Under the pressing need to combat climate change, advancing renewable energy technologies has become a global priority. Perovskite photovoltaics (PV), with their remarkable photoelectric conversion efficiency and low-cost production enabled by solution-processable fabrication at relatively low temperatures, have emerged as a promising solution. However, their transition from laboratory research to practical applications faces significant challenges, including stability issues, toxicity concerns, and limitations in material composition. Addressing these challenges requires the development of new energy materials and the optimization of complex parameter spaces, such as material properties, fabrication conditions, and device architectures. Traditional approaches, based on labor-intensive and time-consuming processes, often require thousands or even millions of experiments to identify optimal solutions. This inefficiency underscores the necessity for innovative approaches to accelerate discovery and development. The emerging intelligent and autonomous platforms, known as Materials Acceleration Platforms (MAPs) and Device Acceleration Platforms (DAPs), offer transformative approaches to the PV field. These systems combine automation, high-throughput experimentation, and artificial intelligence (AI) algorithms to explore and optimize complex process-parameter spaces. MAPs help researchers discover and explore new energy materials through automated synthesis, characterization, and AI-driven experimental design and data analysis. DAPs are developed to optimize perovskite films and devices by precisely controlling and refining fabrication conditions. By significantly reducing the number of required experiments from extensive manual trials to a few well-targeted iterations, these platforms offer a highly efficient pathway for optimizing perovskite materials and devices. The application of such platforms holds significant potential to resolve long-standing challenges and to accelerate the large-scale application of perovskite PV in renewable energy systems. Introducing intercalating cations to form quasi-two-dimensional (quasi-2D) structures proves an effective strategy for stabilizing three-dimensional (3D) perovskites. The first part of the thesis presents a simple efficient, low-cost drop-casting method compatible with an automated platform for high-throughput composition screening. With this high-throughput automation method, more than 300 Ruddlesden–Popper-phase perovskite films were fabricated and evaluated to study the impact of various cations and their doping concentrations on thermal stability. The results found that doping 20–25 mol% of large cations into MAPbI₃ achieves optimal stability, while deviations from this range lead to a "stability bowing" effect due to competing protective and defect-inducing mechanisms. This work highlights the potential of quasi-2D structures to enhance perovskite stability and demonstrates the value of combining automated platforms with simple solution-processed methods for rapid material optimization. Building on the method, the second part systematically investigates two distinct cation systems to explore their potential for enhancing perovskite stability through compositional engineering. The thermal stability of linear alkylammonium cations with varying carbon chain lengths revealed distinct behaviors: longer-chain cations enhance thermal stability by suppressing Ostwald ripening through strong steric hindrance, while shorter-chain cations improve stability due to phase redistribution and the regeneration of 3D/3D-like perovskite phases. The results highlight the critical role of carbon chain length in tailoring stability. In the aromatic-based cation system, films with longer phenylpropylammonium chains exhibited lower thermal stability due to lattice distortions and phase mismatches. Conversely, phenylbutanammonium-based films demonstrated superior stability, attributed to unique crystallization dynamics and phase transitions driven by steric hindrance. These findings reveal a deeper understanding of the interplay between molecular structure and perovskite stability, which provide key insights into the design principles for stable perovskite materials. Achieving high-performance perovskite solar cells (PSCs) in ambient air requires precise optimization of complex fabrication parameters. The final part introduces an automated platform with the one-variation-at-a-time method to systematically optimize key parameters for the antisolvent and two-step deposition methods. By systematically exploring the key parameters, SPINBOT achieved significant improvements in perovskite film quality and device performance. To maximize the platform’s potential, machine learning (ML) is integrated to form an autonomous closed-loop framework. This optimization framework efficiently explored complex optimization spaces and identified optimal parameters for achieving high-performance PSCs. Insights from the optimization process are formalized into a standardized operating procedure (SOP) for reproducible fabrication of high-performance devices in ambient air. These studies highlight the potential of AI-driven automation for advancing perovskite photovoltaics. This thesis demonstrates the transformative potential and role of automation in advancing the research and development of perovskite photovoltaic materials beyond traditional research paradigms. By integrating robotic automation, high-throughput experimentation, and machine learning, these platforms hold huge potential to address critical challenges in optimizing the stability and performance of perovskite materials and devices within complex parameter spaces. The AI-driven autonomous frameworks explored in this thesis not only streamline the discovery and optimization of perovskite materials but also open up new possibilities in the PV community, providing researchers with a powerful tool to accelerate the discovery and optimization of functional thin films.
Jiyun Zhang (Thu,) studied this question.