Understanding and optimizing perovskite optoelectronic devices with multi-dimensional imaging techniques

This thesis explores the application of multi-dimensional optical imaging techniques in understanding and optimizing perovskite-based optoelectronics. Chapters 1 and 2 give the motivation behind this work and background to perovskite optoelectronics and machine learning. Chapter 3 introduces the main experimental techniques. The debated passivation strategies on perovskite solar cells (PSCs) are studied in Chapter 4 through quantitative hyperspectral imaging. Specifically, alkali metal passivation imposes distinct effects on the optical and structural properties of the devices based on different transport layers. It is shown that the formation of secondary phases, either due to the additives in the perovskite precursor or in the transport layer, leads to an increase in nonradiative recombination and local open-circuit voltage loss. This provides important guidance to the development of passivation techniques toward efficient and stable PSCs. Chapter 5 expands the capability of the latest hyperspectral microscopy technique by developing a machine-learning-based image processing algorithm that is suitable for scientific research. The proposed algorithm achieves state-of-the-art denoising performances compared to recent ML models and conventional handcrafted algorithms. It is able to strengthen signals from unknown samples under low illumination conditions by exploiting spectral information and adopting a self-learning approach. This enables fast and low-dose measurements for emerging semiconductor materials with poor stability. Chapter 6 characterizes perovskite materials for a wide range of applications using the imaging platform developed in Chapter 5. The degradation of mixed halide perovskite light-emitting diodes (LEDs) is tracked through <i>in-situ</i> PL and <i>in-operando</i> electroluminescence mapping. We reveal that lateral ion migration under device operation leads to the growth of chloride-rich defective regions that emit poorly. This is the first time lateral halide migration is observed in perovskite LEDs due to locally-varying electric fields. Finally, Chapter 7 summarizes all the findings and discusses future research directions. The research presented in this thesis, with approaches across the multidisciplinary scientific fields of physics, material science, and machine learning, paves the way for computer vision-accelerated development of emerging technologies towards commercialization and scale-up.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2022

Author dc:creator

• Ji, Kangyu

Advisor dc:contributor.advisor

• Stranks, Samuel

• Degradation mechanism
• Imaging
• Light-emitting diodes
• Machine learning
• Optoelectronics
• Perovskite
• Solar cells

Licence

• https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/3c4bbfa2-305b-480f-80c2-75974b3bd25b/download
• https://creativecommons.org/licenses/by-nc-nd/4.0/

Language dc:language

eng

Author Identifier

0000-0002-1278-3212

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/349697