Department of Materials and Interfaces, Weizmann Institute of Science , Rehovoth 76100, Israel.
Acc Chem Res. 2014 Nov 18;47(11):3208-16. doi: 10.1021/ar500144s. Epub 2014 Jun 5.
Molecular crystals are ubiquitous in many areas of science and engineering, including biology and medicine. Until recently, our ability to understand and predict their structure and properties using density functional theory was severely limited by the lack of approximate exchange-correlation functionals able to achieve sufficient accuracy. Here we show that there are many cases where the simple, minimally empirical pairwise correction scheme of Tkatchenko and Scheffler provides a useful prediction of the structure and properties of molecular crystals. After a brief introduction of the approach, we demonstrate its strength through some examples taken from our recent work. First, we show the accuracy of the approach using benchmark data sets of molecular complexes. Then we show its efficacy for structural determination using the hemozoin crystal, a challenging system possessing a wide range of strong and weak binding scenarios. Next, we show that it is equally useful for response properties by considering the elastic constants exhibited by the supramolecular diphenylalanine peptide solid and the infrared signature of water libration movements in brushite. Throughout, we emphasize lessons learned not only for the methodology but also for the chemistry and physics of the crystals in question. We further show that in many other scenarios where the simple pairwise correction scheme is not sufficiently accurate, one can go beyond it by employing a computationally inexpensive many-body dispersive approach that results in useful, quantitative accuracy, even in the presence of significant screening and/or multibody contributions to the dispersive energy. We explain the principles of the many-body approach and demonstrate its accuracy for benchmark data sets of small and large molecular complexes and molecular solids.
分子晶体在科学和工程的许多领域都很普遍，包括生物学和医学。直到最近，我们使用密度泛函理论理解和预测它们的结构和性质的能力还受到缺乏能够达到足够精度的近似交换相关泛函的严重限制。在这里，我们表明，在许多情况下，Tkatchenko 和 Scheffler 的简单、最小经验的对校正方案可以对分子晶体的结构和性质提供有用的预测。在简要介绍该方法之后，我们通过我们最近的工作中的一些示例来展示其优势。首先，我们使用分子复合物的基准数据集展示该方法的准确性。然后，我们通过血红蛋白晶体展示其在结构确定方面的功效，血红蛋白晶体是一个具有广泛的强和弱结合情况的具有挑战性的系统。接下来，我们通过考虑超分子二苯丙氨酸肽固体的弹性常数和水在 brushite 中的自由振动的红外特征，表明它对于响应特性同样有用。在整个过程中，我们不仅强调了该方法的经验教训，还强调了所研究晶体的化学和物理方面的经验教训。我们进一步表明，在许多其他情况下，简单的对校正方案不够准确的情况下，可以通过采用计算成本低廉的多体色散方法来超越它，从而获得有用的、定量的准确性，即使在存在显著的屏蔽和/或多体对色散能的贡献的情况下也是如此。我们解释了多体方法的原理，并展示了它在小分子和大分子复合物以及分子固体的基准数据集上的准确性。
