Organic–inorganic hybrid interfaces play a pivotal role in determining the performance of next-generation optoelectronic devices. The structural organization, orientational ordering, and electronic coupling of molecular adsorbates at these interfaces govern fundamental processes such as charge transfer, exciton dissociation, and energy-level alignment. Despite significant advances in experimental surface science, theoretical prediction of adsorption geometries and interfacial electronic properties for large, conformationally flexible organic molecules remains challenging. The associated potential energy surfaces (PES) are inherently high-dimensional and feature a multitude of shallow local minima, rendering first-principles exploration computationally expensive. The present work addresses this challenge by developing a hybrid first-principles and machine-learning strategy to investigate the adsorption and electronic properties of the merocyanine dye HB238 (2-5-(5-dibutylamino-thiophen-2-ylidenemethyl)-4-tert-butyl-5H-thiazol-2-ylidene-malononitrile) on metallic and insulating substrates. HB238, a donor–acceptor π-conjugated chromophore, is investigated as a model system to elucidate how molecular polarity, structural flexibility, and substrate reactivity collectively influence self-assembly and interfacial electronic coupling. Merocyanines represent an important class of polar organic semiconductors, widely used in organic photovoltaics, photodetectors, and nonlinear optical materials. Their optoelectronic responses including charge-transfer character, absorption spectra, and excitonic coupling are highly sensitive to molecular geometry, intermolecular packing, and the surrounding dielectric environment. The HB238 molecule investigated here, exemplifies this family as it exhibits a pronounced dipole moment and a strong tendency for self-organization in thin films. Experimentally, HB238 has been shown to form ordered chiral monolayers on Ag(100) surfaces, providing an ideal benchmark for validating theoretical models of molecular adsorption and self-assembly. Chapters 3–6 develop a coherent framework for understanding and modeling the optoelectronic properties and surface adsorption behavior of merocyanine dyes, particularly HB238. Chapter 3 benchmarks electronic-structure methods, showing that while TD-DFT overestimates charge-transfer excitations, GW-BSE and TDCP-DFT achieve near-experimental accuracy, establishing reliable tools for studying such systems. Chapter 4 applies dispersion-corrected DFT to HB238 adsorption on Ag(100), reproducing the experimentally observed chiral tetrameric monolayer and emphasizing the importance of stable optimization methods like PreconLBFGS. Chapters 5 and 6 introduce and extend a machine-learning–accelerated MACE×BOSS workflow, enabling efficient exploration of high-dimensional adsorption landscapes on hBN and coinage metal surfaces (Cu, Ag, Au). This approach dramatically reduces computational cost while capturing key interactions, ultimately revealing how substrate properties such as reactivity and electronic screening influence adsorption geometry, intermolecular interactions, and electronic coupling at donor–acceptor interfaces.
Ritu Tomar (Mon,) studied this question.