Precursor Adsorption and Surface-Mediated Decomposition Mechanisms in BN Growth on Si(001): Implications for Low-κ Dielectric Materials

Abstract

The use of two-dimensional materials in the semiconductor industry is growing rapidly, showing promise for dramatically improving device performance. Among the wide range of 2D materials, BN is in the spotlight due to its potential use as an ideal substrate in graphene electronic devices or as intermetal and interlayer dielectric barrier with low-κ. Conventional approaches rely on BN transfer from the growth substrate, typically polycrystalline metals, to the corresponding substrate for characterization or device fabrication, a process not exempt from structural risks or contamination, because it implies the use of postprocessing techniques to obtain samples with the desired characteristics. Therefore, direct growth of BN on semiconductor or dielectric substrates is a necessity for the optimal performance of BN to become a reality. In this work, we use density-functional theory (DFT) and ab initio molecular dynamics (AIMD) to model chemical vapor deposition (CVD) and plasma enhanced CVD (PE-CVD) of a BN monolayer on silicon surfaces. In our simulations, single and multiple compound molecular precursors were modeled, selecting borazine (B₃H₆N₃) and diborane/ammonia (B₂H₆/NH₃) as model systems. Precursor decomposition during plasma-assisted processes determines to a large extent the growing modes of BN, but overall mechanisms for BN growth on semiconductor surfaces are complex and diverse, involving not only surface chemical reactions and kinetic effects but also dynamical effects such as strain-driven precursor dissociation or dissociative chemisorption pathways on surface sites next to a previously deposited molecular precursor. Our simulation results accurately describe not only the initial decomposition reactions and growth stages but also the electronic, thermodynamic, and vibrational signatures of the formed BN monolayer. Then, precursor-dependent bonding characteristics allow for the identification of experimentally measurable relevant properties such as infrared (IR) spectra or κ values, which will help optimize precursor usage and interpret the sometimes obscure experimental information, identifying specific measurements to the corresponding submonolayer BN bonding moieties.