Skip to main content

November 16 | Virtual & Free

SigOpt AI & HPC Summit 2021


Talk: Better Glass Design with Multi-Objective Bayesian Optimization

1 - 1:30pm

Paul Leu, Associate Professor in the Laboratory for Advanced Materials (LAMP) at the University of Pittsburgh, will discuss his experience collaborating with the SigOpt team to accelerate the development of new fabrication strategies for glass to improve its performance in key properties, such as reducing haze or reflectance.

Many modern consumer electronic devices such as smartphones and tablets require the use of glass or plastic materials to protect the device’s delicate display. In order to do so, the glass or plastic material must exhibit important characteristics specific to these applications, such as high transparency to allow light to pass through, minimization of haze to reduce blurriness, or resistance to dirt, grease, water or other substances that could impact performance. 

Historically, nanostructured surface research is slow and fragmented due to the use of trial-and-error design methods that often rely on the knowledge and intuition of researchers. Nanostructured surface experiments require the precise selection of various fabrication parameters, such as the flow rate of various gases, ion etching time, chamber pressure, etc. Numerical simulations exist, but can be slow and inaccurate in the most useful circumstances. Moreover, the fabrication process is time consuming: one fabrication in one of our experimental settings requires 16 hours of chemical vapor deposition.

How can we efficiently search for the desired fabrication parameters and, in the process, speed up nanostructured surface research? The answer is multiobjective Bayesian optimization. Bio-inspiration and advances in micro-/nanomanufacturing processes have enabled the design and fabrication of micro-/nanostructures  to create a variety of functionalities. In this talk, we discuss some of my research group’s recent progress in the creation of multi-functional glass using multi-objective Bayesian optimization. In the first part of the talk, we focus on creating new glass with high transparency (high antireflection), low haze, and oil-repellency properties. We discuss functionalities such as self-cleaning, stain-resistance, and anti-fogging. Next, we discuss optimization of solar module glass for both broadband and broad angle antireflection.   We provide examples of how Bayesian optimization may be integrated with both experiments as well as simulations to rapidly create new multi-functional materials.  Multi-functional glass with high transparency may have a wide variety of applications in solar modules and optoelectronic devices. Finally, we also will discuss the new IUCRC between Pitt and Case Western, MDS-Rely, to help bring the worlds of materials science and data science together.


Add to Calendar 11/16/2021 1:00 pm 11/16/2021 1:30 pm America/Los_Angeles Talk: Better Glass Design with Multi-Objective Bayesian Optimization SigOpt AI & HPC Summit 2021 - Virtual & Free


Paul Leu

Associate Professor, University of Pittsburgh