Solubility and Miscibility of Organic Semiconductors for Efficient and Stable Organic Solar Cells Investigated via Machine Learning and Quantum Chemistry Methods

Abstract

Solubility and miscibility are one of the biggest challenges for the performance of or- ganic solar cells. Exploring these thermodynamical properties help us to understand the phase behaviors and to control the bulk heterojunction microstructure, which are highly correlated to the device performance. In this work, a model to predict the microstructure evaluation of photovoltaic organic blends based on the relative miscibility predictions is shown. Particularly, a numerical approach based on quan- tum chemistry and various models of machine learning (ML) to quickly and reliably determine the solubility parameters of organic semiconductors is presented. The ML implemented was trained using a wide range of chemical compounds. Specifically, ML transforms the molecular surface charge density distribution (sigma-profile) as determined by density functional theory (DFT) calculations under the continuum solvation model into solubility parameters. Noteworthy, the ML can correctly pre- dict the dispersive contributions to the solubility parameters of selected fullerenes and other organic compounds, although no explicit information on the van der Waals forces is present in the sigma-profile. Besides, miscibility by calculating phase di- agrams, interaction parameter, and a proposed figure of merit with this method is analyzed. The theoretical framework based on the numerical approach in the polymeric solution theory to study the microstructure evolution of organic photo- voltaic blends. Here, the solubility parameters are employed for the interaction parameters determination. Significantly, my theoretical predictions are in excel- lent agreement with literature values and our experimental data. Making use of the interaction parameters, a figure of merit (FoM) as a relative metric to evalu- ate the phase diagrams of thin film organic semiconductor blends was calculated. The study is displayed for polymer:fullerene mixtures for the prototypical polymers poly(3-hexylthiophen-2,5-diyl) (P3HT) and poly[(5,6-difluoro-2,1,3 -benzothiazol- 4,7-diyl)-alto (3,3-di(2-octyldodecyl)-2,2,5,2;5,2,2- quaterthio -5,5-diyl)] (PffBT4T- 2OD or also called PCE11). After corroborating the relevance of my approximation with an extensive range of materials with different physical chemical properties, I suggest that this model should be able to communicate the possible design criteria and processing guidelines for existing and new high-performance organic semicon- ductor blends for organic semiconductor applications with a typical stable solid- state morphology. Our work pointed out a basis for utilizing various miscibility tests and future simulation methods that will significantly reduce the common and cumbersome trial-and-error approaches for material synthesis and device fabrication of functional semiconducting blends. Several collaborations were established on my initiative - external collaborations which include Pannonia University (Hungary), Harvard University (USA), Mas- sachusetts Institute of Technology (USA), Max Plank of Colloids and Interfaces (Germany), and contribution for the development and interchanges with certain Colombian universities like Universidad del Valle, Universidad del Quindío, and Universidad de Los Andes. I have organized by my initiative an international con- ference, which was held in December 2017 in Cali - Colombia (’Next Generation of Solar Energy’, www.ngse.info) and a social-scientific project with Colombian’s public school children in the fabrication of organic solar cells that terminate in the outer space in collaboration with NASA, Clubes de Ciencia, and Cubes in the space. For the academic collaboration, multiple topics were done. These topics include; the basis of quantum chemistry calculations, machine learning approaches for determining thermodynamical properties, molecular mechanics (MM+) for the intramolecular interaction studies of pyrene-substituted silicon phthalocyanines, two cases of molecular dynamics (MD) GAFF and OPLS-AA for prediction on the structures behavior in novel materials, effective mass approach for quantum con- finement calculations for the colloidal nano-crystals halide perovskites FAPbX3 (X = Cl, Br, I), collaborative training a drop casting experiment robot optimization, and evaluation of stability based on relative miscibility studies. The summary of my academic work is illustrated in Google scholar (https://scholar.google.de/ cita- tions?user=y2l6lYcAAAAJhl=en) This work is divided mainly in four major sections: In section I, an introduction in which the motivation, general physical aspects related to organic solar cells, thermo- dynamic properties related to the solubility of organic compounds, and the aim and state of the art of this thesis are presented. In section II, the solubility parameters determination for organic semiconductors by utilizing machine-learning approaches is displayed. An extension and application of the solubility parameter in the poly- mer solar cells are presented, here I focus on the interaction parameter and phase separation calculation. The figure of merit (FoM) for relative solubility is proposed. The FoM to study the microstructure stability evaluation is exposed. In section III, the appendices are shown in which the experiments employed, calculated data tables, and the python codes for the miscibility calculator and drying kinetics are displayed. Lastly, in section IV, I attached my curriculum vitae. Key words:Organic semiconductor, solubility parameters, quantum computa- tional methods, machine learning models, artificial neural networks, ab-initio, density functional theory, the Flory-Huggins model, the figure of merit for relative miscibility. Show more