Static Distortion Waves in Epitaxial Non-Commensurate Adlayers

我们建议的主要科学目标是证明静态扭曲波(SDW)在有机吸附物结构形成中起普遍作用。这些代表相当灵活的吸附晶格，由分子之间的弱相互作用力引起，即范德华力和-如果分子中包含合适的原子-氢键。为此，我们将研究在石墨和金属衬底上生长的具有几个不同方面的各种分子的自组装单分子膜。提出了一种独特的方法来定量评估二维非公度分子聚集体(呈现出广泛的最近邻距离分布)中的相互作用，方法是通过低温扫描隧道显微镜结合失真校正的LEED测量来观察SDW。为了使实验结果合理化，将进行(远离平均晶格)的松弛模拟。这些模拟所需的无参数层内和层间相互作用能量场景将通过最先进的色散校正DFT计算获得。从计算/理论的角度来看，SDW特别有趣，原因有很多。重要的是，在SDW系统中，分子吸附在各种不同的吸附位置上。因此，要描述SDW，仅定量地再现能量上最有利的吸附位置是不够的。相反，必须获得描述分子-底物和分子-分子相互作用的大部分势能面的良好表示。作为第一步，我们将通过对势能面进行离散化和“蛮力强迫”来分别映射这些相互作用。除了已建立的方法，我们将测试一些不同的计算方法来评估分子-底物和分子-分子相互作用。作为第二步，我们将使用机器学习方法加快势能面的评估。具体地说，我们将依靠高斯过程回归(GPR)来确定有机分子与无机底物界面的相互作用。通过将实验和计算结果相结合，将获得关于分子间相互作用强度的定量信息。

The main scientific goal of our proposal is to demonstrate that static distortion waves (SDWs) play a universal role in the structure formation of organic adsorbates. These represent rather flexible adsorbate lattices caused by the weak interaction forces between molecules, namely van der Waals forces and – if suitable atoms are contained in the molecules – hydrogen bonds. To that end, we will investigate self-assembled monolayers of a variety of molecules with several distinct aspects, grown on graphitic and metallic substrates. A unique method is proposed to quantitatively assess interactions in two-dimensional non-commensurate molecular aggregates (which exhibit a broad distribution of nearest-neighbor distances) via the observation of SDWs by means of low-temperature STM accompanied by distortion-corrected LEED measurements. To rationalize the experimental results, relaxation simulations (away from the average lattice) will be performed. The parameter-free intra- and interlayer interaction energy landscapes needed for those simulations will be obtained from state-of-the-art dispersion-corrected DFT calculations.From a computational/theoretical point of view, SDWs are particularly interesting for multiple reasons. Importantly, in SDW systems molecules adsorb on a variety of different adsorption sites. To describe SDWs, it is thus insufficient to only reproduce quantitatively the energetically most favorable adsorption site. Rather, it is imperative to obtain a good representation of a large part of the potential energy surfaces describing the molecule-substrate and molecule-molecule interactions. As a first step, we will map those interactions separately by discretizing and “brute-forcing” the potential energy surfaces. Going beyond the established approach, we will test a number of different computational methods for the evaluation of molecule-substrate and molecule-molecule interactions. As a second step, we will speed up the evaluation of the potential energy surfaces using machine learning methods. Specifically, we will rely on Gaussian Process Regression (GPR) to determine interactions at the interface of organic molecules with inorganic substrates. By combining the experimental and calculation results, quantitative information on the strength of the intermolecular interactions will be obtained.