An intelligent perovskite solution printing line

Project Information

INT-PVK-PRINT

Grant agreement ID: 101107885

DOI

10.3030/101107885

Project closed

EC signature date 29 March 2023

Start date 1 September 2023

End date 31 August 2025

Funded under

• Marie Skłodowska-Curie Actions (MSCA)

Total cost

No data

EU contribution

€ 188 590,08

Coordinated by

UNIVERSITA DEGLI STUDI DI ROMA TOR VERGATA
Italy

Periodic Reporting for period 1 - INT-PVK-PRINT (An intelligent perovskite solution printing line)

Reporting period: 2023-09-01 to 2025-08-31

Summary of the context and overall objectives of the project

The project an intelligent perovskite printing line tackles the problem of optimizing and controlling solution-processed hybrid perovskite solution films by employing state-of-the-art in situ characterization and Machine Learning methods. Hybrid perovskite absorbers are an emerging semiconductor technology that will likely have an impact on next-generation photovoltaics (PVs). The most promising commerical applications include 1) hybrid perovksites as a lightweight alternative in flexible, printed thin film PV modules or, in combination with an established PV technology, hybrid perovskites as a new partner in the so-called tandem architecture to enhance the conversion efficiency of PVs beyond the physical limit of a single absorber junction-based modules. Still, hybrid perovskites are challenging to fabricate and control of the required high-quality thin film morphology has proven challenging. The proposed project is set up to find new ways of controlling and optimizing hybrid perovksite thin films by 1) employing in situ characterization that can yield early information on the quality of the produced films 2) calulcuating a feedback from state-of-the-art machine learning models and 3) automatically employing the feedback to either control or optimize the perovskite deposition process in real-time. A successful implementation of the project could facilitate contrl and reproduciblity of this emerging technology and thus boost its commercialization to combat the emission of green house gases by PVs with higher efficiency and a lower price.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

Two implement the project the following achievements were executed

1) The researcher was trained in the application of state-ofthe art machine learning (ML) methods and robot control systems
2) A coating recipe of perovskite precursor thin-film was developed on a fully automated slot-die coating system
3) The slot-die coating system was enhanced with a specular and diffuse reflection module that can successuly identify if i) the perovskite has crystallized and ii) the crystal film was smooth or rough
4) A flow controller was installed to control the gas speed and induce variations of gas speeds during the drying process as a function of the measured in situ transients
5) The parameters of the the slot-die coating system were explored in several optimization runs by also using feed forward gradients of gas speeds to gauge if these could improve the thin film quality
6) Since the quality and repeatability of experiments was not sufficient and batch sizes of the automatically coated substrates was insufficient, ML could not be performed on the investigated system.
7) ML (Reinforcement Learning, Bayesian Optimization and RandomForest Optimization) was however carried out in a simulated drying environment, resulting in a recipe how to control the drying process on limited sample sizes (see results).
8) ML was further applied to green solvent screening and image learning on luminescence images captured during the controlled degradation of perovskite solar cells.
9) A review paper was written to point out how process parameters in the field should be reported to improve the reuse and exploitation of data on perovskite fabrication routines and improve reproduciblity within the field.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

- The main result of the project is a modelling framwork that indicates how semiconductor thin film can be dried in a controlled way on a limited substrate size by using an optimized relative movement between the drying nozzle and the substrate. It is shown, in particulor, how the optimal movement trajectory changes if the coated wet film has different shapes as caused by the employed coating method. This result could be very useful to the industrial application of coating perovskite thin film on top of silicon solar cells, where a reproduciblity homogeneous and rapid drying is of utmost importance. The succussful commercialization of silcion-perovksite tandems, in turn, could have a tangible impact of reducing global CO2 emissions.

- Another result of this work is a review that points out the impact of process parameter on perovskite solution processes in general. Thus, by improving the standard of which process parameters should be reported, this work could improve the data gathered on the technology and facilitate the application of machine learning within the community to inpire new experiments or find new materials. Further, reproduciblity of perovskite solution processes wihin the communiy is facilitated if more process parameters, especially on air streams used for drying, are correctly reported.

- The last result of the work is still ongoing and will be released as a green solvent screening tool based on solublity paramters predicted by state-of-the-art foundational models that can be used to select possible solvent substitutes for hyrbird perovskite processes soon. This could help to reduce the impact of perovskite solution processes onto the environment and make the commercialisation more accessible by reducing risk and safety measures for the operators.

main-image.png