Orientation and Alloying Effects on Creep Strength in Ni-Based Superalloys

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Timothy Michael Smith Jr.

Graduate Program in Materials Science and Engineering

The Ohio State University

2016

Dissertation Committee:

Dr. Michael Mills, Advisor

Dr. Hamish Fraser

Dr. Yunzhi Wang

Dr. Andrew Wessman

Dr. Bern Kohler

Timothy Michael Smith Jr.

2016

Abstract

The creep deformation mechanisms present during creep at intermediate stress and temperatures in ME3 were further investigated using diffraction contrast imaging. Both conventional transmission electron microscopy and scanning transmission electron microscopy were utilized. Distinctly different deformation mechanisms become operative during creep at temperatures between 677-815ºC, and at stresses ranging from 274-

724MPa. Both polycrystalline and single crystal creep tests were conducted. The single crystal tests provide new insight into grain orientation effects on creep response and deformation mechanisms. Creep at lower temperatures (760C) resulted in the thermally activated shearing modes such as microtwinning, stacking fault ribbons and isolated superlattice extrinsic stacking faults (SESFs). In contrast, these faulting modes occurred much less frequently during creep at 815ºC under lower applied stresses. Instead, the principal deformation mode was dislocation climb bypass. In addition to the difference in creep behavior and creep deformation mechanisms as a function of stress and temperature, it was also observed that microstructural evolution occurs during creep at 760C and above, where the secondary  coarsened and the tertiary  precipitates dissolved. Based on this work, a creep deformation mechanism map is proposed, emphasizing the influence of stress and temperature on the underlying creep mechanisms.

Next, the effects of varying crystal orientation and composition on active deformation modes are explored for two different, commercially used Ni-base disk alloys,

ii ME3 and ME501. Understanding these effects will allow for improved predictive deformation modeling and consequently faster advancements in Ni-base alloy development. In order to investigate these effects, compression creep tests were conducted on [001] and [110] oriented single crystal specimens of the disk alloys ME3 and ME501, at different stress/temperature regimes. At 760 C and below, a prominent creep anisotropy exists between the two orientations, with the [110] oriented samples exhibiting superior creep strength. At 815 C, the creep anisotropy disappeared between the two orientations.

Through bright field scanning transmission electron microscopy, it was determined that the existence of creep anisotropy is a result of differences in deformation modes between the different orientations and alloy compositions. Results of phase field modeling in which the interaction of dislocations with realistic precipitate structures is also conducted to further advance predictive creep deformation models.

Furthermore, the local compositional and structural changes occurring in association with stacking faults in ME501 are characterized and related to the possible rate- controlling processes during creep deformation at intermediate temperatures. These rate- controlling processes are not presently understood. In order to promote stacking fault shearing, compression creep tests on specially prepared single crystals of ME501 were conducted at 760°C in the [001] orientation. Scanning transmission electron microscopy

(STEM) imaging was coupled with state-of-the-art energy dispersive X-ray (EDX) spectroscopy to reveal for the first time an ordered compositional variation along the extrinsic faults inside the  precipitates, and a distinct solute atmosphere surrounding the leading partial dislocations. The local structure and chemistry at the extrinsic fault is consistent with the  phase, a D024 hexagonal structure. Density Functional Theory (DFT) iii and high angle annular dark field (HAADF)-STEM image simulations are consistent with local  phase formation and indicate that a displace-diffusive transformation occurs dynamically during deformation.

Additional investigation into the chemical segregation changes associated with faults in ME3 and ME501 is analyzed. Compression creep tests were conducted on [001] oriented samples at 760C in stress regimes where microtwin and stacking fault formations prominently occurred. High resolution EDX was performed in regions where stacking faults had terminated inside of a  precipitate, capturing the process as it was transpiring when the creep test had ended. Again, the presence of elemental segregation was observed along superlattice stacking faults as well as multiple examples of a Co and Cr rich Cottrell atmosphere around the leading Shockley partials. The presence and interaction of newly discovered tertiary  particles with the formation of these faults is explored. These combined observations lead to the creation of a new microtwin formation model incorporating the diffusion processes now known to ensue during twin development.

Finally, a new “phase-transformation strengthening” mechanism that resists high temperature creep deformation in Nickel-based superalloys, where specific alloying elements inhibit the deleterious deformation mode of microtwinning at temperatures above

700 C is introduced. Ultra-high-resolution structure and composition analysis via scanning transmission electron microscopy, combined with density functional theory calculations, reveals that a superalloy with higher concentrations of the elements Titanium,

Tantalum, and Niobium encourage a shear-induced solid-state transformation from the  to  phase along stacking faults in γ′ precipitates, which would normally be the precursors of deformation twins. This nanoscale  phase creates a low energy structure that inhibits iv thickening of stacking faults into twins, leading to significant improvement in creep properties.

To my wife,

And the rest of my family

vi Acknowledgements

I would be remiss if I didn’t acknowledge the many people who helped me reach this milestone in my life. First and foremost, I would like to thank my wife. I understand and am fully aware of the many sacrifices one must make when married to a graduate student. Whether it being the uncertainty of the future or the unpredictable hours spent in the lab you were always supportive and understanding throughout my time at OSU. I am truly blessed to be married to such a loving and wonderful person. Likewise, I would also like to acknowledge my family who emphasized the importance of an education and hard work. I always knew that if anything were to happen I had the support of my family to fall back on and I truly cannot over-emphasize how valuable that backing is. I respectfully dedicate this document to my wife and family with much gratitude and thanks

When I was in undergraduate studies at Wright State University I was fortunate enough to obtain an internship at the Wright Patterson Air Force Base in Dayton Oh. While there, I found myself working under the guidance of Dr. Muratore. I wouldn’t know at the time but this encounter would change the direction of my life. While there, I quickly discovered that the pursuit of a PhD in Materials Science and Engineering was what I wanted to do with my life. This may not have turned out to be true if I also hadn’t been lucky enough to end up working under Dr. Mills at The Ohio State University for my PhD.

I can honestly say that it has been my privilege to work in the Mills group and if I could go back and do this all over again I wouldn’t change a thing. The past and present Mills group members have also had a huge role in teaching and guiding me through my graduate studies. Ray Unocic, Patrick Phillips, Dan Coughlin, Matt Bowers, Hallee Deutchman,

Don McAllister, Lee Casalena, Collin Whitt, and Connor Slone; Thank you. This thanks

vii also extends to all of the great Professors and researchers in the MSE department who I have had the opportunity to collaborate with. In particular, I’d like to thank Dr. McComb,

Dr. Wang, Dr. Fraser, Dr. Ghazisaeidi, and Dr. Windl (all from Ohio State) who have all made contributions towards this document. I also received help and contributions from some great colleagues that have attended school at the same time. Bryan Esser, Nik

Antolin, Shahriar Hooshmand, and Duchao Lv; without your efforts and fruitful discussions very little of this document would be possible. Indeed, the collaboration and support from all of these scientists at Ohio State helped me throughout my graduate studies and I am grateful to have been given the opportunity to work with them.

I need to also thank all of the individuals who helped me in my characterization efforts. Specifically, Robert Williams, Babu Viswanathan, Dan Huber, and Henk Colijn. I appreciate the time and passion each of you showed to help teach me the important characterization techniques that I would subsequently use in my research.

I need to also acknowledge the GE University Strategic Alliance Program and the

NSF DMREF program for the financial support. Ultimately, this research was only possible through the collaborative efforts of GE. I would like to thank everyone at GE for their support and knowledge, I am particularly grateful to Andrew Wessman, Dave Mourer, Tim

Hanlon and Ken Bain. Lastly, I would like to thank my dissertation committee members for taking time out of their busy schedules to be a part of my defense.

I am truly blessed. Thank you all.

viii VITA

26 November 1988 ……. Born – Bellefontaine, OH

2011 ……. B.S., Mechanical Engineering Wright State University, Dayton, OH

2011 – Present .…… Graduate Research Assistant The Ohio State University, Columbus, OH

2013 ……. M.S., Materials Science and Engineering The Ohio State University, Columbus, OH

PUBLICATIONS

[1] C. Muratore, V. Varshney, J.J. Gengler, J.J. Hu, J.E. Bultman, T.M. Smith, P.J. Shamburger, B. Qiu, X. Ruan, A.K. Roy, A. A. Voevodin, “Cross-plane thermal properties of transition metal dihalcogenides,” Applied Physics letters. 2013.

[2] D.S. Stone, J. Migas, A. Martini, T.M. Smith, C. Muratore, A.A. Voevodin, S.M. Aouadi “Adaptive NbN/Ag coatings for high temperature tribological applications,” Surface and Coatings Technology. 2012

[3] T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman, et al., Segregation and  phase formation along stacking faults during creep at intermediate temperatures in a Ni-based superalloy, Acta Mater. 100 (2015) 19–31.

[4] T.M. Smith, M.S. Hooshmand, B.D. Esser, F. Otto, D.W. McComb, E.P. George et al., Atomic-Scale Characterization and Modeling of 60 Dislocations in a High Entropy Alloy, Acta Mater. (2016)

[5] T.M. Smith, L.V. Duchao, T. Hanlon, A. Wessman, Y. Wang, M.J. Mills, Determination of Orientation and Alloying Effects on Creep Response and Deformation Mechanisms in Single Crystals of Ni-Base Disk Superalloys, Superalloys 2016.

[6] T.M. Smith, R.R. Unocic, H. Deutchman, M.J. Mills, Creep Deformation Mechanism Mapping in Nickel Base Disk Superalloys, Mater. High Temp. 33 (2016) 372–383.

ix [7] T.M. Smith, B.D. Esser, N. Antolin, A. Carlsson, R.E.A. Williams, A. Wessman, et al., Phase Transformation Strengthening of High Temperature Superalloys, Nature Communications. 7 13434 (2016).

Field of Study

Major Field: Materials Science and Engineering

x Table of Contents Abstract …………………………………………………………………………………..ii Dedication ……………………………………………………………………………….vi Acknowledgements……………………………………………………………………..vii Vita ………………………………………………………………………………………ix List of Tables …………………………………………………………………………...xiv List of Figures ………………………………………………………………………...... xv Nomenclature ………………………………………………………………………...... xx Chapter 1 - Introduction and Background ...... 1 1.1 Ni-based Superalloys ...... 1 1.2 Deformation mechanisms ...... 8 1.2.1 Orientation Dependence on Deformation Mechanisms ...... 16 1.3 Mechanical Behavior ...... 17 1.3.1 Creep ...... 18 1.3.2 Tension/Compression Asymmetry in Creep ...... 21 1.4 Motivation for Study ...... 22 References ...... 24 Chapter 2 – Creep Deformation Mechanism Mapping in Nickel Base Disk Superalloys...... 30 2.1 Introduction ...... 30 2.2 Materials and Experimental Methods ...... 32 2.2.1 Creep Tests ...... 33 2.2.2 Microstructural Characterization ...... 35 2.3 Results ...... 36 2.3.1 Creep Behavior ...... 36 2.3.2 Microstructural Evolution ...... 39 2.3.3 Deformation Characterization using (C)TEM and (S)TEM ...... 41 2.4 Discussion ...... 49 2.4.1 Deformation Mechanism Map ...... 49 2.4.2 Microtwin Formation ...... 51 2.4.3 Segregation and Cottrell Atmospheres ...... 56 2.5 Conclusion ...... 58 2.6 References ...... 59 Chapter 3 – Orientation and Alloying effects on Creep Response and Deformation Mechanisms in Single Crystal Ni-base Disk Superalloys ...... 65 3.1 Introduction ...... 65 3.2 Materials and Experimental Methods ...... 68 3.2.1 Sample preparation ...... 68 3.2.2 Creep Test Preparation ...... 70 3.2.3 Deformation Characterization Methods ...... 71 3.3 Results ...... 72 3.3.1 Monotonic Compression Creep Response ...... 72 3.3.2 Compression Creep Response: Polycrystalline vs. Single Crystal ...... 75 3.3.3 Transient Stress Tests in ME501 ...... 76 3.3.4 Deformation Analysis Using DC-STEM Imaging ...... 78 3.4 Discussion ...... 83 3.4.1 Orientation Effects on Deformation ...... 83 xi 3.4.2 Dislocation Activity Diagram (DAD) – Predicting De-Correlation ...... 85 3.5 Conclusions ...... 92 References ...... 93 Chapter 4 - Segregation and η Phase Formation Along Stacking Faults During Creep at Intermediate Temperatures in a Ni-Based Superalloy ...... 98 4.1 Introduction ...... 98 4.2 Materials and Experimental Methods ...... 102 4.2.1 Sample Preparation ...... 102 4.2.2 Creep Sample Orientation and Preparation ...... 103 4.2.3 Microscopy Methods ...... 104 4.3 Results ...... 105 4.3.1 Diffraction Contrast STEM ...... 105 4.3.2 Segregation and Ordering along Stacking Faults ...... 107 4.4 Discussion ...... 116 4.4.1. SESF Formation ...... 116 4.4.2 η phase formation ...... 119 4.4.3 Density Functional Theory Simulations ...... 122 4.4.4 HAADF Simulations ...... 127 4.5 Conclusions ...... 130 References: ...... 132 Chapter 5 – Diffusion Processes During Creep at Intermediate Temperatures in Ni- based Superalloys ...... 137 5.1 Introduction ...... 137 5.2 Experimental Methods ...... 139 5.2.1 Creep Sample and Testing ...... 139 5.2.2 Microscopy and Chemical Analysis ...... 140 5.3 Results ...... 141 5.3.1 Evidence of Co and Cr Rich Cottrell Atmospheres Ahead of Stacking Faults in ME3 ...... 141 5.3.2 Evidence of Co and Cr Rich Cottrell Atmospheres Ahead of Stacking Faults in ME501 ...... 145 5.3.3 Evidence of Co-Rich and Cr-Rich Cottrell Atmospheres Along Thickening Twin Boundaries...... 147 5.3.4 Evidence of Segregation Along Microtwins ...... 150 5.4 Discussion ...... 154 5.4.1 New Twin Formation Model ...... 154 5.4.2 Tertiary  particles ...... 158 5.4.3 Tertiary Gamma Particle Effects ...... 161 5.4.4 Rate Limiting mechanisms: Preliminary Discussion ...... 164 5.5 Conclusions ...... 166 References: ...... 167 [26] Titus M, Mottura A, Viswanathan G, Suzuki A, Mills MJ, pollock T. High resolution energy dispersive spectroscopy mapping of planar defects in L12- containing Co-base superalloys. Acta Materialia (2015)...... 170 Chapter 6 – Phase Transformation Strengthening of High Temperature Superalloys...... 171 6.1 Introduction ...... 171 xii 6.2 Methods ...... 173 6.2.1 Microstructural characterization ...... 173 6.2.2 Compression creep sample preparation ...... 173 6.2.3 Electron microscopy characterization ...... 174 6.2.4 DFT calculation parameters ...... 175 6.3 Results ...... 176 6.3.1 Mechanical testing and STEM deformation analysis ...... 176 6.3.2 Atomic-Scale Characterization of SESFs...... 182 6.3.3 Density Functional Theory Calculations ...... 191 6.4 Discussion ...... 194 6.5 Conclusions ...... 197 References: ...... 197 Chapter 7 – Summary and Future Work ...... 203 Bibliography ...... 208

xiii List of Tables

Table 1.1: Summary of different alloying elements and their effects on properties ...... 8 Table 2.1: Secondary and tertiary  precipitate volume fraction and size (in nm) between the polycrystalline and single crystal ME3 alloys ...... 33 Table 2.2: The four different creep test set-ups examined in this study ...... 34 Table 2.3: Calculated stress exponents for the [001] and [110] orientations at 700C, 760C, and 815C...... 39 Table 3.1: Alloy compositions of ME3 and ME501 ...... 68 Table 3.2: Secondary and tertiary  precipitate area fraction and size in ME3 and ME501 ...... 70 Table 3.3. The input parameters from the DAD calculation ...... 88 Table 4.1: Compositions of 160-atom SESF calculation cells ...... 123 Table 4.2: Relaxation results of 160-atom SESF calculation cells with segregated and randomized solute atoms ...... 123 Table 5.1: The quantified concentrations of the four regions highlighted in Figure 5.3...... 144 Table 5.2: The quantified EDX compositions of the Cottrell atmosphere around the twin thickening Shockley partials compared to the composition of the  precipitate...... 150 Table 5.3: The tertiary  volume fractions after heat treated at 815C for 100 and 200 hours in ME3 and ME501 ...... 160 Table 5.4: The calculated diffusivities of Cr, Co, Mo, Ta, and Al inside a  precipitate at 760C .... 164 Table 5.5: Comparison of rate limiting mechanisms ...... 165 Table 6.1: The twin occurrence rate for a 30μm2 area for ME3 and ME501. Twins were found to be statistically more prevalent in ME3 than ME501...... 179 Table 6.2: Amount of strain contributed to twinning for ME3 and ME501. Twinning played a significant role in ME3’s creep performance...... 181 Table 6.3: Experimentally determined k-factors for [110] ME501 ...... 189 Table 6.4: Experimental error ranges for the atomic resolution EDX in Figure 6.8...... 191

xiv List of Figures

Figure 1.1: The effect of Temperature on the flow stress of different alloyed Ni3Al particles [10] ...... 2 Figure 1.2: The percent of deviation in atomic diameter between Ni and possible alloying elements [1,18] ...... 5 Figure 1.3: A diagram a.) depicting weakly coupled dislocations cutting γ′ particles b.) depicting strongly coupled dislocations cutting γ′ particles [29] ...... 9 Figure 1.4: The critical stress vs. particle diameter showing the peak in CRSS between weakly and strongly coupled dislocations and another between strongly coupled dislocation cutting and Orowan looping [20] ...... 10 Figure 1.5: Tertiary volume fraction vs. ΔE showing the change in shearing mechanisms when a critical volume fraction is reached. [29] ...... 14 Figure 1.6: High Res HAADF Image used for the analysis of microtwin formation. The lower area is a 14 layer microtwin while above is a portion of a γ′ particle. [37] ...... 16 Figure 1.7: orientation effect on dissociation of a/2<110> dislocations in Ni-based superalloys. In region a.) the trailing partial cannot propagate through the channel and large stacking faults are created. In b.) both partials are able to propagate though with differing velocities. [39] ...... 17 Figure 1.8: a.) A graph of creep strain for a polycrystalline Ni-based alloy. Note the three distinct regions of strain rate. B.) A graph of creep strain for a single crystalline Ni-based alloy [65] .... 20 Figure 1.9: Creep strains for a [256] and [100] oriented single crystal [69]...... 21 Figure 1.10: a.) Compression and tension creep strain curves for [001] oriented single crystal b.) Compression and tension creep strain curves for [011] oriented crystals [68] ...... 22 Figure 2.1: A representative SEM micrograph of the bi-modal microstructure in polycrystalline and single crystal ME3 ...... 32 Figure 2.2: (a) Creep curves for the following four conditions: 677C/690MPa (red curve), 704C/724MPa (blue curve), 760C/345MPa (green curve), and 815C/345MPa (black curve). In (b), the same creep curves magnified at lower strains...... 36 Figure 2.3: The [001] and [110] compression creep tests at (a) 700C and 552MPa, (b) 760C and 414MPa, (c) 760C and 552Mpa, and (d) 815C and 276MPa ...... 38 Figure 2.4: The post creep microstructure for ME3 after the (a) 677C/690MPa/2.0%/3750hrs strain test (b) 704C/724MPa/0.4%/160hrs strain test (c) 760C/345MPa/0.36%/3000hrs strain test and (d) 815C/345MPa/0.42%//140hrs strain test...... 40 Figure 2.5: (a) DC-STEM image showing continuous planar faulting observed in Specimen I. (b) A HR-STEM image of a microtwin observed in Specimen I [21]...... 41 Figure 2.6 A BF-TEM image revealing microtwins in Specimen II post creep test...... 42 Figure 2.7: BF-TEM micrograph of the creep deformation in specimen III post-test showing a mixture of a/2<110> dislocations and isolated stacking faults in the  precipitates...... 44 Figure 2.8: A BF-TEM image of dislocation activity in specimen IV after being crept at 815C ...... 45 Figure 2.9: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 700C compression creep tests at 552 MPa...... 46 Figure 2.10: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 760C compression creep tests at 414 MPa ...... 47 Figure 2.11: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 760C compression creep tests at 552 MPa ...... 48 Figure 2.12: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 815C compression creep tests at 274 MPa ...... 49

xv Figure 2.13: The mechanism map for several Ni-base disk alloys at different stress and temperature regimes revealing different post-creep deformation modes [15,20,26]...... 50 Figure 2.14: A high resolution HAADF-STEM image revealing a 4 layer microtwin forming from a SESF in ME3 inside a  precipitate...... 55 Figure 2.15: The twin formation model presented by Unocic et al.[21] where (a) a single ½<110> dislocation dissociates into its constituent Shockley partials creating an ISF in the  matrix while the leading partial loops the  precipitate. (b) A second ½<110> cross slips in the matrix onto the {111} plane adjacent to the ESF while an SESF is formed in the precipitate. (c) The same process occurs again to form a three layer twin in the  matrix, and repetition of the process leads to a thickening twin. (d) The new twin formation model where two unlike ½<110> dislocations interact at the interface and dissociate (e) so like-signed Shockley partials can shear into the  precipitate forming a SESF. (f) The same process can occur again to form a 3 layer twin in the precipitate (resulting from shearing events on 4 adjacent {111} planes)...... 56 Figure 2.16: Net intensity elemental maps of a SESF in ME3...... 57 Figure 3.1: SEM backscatter image of (a) ME3 microstructure and (b) ME501 microstructure...... 69 Figure 3.2: 700 C compression creep strain versus time curves from [001] and [110] orientated ME501 and ME3 samples with (a) same stress and (b) comparable strain rates...... 72 Figure 3.3: 760 C compression creep strain versus time curves from [001] and [110] orientated ME501 and ME3 samples with (a) same stress and (b) comparable strain rates...... 74 Figure 3.4: 815 C compression creep curves from [001] and [110] orientated ME501 and ME3 samples with (a) same stress and (b) comparable strain rates...... 75 Figure 3.5: Compression creep strain versus time curves from [001], [110] orientated and polycrystalline ME501 (a) at 700 C and (b) at 760 C...... 76 Figure 3.6: Transient stress test performed on a [001] compression sample of ME501 at 760C. The stress level was changed mid test to observe the change in strain rate...... 77 Figure 3.7: The calculated stress exponents (n) from the transient stress tests performed at all three temperature regimes and both orientations for ME501 are presented along with the plotted stress (MPa) vs strain rate (s-1) points...... 77 Figure 3.8. BF-STEM Zone Axis images for both [001] and [110] orientations after 700 C compression creep tests for (a) ME3 and (b) ME501...... 79 Figure 3.9: BF-STEM Zone Axis images for both [001] and [110] orientations after 760 C compression creep tests for (a) ME3 and (b) ME501...... 80 Figure 3.10: BF-STEM Zone Axis images for both [001] and [110] orientations after 815 C compression creep tests for (a) ME3 and (b) ME501...... 82 Figure 3.11: BF-STEM image of deformation in a polycrystalline sample after a compression creep test at (a) 700C and (b) 815C...... 83 Figure 3.12: Channels in a ME3 precipitate microstructure. The secondary precipitate cross-sections are shown in black. The average channel width is 210 nm. The critical channel width for the activation of de-correlation (red) is 156 nm, while that for stopping even partial looping (blue) is 65 nm...... 89 Figure 3.13: Percolation simulations preformed when dislocations are initiated at grain boundaries. white color indicates a closed zone of full dislocation activation. pink color shows the activation area of de-correlation. The red color shows unsheared area. Percolation distances are indicated by the green arrows...... 90 Figure 3.14: DAD is calculated by using (a) average channel width (210 nm) and (b) the critical channel width for de-correlation (156 nm)...... 91

xvi Figure 4.1: Back-scatter SEM image of the microstructure of ME501 where the  precipitates have been etched away (black) from the  channels (grey)...... 103 Figure 4.2: (a) [001] Zone axis bright field image (b) Near [001] zone axis LAADF-STEM image of isolated stacking faults ...... 106 Figure 4.3: LAADF-STEM images from a comprehensive g⋅b analysis of dislocation types conducted in same area near the [001] zone axis shows two different <110> type dislocations (A and B) crossing each other at the / interface where (a) g[020] and (b) g[200] diffraction vector is excited...... 107 Figure 4.4: HAADF-STEM image obtained along the [110] zone axis revealing a grid-like ordering along a SESF inside a  precipitate...... 108 Figure 4.5: EDX elemental map of a vertical SESF showing segregation along the fault ...... 109 Figure 4.6: Vertically integrated EDS elemental profiles across SESF ...... 111 Figure 4.7: (a) An example of an EDS line scan over the prominently segregated atomic columns. (b) The summed intensity for Ta (purple), Al (red) and the corresponding HAADF intensity profile (black) from the line scan ...... 112 Figure 4.8: (a) Burgers circuit analysis showing Shockley partials leading SESF shear. (b) Center of symmetry analysis on atomic peak positions from same SESF revealing a “CSF-like” fault leading the SESF...... 113 Figure 4.9: (a) The elemental maps of a terminated SESF (b) a vertically integrated EDX line scan across the SESF present in 10a...... 115 Figure 4.10: (a) Two different full dislocations interact at the /. (b) Leading, like-signed Shockley partials enter the precipitate forming a two-layer CSF, which re-orders (red circular arrow) to form a lower energy SESF. (c) Scenario incorporating new observations of both Cr and Co atmosphere surrounding the leading partials of the fault and the formation of η phase immediately trailing the partials...... 117 Figure 4.11: The structure of the Ni6AlNb reportedly seen by Pickering et al. [23] in superalloy 718 plus...... 121 Figure 4.12: Energy difference between the randomized and segregated SESF calculation cells in eV .... 124 Figure 4.13: Percent difference in out-of-plane expansion in randomized and segregated SESF calculation cells ...... 125 Figure 4.14: Planar average Bader charges on Ni as a function of atomic position for segregated and randomized calculation cells. The red dash line is the Bader partial charge of the bulk...... 126 Figure 4.15: [110] view of the segregated SESF structure in L12. Ta and Nb are segregated to Al sites, and Co to Ni sites. [52] The black box represents the area sampled for the HAADF simulation shown in Figure 4.16...... 128 Figure 4.16: (a) Experimental and (b) simulated HAADF image of SESF (c) Experimental and (d) simulated averaged intensity of grid-like ordering outlined by the dashed lines ...... 130 Figure 5.1: Atomic resolution HAADF image of a SESF (upper fault) and SISF (lower fault) terminating inside a  precipitate. The red ovals represent where each fault has ended and the hole was purposely burnt into the sample to help with EDS scan correction...... 142 Figure 5.2: Elemental maps of the terminated SESF and SISF inside a  precipitate ...... 143 Figure 5.3: Four regions highlighted to compare concentration values. 1 = region of  precipitate, 2 = SISF, 3 = SISF Cottrell atmosphere, and 4 = SESF Cottrell atmosphere...... 144 Figure 5.4: High resolution EDS scan of a SISF terminating inside a  precipitate in ME501...... 146 Figure 5.5: Elemental Cr maps of notches where (a) a SESF and (b) a SISF enter into a precipitate...... 147

xvii Figure 5.6: An atomic resolution MAADF image of a twin boundary extending by the shear of adjacent Shockley Partials...... 148 Figure 5.7: Elemental maps of a twin extending by two atomic planes...... 149 Figure 5.8: Elemental maps of microtwin shearing through a  precipitate. The presence of tertiary  is evident in the Co and Cr maps...... 151 Figure 5.9: High resolution elemental maps of twin shearing a  precipitate through the denuded zone area highlighted in Figure 6...... 152 Figure 5.10: (a) Combined Cr, Co, Ni, and Al elemental map of microtwin near the / interface. The red box represents where the integrated line scan was performed. (b) The results of the integrated line scan. The interface of the twin can easily be discerned by the segregation of Co, Cr, and Mo and depletion of Ni and Al from the interface...... 153 Figure 5.11: The Twin formation model recently described by Smith et al. [4], where two unlike ½<110> dislocations interact at the interface and dissociate (b) so like-signed Shockley partials can shear into the  precipitate forming a SESF. (c) The same process occurs again alongside the SESF to form a 3 layer twin inside the  precipitate. (d) This model has now been updated to include the diffusion processes observed in the results section and by Barba et al. (e) where  formers Co, Cr and possibly Mo segregate along the formed SESF. (f) Then when the process occurs again Some of the  formers segregate to the new twin interface from the  channel while the other source of the  formers come from what segregated along the SESF, leaving a pristine  state inside the twin boundaries...... 154 Figure 5.12: (a) A MAADF image of a Twin shearing both a  precipitate and  channel near the / interface. (b) A Cr map showing the two sources of  former segregation along the twin and twin interfaces. Inside the tertiary  denuded zone the source of the segregates is the  channel, however, in the area of the precipitate where there exist tertiary  particles the primary source of  former segregation along the twin is from those particles...... 157 Figure 5.13: The result of the heat treatment study on tertiary  particles in both ME3 and ME501. The particles were found to dissipate quicker in ME3 but after 200 hours most of the tertiary particles were gone for both alloys...... 159 Figure 5.14: Compression creep curves of overaged ME3 vs original ME3. Tested under 414MPa of stress at 760C...... 161 Figure 5.15: (a) An HAADF image of deformation observed in the [001] overaged ME3 post creep. (b) The elemental Cr map clearly illustrating the complete dissolution of tertiary  particles. . 162 Figure 5.16: The HAADF, Al, Cr, and Co maps of a terminated SISF observed in the overaged ME3 sample post-test...... 163 Figure 6.1: Twin energy in pure Ni3Al calculated as a function of number of lattice planes in the calculation cell; the energetic cost of twinning decreases drastically as defects are separated by added lattice planes, as expected...... 176 Figure 6.2: Compression creep data and deformation analysis. (a) [001] Compression creep curves of both ME3 and ME501 at 760C and 552 MPa (blue and green, respectively) and ME3 at 414 MPa (red) to achieve comparable strain rates (5e-9) between the two alloys. [001] Zone axis BF- STEM image revealing isolated SESFs and microtwins in post-crept [001] (b) ME3 and (c) ME501 crystals. HAADF-STEM images showing (d) a microtwin in ME3 and (e) a (SESF) in ME501. The inset in (d) shows a higher magnification of the twin interface in ME3. Scale bars in b and c, 500nm. Scale bar in d, 5nm (inset, 2nm). Scale bar in e, 2nm...... 178 Figure 6.3: ECCI image of (a) ME3 showing 10 different twins and (b) ME501 showing 4 different twins. All scale bars, 500nm...... 179

xviii Figure 6.4: Segregation along stacking faults in ME3 and ME501. HAADF-STEM image obtained on a [110] zone axis revealing (a) segregation along a SESF in ME3 and (b) segregation and “grid- like” ordering of η phase along a SESF in ME501. Integrated EDX line scans showing elemental segregation along a SESF for (c) ME3 and (d) ME501, as indicated in (a) and (b), respectively. Not shown is the enrichment of Co and the depletion in Ni content along the faults in both cases. All scale bars, 2nm...... 183 Figure 6.5: Rotated EDX spectrum image, showing the HAADF layer with 6 repeating units along the SESF (yellow boxes) and 5 additional repeating units (red boxes) taken midway between the first 6...... 186 Figure 6.6: Representative EDX spectra from raw data (top) and summed data (bottom) showing a significant improvement in both counts per peak and signal-to-noise ratio ...... 187 Figure 6.7: HAADF-STEM image of ME501 sample solutionized at 1210°C for 1 hour. Scale bar, 500nm...... 188 Figure 6.8: Quantified atomic resolution EDX of η phase in ɣ′ showing the HAADF-STEM image of the fault exhibiting characteristic ordering of intensity within the fault; Ni sublattice (green); Co (yellow) segregating to Ni sites; Ta and Nb (dark and light blue, respectively) segregating to the Wyckoff 2a sites; Al and Ti (red and magenta, respectively) segregating to the Wyckoff 2d sites. W, Cr, and Mo (purple, orange, and light green, respectively) are fairly noisy. All EDX values are in at%. Scale bar, 0.5nm...... 190 Figure 6.9: SESF to microtwin transformation. (a) DFT cell showing segregation along a SESF in ME3. (b) on the left is an experimental HAADF-STEM image of a two layer SESF being sheared by two Shockley partials to form a three layer twin near a / interface. On the right is a center of symmetry analysis of the HAADF-STEM image. (c) DFT cell showing the resulting twin from the process observed in 4(b). Scale bar, 1nm...... 192 Figure 6.10: DFT calculations of SESF and microtwin formations. (a) Energetic cost of twin formation by shearing along a SESF prior to reordering in Ni3Al, ME501 with a random solid solution, i.e. no segregation (RSS), and ME501 where  has nucleated along the fault as observed experimentally. Note the relatively large energy cost to form a twin along a SESF with  phase. (b) Energy difference due to reordering after twin formation. The small difference found for ME3 suggests that the segregation of  formers (Co, Cr, and Mo) replacing Ni and Al has removed nearest-neighbor violations in the precipitate near the fault, making twinning easier for ME3...... 193 Figure 6.11: Phase transformation strengthening in ME501. (a) Schematic of an isolated SESF in a  precipitate. (b)  formers (Co, Cr, and Mo) segregated along the SESF in ME3. (c) Two more dislocations have interacted at the / interface near the SESF in ME3. (d) Two more Shockley partials shear along the SESF forming a four layer twin that is able to shear both the  and  precipitates. (e)  formers (Co, Ta, Ti, and Nb) segregated along the SESF in ME501. (f) Two more dislocations have interacted at the / interface near the SESF in ME501. (g) Given results in Figure 6.10 the dislocations are not able to form a twin in ME501...... 195

xix Nomenclature

APB Anti-phase boundary BF Bright field CRSS Critical resolved shear stress CSF Complex stacking fault CTEM Conventional transmission electron microscopy DAD Dislocation activity diagram DC Diffraction contrast DFT Density functional theory ECCI Electron channeling contrast imaging EDM Electrical discharge machining EDX/EDS Energy dispersive X-ray spectroscopy FCC Face centered cubic FEM Finite element modelling FIB Focus ion beam GBS Grain boundary sliding GGA Generalized gradient approximation HAADF High angle annular dark field HCP Hexagonal close packed HR High resolution ICME Integrated computational materials science and engineering LAADF Low angle annular dark field LCF Low cycle fatigue MAADF Medium angle annular dark field MGI Materials genome initiative OIM Orientation imaging microscopy PM Powder metallurgy PSB Persistent slip bands QEP Quantum excitation of phonons RSS Random solid solution SEM Scanning electron microscopy (S)ESF Superlattice extrinsic stacking fault SF Stacking faut SFE Stacking fault energy (S)ISF Superlattice intrinsic stacking fault SSS Solid solution strengthening STEM Scanning Transmission electron microscopy TEM Transmission electron microscopy VASP Vienna Ab-initio simulation package VF Volume fraction

Chapter 1 - Introduction and Background

1.1 Ni-based Superalloys

Ni-based superalloys are essential materials found in the hot section of jet turbine engines and therefore, critical to the engines properties such as operating temperature, fuel consumption, and efficiency. These alloys are some of the most complex metals used in the aerospace industry; some can consist of 14 or more alloying elements [1,2]. Many of these alloying elements have a significant effect on the microstructure created through heat treatment and, consequently, a substantial effect on the mechanical properties exhibited by the alloy. Further data is needed on the changes in microstructure each alloying element will create and whether they will have a positive or negative effect on creep or fatigue strength. Additional research on the topic will lead to improved optimization of the design requirements found in the corrosive and hostile environment inside a turbine engine.

The matrix of Ni-based superalloys is referred to as the gamma (γ) phase. The γ phase is primarily a disordered FCC structured solid solution that is mainly composed of Ni; [3–5] however, for certain alloys this may not always hold true, such as when Co becomes the primary element in a matrix. Despite their potential, Co superalloys will not be discussed in this review [6]. The γ matrix is much more ductile throughout various temperature ranges and can deform more easily than the gamma prime (γ′) precipitate phase [7].

Many Ni-based superalloys designed for high temperature applications are strengthened by incorporating inter-metallic phase precipitates with a Cu3Au or L12

1 structure dispersed throughout the matrix. The structure of these precipitates is commonly referred to as Ni3Al, since this is generally the composition, where the Ni atoms are located on the cube faces and Al atoms found on the corners [8]. The size of the precipitate, which is usually controlled through alloying chemistry and heat treatment, has an effect on the mode of precipitation hardening that will occur in the presence of stress.

In 1957, Westbrook et al. reported the first observations of the yield stress anomaly inherent in the L12, Ni3Al structured γ′ precipitate [9]. The anomaly refers to the strengthening seen with an increase in temperature, up to 800°C, of the γ′ precipitates, which is counter to what is usually seen with other metals. Thorton et al. revealed that with certain alloying elements the flow stress could be improved with the increasing temperature. For example, by alloying Nb the flow stress was shown to almost double compared to the original flow stress presented for a pure Ni3Al precipitate [10].

Figure 1.1: The effect of Temperature on the flow stress of different alloyed Ni3Al particles [10]

Figure 1.1 shows the increase in flow stress to temperatures as high as 700°-800°C.

The Kear-Wilsdorf Lock Theory was the first to try to explain this phenomenon. The theory 2 hypothesized that the observed flow stress increase was the result of different mechanisms influencing the flow stress at different temperature regimes. They stated that with an increase in temperature came a higher propensity of segments of gamma prime superpartial dislocations cross slipping onto a {010} plane. These slipped partials then become sessile by a process now termed as Kear-Wisldorf locking, which hypothesized that after cross slipping, the dislocation’s resistance towards further slip would dramatically increase. This resulted in stalled dislocations, which need a trailing APB (Anti-phase boundary) behind them to propagate further [11,12]. This later will be shown to be energetically costly. The theory was found to be inadequate, however, in fully describing the yield stress anomaly, due in part that the anomaly was observed at temperatures too low for the Kear-Wilsdorf locking to be active [13]. In 1984, Paider et al. hypothesized a new theory (PPV) to better account for the behavior. Using recently conducted atomistic studies of screw dislocations, they found that additional forces could help pin the dislocations that cross-slipped onto a

{010} plane due to the superpartial cores dissociating onto the (111) and (1-11) planes.

Again, this theory used assumptions, including the idea that dislocations are unable to deform on the (1-11) octahedral plane. However, this is most likely not true. [13,14]

Subsequent electron microscopy work measured the ratio of APB energies on both the

{010} and {111} planes and the ratio between them did not exceed 31/2 as the PPV theory had stated. It also falls short in explaining the low strain rate dependence of the yield stress

[14,15].

Kirsch later proposed, contrary to the PPV theory, that jogs on {111} planes at higher temperatures could propagate rapidly, resulting in a long screw dislocation on a

(010) plane that limits further dislocation movement [16,17]. Further modeling and

3 experimental research must be conducted to fully understand the mechanisms controlling the yield stress anomaly in Ni-based alloys.

It has already been suggested above that alloying elements used to create a Ni-based superalloy will have an effect on the mechanical behavior, alter the type and volume fraction of precipitates formed, change stacking fault energy, and either improve or limit the upper working temperature of the alloy. Ultimately, the properties of a superalloy can be manipulated by the type and quantity of alloying elements included in its chemistry.

Solid solution strengthening (SSS) is a primary strengthening technique employed for Ni-based superalloys. Adding solid solution strengtheners gives strength to the matrix

γ, thereby improving creep strength and extending rupture life. An important property influencing elements that can be used for SSS is solubility in a Ni solution. The likelihood of solubility can be estimated with a size factor calculation using the equation given below, where di is the atomic diameter of an element i and dNi is the diameter for Nickel [1,18].

100(푑푖 − 푑푁푖) 푠푖푧푒 푓푎푐푡표푟 = 푑푁푖[ ] 푑푁푖

The above equation was used to create the plot in Figure 1.2, which shows the possible alloying elements that can be used for solid solution strengthening.

Figure 1.2: The percent of deviation in atomic diameter between Ni and possible alloying elements [1,18]

The shaded area in Figure 1.2 represents +/- 15% deviation from the atomic diameter of a Ni atom. It is understood that the elements found in this region are soluble in

Ni and can be used for SSS. By improving the strength of the solution, dislocation motion is retarded in the γ matrix. Mo, W, Cr, Fe and Co are preferred solid solution strengtheners due to their high melting temperatures, solubility with Ni, and high hardening coefficients

[1,3]. Recently, Re has been found to be an excellent SSS, but its price makes Re not economical to use in large quantities. Ti, Ta, and Nb could be candidates if it were not for their penchant to partition towards the γ′ precipitates.

Precipitates are used to strengthen Ni-based superalloys by inhibiting dislocation motion through multiple parameters, including APB energy, γ/γ′ coherency, interfacial energy and differences in stacking fault energy between γ and γ′ [18–22]. As previously stated, γ′ precipitates are primarily structured Ni3Al; however, this is not the only precipitate that can be formed. For example, the addition of Ti can result in an ordered hexagonal phase η. This phase, along with most others, is usually not desired due to the

5 reduction in overall ductility and toughness it causes [18]. Not discussed in this review is the formation of the ordered Do22 γ′′ phase, which can also be used since it meets many of the same requirements as γ′.

The coherency between the γ and γ′ lattices will have an effect on the shape and type of precipitate that forms during heat treatment. In many cases, lower mismatch (<

.4%) will promote more spherical precipitates, while larger mismatches can facilitate a cubical or rod shaped γ′. It should be noted that different heat treatments will also play a role in the shape of the γ′. This mismatch can also be controlled by the addition of different alloying agents. In almost all circumstances γ will have a lower lattice parameter than γ′; therefore, the reduction of mismatch between the two lattices is done by either decreasing the lattice parameter of γ′ or increasing it for γ. Typically it is easier to add larger elements in γ; for example, Mo is excellent for improving coherency due to its relatively large radii and its tendency to partition towards γ. Though the mismatch is usually increased by substituting Al in γ′, it is an affordable compromise since the strength and stability of γ′ can be significantly improved by doing so. The substitution of Al with Nb can increase the long range order of γ′ by 25 percent. Ta, Ti and V are also potential Al substitutes in γ′

[1,18].

The addition of carbides has been found to vastly improve an alloy’s creep and fatigue life by mitigating grain boundary sliding (GBS). The most stable forms of carbides are the cubic MC and hexagonal M23C6 carbides. Ti, Ta, Nb, Zr and V all form stable MC type carbides, additions of Mo and W over 8at% will also lead to a MC type carbides [23–

25]. Hf, though not found to be soluble in Ni, has been discovered to help in the formation of carbides and has been used as such. Cr leads to an M23C6 carbide when additions of it

6 exceed 18at%. The addition of B, Mo, and W can be added as slow diffusers to prevent rapid agglomeration of carbides forming at grain boundaries, which would otherwise lead to embrittlement. The addition of Cr23C6 can lead to Cr depleted zones, which can present the possibility of corrosion damage localized in those sites [1].

Cr is important for corrosion prevention due to the formation of an oxide layer on the surface of the alloy. Concentrations of Cr above 20at% usually are needed for optimal protection from corrosion. S, which is difficult to avoid during production and later during operation in turbine engines, can form very destructive phases with Ni such as NiS2 and other compounds with different alloying elements. S is therefore a concern. It will preferably segregate towards grain boundaries; therefore, many elements are added that partition towards grain boundaries where they can “get” S, like manganese. Boron also segregates to grain boundaries and when added in excess will form borides which have been found to improve the creep strength of the alloy by preventing GBS. Table 1.1 summarizes the effects different alloying elements will cause and where they will tend to partition [1,25–27].

7 Table 1.1: Summary of different alloying elements and their effects on properties

Element Partitions to: Effects Mo γ/GB SSS, carbide, mismatch, slow diffuser , oxidation resistance, creep

Re γ SSS, creep W γ/γ′ SSS, MC carbide, reduce mismatch, slow diffuser, creep strength

Cr γ/GB SSS, reduce mismatch, Cr23C6, oxidation resistance, boride former Ti γ′ Increase γ′ strength, η formation, carbide former, boride former Ta γ′ Carbide former Nb γ′ Carbide former, increase γ′ strength, β former V γ′ Carbide former Hf GB Carbide former, prevent GB embrittlement Al γ′ γ′ former, improve oxidation resistance Co γ SSS, hot corrosion resistance C GB Carbide former B GB Boride former, slows carbide formation on GB

1.2 Deformation mechanisms

In order to better understand creep and fatigue lives of different alloys and what can be done to extend them, a more in depth comprehension of how these alloys deform is required. Both γ and γ′ have FCC crystal structures resulting in 12 independently active slip systems. Dislocations glide along these slip systems, thereby inducing plastic deformation. These dislocations can be separated into two groups: line and planar defects.

Planar defects consist of stacking faults, which are the removal (intrinsic) or addition

(extrinsic) of extra planes of atoms and microtwinning. In the most basic sense, the motion of dislocations occurs when lower energy states can be achieved by said movement. The force on a dislocation in the direction of its burgers vector was derived by Peach-Koehler.

The longer a dislocation is the higher the stress needed to move it. Below is the energy per unit length of a dislocation in terms of its burgers vector [28]. 8 퐸 = 훼퐺푏2

Where b is the burgers vector, G is the shear modulus, α is a constant and E is the energy per unit length of the dislocation [28].

Most of a Ni-based alloy’s strength comes from precipitation hardening. As a dislocation approaches the channel between two precipitate particles, it must either bend, cut through, cross-slip over them or cease propagation. The first three require greater force in order to proceed. This implies that larger amounts of force are needed for further dislocation motion and subsequent plastic deformation. When a dislocation approaches a channel width that’s relatively thin, created by small precipitates the most likely mode of precipitation hardening is caused by cutting of the precipitates through weakly coupled dislocations. As the Ni3Al particles become larger, they proceed to be cut but with strongly coupled dislocations [18,29].

Figure 1.3: A diagram a.) depicting weakly coupled dislocations cutting γ′ particles b.) depicting strongly coupled dislocations cutting γ′ particles [29]

Raynor, Silcock, Brown and Ham (RSBH) calculated the critical resolved shear stress (CRSS) τ0 as a function of the diameter of the particles, volume fraction and the APB energy needed to cut small γ′ particles. Conversely, for larger particles Huther and Reppich calculated a hyperbolic decrease for the CRSS of spherical precipitates [20]. By combining 9 these two calculations of the CRSS needed for a dislocation to cut a Ni3Al particle, a better picture of precipitation hardening versus γ′ particles size can be plotted.

Figure 1.4: The critical stress vs. particle diameter showing the peak in CRSS between weakly and strongly coupled dislocations and another between strongly coupled dislocation cutting and Orowan looping [20]

Figure 1.4 suggests that the peak strength is found right at the transition between weak and strong coupled dislocations, though another peak and possibly greater one can be found between strongly paired dislocation cutting and Orowan looping [20].

As the particles continue to increase in size so will the channel width between the particles. Eventually the width becomes large enough that the stress required to loop a particle is less than the stress needed to cut a particle. This transition occurs once a critical diameter is reached.

훾 2푇 ∆휏푐 = ( ) − 푏 푏√2/3푑

In the above equation, T refers to the line tension, d is the diameter of γ′ particles and γ is the APB energy [20]. Orowan looping causes work hardening by forming dislocation loops around the Ni3Al particles, thereby repulsing like-signed dislocations that may follow. This

10 leads to the need for ever increasing stresses for further looping and motion [18]. These theories, though helpful in the basic understanding of precipitate hardening, lack thorough explication by not incorporating all variables that must be considered for more accurate modeling; these variables include orientation dependence, better shape parameters for secondary γ′, the effects of tertiary γ′ size and volume fraction and temperature dependence.

Gamma is a disordered FCC structured phase that makes up the matrix of a Ni-based superalloy. Kear et al. hypothesized that this phase was primarily cut by a/2<110> dislocations located on {111} planes. As previously stated, greater force is required for these dislocations to cut or loop γ′ particles; therefore, dissociation into either a/3<112> super Shockley partials or a/6<112> Shockley partials occurs to circumvent the costly energetic cost of looping or perfect dislocation cutting. These partials result in stacking fault formations between them [11].

Gamma prime is an ordered L12 FCC structured phase; therefore, slip occurs on the

{111} planes. The shearing of these particles is either done by a/2<110>, a/3<112> or a/6<112> dislocations. Cutting by each dislocation will result in different faulting in the γ′ particle. Below are the three different shearing mechanisms found in γ′ precipitates with the subsequent fault found between them [30,31].

a<110>  a/2<110> + APB + a/2<110>

a<110>  a/3<112> + SISF + a/3<112>

a/2<110>  a/6<112> + CSF+ a/6<112>

An APB occurs when a layer is shifted in the Ni3Al structure, resulting in a change in nearest neighbors. In this instance, the change occurs in Al-Al and Ni-Ni bonds. This fault usually exhibits a higher energy than the other stacking faults [32]. A superlattice intrinsic

11 stacking fault (SISF) results from a change in the ABC stacking sequence, though no nearest neighbor violations occur; therefore, it is seen that SISFs are lower in energy than

APBs. Lastly, complex stacking faults (CSF) occur when the top plain of atoms are shifted by an a/6<112> shockley partial. This shift results in a change in sequence order, much like the SISF, and a change in the nearest neighbor, much like an APB. In this way, CSFs are considered to have more energy than APBs, which in turn are more energetic than

SISFs. Superlattice extrinsic stacking faults (SESFs) can also occur in γ′ when a shift by an a/3<-2-11> dislocation is followed by an a/3<-211> or an a/3<11-2> dislocation. SESFs result in a change in the stacking sequence, much like the SISF; however, in this case an extra plane of atoms has been added instead of vice versa. It was originally assumed that

SESFs had similar energies compared to SISFs, though Milligan et al. came to the later conclusion that SESFs actually had lower energy values [32]. Below is a basic conclusion comparing the energies required for different faulting in γ′.

SESF

Understanding when these different deformation mechanisms become active and, even more importantly, why they occur is imperative in better modeling for future alloy development. Many studies have reported a strong correlation between channel width and whether dissociation of a<110> or a/2<110> dislocations occur[5,7,28,33–35]. Numerous studies have also found that as channel widths decrease dissociation is much more probable. In many cases, where wider channel widths exist APB shearing was observed, whereas the same alloy with smaller channel widths revealed shearing by a/6<112>

Shockley partials. Unocic et al. concluded that channel width and stacking fault (SF) energy were the controlling parameters in the shearing of γ′ and while tertiary γ′ played an

12 important role it was not one of the controlling parameters [7]. Yuan et al., [36] in their study of the transition of deformation activity in U720Li, found that SF energy played a large role while tertiary γ′ volume fraction was the controlling mechanism. A model was put forward by them to better describe this transition from APB shearing to SF shearing.

The increase in surface energy resulting from APB shearing (ΔE) was found to be related to the APB energy in γ′ multiplied by the addition of secondary and tertiary γ′. The increase in surface energy for stacking fault formation was the stacking fault energy multiplied by the addition of secondary and tertiary γ′ volume fraction, plus the stacking fault energy in

γ times the volume fraction of γ.

∆퐸퐴푃퐵 = 훾퐴푃퐵(푓훾′퐼퐼퐼 + 휃푓훾′퐼퐼)

푝 푚 ∆퐸푆퐹 = 훾푠푓(푓훾′퐼퐼퐼 + 휃푓훾′퐼퐼) + 훾푠푓푓푚

It was assumed in these two relationships that ΔE is linear for APB formation since the volume fraction of the secondary particles was considered to be a constant. With the volume fraction of secondary γ′ considered constant, the variable in both equations became the volume fraction of the tertiary γ′. This led to a critical volume fraction of tertiary γ′ where the shearing mechanism of γ′ changes from APB shearing to SF shearing when this critical value was approached. Another assumption is the constant channel width or lack of consideration towards a channel width change, which assuredly would cause a change in the active deformation mechanism since it cannot be assumed that channel width and tertiary γ′ volume fractions are linearly related [29].

Figure 1.5: Tertiary volume fraction vs. ΔE showing the change in shearing mechanisms when a critical volume fraction is reached. [29]

Figure 1.5 above shows the two relationships of surface energy for APB shearing and SFE shearing. Below the critical tertiary volume fraction APB shearing is prevalent, while above it SFE shearing is prevalent. Multiple other papers have pointed to an important dependence of tertiary volume fraction in controlling the type of active deformation mechanism [5,7,29,34,37–40].

Stress and temperature have also been shown to be crucial parameters affecting which type of deformation mechanism is active. An increase in temperature or stress will result in an increase of available energy acting on dislocations, allowing more energetic motion to occur. Preuss et al. investigated the temperature effects on precipitation hardening mechanisms and found a particle size/temperature dependence did exist. For small particle sizes, (80nm) cutting was found to be the dominant mechanism in all temperature ranges (room Temperature, 250°C, 750°C). Whereas at 250°C, either Orowan looping or strong coupled dislocation cutting became the dominate mechanism for coarse sized γ′ particles (250nm). By 750°C both the alloys containing coarse and medium sized particles experienced Orowan looping. Despite the advancements in understanding, still more needs to be investigated on the empirical relationship between temperature and 14 precipitation hardening. For example, the authors argued that as dislocations became locked by the yield strength anomaly found in γ′ that dislocation would then favor Orowan looping. This is an inadequate reason, however, considering alloys can experience yield stress anomaly commonly without Orowan looping ever appearing as a mechanism [22].

As has already been stated, deformation mechanisms have shown a large dependence on temperature. Kolbe performed in-situ TEM tests in elevated temperatures and found a transition occurring around 760°C. Prior to this temperature, a/2<110> dislocations were the primary mechanism and their motion was observed to be “jerky”, meaning there was significant waiting periods when dislocations encountered γ′ particles.

Beyond this temperature, long faults that propagated through entire grains were observed.

These were most likely microtwins. At intermediate to high temperatures in both creep and fatigue, microtwinning has been observed [41]. The precursor and mechanisms needed to form microtwinning have long been unknown and elusive to the science community, though recent theories have been proposed. Early observations have shown that SESFs appeared in regions where eventual microtwinning would later be observed [41]. It was therefore concluded that SESF’s were in some way precursors to microtwinning. Knowles and Chen believed that microtwinning formation occurred through the passage of a/3<112>

SESF’s in the γ′ precipitates. It was later shown by Kolbe that twinning most likely was formed through the passage of paired a/6<112> Shockley partials in the gamma prime on adjacent {111} planes [42]. These partials would each form a complex stacking fault, creating a two layer CSF that would be undesirable energetically. Though no evidence was presented, they proposed that thermally activated atomic reordering would occur preventing this energetically costly fault and forming a microtwin instead.

15 Later, Kovarik et al. used high resolution TEM imaging to prove that the partials were in fact a/6<112> Shockley partials and not the a/3<112> super-shockley partials proposed by Knowles and Chen, and that Kolbe was correct in stating that the Shockley partials were propagating in pairs on adjacent {111} planes [37,38].

Figure 1.6: High Res HAADF Image used for the analysis of microtwin formation. The lower area is a 14 layer microtwin while above is a portion of a γ′ particle. [37]

They further suggested that the process of diffusion-mediated reordering was energetically possible according to simulations that revealed a three step process resulting in the exchange of NiAl and AlNi anti-sites. Lastly, another observed deformation process seen at higher temperatures, especially in creep, is climb by-pass. This deformation is strongly temperature dependent and occurs mainly in high temperatures [43–45]. When enough stress and temperature has been added internally to a dislocation, it achieves the ability to cross slip onto other adjacent planes and over a γ′ particle instead of looping or cutting it. Many alloys fail in creep through this deformation mechanism.

1.2.1 Orientation Dependence on Deformation Mechanisms

Douin et al. created a model predicting the behavior of a/2<110> and a/6<112> dislocations when they came into contact with γ′ precipitates. They found a significant orientation dependence in the behavior of dislocations when the applied stresses were originating from different orientations. Stresses oriented between [132] and [-112] 16 directions were found to be far more likely to cause dissociation where the trailing partial would not be able to propagate through the channel forming large stacking faults. Stresses oriented near a [121] direction would allow both partials to propagate through though at different velocities [40].

Figure 1.7: orientation effect on dissociation of a/2<110> dislocations in Ni-based superalloys. In region a.) the trailing partial cannot propagate through the channel and large stacking faults are created. In b.) both partials are able to propagate though with differing velocities. [39]

1.3 Mechanical Behavior

As previously noted, there are many different parameters that can control the creep and fatigue properties of Ni-based superalloys. Microstructure, alloying elements, temperature and orientation anisotropy must all be considered. This section’s goal is to examine how these parameters affect creep properties, as well as what work remains to gain a deeper understanding of the relationship between each variable and the alloys’ mechanical properties.

17 1.3.1 Creep

In polycrystalline alloys, creep can typically be broken down into three distinct regimes: primary, secondary and tertiary creep. In primary creep, a rapid initial transient occurs due to rapid dislocation creation. In secondary creep, a steady state creep strain is reached. This can be explained by new dislocation creation and dislocation annihilation densities that are equable with each other. Lastly, tertiary creep is the final rapid increase in strain as the alloy approaches failure. Many polycrystalline and pure metals while in secondary creep follow what is termed “the creep power law”, which relates creep strain rate to stress and temperature levels [18,46–48].

−푄 ∆휀 = 퐾휎푛exp ( ) 푅푇

Where K is a constant that has shown dependence to stacking fault energy γsf, Q is the activation energy, which is usually found to be close to the activation energy for self- diffusion in nickel and n the stress exponent. At moderate stresses n is equal to 4.5-5, which comprises what is known as the power law region. The value of n has been found to be dependent on time (increasing with time), temperature, and on the type of deformation mechanism that is active [18]. In tertiary creep, the microstructure has usually formed a crack or, in the case of Ni-based alloys, rafting of the γ′ precipitates has formed. Rafting occurs when precipitates diffuse into long, connected lines perpendicular to the tensile axis, which is in contrast to the normally separate, spherical or cuboidal γ′ particles that exist before creep is initiated [44,49–52].

In fact, the size of the γ′ precipitates has been found to play a major role in the creep strength of Ni-based superalloys. Bhowal et al. found that in the superalloy Rene 95 the 18 microstructure consisting of smaller and finer γ′ performed significantly better than the alloy containing larger γ′. It was not clear in their research whether this was due to the size of the γ′ or the change in channel width associated with the different γ′ sizes. A change in deformation mechanism was found to occur when the average channel width reached

.05μm. Below this value, shearing and the formation of stacking and isolated faults were the prevalent deformation mechanisms, whereas Orowan looping was prevalent above

.05μm. Numerous studies have also examined the role of the smaller tertiary γ′ in creep performance and consistently found that tertiary γ′ was in fact beneficial to the creep performance of Ni-based superalloys. Once again, a change in active deformation mechanism was observed with the absence and then presence of tertiary γ′ [53]. Lastly, grain boundary shape also contributed to creep strength. As previously stated, carbides are included to form barriers on grain boundaries to prevent them from sliding. Furthermore, serrated grain boundaries performed better in creep by slowing grain boundary sliding, as well. It has also been shown that larger grains improved the creep resistance in polycrystalline Ni-based superalloys [49,52,54,55].

In single crystal creep, the stress exponent is usually larger (> 8) and the activation energy derived from creep power law is frequently 4-5 times greater than the activation energy for self-diffusion of nickel. This suggests that a breakdown in the creep power law occurs. In fact, the three regimes of creep outlined above do not exist in most cases of single crystal creep. Single crystal creep strain is typically shown to progressively increase with time, never acquiring a steady-state creep strain [56].

19 a b

Figure 1.8: a.) A graph of creep strain for a polycrystalline Ni-based alloy. Note the three distinct regions of strain rate. B.) A graph of creep strain for a single crystalline Ni-based alloy [65]

Many studies on single crystals were conducted to better understand the anisotropy of its strength for different crystalline orientations. Several authors have found that in tensile creep (usually for CMSX-4) the [100] oriented crystals performed well in creep and the

[110] and [111] oriented specimens performed poorly. This is consistent with [100] having the most active slip systems, allowing for the restriction of dislocation motion and, subsequently, better creep life. The [110] oriented crystals had fewer active slip systems and, therefore dislocation motion was less restricted due to the absence of locking mechanisms; this ultimately resulted in poor creep strength. Finally, [111] oriented crystals were found to have the least amount of active slip systems and performed the worst.

Various deformation mechanisms were also found active in the [111] oriented crystals, including cube slip [45,57–59]. The lack of active slip planes has also been found to promote micro-twinning and poor creep strength [60].

20 [256]

[100]

Figure 1.9: Creep strains for a [256] and [100] oriented single crystal [69].

1.3.2 Tension/Compression Asymmetry in Creep

Another property of single crystals, which until recently had not been examined thoroughly, is the asymmetry found between tensile and compressive testing. Kakahi et al. found significant differences in creep strength between tensile and compressive creep that were conducted on the same orientation. In many cases orientations that performed well in tensile tests performed poorly in compression. For example, compression tests conducted on [100] oriented CMSX-4 samples performed rather poorly, where earlier it was shown that tensile tests on [100] oriented crystals performed extremely well [61]. Knowles et al. found only a single slip plane active <112>{111} in compression tests. This led to mechanical twinning formation and to a rapid strain rate increase with time. In comparison,

[100] tensile creep was found to deform through multiple slip systems, with the main active deformation mechanism being dislocation motion in the γ channels. The opposite occurred for testing done on [110] oriented crystals. Poor creep strength and mechanical twinning which favored the {111} <-211> slip system were seen. In compression testing, the [110] oriented crystal had multiple active slip systems, which led to strengthening and improved creep strength. The creep strain on tensile and compression creep tests for both orientations discussed above can be observed in Figure 1.10 below [59][28]. 21

Figure 1.10: a.) Compression and tension creep strain curves for [001] oriented single crystal b.) Compression and tension creep strain curves for [011] oriented crystals [68]

It can be concluded that the type of active deformation mechanism has a large influence on the asymmetry seen in Ni-based superalloys. Kakehi et al. recently tested

PWA 1480 single crystals with increased amounts of Tantalum and were able to significantly reduce the asymmetry previously present in creep strength. The introduction of higher amounts of Tantalum led to the elimination of {111}<112> slip as active slip systems and, consequently, the elimination of mechanical twinning, which was seen to be the cause of the asymmetry earlier. This finding was only for creep performance, an asymmetry was still seen for yield stresses [61].

1.4 Motivation for Study

The need to improve the high temperature properties of Ni-based superalloys is more important than ever, with a focus on producing future jet turbine engines that emit less CO2 by increasing the turbine’s operating temperature. Throughout the past one hundred years, significant advancements on superalloys high temperature properties have been made, with precipitation and solid solution strengthening occurring in the 1940-50’s and grain boundary strengthening in the 1960’s [18]. However, other than new processing 22 techniques such as powder metallurgy, directional solidification, and single crystal casts, no new fundamental strengthening mechanisms have been discovered. This means that the improvements made to each new generation of disk superalloy are incremental, utilizing and building upon the techniques discovered decades ago. Yet, it is apparent that present day research is far from the comprehensive level necessary to fully understand the many parameters affecting the performance of these alloys. As next generation alloys reach increasingly higher operating temperatures, improved knowledge of how alloying chemistry affects the mechanical properties and microstructure will be essential. Presently, literature is lacking in substantial data on the effects different alloying chemistries will have on the microstructure and deformation modes, primarily at the atomic scale.

Fortunately, recent advances in high resolution characterization techniques such as, atomic resolution energy dispersive x-ray spectroscopy (EDX) and scanning transmission electron microscopy (STEM) combined with new, more powerful atomistic simulation and modeling will allow for unprecedented insights into the atomic scale mechanisms responsible for the macro-scale creep deformation in these alloys. [1,26].

In addition, improved comprehension of how varying stress, temperature and orientation regimes result in different active deformation modes, and more importantly why these modes become active, will lead to more advanced models that can accurately predict the performance of next generation alloys. These new models have the potential to save the airline industry millions of dollars by eliminating unnecessary mechanical testing and accelerating future alloy development.

Again, using the new, powerful characterization techniques mentioned above to examine the deformation from crept Ni-based superalloys in different stress and

23 temperature regimes, such as the currently in service alloy ME3, should provide new discoveries and insights yet unknown. This study represents a promising area of research that can produce significant discoveries in the near future and where the possibility of major advancements can still be made.

References

[1] A.K. Jena, Review The role of Alloying Elements in the Design of Nickel-base superalloys, Journal of Materials Science. 19 (1984) 3121–3139.

[2] E.S. Huron, K.R. Bain, D.P. Mourer, T. Gabb, Development of High Temperature Capability P/M Disk Superalloys, Superalloys 2008 (Eleventh International Symposium). (2008) 181–189.

[3] G.P. Sabol, R. Stickler, Microstructure of Nickel-Based Superalloys, Phys. Stat. Sol. 35 (1969) 11–52.

[4] T.P. Gabb, J. Gayda, J. Telesman, A. Garg, N. Glenn, B. Rd, The Effects of Heat Treatment and Microstructure Variations on Disk, Superalloys 2008 (Eleventh International Symposium). (2008) 121–130.

[5] H. Z. Deutchman, P.J. Phillips, N. Zhou, M.K. Samal, S. Ghosh, Y. Wang, M.J. Mills, Deformation mechanisms coupled with phase field and crystal plasticity modeling in a high temperature polycrystalline ni-based superalloy, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 25–33.

[6] M. Heilmiaer, M. Kruger, Metallic Materials for Structural Applications Beyond Nickel-Based Superalloys, JOM. 61 (2009) 61–67.

[7] R.R. Unocic, L. Kovarik, C. Shen, P.M. Sarosi, Y. Wang, J. Li, et al., Deformation Mechanisms in Ni-Base Disk Superalloys at Higher Temperatures, Superalloys 2008 (Eleventh International Symposium). (2008) 377–385.

[8] E. J. Payton, P. J. Phillips, M.J. Mills, Semi-automated characterization of the gamma prime phase in Ni-based superalloys via high-resolution backscatter imaging, Material Science and Engineering. 527 (2010) 2684–2692.

[9] J. Westbrook, In Intermetallic Compounds: Principals and Practice, Transactions of the Metallurgical Society of AIME. 209 (1957).

24 [10] P.H. Thornton, R.G. Davies, T.L. Johnston, The Temperature Dependence of the Flow Stress of the gamma prime Phase Based upon Ni3Al, Metallurgical Transactions. 1 (1970).

[11] B.H. Kear, J.M. Oblak, A.F. Giamei, Stacking faults in gamma prime Ni3(AI,Ti) precipitation hardened nickel-base alloys, Metallurgical Transactions. I (1970) 2477–2486.

[12] P.P. Swu-Jian Liang, The Yield Stress of L12 Ordered Alloys, Acta Materialia. 25 (1976) 485–493.

[13] D.P. Pope, V. Vitek, A Theory of the Anomalous Yield Behavior in L12 Ordered Alloys, Acta Materialia. 32 (1984) 435–448.

[14] A.H.W. Ngan, M. Wen, C.H. Woo, Atomistic simulations of Paidar–Pope–Vitek lock formation in Ni3Al, Computational Materials Science. 29 (2004) 259–269.

[15] A.T. Paxton, Y. Q. Sun, The role of planar fault energy in the yield anomaly in L12 intermetallics, Philosophical Magazine A. 78 (1998) 85–103.

[16] P.B. Hirsch, A New Theory of the Anomalous Yield Stress in L12 Alloys, Philosophical Magazine A. 65 (1992) 569–612.

[17] P.B. Hirsch, A model of the anomalous yield stress for (111) slip in L12 alloys, Progress in Materials Science. 36 (1992) 63–88.

[18] R. Reed, The Superalloys, Cambridge University Press, Cambridge, UK, 2006.

[19] J.J. Schirra, P.L.R. Pratt, E. Hartford, E.S. Huron, K.R. Bain, D.P. Mourer, et al., Effect of microstructure (and heat treatment) on the 649C properties of advanced P/M superalloy disk materials, Superalloys 2004 (Tenth International Symposium). (2004) 341–350.

[20] B. Reppich, Some New Aspects Concerning Particle hardening Mechanisms in Gamma Prime Precipitating Ni-Base Alloys - I. Theortical Concept, Acta Metallurgica. 30 (1982) 87–94.

[21] D.P. Pope, S.S. Ezz, Mechanical properties of Ni 3 AI and nickel-base alloys with high volume fraction of Gamma Prime, International Metals Reviews. 29 (1984) 136–167.

[22] M.D. M. Preuss, J. Q. Fonseca, B. Grant, E. Knoche, R. Moat, The effect of gamma prime particle size on the deformation mechanism in an advance polycrystalline nickel-base superalloy, Superalloys 2008 (Eleventh International Symposium). (2008) 405–414.

25 [23] T.J. Garosshen, G.R. Mccarthy, Low Temperature Carbide Precipitation in a Nickel Base Superalloy, Metallurgical Transactions A. 16 (1985) 1213–1223.

[24] G. Lvov, V.I. Levit, M.J. Kaufman, Mechanism of Primary MC Carbide Decomposition in Ni-Base Superalloys, Metallurgical Transactions A. 35 (2004) 1669–1679.

[25] T.J. Garosshen, Influence of Alloy Chemistry on Carbide Precipitation in a Nickel Base superalloy, Metallurgical Transactions A. 17 (1986) 2075–2077.

[26] T.J. Garosshen, T.D. Tillman, G.P. Mccarthy, Effects of B , C , and Zr on the Structure and Properties of a PM Nickel Base Superalloy, Metallurgical Transactions A. 18 (1987) 69–77.

[27] D. Blavette, E. Cadel, C. Pareige, B. Deconihout, P. Caron, Phase transformation and segregation to lattice defects in Ni-base superalloys., Microscopy and Microanalysis. 13 (2007) 464–83.

[28] D. Hull, D.J. Bacon, Introduction to Dislocations, 4th ed., Elsevier Ltd., Oxford, UK, 2007.

[29] E.S. Huron, R.C. Reed, M.C. Hardy, M.J. Mills, R.E. Montero, P.D. Portella, et al., Controlling the deformation mechanism in disk superalloys at low and intermediate temperatures, Superalloys 2012. (2012) 35–42.

[30] A. Korner, Weak-beam study on superlattice dislocations in a Ni 3 Al alloy in the temperature range of the flow stress anomaly, Philosophical Magazine A. 59 (1989) 1–7.

[31] D. Lin, M.A.O. Wen, Dislocation Structure Due to High Temperature Deformation in the gamma prime Phase of a Nickel-based Superalloy, Materials Science and Engineering A. 3 (1989) 207–214.

[32] W.W. Milligan, S.D. Antolovich, The mechanisms and temperature dependence of superlattice stacking fault formation in the single-crystal superalloy PWA 1480, Metallurgical Transactions A. 22 (1991) 2309–2318.

[33] B. Décamps, S. Raujol, A. Coujou, On the shearing mechanism of γ ′ precipitates by a single ( a / 6 )⟨ 112 ⟩ Shockley partial in Ni-based superalloys, Philosophical Magazine. 84 (2004) 91–107.

[34] S.I. Rao, T.A. Parthasarathy, D.M. Dimiduk, P.M. Hazzledine, Discrete dislocation simulations of precipitation hardening in superalloys, Philosophical Magazine. 84 (2004) 3195–3215.

26 [35] S. Raujol, M. Benyoucef, D. Locq, P. Caron, F. Pettinari, N. Clement, et al., Decorrelated movements of Shockley partial dislocations in the γ-phase channels of nickel-based superalloys at intermediate temperature, Philosophical Magazine. 86 (2006) 1189–1200.

[36] Y. Yuan, Y.F. Gu, T. Osada, Z.H. Zhong, T. Yokokawa, H. Harada, A new method to strengthen turbine disc superalloys at service temperatures, Scripta Materialia. 66 (2012) 884–889.

[37] L. Kovarik, R.R. Unocic, J. Li, P. Sarosi, C. Shen, Y. Wang, et al., Microtwinning and other shearing mechanisms at intermediate temperatures in Ni-based superalloys, Progress in Materials Science. 54 (2009) 839–873.

[38] L. Kovarik, R.R. Unocic, J. Li, M.J. Mills, The intermediate temperature deformation of ni-based superalloys : importance of reordering, JOM. 61 (2009) 42– 48.

[39] P. Villechaise, J. Cormier, T. Billot, J. Mendez, Mechanical Behavior and Damage Processes of Udimet 720Li: Influence of Localized Plasticity at Grain Boundaries, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 15–24.

[40] J. Douin, F. Pettinari-Sturmel, a. Coujou, Dissociated dislocations in confined plasticity, Acta Materialia. 55 (2007) 6453–6458.

[41] M. Kolbe, The High Temperature Decrease of the Critical Resolved Shear Stress in Nickel-Based Superalloys, Materials Science and Engineering A. (2001).

[42] D.M. Knowles, Q.Z. Chen, Superlattice stacking fault formation and twinning during creep in gamma/gamma prime single crystal superalloy CMSX-4, Materials Science and Engineering A. 340 (2003) 88–102.

[43] R.R. Unocic, G.B. Viswanathan, P.M. Sarosi, S. Karthikeyan, J. Li, M.J. Mills, Mechanisms of creep deformation in polycrystalline Ni-base disk superalloys, Materials Science and Engineering: A. 483-484 (2008) 25–32.

[44] Y. Tsukada, Y. Murata, T. Koyama, N. Miura, Y. Kondo, Coupled Analysis of Microstructure Evolution with Creep Deformation in Nickel-based Superalloys by the Phase-field Method, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 111–119.

[45] A. Soula, Y. Renollet, D. Boivin, J.L. Pouchou, D. Locq, P. Caron, et al., Grain Boundary and Intragranular Deformations During High Temperature Creep of a PM Nickel-Based Superalloy, Superalloys 2008 (Eleventh International Symposium). (2008) 387–394.

27 [46] A. Ahmadieh, A.K. Mukhergee, Stress--Temperature--Time Correlation for High Temperature Creep Curves, Material Science and Engineering. 21 (1975) 115–124.

[47] W.D. Jenkins, T.G. Digges, C.R. Johnson, Creep of High-Purity Nickel, Journal of Research of the National Bureau of Standards. 53 (1954) 329.

[48] F.R.N. Nabarro, Do we have an acceptable model of power-law creep?, Materials Science and Engineering: A. 387-389 (2004) 659–664.

[49] E. Vacchieri, A. Costa, Comparison of the mechanical behavior and evaluation of different damage mechanismsin an equiaxed and a single crystal superalloys subjected to creep, LCF and TMF, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 235–244.

[50] R. Giraud, J. Cormier, Z. Hervier, D. Bertheau, K. Harris, J. Wahl, et al., Effect of the Prior Microstructure Degradation of the High Temperature/ Low Dtress Non- Isothermal Creep Behavior of CMSX-4 Ni-based Single Crystal Superally, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 265–274.

[51] V.A. Vorontsov, L. Kovarik, M.J. Mills, C.M.F. Rae, High-resolution electron microscopy of dislocation ribbons in a CMSX-4 superalloy single crystal, Acta Materialia. 60 (2012) 4866–4878.

[52] L. Lou D. Wang, C. Liu, J. Zhang, Evolution of Grain Boundary Precipitates in a Directionlly Solidifed Ni-base Superally during High Temperature Creep, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 363–368.

[53] P.R. Bhowal, E.F. Wright, E.L. Raymond, Effects of cooling rate and morphology on creep and stress-rupture properties of a powder metallurgy superalloy, Metallurgical Transactions A. 21 (1990) 1709–1717.

[54] H.L. Danflou, M. Marty, A. Walder, Formation of serrated grain boundaries and their effect on the mechanical properties in a P/M nickel base superalloy, Superalloys 1992. (1992) 63–72.

[55] M. Yoshiba, O. Miyagawa, Effect Cyclic in Hot of Grain Creep Boundary Strengths Serration on the Fatigue , Creep in Air and and of a Nickel-base Superalloy Corrosive Environment, Transactions. 26 (1986).

[56] C. Carry, J.L. Strudel, Apparent and Effective Creep Parameters in Single Crystals of a Nickel Base Superally Incubation Period, Acta Materialia. 25 (1976) 767–777.

[57] V. Sass, U. Glatzel, M. Feller-Kniepmeier, Anisotropic creep properties of the nickel-base superalloy CMSX-4, Acta Materialia. 44 (1996) 1967–1977.

28 [58] F.H. Latief, K. Kakehi, H. Murakami, K. Kasai, Influence of crystallographic orientation on creep behavior of aluminized Ni-based single crystal superalloys, Superalloys 2012 (12th International Symposium on Superalloys). (2012) 311–320.

[59] D.M. Knowles, S. Gunturi, The role of 〈112〉{111} slip in the asymmetric nature of creep of single crystal superalloy CMSX-4, Materials Science and Engineering: A. 328 (2002) 223–237.

[60] K. Kakehi, Effect of primary and secondary precipitates on creep strength of Ni- base superalloy single crystals, Materials Science and Engineering: A. 278 (2000) 135–141.

[61] K. Kakehi, Tension/compression asymmetry in creep behavior of a Ni-based superalloy, Scripta Materialia. 41 (1999) 461–465.

Chapter 2 – Creep Deformation Mechanism Mapping in Nickel Base Disk Superalloys

This chapter is a modified version of publication [6]

2.1 Introduction

Ni-base disk superalloys are an important class of high temperature structural materials that are primarily utilized for critical hot section engine components in aircraft gas turbine engines. For these applications, Ni-base superalloys are the materials of choice due to their high strength and excellent resistance to creep, fatigue, and oxidation at high temperatures [1,2]. The high temperature mechanical properties are attained through precipitation hardening of an ordered, L12 structured, Ni3Al based intermetallic  precipitate that is coherently embedded in a disordered, face centered cubic (FCC), solid solution  matrix[3,4]. The  precipitates impart resistance to creep and fatigue deformation by the formation of high energy faults if the precipitates were sheared by various dislocation types. Examples of these structural and chemical defects include antiphase boundaries (APB), complex stacking faults (CSF), superlattice intrinsic stacking faults (SISF), and superlattice extrinsic stacking faults (SESF) [5–8]. The latter three all have been found to occur in the intermediate temperature range between 650C and 815C.

Another important deformation mechanism observed in this temperature range is de-correlation of full dislocations in the  channels, where full dislocations separate into their constituent Shockley partials. Raujol et al.[9] found de-correlation active during creep at 700C. Their analysis found that the de-correlated Shockley partials were able to shear

30 both the  and  phases. Considering forces on both the shearing and trailing Shockley partials, Raujol et al. [9] found that channel width was an important parameter that determined whether a dislocation would de-correlate. Chen, et al. [5] further extended these concepts in creating a “dislocation activity diagram” in which transitions between these major deformation modes were established as a function of stress and orientation of the applied shear stress.

Much of the detailed mechanistic understanding and interpretation of rate controlling, time-dependent creep deformation in Ni-base superalloys have been determined directly from in-situ and/or post-test transmission electron microscopy (TEM) observations of different deformation modes. For a given alloy composition, changes in precipitate volume fraction (VF), size and channel width will all have an effect on the type of active deformation modes observed in certain stress and temperature regimes [10–12].

Therefore, the purpose of this study is to perform an in depth TEM/STEM characterization study and determine the mechanisms responsible for creep deformation in a modern, powder metallurgy (PM) turbine disk alloy as a function of stress and temperature. In presenting the results for the polycrystalline samples, it should be noted that some variation in deformation processes were typically observed from one grain to the next, and between grain boundary regions and grain interiors. Thus, some judgment must be exercised in order to establish a predominant deformation mode for a given condition. The results of single crystal testing are also informative since variation in applied stress state is eliminated, and orientation effects can be assessed more directly.

31 2.2 Materials and Experimental Methods

The Ni-base superalloy ME3 is a newer generation powder metallurgy (PM) alloy that was developed specifically for use as a turbine disk in the high-pressure region of aircraft engines. Initial alloy design and development commenced in the mid 1990’s as part of NASA’s High Speed Research/Enabling Propulsion Materials (HSR/EPM) program with GE Aircraft Engines and Pratt & Whitney [13,14]. The goal of the HSR/EPM alloy development program was to design an advanced turbine disk alloy that can be utilized in future supersonic aircraft gas turbine engine applications with the capability of extended service durability at temperatures of up to 650°C or replace turbine disk material which it is expected to outperform such as IN100 or René 88DT. Figure 2.1 shows the typical pre- test microstructure in ME3 [15].

Figure 2.1: A representative SEM micrograph of the bi-modal microstructure in polycrystalline and single crystal ME3

32 Polycrystalline creep specimens used in this study were extracted from different locations in a forged turbine disk where each underwent slightly different cooling rates.

Each specimen had an average ASTM grain size of 6.5. From validated finite element modeling (FEM) of the disk forging thermal heat treatments, the range of initial cooling rates experienced from these creep specimens were from 60-82ºC/min (see Table 2.1). Due to the narrow range in cooling rates, the starting microstructure (i.e.  precipitate morphology and grain size) for each of the creep experiments were nearly identical as shown below in Table 2.1.

In addition to the polycrystal creep specimens, a single crystal of ME3 was obtained from the GE Global research center. The casting had an initial level of chemical segregation greater than the polycrystalline powder metallurgy samples, and thus a long term solution heat treatment was conducted to minimize this segregation. The crystal was then given a standard heat treatment in order to yield a precipitate structure similar to the polycrystalline materials as indicated below.

Table 2.1: Secondary and tertiary  precipitate volume fraction and size (in nm) between the polycrystalline and single crystal ME3 alloys

Sample Secondary  Secondary  Size Tertiary  Tertiary  Size VF (nm) VF (nm) Polycrystal ME3 48% 321 2.7% 26 Single Crystal 47% 350 3.5% 34 ME3

2.2.1 Creep Tests

The polycrystalline samples were provided by GE Aviation (Evendale, OH) and the polycrystalline creep testing was performed by MetCut Research Inc. (Cincinnati, OH).

After a predetermined level of strain had been accumulated, testing was interrupted in order

33 to evaluate the deformation substructure using TEM/STEM characterization. The details

of the creep tests along with some microstructure and processing information is presented

in Table 2.2 below.

Table 2.2: The four different creep test set-ups examined in this study

Sample Cooling Rate (C/min) Temperature (C) Stress (MPa) Total Strain (%) I 62 677 690 2.0 II 86 704 724 0.4 III 60 760 345 0.36 IV 45 815 345 0.42

The orientation of the ME3 single crystal casting was determined using Orientation

Imaging Microscopy (OIM). Rectangular prism specimens were cut with a 1:1:2.5

dimensional ratio in two orientations, [001] and [110], using electrical discharge machining

(EDM). These orientations have previously shown strong anisotropic creep properties at

intermediate temperatures and the presence of multiple, active slip systems allows for

insights into the rate-limiting processes for different shearing events [16–18]. To remove

the damage layer created by the EDM process, the sides of each sample were polished

down to a 1200 fine grit finish.

For the single crystal creep tests, three different temperature regimes were

explored: 700C, 760C, and 815C. The stresses at each temperature were 552MPa, 552

and 413MPa, and 276MPa respectively. Each test was ended once about 0.5% plastic strain

was reached. MTS 810 compression cages with attached linear variable displacement

transducers (LVDTs) that recorded the displacement of the compression plattens were used

to conduct each compression creep test.

34 2.2.2 Microstructural Characterization

Scanning electron microscopy (SEM) was used to characterize the  precipitate size and area fraction prior to creep. For this, samples were prepared from the grip ends of the creep specimens. An etchant consisting of a mixture of 50ml lactic, 30ml nitric, and

2ml hydrofluoric acid was applied to selectively etch the  phase and then imaged in an

FEI Sirion SEM. The imaged microstructures were then quantified using Adobe Photoshop

CS and Fovea Pro 3.0 plug-in or ImageJ [19]. Post-creep characterization was employed using TEM foils prepared from the gauge length by sectioning the specimens 45° with respect to the tensile axis in order to view deformation activity along crystallographic planes that had experienced maximum shear stresses. After sectioning, the foils were polished to a thickness of 100m, slurry drilled, then jet polished using an electrolyte consisting of 10% HClO4 and 90% Methanol at -45°C/15V or an electrolyte consisting of

5% HClO4, 35% 2-n butoxyethanol and 60% Methanol at -45ºC/15V. Focus Ion Beam

(FIB) foils from post mortem single crystal samples were extracted normal to the compression axis using an FEI Helios Nanolab Dualbeam 600 Focused ion beam microscope. By extracting the foils normal to the compression axis one last check could be made to ensure the sample was indeed cut to the right orientation. TEM analysis was then conducted using a Philips CM200 TEM, while the STEM analysis was done on a FEI TF20

Tecnai STEM and a FEI Titan3 80-300KV STEM.

35 2.3 Results

2.3.1 Creep Behavior

2.3.1.1 Polycrystalline Tests

For the polycrystalline creep study, the stress and temperature regimes mentioned in Table 2.2 were explored. These test temperatures between 677-815ºC are above the expected operating temperatures for present turbine disks, but explores a regime of possible relevance for next-generation engines. The results of the creep tests are presented in Figure

2.2 in plots of plastic strain versus time.

Figure 2.2: (a) Creep curves for the following four conditions: 677C/690MPa (red curve), 704C/724MPa (blue curve), 760C/345MPa (green curve), and 815C/345MPa (black curve). In (b), the same creep curves magnified at lower strains.

Of course, the time required to reach the different levels of creep strain depends on the temperature and stress to which each specimen was subjected. For example, when

Specimen I was crept at 677ºC at 690MPa, it took nearly 3750 hours to reach 2.0% creep

36 strain. By raising the test temperature and stress slightly, as in the case of Specimen II

(704ºC and 724MPa), deformation occurred much more rapidly and reached 0.4% strain in only 160 hours. This indicates a strong stress and temperature dependence on creep deformation within this intermediate temperature range. A similar comparison can be made by examining the creep response of Specimen III and IV which was crept to the same stress but at different temperatures, 760ºC and 815ºC, respectively. Creep at the higher temperature resulted in a much faster strain rate. In fact, Deutchman [20] conducted transient stress tests on polycrystalline ME3 samples at 700C with a comparable microstructure and found stress exponents near 15, supporting the strong stress dependence for creep behavior indicated above.

2.3.1.2 Single Crystal Tests

Below are the six single crystal compression creep tests conducted for this study.

Figure 2.3: The [001] and [110] compression creep tests at (a) 700C and 552MPa, (b) 760C and 414MPa, (c) 760C and 552Mpa, and (d) 815C and 276MPa

A prominent anisotropy in creep strength, especially at the lower temperatures, is observed between the [001] and [110] orientations. The [110] oriented samples all revealed strain hardening as the time progressed, whereas the [001] oriented samples exhibited strain softening at the lower temperatures. However, in the higher stress test (552MPa) at 760C, the behavior transitioned such that the [001] orientation performed better than the [110] orientation, and both exhibited a normal primary strain transient. It should be noted however that much faster strain rates were observed at this stress and temperature regime

38 (5x10-7 s-1 compared to about 1x10-9 s-1 for the other compression creep tests) and that different deformation modes may be active in these tests. Interestingly, at the highest test temperature (815C), the anisotropy has become minimal, though the [001] oriented sample still was inferior to the [110] sample. This observed anisotropy implies differences in active deformation modes and strong orientation effects in the intermediate temperature range.

Transient stress experiments were also performed at all three test temperatures for both orientations in which the stress was increased when about 0.5% plastic strain was reached.

The results indicate a strong stress dependence for each case, as shown in Table 2.3. Since the load-shedding between grains of a polycrystal might strongly affect the transient response, these values for single crystals should be more directly related to the intragranular deformation processes active within grains. Once again, large stress exponents are observed, that generally decrease with temperature. However, even the values at 815°C are significantly larger than might be expected for a classical, climb-bypass mechanism.

Table 2.3: Calculated stress exponents for the [001] and [110] orientations at 700C, 760C, and 815C. Orientation Stress Exponent at Stress Exponent at Stress Exponent at 700C 760C 815C [001] 26.4 14.8 9.9 [110] 23.0 5.5 9.0

2.3.2 Microstructural Evolution

After each creep test the microstructures were analyzed again. As shown in Figure

2.3. Upon inspection of the  precipitate microstructure for the specimens that were crept at 677 and 704ºC, the bimodal  precipitate size distribution is still retained following exposure at temperature for times in excess of 3000hrs and 160hrs, as shown in Figure 39 2.4(a-b), respectively. These results show that the microstructure is relatively stable especially for these two test temperatures.

Figure 2.4: The post creep microstructure for ME3 after the (a) 677C/690MPa/2.0%/3750hrs strain test (b) 704C/724MPa/0.4%/160hrs strain test (c) 760C/345MPa/0.36%/3000hrs strain test and (d) 815C/345MPa/0.42%//140hrs strain test.

On the other hand, following the duration of creep at higher temperatures

(760ºC/3000hrs and 815ºC/140hrs) it is quite apparent that microstructural evolution of the

 precipitates, with coarsening of the larger secondary  precipitates and dissolution of tertiary  precipitates, has occurred as shown in Figure 2.4(c-d), respectively. As the tertiary  begins to dissolve into the matrix, the  channel width and effective spacing

40 between secondary  precipitates increases. As a result, it should be much easier for dislocations to bow through and/or circumvent the  precipitate field rather than being forced to cut precipitates in a planar shearing configuration. The same findings were made in the single crystal tests as well, with the tertiary  precipitates dissolving after the 760C and 815C tests but remaining after the 700C test.

2.3.3 Deformation Characterization using (C)TEM and (S)TEM

2.3.3.1 Polycrystalline Specimen I (677°C and 690MPa)

Both DC-TEM and high resolution STEM (HR-STEM) characterization techniques were employed to identify the post creep deformation mechanism of Specimen I that was crept to 2.0% plastic strain at 677ºC and 690MPa. Continuous planar faults that shear through both the  matrix and  precipitates were observed as shown in Figure 2.5(a).

Figure 2.5: (a) DC-STEM image showing continuous planar faulting observed in Specimen I. (b) A HR-STEM image of a microtwin observed in Specimen I [21].

These planar faults extend across entire grains. It is evident that these are indeed microtwins when imaged edge-on in HR-STEM. Figure 2.5(b) shows a HR-STEM image of a microtwin that is present in the  phase. Microtwinning was the prevalent deformation mechanism in all the grains that were examined, and it was rare to observe any other

41 deformation modes. Since this particular creep specimen was carried out to such high strain and the twins extended across individual grains, it was extremely difficult to find discrete locations where the microtwins had initiated and/or terminated. Further study aimed at tracking the nucleation and evolution of microtwins with increasing plastic deformation is needed to further explore how these microtwins nucleate.

2.3.3.2 Polycrystalline Specimen II (704°C and 724MPa)

When crept at a slightly higher temperature and stress, as in the case of Specimen

II at 704ºC and and 724MPa, several deformation mechanisms were observed operating simultaneously: microtwinning and dislocation activity. Using conventional bright field

(BF-TEM) imaging, highly planar, continuous stacking faults, extending through  and  in the same manner as found in Specimen I, were observed with the exception that they do not extend across entire grains, as shown in Figure 2.6.

Figure 2.6 A BF-TEM image revealing microtwins in Specimen II post creep test.

42 Along the length of the continuous stacking faults it can be seen that there are changes in contrast due to the presence of partial dislocations that are on the same {111} type glide planes or on parallel {111} planes. The primary deformation mechanism therefore appears to be microtwinning just as in the case of the Specimen I, which was crept at a lower temperature and stress.

Aside from microtwinning, there also exists a high density of dislocations entirely in the  matrix. Though these dislocations are found in grain interiors, they are mainly concentrated along grain boundaries. The dislocation activity near grain boundaries may be an indication that grain-boundary-localized deformation is already important in this temperature regime.

Direct comparison of polycrystalline response versus single crystal creep would be extremely helpful to determine the importance of grain boundaries to the overall creep behavior in this regime and for this class of superalloys

2.3.3.3 Polycrystalline Specimen III (760ºC and 345 MPa)

The deformation modes active during creep at 760ºC and 345MPa are distinctly different from those found in Specimens I and II. A representative TEM image of the deformation substructure observed is shown below in Figure 2.7.

Figure 2.7: BF-TEM micrograph of the creep deformation in specimen III post-test showing a mixture of a/2<110> dislocations and isolated stacking faults in the  precipitates.

The BF-TEM micrograph reveals a distinctly different shearing configuration in which faults are isolated to the ’ precipitates. Most of the time, the stacking faults do not transcend into the  channels as was the case with the earlier observed microtwins. In addition to the isolated stacking faults, a/2<110> full dislocations are present within the matrix as well. A gR TEM analysis of the stacking faults indicates the faults isolated within the  precipitates are superlattice extrinsic stacking faults (SESF).

2.3.3.4 Polycrystalline Specimen IV (815ºC and 345 MPa)

A different deformation mechanism was observed yet again in specimen IV, which was crept at the highest test temperature (815C). This new prominent deformation feature is dislocation activity, which is shown in Figure 2.8.

Figure 2.8: A BF-TEM image of dislocation activity in specimen IV after being crept at 815C

The dislocations are likely avoiding the  precipitates by a thermally activated climb mechanism. This conclusion is further supported by the fact the dislocations were found unpaired and seem to be situated above or below the  precipitates. Isolated stacking faults and other related precipitate shearing mechanisms were observed with much less frequency.

2.3.3.5 Single Crystal Creep (700C and 552MPa)

BF-STEM analysis of the deformation found different defect modes active between the [001] and [110] oriented ME3 samples as shown in Figure 2.9(a) and Figure 2.9(b).

Figure 2.9: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 700C compression creep tests at 552 MPa.

For the [110] orientation, which possessed superior creep strength, DC-STEM analysis revealed full dislocation activity in the  channels and stacking fault ribbons that can shear both phases. Stacking fault ribbons have been observed in [001] single crystals pulled in tension and involve the movement of a dissociated a<112> dislocation such that SISF and

SESF ribbons move cooperatively. It is envisioned that this complex configuration is the result of interacting ½<110> dislocations with different Burgers vectors. Stacking fault ribbons are usually observed among adjacent  precipitates in order to avoid the high energy APBs found between the SISF and trailing SESF. Reordering is necessary for the formation of both the SISF and SESF and may reduce the ability of the ribbon to shear

[22,23]. In contrast, intrinsic stacking faults (ISFs) in the  channels were found in the

[001] oriented sample. The presence of these stacking faults implies that de-correlation of the ½<110> full dislocations into their constituent 1/6<112> Shockley partials is active.

As well as de-correlation, shearing events such as isolated superlattice extrinsic stacking 46 faults (SESFs) and continuous planar faults, later determined to be microtwins, are present in the [001] oriented sample. The difference in deformation modes between the two orientations correlates with the anisotropy observed in the creep curves.

2.3.3.6 Single Crystal creep (760C and 414MPa)

For the two tests at 760ºC and 414MPa, again different deformation mechanisms were observed between the two orientations, as shown in Figure 2.10.

Figure 2.10: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 760C compression creep tests at 414 MPa

In the [001] oriented sample, de-correlation was no longer present in the 760ºC test. Instead a high frequency of microtwins and some isolated faults were observed. Conversely, the

[110] sample exhibited the same type of deformation mechanisms as the [110] test at

700ºC.

2.3.3.7 Single Crystal creep (760C and 552MPa)

When the stress level was increased at 760C to 552MPa both orientations revealed different deformation modes as compared to the 414MPa stress tests, as shown in Figure

2.11(a) and Figure 2.11(b).

Figure 2.11: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 760C compression creep tests at 552 MPa

At this higher stress, many examples of coupled dislocations shearing the  precipitates, indicative of APB shearing, were observed. The frequency of microtwins and SESFs were significantly reduced as well. In contrast, the [110] sample revealed a large number of isolated stacking faults instead of the stacking fault ribbons observed at the lower stresses.

48 2.3.3.8 Single Crystal creep (815C and 274MPa)

Just as the creep anisotropy decreased between the two orientations at the highest temperature so did the differences in deformation behavior, as shown in Figure 2.12(a) and

Figure 2.12(b).

Figure 2.12: BF-STEM Zone Axis images of the (a) [001] and (b) [110] oriented ME3 samples after the 815C compression creep tests at 274 MPa

For both orientations, shearing events such as microtwinning and stacking fault ribbons were significantly reduced in frequency. Instead, dislocation climb by-pass was observed for both. This is consistent with the results in the polycrystalline tests, and helps explain why the creep anisotropy is minimal at the higher test temperatures.

2.4 Discussion

2.4.1 Deformation Mechanism Map

Presently, limited literature has explored the effects of microstructure, stress, and temperature on creep deformation mechanisms in disk superalloys [5,17]. Moreover, it is imperative that this information is taken into account for newly developed Ni-base 49 superalloys, as in this study, to help with future alloy improvement. The deformation mechanism map, presented in Figure 2.13, provides the post creep deformation mechanisms that have been identified in this and earlier study of polycrystalline ME3, as well as similar disk alloy studies on Rene 88DT [21,25]. The map does not capture the influence of grain size, precipitate size/distribution, nor alloy chemistry, even though such factors are known to be important and contributing factors that control deformation behavior. The map was constructed solely to provide a qualitative representation of the operative creep mechanisms as a function of stress and temperature.

Figure 2.13: The mechanism map for several Ni-base disk alloys at different stress and temperature regimes revealing different post-creep deformation modes [15,20,26].

At high stresses, for both low and intermediate temperatures, coupled ½<110> dislocations are found, indicating that APB shearing of the  precipitates is the dominant deformation mechanism (blue area) [27,28]. In this mechanism, the dislocations travel in pairs since shearing of the  precipitates by the first a/2<110> dislocation would cause a 50 high energy APB to form. This is avoided when the secondary a/2<110> dislocation shears the precipitates restoring the precipitate to its original crystal structure. Since the a/2<110> dislocations are perfect dislocations in the  matrix, they do not leave a fault behind in the matrix, nor in the precipitate [29]. At high temperatures (>800C) and low stresses, thermally activated climb by-pass is the dominant deformation mechanism. In this temperature regime the  precipitates are surmounted by dislocations climbing over and around, leading to significant softening [30]. The mechanics involved for both APB shearing and climb by-pass have been explored in detail. Not presently well understood are the rate-controlling mechanisms in the reordering (light orange) regime of the mechanism map. These mechanisms result in matrix and precipitate shearing modes, which leave stacking faults and/or microtwins in their wake. In the following sections, the possible, creep-rate-limiting processes will be described in more detail for these deformation mechanisms.

2.4.2 Microtwin Formation

Twinning as a deformation mode is conventionally considered to occur only under low temperature, high strain rate conditions, or when the twinning stress is less than the stress required to initiate slip [31]. Nonetheless, twinning has been observed in Ni-base superalloys during high temperature creep deformation with under very low strain rate conditions [32,33].

Ardakani et al. [34] revealed that microtwins form in the single crystal Ni-base

Superalloy SRR99 when crept in tension at 750ºC and 850MPa along the [011] crystallographic orientations. Although they did not provide a mechanism for twin formation, they did mention that it is likely that the twins formed by the dissociation of an 51 a/2[011] dislocations into a a/6[112] Shockley partial and a a/3[121] superlattice Shockley partial, where the a/3[121] Shockley partial is responsible for creating the twin.

Kakehi [11] also observed microtwinning during creep at 700ºC and 820MPA on an experimental Ni-base superalloy with a multimodal  precipitate size distribution. Kakehi adopted the mechanism proposed by Guimier and Strudel [35] stating that shearing by a/6<112> Shockley partials is highly unlikely because the ordered L12 structure would be destroyed, resulting in a high energy configuration with incorrect nearest neighbor atomic bonds. The reasoning behind this argument is valid, since shear by successive a/6<112>

Shockley partials does indeed create a CSF on each shear plane with wrong nearest neighbor bond violations. The passage of a/3<112> partials would, however, preserve the

L12 structure.

Christian and Mahajan [36] hypothesized that deformation twinning in L12 superlattice structures would result in the production of a high energy pseudotwin; but, atomic re-shuffling could convert the pseudotwin structure to that of a true twin, with correct nearest neighbor atomic bonds. Kolbe [37] was the first to propose a diffusion- based crystallographic model that accounted for the thermally activated atomic reordering necessary to produce a true twin structure from a CSF in the  precipitates. Assuming that microtwins formed by successive shear of a/6<112> Shockley partials along adjacent

{111} planes, a high energy CSF would result in the  phase possessing a lower symmetry orthorhombic phase and a true twin structure in the fcc  matrix phase. The atomic reshuffling steps necessary to convert the pseudotwin to a true twin could be facilitated at the elevated temperatures. Viswanathan et al. [25] provided experimental evidence of the

52 microtwin formation process which convincingly show that microtwins propagate as a result of successive shear by a/6<112> partials on adjacent {111} planes.

Based on the detailed TEM characterization of the microtwinning deformation mechanisms that were reported in the Ni-base disk superalloys Rene 88DT and ME3, a quantitative creep model was developed by Karthikeyan et al. [38] The model incorporates the notion of paired a/6<112> Shockley partial dislocations, which shear both the  and  phases on adjacent {111} planes. Initially, high-energy 2-layer CSFs are produced as the precipitates are sheared by these paired Shockley partial dislocations. However, facilitated by the high temperatures, atomic reordering in the wake of the leading partials reverts the rearrangement of the Ni and Al atoms back to the ordered L12 structure. It can be argued that this local diffusion process of reordering is the rate limiting process during creep.

Kovarik et al. [39] later found multiple fault configurations that would energetically be able to reorder to form a true microtwin–all with the common requirement that the net shear that had created them is 1/3<112>. Thus, in addition to CSF/CSF configurations on adjacent {111} planes, CSF/APB configurations can also undergo reordering by a local, conservative diffusion process.

Unocic et al. [21] have proposed that de-correlation is a necessary precursor for microtwinning and that the shearing Shockley partials in the  matrix could interact on adjacent {111} planes and proceed to form a twin after the re-ordering process. However, the results in Figure 2.10(a) indicate that de-correlation is not necessary for microtwins to form. In the [001] oriented sample after deformation at 760°C, a large density of microtwins were present, while de-correlation was not. The two other features present along with microtwinning were full dislocation activity in the  channels, and isolated

53 SESFs in the ’ precipitates. In fact, Knowles and Chen [32] revealed through detailed

TEM work on creep mechanisms in single crystal CSMX-4 that twins are almost always associated with SESFs, which suggested that the formation of microtwins and SESFs may be related. Indeed, Smith et al. [40] using the models described by Kolbe and Kovarik,

[37,39] extended the reordering process to the formation of SESFs. They proposed that two unlike a/2<110> dislocations could interact at the / interface allowing for two a/6<112>

Shockley partials (with the same Burgers vector) to shear into the precipitate on adjacent

{111} planes. A high energy two-layer CSF would then be transformed to the low energy

SESF through reordering. In this way, SESFs can be precursors to microtwins in that the same process could occur again alongside the SESF to form a four-layer twin. They found no evidence of a/3<112> super-Shockley partials responsible for the creation of SESFs as first proposed by Milligan and Antolovich [41]. The ME3 single crystal creep specimens give a rare opportunity to further explore the formation of the microtwins. Indeed, Figure

2.14 provides an example of a microtwin forming directly from a SESF.

Figure 2.14: A high resolution HAADF-STEM image revealing a 4 layer microtwin forming from a SESF in ME3 inside a  precipitate.

It is envisioned that the lateral movement of the step, which is equal in height to two {111} planes, is responsible for transforming the SESF into a thin twin within the  precipitate.

In addition to this type of configuration, others have also been observed such that the

Shockley partials shearing the SESF on either side of the SESF. This implies that there may be even more reordering configurations than those considered by Kovarik et al. [39] while also strengthening the hypothesis that these microtwins are created from the widening of SESFs by Shockley partials. Presently, it remains unclear why the SESFs are isolated to the precipitates while the microtwins are able to shear through both phases. In fact, the SESF shown in Figure 2.14, terminates at the / interface while the twin extends

55 into the  channel. In summary, Figure 2.15 shows schematics of the two twin formation models discussed above.

Figure 2.15: The twin formation model presented by Unocic et al.[21] where (a) a single ½<110> dislocation dissociates into its constituent Shockley partials creating an ISF in the  matrix while the leading partial loops the  precipitate. (b) A second ½<110> cross slips in the matrix onto the {111} plane adjacent to the ESF while an SESF is formed in the precipitate. (c) The same process occurs again to form a three layer twin in the  matrix, and repetition of the process leads to a thickening twin. (d) The new twin formation model where two unlike ½<110> dislocations interact at the interface and dissociate (e) so like-signed Shockley partials can shear into the  precipitate forming a SESF. (f) The same process can occur again to form a 3 layer twin in the precipitate (resulting from shearing events on 4 adjacent {111} planes).

2.4.3 Segregation and Cottrell Atmospheres

Modern advances in EDS systems and microscope stability have allowed unprecedented insights into the chemical structure of these faults, including elemental segregation. The presence of segregation along shearing faults within precipitates indicates that long range diffusion is occurring, driven primarily by lowering fault energy.

Viswanathan et al.[42] found segregation along SISFs in ME3 and CMSX4, tending toward a local composition that is trending toward that of the  matrix. Thus, wrong-nearest neighbors created during the shearing events described in section 2.4.2 would not have as

56 large an energetic penalty. Similar segregated faults where also observed in Co-based alloys as well [43]. In order to examine if segregation has played a roll in the creation of the isolated SESFs observed in this study, high resolution EDX scans were conducted using an image-corrected Titan3 equipped a Super-X EDX detector at 300KV. This technology utilizes four silicon drift detectors located radially around the pole piece for improved counts and performance. Shown in Figure 2.16 are the net intensity elemental maps of an isolated SESF observed after the [001] single crystal test conducted at 760C and 414MPa.

Figure 2.16: Net intensity elemental maps of a SESF in ME3.

The EDX maps in Figure 2.16 reveal that the  formers Co, Cr, and Mo have segregated along the fault, while Ni and Al were depleted. This segregation matches that observed by

Viswanathan at al. [42] for SISFs in ME3 and again points towards a -like environment along the fault.

57 Recently, Smith et al. [ref] also observed prominent Co and Cr Cottrell atmospheres surrounding the region of leading Shockley partials creating the SESF in a  precipitate using EDS analysis. This is further discussed in Chapters 4 and 5. The Cottrell atmosphere, with similar composition to the  phase, may reduce the energy of the complex stacking faults created by the Shockley partials. Thus, it is presently unclear whether reordering, segregation, or translation of these Cottrell atmospheres are the rate controlling process during the  shearing process at intermediate temperatures, particularly since it is likely that all three processes are operative simultaneously.

2.5 Conclusion

A detailed substructure investigation aimed at exploring the effects of stress and temperature on creep deformation mechanisms in the newer generation Ni-base disk superalloy ME3 has been conducted. In order to evaluate its high temperature capabilities, creep experiments were conducted at stresses and temperatures above normal turbine operating conditions for both polycrystalline and single crystal specimens.

Characterization using TEM and STEM methods were used to identify and correlate the deformation mechanisms to macroscopic creep behavior. The following conclusions are made from this study:

1) A number of different deformation mechanisms were found operative for polycrystalline tensile creep and single crystal compression creep at temperatures and stresses ranging between 677-815ºC and 274-724MPa, respectively. The results are summarized in a deformation mechanism map that relates the different deformation modes to differences in stress and temperature. 2) Microtwinning was the observed dominant deformation mode in Specimens I and II and the [001] oriented single crystal at 760C. These microtwins propagate

58 through the motion of paired and like-sign a/6<112> Shockley partials on adjacent {111} planes. A reordering process is then necessary for the trailing fault to convert into a true twin. SESFs may be the precursors to twin formation by extending into twins when like-sign Shockley partials shear alongside the faults 3) In polycrystalline specimen III, which was crept at 760C, microtwins that sheared  and  phases were observed, in conjunction with isolated SESFs in the  precipitates. 4) At higher stresses and temperatures as high as 760°C, shearing of  particles by APB coupled ½<110> dislocations was observed. Thus, at sufficiently high stress, the APB shearing mode remains active, even at these higher temperatures. 5) Thermally activated climb by-pass was observed at all tests conducted at 815C, where unpaired a/2<110> dislocations were observed avoiding the  precipitates, while shearing modes were found to be much less frequent. 6) Orientation effects are prominent at intermediate temperatures (up to 760°C) and influences the active deformation mode. In this case, the [110] oriented samples showed superior creep strength compared to the [001] orientations. 7) This work reveals the need for mechanism-sensitive models of creep for Ni-based superalloy creep that can incorporate multiple deformation modes, and account for the microstructural evolution and orientation effects at these temperatures.

2.6 References

[1] C.T. Sims, A History of Superalloy Metallurgy for Superalloy Metallurgists,

Superalloys 1984. (1984) 399–419. doi:10.7449/1984/Superalloys_1984_399_419.

[2] R. Reed, The Superalloys: Fundamental and Applications, Cambridge University

Press, New York, 2006.

[3] D.P. Pope, V. Vitek, A Theory of the Anomalous Yield Behavior in L12 Ordered

Alloys, Acta Mater. 32 (1984) 435–448.

[4] B.H. Kear, J.M. Oblak, A.F. Giamei, Stacking faults in gamma prime Ni3(AI,Ti) 59 precipitation hardened nickel-base alloys, Metall. Trans. I (1970) 2477–2486.

[5] Q.Z. Chen, D.M. Knowles, Mechanism of <112>/3 slip initiation and anisotropy of

γ′ phase in CMSX-4 during creep at 750°C and 750 MPa, Mater. Sci. Eng. A. 356

(2003) 352–367. doi:10.1016/S0921-5093(03)00148-5.

[6] E.S. Huron, R.C. Reed, M.C. Hardy, M.J. Mills, R.E. Montero, P.D. Portella, et al.,

Controlling the deformation mechanism in disk superalloys at low and intermediate

temperatures, Superalloys 2012. (2012) 35–42.

[7] V.A. Vorontsov, C. Shen, Y. Wang, D. Dye, C.M.F. Rae, Shearing of Gamma Prime

Precipitates by a<112> Dislocation Ribbons in Ni-base Superalloys : A phase field

approach, Acta Mater. 58 (2010) 4110–4119. doi:10.1016/j.actamat.2010.03.041.

[8] V.A. Vorontsov, L. Kovarik, M.J. Mills, C.M.F. Rae, High-resolution electron

microscopy of dislocation ribbons in a CMSX-4 superalloy single crystal, Acta

Mater. 60 (2012) 4866–4878. doi:10.1016/j.actamat.2012.05.014.

[9] S. Raujol, M. Benyoucef, D. Locq, P. Caron, F. Pettinari, N. Clement, et al.,

Decorrelated movements of Shockley partial dislocations in the γ-phase channels of

nickel-based superalloys at intermediate temperature, Philos. Mag. 86 (2006) 1189–

1200. doi:10.1080/14786430500254685.

[10] P.R. Bhowal, E.F. Wright, E.L. Raymond, Effects of cooling rate and morphology

on creep and stress-rupture properties of a powder metallurgy superalloy, Metall.

Trans. A. 21 (1990) 1709–1717.

[11] K. Kakehi, Effect of primary and secondary precipitates on creep strength of Ni-

base superalloy single crystals, Mater. Sci. Eng. A. 278 (2000) 135–141.

doi:10.1016/S0921-5093(99)00579-1.

60 [12] D. Locq, P. Caron, S. Raujol, A. Coujou, N. Clément, On the Role of Tertiary

Gamma Prime Precipitates in the Creep Gebahior at 700C of a PM Cisk Superally,

Superalloys 2004 (Tenth Int. Symp. (2004) 179–187.

[13] T.P. Gabb, J. Telesman, P.T. Kantzos, K. O’Connor, Characterization of the

Temperature Capabilities of Advanced Disk Alloy ME3, 2002.

[14] T.P. Gabb, A. Garg, D.L. Ellis, K.M. O’Connor, Detailed Microstructural

Characterization of the Disk Alloy ME3, 2004.

[15] R. Unocic, On the Creep Deformation Mechanisms of an Advanced Disk Nickel-

Base Superalloy, The Ohio State University, 2008.

[16] P. Lukas, J. Cadek, L. Kunz, Creep of CMSX-4 Single Crystals of different

orientations in tension and compression, Mater. Sci. Eng. 208 (1996) 149–157.

[17] V. Sass, U. Glatzel, M. Feller-Kniepmeier, Creep Anisotropy in the Monocrystalline

Nickel-Base Superalloy CMSX-4, Superalloys 1996 (Eighth Int. Symp. (1996) 283–

290. doi:10.7449/1996/Superalloys_1996_283_290.

[18] K. Kakehi, Tension/compression asymmetry in creep behavior of a Ni-based

superalloy, Scr. Mater. 41 (1999) 461–465. doi:10.1016/S1359-6462(99)00191-8.

[19] M.D. Abramoff, P.J. Magalhaes, S.. Ram, Image Processing with ImageJ,

Biophotonics Int. 11 (2004) 36–44.

[20] H. Deutchman, On the Creep Behavior and Deformation Mechanisms Found in an

Advanced Polycrystalline Nickel-base Superalloy at High Temperatures, 2013.

[21] R.R. Unocic, N. Zhou, L. Kovarik, C. Shen, Y. Wang, M.J. Mills, Dislocation

Decorrelation and Relationship to Deformation Microtwins During Creep of a γ’

Precipitate Strengthened Ni-Based Superalloy, Acta Mater. 59 (2011) 7325–7339.

61 [22] V.A. Vorontsov, C. Shen, Y. Wang, D. Dye, Shearing of γ’ Precipitates by a<112>

Dislocation Ribbons in Ni-Base Superalloys: A Phase Field Approach, Elsevier.

(n.d.).

[23] C.M.F. Rae, N. Matan, R.C. Reed, The role of stacking fault shear in the primary

creep of [001]-oriented single crystal superalloys at 750°C and 750 MPa, Mater. Sci.

Eng. 300 (2001) 125–134.

[24] R.R. Unocic, G.B. Viswanathan, P.M. Sarosi, S. Karthikeyan, J. Li, M.J. Mills,

Mechanisms of creep deformation in polycrystalline Ni-base disk superalloys,

Mater. Sci. Eng. A. 483-484 (2008) 25–32. doi:10.1016/j.msea.2006.08.148.

[25] G.B. Viswanathan, S. Karthikeyan, P.M. Sarosi, R.R. Unocic, M.J. Mills,

Microtwinning During Intermediate Temperature Creep of Polycrystalline Ni-Base

Superalloys: Mechanisms and Modeling, Philos. Mag. 86 (2006) 4823–4840.

[26] T.M. Smith, L.V. Duchao, T. Hanlon, A. Wessman, Y. Wang, M.J. Mills,

Determination of Orientation and Alloying Effects on Creep Response and

Deformation Mechanisms in Single Crystals of Ni-Base Disk Superalloys,

Superalloys 2016. submitted (2016).

[27] B.H. Kear, J.M. Oblak, W. Aircraft, Deformation modes in gamma prime

precipitation hardened nickel-base alloys., J. Phys. 35 (1974).

[28] R.R. Unocic, L. Kovarik, C. Shen, P.M. Sarosi, Y. Wang, J. Li, et al., Deformation

Mechanisms in Ni-Base Disk Superalloys at Higher Temperatures, Superalloys

2008 (Eleventh Int. Symp. (2008) 377–385.

doi:10.7449/2008/Superalloys_2008_377_385.

[29] D. Raynor, J.M. Silcock, Strengthening Mechanisms in γ‘-Precipitating Alloys, Met.

62 Sci. J. 4 (1970) 121–130.

[30] R.H. B. Kear, A. Giamei, J. Silcock, Slip and Climb Processes in Gamma Prime

Hardened Nickel-base Alloys, Scr. Metall. 2 (1968) 287–293.

[31] G. Dieter, Mechanical Metallurgy, McGraw Hill, 1986.

[32] D.M. Knowles, Q.Z. Chen, Superlattice stacking fault formation and twinning

during creep in gamma/gamma prime single crystal superalloy CMSX-4, Mater. Sci.

Eng. A. 340 (2003) 88–102.

[33] G.B. Viswanathan, S. Karthikeyan, P.M. Sarosi, R.R. Unocic, M.J. Mills,

Microtwinning during intermediate temperature creep of polycrystalline Ni-based

superalloys: mechanisms and modelling, Philos. Mag. 86 (2006) 4823–4840.

doi:10.1080/14786430600767750.

[34] M.G. Ardakani, M. McLean, B. a. Shollock, Twin formation during creep in single

crystals of nickel-based superalloys, Acta Mater. 47 (1999) 2593–2602.

doi:10.1016/S1359-6454(99)00145-7.

[35] A. Guimier, J.L. Strudel, Stacking fault formation and mechanical twinning

processes in a nickel-base superalloy during tensile deformation at high temperature,

in: Int. Conf. Strength Met. Alloy. Conf. Proc., 1970.

[36] J.W. Christian, S. Mahajan, Deformation Twinning, Prog. Mater. Sci. 39 (1995).

[37] M. Kolbe, The High Temperature Decrease of the Critical Resolved Shear Stress in

Nickel-Base Superalloys, Mater. Sci. Eng. (2001) 383.

[38] S. Karthikeyan, R. Unocic, P.M. Sarosi, G.B. Viswanathan, M.J. Mills, Modeling

Microtwinning During Creep in Ni-based Superalloys, Scr. Mater. 54 (2006) 1157–

1162.

63 [39] L. Kovarik, R.R. Unocic, J. Li, P. Sarosi, C. Shen, Y. Wang, et al., Microtwinning

and other shearing mechanisms at intermediate temperatures in Ni-based

superalloys, Prog. Mater. Sci. 54 (2009) 839–873.

doi:10.1016/j.pmatsci.2009.03.010.

[40] T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman, et

al., Segregation and η phase formation along stacking faults during creep at

intermediate temperatures in a Ni-based superalloy, Acta Mater. 100 (2015) 19–31.

doi:10.1016/j.actamat.2015.08.053.

[41] W.W. Milligan, S.D. Antolovich, The mechanisms and temperature dependence of

superlattice stacking fault formation in the single-crystal superalloy PWA 1480,

Metall. Trans. A. 22 (1991) 2309–2318.

[42] G.B. Viswanathan, R. Shi, A. Genc, V.A. Vorontsov, L. Kovarik, C.M.F. Rae, et

al., Segregation at Stacking Faults within the γ’ Phase of Two Ni-Base Superalloys

Following Intermediate Temperature Creep, Acta Mater. (2014).

[43] T. pollock M. Titus, A. Mottura, G. Viswanathan, A. Suzuki, M.J. Mills, High

Resolution Energy Dispersive Spectroscopy Mapping of Planar Defects in L1 2 -

containing Co-base Superalloys, Submiss. to Acta Mater. (2014).

Chapter 3 – Orientation and Alloying effects on Creep Response and Deformation Mechanisms in Single Crystal Ni- base Disk Superalloys

This chapter is a modified version of publication [5]

3.1 Introduction

Ni-base superalloys have long been the standard for use in the hot section of turbine engines. However, as the operating temperatures in next generation engines are increased, in order to improve engine efficiency and reduce carbon emissions, new deformation modes may become dominant and strongly affect the creep behavior. Improved understanding of these mechanisms will help progress the understanding of deformation in this regime, and drive improvements in the high temperature properties of new generation

Ni-base superalloys.

Detailed studies of the mechanism changes in these polycrystalline Ni-base alloys in different temperature and stress regimes have proven to be a daunting endeavor, especially at intermediate temperatures between 650-800 C. Presently little is understood about these deformation mechanisms and the parameters that affect their operation. It is widely agreed that below the intermediate temperature range athermal shearing by full

½<110> dislocations, creating anti phase boundaries (APBs) in  precipitates is the active deformation mode and a source of superior creep strength [1–3]. Well above this temperature range, dislocation climb by-pass is usually found to be the primary deformation mode [4]. Multiple studies have found that in the intermediate temperature

65 range (roughly between 650-750 °C), following creep under a range of stresses, reordering processes such as microtwinning and isolated stacking faults become important factors on creep performance [5–8]. Koble [9] first hypothesized that the formation of microtwins involved the shear of Shockley partials pairs on adjacent {111} planes in  precipitates, even though this would create a high energy two layer complex stacking fault (CSF). Kolbe proposed, without experimental evidence, that a reordering process must occur in the wake of the shearing partials, removing wrong nearest neighbor bonds. Kovarik et al. [4] using ab initio calculations found that not only was this reordering process energetically possible but multiple configurations could result in the subsequent low energy twin formation.

Smith et al. [10] using the same reordering process described by Kovarik et al. [4] expanded it to include isolated superlattice extrinsic stacking fault (SESF) formation. In addition to reordering, Smith et al. [10] also reported that elemental segregation along faults and motion of a Cr- and Co-rich Cottrell atmosphere surrounding the shearing

Shockley partials may be necessary for these shearing events to proceed through the  precipitates.

Another important deformation mechanism reported in this temperature range is the change from full dislocation motion in  channels to de-correlation, where full dislocations separate into their constituent Shockley Partials. Raujol et al. [11] found de- correlation to be active during creep at 700 C. In-situ analysis found that the de-correlated

Shockley partials could proceed shearing both the  and  phases. Considering forces on both the shearing and trailing Shockley partials, Raujol et al. [11] found that channel width was an important factor in whether a dislocation would de-correlate or not. Later, Unocic et al. [12] hypothesized that de-correlation may be a precursor to microtwins as a source 66 of the  shearing Shockley partials. They found de-correlated dislocations present in  channels in close proximity to microtwins after tensile tests conducted at 677 C for ME3, a polycrystalline disk alloy currently used commercially.

In polycrystalline alloys, grain-to-grain variations in the observed deformation modes not only depend on stress and temperature levels, but also how the individual grains are oriented relative to the loading axis. These differences between grains are not completely understood. It is possible that these differences reflect a significant dependence on Schmid factor; however, dissecting this dependence from those due to stress, temperature, and strain level is a formidable task. Furthermore, Phillips et al. [13] have shown that Schmid factors alone cannot accurately predict the active slip systems found in each grain. Low Schmid factor slip systems can become active during creep. It is not clear whether this is caused by local stresses found at interfaces like grain boundaries or if it is a grain-to-grain interaction effect.

Single crystal work by Kakehi et al. [14] determined that anisotropic creep properties are exhibited for single crystal alloys and an asymmetry is present depending on whether compressive or tensile load is applied. This variation in grain deformation mode makes creating accurate deformation models for predicting alloy creep performances a complex, yet scientifically rich, endeavor. In order to advance these efforts to understand and model the creep behavior of polycrystalline disk alloys, this study will employ single crystal surrogates to the disk alloys so as to create a more thorough understanding of the orientation, temperature, and stress dependence of deformation modes. The present chapter reports on the behavior in two specific crystal orientations and focuses on two different alloy compositions. The deformation mechanism after modest creep strain is

67 analyzed using advanced scanning transmission electron microscopy (STEM) based techniques. Of particular interest is how deformation varies and why between the different creep tests. In this chapter, we focus on recent studies of specially-grown single crystals of two Ni-base superalloys, used in polycrystalline form for turbine engine disks. The single crystals enable exploration of the orientation and temperature effects on creep response, which is presently not known for this class of alloys. In addition, diffraction contrast

STEM (DC-STEM) imaging is used to evaluate for the first time the operative deformation mechanisms in the intermediate temperature regime, spanning 700-800 °C. Motivated by these experimental results, phase field modeling has been employed to study the interaction of matrix dislocations with realistic precipitate structures.

3.2 Materials and Experimental Methods

3.2.1 Sample preparation

Single crystal analogs of two different disk alloys were obtained from GE Global Research

Center; the commercial alloy ME3, first presented in Chapter 2, and a modified alloy

ME501 [15], with minor compositional changes. Table 3.1 below shows these differences in composition by weight percent.

Table 3.1: Alloy compositions of ME3 and ME501

Alloy Ni Co Cr Mo W Nb Ta Al Ti Hf C B Zr

ME3 Bal. 20.6 13.0 3.8 2.1 0.9 2.4 3.5 3.4 0 0.05 0.03 0.03

ME501 Bal. 18.0 12.0 2.9 3.0 1.5 4.8 3.0 3.0 0.4 0.05 0.03 0.05

Both single crystal castings, which again inherently have a level of chemical segregation greater than their powder metallurgy counterparts, underwent a heat treatment

68 that formed a bi-modal ' precipitate microstructure. To compare the resultant microstructures for both alloys, a sample from each alloy was polished down to a 1200 fine grit polish with a subsequent colloidal silica finish. After a  etchant consisting of 2 mL hydrofluoric acid, 30 mL nitric acid, and 50 mL lactic acid was applied the microstructure of both alloys were imaged using a FEI Sirion scanning electron microscope (SEM). A secondary detector was used over backscatter for improved surface resolution and to avoid subsurface precipitates from being included in the analysis. Figure 3.1 shows the differences between the two microstructures.

Figure 3.1: SEM backscatter image of (a) ME3 microstructure and (b) ME501 microstructure.

Some differences are observed in the microstructures of the two alloys, even though they experienced identical heat treatments and cooling rates. ME501 possesses a finer microstructure compared to ME3, while ME3 appears to have a coarser secondary  precipitates. To quantify these differences each microstructure, SEM images were analyzed using ImageJ [16]. Below Table 3.2 shows the differences in area fraction and precipitate size based on equivalent diameters between the two alloys.

69 Table 3.2: Secondary and tertiary  precipitate area fraction and size in ME3 and ME501

Alloy Secondary  Area Secondary  size Tertiary  Tertiary  Fraction (AF) AF size ME3 47% 350 nm 3-4% 34 nm ME501 52% 300 nm 3-4% 30 nm

3.2.2 Creep Test Preparation

Orientation Imaging Microscopy (OIM) was employed to determine the orientation for both alloy castings, as well as to detect and eliminate spurious grains from the regions used to prepare creep specimens. 7 mm long rectangular prism specimens were extracted for compression creep tests from each casting with a 1:1:2.5 dimensional ratio. Two orientations were extracted, [001] and [110], using electrical discharge machining (EDM).

To remove the damage layer created by the EDM process, each side of the samples were polished down to a 1200 fine grit finish.

Three different temperature regimes were explored in this study: 700 C, 760 C, and 815 C. The stress varied from each test in an attempt to maintain a consistent strain rate at each temperature. For the ME501 samples, the compression samples were crept under 710 MPa, 552 MPa, and 413 MPa, at each respective temperature. The ME3 samples were crept under these stresses as well; however, lower stresses were needed in order to keep strain rates comparable those displayed by the ME501 samples. Therefore, the stresses were reduced to 552 MPa, 413 MPa and 276 MPa, at each respective temperature.

Each test was ended once about 0.5% plastic strain was reached. MTS 810 compression cages with attached linear variable displacement transducers (LVDTs) that recorded the displacement of the compression plattens were used to conduct each compression creep test. Thermocouples (K type) were used to record the temperature for each test. Once the 70 desired strain was reached, the test was immediately ended and the sample kept under load while being cooled rapidly under forced air. This procedure was an attempt to minimize changes in microstructure, chemistry and deformation as the sample reached room temperature.

3.2.3 Deformation Characterization Methods

After each compression test was completed a side of the compression cuboid was polished to remove any oxidation. Transmission electron microscope (TEM) foils were then extracted normal to the compression axis using an FEI Helios Nanolab Dualbeam 600

Focused ion beam microscope. By extracting the foils normal to the compression axis one last check could be made to ensure the sample was indeed cut to the right orientation. The

TEM foils were analyzed using an FEI Tecnai F20 field emission 200 KV STEM. STEM analysis was used in preference to conventional TEM (CTEM) for a multitude of reasons.

An important one is the ability to image on low index zone axes which allows for multiple diffraction conditions to be activated at the same time. This allows for several slip systems to be imaged at the same time giving an improved overall image of the deformation modes active. Other advantages for using STEM include reduced bend contours and the ability to image thicker samples. A study of STEM diffraction contrast by Phillips et al. 17] found that the conventional gb invisibility rules still apply for STEM and STEM images can be evaluated the same as CTEM images.

71 3.3 Results

3.3.1 Monotonic Compression Creep Response

Figure 3.2(a) below shows the [001] and [110] oriented single crystal compression creep responses for ME3 and ME501 tested at 710 MPa. ME3 crept 1-2 orders of magnitude faster than ME501 (with a strain rate of 5x10-7 s-1 compared to ME501’s 1x10-

9 s-1 strain rate). For better analysis on the deformation responses between the two alloys, two more ME3 tests were conducted using a lower stress of 552 MPa to reach similar strain rates similar to those for ME501. At these slower strain rates, a prominent anisotropy in creep strength is observed, shown in Figure 3.2(b), with greater anisotropy between orientations experienced by the ME3 alloy.

Figure 3.2: 700 C compression creep strain versus time curves from [001] and [110] orientated ME501 and ME3 samples with (a) same stress and (b) comparable strain rates.

For the comparable strain rate tests both alloys experienced the better creep strength in the [110] orientation. This coincides with reported compression creep strength values being considerably lower for [001]-oriented crystals than for other orientations [14,18].

72 The poor compression creep performance in this orientation has been attributed to events that cause local stress relaxation, such as  shearing by microtwins or SESFs [19–21].

Additionally, the [001] orientation has the largest number of active slip systems of any orientation, allowing for the activation of more glide systems initially, and resulting in the observed high primary strain rates [22]. Fluctuations in the creep curves are most likely a result of the very low strain rates that are approaching the resolution limit of the LVDT’s.

These small instabilities are believed to be the result of small temperature fluctuations in the furnace and/or other external variables that become apparent in these very low strain rate creep tests.

At increased temperature of 760 C, the stress levels were lowered to 552 MPa in order to maintain strain rates near 1x10-9 s-1 . As shown in Figure 3.3(a) and Figure 3.3(b), similarities in the creep responses were seen in this temperature regime when compared with 700 °C. Once again, ME3 crept at rates significantly faster than ME501 at the same applied stress. Once again additional tests were performed to achieve comparable strain rates, as shown in Figure 3.3(b).

Figure 3.3: 760 C compression creep strain versus time curves from [001] and [110] orientated ME501 and ME3 samples with (a) same stress and (b) comparable strain rates.

Though the anisotropy appears to be reduced compared to the 700 C tests, the creep curves again have different shapes depending on the orientation. For the [001] orientations, the curve revealed a small amount of strain hardening followed by accelerating creep strain rates. In contrast, the [110] oriented crystals revealed continual hardening as the tests progressed. In this case, ME501 performed better, even when ME3 was crept under the reduced stress.

The next set of compression creep tests were performed at 815 C and 413 MPa.

The ME3 alloy predictably exhibited reduced strength, relative to ME501, and new tests were carried out with a stress of 276 MPa to obtain comparable strain rates. Figure 3.4 below reveals the creep curves for all the tests at 815 C. At 815 C, the anisotropy between

[001] and {110} orientations that was observed at 700 C and 760 C has become much smaller, no orientation seems to show superior creep strength over the other. However,

74 ME501 still possesses clearly superior creep strength compared to ME3 when both were crept at 413 MPa.

Figure 3.4: 815 C compression creep curves from [001] and [110] orientated ME501 and ME3 samples with (a) same stress and (b) comparable strain rates.

3.3.2 Compression Creep Response: Polycrystalline vs. Single Crystal

To improve understanding of grain boundary effects on creep strength, new compression tests were conducted on polycrystalline ME501 specimens of similar size.

Below in Figure 3.5 is the comparison between the polycrystalline creep response to that of the ME501 single crystals at (a) 700 C and (b) 760 C.

Figure 3.5: Compression creep strain versus time curves from [001], [110] orientated and polycrystalline ME501 (a) at 700 C and (b) at 760 C.

3.3.3 Transient Stress Tests in ME501

To probe the dependence of stress on creep strain rates and possibly new insights into the type of creep behavior exhibited in the single crystals, transient stress tests were employed on ME501 for both orientations at 700C, 760C and 815C. An example of one of these tests is shown below for a [001] oriented sample at 760C.

Figure 3.6: Transient stress test performed on a [001] compression sample of ME501 at 760C. The stress level was changed mid test to observe the change in strain rate.

By plotting the stress over the strain rate and calculating the slope between the two points, creep stress exponents (n) can be calculated. Below are the results from all of the transient stress tests.

Figure 3.7: The calculated stress exponents (n) from the transient stress tests performed at all three temperature regimes and both orientations for ME501 are presented along with the plotted stress (MPa) vs strain rate (s-1) points. 77

In all cases shown in Figure 3.7 the stress exponents were found to be very high and other than the [110] tests at 760C all of the calculated stress exponents were found to be above 9. This results in activations energies (Q) over three times of the activation energy of self-diffusion in Nickel [1]. This may be due to observed differences between single crystal creep and polycrystalline creep. In single crystal creep the creep curve usually doesn’t possess the three regimes of primary, secondary and tertiary creep and instead continues to creep progressively faster as time advances. The lack of grain boundaries in single crystals also results in the absence of grain boundary sliding and diffusion during creep. Due to these differences, creep power law seems to break down for single crystals and the high measured stress exponents are to be expected [1].

3.3.4 Deformation Analysis Using DC-STEM Imaging

3.3.4.1 Deformation Analysis Using DC-STEM for 700 C tests

Only the samples crept at the target strain rate of 10-9 s-1 will be discussed presently for the STEM deformation analysis in this chapter. Representative, zone axis bright field

STEM (BF-STEM) deformation analysis images are shown in Figure 3.8(a) and Figure

3.8(b) for ME3 and ME501, respectively, and both orientations after 700 C compression creep tests. Despite the alloying and microstructure differences observed in the two alloys, both ME3 and ME501 exhibited similar deformation modes. For the [110] orientation, which possessed superior creep strength, STEM diffraction contrast reveals full dislocation activity in the  channels and stacking fault ribbons active for both alloys. Stacking fault ribbons are usually observed in [001] single crystals pulled in tension and require the interaction of unlike ½<110’s> [23,24]. Another important observation is the intrinsic

78 stacking faults (ISFs) in the  channels found in both the ME3 and ME501 in the [001] orientation. The presence of these stacking faults implies that de-correlation of the ½<110> full dislocations into their constituent 1/6<112> Shockley partials is active. As well as de- correlation, shearing events such as isolated SESFs and microtwins are present in the [001] orientation for both alloys. The difference in deformation modes between the two alloys correlates with the anisotropy observed in the creep curves between the [001] and [110] orientations.

Figure 3.8. BF-STEM Zone Axis images for both [001] and [110] orientations after 700 C compression creep tests for (a) ME3 and (b) ME501. 79

3.3.4.2 Deformation Analysis Using DC-STEM for 760 C tests

Figure 3.9 shows examples of the primary deformation modes observed from ME501 and

ME3 specimens after compression creep tests at 760 C.

Figure 3.9: BF-STEM Zone Axis images for both [001] and [110] orientations after 760 C compression creep tests for (a) ME3 and (b) ME501.

80 There is no evidence of ISFs in the  channels, indicating that de-correlation is no longer observed at 760 C in [001] samples. Instead of dislocation de-correlation, numerous  shearing events are active; however, an apparent difference in the type of shearing events exists between ME3 and ME501. For ME3, the long faults that extend through both the  and  phases are microtwins, confirmed using high angle annular dark field (HAADF) imaging [25]. The presence of numerous microtwins in ME3, reveals that the hypothesis by Unocic et al. [12] that dislocation de-correlation is necessary for microtwin formation may not be exclusively correct. Indeed, other unknown dislocation interactions and pathways must exist that can result in microtwin formation. In the ME501 specimen, only isolated SESFs are present. The formation of these SESF’s was discussed in full by Smith et al. Again for the [110] orientations in both alloys full dislocation activity in the  channels and stacking fault ribbons were active.

3.3.4.3 Deformation Analysis Using DC-STEM for 815 C tests

Figure 3.10 shows the active deformation modes observed in ME501 and ME3 specimens after compression creep tests at 815 C. Interestingly, as the anisotropy between the two different orientations disappeared in the creep curves at 815 C so did the difference in deformation mechanisms. For each alloy at both orientation, climb by-pass becomes the dominant mechanism for creep tests at 815 C while  shearing events were significantly reduced in frequency.

Figure 3.10: BF-STEM Zone Axis images for both [001] and [110] orientations after 815 C compression creep tests for (a) ME3 and (b) ME501.

3.3.4.4 Deformation Analysis Using DC-STEM for 700C and 815 C Polycrystalline Tests

Figure 3.11 shows the deformation modes observed in the polycrystalline ME501 compression creep tests at 700C and 815C. At 700C, the polycrystalline test performed as well if not better than the single crystal creep tests and the observed deformation seemed to support that observation. Some grains showed very little deformation present where the grains that did have deformation appeared to be consigned near the grain boundaries as shown in Figure 3.11(a). Figure 3.11(b) shows the deformation present after the

82 polycrystalline test at 815C, which performed much more poorly than the single crystals.

A high density of deformation was observed in this sample, with dislocation and stacking faults present through entire grains. Very heavy dislocation densities were observed near grain boundaries as well, possibly highlighting the effects of grain boundary sliding occurring at creep in this higher temperature regime.

Figure 3.11: BF-STEM image of deformation in a polycrystalline sample after a compression creep test at (a) 700C and (b) 815C.

3.4 Discussion

3.4.1 Orientation Effects on Deformation

A relationship between the macro-scale creep behavior and nano-scale deformation is clearly present. The creep anisotropy between the [110] and [001] orientations at 700 C and 760 C correlates very well with differences in active shearing mechanisms.

For the [110] orientations in both alloys, stacking fault ribbons were prominent, while full dislocations were found in the  matrix. A stacking fault ribbon is usually

83 observed amid multiple adjacent  precipitates to avoid the high energy APBs. This leaves the constituent ½<112> dislocations separated by perfect crystal, where the latter forms a

SESF. Re-ordering must take place for the formation of the SESF and while reordering is

1 not necessary for the SISF, it may need to occur if the ⁄3<112> super-partials have dissociated [26]. These reordering mechanisms are likely rate-controlling and may slow the ability for the stacking fault ribbon to shear. Multiple studies have reported stacking fault ribbons to be the source of superior creep strength in [001] single crystals crept in tension [23,26–28]. For the [001] samples, both ME3 and ME501 experienced similar active deformation modes. At 700 C, the presence of microtwins in ME3 and isolated stacking faults in ME501 were observed as well as de-correlated dislocations in the  matrix. This is the only time dislocation de-correlation is observed in both ME3 and

ME501. All other cases only revealed the presence of full dislocation activity in the channels. For both ME3 and ME501, the frequency of  shearing events is significantly increased at 760 C compared to 700 C. Again, microtwins were found in ME3 while isolated stacking faults were observed in the ME501 sample.

For the 815 C compression creep experiments, the different  shearing events and exclusively channel-bound, full ½[110] dislocations are replaced with climb by-pass of the full dislocations for all orientations and alloys. Two reasons for this change in deformation mode could be the increased temperature during the creep tests providing more energy for dislocation climb, and the dissolution of the tertiary  precipitates in the channels. As expected, the change in active deformation modes to climb by-pass for all specimens

84 correlates with the disappearance of anisotropy in the creep curves at 815 C, supporting the hypothesis that creep anisotropy is directly related to the type of deformation mode active between orientations. From these results, a qualitative ranking of  shearing modes, from best to worst, can be assembled and is shown below with stacking fault ribbons promoting superior creep properties:

Stacking Fault Ribbons > Isolated SESFs > Microtwinning (1)

This ranking is further supported by the work of Rae et al .[27] who found stacking fault ribbons to be the cause of superior creep strength in [001] single crystals crept in tension, in addition to work by Yamashita et al. [21] who found microtwin formation correlates with poor creep strength. . On the other hand, Unocic, et al. [29] have compared the creep response in tension for polycrystals of fine (rim) versus course (bore) microstructures in ME3 at 677 °C, that the former exhibited microtwinning and better creep strength than the latter at the same stress level. However, trying to correlate deformation to creep response in polycrystals is difficult due to local and grain to grain variations in observed deformation. Without a rigorous statistical study, any conclusion will be confounded by local grain boundary effects and other unaccounted variables.

3.4.2 Dislocation Activity Diagram (DAD) – Predicting De-Correlation

The dislocation movement through the γ matrix channel is one of the major factors determining the creep mechanisms in Ni-based superalloys. When dislocation sources are distributed throughout the microstructure, the -channel width controls the characteristics 85 of the looping mechanisms (full looping, partial looping, or unable to loop). This has been investigated in previous works [12], where an analytical DAD model has been developed to investigate the critical shear stresses of full and partial (de-correlated) dislocation looping. The activation conditions for de-correlation and full dislocation looping have been found to be sensitive to both precipitate microstructure (especially -channel width) and other physical parameters such as shear modulus, lattice friction and work hardening in the channel. At the same time, it has been observed that de-correlation occurs at lower stress levels relative to other possible processes (for example, full dislocation Orowan looping and APB shearing). For de-correlation to occur, the direction of applied shear stress must be such that sufficient stress magnitude is applied on the leading Shockley partial [11].

The dislocation sources in these supersolvus heat treated alloys are sparse, usually associated with intragranular carbides or grain boundary sources in polycrystals, as the initial dislocation density is very small. In this case, the operating deformation mechanism can only be revealed as a result of dislocation percolation, when the dislocations are initiated from the rare sources. In this work, a percolation simulation approach is designed based on the analytic DAD diagram and direct SEM observation of the precipitate microstructure. In this approach, the critical channel widths of dislocation activation must first be calculated by the DAD for a given combination of physical parameters such as shear modulus, lattice friction and work hardening. In the DAD, critical loading conditions for dislocation activation are represented by boundaries of stress magnitude and direction.

The critical channel width is obtained when the external loading condition falls on a boundary shown in the DAD, where fulfillment of the following equations occurs:

86 τcrt(ChW) = σapp (2)

휙crt(ChW) = 휃app (3)

Where ChW is channel width, τcrt is magnitude of critical stress and 휙crt is the direction,

σapp is the magnitude of applied stress and 휃app is the direction. When both the equations are satisfied, they provide the solution of ChW which gives the critical channel width where de-correlation is promoted. In order to apply the concept of the DAD on an actual microstructure, it is insufficient to consider only average microstructure parameters. For instance, a dislocation segment will pass through the matrix continuously until it meets a channel narrower than the critical value. Therefore, a percolation study is needed to determine the critical channel width. In the percolation simulation, a dislocation line is set to proceed from some unknown source and into a precipitate microstructure obtained directly from segmented SEM images. The SEM images are obtained in <111> oriented grains such that the precipitate cross-sections are most accurate to that sampled by dislocations on a {111} shear plane. Finally, the sheared area-fraction and percolation distance are obtained by the simulation, and the activation probability of each deformation mode is estimated.

3.4.2.1 Determining Deformation Mechanism Populations

Using the input parameters in Table 3.3, the critical channel widths for full dislocation activity and for de-correlation are calculated to be 156 nm and 117 nm, respectively.

87 Table 3.3. The input parameters from the DAD calculation Intrinsic stacking fault energy 10 mJ/ m2 [11] Shockley partial Burgers vector b=0.146 nm [1] Shear modulus μ=67.2E3 MPa [1] Lattice friction f=15 MPa (700 C) [30] Line tension α=1.5 [1] Secondary  area fraction 47.2% Tertiary  area fraction 3.5% Applied stress 552 MPa along [001] direction

Due to the distribution in channel widths in the microstructure, there are certain channels (white area in Figure 3.12) that allow activation of only the de-correlation mode.

Therefore, locally the deformation microstructure will be dominated by de-correlation that will spread throughout until encountered by obstacles. An example of the percolation simulations, performed for a given microstructure with dislocations entering from the 4 edges is shown in Figure 3.13. In each case, a full dislocation is placed initially on the edge and an applied load drives the dislocation into the microstructure. The white colored region represents the full dislocation sheared region, which is bounded by narrower channels. These narrower channels allow only the leading partial dislocation to pass through. As a result, de-correlation is activated in the pink colored area. Finally, much narrower channels prohibit movement by any kind of dislocation, which results in unsheared regions (red colored areas).

Figure 3.12: Channels in a ME3 precipitate microstructure. The secondary precipitate cross-sections are shown in black. The average channel width is 210 nm. The critical channel width for the activation of de-correlation (red) is 156 nm, while that for stopping even partial looping (blue) is 65 nm.

The average area fraction of full dislocation shearing is calculated to be 27.4%, while the average area fraction of partial dislocation shearing (de-correlation activation) is calculated to be 64.3%. In addition, the average percolation distance (measured from each

is expected to be only concentrated in the vicinity of their sources. According to these percolation simulations, the de-correlation area dominates for the given microstructure and loading condition, in spite of a rather broad channel width distribution in the microstructure.

Figure 3.13: Percolation simulations preformed when dislocations are initiated at grain boundaries. white color indicates a closed zone of full dislocation activation. pink color shows the activation area of de-correlation. The red color shows unsheared area. Percolation distances are indicated by the green arrows.

It should be emphasized that it is the critical channel width rather than the average channel width that determines the active deformation modes, as demonstrated clearly by the above percolation simulations. Thus, critical channel width rather than average channel width should be used in DAD analysis. For the microstructure shown in Figure 3.12, for example, if the average channel width of 210 nm is used, then the DAD calculation incorrectly predicts full dislocation activity at 700 C in both the [110] and [001] orientations at 552 MPa as shown in Figure 3.14(a).

90 This does not match the experimental findings of de-correlation in the [001] orientation at

700 C. Additionally, the simulations in Figure 3.13 found that for the [001] orientation the smaller channel widths controlled the majority of the deformation by forcing full dislocations to de-correlate in order to propagate further. As expected, when the critical channel width for the activation of de-correlation (156 nm) is used, as shown in Figure

3.14(b), then the DAD predicts de-correlation as the active deformation mode in the [001] oriented sample, whereas full dislocation as the active deformation mode for the [110] oriented sample. This matches the deformation microstructures observed experimentally for both ME3 and ME501 at this temperature. Therefore, when a wide range of channel width values are present in a given microstructure, the DAD analysis should be used in combination with the percolation simulation in order to make an accurate prediction of dislocation activity.

Figure 3.14: DAD is calculated by using (a) average channel width (210 nm) and (b) the critical channel width for de-correlation (156 nm).

Dislocation de-correlation appears to be very temperature and stress dependent. Indeed, de-correlation appears only in a small window of stress and temperature around 700 C. 91 Moreover, de-correlation appears to favor orientations at or near <001> as was observed in this study. As was discovered in the ME501 polycrystalline creep tests at around 725 C grain boundary behavior, such as grain boundary sliding, becomes the dominant creep mechanism, significantly reducing creep strength for polycrystals above this temperature.

Another observation from the creep tests was the dissolution of tertiary  precipitates after the 815 C tests. For example, tertiary  precipitates in ME501 were found to dissipate from 3.3% to .4% area fraction after the [001] 815 C creep test. In addition to the increase in temperature, this tertiary  dissolution may help explain the change in deformation mechanism from precipitate shearing to climb bypass. Lastly, at high stress low temperature tests APB shearing is clearly dominate.

3.5 Conclusions

This study explored the effects orientation and alloying have on deformation activity and the subsequent macro-scale creep behavior in Ni-base disk alloys. Multiple single crystal compression creep tests were conducted in different stress and temperature regimes for two different orientations; [001] and [110] for ME3 and ME501. From these tests and the following STEM deformation analysis the following conclusions can be made.

(i) In compression creep the [110] orientation is superior to the [001] orientation due

to the formation of stacking fault ribbons in [110] and microtwinning/isolated

faulting in [001] orientations.

92 (ii) Above 815 C the creep anisotropy between the two orientations disappears along

with differences in active deformation. All deformation analysis found climb

bypass active at this temperature.

(iii) Dislocation de-correlation is a unique deformation mode found in a small

temperature and stress regime and only in favorable orientations. Dislocation

activity diagram calculations found channel widths to be a significant factor in

whether dislocations will de-correlate in the  channels.

(iv) Polycrystals creep at comparable rates to single crystals of ME3 at 700 °C, but

much faster than single crystals at 760 °C, indicating the onset of grain boundary

dominated creep behavior in this temperature range.

In summary, these results prove grain orientation and alloy composition have a profound effect on deformation modes and subsequent creep behavior and these effects need to be incorporated into future modelling efforts.

References

[1] R.R. Unocic, G.B. Viswanathan, P.M. Sarosi, S. Karthikeyan, J. Li, M.J. Mills,

Mechanisms of creep deformation in polycrystalline Ni-base disk superalloys,

Mater. Sci. Eng. A. 483–484 (2008) 25–32. doi:10.1016/j.msea.2006.08.148.

[2] Y. Yuan, Y. Gu, C. Cui, T. Osada, Z. Zhong, T. Tetsui, et al., Influence of Co content

on stacking fault energy in Ni–Co base disk superalloys, J. Mater. Res. 26 (2011) 93 2833–2837. doi:10.1557/jmr.2011.346.

[3] E.S. Huron, R.C. Reed, M.C. Hardy, M.J. Mills, R.E. Montero, P.D. Portella, et al.,

Controlling the deformation mechanism in disk superalloys at low and intermediate

temperatures, Superalloys 2012. (2012) 35–42.

[4] L. Kovarik, R.R. Unocic, J. Li, P. Sarosi, C. Shen, Y. Wang, et al., Microtwinning

and other shearing mechanisms at intermediate temperatures in Ni-based

superalloys, Prog. Mater. Sci. 54 (2009) 839–873.

doi:10.1016/j.pmatsci.2009.03.010.

[5] G.B. Viswanathan, S. Karthikeyan, P.M. Sarosi, R.R. Unocic, M.J. Mills,

Microtwinning during intermediate temperature creep of polycrystalline Ni-based

superalloys: mechanisms and modelling, Philos. Mag. 86 (2006) 4823–4840.

doi:10.1080/14786430600767750.

[6] H. Z. Deutchman, P.J. Phillips, N. Zhou, M.K. Samal, S. Ghosh, Y. Wang, M.J.

Mills, Deformation mechanisms coupled with phase field and crystal plasticity

modeling in a high temperature polycrystalline ni-based superalloy, Superalloys

2012 (12th Int. Symp. Superalloys). (2012) 25–33.

[7] L. Kovarik, R.R. Unocic, J. Li, M.J. Mills, The intermediate temperature

deformation of ni-based superalloys : importance of reordering, JOM. 61 (2009) 42–

48.

[8] S. Karthikeyan, R.R. Unocic, P.M. Sarosi, G.B. Viswanathan, D.D. Whitis, M.J.

Mills, Modeling microtwinning during creep in Ni-based superalloys, Scr. Mater.

54 (2006) 1157–1162. doi:10.1016/j.scriptamat.2005.11.049.

[9] M. Kolbe, The High Temperature Decrease of the Critical Resolved Shear Stress in

94 Nickel-Base Superalloys, Mater. Sci. Eng. (2001) 383.

[10] T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman, et

al., Segregation and η phase formation along stacking faults during creep at

intermediate temperatures in a Ni-based superalloy, Acta Mater. 100 (2015) 19–31.

doi:10.1016/j.actamat.2015.08.053.

[11] S. Raujol, M. Benyoucef, D. Locq, P. Caron, F. Pettinari, N. Clement, et al.,

Decorrelated movements of Shockley partial dislocations in the γ-phase channels of

nickel-based superalloys at intermediate temperature, Philos. Mag. 86 (2006) 1189–

1200. doi:10.1080/14786430500254685.

[12] R.R. Unocic, N. Zhou, L. Kovarik, C. Shen, Y. Wang, M.J. Mills, Dislocation

Decorrelation and Relationship to Deformation Microtwins During Creep of a γ’

Precipitate Strengthened Ni-Based Superalloy, Acta Mater. 59 (2011) 7325–7339.

[13] P.J. Phillips, R.R. Unocic, L. Kovarik, D. Mourer, D. Wei, M.J. Mills, Low cycle

fatigue of a Ni-based superalloy: Non-planar deformation, Scr. Mater. 62 (2010)

790–793. doi:10.1016/j.scriptamat.2010.01.044.

[14] K. Kakehi, Tension/compression asymmetry in creep behavior of a Ni-based

superalloy, Scr. Mater. 41 (1999) 461–465. doi:10.1016/S1359-6462(99)00191-8.

[15] K.R. Bain, D.P. Mourer, R. DiDomizio, T. Hanlon, L. Cretegny, A.E. Wessman,

United States Patent, US 8.992,699 B2, 2015.

[16] M.D. Abramoff, P.J. Magalhaes, S. Ram, Image Processing with ImageJ,

Biophotonics Int. 11 (2004) 36–44.

[17] P.J. Phillips, M.C. Brandes, M.J. Mills, M. De Graef, Diffraction contrast STEM of

dislocations: Imaging and simulations, Ultramicroscopy. 111 (2011) 1483–1487.

95 doi:10.1016/j.ultramic.2011.07.001.

[18] K. Kakehi, Effect of Plastic Anisotropy on the Creep Strength of Single Crystals of

a Nickel-Based Superalloy, Metall. Trans. A. 31 (2000) 421–430.

[19] C.M.F. Rae, R.C. Reed, Primary creep in single crystal superalloys: Origins,

mechanisms and effects, Acta Mater. 55 (2007) 1067–1081.

doi:10.1016/j.actamat.2006.09.026.

[20] R. Srinivasan, G.F. Eggeler, M.J. Mills, Gamma Prime cutting as rate-conrolling

recovery process during high-temperature and low-stress creep of superalloy single

crystals, Acta Mater. 48 (2000) 4867–4878.

[21] M. Yamashita, K. Kakehi, Tension/compression asymmetry in yield and creep

strengths of Ni-based superalloy with a high amount of tantalum, Scr. Mater. 55

(2006) 139–142. doi:10.1016/j.scriptamat.2006.03.048.

[22] D.M. Knowles, Q.Z. Chen, Superlattice stacking fault formation and twinning

during creep in gamma/gamma prime single crystal superalloy CMSX-4, Mater. Sci.

Eng. A. 340 (2003) 88–102.

[23] V.A. Vorontsov, C. Shen, Y. Wang, D. Dye, Shearing of γ’ Precipitates by a<112>

Dislocation Ribbons in Ni-Base Superalloys: A Phase Field Approach, Elsevier.

(n.d.).

[24] C.M.F. Rae, N. Matan, R.C. Reed, The role of stacking fault shear in the primary

creep of [001]-oriented single crystal superalloys at 750°C and 750 MPa, Mater. Sci.

Eng. 300 (2001) 125–134.

[25] T.M. Smith, B.D. Esser, N. Antolin, A. Carlsson, R.E.A. Williams, A. Wessman, et

al., Phase Transformation Strengthening of High Temperature Superalloys,

96 Accepted. Nature Communications (2016).

[26] V.A. Vorontsov, L. Kovarik, M.J. Mills, C.M.F. Rae, High-resolution electron

microscopy of dislocation ribbons in a CMSX-4 superalloy single crystal, Acta

Mater. 60 (2012) 4866–4878. doi:10.1016/j.actamat.2012.05.014.

[27] C.M.F. Rae, L. Zhang, Primary creep in single crystal superalloys: some comments

on effects of composition and microstructure, Mater. Sci. Technol. 25 (2009) 228–

235. doi:10.1179/174328408X369311.

[28] A. Ma, D. Dye, R.C. Reed, A model for the creep deformation behaviour of single-

crystal superalloy CMSX-4, Acta Mater. 56 (2008) 1657–1670.

doi:10.1016/j.actamat.2007.11.031.

[29] R.R. Unocic, L. Kovarik, C. Shen, P.M. Sarosi, Y. Wang, J. Li, et al., Deformation

Mechanisms in Ni-Base Disk Superalloys at Higher Temperatures, Superalloys

2008 (Eleventh Int. Symp. (2008) 377–385.

doi:10.7449/2008/Superalloys_2008_377_385.

[30] H. Roth, C. Davis, R. Thomson, Modeling solid solution strengthening in nickel

alloys, Metall. Mater. …. 28 (1997) 1329–1335.

http://link.springer.com/article/10.1007/s11661-997-0268-2.

[31] R. Unocic, On the Creep Deformation Mechanisms of an Advanced Disk Nickel-

Base Superalloy, The Ohio State University, 2008.

Chapter 4 - Segregation and η Phase Formation Along Stacking Faults During Creep at Intermediate Temperatures in a Ni-Based Superalloy

This chapter is a modified version of publication [3]

4.1 Introduction

Polycrystalline nickel-based superalloys are essential materials for disks in the hot section of jet turbine engines due to their high strength and microstructural stability at elevated temperatures. Until recently, the fatigue performance of polycrystalline turbine disks has been a primary property of importance. As the operating temperature and engine hold times increase, developing a complete understanding of creep deformation becomes imperative. Recent studies have shown that the prominent deformation modes are strongly temperature-dependent. Operation of ½<110> dislocations in the  matrix and cutting of the  precipitates by pairs of ½<110> dislocations (linked by an antiphase boundary or

APB) is observed predominantly following lower temperature (400-600°C) testing [1–3], for which the temperature dependence of the alloy strength is small. Stacking fault shearing and microtwinning become active at higher temperatures (600-800°C), particularly under low strain rate or creep conditions [4–6]. Understanding the rate-limiting process for these higher temperature deformation modes is important for creating more accurate models of creep behavior and for improving the temperature capability of future alloys.

The mechanistic reason for the onset of time- and temperature-dependence at intermediate temperatures is not obvious since both stacking faults and microtwins are created by the shear movement of partial dislocations on close-packed {111} planes. For 98 example, Chen and Knowles [7,8] proposed a mechanism for creation of superlattice

1 extrinsic stacking faults (SESFs) based on the passage of two different ⁄3<112>

1 superpartials on adjacent {111} planes. The ⁄3<112> superpartials are created by the interaction of two like ½<110> dislocations at the / interface:

1 ̅ 1 ½[110] + ½ [110] = ⁄3 [121] + ⁄3 [211] (1)

There are several concerns with this mechanism: (a) the strong repulsive forces between

1 the parent dislocations that would resist their interaction, (b) the two dissimilar ⁄3<112> dislocations cannot simultaneously experience a large Schmid factor driving the shearing event, and (c) since this is a pure shear process, it should not be strongly temperature dependent.

In early work by Kear, et al. [9] and more recently by Decamps et al. [10,11], a mechanism for the creation of intrinsic and extrinsic faults in L12 precipitates was proposed that involves individual ½<110> dislocations initiating shearing of the  precipitate, thereby forming an antiphase boundary (APB). This high energy fault is presumed to

1 create a favorable situation for nucleation of a ⁄6<112> Shockley partial on the plane of

1 (or adjacent to) the APB, and is equivalent to the net shear of a ⁄3<112> dislocation, which is a partial dislocation in the L12 superlattice. Kear [9] proposed that the nucleation of the partial loop could yield a temperature dependence. However, analysis by Zhou et al. [12] has criticized the Decamps model since the nucleation of the Shockley partial loop

αB on the pre-existing APB would require overcoming a very large activation barrier.

The earliest proposed mechanism for forming superlattice extrinsic stacking faults

(SESFs) was put forward by Kear et al. [13] who postulated that a the combination glide 99 of three differently signed Shockley partials could create the D024 crystal structure of the two layer SESF in an L12 precipitate. Soon after, they introduced the possibility that like- sign Shockley partial pairs could create a SESF; however, this scenario required the interaction of a dipole displacement near the dislocation core [14]. The introduction of the dipole displacement was their attempt to account for the wrong nearest neighbors that would be created in the ordered L12 structure due to the movement of Shockley partials.

However, this concept of a dipole occurring at or near the dislocation core was not developed further and is not supported by more recent high resolution TEM analysis of partial dislocation configurations [15–17].

More recently [18], the concept of “reordering” has been suggested as a rate- limiting process under the assumption that stacking faults are created by the movement of combinations of 1/6<112> Shockley partials rather than 1/3<112> superpartials. For instance, Decamps et al. [19] proposed that the decorrelation of two like ½<110> type

1 dislocations at the interface leads to shearing of the precipitates by ⁄6<112> Shockley partials. These partials form a high energy, complex stacking fault (CSF) during the shearing process. Therefore, if two partials shear a precipitate on adjacent planes, forming a two-layer CSF, the reordering process described by Kolbe et al. [18], and later in more detail by Kovarik et al. [20] could occur, converting the high energy, two-layer CSF into a low energy SESF. Since reordering is a local, diffusion-mediated process, it could potentially be responsible for the onset of time- and temperature-dependent behavior at intermediate temperatures. Indeed, Kovarik, et al. used DFT calculations to explore possible diffusion pathways for reordering, and the associated activation barriers [3]. A

100 simple model of deformation based on reordering as the rate-limiting process was developed by Karthikeyan, et al. [21].

These previous models have been challenged recently by several studies that have revealed the presence of compositional fluctuations at stacking faults in both Ni-base [22] and Co-base [23] superalloys–studies enabled by the advent of advanced, thin-foil, energy dispersive spectroscopy (EDS) capabilities. These results have lent important support to the earlier observation of increased intensity of atomic column positions at stacking faults in alloy ME3 [6] and CMSX-4 [24]. For the Ni-base alloys, the indication from initial EDS analysis is that the composition at superlattice intrinsic faults (SISFs) within the  tend toward that of the FCC  matrix. In the case of the Co-base alloys, the proposed explanation by Titus, et al. [20] is that the local DO19 structure of the SISF is tending toward that of the thermodynamically favored Co3Ti phase. Thus, Titus, et al. [20] argue that the shearing process is the result of a displacive-diffusive transformation at the fault.

The present chapter focuses on an exploratory Ni-base superalloy, and provides the first structural and chemical analysis of superlattice extrinsic stacking faults (SESFs), which are the most prominent shearing mode for this alloy under creep conditions at intermediate temperature. An alternative mechanism is proposed for forming the partials that enable  shearing involving unlike ½<110> matrix dislocations. Definitive evidence for another type of local displacive-diffusive phase transformation, from faulted  to the η phase, is also provided for the first time. In addition, evidence for the formation of a distinctive solute atmosphere surrounding the leading partials of the SESF is also obtained for the first time. In order to explore the origin of this mechanism and confirm that it is sensible, we perform Density Functional Theory (DFT) [25] calculations and find that the 101 observed segregation behavior of the solutes can be explained by a changed bonding environment around the SESF. The atomic structure used in the DFT calculations is validated by simulated HAADF images. Together, these results provide further evidence for the importance of long-range diffusional processes in association with shearing of  particles at intermediate temperatures.

4.2 Materials and Experimental Methods

4.2.1 Sample Preparation

A single crystal analog to a disk alloy ME501 with minor compositional variations relative to the commercial alloy ME3 (with increases in Ta, Hf and Nb and decreases in

Co and Cr as compared to ME3. Refer to Table 3.1 for full comparison) was obtained from

GE Global Research Center in the form of a single, large casting after a heat treatment that formed a bi-modal  precipitate microstructure. Before testing, microstructure analysis on

ME501 was conducted to obtain volume fraction and average size for the secondary and tertiary precipitates. Samples were polished progressively down to a 1200 fine grit using

alloy was then etched with a solution of 2mL hydrofluoric acid, 30mL nitric acid, and

50mL lactic acid that preferentially etched the  precipitates. Using an FEI Sirion scanning electron microscope (SEM), backscattered electron micrographs of the alloy’s microstructure were obtained and then analyzed using ImageJ [26]. For statistical significance, multiple images were obtained and over a thousand particles were included in the analysis. Figure 4.1 shows an example of the microstructure present in ME501. The combined volume fraction of the secondary and tertiary  was approximately 55.5%. The

102 average secondary precipitate diameter was 300nm and the tertiary precipitate size was approximately 30nm.

Figure 4.1: Back-scatter SEM image of the microstructure of ME501 where the  precipitates have been etched away (black) from the  channels (grey).

4.2.2 Creep Sample Orientation and Preparation

Sections were prepared from the bulk single crystal casting and etched with a solution of Walkers etchant to validate that the casting was indeed a single crystal.

Orientation Imaging Microscopy (OIM) was used to determine the orientation of the bulk crystal prior to extracting compression samples. Surfaces were again polished to a 0.05 um colloidal silica finish. An [001] oriented single crystal silicon sample was used for calibration for improved orientation accuracy.

Compression cubes with a 1:1:2.5 dimension ratio were extracted from the bulk crystal using Electrical Discharge Machining (EDM). The sides of the individual compression cubes were then polished to a 1200 fine grit finish to reduce the damage layer created from the cutting process. A monotonic compression creep test was performed on an [001] oriented sample at 760°C and 552MPa until 0.5% plastic strain was reached.

103 Linear variable displacement transducers (LVDTs) were used on an MTS 810 compression cage to directly record the displacement of the compression plattens. Temperature was recorded using two K thermocouples. After the desired plastic strain was reached, the test was stopped and the specimen was very quickly returned to room temperature using a fan.

The compression sample remained under load in order to preserve the deformation substructure that was present at the end of the test, and to minimize and changes to local structure and chemistry associated with dislocations and stacking faults. The maximum cooling rate was 4°C/sec from 760°C to 450°C.

4.2.3 Microscopy Methods

Post creep samples were prepared by polishing off the oxidized layer and then extracting transmission electron microscopy (TEM) foils normal to the compression axis using an FEI Helios Nanolab Dualbeam 600 focused ion beam (FIB). Samples were thinned at 5kV and then further cleaned using a Fischione Nanomill. Thin foils were analyzed using low annular angle dark field (LAADF) diffraction contrast on an FEI Tecnai

F20 field emission 200kV STEM. Dislocation analysis using STEM has many advantages over conventional TEM. It was shown by Phillips et al. [27–29] that conventional g⋅b invisibility rules still apply for STEM and therefore images can be interpreted in the same fashion as conventional TEM (CTEM) images. STEM images exhibit decreased fringe and bend contour contrast and better signal to noise ratios when compared to CTEM [30,31].

Both traditional two beam conditions and higher-order g imaging were employed in the present dislocation analysis [27].

In CTEM, zone axis diffraction contrast imaging is rarely employed due to the low contrast images that correspond to these high absorption orientations. In STEM, the image 104 intensity of the bright field (BF) and annular dark field (ADF) are averaged together allowing for high contrast defect micrographs. The ability to image on low index zone axes gives the capability to view multiple slip systems simultaneously for improved dislocation assessment.

Samples were also extracted normal to <110> orientations (longitudinal to the [001] compression axis) in order to view stacking faults edge-on using high angle annular dark field (HAADF) zone axis imaging conducted on a probe-corrected Titan3™ 80-300kV.

High resolution EDX mapping was conducted at 300 kV using an image-corrected

Titan3™ 60-300KV with a Super-X detector utilizing the Bruker Espirit software. The

Super-X EDX detection system uses four silicon drift detectors that are located radially around the objective pole piece and specimen stage for improved collection performance.

This technique has recently been used to observe segregation along edge-on stacking faults traversing through  precipitates [23,32]. Integrated line scans were conducted and quantified through Cliff-Lorimer analysis [33] using experimental Kα energies for Ni, Co,

Al, Nb, Cr, Mo and Ti. The Cu specimen holder signal was avoided by using the Mα lines for Ta and W since the Lα Ta and W peaks corresponded too closely to a Cu peak to be accurately considered. Deconvolution for the W and Ta Mα peaks, as well as background subtraction was used to reduce the influence of Bremsstrahlung.

4.3 Results

4.3.1 Diffraction Contrast STEM

Post-creep STEM analysis illustrates that the deformation mechanisms primarily consist of precipitate shearing by stacking faults as well as dislocation build-up concentrated in the  channels. In Figure 4.2, examples of these prominent deformation 105 mechanisms are shown using [001] zone axis bright field and dark field images.

Conventional g⋅R analysis utilizing diffraction contrast STEM confirms that the observed stacking faults are primarily extrinsic in nature, lying on inclined {111} planes. This is in agreement with the observations by Decamps et al. [34]. The presence of SESFs on several

{111} plane families demonstrates that multiple slip systems are active. It is noted that the dislocation and stacking fault content is very low in the as-heated-treated condition.

Figure 4.2: (a) [001] Zone axis bright field image (b) Near [001] zone axis LAADF-STEM image of isolated stacking faults

Although most of the observed shearing events are isolated within the  precipitate, some shearing is also observed that extends through both  and . High-resolution HAADF imaging has demonstrated that these are usually microtwins and not extended SESFs.

Extensive g⋅b analysis has revealed the existence of all 6 ½<110> type dislocations; however ½[101̅] and ½[011̅] were the most prominent for the particular sample tested.

106 Figure 3 shows an example of these two dislocation types together at the / interface

(determined from more complete g⋅b analysis, not shown). In this particular example, these dislocations appear to cross each other, presumably at the interface.

Figure 4.3: LAADF-STEM images from a comprehensive g⋅b analysis of dislocation types conducted in same area near the [001] zone axis shows two different <110> type dislocations (A and B) crossing each other at the / interface where (a) g[020] and (b) g[ퟐ̅ퟎퟎ] diffraction vector is excited.

4.3.2 Segregation and Ordering along Stacking Faults

Observations along the [110] zone axis using high resolution HAADF imaging is used presently to characterize the local structure and composition variation at two-layer

SESFs in ME501. As shown in Figure 4.4, Z-contrast HAADF-STEM imaging suggests heavy element segregation due to the appearance of a characteristic, grid-like pattern of brighter atomic columns at the fault.

107

Figure 4.4: HAADF-STEM image obtained along the [110] zone axis revealing a grid-like ordering along a SESF inside a  precipitate.

A similar grid-like ordering has also been observed along SESFs in a CMSX-4 alloy crept at 750C [24]. This feature at the fault is distinctly different from the usual L12 ordering of the  precipitate which can also be observed in Figure 4 from the presence of the alternating contrast of the (001) atomic planes. The brighter planes correspond to the Al-sites where heavier elements such as Ta, Nb and W naturally replace Al; therefore, the ordering seen along the SESF is predominantly caused by segregation of higher Z elements on the Al lattice sites. Since the bright contrast present in HAADF images does not prove

108 segregation, nor quantify the segregating elements, EDX mapping was conducted on multiple SESFs in order to more completely specify the nature of the segregation.

Figure 4.5 shows the elemental EDX maps along a SESF that had vertically sheared a  precipitate.

Figure 4.5: EDX elemental map of a vertical SESF showing segregation along the fault

Two fiducial markers, approximately 2nm diameter holes, were “drilled” into the sample on each side of the SESF by positioning a converged electron beam with a high current density over a specific site for 30 seconds. These fiducial nanoholes aid the software correction for drift during the scan. EDX analysis consistently indicated Al and Ni to be deficient along SESFs. Conversely Co, Nb, Ti and Ta (and to a lesser extent W) all

109 segregated along the fault. The Cr elemental map, though not displaying segregation at the fault, does highlight the appearance of Cr rich particles (which are postulated to be  particles) averaging 2-6nm in diameter inside the secondary  precipitate. The influence of these particles on the creep strength of the alloy warrants further investigation as it is apparent that these small particles can be sheared and may be possible sources of Co within the secondary precipitates.

To quantify the compositional changes along the SESF, elemental profiles were created across the SESF, as shown in Figure 4.6. These intensity data were vertically integrated parallel to the fault in order to improve signal-to-noise, and then converted to composition using uncorrected k-factors. The results confirm the qualitative results obtained from the elemental maps. Both Ni and Al are deficient along the fault, while Ta,

Nb, and Co are enriched at the fault as compared to the bulk precipitate. Both W and Ti also segregate slightly to the fault. It is expected that Co, having similar properties to Ni, would replace Ni along the SESF [35]; whereas, Nb and Ta are known  formers that replace Al [36]. Higher resolution line scans were obtained in an attempt to determine which species were responsible for the grid-like ordering along the fault. These scans were not vertically integrated and were placed directly over the atomic columns exhibiting the higher intensity, grid-like ordering. The combined total intensities for Al and Ta from these scans are shown in Figure 4.7(b). Ta was the only element that appeared to show statistically significant peaks in its line scan that correspond to the higher Z-contrast seen in the HAADF images, while Al was deficient across the fault, although this trend is much clearer in Figure 4.6. Thus, Ta appears to be the only element to clearly correlate with the

HAADF images since it has a relatively large cross-section for ionization, thereby 110 producing a greater signal, as well as the fact that Ta is found in significantly higher concentrations along the fault (~5 wt%) when compared, for example, to Nb (~2 wt%) and

W (~1 wt%).

Figure 4.6: Vertically integrated EDS elemental profiles across SESF

111

Figure 4.7: (a) An example of an EDS line scan over the prominently segregated atomic columns. (b) The summed intensity for Ta (purple), Al (red) and the corresponding HAADF intensity profile (black) from the line scan

For improved understanding of the mechanisms responsible for SESF formation, faults terminating inside a  precipitate were also analyzed. For each terminated SESF, atomic

112 resolution HAADF-STEM images of the leading partial(s) were recorded, and a Burgers circuit analysis was performed, as indicated in Figure 4.8(a), in order to determine the type of dislocations responsible for creating the SESF.

Figure 4.8: (a) Burgers circuit analysis showing Shockley partials leading SESF shear. (b) Center of symmetry analysis on atomic peak positions from same SESF revealing a “CSF-like” fault leading the SESF.

1 Multiple examples revealed that two adjacent and identical ⁄6<112> Shockley partials were responsible for forming the SESFs. Figure 4.8(b), supports the conclusions from the Burgers circuit analysis indicating that a narrow (approximately 0.5nm wide), one-layer fault is in the lead, then transitions to the two-layer fault. Since the leading partial

1 is a ⁄6<112> partial, nearest neighbor violations must be created in their wake [10,11], necessitating a re-ordering process in order to form a true SESF with a localized D024 crystal structure.

113 Compositional EDX analysis was also conducted on these leading partials. Figure

4.9(a) shows examples of elemental maps for an SESF. Figure 4.9(b), displays the integrated elemental profiles taken across the SESF very close to its termination within the precipitate at the leading partial.

114

Figure 4.9: (a) The elemental maps of a terminated SESF (b) a vertically integrated EDX line scan across the SESF present in 10a.

115

Most remarkably, the local region around the leading partials of the SESF are enriched in

Cr, Co and Mo; whereas, the  formers (Al, Ti, Ta, and W) along with Ni, appear to be reduced near the partial. Another important observation is that the segregation present on this SESF is similar to that found on other SESFs (e.g. see Figure 4.6) far from the leading partials. This SESF exhibits segregation of Ta, Nb, Ti, and W along the fault while, Cr,

Al, and Ni were either deficient or not detectably changed. Though it may appear that Mo has segregated along the SESF, further analysis indicates that the amount of Mo estimated to have segregated is within the error in measurement.

4.4 Discussion

4.4.1. SESF Formation

The experimental results have coupled diffraction-contrast and HAADF imaging with state-of-the-art EDX analysis to provide important new insights into the shearing of precipitates by stacking faults. To begin, the frequently observed interaction of different

½<110> dislocations along the / interface combined with the presence of isolated SESFs, leads to a new hypothesis for the formation of isolated SESFs, which expands on the models of Kear and Decamps for this process [11,37]. Figure 4.10(a) and Figure 4.10(b) below illustrates this new model which is analogous to that presented recently by

Voronstov, et al [24] describing SISF formation:

116

Figure 4.10: (a) Two different full dislocations interact at the /. (b) Leading, like-signed Shockley partials enter the precipitate forming a two-layer CSF, which re-orders (red circular arrow) to form a lower energy SESF. (c) Scenario incorporating new observations of both Cr and Co atmosphere surrounding the leading partials of the fault and the formation of η phase immediately trailing the partials.

The Decamps et al. [19] model assumes that the de-correlation of a ½<110> type

1 dislocation at the interface can lead to the shearing of the precipitate by a ⁄6<112>

Shockley partial. This creates a CSF in the precipitate, which is energetically unfavorable

1 and hence another ⁄6<112> Shockley partial is assumed to nucleate and shear the plane

1 above the CSF. In essence, this is equivalent to the shear of a ⁄3<112> super-Shockley partial leaving behind a low energy SESF. However, their model ignores the nearest neighbor violations that would be created, and thus a two layer CSF would in fact be the

117 result. Kolbe [10] realized that the wrong-nearest neighbors at such a two-layer CSF could be eliminated by local diffusion (i.e. re-ordering), thereby creating an SESF.

Figure 4.10, illustrates how this same scenario could occur involving two different ½

<110> dislocations. As stated earlier, the two most prominent dislocations observed at the

/ interface were ½[101̅] (BA) and ½[011̅] (BC). In Figure 4.10(a), both the BA and BC dislocations have become immobilized at the interface on parallel (111) planes. The

Shockley partial 1/6[112̅] (b), which experiences the highest Schmid Factor (0.47) for a single crystal with a [001] stress direction, will shear the precipitate creating the reactions in Equation 2 and the scenario illustrated in Figure 10b.

̅ 1 ̅̅̅̅ 1 ̅ ½[101] ➝ ⁄6 [211] + CSF + ⁄6 [112] (2)

̅ 1 ̅ ̅ 1 ̅ ½[011] ➝ ⁄6 [121] + CSF + ⁄6 [112]

This will form a two layer CSF as described by Kolbe[18]. The operation of these partials has been supported through Burger circuit analysis, as shown in Figure 4.8(a).

Rordering must occur in order to convert the two-layer CSF to an SESF and enable further extension of the fault [18]. This reordering mechanism, illustrated in Figure 4.10 and discussed in detail in Kovarik et al. [3] is potentially rate controlling. Local diffusion is required for reordering to occur. Kovarik, et al. utilized VASP calculations to explore the possible diffusion paths, as well as the energy barriers associated with vacancy formation and migration near the SESF. It was found that the activation energy for reordering should

118 be very similar to that for Ni self-diffusion. In this previous work, the effect of segregation and shear-induced phase transformation was not considered.

An important new observation from the present work is the prominent Cr and Co segregation around the leading partials. The presence of a solute atmosphere of these elements is important in several respects. First, since these are elements favored in the γ

(FCC) matrix, a higher concentration of them at the tip of the fault should significantly lower the CSF energy trailing the first Shockley partial, as well as the two-layer CSF trailing the passage of the second Shockley partial. Indeed, in the limit of a local FCC structure due to the high Cr and Co content, the faults trailing the partials would simply be intrinsic and extrinsic stacking faults containing no wrong nearest-neighbors, and consequently will have much lower energy than their CSF counterparts. The observation of the solute atmosphere therefore brings into question the concept of reordering as a rate- limiting process. Instead, the movement the leading partials may be constrained by the diffusive motion of the solute atmosphere. This possibility has recently been proposed by

Titus, et al. [23] in the context of SISF propagation through  precipitates in Co base superalloys, although no direct evidence for a solute atmosphere was presented in their work.

4.4.2 η phase formation

It will now be shown that the segregation and structural features at the extrinsic faults within the  are actually consistent with the formation of a local η phase. Numerous experimental studies have demonstrated that formation of the  phase, exhibiting the hexagonal close packed (HCP) D024 crystal structure, has detrimental effects on the 119 strength and durability of Ni-based superalloys. It has been shown that changing an alloy’s composition with respect to certain elements will promote the growth of  phase relative to the formation of , to the detriment of creep strength [38–42]. Bouse et al. [43] hypothesized that the  phase volume fraction will increase as the ratio of Ti+Ta+Hf+Nb relative to Al increases. Indirect support of this hypothesis can be derived based on the effect of carbon in both PWA1480 and IN792+Hf alloys which promotes the formation of carbides involving  phase formers (Hf, Ta, Nb, and Ti). It was found that  phase formation was greatly reduced for both alloys. They further hypothesized that at lower temperatures a solid-state transformation from  to  phase can occur. In fact, it was concluded by Zhao et al [44] through thermodynamic calculations that  was not an equilibrium phase above 750°C and would transform to the η phase, which was in equilibrium for Nimonic 263. Asguri et al. [36] proposed that the passage of partials through  precipitates, and the consequent stacking sequence changes, could lead to a favorable local environment for  phase formation. This scenario is reasonable since the formation of an SESF inside a  precipitate will locally form the DO24 crystal structure that the  phase possesses[13,45]; however, no direct evidence for the connection between deformation-induced stacking faults in the precipitates and the formation of  phase has been reported prior to the present work.

Recently, variations in the structure and chemistry of the  phase have been recognized, where ordering of elements on the Al sites has been deduced based on high angle annular dark field (HAADF) Z-contrast imaging and diffraction simulations.

Pickering et al. [46] reported the first instance where Nb atoms were ordering to certain Al

120 sites, leaving Ti and Al to occupy randomly the other sites. Figure 4.11 shows a schematic

Figure 4.11: The structure of the Ni6AlNb reportedly seen by Pickering et al. [23] in superalloy 718 plus.

As noted by Voronstnov et al. [24], the grid-like ordering observed along SESFs in

CMSX-4, and now in the present results on ME501, is similar to the local Z-contrast ordering observed for the η phase found by Pickerington et al. [46] in Figure 4.11. The segregation revealed in the present EDX analysis is in agreement with this hypothesis.

Numerous studies have reported Ta, Nb and Ti segregating to η phase at the expense of the

 phase [39,40,43,45,47]. Concurrently the absence of Cr and the reduction of Al along the faults aligns well with past chemical studies on η [39,40,48]. These observations, combined with the findings of Zhao et al.,[44] leads to the hypothesis that the η phase has nucleated along the two-layer faults analyzed presently. The only difference between the η phase found in ME501 and the one established by Pickerington et al. [46] is the presence of Ta on the Wyckoff 2a sites along with Nb. Furthermore, the local D024 crystal structure along the SESF should be a local nucleation site for the η phase as suggested by Asguri et

121 al. [45]. Indeed, no evidence for “bulk” η phase has been found at other length scales. It is still unknown what importance this  ➝ η phase transformation has on the shearing event, but the presence of segregation and local ordering along the SESFs terminated inside the  precipitate indicates it may be rate-limiting, as suggested schematically in Figure 10c above.

4.4.3 Density Functional Theory Simulations

The DFT code VASP [25] was used to examine if the dependence of the chemical potentials of the solutes on ordering in the SESF would provide the necessary driving force for the observed segregation behavior, and also to explore whether the ordering is driven by strain-minimization or by chemical-bonding effects.

Cells containing 160 atoms were created consisting of 10 FCC-type {111} planes with Ni3Al-type occupation of the lattice sites incorporating a SESF. Calculation cells contained solute atoms of Co, Nb, Mo, Ta, and W and combinations thereof, with compositions listed in Table 4.1. As suggested by experiment, Co occupied Ni superlattice sites, while all other solutes were placed on Al superlattice sites. This composition most closely matched the EDX data from the fault. In order to validate these choices, HAADF simulations as described in Sec. 4.4.4 were performed which reproduced the observed

HAADF images very well, thus supporting the chosen structure. Structural relaxations were performed using a quasi-Newtonian algorithm with electron exchange treated with the generalized gradient approximation (GGA) including spin polarization in the formulation of Perdew, Burke, and Ernzerhof [49]. All calculations were performed using a 6x3x5 Monkorst-Pack k-mesh with plane wave energy cutoffs at least 30% greater than the highest specified in the pseudopotentials used. For each composition calculated, two 122 cells were relaxed: one with the solute atoms forming an η-phase at the defect boundary, and a second in which solutes were randomly dispersed at substitutional sites in the structure. The structures were first relaxed internally with no shape or volume change in the calculation cell, followed by expansion of the calculation cell perpendicular to the fault to minimize the stress in that direction. Energies of the internally relaxed cells as well as the out-of-plane lattice expansion percentages are listed in Table 4.2.

Table 4.1: Compositions of 160-atom SESF calculation cells

System Composition at%Co at%Nb at%Mo&W &Ta Nb 103Ni-22Al-17Co-18Nb 10.6 11.3 0 Mo 103Ni-22Al-17Co-18Mo 10.6 0 11.3 NbMo 103Ni-22Al-17Co-10Nb-8Mo 10.6 6.3 5.0 Ta 103Ni-22Al-17Co-18Ta 10.6 11.3 0 W 103Ni-22Al-17Co-18W 10.6 0 11.3 TaW 103Ni-22Al-17Co-10Ta-8W 10.6 6.3 5.0 NbTa 103Ni-22Al-17Co-10Nb-8Ta 10.6 11.3 0

Table 4.2: Relaxation results of 160-atom SESF calculation cells with segregated and randomized solute atoms

E (eV) Out of plane expansion System Segregated solute Randomized Segregated solute Randomized solute solute Nb -1003.20 -1002.55 1.37% 1.61% Mo -1005.31 -1004.96 0.28% 0.35% NbMo -1004.72 -1004.01 0.78% 1.01% Ta -1042.12 -1041.64 1.02% 1.28% W -1045.83 -1045.78 0.29% 0.36% TaW -1044.46 -1043.90 0.61% 0.85% NbTa -1020.48 -1019.85 1.20% 1.47%

Figure 4.12 and Figure 4.13 show the energy difference and the percentage out-of- plane expansion difference between the randomized and segregated relaxed structures,

123 respectively. Energy differences suggest that the presence of the group 5 solutes (Nb and

Ta) will result in strong segregation effects of solute to the SESF, even for solutes (e.g. W) that have very small driving force for segregation on their own. These energetics are consistent with the chemistry observed by EDX (Sec. 3.2). While both energy and strain are higher in the presence of group-5 solutes, differences in out-of-plane expansion do not correlate strongly to energy differences across atomic periods, leading us to conclude that the energy differences are not solely due to a strain effect that results from the localization of solute atoms. This is supported by studying the distribution of bond lengths in both structures: the average in both the segregated and randomized structures for atoms on each plane parallel to the SESF differs by less than 1% from the equilibrium bond length in

Ni3Al, and the standard deviation and range of bond lengths differs consistently at the limits of the SESF structure in both the segregated and randomized calculation cells.

Figure 4.12: Energy difference between the randomized and segregated SESF calculation cells in eV

124

Figure 4.13: Percent difference in out-of-plane expansion in randomized and segregated SESF calculation cells

In order to determine if there is a chemical difference in bonding between calculation cells with solutes concentrated in the SESF and those with randomized solute positions, which should lead to electron rearrangements, Bader partial charges were calculated and plotted as a function of atomic position relative to the planar defect [50].

Due to the limited number of solute atoms, it is most practical to study these effects on the most abundant species in the structures: Ni. Plotted in Figure 4.14 are the planar average

Bader partial charges on Ni atoms as a function of atomic position for each of the solute systems. The red dash line represents the average Bader partial charge of the bulk structure.

In these plots, we see that although the average Bader charge on Ni across the entire system changes insignificantly between the randomized and segregated calculation cells, there are significant changes in Ni charges between the SESF and the bulk phase when solutes are segregated to the SESF planes. From calculation, we find that the Bader charge on Ni in bulk Ni3Al is ~0.61e, which agrees well with the value we find in the bulk phase when solutes are confined to the SESF planes. On the other hand, the charge on Ni is significantly smaller at the fault, indicating that the bonding is more “metallic” locally, and thus more

125 tolerant to the presence of wrong (i.e. different from the correct neighbors in the ordered intermetallic) nearest neighbors and alloying species. Thus, chemical (bonding) energy gains come both from the decreased distortion of the ideal Ni3Al-like charge transfer outside of the SESF (maximized charge) and decreased penalty from charged-atom interactions for the “wrong” bonding environment in the SESF (minimized charge).

Figure 4.14: Planar average Bader charges on Ni as a function of atomic position for segregated and randomized calculation cells. The red dash line is the Bader partial charge of the bulk. 126 4.4.4 HAADF Simulations

In order to validate the DFT-modeled structure, and thus further support the hypothesis that η phase is forming along the SESF, HAADF-STEM images of the SESF were simulated using the μSTEM program, based on the quantum excitation of phonons

(QEP) model [51]. In the QEP model, a quantum mechanical approach is taken, whereby momentum and energy transfer are included through inelastic scattering: specifically through phonon excitation.

The crystal structure used as input for the image simulation was the relaxed Nb-Ta system described in Sec. 4.4.3 with out-of-plane expansion and solute atoms segregated to the SESF. As seen in Figure 4.15, the Ta and Nb atoms were segregated to the Al sites, while Co was segregated to the Ni sites within the SESF. The structure used was 8 atomic planes thick in the <110> direction and repeated to the appropriate thickness based on the experimental sample to allow for enough randomization of the solute positions, while still balancing computational demands. The black box represents the area sampled for the

HAADF simulation shown in Figure 4.16.

127

Figure 4.15: [110] view of the segregated SESF structure in L12. Ta and Nb are segregated to Al sites, and Co to Ni sites. [52] The black box represents the area sampled for the HAADF simulation shown in Figure 4.16.

The foil thickness was determined to be roughly 35nm thick in the region of interest by comparison between experimental and simulated position averaged convergent beam electron diffraction (PACBED) patterns [53]. To ensure accurate representation of the experimental data in the simulations, the microscope variables from the FEI Titan 80-300

S/TEM with a probe corrector were used. Specifically, a convergence semi-angle of

12mrad, Cs coefficient of 2μm, and C5 of 1mm were used at a zero defocus. Finally, a detection range from 55 to 372mrad was used, which corresponds to HAADF image acquisition using a camera length of 91mm on this microscope.

In order to minimize wrap-around errors during Fourier transforms, a 6×6 tiling of unit cells was used. The structure was sampled at discrete probe positions determined by

128 twice the maximum spatial frequencies allowed by the probe-forming aperture. Images were interpolated via Fourier padding to provide a result that was both aesthetic and representative without changing the simulated data. The temperature dependent mean square displacement term, 〈푢2〉, was described by values of the elemental crystals of each species, as parameterized by Gao et al. [54].

From Figure 4.16, it can be seen that the fault structure, along with the effect of elemental segregation in the simulated and experimental images are in close agreement.

The experimental HAADF image shown is the result of superimposing 12 periodic units of the images parallel to the fault in order to improve the signal-to-noise in the image for comparison with the simulations. The experimental background is noticeably higher than that of the simulation, which could be due to factors such as amorphization at the surface or a thicker experimental sample than that which was simulated.

The excellent agreement between the observed and simulated images for the atomic structure based on the EDX-results, which has been found by the VASP simulations to be energetically favorable, provides additional validation for the η phase structure, including the elemental segregation and sublattice positions previously described. The similar intensity ratios between the high-Z and low-Z atomic columns for both the simulated and experimental cases establishes that the observed ordering is not a by-product of strain or diffraction contrast along the SESF. Therefore, the HAADF simulation confirms that the high intensity grid-like ordering along SESFs is a product of Ta and Nb segregating to those specific Al-sites forming the η phase revealed in Figure 4.11.

129

Figure 4.16: (a) Experimental and (b) simulated HAADF image of SESF (c) Experimental and (d) simulated averaged intensity of grid-like ordering outlined by the dashed lines

4.5 Conclusions

In this chapter, we have examined the to-date little understood rate-controlling processes in Ni-base superalloys during creep deformation at intermediate temperatures by investigating local compositional and structural changes occurring in association with stacking faults. Specifically, a constant stress compression creep test was conducted on an

[001] oriented single crystal analogue of a disk superalloy at 760°C, and the following conclusions can be drawn:

(i) Based upon STEM diffraction contrast analysis following creep testing, shearing of

1 precipitates occurs predominantly by the passage of two identical ⁄6<112> Shockley partials on consecutive {111} planes. This creates two layer extrinsic faults that are isolated to the  particles. 130

(ii) High spatial resolution EDX mapping confirmed elemental segregation of Co, Ta, Nb, and Ti along superlattice extrinsic stacking faults, and depletion of Ni and Al.

(iii) A distinctive solute atmosphere, rich in Cr and Co and lean in Ni and Al, surrounds the leading Shockley partials of SESFs, indicating that diffusive movement of this solute atmosphere must be associated with the shearing process.

(iv) HAADF-STEM imaging revealed ordered contrast along SESFs consistent with a phase change from faulted  to η (DO24 structure). Using high resolution EDX analysis, reduction in Al and Ni and an increase in Co, Ta, Nb, and Ti is consistent detected. In fact,

Ta is prominent on the Wyckoff 2a sites resulting in the grid-like contrast observed using

HAADF-STEM. In addition, the same elemental segregation and local ordering was found present in cases where the SESF had terminated in the precipitate.

(v) VASP calculations, validated by HAADF image simulations, confirm that the observed structure along the SESF, consistent with the structure of the η phase, and the observed segregation of solutes are energetically favorable. For the validated structure, a decrease in electron transfer between the atoms was found at the SESF, thus facilitating enhanced accommodation of solutes, which in turn should stabilize the observed structure.

In combination, these suggest that a displacive-diffusive phase transformation from

  η can occur along SESFs, and may indeed represent a novel, rate-limiting process during creep deformation.

131 References:

[1] Hirsch PB. Philos Mag A 1992;65:569.

[2] Pope DP, Vitek V. Acta Mater 1984;32:435.

[3] Phillips PJ, Unocic RR, Kovarik L, Mourer D, Wei D, Mills MJ. Scr Mater

2010;62:790.

[4] Lerch B, Gerold V. Acta Met 1985;33:1709.

[5] Unocic RR, Kovarik L, Shen C, Sarosi PM, Wang Y, Li J, Ghosh S, Mills MJ.

Superalloys 2008 (Eleventh Int Symp 2008:377.

[6] Kovarik L, Unocic RR, Li J, Sarosi P, Shen C, Wang Y, Mills MJ. Prog Mater Sci

2009;54:839.

[7] Knowles DM, Chen QZ. Mater Sci Eng A 2003;340:88.

[8] Chen QZ, Knowles DM. Mater Sci Eng A 2003;356:352.

[9] Kear BH, Oblak JM, Giamei AF. Metall Trans 1970;I:2477.

[10] Hirth JP, Lothe J. Theory of Dislocations, 2nd ed. New York: Wiley; 1968.

[11] Décamps B, Morton AJ, Condat M. Philos Mag A 1991;64:641.

[12] Zhou N, Shen C, Mills MJ, Li J, Wang Y. Acta Mater 2011;59:3484.

[13] Kear BH, Giamei A, Silcock J, Ham RK. Scr Metall 1968;2:287.

132 [14] Kear BH, Giamei A, Leverant GR, Oblak JM. Scr Metall 1969;3:455.

[15] Kanchanbagh PO. Acta Met Mater 1991;39:1515.

[16] Rae CMF, Matan N, Reed RC. Mater Sci Eng 2001;300:125.

[17] Lin D, Wen M. Mater Sci Eng A 1989;3:207.

[18] Kolbe M. Mater Sci Eng 2001:383.

[19] Décamps B, Raujol S, Coujou A, Pettinari-Sturmel P, Clement N, Locq D, Caron P.

Philos Mag A 2004:91.

[20] Kovarik L, Unocic RR, Li J, Mills MJ. JOM 2009;61:42.

[21] Karthikeyan S, Unocic RR, Sarosi PM, Viswanathan GB, Whitis DD, Mills MJ. Scr

Mater 2006;54:1157.

[22] Viswanathan G, Shi R, Genc A, Vorontsov V, Kovarik L, Rae C. Scr Mater

2015;94:5.

[23] Titus M, Mottura A, Viswanathan G, Suzuki A, Mills MJ, pollock T. Submission to

Acta Mater 2014.

[24] Vorontsov V, Kovarik L, Mills MJ, Rae CMF. Acta Mater 2012;60:4866.

[25] Kresse G, Furthmüller J. Phys Rev B 1996;54.

[26] Abramoff MD, Magalhaes PJ, Ram S. Biophotonics Int 2004;11:36.

133 [27] Phillips PJ, Brandes MC, Mills MJ, De Graef M. Ultramicroscopy 2011;111:1483.

[28] Phillips PJ, Mills MJ, De Graef M. Philos Mag 2011;91:2081.

[29] Phillips PJ, De Graef M, Kovarik L, Agrawal a., Windl W, Mills M. Microsc

Microanal 2012;18:676.

[30] Maher DM, Joy DC. Ultramicroscopy 1976;1:239.

[31] Humphreys CJ. Ultramicroscopy 1981;7:7.

[32] D’Alfonso AJ, Freitag B, Klenov D, Allen LJ. Phys Rev B 2010;81:100101.

[33] Hoeft H, Schwaab P. X-ray Spectrom 1988;17:201.

[34] Decamps B, Condat M. Scr Metall 1987;21:607.

[35] Xu J, Lin W. Phys Rev B 1993;48:4276.

[36] Blavette D, Cadel E, Pareige C, Deconihout B, Caron P. Microsc Microanal

2007;13:464.

[37] Décamps B, Raujol S, Coujou A. Philos Mag 2004;84:91.

[38] Brooks CR, Wang YM. metallography 1989;86:57.

[39] Evans ND, Maziasz PJ, Swindeman RW, Smith GD. Scr Mater 2004;51:503.

134 [40] Long F, Yoo YS, Jo CY, Seo SM, Song YS, Jin T, Hu ZQ. Mater Sci Eng A

2009;527:361.

[41] Mukherji D, Strunz P, Genovese DDEL, Gilles R, Rösler J, Wiedenmann A. Metall

Trans A 2003;34:2781.

[42] Pollock TM, Rene N. J Propuls Power 2006;22:361.

[43] Bouse GK. Superalloys 1996 (Eighth Int Symp 1996.

[44] Zhao J, Ravikumar V, Beltran AM. Metall Trans A 2001;32:1271.

[45] Asgari S, Sharghi-moshtaghin R, Sadeghahmadi, Mehdi Pirouz P. Philos Mag

2013;93.

[46] Pickering EJ, Mathur H, Bhowmik A, Messé OMDM, Barnard JS, Hardy MC,

Krakow R, Loehnert K, Stone HJ, Rae CMF. Acta Mater 2012;60:2757.

[47] Zhao S, Xie X, Smith GD, Patel SJ. Mater Sci Eng A 2003;355:96.

[48] Zhao J-C, Henry MF. JOM Appl Technol 2002;54:37.

[49] Perdew JP, Burke K, Ernzerhof M. Phys Rev Lett 1996;77:3865.

[50] Sanville E, Kenny SD, Smith R, Henkelman G. J Comput Chem 2007;28:899.

[51] Allen LJ, D’Alfonso AJ, Findley SD. Ultramicroscopy 2014.

135 [52] Images and video generated using CrystalMaker®: a crystal and molecular

structures program for Mac and Windows. CrystalMaker Software Ltd, Oxford E

(www. crystalmaker. com)

[53] Lebeau JM, Findlay SD, Allen SJ, Stemmer S. Ultramicroscopy 2010;110:118.

[54] Gao HX, Peng LM. Acta Crystallogr 1999;A55:926.

136

Chapter 5 – Diffusion Processes During Creep at Intermediate Temperatures in Ni-based Superalloys

5.1 Introduction

Ni-base Superalloys are an essential material used primarily in the hot section of jet turbine engines. With the creation of each new generation of turbine engine the goal remains the same: increase the operating temperature of the engine thereby, reducing CO2 emissions and fuel costs. This increase in temperature promotes a change in deformation modes in Ni-base superalloys that are still not quite understood. Multiple studies have found that a shift from a-thermal APB shearing towards re-order mediated precipitate shearing occurs during creep at temperatures above 700C [1–5]. One of these precipitate shearing modes is microtwinning, which has been found to adversely affect creep properties in Ni-based disk superalloys [6–9]. Thus, improved understanding of the rate- limiting mechanisms responsible for the formation of microtwins is needed to further improve the high temperature properties of these superalloys.

Mechanical microtwinning was first reported in a waspalloy alloy by Guimier and

Strudel [10]. Interestingly, it was found that a similar alloy tested using comparable parameters but without the presence of  precipitates did not experience twinning. This revealed that  precipitates played an important role in the formation of microtwins [11].

Later work found that microtwinning was orientation dependent, occurring in [110] and

[111] oriented single crystals under tensile creep and [001] oriented crystals under

137 compression creep [4,9,11–13]. Knowles and Chen, found that microtwinning occurred in orientations that also promoted superlattice extrinsic stacking fault (SESF) formation, implying the formation of both SESFs and microtwins were connected. This was supported in future studies; however, early work incorrectly predicted that these faults were created by adjacent a/3<112> super-Shockley partials shearing the  precipitates [12,14]. Rather, later experimental evidence revealed that SESFs and microtwins were created by like-sign a/6<112> Shockley partials shearing on adjacent {111} planes [2,13][15]. In order for this shearing process to end with a low energy SESF, a reordering process, first predicted by

Kolbe, must occur in the fault created behind the shearing Shockley partials [16]. Kovarik et al. expanded upon Kolbe’s theory, using density functional theory analysis, and revealed an energetically favorable diffusion step process that could remove wrong nearest-neighbor violations converting the high energy two layer CSF into a low energy SESF[17]. Later,

Smith et al., first reported the presence of a Co and Cr Cottrell atmosphere around shearing

Shockley partials in a  precipitate, which they predict is necessary to further reduce the energy of two layer CSF and promote precipitate shearing [18].

Kovarik et al. also predicted the presence of segregation along SESFs and microtwin interfaces using atomic resolution high angle annular dark field (HAADF) imaging [17]. Indeed, elemental segregation has been recently observed along superlattice intrinsic (SISF) and extrinsic faults in both Co and Ni based superalloys after intermediate temperature creep [18–20]. Smith et al. discovered that the type of elemental segregation along SESFs controlled whether the fault would extend into a twin or not, thereby affecting the overall creep properties of the alloy. They predicted that the formation and extension of microtwins were reliant on the segregation of Co and Cr along the twin’s interface.

138 Furthermore, Barba et al. [21], using high resolution atom probe, found unequivocally the presence of Co and Cr along microtwins in a Ni-base superalloy, with higher concentrations along the twin interface. They also predicted that the presence of a Co and

Cr Cottrell atmospheres surrounding the Shockley partials shearing the  precipitate to thicken the twin.

The purpose of this study is to provide new insights into the formation of microtwins and the diffusional processes that necessitate it. Using high resolution Super-

X EDX and density fuctional theory calculations, a new microtwin formation model is presented highlighting the role of elemental segregation. Other observations include the presence and effect that newly discovered tertiary  particles inside secondary  precipitates have on creep performance as well as new evidence of Cottrell atmospheres around shearing Shockley partials inside  precipitates. Together, these new insights will help the creation of improved deformation models and ultimately improves high temperature capabilities in disk superalloys.

5.2 Experimental Methods

5.2.1 Creep Sample and Testing

Again, a single crystal analog of the currently used commercial disk alloy ME3 and

ME501 was obtained from the GE research center. The composition of both alloys can be found in Table 3.1. After a heat treatment that resulted in a bimodal  precipitate distribution, an [001] oriented rectangular prisms with a 1:1:2.5 dimension ratio were extracted using electrical discharge machining (EDM). The sides of the sample were then polished to a 1200 fine grit using SiC polishing pads to remove the subsequent damage

139 layer. A 414 MPa and 552MPa (ME3 and ME501 reprectively) monotonic compression creep test was performed at 760 C on the [001] oriented samples in order to promote reorder mediated precipitate shearing modes; specifically, microtwinning and isolated faulting. The test was performed using a MTS 810 Compression creep frame with two linear variable displacement transducers (LVDTs) to record plastic strain. Both tests were ended when about 0.5% plastic strain was reached and the specimens were quickly fan- cooled to room temperature in order to capture the deformation and microstructure present at the end of the test as well as to minimize diffusional changes during cool-down.

5.2.2 Microscopy and Chemical Analysis

Post creep, the samples were again polished to a fine 1200 grit using SiC polishing paper and then followed by a .05 um colloidal silica finish. In order to confirm that

Microtwinning had occurred during the ME3 creep test, electron channeling contrast imaging (ECCI) was conducted using a FEI Sirion scanning electron microscope (SEM) while utilizing the backscatter detector at high accelerating voltages. ECCI allows for large areas to be imaged for improved statistical deformation analysis [22,23]. To validate the

ECCI results, which found microtwinning to be the prominent deformation mode, transmission electron microscopy (TEM) samples were created using a FEI Helios Nanolab

Dualbeam 600 focused ion beam (FIB). The FIB foils were further cleaned to remove ion damage using a Fischione Nanomill at 900eV. First foils were extracted orthogonal to the compression axis to confirm the orientation of the sample and for STEM analysis. These foils were analyzed using low angle annular dark field (LAADF) detectors on an FEI

Tecnai F20 STEM at 200 kV and again found evidence for microtwinning.

140 For atomic resolution HAADF imaging and high resolution EDX, [110] oriented

FIB foils were extracted from the post compression creep samples in order to image the microtwins and superlattice stacking faults (SFFs) edge on. HAADF Imaging was conducted on a FEI probe-corrected Titan3 80–300 kV STEM. While the EDX was performed on an image-corrected Titan3 60–300 kV with Super-X detector technology utilizing the Bruker Esprit software. All quantified and summed EDX spectra were done using cliff-lorimer k factors and the Esprit software.

5.3 Results

5.3.1 Evidence of Co and Cr Rich Cottrell Atmospheres Ahead of Stacking Faults in ME3

To further explore and confirm the existence of Cottrell atmospheres around shearing Shockley partials, both SISFs and SESFs terminating inside a  precipitate were sought out. Figure 5.1 shows an example of both a SESF and SISF terminating inside a  precipitate in ME3.

141

Figure 5.1: Atomic resolution HAADF image of a SESF (upper fault) and SISF (lower fault) terminating inside a  precipitate. The red ovals represent where each fault has ended and the hole was purposely burnt into the sample to help with EDS scan correction.

In Figure 5.1, the red rectangles represent where each fault is, with the upper fault being a

SESF and the lower one a SISF. The red ovals highlight the areas that the faults have terminated. This area is more easily recognized when imaged in a medium angle annular dark field (MAADF) condition that highlights diffraction contrast and strain instead of Z- contrast. The electron beam hole was purposely burnt into the sample to act as a fiducial marker and help the esprit software with the drift correction during the EDS scan. High resolution EDX mapping was conducted in the area shown in Figure 5.1 and the results are revealed below in Figure 5.2.

142

Figure 5.2: Elemental maps of the terminated SESF and SISF inside a  precipitate

Several Observations can be made from the elemental maps in Figure 5.2.

Importantly, the segregation along both the SISF and SESF appear to be similar, with Co and Cr segregating to the fault replacing Al and Ni. Another observation, is the presence of Co and Cr-rich Cottrell atmospheres around the shearing Shockley partials for both faults. The observed Cottrell atmosphere ahead of the SISF is particularly significant as its the first time a Cottrell atmosphere has been observed without the influence of a tertiary  particle nearby, which can be seen most prominently in the Cr map near the SESF. To further examine the extent of segregation in the Cottrell atmospheres and SISF, the quantified elemental concentrations of the four areas highlighted in Figure 5.3 are given in

Table 5.1 below.

143

Figure 5.3: Four regions highlighted to compare concentration values. 1 = region of  precipitate, 2 = SISF, 3 = SISF Cottrell atmosphere, and 4 = SESF Cottrell atmosphere.

Table 5.1: The quantified concentrations of the four regions highlighted in Figure 5.3.

Element Area 1: Area 2: Area 3: Area 4:  precipitate SISF SISF Cottrell SESF Cottrell Atmosphere Atmosphere Al 4.24  0.2 3.90  0.2 3.79  0.2 3.50  0.2

Ni 70.79  2.2 69.93  2.2 65.20  2.1 63.69  2.0

Co 11.75  0.4 12.95  0.4 15.52  0.6 16.07  0.6

Cr 1.92  0.1 2.53  0.1 5.13  0.2 6.08  0.3

Ti 6.48  0.2 6.53  0.2 5.75  0.3 5.44  0.2

Nb 0.77  0.1 0.47  0.1 0.43  0.1 0.95  0.2

Mo 0.58  0.1 0.68  0.1 1.14  0.2 1.29  0.2

W 1.28  0.3 1.05  0.2 1.18  0.3 1.25  0.3

Ta 2.18  0.5 1.96  0.4 1.86  0.4 1.73  0.4

144 The following trends are observed in Table 5.1: first, all segregation appears to show an increase in  formers (Co, Cr and Mo) and a decrease of the  formers (Ni, Al,

Ti). However, no change in composition is observed for Nb, Ta, and W. The greatest amount of segregation occurs around the Cottrell atmospheres of both the SISF and SESF.

In fact, the weight percentage of Cr in the leading SISF Cottrell atmosphere is over double that found along the SISF (2.53 wt% compared to 5.13 wt%).

5.3.2 Evidence of Co and Cr Rich Cottrell Atmospheres Ahead of Stacking Faults in ME501

The same analysis was conducted in the alloy ME501. Both a SESF and SISF terminating inside a precipitate  were found and high resolution EDX scans were conducted. The SESF was discussed in detail in chapter 4. For the SISF, the results from the high resolution EDX scan is presented in Figure 5.4.

145

Figure 5.4: High resolution EDS scan of a SISF terminating inside a  precipitate in ME501.

The presence of a prominent CO and Cr Cottrell atmosphere matches the results found in

ME3. In this case, it does appear that the Cottrell atmosphere is near a tertiary  particle but the contrast is much more prominent than that seen in the particle located in the upper right of the map.

EDX maps were also taken in the area where the SESF and SISF entered the precipitate to see if any evidence of diffusion processes were also noticed there. Below are the results for the SESF and SISF in ME501.

146

Figure 5.5: Elemental Cr maps of notches where (a) a SESF and (b) a SISF enter into a precipitate.

In both cases, a notch is observed at the / interface where the two stacking faults entered into the precipitate. In Figure 5.5(a), a SESF has entered into the precipitate in the area near a notch though it is unclear is this “notch” existed before the fault and caused the dislocation to interact there or if the dislocations interacting at that location created the notch. In Figure 5.5(b), it is very clear that the notch formation is the consequence of having dislocations interacting near the interface. Figure 5.5(a) and Figure 5.5(b) represent large diffusion processed occurring at fault formation where different shaped notches formed. In fact, recently a study found that this difference in notch shape could be used to predict the type of dislocations responsible for the stacking fault [24].

5.3.3 Evidence of Co-Rich and Cr-Rich Cottrell Atmospheres Along Thickening Twin Boundaries

Below in Figure 5.6, is an example of a twin being thickened by two atomic planes through the shear of adjacent 1/6<112> Shockley partials. 147

Figure 5.6: An atomic resolution MAADF image of a twin boundary extending by the shear of adjacent Shockley Partials.

In Figure 5.6, as has been reported in earlier literature [2,15,17], the twin is being thickened by the cooperative shear of adjacent 1/6<112> Shockley partials. The red oval represents the area where the shearing Shockley partials are located and the change of where the twin interface is, which is represented by the red lines. High resolution EDX mapping was also conducted on this area and the results are shown below in Figure 5.6. The red ovals in

Figure 5.6 and Figure 5.7 represent the same area, which is where the twin thickening

Shockley partials are located at the microtwin/ interface.

148

Figure 5.7: Elemental maps of a twin extending by two atomic planes.

Figure 5.7 reveals for the first time a prominent Co and Cr Cottrell atmosphere around the twin thickening Shockley partials shearing alongside a twin interface. Subtle segregation along the twin interface can also be seen with what appears to be an increase of Co and Cr in conjunction with a depletion of Ni and Al. In contrast, there does not appear to be a uniform increase of Co and Cr segregation through the whole thickness of the twin.

The rest of the elemental maps are not shown due to high noise levels and/or no evidence of segregation. The composition around the partials, again highlighted by the red oval in

Figure 5.7, were quantified and compared to the composition of the  as shown below in

Table 5.2.

149 Table 5.2: The quantified EDX compositions of the Cottrell atmosphere around the twin thickening Shockley partials compared to the composition of the  precipitate.

Element  precipitate Twin Cottrell Atmosphere Al 4.24  0.2 3.13  0.1 Ni 70.79  2.2 62.35  1.9 Co 11.75  0.4 17.21  0.6 Cr 1.92  0.1 6.91  0.2 Ti 6.48  0.2 5.32  0.2 Nb 0.77  0.1 0.77  0.1 Mo 0.58  0.1 1.19  0.2 W 1.28  0.3 1.24  0.3

Again, significant Co and Cr segregation around the microtwin thickening

Shockley partials is evident. In fact, the concentration is similar to those found around the

SISF and SESF Cottrell atmospheres.

5.3.4 Evidence of Segregation Along Microtwins

To better understand the amount and type of segregation present inside a microtwin, low magnification EDX maps was taken of a  precipitate with a microtwin that had sheared through the center of it. The results are shown below in Figure 5.8.

150

Figure 5.8: Elemental maps of microtwin shearing through a  precipitate. The presence of tertiary  is evident in the Co and Cr maps.

Some noteworthy observations can be seen in Figure 5.8. The presence of tertiary

 particles can be seen in the Co and Cr elemental maps. Instead of these particles being present throughout the bulk of the precipitate there appears to be a denuded zone along the edge of the precipitate where they most likely dissipated during the heat treatment and creep test. Also shown in these maps is the presence of Co and Cr segregation along the microtwin, yet, the segregation does not appear uniform. In the region where the tertiary 

151 particles are present the twin has Co and Cr along the interface and inside the twin region, though the intensity of the Co and Cr signal inside the twin is not consistent. In the region of the precipitate without the tertiary  particles, the twin doesn’t appear to have any segregation at all. To confirm if this is accurate a higher magnification EDX scan was taken in the denuded tertiary  zone of the precipitate represented by the white box shown in the

HAADF image in Figure 5.8 with the results revealed in Figure 5.9.

Figure 5.9: High resolution elemental maps of twin shearing a  precipitate through the denuded zone area highlighted in Figure 6.

152 Though there did not appear to be any segregation along the twin in the denuded tertiary  zone of the precipitate along the twin in the low magnification EDX scan shown in Figure 8, the higher resolution EDX scan reveals that this was not the case. Figure 5.9 shows that Co, Cr and possibly Mo segregation is present along the twin interface but a lack of segregation inside the twin. This is better characterized through integrated line scans across the twin as shown below in Figure 5.10.

Figure 5.10: (a) Combined Cr, Co, Ni, and Al elemental map of microtwin near the / interface. The red box represents where the integrated line scan was performed. (b) The results of the integrated line scan. The interface of the twin can easily be discerned by the segregation of Co, Cr, and Mo and depletion of Ni and Al from the interface.

The combined elemental map in Figure 5.10(a) better illustrates the segregation present along the twin interfaces. The red box represents the area used for the integrated line scan, the results of which are shown in Figure 5.10(b). The Integrated line scan shows significant chemical segregation along both microtwin interfaces with little to no segregation inside the twin. This is in contrast to the segregation seen along SESFs and

SISFs as well as previous work done on microtwin segregation [18,21].

153 5.4 Discussion

5.4.1 New Twin Formation Model

From the new EDX mapping results, a new model on twin formation in Ni-base superalloys can be created, inspired by previous models proposed by smith et al. [4] and Barba et al.

[21] Below in Figure 5.11 is a diagram detailing the new model.

Figure 5.11: The Twin formation model recently described by Smith et al. [4], where two unlike ½<110> dislocations interact at the interface and dissociate (b) so like-signed Shockley partials can shear into the  precipitate forming a SESF. (c) The same process occurs again alongside the SESF to form a 3 layer twin inside the  precipitate. (d) This model has now been updated to include the diffusion processes observed in the results section and by Barba et al. (e) where  formers Co, Cr and possibly Mo segregate along the formed SESF. (f) Then when the process occurs again Some of the  formers segregate to the new twin interface from the  channel while the other source of the  formers come from what segregated along the SESF, leaving a pristine  state inside the twin boundaries.

The new microtwin formation model presented in Figure 5.11 includes the diffusion processes now known to occur during twin formation and extension. The proposed mechanics for the twin formation (ie. the type of dislocations and interactions) has not

154 changed. It’s still assumed that the formation of a SESF created by the interaction of two unlike sign 1/2 <110> dislocations interacting at the / interface, is the precursor to microtwin formation as shown in Figure 5.11(a) and Figure 5.11(b). This same process then can occur alongside the SESF to form a true three-layer twin, as shown in Figure

5.11(c). It been recently revealed that the segregation of  formers Co, Cr, and Mo along the SESF promotes further shearing along the fault through the removal of possible nearest neighbor violations and consequentially microtwin formation. [nature] This type of segregation has also observed for ME3 prior to this study and further confirmed in Figure

2, and is now represented in the model as shown in Figure 5.11(d) and Figure 5.11(e). As proposed by Barba et al., our findings confirm the existence of Co and Cr-rich Cottrell atmospheres around the leading Shockley Partials, both for the SESF and the twin extending partials here represented in the model as light blue ovals. In contrast to the Barba model, new EDX evidence did not show Co and Cr segregation through the bulk of the twin and instead only along the twin interface. As represented in Figure 5.11(e), the bulk of the twin returns to a -like composition as the segregation is no longer necessary, after sufficient time, as the re-ordering process has been completed resulting in a true twin atomic structure. Therefore, the  former segregates diffuse to the area of the new twin boundary where the reordering process is still in process to help lower the fault energy and allow the Shockley partials to continue shearing through the precipitate. This leaves a twin with only segregation along its boundaries. This diffusion process should lower the energy of the fault in less time than waiting for more  formers to diffuse from the  channels as the length of diffusion length scales are much shorter.

The reason that Barba et al. [21] may have reported segregation throughout the bulk 155 of microtwins may be due to the other source of  formers shown to segregate to the twin boundaries, the tertiary  particles inside the  precipitates. In Figure 5.8 and Figure

5.12(b), segregation of Co and Cr are shown inside the bulk twin near the area of the precipitate where tertiary  particles are present. This excess presence of  formers near the twin interface means that the segregation of Co and Cr from the area of the old interface is not necessary to lower the new fault energy and remains inside the reordered twin. The end result is a twin that appears to have  formers segregated distributed non-uniformly through the thickness of the twin except for the part of the twin present in the tertiary  denuded zone, where only the twin interfaces have segregation. This is represented in detail below in Figure 5.12(a) and Figure 5.12(b).

156

Figure 5.12: (a) A MAADF image of a Twin shearing both a  precipitate and  channel near the / interface. (b) A Cr map showing the two sources of  former segregation along the twin and twin interfaces. Inside the tertiary  denuded zone the source of the segregates is the  channel, however, in the area of the precipitate where there exist tertiary  particles the primary source of  former segregation along the twin is from those particles.

The abundance and distance of both  former sources leads to variation of segregation along the microtwin. The area of the microtwin that is sheared through the tertiary  denuded zone of the  precipitate primarily gets its source of  formers from either the  channels or the segregation that had previously occurred earlier along the fault

157 (SESF or old twin boundary). In the region of the  precipitate where the tertiary  particles are present, the microtwin has a diffuse segregation of Co and Cr along the inside of the twin as well as, though the amount is not constant and varies depending on the distribution of tertiary  particles the twin interacts with.

5.4.2 Tertiary  particles

The presence of the denuded tertiary  particles near the / interface inside the precipitate implies that the small gamma particles are able to diffuse back into the  channels when given enough time at high enough temperatures. In order to examine this process a piece of ME3 and ME501 was heat treated at 815C for 100 and 200 hours, after which EDX maps were performed to observe the changes in the tertiary  particle sizes and morphologies. Below in Figure 5.13, shows the elemental Cr maps for both alloys after each heat treatment.

158

Figure 5.13: The result of the heat treatment study on tertiary  particles in both ME3 and ME501. The particles were found to dissipate quicker in ME3 but after 200 hours most of the tertiary particles were gone for both alloys.

Surprisingly, there were observed differences in stabilities of the tertiary  particles in both alloys. The gamma particles in ME501 were still present after a 100 hour heat treatment at

815C while almost all of the particles were dissipated in ME3. The particles were also observed to coarsen in ME501 after 100 hours. This difference may have to do with the higher amounts of Nb and Ta decreasing diffusion rates inside the  particles for ME501 compared to ME3. After 200 hours, there was no evidence of tertiary particles in ME3 and in most cases ME501. However, as shown in Figure 5.13 there still appears to be some tertiary particles that exist in the middle of the precipitate. It is unclear at this moment if the presence of more stable tertiary gamma particles has any effect on the creep strength between both alloys but it has already been shown that these particles do play a role in the 159 deformation mechanisms present at these temperatures. Also, examined at the same time were the tertiary  volume fractions after each heat treatment for both alloys. Interestingly, the results were opposite to those just discussed for the tertiary  particles, with the tertiary

 precipitates more stable in ME3 compared to ME501. Table 5.3 below shows the measured tertiary  volume fractions after each heat treatment.

Table 5.3: The tertiary  volume fractions after heat treated at 815C for 100 and 200 hours in ME3 and ME501

Alloy 0 hours at 815ºC 100 hours at 815ºC 200 hours at 815ºC

ME501 3.3% .69% 0%

ME3 3-4% 1.24% 0%

After 100 hours, there was almost twice the volume fraction of tertiary  precipitates in ME3 compared to ME501 (1.24% vs. .69%) even though the volume fractions were close to equal before the heat treatment. After 200 hours, all tertiary  precipitates had dissolved for both alloys. This observation does allow for the ability to test the effect that both tertiary  and  particles may have on creep properties. By heat treating

ME3 to 100 hours a microstructure devoid of tertiary  particles but with tertiary  precipitates can be tested. While heat treating ME501 a little more than 100 hours, a microstructure with tertiary  particles but devoid of tertiary  precipitates can be tested.

160 5.4.3 Tertiary Gamma Particle Effects

To remove the confounding effects that the tertiary  have on segregation events inside  precipitates, ME3 samples were aged for 300 hours at 815C. This ageing step removed all the tertiary  and  particles, as is shown in Figure 5.13. [001] and [110] oriented samples were then crept under 414 MPa of stress at 760C to compare the creep properties of the new overaged samples to the original tests. The results are shown below in Figure 5.14.

Figure 5.14: Compression creep curves of overaged ME3 vs original ME3. Tested under 414MPa of stress at 760C.

As shown in Figure 5.14, the overaged ME3 samples, without the tertiary  and  particles, performed poorly compared to the original ME3 microstructures. Interestingly, both orientations still retained the overall creep curve shape exhibited by the original tests. The same process to extract TEM foils from the original creep samples was then employed for the [001] overaged sample. As shown in Figure 5.15(a), the deformation observed in the

161 overaged sample exhibited many of the same mechanisms as observed for the original

[001] oriented sample crept at 760C. For example, numerous stacking faults and twins were observed throughout the TEM foil. One notable difference was an increase frequency of observed SISFs in the overaged samples. One of these SISFs can be seen in Figure

5.15(b), where it has terminated inside of a  precipitate.

Figure 5.15: (a) An HAADF image of deformation observed in the [001] overaged ME3 post creep. (b) The elemental Cr map clearly illustrating the complete dissolution of tertiary  particles.

The presence of a terminated fault inside a tertiary  free precipitate allows for the observation of segregation and Cottrell atmospheres without the convolution of the surrounding  particles. Below is the EDX map of the terminated SISF from Figure 5.15(a).

162

Figure 5.16: The HAADF, Al, Cr, and Co maps of a terminated SISF observed in the overaged ME3 sample post-test.

This new EDX maps shown in Figure 5.16 proves that the observed Cottrell atmospheres in the original EDX maps are in fact real and not an artifact from the surrounding tertiary

 particles. Indeed, the Cottrell atmosphere exhibits a very noticeable “comma” shape that has not been observed in the other EDX maps. The segregation along the fault also reveals that the tertiary  particles are not major contributors of the segregated  formers (Co and

Cr) noticed along the faults in ME3. Finally, this means that we have three primary mechanisms that must be considered in determining what may be the rate-limiting process for fault and twin formation; Cottrell atmospheres, reordering, or segregation/phase transformations along faults.

163 5.4.4 Rate Limiting mechanisms: Preliminary Discussion

Currently very little work has been done exploring the diffusivity of elements inside a  precipitate, therefore, Thermocalc [25] was employed to estimate the diffusivities of five different elements using the composition of ME3 at 760C.

Table 5.4: The calculated diffusivities of Cr, Co, Mo, Ta, and Al inside a  precipitate at 760C

Element Diffusion Coefficient m2/s

Cr 2.5e-19 Co 2.7e-19 Mo 1.7e-19 Ta 5.0e-19 Al 4.4e-17

Using the above calculated diffusivities in Table 5.4, estimated specie velocities can be determined to decide if reordering (Al diffusion) or phase transformation along the fault

(Ta diffusion for  phase or Cr/Co diffusion for  phase) is the more rate-limiting process.

Recently, an equation was presented by Titus et al. [26] that estimates the velocity of a

Cottrell atmosphere surrounding dislocations and is shown below.

(휏푏)퐷푘푇 휐 = 2 (1) 2.1퐶0훽

Here  is the relative shear stress on the atmosphere (195.3MPa), D is the Diffusivity of Cr presented in Table 5.4, k is the Boltzman constant, T is 1033K, C0 represents 5at% Cr segregation (4푒27 푚−3), and  is the interaction energy between the solute and the

Shockley partial dislocations (1.6푒−29 푁푚2). Considering a 5% Cr enriched Cottrell

164 atmosphere and using the equation above, all three mechanism speeds (reordering, Cottrell atmospheres, and fault phase transformations) can be estimated.

Table 5.5: Comparison of rate limiting mechanisms

Process Rate (nm/s) Cottrell atmosphere (5at% Cr) .07 η phase transformation (Ta Diffusion) .7  phase transformation (Cr and Co diffusion) .5 Re-ordering (Al diffusion) 7

Comparing the calculated velocities revealed in Table 5.5, notable differences can be found between the three processes. The mechanism that occurs the fastest is reordering due to the fast diffusivities of Al and Ni inside  precipitate. Second, the diffusion of Ta, Co, and Cr, the main elements responsible for the phase transformations along faults, are an order of magnitude slower than the reordering element of Al. Therefore, it appears that the reordering process will occur first, immediately after the faults are formed, followed by the segregation/phase transformation mechanism. Lastly, the Co and Cr rich Cottrell atmosphere is apparently the slowest and the principle rate-limiting mechanism associated with fault formation. This initial conclusion makes sense due to the increased complexity of the process (i.e. dislocation shearing along with diffusion processes) as well as a larger amount of elemental segregation needed. The segregation inside the Cottrell atmospheres was consistently double that found along faults and twin interfaces (refer to Table 5.1).

165 5.5 Conclusions

This study investigated the diffusive and rate controlling processes during creep in a Ni-base superalloy at intermediate temperatures using high resolution EDX mapping and

DFT analysis. By investigating the local compositional changes occurring during fault and microtwin formation, new insights were obtained and the following conclusions could be drawn.

(i) A prominent solute atmosphere of Co and Cr replacing Ni and Al exists around  shearing Shockley partials for both SISF, SESF and Twin formation and need to be incorporated in new deformation models

(ii) A new microtwin formation model has been proposed incorporating the diffusive and segregating processes that are now understood to exist and promote microtwinning.

(iii) Tertiary  particles inside the  precipitates have been shown to play a significant role in providing  formers necessary to promote twin formation and thickening.

(iv) Overaged microstructures without the presence of tertiary  and  particles consistently performed poorly compared to the original samples.

(v) The Co and Cr rich Cottrell atmospheres were determined to be the rate limiting process during fault formation.

166 In combination, these new conclusions suggest that diffusion processes during intermediate temperature creep play a pivotal role in twin formation and therefore the alloys overall creep properties.

References:

[1] M. Legros, N. Clément, P. Caron, A. Coujou, In-situ observation of deformation

micromechanisms in a rafted γ/γ’ superalloy at 850 °C, Mater. Sci. Eng. A. 337

(2002) 160–169. doi:10.1016/S0921-5093(02)00037-0.

[2] G.B. Viswanathan, S. Karthikeyan, P.M. Sarosi, R.R. Unocic, M.J. Mills,

Microtwinning during intermediate temperature creep of polycrystalline Ni-based

superalloys: mechanisms and modelling, Philos. Mag. 86 (2006) 4823–4840.

doi:10.1080/14786430600767750.

[3] G.B. Viswanathan, P. Sarosi, D. Whitis, M. Henry, M.J. Mills, Investigation of

Creep Deformation Mechanisms at Intermediate Temperatures in René 88 DT

Superalloys, Acta Mater. 53 (2005) 3041–3057.

[4] T.M. Smith, R.R. Unocic, H. Deutchman, M.J. Mills, Creep Deformation

Mechanism Mapping in Nickel Base Disk Superalloys, Mater. High Temp. 33

(2016) 372–383.

[5] I.S. Kim, B.G. Choi, H.U. Hong, Y.S. Yoo, C.Y. Jo, Anomalous deformation

behavior and twin formation of Ni-base superalloys at the intermediate

temperatures, Mater. Sci. Eng. A. 528 (2011) 7149–7155.

doi:10.1016/j.msea.2011.05.083.

[6] D.M. Knowles, S. Gunturi, The role of 〈112〉{111} slip in the asymmetric nature

167 of creep of single crystal superalloy CMSX-4, Mater. Sci. Eng. A. 328 (2002) 223–

237. doi:10.1016/S0921-5093(01)01688-4.

[7] K. Kakehi, Influence of Secondary Precipitates and Crystallographic Orientation on

the Strength of Single Crystals of a Ni- Based Superalloy, Metall. Trans. A. 30

(1999).

[8] B. Décamps, A.J. Morton, M. Condat, On the mechanism of shear of γ′ precipitates

by single (a/2)⟨110⟩ dissociated matrix dislocations in Ni-based superalloys, Philos.

Mag. A. 64 (1991) 641–668. doi:10.1080/01418619108204866.

[9] M. Yamashita, K. Kakehi, Tension/compression asymmetry in yield and creep

strengths of Ni-based superalloy with a high amount of tantalum, Scr. Mater. 55

(2006) 139–142. doi:10.1016/j.scriptamat.2006.03.048.

[10] A. Guimier, J.L. Strudel, Stacking fault formation and mechanical twinning

processes in a nickel-base superalloy during tensile deformation at high temperature,

in: Int. Conf. Strength Met. Alloy. Conf. Proc., 1970.

[11] M.G. Ardakani, M. McLean, B. a. Shollock, Twin formation during creep in single

crystals of nickel-based superalloys, Acta Mater. 47 (1999) 2593–2602.

doi:10.1016/S1359-6454(99)00145-7.

[12] D.M. Knowles, Q.Z. Chen, Superlattice stacking fault formation and twinning

during creep in gamma/gamma prime single crystal superalloy CMSX-4, Mater. Sci.

Eng. A. 340 (2003) 88–102.

[13] R.R. Unocic, N. Zhou, L. Kovarik, C. Shen, Y. Wang, M.J. Mills, Dislocation

Decorrelation and Relationship to Deformation Microtwins During Creep of a γ’

Precipitate Strengthened Ni-Based Superalloy, Acta Mater. 59 (2011) 7325–7339.

168 [14] W.W. Milligan, S.D. Antolovich, The mechanisms and temperature dependence of

superlattice stacking fault formation in the single-crystal superalloy PWA 1480,

Metall. Trans. A. 22 (1991) 2309–2318.

[15] S. Karthikeyan, R.R. Unocic, P.M. Sarosi, G.B. Viswanathan, D.D. Whitis, M.J.

Mills, Modeling microtwinning during creep in Ni-based superalloys, Scr. Mater.

54 (2006) 1157–1162. doi:10.1016/j.scriptamat.2005.11.049.

[16] M. Kolbe, The High Temperature Decrease of the Critical Resolved Shear Stress in

Nickel-Base Superalloys, Mater. Sci. Eng. (2001) 383.

[17] L. Kovarik, R.R. Unocic, J. Li, P. Sarosi, C. Shen, Y. Wang, et al., Microtwinning

and other shearing mechanisms at intermediate temperatures in Ni-based

superalloys, Prog. Mater. Sci. 54 (2009) 839–873.

doi:10.1016/j.pmatsci.2009.03.010.

[18] T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman, et

al., Segregation and η phase formation along stacking faults during creep at

intermediate temperatures in a Ni-based superalloy, Acta Mater. 100 (2015) 19–31.

doi:10.1016/j.actamat.2015.08.053.

[19] T. pollock M. Titus, A. Mottura, G. Viswanathan, A. Suzuki, M.J. Mills, High

Resolution Energy Dispersive Spectroscopy Mapping of Planar Defects in L1 2 -

containing Co-base Superalloys, Submiss. to Acta Mater. (2014).

[20] G. Viswanathan, R. Shi, A. Genc, V. Vorontsov, L. Kovarik, C. Rae, Segregation at

stacking faults within the γ’ phase of two Ni-base superalloys following intermediate

temperature creep, Scr. Mater. 94 (2015) 5–8.

[21] D. Barba, S. Pedrazzini, A. Vilalta-Clemente, A.J. Wilkinson, M.P. Moody, P.A.J.

169 Bagot, et al., On the composition of microtwins in a single crystal nickel-

based superalloy, Scr. Mater. 127 (2017) 37–40.

doi:10.1016/j.scriptamat.2016.08.029.

[22] D.G. Coates, Kikuchi-like Reflection Patterns Obtained with the Scanning Electron

Microscope, Philos. Mag. 16 (1967).

[23] J. Ahmed, A.J. Wilkinson, S.G. Roberts, Characterizing dislocation structures in

bulk fatigued copper single crystals using electron channelling contrast imaging

(ECCI), Philos. Mag. Lett. 76 (1997) 237–245. doi:10.1080/095008397178986.

[24] M. Huang, Z. Cheng, J. Xiong, J. Li, J. Hu, Z. Liu, et al., Coupling between Re

segregation and interfacial dislocations during high-temperature, low-stress creep of

a nickel-based single-crystal superalloy, Acta Mater. 76 (2014) 294–305.

doi:10.1016/j.actamat.2014.05.033.

[25] Andersson J.O., Helander T., Höglund L., Shi P.F., and Sundman B., (2002).

Thermo-Calc and DICTRA, Computational tools for materials science. Calphad,

26, 273-312

[26] Titus M, Mottura A, Viswanathan G, Suzuki A, Mills MJ, pollock T. High

resolution energy dispersive spectroscopy mapping of planar defects in L12-

containing Co-base superalloys. Acta Materialia (2015).

170

Chapter 6 – Phase Transformation Strengthening of High Temperature Superalloys

This chapter is a modified version of publication [7]

6.1 Introduction

The relentless drive for energy efficiency in power generation and propulsion places development of high-performance materials at the forefront of materials science.

Turbine engine efficiency and reduction in carbon emissions are directly related to engine operating temperature. With increasing temperatures, materials start to plastically deform under load, a process known as creep, which eventually sets the most severe limits to materials performance.[1] Therefore, increased performance in aircraft engines and land- based power generators require the development of a new generation of high-temperature structural materials that are resistant to creep. Among these materials, Ni-based superalloys are presently enabling with their unique combination of creep, fatigue and corrosion resistance.1 Superalloys have a face centered cubic (fcc), solid solution matrix (γ phase) with coherent precipitates (γ′ phase) of the Cu3Au structure which constitute around 50 volume percent of the microstructure. The γ′ phase provides superb resistance against shearing via lattice dislocation movement, and thus remarkable strength at temperatures as high as 700 °C – a crucial capability for turbine disk components.

Currently three different strengthening mechanisms are understood and employed to improve the high temperature performance of alloys: solid solution strengthening,

171 precipitation hardening, and grain boundary strengthening. Previous studies have explored how to maximize the potential from all three of these “classical” strengthening mechanisms. Since the characterization of the γ′ phase in Ni-base superalloys by Bradley and Taylor in 1937,[2,3] the development of high temperature alloys has proceeded in an incremental fashion with new progress focusing directly on the shortcomings of the previous generation of alloys. Understanding the effect of specific elements in the compositionally complex superalloys remains a qualitative and highly empirical endeavor.

While significant advances have been made in the prediction of microstructures and phase stability based on thermodynamic and kinetic databases,[4–6] the ability to predict consequent mechanical properties for a given alloy and microstructure persists as a major challenge for the materials genome initiative (MGI).[7] A significant shortcoming to computationally-directed high temperature alloy development is the lack of quantitative understanding of the detailed deformation mechanisms controlling high temperature behavior for various alloy compositions, temperatures and applied stresses.

A primary goal of the present research is to provide quantitative insight into the effect of various alloying elements on the operative deformation mechanisms under conditions that are relevant to advanced engine designs, and in alloys that are closely related to those presently used for advanced turbine disk applications. This has been achieved by application of integrated computational materials science and engineering

(ICME) involving the coupling of aberration-corrected atomic-resolution imaging with state-of-the-art energy dispersive X-ray (EDX) spectroscopy, and density functional theory

(DFT) calculations. This coupled study has resulted in the discovery of a high-temperature strengthening mechanism which we refer to as “phase transformation strengthening.” The

172 identification of this mechanism along with the new mechanistic insights represent an enabling leap forward in high temperature alloy design.

6.2 Methods

6.2.1 Microstructural characterization

This study used the single crystal analogs of the disk alloy ME3 and ME501 from the previous chapters, Again, for both alloys compositions refer to table Table 3.1. Before testing, microstructure analysis on ME3 and ME501 was conducted to obtain volume fraction and average size for secondary precipitates. After polishing with SiC pads and 0.05

m colloidal silica, each alloy was etched with a solution of 2 mL hydrofluoric acid, 30 mL nitric acid, and 50 mL lactic acid that preferentially etched the  precipitates. Using an

FEI Sirion scanning electron microscope (SEM), backscattered electron micrographs of the alloys’ microstructures were obtained and then analyzed using ImageJ.[8]

6.2.2 Compression creep sample preparation

Compression cuboids with a 1:1:2.5-dimension ratio were extracted from both bulk crystals using Electrical Discharge Machining (EDM). Monotonic compression creep tests were performed on both [001] and [110] oriented samples at 760C until 0.5% plastic strain was reached. For the ME3 tests a stress of 414 MPa was used while a stress of 552 MPa was used for the ME501 samples in order to obtain deformation under comparable strain rates (9e-5 s-1). Linear variable displacement transducers (LVDTs) were used on an MTS

810 compression cage to record the displacement of the compression cubes. Temperature was recorded using two K thermocouples. After the desired plastic strain was reached, the

173 test was stopped and the specimen was very quickly returned to room temperature using a fan. All compression samples remained under load in order to preserve the deformation substructure that was present at the end of the test, and to minimize and changes to local structure and chemistry associated with dislocations and stacking faults.

6.2.3 Electron microscopy characterization

Electron channeling contrast imaging (ECCI) was used to obtain statistically significant occurrence values for microtwinning in both alloys. Contrast from ECCI can provide information on local crystal distortions, for example, near-surface dislocations and faults such as twins.[9] Faults that appeared to shear both  and  phases were considered twins. After the oxidized layer was polished off of post-creep samples, transmission electron microscopy (TEM) foils normal to the compression axis were extracted using an

FEI Helios Nanolab DualBeam 600 focused ion beam (FIB). Samples were thinned at 5 kV and then further cleaned using a Fischione Nanomill. Thin foils were analyzed using bright field (BF) and low angle annular dark field (LAADF) detectors on an FEI Tecnai

F20 STEM at 200 kV. Samples were also extracted normal to <110> orientations in order to view stacking faults edge-on using high angle annular dark field (HAADF) zone axis imaging conducted on a probe-corrected Titan3 80–300 kV STEM. All atomic resolution

HAADF-STEM images were corrected for scan distortions[10]. High spatial resolution

EDX line scans were conducted at 300 kV using an image-corrected Titan3 60–300 kV with a Super-X detector utilizing the Bruker Esprit software. The Super-X EDX detection system uses four silicon drift detectors that are located radially around the objective pole piece and specimen stage for improved collection performance.

174 Atomic resolution EDX maps were collected using a double aberration corrected

FEI Themis with Super-X EDX detector at 300kV. The probe current was set to 50 pA and the dwell time to 25 μs/pixel with a total spectrum collection time of 641 seconds. Raw

EDX data were extracted from the original spectral map and summed over a defined repeat unit based on the  phase crystal structure. The summed spectrum image was quantified using the Bruker Esprit software and experimentally determined Cliff-Lorimer k-factors from a solutionized ME501 sample.

6.2.4 DFT calculation parameters

Energetics of twin formation in pure Ni3Al, ME501, and ME3 alloys were studied using spin-polarized first principles calculations utilizing the Vienna Ab-initio Simulation

Package (VASP).[11] Building on previous work on the energetics of SESF formation, supercells containing twin configuration both before and after the nearest-neighbor reordering process were relaxed using a quasi-Newtonian algorithm with electron exchange treated with the generalized gradient approximation (GGA) including spin polarization in the formulation of Perdew, Burke, and Ernzerhof[12]. All calculations were performed using a 3×7×6 Monkhorst-Pack k-mesh with plane wave energy cutoffs at least

30% greater than the highest specified in the pseudopotentials used[13]. Brillouin zone integration was performed using first-order Methfessel-Paxton smearing with a smearing width of 0.2 eV[14]. Computational cell size and shape were not allowed to relax for any but the pure Ni3Al structures, but structures were relaxed internally until energy differences were less than 10-4 eV. Due to limitations in computational resources, a full convergence study of twin formation energy in ME3 and ME501 with respect to calculation cell size 175 was not possible. In order to establish that the relatively high twin formation energies were a result of cell size limitations, we conducted a partial convergence study on twin formation energy on the simpler pure Ni3Al system. Results are displayed below in Figure 6.1 and indicate that no fewer than 20 {111} planes would be necessary to obtain converged formation energies consistent with twinning being a low energy defect in Ni3Al.

Figure 6.1: Twin energy in pure Ni3Al calculated as a function of number of lattice planes in the calculation cell; the energetic cost of twinning decreases drastically as defects are separated by added lattice planes, as expected. 6.3 Results

6.3.1 Mechanical testing and STEM deformation analysis

In order to demonstrate the effect of the new strengthening mechanism, we examine two similar Ni-base superalloys, ME3 and ME501, for which the main difference important for our purposes is the amount of  phase formers (Nb, Ta, W, Hf, Ti), which is 9.1 wt-%

176 for ME3 and 13% for ME501 (see chapter 3 for complete information on the two alloys).

Figure 6.2 shows the compression creep response for ME3 and ME501 at 760 C for the

[001] orientation, i.e. the time-dependent plastic strain at constant load. Minimizing these plastic strains is critical to the high dimensional stability required of turbine engine disk materials. The creep curves in Figure 6.2(a) reveal the remarkably improved creep resistance of ME501 compared with ME3 at 760 C and 552 MPa (the green and blue curves, respectively). For accurate evaluation of the deformation mechanisms between the two alloys, the ME3 compression creep stress was repeated at 414 MPa to obtain more comparable strain rates (red curve). Post creep scanning transmission electron microscopy

(STEM) analysis revealed for both alloys the presence of dislocations with Burgers vectors of the type ½<110> dislocation in the  matrix, and faulting in the  precipitates, as can be seen in the [001] zone axis bright field (BF) images shown in Figure 6.2(b) and Figure

6.2(c) for ME3 and ME501, respectively. High resolution, high angle annular dark field

(HAADF) STEM shows that the faults extending through both the  and  phases observed in ME3 were microtwins, as exemplified in Figure 6.2(d). However, these microtwins were absent in deformed ME501, while many instances of isolated stacking faults could be seen, which were constrained to the  phases. High resolution HAADF-STEM has also revealed that these isolated faults are extrinsic faults (i.e. superlattice extrinsic stacking faults or

SESFs) as shown in Figure 6.2(e) consisting of stacking faults on adjacent close-packed

{111} planes of the  phase.

177

Figure 6.2: Compression creep data and deformation analysis. (a) [001] Compression creep curves of both ME3 and ME501 at 760C and 552 MPa (blue and green, respectively) and ME3 at 414 MPa (red) to achieve comparable strain rates (5e-9) between the two alloys. [001] Zone axis BF-STEM image revealing isolated SESFs and microtwins in post-crept [001] (b) ME3 and (c) ME501 crystals. HAADF-STEM images showing (d) a microtwin in ME3 and (e) a (SESF) in ME501. The inset in (d) shows a higher magnification of the twin interface in ME3. Scale bars in b and c, 500nm. Scale bar in d, 5nm (inset, 2nm). Scale bar in e, 2nm.

Initial analysis using BF-STEM found microtwins to be more active in ME3 compared to ME501, but the sample size reasonably obtained through STEM diffraction analysis of focus ion beam foils is not enough to conclusively make this assumption.

Therefore, a technique called electron channeling contrast imaging (ECCI) was employed on both post creep and un-deformed samples to create a larger deformation mode sample size.[15,16] By using the backscattered electron detector at low magnifications, electron channeling pattern lines can be observed. These lines indicate diffraction conditions, that when activated through tilting the bulk sample, can allow diffraction contrast imaging of dislocations and/or faults. The ability to image defects at low magnifications from large bulk samples allows for a better statistical analysis. This technique was implemented on the [001] ME3 and ME501 samples to better understand the regularity of microtwins in

178 both alloys both before and after a creep test. Below in Figure 6.3 is an example of two

ECCI images showing defects in ME3 and ME501 from post-creep samples.

Figure 6.3: ECCI image of (a) ME3 showing 10 different twins and (b) ME501 showing 4 different twins. All scale bars, 500nm.

Faults that extend through both  and  phases were classified as twins; whereas, those that were isolated to a precipitate were deemed stacking faults. Below in Table 6.1 are the final statistics from the ECCI investigation. As was discovered in the STEM diffraction work, microtwins were much more prevalent in ME3 than in ME501.

Table 6.1: The twin occurrence rate for a 30μm2 area for ME3 and ME501. Twins were found to be statistically more prevalent in ME3 than ME501.

Sample Twin avg. Twin std. Twin std. err. N Un-deformed ME501 4.20 2.78 0.88 10 ME501 4.23 1.01 0.28 13 Un-deformed ME3 1.72 0.65 0.20 10 ME3 9.69 1.97 0.55 13

179 Two sample, student t tests were employed to investigate the null hypothesis that no twins were formed during the creep test (H0: 1-2=0) where 1 is the mean twin occurrence found in un-deformed samples and 2 is the mean twin occurrence observed in post-creep samples. A 0.05 statistical threshold was chosen for this study. Rejecting the null hypothesis would indicate that twins were either created or destroyed during the creep test, while not rejecting would reveal that no significant number of twins were formed during the test. The p value for HA: 1-20 between the un-deformed and crept ME501 samples was found to be 0.97 which signifies the null hypothesis is not rejected and no significant number of twins were formed in ME501 during the creep test. However, for

ME3 the p value for HA: 1-20 was less than 0.0001 indicating that microtwins had in fact been formed during the creep test.

Viswanathan et al.[17] revealed a method to calculate the relative strain contribution from twinning using the equation below.

푁 ∆ 훾 = 푡푤푖푛 푎푣푔 Equation (1) 퐿

Using the ECCI images, for example those in Figure 6.3, a line can be drawn across the image with a length (L) that will cross a number of twins (Ntwin). Using the average width of twins (avg) determined from high-resolution HAADF-STEM images of edge-on twins, the strain contribution from twins () was calculated in ME3 as shown below in Table 6.2.

The below values take into account the percentage of twins that were present before the creep test began.

180

Table 6.2: Amount of strain contributed to twinning for ME3 and ME501. Twinning played a significant role in ME3’s creep performance.

Alloy Strain from twinning % of Total strain from twinning ME3 0.0029 42% ME501 N/A 0%

Twinning was found to contribute around 42% of the overall strain in ME3 while none of the strain in ME501 could be attributed to twinning. Therefore, the increase in creep strength found in ME501 can be directly related to inhibition of creep by twinning.

However, not all of the creep property differences between the two alloys can be explained solely through this phase transformation strengthening mechanism. The higher amount of

 formers in ME501 led to a larger volume fraction of secondary precipitates (52% compared to 47% in ME3), and the finer microstructure may have also contributed to the creep strength differences.. It is emphasized that this comparison of twinning propensity is based on comparison of behavior at significantly different stress levels for the two alloys, in order to achieve similar creep rates. In this way, we have attempted attempted to compensate for the volume fraction differences. Furthermore, while comparing coarse and fine microstructures in ME3, Unocic et al.[18] found that the finer microstructure possessed better high temperature properties compared to the coarser microstructure. They attributed the improvement of properties to the smaller channel widths found in the fine microstructure, which inhibited dislocation motion. However, these tests were conducted on polycrystalline samples and the relationship between microstructure and creep

181 properties is still not clear. Diologent and Caron[19] found that increasing precipitate size, while keeping the volume fraction constant, improved the primary creep properties for single crystal AM1 and MC544. This improvement was attributed to a decrease of precipitate shearing dislocations in the larger precipitate samples. In this study, single crystal samples were tested, removing the complicating and confounding variables of grain boundaries, secondary phases along grain boundaries (ie carbides and borides), and grain size in order to more effectively explore the effects small alloying differences have on creep properties.

6.3.2 Atomic-Scale Characterization of SESFs

In order to understand these differences in deformation modes, we start by analyzing isolated SESFs using HAADF-STEM as shown in the <110> zone axis images for both ME3 and ME501 in Figure 6.4(a) and Figure 6.4(b), respectively. In both cases, a distinct local composition at the faults is observed. As described in a preliminary study,[20] the ordering observed along the fault in ME501 can be attributed to a shear- induced phase transformation from  to  phase along the fault. Bulk  phase possesses a hexagonal D024 crystal structure (P63/mmc, a=.5096, c=.8304, ==90, =120), the same found locally along a SESF[20,21]. In ME3, brighter contrast, attributed to a larger average atomic number at the SESFs is also observed; however, the lack of atomic ordering along the fault implies that a different segregation event has occurred.[22–24]

182

Figure 6.4: Segregation along stacking faults in ME3 and ME501. HAADF-STEM image obtained on a [110] zone axis revealing (a) segregation along a SESF in ME3 and (b) segregation and “grid-like” ordering of η phase along a SESF in ME501. Integrated EDX line scans showing elemental segregation along a SESF for (c) ME3 and (d) ME501, as indicated in (a) and (b), respectively. Not shown is the enrichment of Co and the depletion in Ni content along the faults in both cases. All scale bars, 2nm.

In addition, the concentration profiles across the SESFs from vertically integrated

EDX line scans also show distinct differences. For ME501, enrichment of Nb, Ta, and Ti to the fault is found (Figure 6.4(d)) – elements that are known to favor the formation of  phase, which is ordinarily considered to be a deleterious phase compromising strength and

183 ductility when present as a micrometer length-scale phase[25]. The small amount of segregation needed to nucleate the  phase along the SESF implies that a threshold may exist where  phase is promoted over . In fact, one study found that when the C content was lowered from 0.15 wt% to 0.08 wt% in IN792, thereby increasing the content of carbide formers (Ta, Nb, and Ti) in the bulk alloy, that  phase would begin to form[25].

It may be that the composition of ME501 is approaching this threshold. In ME3, Cr and

Mo is observed to segregate along the fault (Figure 6.4(c)) which are known to be elements that favor the formation of the  phase.[26]

In previous studies of single crystal Ni-based superalloys, it has been found that orientations which promote microtwinning result in poor creep performance.[27–30] In fact, microtwinning has been speculated as the source of the tension/compression anisotropy observed in single crystal creep tests for Ni-based superalloys, with the directions that encourage microtwinning exhibiting inferior creep strength.[31] This is consistent with the poor creep strength exhibited in this study by ME3 compared to ME501, as microtwinning is observed only in ME3. In order to more deeply understand the relationship between SESFs and microtwinning, as well as the role of atomic arrangement, site-specific analysis of the distribution of the different elements is essential. The most commonly used method in alloy research for this task is energy-dispersive X-ray spectroscopy (EDX). However, due to the stochastic nature of signal generation, even in thin foils, the spatial resolution of EDX has been previously limited to the nanometer scale for superalloys due to interaction volume and large number of alloying elements present.[32] However, the recent development of advanced, high sensitivity X-ray detector systems[33] has opened the door to use X-ray emission to characterize materials at 184 previously unattainable spatial resolution. In fact, we demonstrate here for the first time in a structural metal alloy, that atomic resolution EDX maps can be obtained, as shown in

Figure 6.8, and provide quantified, site-specific segregation of solute atoms which definitively confirm that the  phase has formed locally at the stacking faults in ME501.

As mentioned in the Methods section, 256 x 256 pixel atomic resolution EDX maps were collected in a region including an SESF in the ME501 alloy. Given the fact that the SESF forms the η phase, it can be asserted that site-specific segregation should exhibit the same periodicity dictated by the η phase crystal structure.[34,21] Following on this, repeating units along the SESF can then be extracted from the total spectrum image and summed to improve the signal-to-noise ratio. Because the SESF was not oriented parallel to any side of the original spectrum image, the entire spectrum image was rotated using a bicubic interpolation from the Matlab 2015a Image Processing Toolbox.[35] As seen Figure 6.5,

11 repeating units, corresponding to two unit cells in the [112̅0] projection were extracted from the total spectrum image and then summed. Given the limited amount of SESF length available, six sequential repeat units were taken (yellow boxes), with five additional repeat units taken midway between each of the previous six. This results in overlapping data sets; however, this is still a valid process because of the symmetry of the region contained. If, for example, this assumption were incorrect, and site-specific segregation were not occurring, the resulting summed data would exhibit a more uniform composition.

185

Figure 6.5: Rotated EDX spectrum image, showing the HAADF layer with 6 repeating units along the SESF (yellow boxes) and 5 additional repeating units (red boxes) taken midway between the first 6.

By summing the 11 regions mentioned above, the EDX spectra at each pixel were vastly improved. Figure 6.6(a) shows a representative spectrum from the raw data with very few counts per peak, which is typical of atomic resolution EDX data. Figure 6.6(b) shows a representative spectrum from the summed spectrum image, displaying nearly an order of magnitude higher maximum peak counts and an even greater increase in total integrated peak counts. The summed spectrum image, with a better signal-to-noise ratio than the raw data, was then imported into Esprit and quantified using experimentally determined Cliff-Lorimer k-factors

186 2.5

1.5

C 1

0.5

0 0 2 4 6 8 10 12 Energy (keV) 25

C 10

0 0 2 4 6 8 10 12 Energy (keV)

Figure 6.6: Representative EDX spectra from raw data (top) and summed data (bottom) showing a significant improvement in both counts per peak and signal-to-noise ratio

187 K-factors were experimentally found by solutionizing the ME501 alloy at 1210°C for

1 hour, seen in Figure 6.7. Some variation in HAADF-STEM image intensity suggestions that there may be a compositional modulation that was not fully solutionized; however, the an EDX spectrum was then taken from a large area on a ⟨110⟩ zone axis just like the atomic resolution EDX data to ensure that the proper channeling condition was taken into account.

Experimental k-factors (wt%) for this alloy and the associated error with each can be seen in Table 6.3.

Figure 6.7: HAADF-STEM image of ME501 sample solutionized at 1210°C for 1 hour. Scale bar, 500nm.

188

Table 6.3: Experimentally determined k-factors for [110] ME501

Element X-ray Series ⟨110⟩ k-factor Error Al K 1.0334 ± 0.01 Co K 0.8694 ±0.005 Cr K 0.8372 ±0.005 Mo L 1.6450 ±0.02 Nb L 3.9240 ±0.18 Ni K 1.0000 -- Ta M 5.8734 ±0.15 Ti K 0.8507 ±0.01 W M 2.4190 ±0.05

After the summed spectrum image was quantified using the Bruker Esprit software package, along with the experimentally determined k-factors, the final atomic resolution

EDX maps were constructed, as seen in Figure 6.8. Error in the EDX quantification is shown in at% in Table 6.4. Values displayed in Table 6.4 are the minimum and maximum values (in at%) and their associated standard deviations.

189

Figure 6.8: Quantified atomic resolution EDX of η phase in ɣ′ showing the HAADF-STEM image of the fault exhibiting characteristic ordering of intensity within the fault; Ni sublattice (green); Co (yellow) segregating to Ni sites; Ta and Nb (dark and light blue, respectively) segregating to the Wyckoff 2a sites; Al and Ti (red and magenta, respectively) segregating to the Wyckoff 2d sites. W, Cr, and Mo (purple, orange, and light green, respectively) are fairly noisy. All EDX values are in at%. Scale bar, 0.5nm.

190

Table 6.4: Experimental error ranges for the atomic resolution EDX in Figure 6.8.

Element Minimum value (at%) Maximum value (at%) Al 1.0 ± 1.0 21.1 ± 5.1 Co 2.2 ± 1.0 15.3 ± 2.8 Cr 0.3 ± 0.4 4.8 ± 1.6 Mo 0.4 ± 0.4 5.8 ± 2.1 Nb 0.9 ± 1.0 12.6 ± 5.5 Ni 45.6 ± 4.7 81.6 ± 7.8 Ta 0.8 ± 0.8 21.0 ± 6.8 Ti 0.5 ± 0.6 15.6 ± 3.0 W 0.3 ± 0.3 7.5 ± 2.4

At the stacking fault, it can be seen that Ta and Nb segregate preferentially to the indicated

(circled) positions, which are the Wyckoff 2a positions of the bulk  phase. Furthermore,

Al and Ti are observed to segregate to the Wyckoff 2d positions of the bulk  phase, while

Co can be seen to segregate to the Ni sublattice. Not only do these site-specific results account for the Z-contrast intensity observed in the HAADF-STEM images, they also confirm the assertion that the  phase has nucleated along the SESF, and matches the ordering first described by Pickering et al.[21] in a different alloy (718 plus), also containing elevated Ti, Nb, and Ta content.

6.3.3 Density Functional Theory Calculations

Taken together, Figure 6.4 and Figure 6.8 indicate, for the first time that elemental segregation to stacking faults occurs differently in these two important, commercial alloys.

Elemental segregation and nucleation of the  phase occurs along the faults in ME501.

However, the segregation in ME3 indicates a distinctly different trend for forming a -like phase at the stacking faults, with Cr and Mo replacing Al. Several reports, this analysis included, have found microtwinning to occur in orientations that favor SESF 191 formation.[36,31,29] This correlation can be rationalized mechanistically by noting that additional shearing events by Shockley partial pairs will lead to thickening of stacking faults into microtwins as shown below in Figure 6.9.

Figure 6.9: SESF to microtwin transformation. (a) DFT cell showing segregation along a SESF in ME3. (b) on the left is an experimental HAADF-STEM image of a two layer SESF being sheared by two Shockley partials to form a three layer twin near a / interface. On the right is a center of symmetry analysis of the HAADF-STEM image. (c) DFT cell showing the resulting twin from the process observed in 4(b). Scale bar, 1nm.

In order to understand how the different segregation/phase transformation phenomena observed between the two alloys affect formation of the detrimental microtwinning deformation mode, DFT calculations were performed on simulation cells such as those in Figure 6.9(a) and Figure 6.9(c), created using knowledge gleaned from the site specific EDX maps in Figure 6.8 and HAADF-STEM images in Figure 6.4, and following the procedures described by Kresse and Furthmüller[11]. Figure 6.10(a) displays the energetics of the twin formation process in the Ni-based superalloys studied.

192

Figure 6.10: DFT calculations of SESF and microtwin formations. (a) Energetic cost of twin formation by shearing along a SESF prior to reordering in Ni3Al, ME501 with a random solid solution, i.e. no segregation (RSS), and ME501 where  has nucleated along the fault as observed experimentally. Note the relatively large energy cost to form a twin along a SESF with  phase. (b) Energy difference due to reordering after twin formation. The small difference found for ME3 suggests that the segregation of  formers (Co, Cr, and Mo) replacing Ni and Al has removed nearest-neighbor violations in the precipitate near the fault, making twinning easier for ME3.

Once the SESF configuration has formed, it is necessary for two additional partial dislocations to shear adjacent planes to create a four-layer microtwin configuration, as shown above in Figure 6.9; however, in the absence of diffusional rearrangement of the elemental segregation, this process creates a plane of atoms with high-energy, nearest- neighbor violations. We consider this configuration as an energy barrier for thickening into twins in the ordered  superlattice: as shown in Figure 6.10(a), for pure Ni3Al this barrier is 364 mJ/m2, for alloy ME501 where the solutes are distributed as a random solid solution

(indicated as “RSS”) the barrier is 481 mJ/m2, and for the segregated SESF with  phase

193 the barrier is 743 mJ/m2. It may be concluded that the ordered  phase observed at the

SESFs in ME501 creates a significant barrier for microtwin formation in this alloy system.[20]

6.4 Discussion

Following the hypothetical shearing of planes adjacent to the SESF, atomic reordering is necessary to create low-energy microtwins.[37–39] In order for a SESF and twin to form by movement of Shockley partial dislocations, a reordering process must occur to eliminate high-energy, wrong nearest-neighbor bonds along the fault and twins.

Kolbe[39] and Kovarik et al.[37] both described the mechanism in which a fault created by the shear of like-signed Shockley partials on adjacent {111} planes in the  phase can reorder by a local diffusional process and eliminate wrong nearest-neighbor violations. In

Figure 6.10(b), where the energetics of this process are also considered, a large release of energy associated with these atomic rearrangements is found: for pure Ni3Al, this release is on the same order as the additional energy for shearing of partials, i.e. 375 mJ/m2. For the ME3 alloy case, this process releases almost no energy at all: only 15 mJ/m2, much less than the energy associated with twin formation. This reduction in energy in ME3 is a result of the loss of nearest neighbor violations in the  precipitate – a significant contributor to the high strength of superalloys – from the segregation of  formers along the fault.

194

Figure 6.11: Phase transformation strengthening in ME501. (a) Schematic of an isolated SESF in a  precipitate. (b)  formers (Co, Cr, and Mo) segregated along the SESF in ME3. (c) Two more dislocations have interacted at the / interface near the SESF in ME3. (d) Two more Shockley partials shear along the SESF forming a four layer twin that is able to shear both the  and  precipitates. (e)  formers (Co, Ta, Ti, and Nb) segregated along the SESF in ME501. (f) Two more dislocations have interacted at the / interface near the SESF in ME501. (g) Given results in Figure 6.10 the dislocations are not able to form a twin in ME501.

The implications from the DFT calculations in Figure 6.10 are summarized in

Figure 6.11. In the case of ME3,  formers (Co, Cr, Mo) segregate to the fault, transforming the fault to a -like region, as shown in Figure 6.11(b). New Shockley partials interact at the / interface where the SESF has formed. These partials are able to enter the  precipitate and shear along the SESF, with little energy penalty. The formation of a -like phase along the SESF removes nearest-neighbor violations, promoting further shearing by partials due to the subsequent lower energy barrier. Consequently, twins which shear through both  and  phases (Figure 6.11(c) and Figure 6.11(d)) can form, thereby defeating the effectiveness of the strengthening  precipitates. This explains the high frequency and of twins observed in ME3.

In the case of ME501,  phase formers (Nb, Ta and Ti) segregate to the SESF as

195 shown in Figure 6.11(e). When Shockley partials interact with the fault at the / interface, the lack of microtwinning implies that shearing along the fault is a rare occurrence and

DFT calculations confirm that a very large energy barrier prevents these partials from shearing into the precipitate (Figure 6.10(a) and Figure 6.10(b)). Therefore, the  phase formation along isolated SESFs represents a new strengthening mechanism by limiting the formation of microtwinning, greatly improving creep strength and high temperature capabilities.

Conceptually, the distinct interface structures observed in both alloys are similar to the formation of interface “complexions” which have been identified in the case of grain boundaries in several ceramic and metallic systems.[40,41] Complexions are “interface- stabilized states”[42] that have a structure and composition different from that of the matrix and remain confined in the region where they form. This concept has recently been extended to “linear complexions” along edge dislocations in a ferritic steel in which the core region of the dislocation reveals an FCC structure[42]. Common to this previous work is that these special structural states arise during thermal exposure, and thus appear to be thermodynamically stable when localized to the defects, and furthermore do not tend to grow (i.e. thicken in the case of grain boundaries) with time. Thus, the stacking faults which are compositionally “-like” in the case of alloy ME3, and form a one unit cell thick  phase in alloy ME501, both appear to have the attributes of “complexions.” However, a new aspect to the present observations is that these special structural states have developed dynamically under applied stress during high temperature deformation, as opposed to under static annealing, and thus may be accurately described as “dynamic complexions” which have a direct and profound impact on the strength of the superalloy. 196 6.5 Conclusions

The high-temperature strengthening mechanism discussed here has two important, beneficial aspects. Firstly, the nucleation of a secondary ordered phase at stacking faults doubles the energy required to operate additional partial dislocations that are required for microtwinning; and secondly, the inhibited segregation of  formers to the stacking fault, thereby creating a local -like phase at the fault, which would promote microtwinning. This

“phase transformation strengthening” mechanism operates in conjunction with the strengthening from secondary phase precipitates in / alloys, and is heretofore an unidentified strengthening mechanism. Indeed, this mechanism may be further manipulated through alloying and processing to further improve the high-temperature properties of next-generation superalloys for critical structural applications.

References:

[1] R. Reed, The Superalloys: Fundamental and Applications, Cambridge University

Press, New York, 2006.

[2] A.J. Bradley, A. Taylor, An X-Ray Analysis of the Nickel-Aluminium System, Proc.

R. Soc. London. Ser. A. Math. Phys. Sci. 159 (1937) 56–72.

[3] C.T. Sims, A History of Superalloy Metallurgy for Superalloy Metallurgists,

Superalloys 1984. (1984) 399–419. doi:10.7449/1984/Superalloys_1984_399_419.

[4] S.L. Chen, S. Daniel, F. Zhang, Y. a. Chang, X.Y. Yan, F.Y. Xie, et al., The

PANDAT software package and its applications, Calphad Comput. Coupling Phase

Diagrams Thermochem. 26 (2002) 175–188. doi:10.1016/S0364-5916(02)00034-2.

[5] J.O. Andersson, T. Helander, L. Höglund, P. Shi, B. Sundman, Thermo-Calc & 197 DICTRA, computational tools for materials science, Calphad Comput. Coupling

Phase Diagrams Thermochem. 26 (2002) 273–312. doi:10.1016/S0364-

5916(02)00037-8.

[6] N. Saunders, A.P. Miodownik, CALPHAD Calculation of Phase Diagrams: A

Comprehensive Guide, Elsevier Science Inc., New York, 1998.

[7] A. Jain, S.P. Ong, G. Hautier, W. Chen, W.D. Richards, S. Dacek, et al.,

Commentary: The Materials Project: A materials genome approach to accelerating

materials innovation, APL Mater. 1 (2013) 11002. doi:10.1063/1.4812323.

[8] M.D. Abramoff, P.J. Magalhaes, S.. Ram, Image Processing with ImageJ,

Biophotonics Int. 11 (2004) 36–44.

[9] R. Castaing, Electron Probe Microanalysis, in: L. Marton (Ed.), Adv. Electron.

Electron Phys., Academic Press, New York, 1960: pp. 317–386.

[10] C. Ophus, J. Ciston, Correcting nonlinear drift distortion of scanning probe

microscopy from image pairs with orthogonal scan directions, Ultramicroscopy.

(2015). doi:10.1016/j.ultramic.2015.12.002.

[11] G. Kresse, J. Furthmu, Efficient iterative schemes for ab initio total-energy

calculations using a plane-wave basis set, Phys. Rev. B. 54 (1996).

[12] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made

Simple, Phys. Rev. Lett. 77 (1996) 3865–3868. doi:10.1103/PhysRevLett.77.3865.

[13] E. Sanville, S.D. Kenny, R. Smith, G. Henkelman, Improved grid-based algorithm

for Bader charge allocation., J. Comput. Chem. 28 (2007) 899–908.

doi:10.1002/jcc.20575.

[14] M. Methfessel, A.T. Paxton, High-precision sampling for Brillouin-zone integration

198 in metals, Phys. Rev. B. 40 (1989) 3616–3621. doi:10.1103/PhysRevB.40.3616.

[15] D.G. Coates, Kikuchi-like Reflection Patterns Obtained with the Scanning Electron

Microscope, Philos. Mag. 16 (1967).

[16] J. Ahmed, A.J. Wilkinson, S.G. Roberts, Characterizing dislocation structures in

bulk fatigued copper single crystals using electron channelling contrast imaging

(ECCI), Philos. Mag. Lett. 76 (1997) 237–245. doi:10.1080/095008397178986.

[17] G.B. Viswanathan, S. Karthikeyan, P.M. Sarosi, R.R. Unocic, M.J. Mills,

Microtwinning During Intermediate Temperature Creep of Polycrystalline Ni-Base

Superalloys: Mechanisms and Modeling, Philos. Mag. 86 (2006) 4823–4840.

[18] R.R. Unocic, L. Kovarik, C. Shen, P.M. Sarosi, Y. Wang, J. Li, et al., Deformation

Mechanisms in Ni-Base Disk Superalloys at Higher Temperatures, Superalloys

2008 (Eleventh Int. Symp. (2008) 377–385.

doi:10.7449/2008/Superalloys_2008_377_385.

[19] F. Diologent, P. Caron, On the creep behavior at 1033K of new generation single-

crystal superalloys, Mater. Sci. Eng. A. 385 (2004) 245–257.

doi:10.1016/j.msea.2004.07.016.

[20] T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman, et

al., Segregation and η phase formation along stacking faults during creep at

intermediate temperatures in a Ni-based superalloy, Acta Mater. 100 (2015) 19–31.

doi:10.1016/j.actamat.2015.08.053.

[21] E.J. Pickering, H. Mathur, a. Bhowmik, O.M.D.M. Messé, J.S. Barnard, M.C.

Hardy, et al., Grain-boundary precipitation in Allvac 718Plus, Acta Mater. 60 (2012)

2757–2769. doi:10.1016/j.actamat.2012.01.042.

199 [22] S.J. Pennycook, Z-CONTRAST TRANSMISSION ELECTRON MICROSCOPY :

Direct Atomic Imaging of Materials, Annu. Rev. Mater. Sci. 22 (1992).

[23] S.J. Pennycook, D.E. Jesson, High-resolution Z-contrast imaging of crystals,

Ultramicroscopy. 37 (1991) 14–38. doi:10.1016/0304-3991(91)90004-P.

[24] M.M.J. Treacy, a Howie, C.J. Wilson, Z contrast of platinum and palladium

catalysts, Philos. Mag. A. 38 (1978) 569–585. doi:10.1080/01418617808239255.

[25] G.K. Bouse, Eta and Platelet Phases in Investment Cast Superalloys, Superalloys

1996 (Eighth Int. Symp. (1996).

[26] A.K. Jena, M.C. Chaturvedi, Review The role of Alloying Elements in the Design

of Nickel-base superalloys, J. Mater. Sci. 19 (1984) 3121–3139.

[27] D.M. Knowles, S. Gunturi, The role of 〈112〉{111} slip in the asymmetric nature

of creep of single crystal superalloy CMSX-4, Mater. Sci. Eng. A. 328 (2002) 223–

237. doi:10.1016/S0921-5093(01)01688-4.

[28] K. Kakehi, Effect of primary and secondary precipitates on creep strength of Ni-

base superalloy single crystals, Mater. Sci. Eng. A. 278 (2000) 135–141.

doi:10.1016/S0921-5093(99)00579-1.

[29] B. Décamps, A.J. Morton, M. Condat, On the mechanism of shear of γ′ precipitates

by single (a/2)⟨110⟩ dissociated matrix dislocations in Ni-based superalloys, Philos.

Mag. A. 64 (1991) 641–668. doi:10.1080/01418619108204866.

[30] K. Kakehi, Influence of Secondary Precipitates and Crystallographic Orientation on

the Strength of Single Crystals of a Ni- Based Superalloy, Metall. Trans. A. 30

(1999).

[31] K. Kakehi, Tension/compression asymmetry in creep behavior of a Ni-based

200 superalloy, Scr. Mater. 41 (1999) 461–465. doi:10.1016/S1359-6462(99)00191-8.

[32] C. Notthoff, M. Winterer, A. Beckel, M. Geller, J. Heindl, Spatial high resolution

energy dispersive X-ray spectroscopy on thin lamellas., Ultramicroscopy. 129

(2013) 30–5. doi:10.1016/j.ultramic.2013.02.008.

[33] L.J. Allen, A.J. D’Alfonso, B. Freitag, D.O. Klenov, Chemical mapping at atomic

resolution using energy-dispersive x-ray spectroscopy, MRS Bull. 37 (2012) 47–52.

doi:10.1557/mrs.2011.331.

[34] S. Asgari, R. Sharghi-moshtaghin, P. Sadeghahmadi, Mehdi Pirouz, On phase

transformations in a Ni-based superalloy, Philos. Mag. 93 (2013).

doi:10.1080/14786435.2013.773106.

[35] Matlab and Imaging Processing Toolbox Release 2015a, (n.d.).

[36] D.M. Knowles, Q.Z. Chen, Superlattice stacking fault formation and twinning

during creep in gamma/gamma prime single crystal superalloy CMSX-4, Mater. Sci.

Eng. A. 340 (2003) 88–102.

[37] L. Kovarik, R.R. Unocic, J. Li, P. Sarosi, C. Shen, Y. Wang, et al., Microtwinning

and other shearing mechanisms at intermediate temperatures in Ni-based

superalloys, Prog. Mater. Sci. 54 (2009) 839–873.

doi:10.1016/j.pmatsci.2009.03.010.

[38] V.A. Vorontsov, L. Kovarik, M.J. Mills, C.M.F. Rae, High-resolution electron

microscopy of dislocation ribbons in a CMSX-4 superalloy single crystal, Acta

Mater. 60 (2012) 4866–4878. doi:10.1016/j.actamat.2012.05.014.

[39] M. Kolbe, The High Temperature Decrease of the Critical Resolved Shear Stress in

Nickel-Base Superalloys, Mater. Sci. Eng. (2001) 383.

201 [40] P.R. Cantwell, M. Tang, S.J. Dillon, J. Luo, G.S. Rohrer, M.P. Harmer, Grain

boundary complexions, Acta Mater. 62 (2014) 1–48.

doi:10.1016/j.actamat.2013.07.037.

[41] P.R. Cantwell, S. Ma, S.A. Bojarski, G.S. Rohrer, M.P. Harmer, Expanding time–

temperature-transformation (TTT) diagrams to interfaces: A new approach for grain

boundary engineering, Acta Mater. 106 (2016) 78–86.

doi:10.1016/j.actamat.2016.01.010.

[42] M. Kuzmina, M. Herbig, D. Ponge, S. Sandlobes, D. Raabe, Linear complexions:

Confined chemical and structural states at dislocations, Science (80-. ). 349 (2015)

1080–1083. doi:10.1126/science.aab2633.

202

Chapter 7 – Summary and Future Work

The first study of this dissertation explored the effects that temperature and orientation have on creep properties in ME3, a commercially used Ni-base superalloy. Both polycrystalline and single crystal compression creep samples were employed. It was discovered that the [110] orientation exhibited superior creep properties over the [001] orientation when tested under compression creep as well as a temperature dependence on active deformation mechanisms. This work was organized to create the first published deformation mechanism map as shown in Figure 2.13. This study finished by proposing a new twin formation model using the new insights from the deformation analysis.

This work was then broadened to including a new Ni-base superalloy, ME501, which had a higher weight percent of  formers (Nb, Ti, and Ta). Single crystal compression creep samples were tested for both ME3 and ME501 in different stress and temperature regimes where ME501 displayed superior creep properties for all orientations and temperature regimes. Again, a creep anisotropy existed between the two orientations tested, with the [110] oriented samples presenting superior creep strength over the [001] oriented samples. This was found to be due to a difference in active deformation modes.

Interestingly, this anisotropy disappeared in the 815C creep tests. This correlated with dislocation climb by-pass being the prominent deformation mode for all orientations. The dislocation activity diagram was used to explore why dislocation de-correlation is present

203 in the [001] oriented samples at 700C. It was discovered that the stress used in the 700C tests produced favorable conditions for decorrelation in [001] samples. Critical channel widths were also found to play a large factor in whether dislocations de-correlated.

In chapter 3, it was found that SESFs were prevalent in the [001] oriented samples when crept at 760C for both ME3 and ME501. Therefore, in chapter 4 a more extensive investigation was conducted to discover the rate limiting mechanisms associated with

SESF formation. Atomic resolution HAADF imaging found that these stacking faults in

ME501 exhibited a unique ordering contrast that had previously been reported for a sub- ordered  phase in a 718-plus Ni-base superalloy. Through high resolution EDX analysis combined with VASP and HAADF simulations, it was concluded that there was a shear induced solid-state phase transformation from  to  phase along SESFs in ME501. It was further determined that the added amounts of Ta, Nb, and Ti (also  phase formers) was the cause of this phase transformation. In addition, the presence of a Co and Cr rich Cottrell atmosphere was observed around the leading Shockley partials of a SESF. Lastly, through

LAADF-STEM analysis a new SESF formation model was presented. This model predicts that two unlike sign ½<110> dislocations can interact at the / interface, where they dissociate allowing like sign 1/6<112> Shockley partials to shear the  precipitate on adjacent {111} planes. After a reordering process, an isolated SESF will exist in the  precipitate.

The presence of segregation along SESFs in ME501 as well as a prominent Cottrell atmosphere around shearing Shockley partials meant that new analysis was needed to look at segregation and diffusion processes in other creep mechanisms in ME3 and ME501. It

204 was discovered that a prominent Co and Cr Cottrell atmosphere existed around all shearing

Shockley partials in  precipitates, for both ME3 and ME501. Further analysis was also conducted to explore the effects and stability of the newly discovered tertiary  particles inside  precipitates. These particles appear to be sources of Co and Cr for segregation along SESFs and microtwins in ME3, though it is still unclear if these particles improve or worsen creep properties. Lastly, a new twin formation model is proposed that builds on the one created in chapter 2, but now includes the diffusion processes captured in this new study.

The last analysis of this dissertation looks to explain the improved creep properties observed in ME501 compared to ME3. In doing so, a new strengthening mechanism was discovered. The phase transformation from  to  phase along SESFs in ME501 actually inhibits the further extension of the fault into a microtwin. This suppression of microtwin formation creates a more creep resistant alloy. It is believed that this new mechanism can be improved and used to further increase the strength of future Ni-base Superalloys.

This thesis focused on providing new insights into the deformation mechanisms active during creep at intermediate temperatures in Ni-base Superalloys. By testing single crystal disk superalloys in compression many variables were able to be controlled such as grain size, grain boundaries, secondary precipitates (carbides, borides and oxides) allowing for better analysis on the effects that orientation and alloying have on creep properties.

Through new high resolution characterization techniques (HAADF-STEM, BF-STEM,

DC-STEM, and EDX), combined with DFT modelling and HAADF simulation, pioneering observations were acquired and supported which hopefully will allow for improved

205 deformation models and future alloy design. Even with these new discoveries further work needs to be pursued and questions still linger.

New compression creep tests should be conducted on orientations different from

[001] and [110] that were tested here. Though these orientations provided new information on orientation dependence there are still many questions that need to be examined. Right now the dislocation activity diagram (DAD) only has these two orientations to work with.

In order to optimize this model, new creep tests should be tested and characterized. New insights in deformation mechanisms will surely be found in these new tests as well further improving future modelling efforts. The same orientations should also be tested in tension to examine if tension/compression asymmetry exists in these alloys as has been observed for other alloy systems.

The [001] orientations in compression at 760C were particularly interesting in this study due to their propensity for stacking faults and microtwin formation. Using this knowledge, future work should be done to explore the relationship between fault formation and plastic creep strain. New compression creep tests on ME3 and ME501 can be done to varying strains to examine what deformation is present after each test. This way, the shape of the compression creep curves can be further explained in relation to active deformation occurring during the test. This will also have an impact on future crystal plasticity modelling efforts.

Another future effort should be to investigate the effects that the observed tertiary

 particles may have on creep properties in these alloys. An idea was already presented in chapter 5, where heat treating ME501 and ME3 at 815C for 100 hours, may result in two different microstructures. One where the tertiary  precipitates have dissolved but the 206 tertiary  particles still remain (ME501) and the other where the tertiary  precipitates are still present but the tertiary  particles have disappeared. Examining the creep performance from these two new microstructures and comparing them to the previous un-heat treated tests may reveal the effects, if any, that these tertiary  particles may have on creep properties.

Lastly, this study certainly made apparent the significance that elemental segregation may have on creep properties at intermediate temperatures. This combined with new higher resolution EDX technologies coming around the horizon may reveal even more surprising observations. Further analysis on the presence of Co and Cr rich Cottrell atmospheres should be explored as well as segregation along stacking faults in new Ni- base superalloys. These new findings will provide the knowledge necessary to create new, refined disk superalloys that ultimately will push the operating temperature of new jet turbine engines to new levels.

207

Bibliography

A. Ahmadieh, A.K. Mukhergee, Stress--Temperature--Time Correlation for High

Temperature Creep Curves, Material Science and Engineering. 21 (1975) 115–124.

A. Guimier, J.L. Strudel, Stacking fault formation and mechanical twinning processes in a

nickel-base superalloy during tensile deformation at high temperature, in: Int. Conf.

Strength Met. Alloy. Conf. Proc., 1970.

A.H.W. Ngan, M. Wen, C.H. Woo, Atomistic simulations of Paidar–Pope–Vitek lock

formation in Ni3Al, Computational Materials Science. 29 (2004) 259–269.

A. Jain, S.P. Ong, G. Hautier, W. Chen, W.D. Richards, S. Dacek, et al., Commentary: The

Materials Project: A materials genome approach to accelerating materials innovation,

APL Mater. 1 (2013) 11002. doi:10.1063/1.4812323.

A. K. Jena, Review The role of Alloying Elements in the Design of Nickel-base superalloys,

Journal of Materials Science. 19 (1984) 3121–3139.

A. Korner, Weak-beam study on superlattice dislocations in a Ni 3 Al alloy in the

temperature range of the flow stress anomaly, Philosophical Magazine A. 59 (1989) 1–7.

A. Ma, D. Dye, R.C. Reed, A model for the creep deformation behaviour of single-crystal

superalloy CMSX-4, Acta Mater. 56 (2008) 1657–1670.

doi:10.1016/j.actamat.2007.11.031.

208 A. Soula, Y. Renollet, D. Boivin, J.L. Pouchou, D. Locq, P. Caron, et al., Grain Boundary

and Intragranular Deformations During High Temperature Creep of a PM Nickel-Based

Superalloy, Superalloys 2008 (Eleventh International Symposium). (2008) 387–394.

A.T. Paxton, Y. Q. Sun, The role of planar fault energy in the yield anomaly in L12

intermetallics, Philosophical Magazine A. 78 (1998) 85–103.

A.J. Bradley, A. Taylor, An X-Ray Analysis of the Nickel-Aluminium System, Proc. R. Soc.

London. Ser. A. Math. Phys. Sci. 159 (1937) 56–72.

A.J. D’Alfonso, B. Freitag, D. Klenov, L.J. Allen, Atomic-resolution chemical mapping

using energy-dispersive x-ray spectroscopy, Phys. Rev. B. 81 (2010) 100101.

doi:10.1103/PhysRevB.81.100101.

B. Décamps, A.J. Morton, M. Condat, On the mechanism of shear of gamma prime

precipitates by single (a/2)<110> dissociated matrix dislocations in Ni-based superalloys,

Philos. Mag. A. 64 (1991) 641–668. doi:10.1080/01418619108204866.

B. Lerch, V. Gerold, Room Temperature Deformation Mechanisms in Nimonic 80A, Acta

Met. 33 (1985) 1709–1716.

B. Reppich, Some New Aspects Concerning Particle hardening Mechanisms in Gamma

Prime Precipitating Ni-Base Alloys - I. Theortical Concept, Acta Metallurgica. 30 (1982)

87–94.

209 B.H. Kear, A.F. Giamei, G.R. Leverant, J.M. Oblak, Viscous slip in the L12 lattice, Scr.

Metall. 3 (1969) 455–460. doi:10.1016/0036-9748(69)90130-6.

B.H. Kear, A. Giamei, J. Silcock, R.K. Ham, Slip and Climb Processes in Gamma Prime

Hardened Nickel-base Alloys, Scr. Metall. 2 (1968) 287–293.

B.H. Kear, J.M. Oblak, A.F. Giamei, Stacking faults in gamma prime Ni3(AI,Ti)

precipitation hardened nickel-base alloys, Metallurgical Transactions. I (1970) 2477–

2486.

B.H. Kear, J.M. Oblak, W. Aircraft, Deformation modes in gamma prime precipitation

hardened nickel-base alloys., J. Phys. 35 (1974).

C. Carry, J.L. Strudel, Apparent and Effective Creep Parameters in Single Crystals of a

Nickel Base Superally Incubation Period, Acta Materialia. 25 (1976) 767–777.

C. Notthoff, M. Winterer, A. Beckel, M. Geller, J. Heindl, Spatial high resolution energy

dispersive X-ray spectroscopy on thin lamellas., Ultramicroscopy. 129 (2013) 30–5.

doi:10.1016/j.ultramic.2013.02.008.

C. Ophus, J. Ciston, Correcting nonlinear drift distortion of scanning probe microscopy from

image pairs with orthogonal scan directions, Ultramicroscopy. (2015).

doi:10.1016/j.ultramic.2015.12.002.

C.J. Humphreys, Fundamental Concepts of STEM Imaging, Ultramicroscopy. 7 (1981) 7–

12.

210 C.M.F. Rae, L. Zhang, Primary creep in single crystal superalloys: some comments on

effects of composition and microstructure, Mater. Sci. Technol. 25 (2009) 228–235.

doi:10.1179/174328408X369311.

C.M.F. Rae, N. Matan, R.C. Reed, The role of stacking fault shear in the primary creep of

[001]-oriented single crystal superalloys at 750°C and 750 MPa, Mater. Sci. Eng. 300

(2001) 125–134.

C.M.F. Rae, R.C. Reed, Primary creep in single crystal superalloys: Origins, mechanisms

and effects, Acta Mater. 55 (2007) 1067–1081. doi:10.1016/j.actamat.2006.09.026.

C.R. Brooks, Y.M. Wang, Effect on the Microstructure of Aging Hastelloy B2 from 550 to

850C for 1200 Hours, Metallography. 86 (1989) 57–86.

C.T. Sims, A History of Superalloy Metallurgy for Superalloy Metallurgists, Superalloys

1984. (1984) 399–419. doi:10.7449/1984/Superalloys_1984_399_419.

D. Barba, S. Pedrazzini, A. Vilalta-Clemente, A.J. Wilkinson, M.P. Moody, P.A.J. Bagot, et

al., On the composition of microtwins in a single crystal nickel-based superalloy, Scr.

Mater. 127 (2017) 37–40. doi:10.1016/j.scriptamat.2016.08.029.

D. Blavette, E. Cadel, C. Pareige, B. Deconihout, P. Caron, Phase transformation and

segregation to lattice defects in Ni-base superalloys., Microscopy and Microanalysis. 13

(2007) 464–83.

D. Hull, D.J. Bacon, Introduction to Dislocations, 4th ed., Elsevier Ltd., Oxford, UK, 2007.

211 D. Lin, M. Wen, Dislocation Structure Due to High Temperature Deformation in the gamma

prime Phase of a Nickel-based Superalloy, Materials Science and Engineering A. 3

(1989) 207–214.

D. Locq, P. Caron, S. Raujol, A. Coujou, N. Clément, On the Role of Tertiary Gamma Prime

Precipitates in the Creep Gebahior at 700C of a PM Cisk Superally, Superalloys 2004

(Tenth Int. Symp. (2004) 179–187.

D. Mukherji, P. Strunz, D.D.E.L. Genovese, R. Gilles, J. Rösler, A. Wiedenmann,

Investigation of Microstructural Changes in INCONEL 706 at High Temperatures by In-

Situ Small-Angle Neutron Scattering, Metall. Trans. A. 34 (2003) 2781–2792.

D. Raynor, J.M. Silcock, Strengthening Mechanisms in gamma prime-Precipitating Alloys,

Met. Sci. J. 4 (1970) 121–130.

D. Ye, D. Ping, Z. Wang, H. Xu, X. Mei, C. Xu, et al., Low cycle fatigue behavior of nickel-

based superalloy GH4145/SQ at elevated temperature, Materials Science and

Engineering: A. 373 (2004) 54–64.

D.G. Coates, Kikuchi-like Reflection Patterns Obtained with the Scanning Electron

Microscope, Philos. Mag. 16 (1967).

D.M. Knowles, Q.Z. Chen, Superlattice stacking fault formation and twinning during creep

in gamma/gamma prime single crystal superalloy CMSX-4, Materials Science and

Engineering A. 340 (2003) 88–102.

212 D.M. Knowles, S. Gunturi, The role of <112>{111} slip in the asymmetric nature of creep

of single crystal superalloy CMSX-4, Mater. Sci. Eng. A. 328 (2002) 223–237.

doi:10.1016/S0921-5093(01)01688-4.

D.M. Maher, D.C. Joy, The Formation and Interpretation of Defect Images from Crystalline

Materials in a Scanning Transmission Electron Microscope, Ultramicroscopy. 1 (1976)

239–253.

D.P. Pope, S.S. Ezz, Mechanical properties of Ni 3 AI and nickel-base alloys with high

volume fraction of Gamma Prime, International Metals Reviews. 29 (1984) 136–167.

D.P. Pope, V. Vitek, A Theory of the Anomalous Yield Behavior in L12 Ordered Alloys,

Acta Materialia. 32 (1984) 435–448.

E. J. Payton, P. J. Phillips, M.J. Mills, Semi-automated characterization of the gamma prime

phase in Ni-based superalloys via high-resolution backscatter imaging, Material Science

and Engineering. 527 (2010) 2684–2692.

E. Sanville, S.D. Kenny, R. Smith, G. Henkelman, Improved grid-based algorithm for Bader

charge allocation., J. Comput. Chem. 28 (2007) 899–908. doi:10.1002/jcc.20575.

E. Vacchieri, A. Costa, Comparison of the mechanical behavior and evaluation of different

damage mechanismsin an equiaxed and a single crystal superalloys subjected to creep,

LCF and TMF, Superalloys 2012 (12th International Symposium on Superalloys). (2012)

235–244.

213 E.J. Pickering, H. Mathur, A. Bhowmik, O.M.D.M. Messé, J.S. Barnard, M.C. Hardy, et al.,

Grain-boundary precipitation in Allvac 718Plus, Acta Mater. 60 (2012) 2757–2769.

doi:10.1016/j.actamat.2012.01.042.

E.S. Huron, K.R. Bain, D.P. Mourer, T. Gabb, Development of High Temperature Capability

P/M Disk Superalloys, Superalloys 2008 (Eleventh International Symposium). (2008)

181–189.

E.S. Huron, R.C. Reed, M.C. Hardy, M.J. Mills, R.E. Montero, P.D. Portella, et al.,

Controlling the deformation mechanism in disk superalloys at low and intermediate

temperatures, Superalloys 2012. (2012) 35–42.

F. Diologent, P. Caron, On the creep behavior at 1033K of new generation single-crystal

superalloys, Mater. Sci. Eng. A. 385 (2004) 245–257. doi:10.1016/j.msea.2004.07.016.

F. Long, Y.S. Yoo, C.Y. Jo, S.M. Seo, Y.S. Song, T. Jin, et al., Formation of  and  phase

in three polycrystalline superalloys and their impact on tensile properties, Mater. Sci.

Eng. A. 527 (2009) 361–369. doi:10.1016/j.msea.2009.09.016.

F.H. Latief, K. Kakehi, H. Murakami, K. Kasai, Influence of crystallographic orientation on

creep behavior of aluminized Ni-based single crystal superalloys, Superalloys 2012 (12th

International Symposium on Superalloys). (2012) 311–320.

F.R.N. Nabarro, Do we have an acceptable model of power-law creep?, Materials Science

and Engineering: A. 387-389 (2004) 659–664.

214 G. Dieter, Mechanical Metallurgy, McGraw Hill, 1986.

G. Kresse, J. Furthmuller, Efficient iterative schemes for ab initio total-energy calculations

using a plane-wave basis set, Phys. Rev. B. 54 (1996).

G. Lvov, V.I. Levit, M.J. Kaufman, Mechanism of Primary MC Carbide Decomposition in

Ni-Base Superalloys, Metallurgical Transactions A. 35 (2004) 1669–1679.

G.B. Viswanathan, P. Sarosi, D. Whitis, M. Henry, M.J. Mills, Investigation of Creep

Deformation Mechanisms at Intermediate Temperatures in René 88 DT Superalloys, Acta

Mater. 53 (2005) 3041–3057.

G.B. Viswanathan, R. Shi, A. Genc, V.A. Vorontsov, L. Kovarik, C.M.F. Rae, et al.,

Segregation at Stacking Faults within the Gamma Prime Phase of Two Ni-Base

Superalloys Following Intermediate Temperature Creep, Acta Mater. (2014).

G.B. Viswanathan, S. Karthikeyan, P.M. Sarosi, R.R. Unocic, M.J. Mills, Microtwinning

During Intermediate Temperature Creep of Polycrystalline Ni-Base Superalloys:

Mechanisms and Modeling, Philos. Mag. 86 (2006) 4823–4840.

G.K. Bouse, Eta and Platelet Phases in Investment Cast Superalloys, Superalloys 1996

(Eighth Int. Symp. (1996).

G.P. Sabol, R. Stickler, Microstructure of Nickel-Based Superalloys, Phys. Stat. Sol. 35

(1969) 11–52.

215 H. Deutchman, On the Creep Behavior and Deformation Mechanisms Found in an Advanced

Polycrystalline Nickel-base Superalloy at High Temperatures, 2013.

H. Hoeft, P. Schwabb, Investigations towards optimizing EDS analysis by the Cliff-Lorimer

method in scanning transmission electron microscopy, X-Ray Spectrom. 17 (1988) 201–

208.

H. Kagawa, Y. Mukai, The Effect of Crystal Orinetation and Temperature on Fatigue Crack

Growth of Ni-based Single Crystal Superally, Superalloys 2012 (12th International

Symposium on Superalloys). (2012) 225–233.

H. Roth, C. Davis, R. Thomson, Modeling solid solution strengthening in nickel alloys,

Metall. Mater. …. 28 (1997) 1329–1335.

http://link.springer.com/article/10.1007/s11661-997-0268-2.

H. X. Gao, L.M. Peng, Parameterization of the temperature dependence of Debye-Waller

factors, Acta Crystallogr. A55 (1999) 926–932.

H. Z. Deutchman, P.J. Phillips, N. Zhou, M.K. Samal, S. Ghosh, Y. Wang, M.J. Mills,

Deformation mechanisms coupled with phase field and crystal plasticity modeling in a

high temperature polycrystalline ni-based superalloy, Superalloys 2012 (12th

International Symposium on Superalloys). (2012) 25–33.

H.L. Danflou, M. Marty, A. Walder, Formation of serrated grain boundaries and their effect

on the mechanical properties in a P/M nickel base superalloy, Superalloys 1992. (1992)

63–72.

216 Images and video generated using CrystalMaker®: a crystal and molecular structures

program for Mac and Windows. CrystalMaker Software Ltd, Oxford, Images and video

generated using CrystalMaker®: a crystal and molecular structures program for Mac and

Windows. CrystalMaker Software Ltd, Oxford, England (www.crystalmaker.com).

I.S. Kim, B.G. Choi, H.U. Hong, Y.S. Yoo, C.Y. Jo, Anomalous deformation behavior and

twin formation of Ni-base superalloys at the intermediate temperatures, Mater. Sci. Eng.

A. 528 (2011) 7149–7155. doi:10.1016/j.msea.2011.05.083.

J. Ahmed, A.J. Wilkinson, S.G. Roberts, Characterizing dislocation structures in bulk

fatigued copper single crystals using electron channelling contrast imaging (ECCI),

Philos. Mag. Lett. 76 (1997) 237–245. doi:10.1080/095008397178986.

J. Dahal, K. Maciejewski, H. Ghonem, Grain Boundary Deformation and Fracture

Mechanisms in Dwell Fatigue Crack Growth in Turbine Disk Superally ME3,

Superalloys 2012 (12th International Symposium on Superalloys). (2012) 149–158.

J. Douin, F. Pettinari-Sturmel, a. Coujou, Dissociated dislocations in confined plasticity,

Acta Materialia. 55 (2007) 6453–6458.

J. M. Lebeau, S. D. Findlay, L. J. Allen, S. Stemmer, Position averaged convergent beam

electron diffraction: theory and applications., Ultramicroscopy. 110 (2010) 118–125.

J. Telesman, T.P. Gabb, A. Garg, P. Bonacuse, J. Gayda, N. Glenn, et al., Effect of

Microstructure on Time Dependent Fatigue Crack Growth Behavior In a PM Turbine

Disk Alloy, Superalloys 2008 (Eleventh International Symposium). (2008).

217 J. Westbrook, In Intermetallic Compounds: Principals and Practice, Transactions of the

Metallurgical Society of AIME. 209 (1957).

J. Xu, W. Lin, Electronic Structure and Phase Stability of A3Ti (A=Fe, Co, Ni, and Cu),

Phys. Rev. B. 48 (1993) 4276–4286.

J. Zhao, V. Ravikumar, A.M. Beltran, Phase Precipitation and Phase Stability in Nimonic

263, Metall. Trans. A. 32 (2001) 1271–1282.

J.-C. Zhao, M.F. Henry, The thermodynamic prediction of phase stability in multicomponent

superalloys, JOM Appl. Technol. 54 (2002) 37–41. doi:10.1007/BF02822603.

J.J. Schirra, P.L.R. Pratt, E. Hartford, E.S. Huron, K.R. Bain, D.P. Mourer, et al., Effect of

microstructure (and heat treatment) on the 649C properties of advanced P/M superalloy

disk materials, Superalloys 2004 (Tenth International Symposium). (2004) 341–350.

J.O. Andersson, T. Helander, L. Höglund, P. Shi, B. Sundman, Thermo-Calc & DICTRA,

computational tools for materials science, Calphad Comput. Coupling Phase Diagrams

Thermochem. 26 (2002) 273–312. doi:10.1016/S0364-5916(02)00037-8.

J.P. Hirth, J. Lothe, Theory of Dislocations, 2nd ed., Wiley, New York, 1968.

J.P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made Simple,

Phys. Rev. Lett. 77 (1996) 3865–3868. doi:10.1103/PhysRevLett.77.3865.

J.W. Christian, S. Mahajan, Deformation Twinning, Prog. Mater. Sci. 39 (1995).

218 K. Kakehi, Effect of Plastic Anisotropy on the Creep Strength of Single Crystals of a Nickel-

Based Superalloy, Metall. Trans. A. 31 (2000) 421–430.

K. Kakehi, Effect of primary and secondary precipitates on creep strength of Ni-base

superalloy single crystals, Materials Science and Engineering: A. 278 (2000) 135–141.

K. Kakehi, Influence of Secondary Precipitates and Crystallographic Orientation on the

Strength of Single Crystals of a Ni- Based Superalloy, Metall. Trans. A. 30 (1999).

K. Kakehi, Tension/compression asymmetry in creep behavior of a Ni-based superalloy,

Scripta Materialia. 41 (1999) 461–465.

K.R. Bain, D.P. Mourer, R. DiDomizio, T. Hanlon, L. Cretegny, A.E. Wessman, United

States Patent, US 8.992,699 B2, 2015.

L. Kovarik, R.R. Unocic, J. Li, M.J. Mills, The intermediate temperature deformation of ni-

based superalloys: importance of reordering, JOM. 61 (2009) 42–48.

L. Kovarik, R.R. Unocic, J. Li, P. Sarosi, C. Shen, Y. Wang, et al., Microtwinning and other

shearing mechanisms at intermediate temperatures in Ni-based superalloys, Progress in

Materials Science. 54 (2009) 839–873.

L. Lou D. Wang, C. Liu, J. Zhang, Evolution of Grain Boundary Precipitates in a

Directionlly Solidifed Ni-base Superally during High Temperature Creep, Superalloys

2012 (12th International Symposium on Superalloys). (2012) 363–368.

219 L.J. Allen, A.J. D’Alfonso, B. Freitag, D.O. Klenov, Chemical mapping at atomic resolution

using energy-dispersive x-ray spectroscopy, MRS Bull. 37 (2012) 47–52.

doi:10.1557/mrs.2011.331.

L.J. Allen, A.J. D’Alfonso, S.S. Findlay, Modelling the inelastic scattering of fast electrons,

Ultramicroscopy. (2014).

M. Condat, B. Decamps, Shearing of gamma prime precipitates by single a/2<110> matrix

dislocations in a gamma/gamma prime Ni-based superalloy, Scr. Metall. 21 (1987) 607–

612.

M. Heilmiaer, M. Kruger., Metallic Materials for Structural Applications Beyond Nickel-

Based Superalloys, JOM. 61 (2009) 61–67.

M. Huang, Z. Cheng, J. Xiong, J. Li, J. Hu, Z. Liu, et al., Coupling between Re segregation

and interfacial dislocations during high-temperature, low-stress creep of a nickel-based

single-crystal superalloy, Acta Mater. 76 (2014) 294–305.

doi:10.1016/j.actamat.2014.05.033.

M. Kolbe, The High Temperature Decrease of the Critical Resolved Shear Stress in Nickel-

Based Superalloys, Materials Science and Engineering A. (2001).

M. Kuzmina, M. Herbig, D. Ponge, S. Sandlobes, D. Raabe, Linear complexions: Confined

chemical and structural states at dislocations, Science (80-. ). 349 (2015) 1080–1083.

doi:10.1126/science.aab2633.

220 M. Legros, N. Clément, P. Caron, A. Coujou, In-situ observation of deformation

micromechanisms in a rafted / superalloy at 850 °C, Mater. Sci. Eng. A. 337 (2002)

160–169. doi:10.1016/S0921-5093(02)00037-0.

M. Methfessel, A.T. Paxton, High-precision sampling for Brillouin-zone integration in

metals, Phys. Rev. B. 40 (1989) 3616–3621. doi:10.1103/PhysRevB.40.3616.

M. Petrenec, K. Obrtlík, J. Polák, Inhomogeneous dislocation structure in fatigued

INCONEL 713 LC superalloy at room and elevated temperatures, Materials Science and

Engineering: A. 400-401 (2005) 485–488.

M. Segers, J.J. Moverare, K. Simonsson, S. Johansson, Deformation and Damage

Mechanisms During Thermomechanical Fatigue of a Single-Crystal Superally in the

<001> and <011> Directions, Superalloys 2012 (12th International Symposium on

Superalloys). (2012) 215–223.

M. Yamashita, K. Kakehi, Tension/compression asymmetry in yield and creep strengths of

Ni-based superalloy with a high amount of tantalum, Scr. Mater. 55 (2006) 139–142.

doi:10.1016/j.scriptamat.2006.03.048.

M. Yoshiba, O. Miyagawa, Effect Cyclic in Hot of Grain Creep Boundary Strengths

Serration on the Fatigue , Creep in Air and and of a Nickel-base Superalloy Corrosive

Environment, Transactions. 26 (1986).

M.D. Abramoff, P.J. Magalhaes, S.. Ram, Image Processing with ImageJ, Biophotonics Int.

11 (2004) 36–44.

221 M.D. M. Preuss, J. Q. Fonseca, B. Grant, E. Knoche, R. Moat, The effect of gamma prime

particle size on the deformation mechanism in an advance polycrystalline nickel-base

superalloy, Superalloys 2008 (Eleventh International Symposium). (2008) 405–414.

M.G. Ardakani, M. McLean, B. a. Shollock, Twin formation during creep in single crystals

of nickel-based superalloys, Acta Mater. 47 (1999) 2593–2602. doi:10.1016/S1359-

6454(99)00145-7.

M.M.J. Treacy, A. Howie, C.J. Wilson, Z contrast of platinum and palladium catalysts,

Philos. Mag. A. 38 (1978) 569–585. doi:10.1080/01418617808239255.

Matlab and Imaging Processing Toolbox Release 2015a.

N. Saunders, A.P. Miodownik, CALPHAD Calculation of Phase Diagrams: A

Comprehensive Guide, Elsevier Science Inc., New York, 1998.

N. Zhou, C. Shen, M.J. Mills, J. Li, Y. Wang, Modeling displacive–diffusional coupled

dislocation shearing of  precipitates in Ni-base superalloys, Acta Mater. 59 (2011)

3484–3497. doi:10.1016/j.actamat.2011.02.022.

N.D. Evans, P.J. Maziasz, R.W. Swindeman, G.D. Smith, Microstructure and phase stability

in INCONEL alloy 740 during creep, Scr. Mater. 51 (2004) 503–507.

doi:10.1016/j.scriptamat.2004.05.047.

P. Lukas, J. Cadek, L. Kunz, Creep of CMSX-4 Single Crystals of different orientations in

tension and compression, Mater. Sci. Eng. 208 (1996) 149–157.

222 P. Phillips, M. De Graef, L. Kovarik, a. Agrawal, W. Windl, M. Mills, Low angle ADF

STEM defect imaging, Microsc. Microanal. 18 (2012) 676–677.

doi:10.1017/S1431927612005235.

P. Villechaise, J. Cormier, T. Billot, J. Mendez, Mechanical Behavior and Damage Processes

of Udimet 720Li: Influence of Localized Plasticity at Grain Boundaries, Superalloys 2012

(12th International Symposium on Superalloys). (2012) 15–24.

P.B. Hirsch, A model of the anomalous yield stress for (111) slip in L12 alloys, Progress in

Materials Science. 36 (1992) 63–88.

P.B. Hirsch, A New Theory of the Anomalous Yield Stress in L12 Alloys, Philosophical

Magazine A. 65 (1992) 569–612.

B. Decamps, S. Raujol, A. Coujou, F. Pettinari-Sturmel, N. Clement, D. Locq, Philos. Mag.

A. (2004) 91–104.

P.H. Thornton, R.G. Davies, T.L. Johnston, The Temperature Dependence of the Flow Stress

of the gamma prime Phase Based upon Ni3Al, Metallurgical Transactions. 1 (1970).

P.J. Phillips, M.C. Brandes, M.J. Mills, M. De Graef, Diffraction contrast STEM of

dislocations: Imaging and simulations, Ultramicroscopy. 111 (2011) 1483–1487.

doi:10.1016/j.ultramic.2011.07.001.

P.J. Phillips, M.J. Mills, M. De Graef, Systematic row and zone axis STEM defect image

simulations, Philos. Mag. 91 (2011) 2081–2101. doi:10.1080/14786435.2010.547526.

223 P.J. Phillips, R.R. Unocic, L. Kovarik, D. Mourer, D. Wei, M.J. Mills, Low cycle fatigue of

a Ni-based superalloy: Non-planar deformation, Scripta Materialia. 62 (2010) 790–793.

P.O. Kanchanbagh, Stacking Fault Formation in gamma Prime Phase During Monotonic

Deformation of In738LC at Elevated Tempearature, Acta Met. Mater. 39 (1991) 1515–

1524.

P.P. Swu-Jian Liang, The Yield Stress of L12 Ordered Alloys, Acta Materialia. 25 (1976)

485–493.

P.R. Bhowal, E.F. Wright, E.L. Raymond, Effects of cooling rate and morphology on creep

and stress-rupture properties of a powder metallurgy superalloy, Metallurgical

Transactions A. 21 (1990) 1709–1717.

P.R. Cantwell, M. Tang, S.J. Dillon, J. Luo, G.S. Rohrer, M.P. Harmer, Grain boundary

complexions, Acta Mater. 62 (2014) 1–48. doi:10.1016/j.actamat.2013.07.037.

P.R. Cantwell, S. Ma, S.A. Bojarski, G.S. Rohrer, M.P. Harmer, Expanding time–

temperature-transformation (TTT) diagrams to interfaces: A new approach for grain

boundary engineering, Acta Mater. 106 (2016) 78–86.

doi:10.1016/j.actamat.2016.01.010.

Q.Z. Chen, D.M. Knowles, Mechanism of <112>/3 slip initiation and anisotropy of  phase

in CMSX-4 during creep at 750°C and 750 MPa, Mater. Sci. Eng. A. 356 (2003) 352–

367. doi:10.1016/S0921-5093(03)00148-5.

224 R. Castaing, Electron Probe Microanalysis, in: L. Marton (Ed.), Adv. Electron. Electron

Phys., Academic Press, New York, 1960: pp. 317–386.

R. Giraud, J. Cormier, Z. Hervier, D. Bertheau, K. Harris, J. Wahl, et al., Effect of the Prior

Microstructure Degradation of the High Temperature/ Low Dtress Non-Isothermal Creep

Behavior of CMSX-4 Ni-based Single Crystal Superally, Superalloys 2012 (12th

International Symposium on Superalloys). (2012) 265–274.

R. Reed, The Superalloys, Cambridge University Press, Cambridge, UK, 2006.

R. Reed, The Superalloys: Fundamental and Applications, Cambridge University Press, New

York, 2006.

R. Srinivasan, G.F. Eggeler, M.J. Mills, Gamma Prime cutting as rate-conrolling recovery

process during high-temperature and low-stress creep of superalloy single crystals, Acta

Mater. 48 (2000) 4867–4878.

R. Unocic, On the Creep Deformation Mechanisms of an Advanced Disk Nickel-Base

Superalloy, The Ohio State University, 2008.

R.R. Unocic, G.B. Viswanathan, P.M. Sarosi, S. Karthikeyan, J. Li, M.J. Mills, Mechanisms

of creep deformation in polycrystalline Ni-base disk superalloys, Materials Science and

R.R. Unocic, L. Kovarik, C. Shen, P.M. Sarosi, Y. Wang, J. Li, et al., Deformation

Mechanisms in Ni-Base Disk Superalloys at Higher Temperatures, Superalloys 2008

(Eleventh International Symposium). (2008) 377–385.

225 R.R. Unocic, N. Zhou, L. Kovarik, C. Shen, Y. Wang, M.J. Mills, Dislocation Decorrelation

and Relationship to Deformation Microtwins During Creep of a  Precipitate

Strengthened Ni-Based Superalloy, Acta Mater. 59 (2011) 7325–7339.

R.W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd ed.,

Wiley, 1989.

S. Asgari, R. Sharghi-moshtaghin, P. Sadeghahmadi, Mehdi Pirouz, On phase

transformations in a Ni-based superalloy, Philos. Mag. 93 (2013).

doi:10.1080/14786435.2013.773106.

S. Karthikeyan, R.R. Unocic, P.M. Sarosi, G.B. Viswanathan, D.D. Whitis, M.J. Mills,

Modeling microtwinning during creep in Ni-based superalloys, Scr. Mater. 54 (2006)

1157–1162. doi:10.1016/j.scriptamat.2005.11.049.

S. Raujol, M. Benyoucef, D. Locq, P. Caron, F. Pettinari, N. Clement, et al., Decorrelated

movements of Shockley partial dislocations in the -phase channels of nickel-based

superalloys at intermediate temperature, Philosophical Magazine. 86 (2006) 1189–1200.

S. Zhao, X. Xie, G.D. Smith, S.J. Patel, Microstructural stability and mechanical properties

of a new nickel-based superalloy, Mater. Sci. Eng. A. 355 (2003) 96–105.

doi:10.1016/S0921-5093(03)00051-0.

S.D. Antolovich, S. Liu, R. Baur, Low Cycle Fatigue Behavior of Rene 80 at Elevated

Temperature, Metallurgical Transactions A. 12A (1981) 473–481.

226 S.I. Rao, T.A. Parthasarathy, D.M. Dimiduk, P.M. Hazzledine, Discrete dislocation

simulations of precipitation hardening in superalloys, Philosophical Magazine. 84 (2004)

3195–3215.

S.J. Pennycook, D.E. Jesson, High-resolution Z-contrast imaging of crystals,

Ultramicroscopy. 37 (1991) 14–38. doi:10.1016/0304-3991(91)90004-P.

S.J. Pennycook, z-contrast transmission electron microscopy: direct Atomic Imaging of

Materials, Annu. Rev. Mater. Sci. 22 (1992).

S.L. Chen, S. Daniel, F. Zhang, Y. a. Chang, X.Y. Yan, F.Y. Xie, et al., The PANDAT

software package and its applications, Calphad Comput. Coupling Phase Diagrams

Thermochem. 26 (2002) 175–188. doi:10.1016/S0364-5916(02)00034-2.

T. pollock M. Titus, A. Mottura, G. Viswanathan, A. Suzuki, M.J. Mills, High Resolution

Energy Dispersive Spectroscopy Mapping of Planar Defects in L1 2 -containing Co-base

Superalloys, Submiss. to Acta Mater. (2014).

T.J. Garosshen, G.R. Mccarthy, Low Temperature Carbide Precipitation in a Nickel Base

Superalloy, Metallurgical Transactions A. 16 (1985) 1213–1223.

T.J. Garosshen, Influence of Alloy Chemistry on Carbide Precipitation in a Nickel Base

superalloy, Metallurgical Transactions A. 17 (1986) 2075–2077.

227 T.J. Garosshen, T.D. Tillman, G.P. Mccarthy, Effects of B, C , and Zr on the Structure and

Properties of a PM Nickel Base Superalloy, Metallurgical Transactions A. 18 (1987) 69–

77.

T.M. Pollock, N. Rene, Nickel-Based Superalloys for Advanced Turbine Engines:

Chemistry, Microstructure, and Properties, J. Propuls. Power. 22 (2006) 361–374.

T.M. Smith, B.D. Esser, N. Antolin, A. Carlsson, R.E.A. Williams, A. Wessman, et al.,

Phase Transformation Strengthening of High Temperature Superalloys, Accepted. Nature

Communications (2016).

T.M. Smith, B.D. Esser, N. Antolin, G.B. Viswanathan, T. Hanlon, A. Wessman, et al.,

Segregation and  phase formation along stacking faults during creep at intermediate

temperatures in a Ni-based superalloy, Acta Mater. 100 (2015) 19–31.

doi:10.1016/j.actamat.2015.08.053.

T.M. Smith, L.V. Duchao, T. Hanlon, A. Wessman, Y. Wang, M.J. Mills, Determination of

Orientation and Alloying Effects on Creep Response and Deformation Mechanisms in

Single Crystals of Ni-Base Disk Superalloys, Superalloys 2016. submitted (2016).

T.M. Smith, R.R. Unocic, H. Deutchman, M.J. Mills, Creep Deformation Mechanism

Mapping in Nickel Base Disk Superalloys, Mater. High Temp. 33 (2016) 372–383.

T.P. Gabb, A. Garg, D.L. Ellis, K.M. O’Connor, Detailed Microstructural Characterization

of the Disk Alloy ME3, 2004.

228 T.P. Gabb, G. Welsh, The High temperature Deformation in Cyclic Loading of a Single

Crystal Nickel-Base Superalloy, Acta Metall. 37 (1989) 2507–2516.

T.P. Gabb, J. Gayda, J. Telesman, A. Garg, N. Glenn, B. Rd, The Effects of Heat Treatment

and Microstructure Variations on Disk, Superalloys 2008 (Eleventh International

Symposium). (2008) 121–130.

T.P. Gabb, J. Telesman, P.T. Kantzos, K. O’Connor, Characterization of the Temperature

Capabilities of Advanced Disk Alloy ME3, 2002.

V.A. Vorontsov, L. Kovarik, M.J. Mills, C.M.F. Rae, High-resolution electron microscopy

of dislocation ribbons in a CMSX-4 superalloy single crystal, Acta Mater. 60 (2012)

4866–4878. doi:10.1016/j.actamat.2012.05.014.

V. Sass, U. Glatzel, M. Feller-Kniepmeier, Anisotropic creep properties of the nickel-base

superalloy CMSX-4, Acta Materialia. 44 (1996) 1967–1977.

V. Sass, U. Glatzel, M. Feller-Kniepmeier, Creep Anisotropy in the Monocrystalline Nickel-

Base Superalloy CMSX-4, Superalloys 1996 (Eighth Int. Symp. (1996) 283–290.

doi:10.7449/1996/Superalloys_1996_283_290.

V.A. Vorontsov, C. Shen, Y. Wang, D. Dye, C.M.F. Rae, Shearing of Gamma Prime

Precipitates by a<112> Dislocation Ribbons in Ni-base Superalloys: A phase field

approach, Acta Mater.

229 V.A. Vorontsov, L. Kovarik, M.J. Mills, C.M.F. Rae, High-resolution electron microscopy

of dislocation ribbons in a CMSX-4 superalloy single crystal, Acta Materialia. 60 (2012)

4866–4878.

W.D. Jenkins, T.G. Digges, C.R. Johnson, Creep of High-Purity Nickel, Journal of Research

of the National Bureau of Standards. 53 (1954) 329.

W.W. Milligan, S.D. Antolovich, The mechanisms and temperature dependence of

superlattice stacking fault formation in the single-crystal superalloy PWA 1480,

Metallurgical Transactions A. 22 (1991) 2309–2318.

Y. Tsukada, Y. Murata, T. Koyama, N. Miura, Y. Kondo, Coupled Analysis of

Microstructure Evolution with Creep Deformation in Nickel-based Superalloys by the

Phase-field Method, Superalloys 2012 (12th International Symposium on Superalloys).

(2012) 111–119.

Y. Yuan, Y. Gu, C. Cui, T. Osada, Z. Zhong, T. Tetsui, et al., Influence of Co content on

stacking fault energy in Ni–Co base disk superalloys, J. Mater. Res. 26 (2011) 2833–

2837. doi:10.1557/jmr.2011.346.

Y. Yuan, Y.F. Gu, T. Osada, Z.H. Zhong, T. Yokokawa, H. Harada, A new method to

strengthen turbine disc superalloys at service temperatures, Scripta Materialia. 66 (2012)

884–889.

230