THE COMPUTATIONAL NANOMATERIALS LABORATORY
Glass dissolution
​The fundamental challenge
In crystalline materials, like most minerals, a single unit cell reflects the entire lattice, allowing theoretical frameworks that rely on the regularities of the dissolving surface, like the terrace–ledge–kink (TLK) model, to successfully capture the governing dissolution mechanisms. In contrast, glasses are structurally disordered solids that display a wide range of local atomic environments over nanometer length scales. As a result, predictive descriptors that linking structure or chemical composition to to behavior remain elusive. In our view, glasses require a fundamentally different approach; one that integrates a statistical understanding of the ensemble of local configurations.
​
Why it matters​
Beyond being a fascinating and fundamental scientific challenge, glass dissolution is central to a range of technological and environmental applications. In nuclear waste immobilization, borosilicate glasses must remain stable over geological timescales. In biomedical implants, bioactive glasses are designed to dissolve in controlled ways to stimulate tissue regeneration. In optical and electronic devices, even trace corrosion can impair performance and longevity. Across these applications, predicting when and how a glass will degrade remains a major open problem.
​
We are particularly interested in glasses in the context of sustainable infrastructure, where the reactivity of supplementary cementitious materials (SCMs)—sustainable alternatives that can partially replace Portland cement—depends on how amorphous calcium aluminosilicate (CAS) phases dissolve in water and participate in hydration reactions. Our limited understanding of this dissolution process remains a critical bottleneck to identifying and engineering more effective replacements for Portland cement.
​
Our approach​​
In this project, we are deploying a multiscale simulation framework to understand how the chemical composition and structure of CAS glasses influence their dissolution kinetics. The approach integrates free energy barriers from electronic structure calculations into a graph-based kinetic Monte Carlo model to capture dissolution over realistic time and length scales. The graphs are mapped from atomistic simulations of realistic glass structures. This project is being conducted in collaboration with Dr. Prannoy Suraneni at the University of Miami, and it is sponsored by the Engineering for Civil Infrastructure (ECI) program at NSF.
High-entropy alloy (HEA) electrocatalysts
The fundamental challenge
​A catalyst’s performance is closely tied to how strongly reaction intermediates adsorb onto its surface—a property that can be computed using density functional theory (DFT). Conventional catalysts, typically composed of one or two metals, offer a limited number of active sites. While this restricts the tunability, and therefore the scope of application, it makes them amenable to systematic computational screening. In contrast, high-entropy alloys (HEAs), which are solid solutions of four or more metallic elements, span a vast compositional space and exhibit even greater configurational diversity at the surface. This richness enables nearly unlimited tunability of catalytic properties, but also generates an enormous number of possible active sites, making exhaustive screening by conventional simulations or experiments infeasible.​​
​
​Why it matters​
​​High-entropy alloys (HEAs) composed of earth-abundant, non-toxic metals are emerging as promising candidates for sustainable electrocatalysis. Unlike noble-metal catalysts, which are expensive and scarce, HEAs offer a pathway to affordable alternatives with comparable or superior performance. Notably, HEAs often exhibit enhanced electrochemical stability that exceeds that of their individual components. This makes them ideal for applications where cost-effectiveness is critical.
​
We are particularly interested in developing electrocatalysts that selectively generate strong oxidants like hydroxyl radicals (•OH), which are crucial for advanced oxidation processes (AOPs) in wastewater treatment applications. Unlike the oxygen evolution reaction (OER), which most electrocatalyst research aims to enhance, the goal here is to suppress OER in favor of •OH production. This presents a significant knowledge gap, as the design principles for favoring •OH generation over OER are less understood. Boron-doped diamond (BDD) anodes are among the few materials known to efficiently produce free •OH radicals, however, BDD electrodes are expensive and not widely adopted, highlighting the need for alternative, cost-effective materials that can achieve similar performance in selective •OH generation.
​
Our approach
​​We have developed an integrated computational framework that combines density functional theory (DFT) and machine learning (ML) to characterize the elctrocatalytic properties of high-entropy alloys (HEAs). DFT is used to compute adsorption energies of oxygen intermediates—O, HO, and HOO*—across diverse local surface environments. We then train ML regression models to predict adsorption energies in any given local atomic configuration and use them to sample large synthetic ensembles, reconstructing adsorption energy distributions with high fidelity. This approach provides a scalable path to understanding catalysis on configurationally disordered surfaces. This project is being conducted in collaboration with Dr. Francisco Raymo at the University of Miami, Dr. Natalia Soares Quinete at Florida International University, and it is sponsored by the Environmental Engineering program at NSF.
Morphogenesis of bacterial biofilms
The fundamental challenge
​Bacterial biofilms are complex, adaptive systems whose development is shaped by coupled biological and physicochemical processes—such as cell proliferation, gene expression, nutrient and oxygen diffusion and uptake, or the mechanical behavior of the extracellular matrix. Nonlinear feedbacks among these processes give rise to emergent collective behaviors across multiple spatial and temporal scales, including wrinkling and spatial patterns of phenotypic expression. However, most existing models remain disciplinary, focusing on individual components and failing to capture the integrated dynamics needed to explain and predict biofilm structure and function. As a result, our ability to predict how biofilms respond to interventions or environmental changes remains severely limited. Bridging this gap requires multiscale, integrative models that resolve the feedback between biological activity and physical constraints.​​
​
​Why it matters​
​​Biofilms are the predominant mode of microbial life on Earth and play critical roles across health, industry, and the environment. In medical contexts, biofilms contribute to persistent infections by resisting antibiotic treatment. In industrial settings, they can cause biofouling in water systems and pipelines.
​​
We are particularly interested in how the structure and morphology of a biofilm influence its function—affecting nutrient transport, mechanical stability, dispersal, and resistance to external stresses. A deeper understanding of how biofilm architecture emerges and evolves is essential for developing strategies to control harmful biofilms and engineer beneficial ones.
​
Our approach
​​We are developing a biophysical modeling framework to capture the morphogenesis of bacterial biofilms by integrating biomass growth, nutrient transport, phenotypic expression, and mechanical interactions with the substrate. The biofilm is modeled as a growing elastic sheet whose expansion is driven by local nutrient availability and modulated by substrate friction and adhesion. These mechanical interactions are spatially heterogeneous, enabling the model to account for variations in environmental conditions and surface properties. Phenotypic expression and other biological functions—such as matrix production and metabolic activity—are incorporated using probabilistic rules that depend on local nutrient levels, mechanical stress, and cell lineage history. This allows the model to simulate spatial patterns of differentiation that emerge alongside physical deformations. This project is being conducted in collaboration with Dr. Diana Fusco at Cambridge, and Dr. Carolina Tropini at UCB, and it is sponsored by the Human Frontier Science Program (HFSP).