UCLA

Nanostructured materials are of critical importance in modern electronic devices. Semiconducting channels of sub-10~nm critical dimension are the primary active components of the smallest transistors. Electric current is transported to these transistors through equally small metallic vias. Crystalline defects play a critical role in the performance of such small devices. An individual vacancy-interstitial point defect, whether introduced through fabrication or through radiation damage, can dramatically alter the performance of a semiconductor device. In metallic interconnects, electromigration (EM) at high current densities causes a flux of atomic vacancies, eventually leading to failure.

Aberration-corrected scanning transmission electron microscopy (STEM) can resolve individual point defects within nanostructured devices, but is blind to the electronic impact of such defects. In the first half of this dissertation, we locate and characterize electrically-active vacancy-interstitial point defects within gallium arsenide nanowire devices using high-resolution STEM electron beam-induced current (EBIC). We directly measure the radius of the 9.6 +/- 0.4 nm e-h generation volume of the STEM beam within the nanowire, which sets the limit of EBIC's electronic resolution. This high resolution allows us to directly map a decrease in minority-carrier diffusion length, due to increased surface recombination, across the width of the 135 nm diameter nanowire device.

If the primary beam energy is raised to 300 kV, vacancy-interstitial defects can be precisely introduced with the electron beam. In real time, the electronic impact of these inserted defects is subsequently recorded with EBIC. In some cases defect insertion events can be localized to within a single sub-nm pixel, by recording abrupt changes in EBIC signal as the beam rasters. The location of these defects, obvious in the EBIC image, is completely invisible in typical STEM imaging channels.

Cobalt is being investigated as a next-generation interconnect material to replace copper, due to its superior EM resistance at small critical dimensions. However, cobalt's EM behavior is complex and poorly understood. In the second half of this dissertation we use electron-energy loss spectroscopy (EELS) to monitor EM-induced stress and thickness changes in cobalt nanowires under bias in situ. EM is strongly dependent on temperature, and nanowire devices under high current density can Joule heat significantly. To account for increases in temperature influencing EM we develop high-resolution techniques such as plasmon-energy expansion thermometry (PEET) and 4D-STEM to measure temperature directly within nanoscale interconnects. Not only can strain due to Joule heating be measured with nanoscale spatial resolution, but so can strain due to the electron-wind force, the root cause of EM. Bias-dependent changes in plasmon energy allows us to measure cobalt's effective ionic charge Z*= +0.62 +/- 0.09 at $400 +/- 20 degrees C.

Under high current density, the nanowire heats significantly due to Joule heating, and the grain structure changes dramatically. We observe secondary grain growth of the hcp phase that is accelerated by EM: Grains on the anode of the nanowire are consistently larger than grains on the cathode. Control of secondary grain growth with an electric current may allow engineering of grains which are larger compared to grains achieved by an equivalent anneal, and may further increase a cobalt nanowire's EM resistance. This possibility, along with the STEM-EBIC techniques developed in gallium arsenide, pave the way towards more failure-resistant nanoscale devices.