2 EXPERIMENTAL ELECTRON DENSITY STUDIES – THE BASICS
2.5 Functions and Moments Derived from Electron Density Distributions
Once a charge density distribution has been obtained experimentally, various chemical and physical properties can be derived. These properties directly depend on the EDD.
The properties used in this thesis will be presented in the following two subchapters.
2.5.1 Static Deformation Density
A direct inspection of the modeled density ρ(r) itself is not very meaningful in almost all cases because the density is dominated by the core electrons and the effects of bonding are only slightly visible. Therefore, difference densities are widely applied to amplify the features of bonding. A commonly used function is the static deformation density ∇ρstatic(r), which is defined as the difference between the thermally averaged density from the multipole model ρMM(r) and the spherically averaged density from the independent atom model ρIAM(r).
Δρstatic(r)=ρMM(r)−ρIAM(r) (Eq. 2‐18)
This deformation density is based on the functions and populations from the aspherical multipole refinement and does therefore not include the effect of thermal smearing.
In a deformation density map, accumulations of density are visible in bonding as well as in the lone‐pair regions. This is expected, as these features are only described within the MM and not within the IAM. Thus, deformation density maps can be used to confirm long‐used chemical concepts. Additionally, these maps have great diagnostic potential and are routinely used to check the quality of an analysis by a comparison of the static deformation densities from X‐ray data with that calculated theoretically. By comparing experimental densities with those from periodic theoretical calculations, shortcomings in either method become apparent. For example, expected features cannot always be seen. Elements with more than half‐
filled valence shells lack bonding features in the deformation densities due to the
neutral spherically averaged reference atom which contains more than one electron in each orbital of the valence shell.[144,145]
The static deformation density is, in contrast to FOURIER densities, not limited to the finite resolution of the experimental data set. This leads to a high dependence on the basis set of functions applied in the refinement, and thus introduces bias. To reduce this bias, special care has to be taken for the quality of the reference molecule.
2.5.2 Electrostatic Potential
Nucleophilic and electrophilic regions in a molecule represent possible reaction sites for electrophiles or nucleophiles, respectively. As the electrostatic potential (ESP) provides information on their spatial arrangement in a molecule, its determination is of particular chemical interest.
The ESP at a given point in space is defined as the energy required for bringing a positive unit of charge from infinite distance to this point. It can be calculated, independent from the crystal environment, applying the formalism of SU and COPPENS.[146] For an atom, composed from a positive charge at a certain point, the nucleus, and a continuous distribution of negative electronic charge around this point,
Here, Z is the charge of the nucleus j located at R. The first term of the equation describes the nuclei contributions, from which the second part, the electron distribution term, is subtracted.
As electrostatic forces are relatively long‐range forces, they determine the path a reactant uses to reach the reactive parts of a molecule. Hence, in chemical terms, nucleophilic reagents are attracted to regions with positive potential while electrophiles approach the negative.
2.5.3 d‐orbital Populations
The radial functions applied within the multipole refinement resemble the radial functions of the orbitals. However, they are not equivalent. Therefore, the multipole populations do in general not directly correspond to the orbital populations in a given system. Nevertheless, for transition metals, there is a possibility to calculate the d‐orbital population from the least‐squares multipole population coefficients.[147]
This relation is based on the assumption, that the d‐orbitals can be represented by single SLATER type orbitals and that the overlap between metal atom and ligand orbitals is small. Therefore, the multipolar density at the transition metal atom can – to a good approximation – be calculated from the population of the outer valence shells of the atom. The relationship between d‐orbital occupancies and multipole population parameters is derived from the equivalence of two alternative descriptions of the atomic electron density.
with φ(di) being the atomic d‐orbital basis set. In the multipolar expression, the 3d‐density is expressed in terms of the density‐normalized spherical harmonic functions dlmp as
∑ ∑ ∑
Assuming, that the radial dependence of the density is equal in both descriptions, equating Eq. 2–20 with 2–21 leads to
PijYij =Plmpdlmp (Eq. 2‐22)
Here, Yij is the 15‐element column vector of the angular part of the φ(di)φ(dj) orbital products, Pij is the row vector of the 15 unique elements of the symmetric 5 x 5 matrix of the coefficients of the 15 spherical harmonic density functions dlmp with l = 0, 2, or 4. Density functions with other than even l values do not contribute to the d‐orbital density.
The spherical harmonic functions constitute a complete set of functions in the spherical point group. A product of two harmonics must therefore be a linear combination of spherical harmonic functions that can in general be written as
Yij = Ldlmp. The elements of the matrix L are the coefficients of each density function. A
complete set of the equations for l ≤ 2 is tabulated in the literature.[129]
The equivalence for the density in Eq. 2–22 can also be written as
PijYij =PijLdlmp =Plmpdlmp (Eq. 2‐23)
Therefore, the relation between the coefficients Pij and Plmp can be formulated as
Plmp=PijL (Eq. 2‐24)
or in a different way be defined as
PlmpT =LTPijT ≡MPijT (Eq. 2‐25)
The d‐orbital occupancies are derived from the experimental multipole populations by the inverse expression[147]
PijT =M−1PlmpT (Eq. 2‐26)
The matrix M–1 is a 15 x 15 matrix and also tabulated in the literature.[129] In all but triclinic point groups, site‐symmetry restrictions limit the allowed functions.