Felix Laufer

Long-term stability of the perovskite top cell remains a hurdle to commercializing perovskite–silicon tandem (PST) solar cells. While accelerated tests provide valuable insights into degradation kinetics, they fail in predicting real-world degradation behavior. Keeping stressors constant, accelerated tests neglect dynamic conditions in actual operational environments, like diurnal and seasonal temperature and irradiance variability. We address this challenge by integrating a degradation function into our energy yield (EY) modelling software which integrates degradation in collection efficiency (and thus photocurrent) over time due to light and heat exposure, bridging the gap between accelerated testing and in real-world stability assessment. By linking the EY model to measurable material parameters like activation energy governing degradation pathways, this approach enables physically grounded degradation modelling. Based on degradation observed under accelerated tests, the model predicts PST operational lifetimes in diverse climates, highlighting the substantial discrepancy between lifetimes measured under accelerated testing and real-world locations. Applied to a PST solar cell, we show that an operational lifetime (T_90,Agg) of about 1400 h under ISOS-L2 (1 Sun, 85 °C), translates to several months in-field (about 26 months in arid Phoenix and 42 months in temperate Seattle), demonstrating strong climate dependence. We also provide a practical mapping from ISOS-L2 to real-world lifetimes and estimate the minimum stability threshold needed for deployment as around 4000 h T_90,Agg under ISOS-L2, translating to more than 5 years of operation across the investigated locations. After device-specific parameterization from appropriate aging tests, this device-agnostic framework allows stability-aware EY modelling to predict real-world degradation.

This study employs in situ probing to control the crystallization dynamics of Sn–Pb mixed perovskites during vacuum-assisted growth and annealing, enabling scalable fabrication of high-quality perovskite films and all-perovskite tandem photovoltaics.

Vapor phase deposition methods are readily able to achieve uniform coverage of large‐area substrates and are widely considered promising for industrial‐scale perovskite solar cell fabrication. However, as perovskite‐silicon tandem solar cells approach commercialization, practical considerations of manufacturing throughput come into play. Here, it is shown that the inherent sublimation characteristics of the organic precursor formamidinium iodide (FAI) make increasing the deposition rate of FA‐based co‐evaporated perovskites negatively impact replicability and lead to a substantial decrease in power conversion efficiency (PCE). These losses are linked to reduced film homogeneity and the emergence of carbon‐rich regions within the perovskite layer. To mitigate these rate‐induced effects, two approaches are explored: source layout optimization and material preconditioning. Utilizing dual FAI sources rather than a single FAI source reduces the relative PCE drop from ≈23% _rel to ≈9% _rel at a deposition rate of ≈18 nm min ⁻¹ (14.8% PCE @ maximum power point (MPP)) compared to the baseline rate of 5 nm min ⁻¹ (16.2% PCE @MPP). Alternatively, preconditioning a single FAI source reduces the performance losses from ≈31% _rel to ≈26% _rel at a deposition rate of ≈21 nm min ⁻¹ . These findings underscore the importance of tailored source strategies to enable high‐rate FA‐based co‐evaporated perovskites without compromising device performance.

Photoluminescence‐based characterization techniques are widely employed in perovskite solar cell research, offering a noninvasive and contactless means of obtaining information about the implied open‐circuit voltage ( ) and, hence, the absorber quality. Driven by the idea of developing a robust yet quantitative in‐line imaging method for perovskite/Si tandem solar cells, we have extended an intensity‐dependent photoluminescence method initially reported for single‐junction solar cells. This method enables local quality assessment of the perovskite thin‐film absorbers processed over planar and textured silicon bottom solar cells with high spatial resolution. A single effective parameter k, also called optical ideality factor, is extracted, which accounts for the complex superposition of locally competing recombination processes. This work demonstrates that our method, called k‐imaging, is a robust and versatile characterization tool for perovskite/Si tandem solar cells that allows the assessment of the general thin‐film absorber quality as well as of specific defects for both scientific and industrial applications. Consequently, it accelerates perovskite research and paves the way for highly reproducible perovskite deposition processes toward commercialized perovskite/Si tandem solar cells.

Vapor phase deposition processes hold great potential for industrializing the deposition of perovskite-based absorbers, offering a pathway to commercialization.

Augmenting characterization methods with deep learning and other machine learning methods allows the identification of material inconsistencies, device performance predictions, and the generation of in situ AI recommendations.

Key advances on triple-junction perovskite–perovskite–Si solar cells with an unprecedented efficiency of 24.4% and enhanced long-term thermal stability are reported via the development of high-performance middle perovskite solar cell.

Large‐area processing of perovskite semiconductor thin‐films is complex and evokes unexplained variance in quality, posing a major hurdle for the commercialization of perovskite photovoltaics. Advances in scalable fabrication processes are currently limited to gradual and arbitrary trial‐and‐error procedures. While the in situ acquisition of photoluminescence (PL) videos has the potential to reveal important variations in the thin‐film formation process, the high dimensionality of the data quickly surpasses the limits of human analysis. In response, this study leverages deep learning (DL) and explainable artificial intelligence (XAI) to discover relationships between sensor information acquired during the perovskite thin‐film formation process and the resulting solar cell performance indicators, while rendering these relationships humanly understandable. The study further shows how gained insights can be distilled into actionable recommendations for perovskite thin‐film processing, advancing toward industrial‐scale solar cell manufacturing. This study demonstrates that XAI methods will play a critical role in accelerating energy materials science.

Remarkable progress in efficiency and stability has been demonstrated for the next generation photovoltaic (PV) technology of thin-film perovskite solar cells (PSCs) on the laboratory scale. Small-area PSCs already exceed 25 % of power conversion efficiency (PCE) since they were first investigated about one decade ago [1, 2]. However, besides the unquestionable potential of perovskite-absorber films for efficient thin-film PV, there are significant challenges when scaling the technology. The reason is that established, industrial deposition techniques such as slot-die coating are difficult to control during the entirety of the complex perovskite film formation - causing low operational stability when fabricating larger areas [3]. In response, this work leverages prior studies of our group on a quantitative model of the drying dynamics based on local heat transfer measurements [9], for controlling the crystallization of slot-die coated perovskite absorber layers dried via gas-quenching. More specifically, we apply and validate and this knowledge by systematically implementing and evaluating the upscaling strategy for fabricating slot-die coated modules on an area of 100 cm2 with an elongated slot nozzle. In situ monitoring with a CCD camera is used to critically evaluate the position of the crystallization front on the sample. This position depends directly on the drying dynamics and thus the applied process parameters. Remarkably, we succeed in demonstrating the accurate correlation of the position of the drying front with the evolving thin-film morphology as predicted by the drying models. We further show that these morphological differences have a direct impact on the achievable PCE of perovskite test devices. These methological findings denote a corner stone toward scaling perovskite fabrication preventing expensive brute force optimization. Thus, the transition from perovskite PV to large scale commercially viable manufacturing plants can be significantly facilitated.

Large‐area processing remains a key challenge for perovskite solar cells (PSCs). Advanced understanding and improved reproducibility of scalable fabrication processes are required to unlock the technology's economic potential. In this regard, machine learning (ML) methods have emerged as a promising tool to accelerate research and unlock the control needed to produce large‐area solution‐processed perovskite thin films. However, a suitable dataset allowing the analysis of a scalable fabrication process is currently missing. Herein, a unique labeled in situ photoluminescence (PL) dataset for blade‐coated PSCs is introduced and explored with unsupervised k‐means clustering, demonstrating the feasibility to derive meaningful insights from such data. Correlations between the obtained clusters and the measured performance of PSC reveal that the in situ PL signal encodes information about the perovskite thin‐film quality. Detrimental mechanisms during thin‐film formation are detected by identifying spatial differences in PL patterns and, consequently, of device performance. In addition, k‐nearest neighbors is applied to predict the performance of PSCs, motivating further investigations into ML‐based in‐line process monitoring of scalable PSC fabrication to detect, understand, and ultimately minimize process variations across iterations.