Can a multipole analysis faithfully reproduce topological descriptors of a total charge density?

Abstract

Total charge densities &(r) of solid NH 3 have been derived using an ab initio crystalline molecular-orbital approach and also from multipole re®nement of the structure factors obtained from the same charge density. Comparison of the topological features of these charge densities, as de®ned by the quantum theory of atoms in molecules, has been used to probe the ability of the multipole analysis to reproduce exactly known total charge-density distributions. For the most part, multipole re®nement satisfactorily returns the features of the original density, although the ®t to theoretical data is not as good as that to the experimental data. The one topological parameter that is poorly reproduced is the Laplacian r 2 &r b at NH bond critical points.

On the transferability of QTAIMC descriptors derived from X-ray diffraction data and DFT calculations: substituted hydropyrimidine derivatives

“…Tsirelson et al (2006) also made an accurate joint low-temperature X-ray diffraction and nonempirical quantum chemical study of bonding in (3), and obtained extensive bond and atomic quantum-topological descriptors. A comparison of the QTAIMC results obtained for (3) from accurate X-ray diffraction measurements, threedimensional periodic calculations (CRYSTAL98; Saunders et al, 1998) and single-molecule DFT B3LYP/6-311G(d,p) calculations confirmed earlier findings (Chandler & Spackman, 1982;Parini et al, 1985;Iversen et al, 1997;de Vries et al, 2000;Volkov et al, 2000;Volkov & Coppens, 2001;Flaig et al, 2002;Bytheway et al, 2002;Tsirelson et al, 2006) that the main differences between the experimentally and theoretically obtained properties are observed for the longitudinal ('along the bond path') ED curvature, 3 . This may be attributed to a deficiency of the ED description with current multipole models.…”

“…The agreement of the Laplacian values at the BCPs from a single-molecule theoretical calculation and the experiment is significantly worse, as expected. This is mainly due to the above-mentioned deficiency of the current multipole models (Chandler & Spackman, 1982;Parini et al, 1985;Iversen et al, 1997;de Vries et al, 2000;Volkov et al, 2000;Volkov & Coppens, 2001;Bytheway et al, 2002;Tsirelson et al, 2006). The largest Laplacian differences have been observed for the polar double C O bonds in both the acetyl and carboxyethyl 5-substituents, and for the carbonyl group in position 2 of the heterocyclic ring; the maximum difference of 1.36 a.u.…”

On the error in the nucleus-centered multipolar expansion of molecular electron density and its topology: A direct-space computational study

“…(c) Our direct-space analysis corresponds to infinite Fourier-space and thus not affected by data-resolution or fitting issues. (d) The restricted flexibility of the HC-PA RDFs adapted in experimental studies has always been considered as the major limiting factor of the accuracy of X-ray charge densities, 46,47 and considerable efforts have been made to overcome this issue but usually within the single-Slater framework 48 to keep the number of parameters to be fitted manageable. The analyses of E-PA RDFs projected out of extended-basis QC-EDs however suggest that the structure of these functions is, as expected, more complex than that of the Slater functions (obviously, they cannot be represented by single-Slater functions).…”

Is there a future for topological analysis in experimental charge-density research?

“…After a rapid expansion of the field, they were ultimately criticized as such (Dunitz & Gavezzotti, 2005; 'descriptive but not predictive'), whereas other studies were criticized for unnecessary effort (Spackman, 1999) and avoidable inaccuracies (Spackman et al, 2007). Computational tools were frequently used uncritically with respect to the accuracy (Bytheway et al, 2002) they can provide. Adding to these points, I would criticize the lack of vision for exploring and expanding into new research questions rather than a lack of technical ability; 5 this becomes most obvious in purely descriptive research.…”