TMS Logo TMS Outstanding Student
Paper Contest Winner--
1998 Undergraduate Division

Terminator III: Design of Tin-Based Biomimetic Self-Healing Alloy Tensile Specimens

Priscilla Bernikowicz
Northwestern University, Evanston, Illinois

CONTENTS

A biomimetic alloy composite composed of a ternary Sn-based matrix reinforced by TiNi shape memory alloy wires has been shown to be capable of closing cracks and healing them when heated to a partially liquid state. The goal of this research was to improve techniques for making tensile samples, to introduce cracks into them, and to get crack clamping, closure, healing, with a final strength >50% of the initial strength. The roles of porosity, composite wire content, and phase fraction liquid during healing were assessed using metallography as well as the Thermo-Calc thermochemical software system. Strength recovery of 83-88% was found in 1.3% and 2.2% wire content samples. Because of poor wetting between alloy and reinforcement wires, 2.2% wire samples had large voids, even though they had higher recovery. The 1.3% wire samples were found to have less porosity and were able to recover strength to within 5% of the 2.2% wire samples.

INTRODUCTION AND BACKGROUND

A smart structure is a non-biological physical structure having a definite purpose, a means to achieve that purpose, and functioning pattern that mimics that of biological materials.1 A smart structure has three tasks: receiving information, processing the information, and acting upon the decision. Most materials can be subsets of smart structures, composed of sensors, actuators, and a microprocessor to control the system.2 Examples of smart structures are the self-forming and self-repairing polymer reinforcedcomposites that mimic human bone formation and repair.3

The benefit of the field of biomimetics and smart materials is the favorable combinations of properties that biological materials possess. The strength and toughness found in the ceramic/polymer Nacre structure of the abalone sea shell provides a useful functioning pattern for materials today.4 The Nacre brick-and-mortar morphology has hard calcium carbonate plates epitaxially bonded to thin layers of a ductile protein.5 This structure combines the hardness of ceramics with the ductility of polymers. By forming crack bridges, the polymer is capable of inhibiting crack propagation. This increases the toughness of the material while retaining the strength of the ceramic.6 Such a combination is what materials scientists would like to mimic.

Applications potentially extend to steels where cracking at extreme temperatures is encountered. A "smart steel" would be able to receive the information that it is cracked, process this information, and then act by healing itself. Candidate actuators for such a smart structure are the shape memory alloys (SMA). The SMA phase of a self-healing alloy composite would act as the ductile polymer in Nacre. In its austenite phase, the alloy is given a "permanent" shape, then cooled to the martensitic phase. It is then subjected to tensile prestrain (crack propagation) where the SMA bridges cracks through the strains of the stress-induced martensitic phase transformation.7 When heated above the austenite finish temperature (Af) the SMA shrinks to its shape-memorized length in the austenite phase thereby closing the cracks.8,9 Compressive stress induced along the shrinkage direction can also contribute to the tensile properties of the composite.10 The initial bridging of cracks by the SMA wires can also allow stable crack growth instead of catastrophic failure during initial fracture.11

This project is a work in progress with the motivation of designing a "smart steel" composite consisting of a ferritic superalloy reinforced by an austenitic SMA to provide crack bridging, closure, and healing.4 The goal is ultimately a smart structure which could not only actuate, but also sense and process without outside intervention. Feasibility studies using a Sn-Bi-In matrix with a NiTi SMA has led to 80% recovery of plastic deformation and crack closure and clamping.4 This paper discusses the design of a tensile sample with the Sn-Bi-In matrix and the results gained from tensile tests and metallography of these samples.

MATERIALS AND PROCEDURES

Materials and Process Design

Students in Northwestern University's new freshman-level course Engineering Design and Communication (EDC) in the past two years have done work on the Sn-Bi-In system in parallel with graduate research on the steel alloy system.4 Commissioned with the task of designing and implementing a solder based composite that would exhibit crack healing, the EDC students embarked on many endeavors to arrive at an the optimal alloy.12

Starting with a binary matrix alloy of Sn-Bi, the students used the Thermo-Calc software13 to determine what third alloy component composition and heating temperature would give a liquid phase fraction that would be just enough to promote crack welding. It was first found that adding Indium allowed for a larger range of melting temperature, giving a melting range of 456-483 K. The compositions of 85.5%Sn, 11%Bi, 3.5% In and 87% Sn, 2.6% Bi, 10.4% In were arrived at, giving fractional melting at 463K of ~25-30%. It was found, however, that the second lower Bi composition, when 3 point bend tested, was not brittle enough for good crack closure. As a result, the 85.5% Sn, 11% Bi, 3.5% In composition was tne recommended.14

SEM mircoanalysis were also done to verify the sample composition before and after melting. These revealed that the compositions were not changing much with melt oxidation.14

For a sample mold, the EDC group used a steel U-frame (length = 11.5 mm, height = 1 mm, width = 1.6 mm, thickness = 0.16 mm). A piece of aluminum foil was shaped and placed at each end of the mold. A layer of 0.2 mm Ti-Ni wires, which were knotted at each end, was laid down on the foil and taped to the table. Successive layers of foil and wire were laid down. The metal alloy was prepared, melted, and poured into the mold and allowed to cool.

This procedure for preparing samples had low reproducibility. The tension applied to the wires ended up being a compound force that tended to draw the wires towards the bottom of the sample and made it difficult to arrange the wires in a uniform fashion. The technique was adapted to using high temperature clay in lieu of foil, but that allowed wires to migrate towards the bottom of the sample.

This previous work on the mold and design was sufficient for the production of bend test samples, yet when it was applied to a tensile application, the problems were magnified. The small area within which the wires could rest and the demand for a higher concentration of wires made it necessary to look into better procedures. It was decided that a graphite mold (length = 15.5 cm, width = 10 cm, height = 1.5 cm, molded section: 0.6 cm height, 1.3 cm width, gauge length = 3 cm, gauge width = 0.7 cm) would be used to improve the quality of the samples produced. Graphite pieces (height = 0.6 cm, width = 1.3 cm, hole diameter = ~0.22 mm) with holes (9 holes, 15 holes, 25 holes) machined through them were used as end pieces through which the wire was threaded. The wires were held in tension by a machined steel arm (length = 11 cm, arms =~ 2 cm, width = 1 cm, tooth width = 0.2 cm) held fast by the end pieces. This allowed for a uniform arrangement of the wires.

Specimen Preparation

The 0.1 mm diameter Ni-Ti (~50% Ni, Af=348.5 K) wire was aged at 500°C for 5 minutes using a salt pot furnace to provide shape memory behavior. After cooling to room temperature, the wire was threaded through a hole on one end cap and knotted twice: once on each end, being careful not to let the knot rest in the gauge section of the mold. Knotting was necessary due to poor wetting between the matrix and the SMA.

The knots provided mechanical anchor points to hold the wires in place during testing11. It was then threaded through the other end cap in the mirror hole position. It was threaded into the adjacent hole and knotted twice again. This continued through all the holes in the end pieces. The wires, through each pass, were tightened so as to support the end caps on the tension arm (Figure 2).

Once this was achieved, another set of end caps was placed on the ends of the mold hole to prevent any leakage. The mold was sprayed with mold release to prevent sample breakage or cracking of the graphite mold during removal of the sample. The ends of the tension arm were also sprayed to allow easy removal of the tension arm once the sample had cooled. 85.5 g of Tin, 11 g Bismuth, 3.5 g of Indium were weighed out and melted on a hot plate at a temperature of ~488 K under a fume hood to prevent inhalation of vapors. The mold was also preheated on the hot plate for 5 minutes. The metal was poured into the mold and allowed to settle (Figure 3).

Then the tension arm with the wires stretched on it with the end caps was immersed into the molten metal (Figure 4). The hot plate was turned off, the sample was allowed to cool and was then removed (Figure 5). This new procedure allowed for precise specifications to be reproduced each time, thus eliminating previous variation in sample preparation. The sample was heated for 24 hours at 453 K ± 5 K in order for it to reach the same microstructure it would achieve during healing treatments.

Tensile Testing

The sample was polished and then mechanically tested in an MTS-SinTech 20/G Materials Testing Workstation at a rate of 0.25mm/min. After failure, the sample was treated with solder flux at the crack interface and heated for 24 hours at 453 K. It was removed from the oven and if it had rehealed, it was tensile tested again.

Details of each sample are as follows:

  1. Sample 1 was one of the first samples to be made using the graphite mold. At this point, however, the idea of the tensile arm had not yet been formulated. Sample 1 was made with 9 of the 0.2 mm diameter wires (as previously used for bend tests: 45.8Ti-42.71Ni-11.7Cu from Raychem Corporation). Because no tension was exerted onthe wires, the sample was not usable.
  2. Sample 2 was made with 9 thin (0.1 mm) wires and then heated for 24 hours at 453 K. During its removal from the mold, however, the specimen cracked. To prevent such adhesion between the alloy and the graphite, Sprayon Premium Mold Release S00303 Food Grade Silicone was employed in subsequent sample preparations. This cracked sample was placed in the oven for 24 hours after solder flux was applied to the crack. After removal, the crack had closed yet it had not yet rehealed. Sample making was modified to use 25 wires in order to promote crack healing.
  3. Sample 3 used 25 thin wires that were arranged using the tension arm. The mold was also sprayed with mold release. This was one of the first successfully made samples.
  4. Sample 4 was made in the same way as Sample 3 and both were heated for 24 hours and then tested. After recording the data from this first tensile test, the samples were used to record the crack closure phenomena using a hot stage microscope, a video camera, and a heat gun. The video clip shows the crack closure, yet this way of heating the sample has many problems. The temperature of the sample was not recorded and the heat gun provides heat from only one direction; thus rehealing and grain size are not uniform. The samples were then placed in an oven and heated for 24 hours. After removal, they had both completely healed. They were then tested again and the data recorded.
  5. When Sample 5 was made with 25 wires, it was noted that a large void had formed due to poor melt flow. When observing Sample 3 metallographically, a void was found in Sample 3 that was ~30% of the cross-sectional area. it was noted as well that re-heating after the first test caused the cross-sectional length and width to increase, suggesting expansion of trapped air. Three possible solutions to this problem were (a) to subject subsequent samples to high frequency vibrations after melting using a metal engraver on the mold to get large voids out of the sample, (b) to coat the wires with solder flux in order to promote the flow of molten metal along the wires, and (c) to reduce the amount of wires in the sample. All three methods were tried.
  6. Sample 6 was made using 25 wires and a void was also observed at both ends of the sample.
  7. Sample 7 was made with no wires in order to take micrographs of the matrix alloy before anda fter heating.
  8. Sample 8 was made with 15 wires using the method of vibrating the mold while pouring to get bubbles out of the sample. The high frequency, however, caused the end pieces to shake out of place and molten metal leaked out.
  9. Sample 9 was a successful attempt at vibrating, fluxing the wires, and using only 15 wires. It was tested, reheated, found to have completely healed, and tested again.
  10. Sample 10 was the same as Samples 8 and 9 except that it broke during polishing and was no longer useful to be tested. It was, however, observed under the microscope to see the crack surface and also to see the distribution of wires in the matrix.

Polishing and Etching

Cross-sections of the samples were polished and etched using a 5% HCl solution for 5 seconds.

RESULTS AND DISCUSSION

The alloy, after being completely melted, is allowed to slowly cool to room temperature, ~300 K. Thermo-Calc predicts that at 300 K there is almost all of one phase of solid BCTA5 Sn solution with small percentages of Bi and In, while 6% of the solid alloy is a solid solution of TetragonalA6 that is Bi-rich. Computed compositions are presented in Figure 6, and the as-solidified microstructure is shown in Figure 7.

The samples, after fabrication, are heated isothermally for 24 hours at 453 K. This is also the temperature at which the samples are re-healed. Heating initially for 24 hours at 453 K allows the strengths of the samples to be compared before and after fracture from essentially the same microstructural state. The table of Figure 8 gives the computed phases and phase fractions present in the composite at 463K and the ternary diagram of Figure 9 shows the 2 phase field that the alloy is in at the same temperature.

Thermo-Calc predicts that, at the normal curing temperature of 463 K, there will be solid BCT_A5 in equilibrium with .339 percent liquid which contains most of the Bi and In. From the micrograph of Figure 10, however, volume fraction of liquid is determined to be .22.

The predicted fraction liquid versus temperature using Thermo-Calc is shown in Figure 11. The apparent difference in estimated and measured. A thermocouple placed in the oven showed there was ~10 degree lower temperature that varied with depth into the oven. Around ~453 K a ~.22 phase fraction of liquid is predicted.

A micrograph of Sample 9 taken at the center of the cross-section and at its edge shwon in Figure 12 rendered a phase fraction of 0.18. The tensile test results of Samples 3,4, and 9 are summarized in Figure 13 with the area % of wires corresponding to the 15 and 25 wire composite samples. From the summary of Figure 14, it is seen that the sample with higher % cross-sectional area of wires was better able to recover its strength after healing. However, it was found that peak stresses decreased. Due to poor wetting between SMA and matrix alloy, porosity in the composite was found to be greater with the larger percent area of wires. The load displacement curves for 15 and 25 wire samples are shwon in Figure 17 and Figure 18.

CONCLUSIONS

From the data, it is clear that samples with fraction of wires similar to those with 15 wires leads to rehealing without problems with voids, but more wires may lead to better recovery of original strength. Applying solder flux to the wires, using high frequencies, and decreasing the density of wires were found to help in eliminating large voids in the samples. The current methodology could also be improved in order to decrease amount of hand work. Techniques used by other researchers include pressurized molding, which would prevent voids within the sample from forming and encouraging bonding by diffusion between SMA and the matrix alloy15 instead of anchoring by hand-made knots. Shortening the length of time required to re-heal is also desirable. This could be acheived with a higher matrix melting point to promote faster kinetics. Tensile tests may use an extensometer to improve the accuracy of data obtained. It is also noted that the alloy, being very brittle, cracked all the way through each time because it was hard to detect when microcracks were forming. Ideally, one would want to only introduce microcracks in the sample and reheal it before complete failure.

The Terminator III materials engineering project represents a systematic, technical approach to an envisioned phenomenon whose motivated by the desire to cause an inorganic system to act in such a way that it appears to have a life of its own, i.e. be self-sustaining and self-regulating. On the basic level, the Sn-Bi-In system study is a feasibility study. Having had the success of rehealing tensile samples to 80% strength that were completely broken apart, the next step is to give that system more intelligence.

The feasibility studies that branch beyond the steel-based composite system point in many directions. Instead of heating externally, the system could heat itself internally by resistance which would promote uniform growth of grains in the alloy and may decrease re-healed grain size. It could also be self-sensing, being able for example to determine its temperature with an internal thermocouple. It could be programmed with an attached logic system to adjust its resistive temperature based upon what it detects. Instead of externally applying flux to cause the liquid phase to disperse itself, it could have a recyclable flux within its microstructure. The extension from the smart SMA material to a smart SMA reinforced composite has been accomplished, yet what remains to be done is to turn the composite into a smart system with the inclusion of sensing and processing systems within a composite-based device.

FIGURES
Figure 1: The wires are surrounded by voids. The dark circles are not necessarily all void, however, since the wire diameter is roughly the same length as the scale provided. Some wires were found to be white circles that were polished, while others, which did not extend out of the sample remained dark colored, as seen in the dark figure above the two white circles. This may either be a knot or two wires next to each other.

Figure 2: The wires, through each pass, were tightened so as to support the end caps on the tension arm.

Figure 3: The metal was poured into the mold and allowed to settle.

Figure 4: The tension arm with the wires stretched on it with the end caps was immersed into the molten metal.

Figure 5: The hot plate was turned off, the sample was allowed to cool and was then removed.

Figure 6: The computed compositions.

Figure 7: The as-solidified microstructure.

Figure 8: Computed phases and phase fractions.

Figure 9: The composition of the alloy of interest is represented by the circle. Tie-line 5-6 shows the distribution between 2 phases: BCT_A5 solid and liquid. The tie line indicates that BCT_A5 is the predominant phase.

Figure 10: Optical micrograph showing large Sn-rich grains and the grain boundary regions, which were once liquid, that have solidified In- and Bi-rich phases. Due to the long anneal time, the grains are very large.

Figure 11: The predicted fraction liquid versus temperature using Thermo-Calc.

Figure 12: Center of Sample 9 cross-section.

Figure 13: The tensile test results of Samples 3,4, and 9 are summarized.

Figure 14: The sample with higher % cross-sectional area of wires was better able to recover its strength after healing.

Figure 15: Sample 2, a 9 wire sample with porosity seen as black dots.

Figure 16: S3, a 15 wire sample - porosity around wires is larger. The huge void in the upper left hand corner is an air pocket made by three wires.

Figure 17: 25 wire (2.2%) sample showing before and after tensile test behavior.

Figure 18: 15 wire (1.3%) sample showing before and after tensile test behavior.


ACKNOWLEDGEMENTS

Thanks to Professor Gregory B. Olson for allowing me to work on this project and giving much thought, time, and encoragement to the project. Thanks to Mark Sinew for helping brainstorm solutions, helping with video and testing, and being a great friend. Thanks to Kathleen Stair for help and use of her facilities. Thanks to Gautam Ghosh, Rachel Bishop, Jim Wright, Chris Kantner, Rick Kraemer, Christopher Scharff, Charlie Kuehmann, Brad Files, and all of the EDC undergraduates.

References
1. W.B. Spillman Jr., J.S. Sirkis, P.T.Gardiner, "The Field Of Smart Structures As Seen By Those Working In It: Survey Results", SPIE, 2444, (1995) 18-27
2. J. Hodgkinson, "What Are Smart Materials Anyway?", Materials World. (August 1993) 449
3. C. Dry, C. Warner, "Biomimetic Bone-Like Polymer Cementitious Composite", SPIE, 3040, (1997) 251-256.
4. B. Files, G.B. Olson, "Terminator 3: Biomimetic Self-Healing Alloy Composite", Proc. Second International Conference on Shape Memory Superelastic Technologies: Engineering and Biomedical Applications, Pacific Grove, CA, (1997)
6. M. McCallum, A, McGeorge, A. Witney, "Terminator 3+: The Biomimetic Smart Steel", (1995)
7. J. S. Paine, C. A. Rogers, "Shape Memory Alloys for Damage Resistant Composite Structures", SPIE 2427 Active Materials and Smart Structures, (1995), 358-370
8. Y. Furuya, A. Sasaki, M. Taya, "Enhanced Mechanical Properties of TiNi Shape Memory Fiber/Al matrix Composite", Materials Transactions, JIM, 34, No. 3 (1993), 24-227
9. M. Taya, Y. Furuya, Y. Yamada, R. Watanabe, S. Shibata, T. Mori, "Strengthening Mechanisms of TiNi Shape Memory Fiber/Al Matrix Composites", SPIE, 1916, (1993), 373-383
10. Y. Yamada, M. Taya, R. Watanabe, "Strengthening of Metal Matrix Composite by Shape Memory Effect", Materials Transactions, JIM, 34, No. 3 (1993), 254-260
11. B. Files, "Design of a Biomimetic Self-healing Superalloy Composite", Northwestern University Dissertation Thesis, (1997)
12. C. Forbell, M. Barney, C. Scharff, W. Lai, "Tin Based Self-healing Alloy", Engineering Design and Communication Class Report (1997)
13. B. Sundman, B. Jansson, J.O. Anderson, "The Thermo-Calc Databank System", CALPHAD 9 (1985) 153
14. E. Tao, M. Price, J. Asahara, K. Benes, T. Key, "Terminator III", Engineering Design and Communication Class Report (1998).
15. H.C. Cao, B.J. Dalgleish. H.E. Deve, C. Elliott, A.G. Evans, R. Mehrabian, G. R. Odette, "A Test Procedure for Characterizing the Toughening of Brittle Intermetallics by Ductile Reinforcements," Acta Metallica, 37, no 11, 2969-1977.

Search Student Member Application Other Best Papers Student Members' Page TMS OnLine