Abstract:

This dissertation presents a collection of studies addressing key challenges in liquid phase electron microscopy. Motivating this work is seeking the answers to two primary questions:

What is happening chemically in a liquid sample being imaged with an electron beam?

It is well known the electrons used to image also initiate a vast array of chemical reactions. For example, electron irradiated water has no less than 80 kinetically characterized reactions on the books. I will present a principled workflow as a step to answer this question.

Is that signal, or noise?

One of the primary ways to lessen interactions between the sample and the imaging electrons is to simply use less electrons. The tradeoff is low signal relative to the noise. When looking through these noisy datasets for many hours, well after the experiment has finished, one might catch themselves seeing signal where there is none. I will present a data driven image analysis workflow targeted at denoising large and low signal electron microscopy datasets of dynamic samples.