Interfacial Reactions in Model NiTi Shape Memory Alloy Fiber-Reinforced Sn Matrix “Smart” Composites
- First Online:
- Cite this article as:
- Coughlin, J., Williams, J., Crawford, G. et al. Metall and Mat Trans A (2009) 40: 176. doi:10.1007/s11661-008-9676-1
In this article, the microstructure properties of a novel Pb-free solder composite were examined. A binary nickel-titanium shape memory alloy (SMA) fiber was used to reinforce the Sn-rich matrix, to take advantage of the superelastic properties of the fiber. The objective of this study was to understand long-term, high-temperature interfacial growth in a model NiTi fiber-reinforced Sn matrix composite solder system. The microstructure was quantified by scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS), and wavelength dispersive spectroscopy (WDS). The mechanical properties of the interfacial zone, e.g., Young’s modulus and hardness, were characterized by nanoindentation. The evolution of the reaction products with time and the relationship between composition and local mechanical properties are discussed.
Health concerns over Pb toxicity in traditional Pb-Sn solders have prompted the need for Pb-free solders in electronic packaging.[1–3] In general, Sn-rich solder alloys based on binary and ternary eutectics of Ag and Cu exhibit significant advantages in creep and thermal fatigue resistance over Pb-Sn solders.[4–6] However, these replacement alloys exhibit inferior ductility to Pb-Sn alloys. When these solder joints are subjected to mechanical stress during assembly, packaging, or in service, the poor ductility may result in failure of the component.[7,8]
The damage tolerance of these solder alloys may be improved by adding shape memory alloy (SMA) fibers to the Sn-rich solder matrix. Shape memory alloys exhibit excellent damage tolerance, up to strains as high as 10 pct with no significant degree of permanent deformation. This is accomplished via two-phase-transformation mechanisms: (a) superelastic transformation and (b) thermally induced transformation. The superelastic transformation occurs when the SMA is strained above the austenite start temperature and transforms from austenite to martensite. After the sample is loaded above the critical transformation stress of the alloy, strain can be recovered simply by unloading and it reverts back to austenite.[9,10] For the thermally induced transformation, plastic deformation can be recovered upon the application of heat, as the material returns to its original dimensions. A superelastic transformation may be desirable in a Pb-free solder composite, because it does not require a heating cycle to recover deformation. Thus, if the SMA/solder composites are subjected to an external stress along the direction of the wire, a superelastic transformation will occur in the SMA fibers. When the external stress is removed, the SMA fibers should return to their original dimensions.
The NiTi SMAs are a good choice as reinforcement in composites due to their relatively low cost and unique properties.[11–19] Binary 50-50 at. pct NiTi alloys can have transformation temperatures just below 0 °C and a superelastic range between freezing and 250 °C, making them excellent candidates for consumer electronics. When considering NiTi fiber-reinforced Sn matrix composites, the microstructure and properties of the interface between the solder and SMA fibers are very important. Bond strength between the matrix and reinforcement must be strong enough to provide load transfer, which is directly dependent on interfacial growth characteristics of the materials in the system. Therefore, it is necessary to have an in-depth understanding of interfacial reactions that take place during melting of the solder (liquid Sn) and NiTi.
The microstructure of NiTi fiber/solder composites has not been studied in detail.[19–23] Dutta and co-workers fabricated Sn-3.8Ag-0.7Cu solder with a single 1-mm-diameter NiTi fiber and a fiber volume fraction of about 11 pct, cast at 533 K. It was originally reported that an intermetallic of Ni3Sn4 was observed at the NiTi wire/solder interface for a Sn-rich sample held at the reflow temperature of 533 K for approximately 3 minutes. This analysis was reversed in subsequent articles,[21,22] where they identified a small reaction zone, via energy dispersive spectroscopy (EDS) line-scan analysis, consisting of interdiffused Sn and Ti. A clear understanding of the interaction between NiTi and Sn is required. The objective of this study is to understand long-term, high-temperature interfacial growth in a model NiTi fiber-reinforced Sn matrix composite solder system. The microstructure was quantified by scanning electron microscopy (SEM), EDS, and wavelength dispersive spectroscopy (WDS). The mechanical properties of the interfacial zone, e.g., Young’s modulus and hardness, were characterized by nanoindentation.
2 Materials and experimental procedure
In order to characterize the microstructure, samples were sectioned perpendicular to the fiber direction and polished to a final finish of 0.3-μm alumina suspension. Image analysis software (ImageJ, National Institutes of Health, Bethesda, MD) was used to determine the area fraction of small Ti-rich particles in the NiTi fiber. Microstructural analysis was conducted using a field emission electron microscope (Hitachi S-4700, Tokyo) at a beam voltage of 15 kV. Backscattered electron imaging was used to increase phase contrast, and EDS point analysis was used to determine phase composition. The WDS analysis was performed on an SEM with a tungsten filament (JEOL1 JXA-8600 Superprobe) at 15 kV. This was done to verify the EDS data and obtain more accurate quantitative chemical analysis. The standards used for the quantitative WDS analysis were TiO2 for Ti, pure Ni for Ni, and pure Sn for Sn. A PET crystal was used for analysis of Ti Kα and Sn Lα peaks, while a ZnF crystal was used for Ni Kα.
A commercial nanoindenter (MTS Nano XP-II, Minneapolis, MN) was used to measure Young’s modulus and hardness of the different phases. Indentations were performed with a Berkovich indenter (three-sided pyramid) to a depth of 900 nm at a nominal strain rate of 0.05/s. The properties of the NiTi fiber, Sn matrix, and the reaction products between fiber and matrix were examined. Approximately 50 indentations were conducted in each component of the microstructure. In order to more accurately probe the mechanical response of small microstructural features with limited contribution from the surrounding microstructure, a dynamic contact module (DCM) was used. The DCM has an increased displacement resolution, which allows for accurate measurements to be obtained at small indentation depths (10 to 100 nm). These experiments were also conducted with a Berkovich indenter to a final depth of 400 nm. All indentations were performed using the continuous stiffness measurement technique. Here, a small high-frequency harmonic is superimposed over the indentation load to continuously measure the contact stiffness of the sample during loading. From this contact stiffness, the Young’s modulus of the material and the hardness can be determined instantaneously as a function of depth. Calibration of the instrument was performed on a reference sample of fused silica. Modulus and hardness data from specific locations were correlated with compositions obtained via EDS and WDS to determine the effect of atomic composition on mechanical properties.
3 Results and discussion
3.1 Microstructure Evolution of NiTi in Liquid Sn
3.2 Mechanical Properties of Interfacial Products by Nanoindentation
Immersion of NiTi fiber in liquid Sn at 300 °C for up to 168 hours resulted in two interfacial reaction layers. The EDS and WDS analyses show that the first layer to form is a ternary solid solution of Sn-Ti-Ni. At the periphery of this layer, where the concentration of Ni is close to zero, Sn reacts with Ti to form Sn3Ti2.
Nanoindentation revealed that the hardness and modulus of the Sn-Ti-Ni phase are directly related to composition. The Sn3Ti2 phase had a higher modulus and hardness than the Sn-Ti-Ni phase at a similar composition, suggesting that it is indeed a stoichiometric intermetallic compound.
JEOL is a trademark of Japan Electron Optics Ltd., Tokyo.
One of the authors (JPC) acknowledges Intel Corporation for an ASU–Intel graduate fellowship. The authors acknowledge the financial support for this research from Intel (Drs. D. Suh, R. Mahajan, and V. Wakharkar). The authors also thank (1) Gordon Moore from the Department of Chemistry and Biochemistry at Arizona State University for his help with the WDS and (2) Memry Corporation for providing the NiTi fibers used in this study.