Quantitative Prediction of Perovskite Degradation over a Broad Range of Humidity, Oxygen, and Temperature Using Machine Learning and Training Data from Photoluminescence, Photoconductivity, and Optical Properties
Hugh Hillhouse a b, Ryan Stoddard a b, Wiley Dunlap-Shohl a b
a University of Washington, US, Seattle, United States
b University of Washington, US, Seattle, United States
nanoGe Perovskite Conferences
Proceedings of International Conference on Perovskite Thin Film Photovoltaics and Perovskite Photonics and Optoelectronics (NIPHO20)
Sevilla, Spain, 2020 February 23rd - 25th
Organizer: Hernán Míguez
Invited Speaker, Hugh Hillhouse, presentation 042
Publication date: 25th November 2019

Quantitatively predicting the degradation of perovskite materials and photovoltaic devices under different environmental stresses is vital to developing accelerated aging test and developing long-term stable photovoltaic devices. Previously, we have developed methods to quantitatively measure photoluminescence from thin film perovskite materials to determine the quasi-Fermi level splitting (QFLS), effective temperature, and distribution of sub-bandgap states [1,2]. We have shown that the material QFLS accurately predicts device Voc [2] for a wide-range of perovskites (bandgaps from 1.2 to 1.9 eV) when significant voltage losses from HTLs, ETLs, and contacts are not present, and we have used the method to develop several record Voc devices [3,4], particularly for high bandgap guanidinium-based perovskites [3] and 2D/3D perovskites [4]. We have also shown that an estimate of the diffusion length can be obtained from photoconductivity measurements [2]. Here, we present new results from high-throughput measurement of thousands of perovskite material compositions (and a smaller subset of full photovoltaic devices) over a broad range of environmental stresses including temperature, humidity and oxygen. In addition to spatially averaged quantities (like the QFLS and diffusion length), high spatial resolution and time resolution quantitative photoluminescence videos reveal the role of spatial heterogeneity, photo-brightening, and blinking. Using machine learning methods and this large training data set, we are able to quantitatively predict the time it takes for selected material or device properties to degrade to 75% of their initial value for selected important perovskite compositions to an accuracy better than 10% [submitted for publication]. The presentation will summarize the methods we have develop and show how they may be used to design long-term stable perovskite materials and devices.

We acknowledge primary financial support from the U.S. Department of Energy award DE-EE0008563 and DE-EE0006710. We also acknowledge support from the U.S. National Science Foundation award DMR-1807541.

© Fundació Scito
We use our own and third party cookies for analysing and measuring usage of our website to improve our services. If you continue browsing, we consider accepting its use. You can check our Cookies Policy in which you will also find how to configure your web browser for the use of cookies. More info