Understanding orderly and disorderly behavior in 2D nanomaterials could enable bespoke design, tailored by AI

(Press-News.org) Since their discovery at Drexel University in 2011, MXenes — a family of nanomaterials with unique properties of durability, conductivity and filtration, among many others — has become the largest known and fastest growing family of two-dimensional nanomaterials, with more than 50 unique MXene materials discovered to date. Experimentally synthesizing them and testing the physical properties of each material has been the labor of tens of thousands of scientists from more than 100 countries. But a recent discovery by a multi-university collaboration of researchers, led by Drexel University researcher Yury Gogotsi, PhD, and Drexel alumnus Babak Anasori, PhD, who is now an associate professor at Purdue University, that sheds light on the thermodynamics undergirding the materials’ unique structure and behavior, could be the key to supercharging this endeavor with artificial intelligence technology. The discovery was recently reported in the journal Science.

The paper, “Order to disorder transition due to entropy in layered 2D carbides,” lays out the foundational parameters governing how the atoms in layered nanomaterials are naturally assembled — looking specifically at how the thermodynamic forces that describe energy disbursal (enthalpy) and disordering of atoms within materials (entropy) apply to interactions between the atom-thick layers that make up MXenes.

Synthesizing MXene materials has been an iterative process of experimentation and verification over nearly a decade and a half since they were first discovered. The materials glean their multitude of properties from the combination of atom-thick layers of which they’re composed. Slight changes in the chemistry or sequence of layers produces an entirely new MXene, typically with an entirely new set of physical properties. 

Due to the complexity of the chemical interactions within the layers of MXenes, the march toward new discoveries has proceeded in small-but-significant increments. According to Gogotsi, distinguished university and Bach professor in Drexel’s College of Engineering, who was one of the lead investigators of the research along with partners from Purdue University, Vanderbilt University, the University of Pennsylvania, Argonne National Laboratory and the Institute of Microelectronics and Photonics in Warsaw, Poland, this breakthrough will not only direct the focus of future inquiries, but it could also allow researchers to avail themselves of high-powered computing and AI technology to take some bigger steps.

“This is exactly where AI will become an enabling technology,” Anasori said. “Guidance from computational science, machine learning and AI will be crucial for navigating the infinite sea of new materials, guiding their development and helping to select the structures and compositions with required properties for specific technologies. I look at this work as opening new avenues in the atomistic design of materials.”

While researchers have used machine learning and computer modeling for decades to posit and discover new materials, recent breakthroughs in microchip technology have taken the predictive capabilities of AI to a new level.

It’s potential for materials science research though apparent — given its complexity and the preponderance of experimental data — has yet to be fully realized. According to Gogotsi, this is due in part to insufficient research on the chemical behavior of the new materials, which is required to train the AI programs and provide the framework needed to harness their predictive power.

“Much of our research thus far has focused on theoretical design, synthesis and testing MXenes to prove their potential in an array of useful applications,” Gogotsi said. “But to capitalize on the exciting potential of AI technology, we need to retrace our steps and explain the electrochemical forces that created these materials and the structures that give them their physical properties.”

To arrive at its finding, the team synthesized 40 MXene materials, 30 of them new, with varying numbers of layers and metals in each — up to nine different metallic elements in a lattice — to observe variations in atomic structure created by the addition of new elements. Shifts in how atoms fall into order within the material structure serve as indicators of the presiding thermodynamic forces.

By first making a theoretical calculation of their atomic structure, then physically examining each of the materials, layer by layer, using dynamic secondary ion mass spectrometry (SIMS), the researchers observed that MAX phases containing up to six different metals tended toward an orderly, predictable arrangement (enthalpic preference), while those containing seven or more metals demonstrated no such preference or a tendency toward perfectly random mixing of atoms (entropic stabilization).

They also observed how electrical resistance and infrared radiation penetration varied between materials as layers — and number of different metals in the structure — increased, an indication of how the material disbursed energy internally due to its atomic structure. These observations allow the researchers to formulate a principle for making both MXenes and their parent materials, MAX phases, with perfectly mixed atomic structures.

“This study indicates that short-range ordering — the arrangement of atoms over a short distance of a few atomic diameters — in high-entropy materials determines the impact of entropy vs. enthalpy on their structures and properties,” said Brian Wyatt, PhD, a postdoctoral researcher at Purdue and first author of the paper. "For the broad scientific community, this work represents major progress in understanding the role of enthalpy and entropy in the formation and order-disorder transitions in these high-entropy materials. Within layered ceramics and 2D material research, this expands the families of these materials and their potential applications.”

Training AI programs with this data to predict whether certain materials could be stably synthesized and tailored for specific technologies is an exciting prospect for the future of materials science, according Anasori.

“We want to continue pushing the boundaries of what materials can do, especially in extreme environments where current materials fall short,” Anasori said. “The ultimate objective is to create materials that can outperform anything currently known to humanity in these demanding conditions. Whether it is enabling clean energy, longer EV range in extreme cold or extreme heat in space, or crafting materials that function in space or deep-sea conditions, I hope our work can help enable the next generation of technologies.”

END