Reply to ‘comment on “extending Hirshfeld‐I to bulk and periodic materials”’

Abstract

The issues raised in the comment by T.A. Manz are addressed through the presentation of calculated atomic charges for NaF, NaCl, MgO, SrTiO 3 and La 2 Ce 2 O 7 , using our previously presented method for calculating Hirshfeld-I charges in Solids [J. Comput. Chem.. doi: 10.1002/jcc.23088]. It is shown that the use of pseudo-valence charges is sufficient to retrieve the full all-electron Hirshfeld-I charges to good accuracy. Furthermore, we present timing results of different systems, containing up to over 200 atoms, underlining the relatively low cost for large systems. A number of theoretical issues is formulated, pointing out mainly that care must be taken when deriving new atoms in molecules methods based on "expectations" for atomic charges. PACS numbers:Our recent extension of the Hirshfeld-I method to solids and periodic systems 1 , allows for the calculation of atoms in molecules (AIM) in solid state codes using common techniques as pseudo-potentials and plane wave basis sets. As a Hirshfeld-type method, 2 it relies on atomic reference densities that are used to construct AIM weight functions w(r) that allow to extract an AIM density function from a computed system electron density distribution (EDD). As was shown by us, when using pseudo-potentials and plane waves, some issues appear that require attention when generating atomic reference densities.1 However, we showed that the delocalization problem can be handled, and that it is actually part of the larger conceptual problem of defining reference densities for anions. In addition, we showed that the use of pseudo-potentials, and their resulting pseudo-valence charges, can provide the all-electron values for the atomic charges. Finally, we showed that the method scales favorably for large systems.In a comment to our paper, 3 T. A. Manz raises some questions which we address here in detail. His criticisms and suggestions can be summarized as:1. Hirshfeld-I atomic charges do not give "chemically feasible" atomic charges and improved performance of Atoms in Molecules (AIM) methods may be obtained by combining spherical averaging and Hirshfeld-I methods. 3. The presented computational scaling was insufficiently justified.In the following paragraphs we address each of these comments, from a theoretical as well as a computational perspective.

Hirshfeld-I charges in linear combination of atomic orbitals periodic calculations

“…According to the comment provided in Sec. 2.2, these are in quite good agreement with those reported by Vanpoucke et al, [43] even considering a radically different implementation. This confirms that HI charges for this periodic system are similar to QTAIM, which differs significantly from Manz and Sholl [27,42] findings.…”

“…This result prompted Manz to suggest that HI, at least with the anionic reference densities, overestimates net atomic charges. [42] Vanpoucke et al implementation, on the other hand, shows a lost of only 1.6|e| at the LDA/PAW level [43] which is closer to value obtained with QTAIM method. In view of such a controversy, we here include results on the same system computed with the present implementation.…”

“…Its particularity with respect to other periodic implementations [11,26,42,43] is that it is suitable for both HartreeFock and Kohn-Sham wave-functions spanned in terms of a basis set of AOs. The method was implemented in the Crystal code and its performance is here illustrated by several examples.…”

A Flexible Photoactive Titanium Metal–Organic Framework Based on a [TiIV3(μ3‐O)(O)2(COO)6] Cluster

“….H irshfeld-I charges show the Ti atoms in as ingle cluster to be differentiated as two Ti IV atoms and one Ti III atom, with charges of 2.87 and 2.49 e, respectively, [54,55] in accordance with the EPR observation of aT i III species.The calculated electronic structure of COK-69 3OH (Figure 3) shows as emiconducting band gap upon reduction, comparable to that of Si in size,inagreement with the experimental absorption spectrum.…”

Quasi-1D physics in metal-organic frameworks: MIL-47(V) from first principles

“…The atomic charges in the systems are calculated by using the Hirshfeld-I approach [67–68] as implemented in our in-house-developed code HIVE [69–71]. The atom-centered spherical integrations [72] are done by using Lebedev–Laikov grids [73] of 1202 grid points per shell, and a logarithmic radial grid.…”