Microscale Phase Transformation in Shape Memory Alloy Actuators by Yue Gong A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Mechanical Engineering) in The University of Michigan 2017 Doctoral Committee: Associate Professor Samantha Hayes Daly, Co-Chair Professor Katsuo Kurabayashi, Co-Chair Professor Diann Erbschloe Brei Dr. Nilesh Mankame Professor John Andrew Shaw Yue Gong [email protected] ORCID iD: 0000-0002-8418-6660 © Yue Gong 2017 To grandmothers: Mrs. Zhuzhen Wang and Mrs. Xiuqiao Wu ii Acknowledgements I’d like to thank my adviser Professor Samantha H. Daly. Thank you for giving me the invaluable opportunity to work on this project. Thank you for the time, energy, and patience you invested in me; what I have learned from you and this experience will be instrumental moving forward, both professionally and in life. I’d also like to thank my graduate chair, co-chair, and committee member Professor Katsuo Kurabayashi, for your tireless support and care. Thank you to my committee members Professor John Shaw, Professor Diann Brei, and Dr. Nilesh Mankame for your support and guidance in the GM/UM CRL. I have been very fortunate to be funded by the program and to be part of this supportive, dynamic, collaborative and innovative team. Thanks to all my dearest colleagues: Kaitlyn Mallett, Dr. Keqin Cao, Dr. Ben Reedlunn, Dr, Kyubum Kim, Dr. Adam Kammers, Dr. Jared Tracy, Dr. Michael Kimiecik, Dr. Jason Geathers, Dr. Zhe Chen, Daniel Biggs, Clover Thebolt, Tizoc Cruz-Gonzalez, Chenyang Li, Alan Githens, Marissa Linne, Will LePage, Maya Nath, Corey Bowin, Laura Giñer, Michelle Harr, and Koray Benli. Thanks to the undergraduate students Monica Piñon and Michael Compagner who I had the privilege to work with. Thanks to my friends around Ann Arbor and from H O campus ministry for your love 2 and prayers: Furaha Kariburyo Yay, Melissa Hernandez Duran, Kaitlyn Thunderstorms, Dr. Shinuo Weng, Yuqing Kong, Biqiao Zhang, Lu Tian, Zach Barnes, Cindy and Jonathan Fellows, Jason Kohler, Julie and Chris Payne, and Tammy and Nino Guarisco. Thanks to my mentors, Glenn Galler, Gary Mull, and Winnie Song. Thanks to my parents Dr. Lingling Wang and Dr. Min Gong. You are such cool parents with so much unconventional wisdom. I am grateful for your unconditional support and for caring more about whether I get enough sleep than anything else. I look forward to the mornings because that is the time we can see each other over video calls. I am incredibly blessed to have all of my friends and family. iii Table of Contents Dedication ................................................................................................................................... ii Acknowledgements ...................................................................................................................... iii List of Tables ............................................................................................................................... vi List of Figures .............................................................................................................................. vii Abstract ................................................................................................................................ xvi Chapter 1. Introduction and Background ............................................................................... 1 Section 1.1. A Brief History of Nickel-Titanium (NiTi) ............................................................. 1 Section 1.2. Microstructure of Martensite ................................................................................... 4 Section 1.3. Stress-Strain-Temperature Response of NiTi ......................................................... 9 Section 1.3.1. Transformation Temperatures and Nomenclature ............................................ 9 Section 1.3.2. Superelasticity (SE) ........................................................................................ 10 Section 1.3.3. Shape Memory Effect ..................................................................................... 14 Section 1.4. Cyclic Loading of Superelastic Nitinol ................................................................ 21 Section 1.5. Effect of Texture ................................................................................................... 23 Section 1.6. Effects of Composition and Heat Treatment ......................................................... 26 Section 1.7. Shape Memory Effect at Small Length Scales ...................................................... 28 Chapter 2. Materials and Methods ......................................................................................... 38 Section 2.1. SEM-DIC Sample Preparation and Testing .......................................................... 38 Section 2.2. SEM-DIC Error Analysis ...................................................................................... 44 Chapter 3. Microscale Characteristics of the Shape Memory Effect in Fine-Grained NiTi Wires ................................................................................................................................. 52 Section 3.1. Spatial Heterogeneity of Strain During 1st Actuation Cycle ................................. 52 Section 3.2. Cycle-to-Cycle Similarity During Subsequent Actuation ..................................... 59 Section 3.3. Chapter Summary .................................................................................................. 64 iv Chapter 4. Microscale Repeatability in Fine-Grained Shape Memory NiTi ...................... 67 Section 4.1. Introduction and Background ................................................................................ 67 Section 4.2. Material and Experimental Procedure ................................................................... 67 Section 4.3. Results and Discussion .......................................................................................... 69 Section 4.3.1. Linearity between Strains Accommodated by Detwinning with Cycling ...... 69 Section 4.3.2. High Strain Regions in Detwinned State Associated with the Accumulation of Residual Strain ....................................................................................................................... 74 Section 4.3.3. Higher Strains in the Detwinned State Are Weakly Correlated with a Subsequent Faster Accumulation of Residual Strain ............................................................ 76 Section 4.4. Conclusions ........................................................................................................... 82 Chapter 5. Microscale Repeatability in Coarse-Grained Shape Memory NiTi ................. 83 Section 5.1. Materials and Experimental Method ..................................................................... 83 Section 5.2. Results and Discussion .......................................................................................... 87 Section 5.2.1. Strain Heterogeneity ....................................................................................... 87 Section 5.2.2. Effect of Grain Diameter on Strain Heterogeneity ......................................... 90 Section 5.2.3. Effect of Grain Orientation on Sub-Grain Strain Heterogeneity .................... 95 Section 5.2.4. Transformation Similarity During Thermal-mechanical Cycles .................. 102 Section 5.3. Conclusions ......................................................................................................... 108 Chapter 6. Summary and Comments on Future Work ...................................................... 110 Section 6.1 Summary .............................................................................................................. 110 Section 6.2. Future Work ........................................................................................................ 112 Section 6.2.1. Statistical Analysis on Microstructural Dependency .................................... 112 Section 6.2.2. The Effect of Second Phase Precipitates ...................................................... 113 Section 6.2.3. Performance Prediction Using Magnitude of Detwinning Heterogeneity .... 115 v List of Tables Table 4.1 First Dataset (shown in Figure 4.2) Adjusted R2 values ............................................... 73 Table 4.2 Second Dataset (shown in Figure 4.3) Adjusted R2 values .......................................... 73 Table 4.3. Third Dataset Adjusted R2 values ................................................................................ 73 vi List of Figures Figure 1.1. A brief summary of the timeline of shape memory alloy discoveries. ........................ 2 Figure 1.2. DSC data showing phase transformation in a Nitinol test specimen at a scanning rate of 5˚C/min, with schematics of the unit cells associated with the austenite, martensite, and rhombohedral phases of NiTi. Note that the rhombohedral, or R-phase, is not always present. [8] ......................................................................................................................................................... 3 Figure 1.3: TEM micrograph of nano-scale twins in a heat treated shape memory NiTi wire. [88] ......................................................................................................................................................... 7 Figure 1.4: Micrographs of activated variants at 2% strain in one set of grains at different cycle numbers during superelastic cycling (cycle 1, 6 and 10 from left to right). Although the same habit plane variants are activated each loading cycle, the spatial position of the variants within the grains varies from cycle to cycle. [35] ...................................................................................... 8 Figure 1.5: Stress-induced transformation between austenite and detwinned martensite that underlies the superelastic effect in Nitinol. (right: [8]) ............................................................... 11 Figure 1.6: Examples of NiTi applications utilizing superelasticity: 1. endoscopy tools, 2. spinal implants, 3. vena cava filters, 4. NiTi clips, 5. glass frames, 6. arterial stents, 7. surgical files, 8. orthodontics arch wires. [39], [40], [89] ....................................................................................... 13 Figure 1.7: Transformation path underlying the shape memory effect in NiTi. ........................... 15 Figure 1.8: Recent SME applications in auto industry. a) shape memory alloy resetable spring lift for pedestrian protection [90]; b) SMA wire providing smooth actuation for infotainment elements in cars [91]; c) a prototype of latched active arch seal for car doors [92]. ................... 19 Figure 1.9: SME applications. Aviation and valve coupling. a) aircraft wings; b) MEMs fluid control; c) airplane engine chevrons; d) temperature sensing valves [40] ................................... 20 Figure 1.10: Variation of SMA transformation temperatures with composition (annealing at 693, 873 and 1023 K for 15 min) [66]. A is the austenite start temperature during heating, M and R s s s vii are the martensite start temperature and R-phase start temperature respectively during cooling. (also shown in Figure 1.2. ............................................................................................................. 26 Figure 1.11: Effect of ageing temperature and time on the transformation temperature of Ti- 50.8% at Nitinol wire with a starting A temperature of 11°C. [70] ............................................. 27 f Figure 2.1: Nanoparticle size is controlled by the ratio of sodium citrate to HAuCl . Data points 4 originally reported by Frens [5] are plotted and fit with an exponential relationship for ease of nanoparticle fabrication of the appropriate size. ........................................................................... 41 Figure 2.2: An example of a self-assembly of approximately 100nm diameter gold nanoparticles used for microscale deformation mapping via SEM-DIC. The larger white dots at the four corners of the field of view are platinum markers that were FIB-deposited to designate the field of view. ......................................................................................................................................... 41 Figure 2.4: Schematic of the microwire specimen geometry, test field of view, and uniaxial loading condition. The 34µm by 34µm field of view was located on the flat surface as illustrated. Gage length represents the distance between the crossheads of the in-situ load frame, measured by caliper with a resolution 0.03mm. ............................................................................................ 42 Figure 2.3: In-SEM testing was performed using a tensile stage (Kammrath and Weiss) with a tension-compression load cell (Honeywell model# 3108-10) in the scanning electron microscope (Tescan MIRA3) as shown above. ................................................................................................ 42 Figure 2.5: Standard deviations of the vertical and horizontal displacements that result from examination of an Inconel test specimen with no strain applied under varying the SEM (model Tescan Mira3D) imaging parameters. ........................................................................................... 46 Figure 2.6: Noise introduced using various scan speed (column) and integration (row), demonstrated in the form of principal strain. The time data for each column represent the time used to collect a single frame without considering integration. Noise of more than 0.02 principal strain was introduced to the system with scan speed 1. Noise is minimized using scan speed 4. There are clear cross hatch patterns using scan speeds 5 and 6, hence demonstrating increased resultant noise in principal strain. ................................................................................................. 50 viii Figure 3.1: Microscale strain maps corresponding to the designated points in the stress–strain temperature curve. There was a pronounced heterogeneity in the strain distribution evident at the microscale, caused by detwinning of the martensite phase upon application of load. This heterogeneous strain distribution remained constant throughout the entire actuation cycle, although the strain magnitude changed. Residual strain concentrated at locations where strain accumulation from detwinning and plasticity were significant. ................................................... 55 Figure 3.2: A heat map of the strain at each data point in cycle 1 is shown, where the color indicates the normalized count of the number of data points with those x- and y- values. The x- axis is the value of the point at strain map C (detwinned martensite at maximum load), and the y- axis is the value at that point of strain map of F subtracted from strain map C (where f is the twinned martensite after heating and subsequent cooling). Deformation was largely recovered after heating, and the recovered strain between maps C and F showed a strong linear dependence on the maximum strain of the detwinned martensite (map C). ..................................................... 57 Figure 3.3: The strain maps acquired in the detwinned martensite at maximum applied load (map C) as well as in the twinned martensite after mechanical unloading, heating, and cooling (map F). The same microscale pattern of strain reappeared at both the maximum load and the unloaded specimen, with an intensification of the residual strain in map F from cycle 1 to cycle 5. The averaged strain over the field of view for both maps increased from cycle 1 to cycle 5, from a strain of 0.0543 to 0.0561 (map C) and from 0.0039 to 0.0080 (map F), respectively. ............... 60 Figure 3.4: A heat map of the point-by-point correlation coefficients between microscale strain maps C for cycles 1 and 2 (in Figure 3.3). There is a strong point-to-point similarity between cycles 1 and 2, as observed in Figure 3.3 and tabulated here. The “Correlation Coefficient” axis refers to the correlation coefficient between the strains at each selected data point in cycles 1 and 2, and the “ε ” axis refers to the strain at that same data point in cycle 1. The z-axis, “Normalized 1 Count”, tabulates the number of pixels that fall into each of these axes values. A small number of points that failed digital image correlation, or did not show a correlation to each other, constitute a short peak around a correlation coefficient of 0. ........................................................................ 62 Figure 4.1: Mechanical loading of the wire was paused every 100µm displacement for nominally 10 minutes to allow for material relaxation before a SEM image was taken. This figure is ix