Comment on “Accurate relativistic effective potentials for the sixth-row main group elements ” [J. Chem. Phys. 107, 9975

Comment on " Accurate relativistic effective potentials for the sixth-row main group elements " [J. Abstract Critical remarks in respect to the Generalized Relativistic Effective Core Potential in the recent article of S. A. Wildman at al. are discussed and shown to be incorrect.'s article [1] deals with a problem of the Relativistic Effective Core Potential (RECP) generation for precise calculations. In this article, new RECPs for the Tl, Pb, Bi, At, and Rn atoms which are more accurate than the previous RECPs [2] are constructed and tested in molecular calculations. However, some statements made in the article are incorrect. As it follows from article [1], the problems with the old RECPs [2] ". .. are due to the inappropriate partitioning of the spinors used to generate the f 5/2 and f 7/2 potentials " that, in its turn, is connected with the restriction on the number (n ≤ 2) of inflections for pseu-dospinors in the RECP generation procedure developed in [3]. However, this does not mean that " Given the comparable approaches in terms of orbital/spinor partitioning, one would expect to see very similar behavior and problems with the potentials developed by other groups 16. .. " (i.e., Ref. [4] here). Generalized RECP (GRECP) generation procedure [4] is based on procedure [3] but has essential distinctions. Some of them are requirement of a smooth shape for the potential and absence of limitations on the number of inflections for pseudospinors [4]. We did not concentrate on such technical details in our papers, because they are not so important as the principal questions concerning the optimal form of the RECP operator. One can see from comparison of Fig. 1 in this comment with Fig. 1(B) in Ref. [1], however, that the new 5f 5/2 pseudospinor of S. A. Wildman et al., in contrast to the old one, is rather close to the 5f 5/2 pseudospinor constructed within the GRECP generation procedure [4]. Some comments should be made on the authors' statement that in paper [4] ". .. it has occasionally been suggested that the small components are neglected in this procedure. .. " (i.e. in the GRECP generation procedure). More correctly (that is evident from our papers, see e.g. [5]), we neglect contribution from the small components in the valence region to the two-electron integrals and 1