UC Berkeley

The In situ Compression of Annealed Molybdenum Nanopillars in the Transmission Electron Microscope

by

Matthew Lowry

Doctor of Philosophy in Engineering - Materials Science and Engineering

University of California, Berkeley

Professor Andrew Minor, Co-chair

Professor J.W. Morris, Jr., Co-chair

The compression of focused ion beam (FIB) machined metallic pillars with submicron diameters offers a unique opportunity to explore dislocation behavior in a limited volume. This data is of particular interest due to the recent observation of an apparent size effect in the compression of metallic pillars. More specifically, metallic pillars with diameters on the order of microns and below have shown elevated yield stresses, elevated flow stresses, stochastic yield point fluctuations, and large strain bursts relative to the bulk. By harnessing the resolution of the transmission electron microscope (TEM) and performing the compressions in situ in the TEM, it is possible to directly observe dislocation behavior during compression and thereby help to explain the source of the size effect. Moreover, the results from in situ compressions can be used as a comparison point for theoretical models that also attempt to explain the dislocation behavior leading to the size effect.

Standing in the way of performing such compressions in situ in the TEM is the high density of defects imparted by FIB milling, in particular the high density of dislocations. The dislocations are too dense to see individual dislocation behavior, limiting the utility of the TEM. This dissertation details efforts to remove dislocations from FIB-milled molybdenum by in situ annealing in the TEM, thereby unlocking the FIB as a sample preparation technique for in situ TEM studies of mechanical properties.

To that end, three classes of pillars were prepared from molybdenum by FIB-milling and subsequent in situ annealing in the TEM: as-fabricated pillars, fully annealed pillars, and partially annealed pillars. As-fabricated pillars were simply milled and then directly compressed without any annealing. The fully annealed pillars, on the other hand, were annealed such that nearly all dislocations were driven from the pillars, leaving either zero or a few isolated dislocations. The pillars in the final class, those that were partially annealed, were annealed such that the dislocation density was reduced but a significant density remained. Each class of pillar displayed a distinct mechanical response under subsequent in situ compression in the TEM, in which load-displacement data was collected in tandem with visual data from the TEM.

It was found that as-fabricated pillars behaved much the same as ex situ tests performed by others, displaying the size effect already reported. Moreover, the high density of defects made it difficult to interpret dislocation behavior, as it was not possible to discern the motion of individual dislocations. On the other end of the spectrum, the fully annealed pillars behaved as pristine crystals, loading elastically until sudden catastrophic collapse. Using a Hertzian analysis, it was found that fully annealed pillars sustained stresses approaching the ideal strength of molybdenum. Finally, the partially annealed pillars behaved in an intermediate manner, displaying limited plasticity akin to fully annealed pillars when the dislocation density was low, and widespread plasticity corresponding to the size effect when the density was higher. Ultimately, it was found that through annealing, it is possible to explore dislocation behavior in pillars with a wide range of dislocation densities. Moreover, the FIB can also be used in future studies of mechanical behavior in limited volumes without necessarily being limited by the initial dislocation density imparted by the FIB.