Tensile Strength of Silicon Nanowires

Open Access
Yashinski , Melisa Sue
Graduate Program:
Materials Science and Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
August 26, 2009
Committee Members:
  • Christopher Muhlstein, Thesis Advisor
  • silicon
  • nanomechanical testing
  • nanowires
With advances in the fabrication of nanostructures and their broad range of applications, nanomechanical characterization is of high importance. Some current nanomechanical testing techniques include vibrational resonance, bend testing, and tensile testing. However there is a huge variability in the nanowire data recorded utilizing these techniques. The nanowires typically reveal structural properties that are identical to bulk, indicating that their elastic properties should also match bulk. Often the data reveal elastic moduli covering a large range of values, some so low or so high that they are nonphysical. The inconsistencies in the data bring into question the validity and reproducibility of these nanomechanical testing techniques and indicate that there exists a basic experimental problem inhibiting the tests from producing accurate data. This study uses silicon, a well understood, model material, in order to probe the technique of tensile testing nanowires using a microelectromechanical system (MEMS) and determine where in the experimental setup that these problems are occurring. In this study, the mechanical properties of silicon nanowires were explored using a MEMS tensile testing device. The nanowires were grown by the vapor-liquid-solid (VLS) method off the top of an anodized alumina membrane. Silicon was chosen as an ideal model material because its bulk properties are well understood and should translate to the nanoscale, ensuring that the MEMS is operating properly. In order to thoroughly examine the material properties of silicon, a theoretical model using higher order elastic constants was used to model the constitutive behavior of silicon, which revealed a nonlinear elastic region at high strains and the ultimate strength of silicon in various crystallographic directions. The model should hold true for nanoscale silicon that is still large enough to contain hundreds of atoms in its cross-section. The nanowires tested in this study had a very low, controlled defect density, indicating that they should have high strengths approaching the theoretical limit. The diameter and orientation of the nanowires were first characterized in a transmission electron microscope (TEM). The diameters ranged from 100 to 500 nm and the growth directions included single crystalline [100], [110], [111], [112], and bicrystalline [112]. The MEMS device used for tensile testing the nanowires consisted of a thermal actuator, a load cell comb drive, and a specimen gap. Testing was performed in vacuum after manipulation in a scanning electron microscope (SEM) equipped with a Ga focused ion beam. The nanowire was anchored to the MEMS specimen gap with a FIB deposited Pt composite (45 atomic% Pt, 28% Ga, 24% C, 3% O). A total of 15 tensile tests were performed, only six of which contained both stress and strain data. The remaining 9 tests were performed on MEMS designs that did not accurately track the force in the nanowire. From the six tests with both stress and strain data, the strength of the nanowires ranged from 3.88 to 10.1 GPa and the strain to failure from 4.86% to 6.16%. The elastic moduli, E, determined from the least squares linear fits of the data at low strains ranged from 64.4 to 159.1 GPa for [112] oriented nanowires (Bulk [112], E = 174GPa). The elastic modulus measured for a [111] nanowire was 116.4 GPa, compared to 188 GPa for bulk. The stress-strain curves from these experiments were also compared to the theoretical behavior of silicon and it was found that the tests with elastic moduli near that of bulk mapped well onto the theoretical until very high strains, where the experimental stress was lower than the theoretical. This indicated that an experimental problem was occurring at the high strains. Other tensile tests that had shown elastic moduli significantly lower than that of bulk did not map well onto any part of the theoretical model, indicating that the experimental problem was occurring throughout the experiment. The strains measured for the remaining nine tests ranged from 4.04% to 8.41%, which from the theoretical model correspond to strengths ranging from 6.67 to 12.5 GPa. Fractography was also performed using a field emission SEM (FE-SEM) on a few of the nanowires failed during the tensile tests. The surfaces contained rough bumpy features that were unlike anything seen before in bulk silicon. Scanning transmission electron microscopy (STEM) was then used to image the failed nanowires and a Pt sheath covering the nanowire and fracture surface was revealed. The sheath presumably was created from overspray and wicking during the deposition of the Pt composite anchors, and was deformed plastically during the tensile test and, after failure, collapsed over the fracture surface. This sheath also implies a compliance issue and inaccuracy in the strain measurements taken during the tests and explains why the experimental stress-strain curves deviated from the theoretical model. Therefore nanowire mechanical testing techniques that use a deposited material in order to fix the nanowire in place provide unreliable strain measurements due to compliance and an inability to observe the nanowire directly due to a sheath of the material covering the nanowire surface. The solution to this problem would be the development of a method that enables the strain in the sample to be monitored directly.