TMCnet News
An Inverse Approach to Materials Discovery and Design [Chemical Engineering Progress](Chemical Engineering Progress Via Acquire Media NewsEdge) AIChE JOURNAL Highlight The success of many emerging technologies hinges on the discovery of new materials that exhibit a combination of desired properties (e.g., mechanical, optical, catalytic, etc.). Identification, synthesis, and fabrication of suitable materials for these applications remain challenging, relying on resource-intensive exploratory science. The classical (or forward) discovery approach involves inspecting the properties of a diverse array of existing materials and selecting the one that best fits the target characteristics, or fabricating and characterizing new materials through combinatorial analysis of promising material precursors and synthesis techniques. From an engineering design perspective, a better approach might be to adopt an inverse strategy that begins with the target macroscopic properties in mind and then systematically narrows the list of possible material precursors and synthetic approaches consistent with these properties through an optimization technique. Such inverse design methods are the topic of the August AIChE Journal Perspective article, "Inverse Methods for Material Design," by Avni Jain, Jonathan Bollinger, and Thomas Truskett of the Univ. of Texas at Austin. The inverse design method described by Jain, et al., relies on the following concept: Macroscopic behavior can be tuned via the underlying structural characteristics of a material exhibited at multiple length scales (from the mesoscopic to the molecular). It uses appropriate physical models, which can range from continuum fluid mechanics to quantum mechanics, depending on the application, to relate structural characteristics to target properties. Once these relationships are identified, computational optimization strategies (e.g., simulated annealing) can be employed to iteratively search for material structures that exhibit the target properties. Given the desired structures, inverse strategies can be extended to identify the best synthesis or fabrication protocols for these materials. Several recent studies illustrate how computational inverse strategies address complex materials design problems, the authors write. For example, researchers at the Technical Univ. of Denmark used finite-element-based topology optimization to create thin-film dielectric material morphologies to produce specific and high-intensity color responses (when exposed to light), without the use of pigments. In another study, researchers used inverse strategies to identify structural characteristics common in disordered photonic band-gap (PBG) materials. The authors provide several examples in which the material designs displayed characteristics that could not have been easily anticipated prior to optimization. Computer simulations and experiments have validated the material designs. Although the inverse approach can be used to separately identify structures with desired properties and synthesis methods to create particular structures, the authors note that this approach may not always be reasonable or advantageous. "Especially when macroscopic properties can be related to the chemical makeup of the material's precursors, it may be possible to tune the precursors directly for the desired property, without explicitly specifying structural connections," Truskett says. "Additionally, in some problems, such as in quantum-mechanical-based catalyst design, one can discover novel microscopic components for a target property within defined structural constraints, which may be known a priori due to fabrication-related limitations or otherwise," he elaborates. Despite the advantages, in terms of industrial application, inverse strategies remain in their infancy, the authors note. For some fields, the challenges are shortcomings in the physical models needed to connect material properties with underlying structures and fabrication-relevant variables. For example, in designing colloidal particles that self-assemble into desired superlattices, very few thermodynamic models adequately predict the physics by which nanoand micro-scale particles interact with one another. Furthermore, existing models do not necessarily describe the physics in terms of practical parameters that scientists and engineers can easily control during particle synthesis. Such knowledge gaps, while certainly addressable in the next decade, currently limit the number of applications for which inverse computational methods can provide quantitative design guidance. To push inverse methods to the forefront of solving material design problems, engineers and scientists must invest more time in theoretical research and development, the authors conclude. The availability of rapidly solvable and accurate physical models that describe connections among properties, structures, and assembly will accelerate the utilization of such models in materials science research. When combined with experiments that test solutions obtained by inverse optimization methods, a feedback loop can be created that both iteratively improves physical models and provides insightful jumping-off points for experimental work. In streamlining the discovery and design of new materials for an ever-expanding portfolio of technologies, inverse strategies will surely play an important role. (c) 2014 American Institute of Chemical Engineers |