An international team of researchers demonstrated a high-throughput platform to aid in the discovery of high-performance organic solar cells suitable for scalable manufacturing. They used digital twin technology and roll-to-roll (R2R) printing in a closed-loop system.
Dubbed MicroFactory, the new platform was used to fabricate, characterize, and analyze 11,800 non-fullerene acceptor (NFA) organic PV devices within a 24-hour period. After analyzing the initial devices, the team used a large dataset of fabrication and characterization parameters to train and optimize machine learning models.
In a subsequent iteration 1,200 devices with improved PCEs were developed based on “machine learning for the inverse generation of parameters”, according to Leonard Ng Wei Tat, the research’s co-corresponding author, who highlighted that champion efficiencies of up to 9.35% were recorded, exhibiting a “1% improvement in just one cycle.”
It was necessary to modify off-the-shelf equipment to make the processing and characterization systems needed for the study. “R2R equipment is commercially available but most of them are meant for graphics printing purposes and are mostly not fit for purpose for printing solar cells,” Ng told pv magazine.
“What we were essentially doing is applying a centuries-old, mature, technology toward the high-throughput fabrication of PV cells. The concept of printing solar cells is simple, depositing layer after layer of functional material until we build a heterostructure that can act as the different components required of a solar cell,” said Ng.
The equipment consisted of slot-die coating and annealing sub-systems with integrated sensors. “We fabricate the PV cells by depositing functional layers onto a strip of polyethylene terephthalate (PET) with a patterned transparent conducting electrode (TCE),” stated the research team. The functional layers consisted of a conductive polymer layer, PEDOT:PSS, a silver grid layer, and zinc oxide nanoparticles.
Multiple sensors collected the data of 36 process parameters, which are stored in a database on a remote data server to be used in the digital twin models. “These models suggested specific alterations to vital fabrication parameters, especially the donor-acceptor (D:A) ratio, and also allowed the incorporation of new, reported scientific knowledge that included the introduction of new interface layers,” stated the research team.
The ability to collect a large amount of data enabled analysis and identification of trends and performance factors. As an example, Ng pointed to the discovery that humidity control was much more important than temperature control in ensuring good quality devices. “This correlates a lot with the observed trends of our fabricated solar cells being better performing in the low-humidity conditions of winter rather than summer, despite producing them in the same air-conditioned environment during both seasons,” said Ng.
The scientists stressed that the iterative approach, informed by machine learning insights, represents a strategic optimization as a digital-twin-driven alternative to traditional design of experiments. “For instance, large scale manufacturers of solar modules can quickly create simple digital twins of their processes in order to build large datasets to identify factors that really move the needle to enhance productivity and yield,” said Ng.
The research team required interdisciplinary skillset including material science, hardware and software development skills, and machine learning knowledge. “Most researchers are familiar with one domain only and it requires a lot of coordination and effort to contextualize things for each other,” said Ng.
The details of the study are discussed in “A printing-inspired digital twin for the self-driving, high-throughput, closed-loop optimization of roll-to-roll printed photovoltaics,” published in Cell Reports Physical Science. The research team members are from Australia's Commonwealth Scientific and Industrial Research Organisation (CSIRO), South Korea's Pukyong National University and Singapore's Nanyang Technological University.
Looking ahead the researchers are investigating new material and device architectures for higher efficiency flexible solar cells, as well as continuing to apply artificial intelligence (AI) technologies, digital-twin, and inverse parameter generation capabilities in other processes, such as batch processing, and traditional solar fabrication. “Eventually, we hope to develop a unified system that will be able to connect multiple machines, factories and laboratories across the world which will allow for the application of more advanced AI,” said Ng.
This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: editors@pv-magazine.com.
By submitting this form you agree to pv magazine using your data for the purposes of publishing your comment.
Your personal data will only be disclosed or otherwise transmitted to third parties for the purposes of spam filtering or if this is necessary for technical maintenance of the website. Any other transfer to third parties will not take place unless this is justified on the basis of applicable data protection regulations or if pv magazine is legally obliged to do so.
You may revoke this consent at any time with effect for the future, in which case your personal data will be deleted immediately. Otherwise, your data will be deleted if pv magazine has processed your request or the purpose of data storage is fulfilled.
Further information on data privacy can be found in our Data Protection Policy.