Publication date: 17th February 2025
Achieving long term operational stability of perovskite cells under real-world conditions continues to be a major concern. Many mechanisms have been shown to cause degradation, recent examples being exposure to both water and oxygen [1], screening of the internal field driving charge extraction through ion migration [2] and interfacial recombination at the SnOx/bathocuproine interface in the hole blocking layer [3].
Here, we show how degradation mechanisms in a solar cell can be identified from experimental measurements by creating a digital twin, a virtual representation of an object designed to produce an accurate reflection of a physical object [https://www.ibm.com/think/topics/what-is-a-digital-twin]. Through simulations performed in real time, our digital twin can analyse performance changes due to degradation under operation and suggests potential mitigations. Our digital twin is a combination of the device transport model IonMonger [4] and machine learning [5]. It can be used to test hypotheses about the physical processes responsible for degradation. These processes include the role of mobile iodide vacancies in influencing charge transport across the interface through charge accumulation/depletion at interfaces, trap assisted recombination at the interfaces, and contributions of other impurities. The TOC Figure shows example results from our digital twin along with perovskite solar cells fabricated in the Cameron lab https://people.bath.ac.uk/chppjc/research.html.
TOC Figure Left: Perovskite solar cells fabricated in the Cameron lab. Right: Illustrative example joint distributions of two IonMonger input parameters for a degrading device as derived by Bayesian Parameter Estimation.