.E. Fuchs, The effe ct of long-term the rmal expos microstructure and properties of CMSX10 single crystal Ni-base sup Materials Science and Engineering A vol. 381, pp. 143-153, Sep. 2004. [6 and F. Alberts, Tensile behavior of a new single-crystal nickel-based (CMSX-4) at room and elevated temperatures, Journal of Materials Engineerin and Performance vol. 3, pp. 73-81, Feb. 1994. [65] P. Veyssire, Yield stress anomalies in ordered alloys: a review of microstructural findings and related hypotheses, Materials Science and Engineering A, vol. 309-310, pp. 44-48, Jul. 2001. [66] E. Nembach, The high temperature peak of the yield strength of [gamma]'strengthened superalloys, Materials Science and Engineering: A vol. 429, pp. 277-286, Aug. 2006. [67] N.N. Stepanova, D.P. Rodionov, V.A. Sazonova, and E.N. Khlystov, Struc formation in <001> single crystals of a nickel-based superalloy solidified with TiC powder addition, Materials Science and Engineering A vol. 284, pp. 88 2000. [68] Z. Huda, Metallurgical failure analysi s for a blade failed in a gas-turbine eng a power plant, Materials & Design vol. 30, pp. 3121-3125, Sep. 2009. [6 Gas Turbine Engine Components, Journal of Failure Analysis and Prevention vol. 8, pp. 281-288, Apr. 2008. [7 metal cooling directionally solidified nickel-base superalloy by carbides, Journal of Alloys and Compounds vol. 475, pp. 648-651, May 2009. 252

PAGE 253

[71] L. He, Q. Zheng, X. Sun, H. Guan, Z. Hu, A. Tieu, C. Lu, a nd H. Zhu, Effect carbides on the creep properties of a Ni-base s uperalloy M963, Materials Science and Engineering A vol. 397, pp. 297-304, Apr. 2005. of 2] P. Zhou, J. Yu, X. Sun, H. Guan, and Z. Hu, Role of yttrium in the microstructure ta [74] gle crystal superalloys: Origins, mechanisms and effects, Acta Materialia vol. 55, pp. 1067-1081, Feb. 2007. [75] neration lloys, Materials Science and Engineering A vol. 385, pp. 245-257, Nov. 2004. [76] rimary to Secondary Creep, Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 127-138. [77] se ylvania: TMS, 2004, pp. 197206. [78] vol. 43, pp. 1375-1384, Apr. 1995. [79] ep of ientati on and temperature, Acta Materialia vol. 47, pp. 1549-1563, Mar. 1999. [80] 3C6 rization vol. 60, pp. 1517-1521, Dec. 2009. [81] ct sed single cr ystal superalloy, Materials Science and Engineering A vol. 385, pp. 105-112, Nov. 2004. [82] of y, Journal of Alloys and Compounds vol. 457, pp. 185-190, Jun. 2008. [7 and mechanical properties of a bor on-modified nickel-based superalloy, Scrip Materialia vol. 57, pp. 643-646, Oct. 2007. [73] Mechanical Properties: Testing Euopean Space Agency, 2006. C. Rae and R. Reed, Primary creep in sin F. Diologent and P. Caron, On the creep behavior at 1033 K of new ge single-crystal supera G. Drew, R. Reed, K. Kakehi, and C. Rae, Single Crystal Superalloys: The Transition from P D. Shah, S. Vega, S. Woodard, and A. Cetel, Pri mary Creep in Nickel-Ba Superalloys, Superalloys 2004 Champion, Penns L. Pan, I. Scheibli, M. Henderson, B. Shollock and M. McLean, Asymmetric creep deformation of a single crystal superalloy, Acta Metallurgica et Materialia N. Matan, D.C. Cox, P. Carter, M.A. Rist, C.M.F. Rae, and R. C. Reed, Cre CMSX-4 superalloy single crystals: effe cts of misor D. Wang, J. Zhang, and L. Lou, F ormation and stability of nano-scaled M2 carbide in a directionally so lidified Ni-base superalloy, Materials Characte L.R. Liu, T. Jin, N.R. Zhao, Z.H. Wang, X.F. Sun, H.R. Guan, and Z.Q. Hu, Effe of carbon addition on the creep properties in a Ni-ba J. Liu, T. Jin, W. Li, X. Sun, H. Guan, and Z. Hu, Creep fracture mechanism TLP joint of a Ni-base si ngle crystal superallo 253

PAGE 254

[83] D.W. MacLachlan and D.M. Knowle s, Modelling and prediction of the rupture behaviour of singl e crystal superalloys, Materials Science and Engineering A, vol. 302, pp. 275-285, Apr. 2001. stress to the oys, vol. 54, pp. 288-309, Feb. 2006. n of e l Materials Science v ol. 47, pp. 471-481, Dec. 2009. [88] vol. 454-455, pp. 461-466, Apr. 2007. ecia l emphasis on deformation modes and oxidation, Engineering Failure Analysis vol. 16, pp. 2668-2697, Dec. 2009. [90] age on the high temperature fatigue life of an aircraft turbine disc, Engineering Failure Analysis 1] M. Kim, D. Kim, and O. Oh, Effect of [gamma]' prec ipitation during hot isostatic 2] K. Harris, G. Erickson, S. Sikkenga, W. Brentnall, J. Aurrecoechee, and K. ified Vane Applications in Advanced Turbine Engines, Journal of Materials En gineering and Performance vol. 2 (4), pp. 481[93] itio ns on the High-Cycle Fatigue Properties of the Single-Crystal Nickel-Ba se Superalloy PWA 1483, Metallurgical and Materials Transactions A vol. 38A, pp. 1177-1183, Jun. 2007. [84] L. Zhao, N. O'Dowd, and E. Busso, A coupled kinetic-constitutive approach study of high temperature crack initiation in single crystal nickel-base superall Journal of the Mechanics and Physics of Solids [85] K. Cheng, C. Jo, T. Jin, and Z. Hu, Influence of applied st ress on the [gamma] directional coarsening in a single crystal superalloy, Materials & Design vol. 31, pp. 968-971, Feb. 2010. [86] R. Reed, D. Cox, and C. Rae, Dam age accumulation during creep deformatio a single crystal superalloy at 1150oC, Materials Science and Engineering: A vol. 448, pp. 88-96, Mar. 2007. [87] T. Tinga, W. Brekelmans and M. Geers, Directional coarsening in nickel-bas superalloys and its effect on the mechanical properties, Computationa L. Shui, T. Jin, S. Ti an, and Z. Hu, Influence of prec ipitate morphology on tensile creep of a single crystal nickel-base superalloy, Materials Science and Engineering: A [89] A. Pineau and S.D. Antolovich, Hi gh temperature fatigue of nickel-base superalloys A review with sp E. Silveira, G. Atxaga, and A. Irisarri, Influence of the level of dam vol. 16, pp. 578-584, Mar. 2009. [9 pressing on the mechanical property of a nickel-based superalloy, Materials Science and Engineering: A vol. 480, pp. 218-225, May 2008. [9 Kubarych, Development of Two Rheniu m-Containing Superalloys for SingleCrystal Blade and Directionally Solid 487, Aug. 1993. M. Lamm and R. Singer, The Effect of Casting Cond 254

PAGE 255

[94] R.J. Morrissey, R. John, and W. John Porter III, Fatigue variability of a single crystal superalloy at elevated temperature, International Journal of Fatigue, vol. 31, pp. 1758-1763, Nov. 2009. [95] D. DeLuca and C. Annis, Fatigue in Single Crystal Nickel Superalloys West Palm Beach, FL: United Technologies Pratt and Whitney, 1995. 6] C. Zhaokuang, Y. Ji njiang, S. Xiaofeng, G. Hengrong, and H. Zhuangqi, High 7] L. Xiao, D. Chen, and M. Chaturvedi, Effect of boron and carbon on 2, f IN 718 super alloy: Part I. Deformation behavior, Materials Science and Engineering: A, vol. 437, pp. 157-171, Nov. 2006. [99] Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 275-281. [100] ashman, C ompeting failure modes and complex S-N curves in fatigue of structural materials, International Journal of 01] K. Clay, J. Jackson, P. Quested, and R. Morrell, NPL Measurement Good l 03] M. Okazaki and M. Sakaguchi, Thermo-mechanical fatigue failure of a single 04] M. Okazaki, I. Ohte ra, and Y. Harada, Damage repai r in CMSX-4 alloy without [9 cycle fatigue behavior of a directionally solidified Ni-base superalloy DZ951, Materials Science and Engineering: A, vol. 496, pp. 355-361, Nov. 2008. [9 thermomechanical fatigue of IN 718 superalloy: Part II: Deformation microstructures, Materials Science and Engineering: A vol. 437, pp. 172-18 Nov. 2006. [98] L. Xiao, D. Chen, and M. Chaturvedi, Effect of boron and carbon on thermomechanical fatigue o L. Xiao, D. Chen, and M. Chaturvedi, Effect of boron concentration on fatigue crack propagation resistance and low cycle fatigue properties of Inconel 718, K. Ravi Chandran, P. Chang, and G. C Fatigue vol. 32, pp. 482-491, Mar. 2010. [1 Practice Guide No. 112 Improving SingleCrystal Orientation Determination for Advanced Nickel-Based Alloys Middlesex, England: National Physical Laboratory, 2009. [102] L.R. Liu, T. Jin, N. R. Zhao, Z.H. Wang, X.F. S un, H.R. Guan, and Z.Q. Hu, Microstructural evolution of a single cr ystal nickel-base superalloy during therma exposure, Materials Letters vol. 57, pp. 4540-4546, Nov. 2003. [1 crystal Ni-based superalloy, International Journal of Fatigue, vol. 30, pp. 318-323 Feb. 2008. [1 fatigue life reduction penalty, Metallurgical and Materials Transactions A vol. 35A, pp. 535-542, Feb. 2004. 255

PAGE 256

[105] J. Chang, Y. Yun, C. Choi, and J. Kim, Developm ent of Microstructure and Mechanical Properties of a Ni-Base Singl e-Crystal Superalloy by Hot-Isostatic Pressing, Journal of Materials E ngineering and Performance vol. 12 (4), pp. 420[106] to remove porosity & creep damage, Materials & Design vol. 3, pp. 461-469, Jun. 1982. [107] cture Materials Chemistry and Physics vol. 84, pp. 284-290, Apr. 2004. [108] : ASTM International, 2008. [109] f Ni570[110] Y. Liu, R. Hu, J. Li, H. Kou, H. Li, H. Chang, and H. Fu Hot working characteristic of as-cast and homogenized Ni-Cr-W superalloy, Materials Science and Engineering: A vol. 508, pp. 141-147, May 2009. [111] Z. Dong, X. Li, G. Yuan, Y. Cong, N. Li, Z. Hu, Z. Jiang, and A. Westwood, Fabrication of protective tantalum ca rbide coatings on carbon fibers using a molten salt method, Applied Surface Science vol. 254, pp. 5936-5940, Jul. 2008. [112] J. Yang, Q. Zheng, X. Sun, H. Guan, and Z. Hu, Relative stability of carbides and their effects on the properties of K465 superalloy, Materials Science and Engineering: A vol. 429, pp. 341-347, Aug. 2006. [113] J. Yang, Q. Zheng, X. Sun, H. Guan, and Z. Hu Topologically close-packed phase precipitation in a nickel-base superalloy during thermal exposure, Materials Science and Engineering: A, vol. 465, pp. 100-108, Sep. 2007. [114] B. Yan, J. Zhang, and L. Lou, Effect of boron additi ons on the microstructure and transverse properties of a directionally solidified superalloy, Materials Science and Engineering: A vol. 474, pp. 39-47, Feb. 2008. [115] L. Zheng, C.Q. Gu, and Y.R. Zheng, Inv estigation of the solid ification behavior of a new Ru-containing cast Ni-base superalloy with high W content, Scripta Materialia vol. 50, pp. 435-439, Feb. 2004. 425, Aug. 2003. R. Stevens and P. Flewitt, Hot isos tatic pressing H.Y. Bor, C. Hsu, and C. N. Wei, Influence of hot isos tatic pressing on the fra transitions in the fine grain MAR-M247 superalloys, Standard Test Method for Determining Volume Fraction by Systematic Manual Point Count, ASTM Standards 2008 West Conshohoken, PA S. Das, J. Seol, Y. Kim, and C. Park, Structure and mechanical properties o Cr alloy produced by single roll strip casting, Materials & Design vol. 31, pp. 573, Jan. 2010. 256

PAGE 257

[116] A. Fukunaga, S. C hu, an structure, and superconducting properties of tantalum carbide nanorods and nanoparticles, Temperature Synthesis of Tantalum Carbide Powders, Journal of American behaviors and effect factors of singl e crystal nickel-base superalloys, Materials decomposition behavior in Ni-base superalloy GTD 111, Materials Science and [120] e volume fractions, Acta Metallurgica vol. 22, pp. 577-588, May 1974. [121] F. Pyczak, H. Bie rmann, H. Mughrabi, A. Vo lek, and R. Singer, CBEDases in Ni-base superalloys, Superalloys 2000 Champion, Pennsylvania: TMS, 2000, Publishing Company, 1994. [123] R.W. Powers and M.V. Doyle, Diffusi on of interstitial solutes in the Group V f Applied Physics vol. 30, Apr. 1959. [124] M. Singleton and P. Nash, T he C-Ni (Carbon-Ni ckel) System, Journal of Phase Equilibria vol. 10, pp. 121-126, Apr. 1989. [125] D. Rafaja, W. Lengauer, and H. Wiesenberger, Non-metal diffusion coefficients for the Ta-C and Ta-N systems, Acta Materialia vol. 46, pp. 3477-3483, Jun. 1998. [126] M.J. Donachie and S.J. Donachie, Superalloys: A Technical Guide Materials Park, OH: ASM International, 2002. [127] B. Bokstein, A. Epishin, T. Link, V. Esin, A. Rodin, and I. Svetlov, Model for the porosity growth in single-crystal nickel-base superalloys during homogenization, Scripta Materialia vol. 57, pp. 801-804, Nov. 2007. d M.E. McHenry, Syn thesis Journal of Materials Research vol. 13, pp. 2465-2471, Sep. 1998. [117] T. Kim and M.S. Wool dridge, Catalytically A ssisted Self-Propagating HighCeramics Society vol. 84, pp. 976-982, 2001. [118] Y. Xingfu, T. Sugui, W. Minggang, Z. Shu, L. Xudong, and C. Shusen, Creep Science and Engineering: A vol. 499, pp. 352-359, Jan. 2009. [119] B.G. Choi, I.S. Kim, D.H. Kim, and C.Y. Jo, T emperature dependence of MC Engineering: A vol. 478, pp. 329-335, Apr. 2008. D. Chellman and A. Ardell, The coars ening of [gamma]' prec ipitates at larg measurement of residual internal stra ins in the neighbourhood of TCP-ph pp. 367-376. [122] R.E. Reed-Hill and R. Abbaschian, Physical Metallurgy Principles Boston: PWS transition metals, Journal o 257

PAGE 258

258 [128] H. Marakami, T. Yamagata, H. Harada, and M. Yamazaki, “The influence of Co on creep deformation anisotropy in Ni-base singl e crystal superalloys at intermediate temperatures,” Materials Science and Engineering A vol. A223, pp. 54-58, 1997. [129] R. Unocic, K. Unocic, P. Phillips, H. Deutchman, and M. Mills, “FIB/TEM Method to Study Environmental Effects of Fa tigue Crack Initiation in Superalloys,” MS&T 2009 Pittsburgh, PA: TMS, 2009. [130] W. Evans, R. Lancaster, A. Steele, and N. Jones, “The Characterisation and Prediction of LCF Behaviour in Ni ckel Single Crystal Blade Alloys,” Superalloys 2008 Champion, Pennsylvania: TMS, 2008, pp. 425-431. [131] P. Reed and M. Miller, “Comparison of Low Cycle (Notch) Fatigue Behavior at Temperature in Single Crystal Turbine Blade Materials,” Superalloys 2008 Champion, Pennsylvania: TMS, 2008, pp. 527-533. [132] E. Ross and C. Sims “Nickel-base alloys,” Superalloys II New York: John Wiley & Sons, 1987, pp. 97-133. [133] K. Cheng, C. Jo, D. Kim, T. Jin, and Z. Hu, “Influence of local chemical segregation on the [gamma]' directional coarsening behavior in single crystal superalloy CMSX-4,” Materials Characterization vol. 60, pp. 210-218, Mar. 2009. [134] H. Nguyen, B. Pletka, and R. Heckel, High Temperature Alloys: Theory and Design New York: TMS-AIME, 1984. [135] S.T. Wlodek, “The St ability of Superalloys,” Long-Term Stability of High Temperature Materials Warrendale, PA: TM S, 1999, pp. 1-38. [136] C.M.F. Rae, N. Matan, and R.C. Reed, “The role of stacking fault shear in the primary creep of [001]-o riented single crystal superalloys at 750C and 750 MPa,” Materials Science and Engineering A vol. 300, pp. 125-134, Feb. 2001. [137] D. Hayhurst, J. Lin, an d R. Hayhurst, “Failure in notched tension bars due to hightemperature creep: Interaction between nuc leation controlled cavity growth and continuum cavity growth,” International Journal of Solids and Structures vol. 45, pp. 2233-2250, Apr. 2008. [138] Q. Yu, Z. Yue, and Z. Wen, “Creep damage evolution in a m odeling specimen of nickel-based single crystal s uperalloys air-cooled blades,” Materials Science and Engineering: A vol. 477, pp. 319-327, Mar. 2008. [139] W. Zhang, Y. Li, G. Liu, A. Zhao, C. Tao, J. Tian, and G. Yao, “Recrystallization and fatigue failure of DS alloy blades,” Engineering Failure Analysis vol. 11, pp. 429-437, Jun. 2004.

PAGE 259

259 [140] M.L. Bhatia, “Recrystal lisation--Some applied aspects,” Progress in Materials Science vol. 42, pp. 59-77, 1997. [141] S. Chen and G. Gottstein, “Strai n softening, grain boundary migration and dynamic recrystallization of nickel duri ng high temperature low cycle fatigue,” Acta Metallurgica vol. 36, pp. 3093-3101, Dec. 1988. [142] P. Wright, M. Jain, an d D. Cameron, “High Cycle Fa tigue in a Single Crystal Superalloy: Time Dependence at Elevated Temperature,” Superalloys 2004 Champion, Pennsylvania: TMS, 2004, pp. 657-666. [143] S. Xu, X. Wu, E. Han, W. Ke, and Y. Katada, “Cra ck initiation mechanisms for low cycle fatigue of type 316Ti stainless steel in high temperature water,” Materials Science and Engineering: A vol. 490, pp. 16-25, Aug. 2008. [144] W. Walston, I. Bernstein, and A. Thompson, “The role of gamma/gamma-prime eutectic and porosity on the tensile behav ior of a single-crystal nickel-base superalloy,” Metallurgical Transactions A Ph ysical Metallurgy and Materials Science vol. 22A, pp. 1443-1452, Jun. 1991.

PAGE 260

260 BIOGRAPHICAL SKETCH Andrew Wasson was born on Clark Air Forc e Base in the Philippines and grew up in places across the United States from Ala ska to Florida and many others in between. Andrew graduated with honors from Niceville High School in Niceville, Florida in 2002. He went on to attend the University of Fl orida in Gainesville, Florida and began his studies majoring in chemistry. He soon felt a draw towards engineering and changed his major to materials science and engineer ing with a specialty in metals, and he graduated cum laude in 2006 with his bachelorÂ’s degree. Andrew chose to remain at the Univer sity of Florida for graduate school to continue working with his senior research advisor, Dr. Gerhard Fuchs, on high temperature alloys. During his gr aduate program, he completed two summer internships at Siemens Energy in Orlando, FL working on materials used in gas turbine power generating engines. He received his Ph.D from the University of Florida in the summer of 2010 and began employment in Hous ton, Texas working as a materials engineer in the energy industry.