Hybrid Interfaces in Thermodynamic Equilibrium (HI-TEq)

相图包含了基础研究和应用研究的关键信息，揭示了压力(P)和温度(T)条件下特定相的形成。然而，对于有机-无机界面，实验相图(作为T和p的函数)实际上是不存在的。同时，依赖于从头算热力学的相图计算通常存在严重的近似性，因为振动和热效应往往被忽略。这个项目的目标是开发获得有机-无机界面的p-T相图的策略。在实验方面，这需要开发出在明确定义的条件下使界面接近热力学平衡的方法。在理论方面，需要评估热效应和振动效应对不同晶型的相对能量的影响。我们假设，包括这些影响对于准确的预测是不可或缺的，并设计出一种方法，在机器学习的帮助下将这些影响纳入其中。为了从相图中提取最大的洞察力，我们还提供了对哪些相互作用稳定特定相的深入分析。界面将在接近热力学平衡的新型真空室中生长，该真空室内装有几乎封闭的低温屏蔽层，并使用小分子来促进理论评估。这些结构将用经过失真校正的低能电子衍射进行原位研究。相图的计算预测和评价是一种依赖于密度泛函理论和贝叶斯线性回归形式的机器学习相结合的专门的结构搜索算法。有机-无机界面的p-T相图是目前表面科学图谱上亟待填补的盲点。到目前为止，由于缺乏足够的实验设备，以及由于确定可能在实验可达区域内显示相变的系统的计算困难，这一努力无法进行。只有随着现代DFT方法和机器学习技术的出现，这样的系统现在才能被解决。该项目将由德国耶拿州立大学和奥地利格拉茨理工学院合作执行。这些实验将在托尔斯滕·弗里茨和罗曼·福克的团队中进行，他们是生长有机-无机界面和确定其结构的专家。理论研究将由奥利弗·T·霍夫曼(Oliver T.Hofmann)和埃格伯特·佐杰(Egbert Zojer)进行，霍夫曼是机器学习和DFT能带结构计算方面的专家，将协调该项目，而埃格伯特·佐杰(Egbert Zojer)在模拟有机半导体及其界面方面拥有丰富经验。

Phase diagrams contain crucial information for both fundamental and applied research, revealing under which pressure (p) and temperature (T) conditions specific phases form. However, for organic-inorganic interfaces, experimental phase diagrams (as function of T and p) are de facto non-existent. At the same time, the computation of phase diagrams, which relies on ab-initio thermodynamics, typically suffers from severe approximations, as vibrational and thermal effects are often neglected.The objective of this project is to develop strategies for obtaining p-T-phase diagrams for organic-inorganic interfaces. On the experimental side, this requires developing approaches to grow interfaces close to thermodynamic equilibrium under well-defined conditions. On the theoretical side, the role of thermal and vibrational effects on the relative energies of different polymorphs need to be assessed. We hypothesize that including these effects is indispensable for accurate predictions and devise a way to incorporate them with the aid of machine learning. To extract maximum insight from the phase diagrams, we also provide an in-depth analysis of which interactions stabilize specific phases.The interfaces will be grown near thermodynamic equilibrium in a novel vacuum chamber housing an almost closed cryoshield and using small molecules to facilitate the theoretical evaluation. The structures will be investigated in situ using distortion-corrected low-energy electron diffraction. The computational prediction and evaluation of phase diagrams a specialized structure search algorithm relying on a combination of density functional theory and machine learning in the form of Bayes Linear Regression.p-T-phase diagrams for organic-inorganic interfaces are presently a blind spot on the map of surface science that urgently needs to be filled. So far, this endeavor could not be undertaken, due to the lack of adequate experimental equipment and due to the computational difficulty to determine systems that are likely to show phase transitions in the experimentally accessible region. Only with the advent of modern DFT methods and machine-learning techniques, such systems can now be tackled.The project will be executed by a collaboration between the FSU Jena, Germany and the TU Graz, Austria. The experiments will be performed in the group of Torsten Fritz and Roman Forker, who are experts for growing organic-inorganic interfaces and the determination of their structures. The theoretical studies will be performed by Oliver T. Hofmann, who is an expert in machine learning and DFT band structure calculations and who will coordinate the project, and by Egbert Zojer, who has extensive experience in modelling organic semiconductors and their interfaces.