Rapid and significant gains against climate change require the development of new, environmentally friendly and energy efficient materials. One of the richest veins in the hope of creating such useful compounds is the vast chemical space, where molecular combinations that offer remarkable optical, conductive, magnetic, and thermal transfer properties await discovery.
But finding these new materials is slow.
“Although computational modeling has enabled us to discover and predict the properties of new materials much faster than experiments, these models are not always reliable,” said Heather J. “To speed up the computational discovery of matter, we need better methods to remove uncertainty and make our predictions more accurate.”
Kulik’s lab team set out to address these challenges with a team including Chenru Duan PhD ’22.
A tool for building trust
Kulik and his group focus on transition metal complexes—molecules consisting of metals found in the middle of the periodic table surrounded by organic ligands. These complexes can be extremely reactive, giving them a central role in the catalysis of natural and industrial processes. By modifying the organic and metallic components of these molecules, scientists can create materials with properties that could improve applications such as artificial photosynthesis, solar energy absorption and storage, higher-efficiency OLEDS (organic light-emitting diodes), and device miniaturization.
“The characterization of these complexes and the discovery of new materials is currently slow, often driven by the intuition of the researcher,” says Kulik. “And the process involves compromises. you might find a material that has good luminescent properties, but the metal in the center might be something like iridium, which is extremely rare and toxic.”
Researchers trying to identify non-toxic, Earth-abundant transition metal complexes with beneficial properties tend to pursue a limited number of features with only modest confidence that they are on the right track. “People keep iterating on a particular ligand and getting stuck in local areas of opportunity rather than making large-scale discoveries,” says Kulik.
To address these screening inefficiencies, Kulick’s team developed a new approach, a machine-learning-based “recommendation,” that lets researchers know the optimal model to conduct their search. Their description of this tool was the subject of an article Computational Science of Nature in December.
“This method outperforms all previous approaches and can tell people when to use methods and when they will be reliable,” says Kulik.
The team, led by Duan, began exploring ways to improve the conventional screening approach, density functional theory (DFT), which is based on computational quantum mechanics. He built a machine learning platform to determine how accurate density functional models are in predicting the structure and behavior of transition metal molecules.
“This tool identified which density functionals are most reliable for specific material complexes,” says Kulik. “We verified this by testing the tool against materials it had never encountered before, where it actually chose the most accurate density functionals to predict the material properties.”
A major breakthrough for the team was its decision to use electron density, a fundamental quantum mechanical property of atoms, as input to machine learning. This unique identifier, along with the use of a neural network model for mapping, creates a powerful and efficient aid for researchers who want to determine if they are using the appropriate density functional to characterize their target transition metal complex. “Computations that would take days or weeks, making computational screening nearly impractical, can instead take just hours to produce a reliable result.”
Kulik has incorporated the tool into molSimplify, an open-source tool on the lab’s website that enables researchers anywhere in the world to predict the properties and model transition metal complexes.
Optimization for multiple properties
In a related study they showed in a recent publication JACS Au:Kulik’s group demonstrated an approach to rapidly introduce transition metal complexes with specific properties across a large chemical space.
Their work leapt from a 2021 paper showing that agreement on the properties of a target molecule among a group of different density functionals significantly reduced the uncertainty of model predictions.
Kulik’s team used this insight to demonstrate the first multi-objective optimization. In their study, they successfully discovered molecules that were easy to synthesize with significant light-absorbing properties using earth-abundant metals. They searched 32 million candidate materials, one of the largest areas ever searched for this application. “We isolated complexes that are already known in experimentally synthesized materials, and we recombined them in new ways, which allowed us to maintain a certain synthetic realism,” says Kulik.
After collecting DFT results on 100 compounds in this giant chemical domain, the group trained machine learning models to make predictions on the entire 32 million compound space with the goal of achieving their specific design goals. They repeated this process generation after generation to eliminate compounds with the obvious properties they wanted.
“In the end, we found the nine most promising compounds and found that the specific compounds we selected through machine learning contained fragments (ligands) that had been experimentally synthesized for other applications requiring optical properties with favorable light absorption spectra,” he says. Kulik:
While Kulick’s primary focus involves overcoming the limitations of computational modeling, his lab takes full advantage of its proprietary tools to streamline the discovery and design of new, potentially impactful materials.
In one notable example, “We are actively working on optimizing metal-organic frameworks for the direct conversion of methane to methanol,” says Kulik. “It’s a holy grail reaction that people have wanted to catalyze for decades but haven’t been able to do effectively.”
The possibility of a quick way to turn a very powerful greenhouse gas into an easily transportable liquid is a big attraction for Kulik. “It represents one of the challenges that multi-site optimization and screening of millions of candidate catalysts has to deal with, an outstanding challenge that has existed for so long.”