research

Disordered Hyperuniformity (DHU) is a recently discovered novel state of condensed matter. DHU materials behave like a perfect crystal on large length scales in the way how they suppress density fluctuations and behave like amorphous materials on local scales without any Brag peaks. Such unique feature has led to many novel physical properties in DHU materials, including large isotropic photonic band gaps, optimal transport and mechanical properties.

We have been involved in the introduction of the notation of DHU, and discovered this state in disordered particle packings, avian photo-receptor patterns, heterogeneous materials, laminar flow fields, and more recently, amorphous 2D materials. We are currently developing theory of dynamic DHU random fields, field theory of hyperuniformity, generalizing the notation of DHU to quantum materials and quantum states, and understand how DHU correlations give rise to unique quantum transport properties in these materials.

Collective cell migration is crucial to many physiological and pathological processes such as embryo development, wound healing, and cancer invasion. Recent experimental studies have indicated that the active traction forces generated by migrating cells in fibreous extracellular matrix (ECM) can mechanically remodel the ECM, giving rise to bundle-like meso-structures bridging individual cells. Such fiber bundles also enable long-range propagation of cellular forces, leading to correlated migration dynamics regulated by the mechanical communication among the cells.

Motivated by these experimental discoveries, we develop novel computational frameworks explicitly incorporating complex cell-ECM interactions for modeling the rich spectrum of emergent multi-cellular migration dynamics regulated by ECM-mediated mechanical communications.

Disordered systems abound in condensed matter physics and materials science. Examples include ionic liquids, glassy state of matter, granular materials, porous media, and most composites and alloys. Statistical quantification of such systems, a first step towards any comprehensive theories, remains very challenging.

We have developed a framework based on the widest class of statistical morphological descriptors, including the standard n-point correlation functions, various surface functions and cluster functions, encoding different geometrical and topological information. Recently, we devised a novel set of n-point polytope functions and employ neural architecture search to identify and extract most sensitive higher order statistics for given systems. We are also developing theories for the spatial-temporal correlation functions for representation learning for 4D Materials.

Heterogeneous materials are ubiquitous in nature and synthetic situations. Examples include alloys, ceramics, composites, sandstones, porous media, and bone structures. Over a broad range of length scales, such materials exhibit a rich variety of microstructures with varying degrees of disorder, which also determine their effective bulk properties.

We have devised highly efficient reconstruction procedure that enables us to produce virtual 3D microstructures from limited morphological information obtained from tomography or scattering experiments. This allows us to characterize and model material microstructure evaluation over time, i.e., the 4D materials science. Subsequent analysis can be performed on the virtual microstructure to obtain effective material properties and material behavior under extreme conditions.

Composite materials are superior to any of the components by possessing novel and optimized properties. Optimized multi-functional composites are of great importance in many industrial applications, including high-efficiency solid oxide fuel cells and ultra-light weight zero-thermal expansion composites for space shuttle.

We have developed an optimization procedure that enables one to systematically design composites with required or optimized properties, by setting competitions between a variety of desired properties. The designed composite microstructures subsequently can be fabricated using 3D lithography or other techniques. It is also interesting to explore the origins of many biological structures, which a multi-functional composites, from an optimization viewpoint.

Soft matters are composed of mesoscale building blocks, which are themselves made of a large number of atoms and molecules. Examples of soft matters include colloids, micro-emulsions, gels and foams, for which ''h-bar" plays virtually no role yet thermal fluctuations can significantly distort the system (thus, they are "soft"). The study of soft matters had resulted in an emerging field of mesoscale physics.

We have devised efficient molecular dynamics and Monte-Carlo simulation techniques that enable us to systematically study how the characteristics of the fundamental building blocks (e.g., the shape of nanoparticles) affect the structure and properties of their large scale assembly. This in turn allows us to design tailored building blocks, which can self-assemble into large scale materials with desirable structure and properties.

Granular materials are systems composed of discrete grains that interact via short-range interactions. Examples of granular materials include concrete, sandstone, rocket propellant, table salt and sleeping pills. Depending on the external conditions (e.g., volume and stress), a granular material can behave both like a fluid and a solid, linked to one another by the so-called "jamming transition." A jammed granular material is a far-from equilibrium system.

We have developed efficient discrete-element method to investigate the flow and jamming of granular materials composed of anisotropic grains. This enables us to tailor the shape and size distributions of the individual grains, leading to optimized granular materials for specific applications, e.g., ultra-dense concrete and high combustion-efficiency propellant.

The dynamics of malignant brain tumor growth and invasion are still medical mysteries. In spite of aggressive conventional and advanced treatments, the prognosis remains uniformly fatal with a median survival time for patients with most cancers. The rapid growth and resilience of tumors suggest that they are emerging opportunistic systems, that not only adapt to their environment but also change their environment for survival purposes. If this hypothesis holds true, a growing tumor must be investigated and treated as a self-organizing complex dynamical system.

We have developed a powerful cellular automaton (CA) simulation tool to model the growth and invasion dynamics of solid tumors. Our CA model enables us to investigate the emergence of sub tumor species, the effects of treatment, angiogenesis and vasculature evolution as well as the effects of physical confinement on tumor growth, invasion and metastasis.

Understanding cell mechanics, cell-cell interaction and cell motility are crucial to the accurate modeling of the invasion of malignant tumor cells. We are currently developing multi-scale simulation techniques that can take into account the cell level heterogeneity for tumor invasion modeling yet allow the development of a full-sized tumor mass containing billions of malignant and stromal cells. In addition, we are applying the powerful techniques in heterogeneous material theory to characterize tumor micro-environments.