Sabyasachi Tiwari Wins Award To Develop Cloud Platform for Quantum Materials Simulations

Semiconductors, a cornerstone of modern technology, are the materials that enable the electronic device you are using to read this article. Yet, how they function remains a mystery to most end-users, whose primary concern is whether their device works or not. 

Beneath the surface, their behavior resembles a mouse navigating a maze — electrons scurrying through an obstacle course, but where each hurdle is incessantly vibrating. According to solid-state physics, electrons do not crash into physical hurdles so much as get jostled by the trembling obstacle course, occasionally losing or gaining energy. These vibrations are called phonons, and in order to enhance a semiconductor’s performance, scientists must understand how electrons interact with them. 

Phonons frequently flick electrons off their tracks. When this happens, electrons are forced to swerve in new directions. However, sometimes electrons push back on phonons, causing them to vibrate more powerfully. This “electron-phonon coupling” governs many material properties, such as carrier mobility: the speed at which electrons, boosted by an electric field, zoom through their obstacle course. 

Sabyasachi Tiwari, a research scientist with the Center for Quantum Materials Engineering at the Oden Institute for Computational Engineering and Sciences, has developed a simplified method to swiftly compute these interactions  — work compelling enough to win him the Texas Proof of Concept Award from The University of Texas at Austin. The award provides funding and support to help researchers bring their discoveries to market. 

For Tiwari, this means developing a commercial cloud platform to simulate quantum materials using a method he formulated. “I started this work out of desperation and necessity,” he says. 

With this cloud platform, instead of running calculations through specialized software on a powerful supercomputer, users will upload their material’s structure to a website, and in return, the website will promptly spit out the material’s properties that depend on electron-phonon coupling. 

Tiwari had no choice but to develop this method after having spent many years powering through the clumsy process, motivated by the “holy grail of materials science: designing materials for target properties.” 

Without Tiwari’s simplified method, calculating the electron-phonon interactions encoded in a grid of numbers, known as a matrix, is an incredibly arduous process. First, a scientist must acquire a file detailing the material’s structure. Next, they need to use specialized software to determine the phonon and electron states, which are stored in a slim, manageable matrix. 

Then, the bottleneck appears: transforming the coarse matrix into a dense one. To do so, a basis must be selected, which serves as a fundamental framework for representing all of the information contained in the matrix. This leads to an experience we can all relate to — the tedious process of trial and error.

“But there is no guarantee that researchers will eventually get something that makes sense,” says Tiwari. “People have given up research projects because they couldn't find a basis that would satisfy their calculations.” If you are one of the lucky few who make it past the trial-and-error phase, you can plug these matrices into a software package to compute key electron–phonon properties such as carrier mobility, optical absorption, and the superconducting gap. The software package, Electron-phonon Wannier (EPW), was developed over many years by a broad scientific community, initiated by Oden Institute principal faculty member Feliciano Giustino and extended by Tiwari and other collaborators.

Tiwari says, “To be able to perform these advanced calculations, you need many years of training. Even new Ph.D. students take about a year or two to get into the flow.” But now, thanks to Tiwari's technique, this entire pipeline is reduced to a single step: provide the initial structure and then sit back and relax while the computer handles the rest. 

The main ingenuity occurs in the bottleneck. Instead of asking the user to guess the correct basis through trial and error, Tiwari’s technique removes all user input and instead lets artificial intelligence (AI) take the wheel, a procedure outlined in a paper recently accepted by npj Computational Materials. By systematically pursuing the entire landscape of possible bases, AI automatically locates the optimal basis. Moreover, it finds a good basis from coarser matrices, whereas previously, denser (thus more expensive and time-consuming) matrices were required.

Now, experimentalists, professionals in adjacent fields, and young computational scientists who once lacked the years of specialized training can simulate electron-phonon coupling. “Even an experimentalist who does not know much about the theory will be able to obtain these properties just by providing the input structure of a material,” says Tiwari. “That's our end goal.”

He hopes his easy, hands-free process will enable scientists to discover a plethora of new materials with desirable qualities, such as an updated semiconductor to replace the waning silicon transistor, which has hit a plateau, unable to shrink any further without encountering quantum effects that degrade transistor performance. Tiwari and many other scientists hope to develop a new semiconductor that can overcome this limitation while remaining industrially scalable. 

The funding accompanying the Texas Proof of Concept Award will allow Tiwari to commercialize his simplified method, which he hopes to host on a cloud-based software platform. Calculations that used to take weeks to months will soon take hours to days. The barrier that once demanded years of training will soon require nothing more than a link to Tiwari’s website.