Ab initio Free Energy Calculations with Chemical Accuracy for Molecule-Surface Interactions

从原子学的角度理解多相催化以及纳米多孔材料中与能量相关的气体的存储和分离，需要对大周期体系的自由能进行量子力学从头算预测，并具有化学精度(4kJ/mol)。我们已经证明，这可以通过(I)使用高能级和低能级的混合计算，结合反应位置的精确波函数方法，在全(伪)周期系统上使用精度较低但计算效率较高的密度泛函理论计算，以及(Ii)以一种负担得起的方式计算非简谐振动能，以改进势能面的采样。到目前为止，还不存在这样的方法。建议(1)开发一种健壮且广泛适用的方法，该方法将免费提供给社会，冒着我们应用于计算非谐性的近似并不适用于所有类型的表面键合的风险，并且关键的技术步骤--简正模式的曲线表示--不能充分自动化，(2)建立分子-表面相互作用的数据库，其与实验的焓和/或熵相一致，因此，适合于基准近似方法，以及(3)使用该方法来获得选定应用中基本步骤的动力学和热力学数据，例如，沸石的催化作用以及金属有机骨架的气体储存和分离。

Atomistic understanding of heterogeneous catalysis as well as of storage and separation of energy-relevant gases in nanoporous materials requires the quantum mechanical ab initio prediction of free energies for large periodic systems with chemical accuracy (4 kJ/mol). We have shown that this can be achieved by (i) use of hybrid high level - low level calculations combining accurate wave-function methods for the reaction site with less accurate, but computationally more efficient density functional theory calculations on the full (pseudo) periodic system, and (ii) calculating anharmonic vibrational energies in an affordable way to improve the sampling of the potential energy surface. Such a method did not exist so far. It is proposed (1) to develop a robust and broadly applicable methodology that will be freely available to the community, taking the risk that the approximations we apply to calculate anharmonicities are not valid for all types of surface bonding and that a crucial technical step - curvilinear representation of normal modes - cannot be made sufficiently automatic, (2) to build a data base of molecule - surface interactions for which agreement is achieved with experimental enthalpies and/or entropies and which are, thus, suitable for benchmarking approximate methods, and (3) to use this methodology to obtain kinetic and thermodynamic data for elementary steps in selected applications, e.g., catalysis by zeolites as well as gas storage and separation by metal-organic frameworks.