[Paper Review] A Digital Twin to overcome long-time challenges in Photovoltaics

Larry Lüer, Ian Marius Peters+10 more|arXiv (Cornell University)|May 12, 2023

Machine Learning in Materials ScienceMaterials Science3 citations

TL;DR

This paper proposes a Digital Twin framework for photovoltaic (PV) materials that integrates machine learning, physics-based models, and high-throughput experimentation to overcome the longstanding challenge of linking molecular structure to device performance. By enabling inverse molecular design through data-driven structural prediction, the framework accelerates the discovery of materials that simultaneously meet performance, longevity, and recyclability requirements.

ABSTRACT

The recent successes of emerging photovoltaics (PV) such as organic and perovskite solar cells are largely driven by innovations in material science. However, closing the gap to commercialization still requires significant innovation to match contradicting requirements such as performance, longevity and recyclability. The rate of innovation, as of today, is limited by a lack of design principles linking chemical motifs to functional microscopic structures, and by an incapacity to experimentally access microscopic structures from investigating macroscopic device properties. In this work, we envision a layout of a Digital Twin for PV materials aimed at removing both limitations. The layout combines machine learning approaches, as performed in materials acceleration platforms (MAPs), with mathematical models derived from the underlying physics and digital twin concepts from the engineering world. This layout will allow using high-throughput (HT) experimentation in MAPs to improve the parametrization of quantum chemical and solid-state models. In turn, the improved and generalized models can be used to obtain the crucial structural parameters from HT data. HT experimentation will thus yield a detailed understanding of generally valid structure-property relationships, enabling inverse molecular design, that is, predicting the optimal chemical structure and process conditions to build PV devices satisfying a multitude of requirements at the same time. After motivating our proposed layout of the digital twin with causal relationships in material science, we discuss the current state of the enabling technologies, already being able to yield insight from HT data today. We identify open challenges with respect to the multiscale nature of PV materials and the needed volume and diversity of data, and mention promising approaches to address these challenges.

Motivation & Objective

• To address the persistent gap between material innovation and commercialization in emerging photovoltaics like perovskites and organics.
• To overcome the lack of design principles linking chemical motifs to functional microscopic structures in PV materials.
• To enable experimental inference of microscopic structures from macroscopic device properties, which remains a major bottleneck in materials development.
• To establish a unified framework that combines materials acceleration platforms (MAPs), digital twin concepts, and multiscale modeling for predictive materials design.
• To facilitate the simultaneous optimization of performance, longevity, and recyclability in next-generation PV devices.

Proposed method

• Integrate machine learning models trained on high-throughput (HT) experimental data to parametrize quantum chemical and solid-state models.
• Use physics-derived mathematical models to describe charge transport, defect formation, and degradation mechanisms at multiple scales.
• Implement a digital twin architecture that iteratively refines models using experimental feedback from HT characterization of PV materials.
• Leverage the digital twin to invert device-level performance data and predict optimal molecular structures and processing conditions.
• Combine multiscale modeling with data-driven learning to generalize structure-property relationships across diverse material systems.
• Utilize causal modeling to trace relationships from chemical structure to macroscopic device behavior through intermediate structural and energetic parameters.

Experimental results

Research questions

• RQ1 How can a digital twin framework bridge the gap between molecular structure and macroscopic photovoltaic performance?
• RQ2 What role do high-throughput experimental datasets play in calibrating and validating physics-based and machine learning models in PV materials?
• RQ3 Can inverse molecular design be achieved by inferring optimal chemical structures from device-level performance metrics?
• RQ4 How can the multiscale nature of PV materials—spanning electronic, atomic, and morphological scales—be effectively modeled and integrated?
• RQ5 What data diversity and volume are required to train robust, generalizable models for predictive materials discovery in photovoltaics?

Key findings

• The proposed digital twin framework enables the inference of microscopic structural parameters from macroscopic device properties through integrated modeling and data-driven learning.
• High-throughput experimentation provides the necessary data volume and diversity to train and refine quantum chemical and solid-state models for accurate predictions.
• The integration of machine learning with physics-based models allows for the generalization of structure-property relationships across different PV material families.
• The framework supports inverse molecular design by predicting optimal chemical structures and processing conditions that satisfy multiple performance, stability, and recyclability criteria simultaneously.
• Current enabling technologies already allow insight extraction from high-throughput data, but challenges remain in data quality, scale, and multiscale model coherence.
• The approach is positioned as a transformative pathway toward accelerating the commercialization of next-generation photovoltaics.