Most of you probably know that electron microscopes are pretty good at imaging materials and detailing their properties, possessing resolution which is 1,000 times greater than a light microscope. However, it has some disadvantages which need to be overcome or fixed.

Often, when working on an upgrade, scientists focus their efforts on the hardware, which is always really costly. Yet, researchers at the U.S. Department of Energy’s (DOE) Argonne National Laboratory are giving us the opportunity to see how advanced software developments are able to push their performance even further.

Recently, a way to improve the resolution and sensitivity of an electron microscope by using an AI framework was introduced by some Argonne researchers. Their approach, which was published in npj Computational Materials, gives scientists a chance to get even more detailed information about materials and the microscope itself.  Tao Zhou, who is an Argonne assistant scientist and lead author, said:

“Our method is helping improve the resolution of existing instruments so people don’t need to upgrade to new expensive hardware so often.”

When a material is being exposed to a beam of electron waves this is the moment when Images are being formed. Because electrons act like waves when they travel, these waves interact with the material. As a result, this interaction is captured by a detector and is measured. Finally, these measurements are used to construct a magnified image.

Also, electron microscopes are able to capture information about material properties, such as magnetization and electrostatic potential. All of this information is stored in a property of the electron wave which is also known as a phase. The phrase describes the location or the timing of a point within a wave cycle.

Soon after all the measurements are taken, information about the phase is seemingly lost. That’s why scientists cannot access information about magnetization or electrostatic potential from the images they acquire. Charudatta Phatak, who is Argonne material scientist and group leader, as well as a co-author of the paper, said:

“Knowing these characteristics is critical to controlling and engineering desired properties in materials for batteries, electronics, and other devices. That’s why retrieving phase information is important.”

One of the most decades-old problems is retrieving phase information.  Eventually, it all started with X-ray imaging and now it is spread by other fields, including microscopy. With the aim of solving this problem, Mathew Cherukara, who is Phatak, Zhou, and Argonne computational scientist and group leader propose leveraging tools built to train deep neural networks, a form of AI.

What are neural networks and how do they work?

They are a series of algorithms that are designed to mimic the human brain and nervous system. When receiving a series of inputs and output, it seems that these algorithms seek to map out the relationship between the two. Cherukara said:

“Tech companies like Google and Facebook have developed packages of software that are designed to train neural networks. What we’ve essentially done is taken those and applied them to the scientific challenge of phase retrieval.”

Zhou pointed out:

“Normally when you’re trying to retrieve the phase, you presume you know your microscope parameters perfectly. However, that knowledge might not be accurate. With our method, you don’t have to rely on this assumption. Instead, you actually get the conditions of your microscope — that’s something other phase retrieval methods can’t do.”

The method they are using also improves the resolution and sensitivity of existing equipment. What does it mean? That means that researchers will be able to recover tiny shifts in phase, and in turn, get information about small changes in magnetization and electrostatic potential, all without requiring costly hardware upgrades. Zhou added:

“Just doing a software upgrade we were able to improve the spatial resolution, accuracy, and sensitivity of our microscopy,”

In his words, the fact that they didn’t need to add any new equipment to leverage these benefits is a huge advantage from an experimentalist’s point of view.