Home
Solar Power for Affordable Housing through Computational Design of Low-Cost/High-Efficiency Solar Cells
This project is part of the Intro to HPC Bootcamp at Lawrence Berkeley National Laboratory in August 2023 and hosted by the Department of Energy (DOE) Advanced Scientific Computing Research (ASCR) Computing Facilities.
World leaders argue that societies are in the transition to a Third Industrial Revolution (Rifkin, 2018). This period is characterized by the rapid deployment of technologies for transport, automatization, and clean energy production and storage. In the case of electricity, which fuels many activities from transportation to cloud services. Its sustainable production and storage are important to fight climate change and promote more egalitarian societies. Today frontline communities and individuals could own and control the sources of renewable energy, for example harvesting solar energy. There are many types of solar cells, some are more efficient, some are cheaper. Organic dye sensitized solar cells (ODSSCs) are a promising new technology for clean energy production. ODSSCs are made from organic materials that are inexpensive and easy to manufacture. They are also flexible and lightweight, making them ideal for use in portable devices and building-integrated photovoltaics. ODSSCs are less efficient than silicon-based, but cheaper. The cost of 1 kWh produced with ODSSCs could be less than $0.10. In this project, we discuss economic and affordable energy production that could bring energy justice by collecting solar energy with ODSSCS solar cells and Artificial Intelligence to find eco-friendly materials and solve pressing problems. We will use data science, visualization, and machine learning approaches to study a database of molecules for ODSSCs. We will use these tools to explore molecular data sets; we will discuss the sources of the data and identify trends, furthermore, we will analyze molecular descriptors and apply machine learning to predict properties of unknown molecules. This project is intended for students with an interest in data science, renewable energy, or materials science. Prior experience in computational chemistry or data science is not required, although it is advisable to have some knowledge of linear algebra, the basics of programming, and python language.
- Explore an autogenerated data set of dyes candidates for DSSC. Understand patterns and correlations. Augment data set with descriptors and fingerprints.
- Visualize the data using dimensional reduction. Find patterns in the composition and correlate them with properties.
- Find families for molecules with similar properties using clustering techniques. Find what molecular groups contribute to better light absorption.
- Predict molecular properties using machine learning approaches. Compare predictive performance and computational cost of machine learning models.
- Use hardware acceleration to train machine learning models over large amounts of data.
- Provide a list of the best dyes to cover most of the solar spectra.
- How to fabricate a dye sensitized solar cell
- 3rd Industrial Revolution by Jeremy Rifkin
