Size effects on mechanical and thermal properties of thin films

Open Access
Alam, Md Tarekul
Graduate Program:
Mechanical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
June 15, 2015
Committee Members:
  • Md Amanul Haque, Dissertation Advisor
  • Md Amanul Haque, Committee Chair
  • Zoubeida Ounaies, Committee Member
  • Donghai Wang, Committee Member
  • Charles E Bakis, Committee Member
  • Thin films
  • nano scale
  • thermal
  • mechanical
  • size effect
  • coupling of properties
  • micro scale
Materials, from electronic to structural, exhibit properties that are sensitive to their composition and internal microstructures such as grain and precipitate sizes, crystalline phases, defects and dopants. Therefore, the research trend has been to obtain fundamental understanding in processing-structure-properties to develop new materials or new functionalities for engineering applications. The advent of nanotechnology has opened a new dimension to this research area because when material size is reduced to nanoscale, properties change significantly from the bulk values. This phenomenon expands the problem to ‘size-processing-structure-properties-functionalities’. The reinvigorated research for the last few decades has established size dependency of the material properties such as thermal conductivity, Young’s modulus and yield strength, electrical resistivity, photo-conductance etc. It is generally accepted that classical physical laws can be used to scale down the properties up to 25-50 nm length-scale, below which their significant deviation or even breakdown occur. This dissertation probes the size effect from a different perspective by asking the question, if nanoscale size influences one physical domain, why it would not influence the coupling between two or more domains? Or in other words, if both mechanical and thermal properties are different at the nanoscale, can mechanical strain influence thermal conductivity? The hypothesis of size induced multi-domain coupling is therefore the foundation of this dissertation. It is catalyzed by the only few computational studies available in the literature while experimental validations have been non-existent owing to experimental challenges. The objective of this research is to validate this hypothesis, which will open a novel avenue to tune properties and functionalities of materials with the size induced multi-domain coupling. Single domain characterization itself is difficult at the nanoscale due to specimen preparation, setting up proper boundary conditions and the required resolution in measurement. The scope of this work requires multiple property characterization simultaneously, which makes it even more challenging. To accomplish this goal, several micro electro mechanical system (MEMS) based testing devices have been developed in this work. These devices can integrate sample, tiny heaters, electrodes, actuators and sensors to measure not just one property, but multiple properties, at the same time. An additional advantage of the setup is its inherently small size that allows high-resolution microscopy techniques such as transmission electron microscope (TEM) to visualize the events taking place inside the specimen during the experiments. These devices were then used to characterize both single and multi-domain studies as highlighted below. As a first step towards multi-domain characterization, this research explored the mechanical behavior (microstructural effects on deformation and failure) of metallic thin films. For example, in-situ TEM study of aluminum thin films under fatigue loading showed that deformation mechanism in nanoscale is remarkably different from that of bulk scale. In bulk material, dislocation based activity due to fatigue loading leads to formation of persistent slip bands (PSB) and failure of material through crack formation. However, 200 nm thick aluminum specimens with average gain size of 80 nm showed no signs of dislocation activities. Instead, elastic or reversible rotation of the grains is observed; a deformation mechanism that is proposed to make the specimens insensitive to stress concentration and fatigue. Consequently, not a single sample failed even after a million cycle of loading at 0.1% and 0.4% tensile strain. As a continuation of the single domain studies towards multi-domain characterization, influence of stress and temperature on the grain size of nickel thin films is studied next. Grain size plays a major role on the mechanical and thermal properties of thin films. Commonly, thermal annealing is used in the literature to control grain size. Irradiation also has strong effects on grain size. Less known is the phenomenon of mechanical stress assisted grain growth. The in-situ TEM experiments on 75 nm thick, 5-10 nm average grain size platinum specimens show rapid grain growth above 290 MPa stress and 0.14% strain. The associated strain energy is found to be higher than that required for stable interface motion but lower than the stress required for unstable grain boundary motion. This is attributed to geometrical incompatibility arising out of crystallographic mis-orientation in adjoining grains, or in other words, geometrically necessary grain growth. Since temperature plays a strong role in grain boundary mobility, this study was expanded to thermo-mechanical loading scenario. The experimental results suggest a unique synergy – as evident from large grain growth in 100 nm thick, 10 nm average grain size nickel at only 325 MPa stress and 373 K temperature. In comparison, the yield stress and melting point (Tm) are about 1.8 ± 0.1 GPa and 1730 K. Such low temperature – low stress synergy towards large grain growth is remarkably different from the literature, which involves high temperature (T/Tm > 0.5) to achieve any appreciable grain growth. To study the thermal transport, a novel approach based on infrared microscopy is developed. This technique uses energy conservation principle to establish the mathematical model and combines the model with experimentally measured sample temperature to determine the thermal conductivity of thin film. This method is used successfully to measure the thermal conductivity of ultra-thin films (10 nm and 20 nm thick hexagonal boron nitride) as well as regular thin films (200 nm amorphous silicon film, 200-500 nm thick carbon doped oxide). This work also presents an experimental technique to measure heat transfer coefficient in micro and nano scale. The heat transfer coefficient for air is roughly two orders of magnitude higher than the bulk value and found to be dependent on the size and temperature of sample. It is suggested that the dominant heat transfer mode in this case is solid to air heat conduction. This work gives an empirical relationship to predict the heat transfer coefficient for different sample size and temperature. To investigate the central hypothesis, i.e. size induced coupling between mechanical strain and thermal conductivity of material, multi-domain experiments are carried out. A major contribution of this dissertation is the experimental validation of this hypothesis. It has been shown that thermal conductivity of material in nano scale can be tuned by mechanical stress. Thermal conductivity of 50 nm thick freestanding silicon nitride thin film is found to be decreased by an order of magnitude with 2.4% tensile strain. Classical thermal physics does not predict such change. It is hypothesized that energy transfer in amorphous insulating solids depends heavily on the vibration localization, i.e. hopping mode, which is sensitive to mechanical strain. Increased localization of vibrations under tensile strain is the cause of thermal conductivity reduction. Another work on 200 nm thick freestanding amorphous silicon thin film shows the opposite trend i.e. increase of thermal conductivity with applied tensile strain. This is also novel because amorphous materials are not predicted to show any such effect. A closer look in the microstructure of the material during the experiment indicates that stress induced grain growth or phase transformation might be the major contributor for this increase in thermal conductivity. Another key contribution of this dissertation is the observation of the mechanical stress dependent dielectric breakdown behavior in carbon doped oxide thin film. When electrical voltage is applied across a dielectric film, a very small amount of leakage current is measured across the material. The amount of leakage current increases with the increase of applied electrical field. As applied electrical field is increased gradually, the material fails at some point, which is known as the “breakdown” of the dielectric material. In this research work, it has been shown that mechanical stress alone can influence the amount of leakage current of the material, even though the applied electrical field is kept same throughout the process. This behavior is ascribed to the de-trapping of electrons under mechanical strain. This dissertation showed experimental evidence of length scale induced coupling between thermal, mechanical and electrical domains. The technological aspect of this research impacts micro and opto electronics, energy conversion, sensors and actuators and in general, nanotechnology applications. However, it is also very fundamental in nature, and the new phenomena that are explored will enrich knowledge of material behavior and lay foundation for the future work on multi-physics of material at nano scale.