Intramolecular Interactions

The geometry of the 2,6-diisopropylphenyl (Dipp) groups of the carbenes in 3 suggests an interaction of the isopropyl groups with the lone pairs of the central silicon atom. The isopropyl groups are not in plane with the phenyl ring but are bent by 5.7–6.9° pointing towards the silicon atom.

In order to investigate the intramolecular non-covalent interactions between the central silicon atom and the hydrogen atoms of the isopropyl groups in 3 the reduced density gradient (RDG) can be a useful tool. The RDG depicts the deviation of the ED from a homo-genous electron gas and thus usually assumes large values at points with low ED e.g. far away from atoms. However, in regions of non-covalent interaction, in which the ED is also very low, the RDG assumes low values.^[104-106] The RDG around Si1 was calculated using the program NCImilano^[104]. A plot of the RDG against the ED is shown in Figure 65a. Three regions with a low RDG can be identified. An assignment of the different interactions is possible by mapping the ED on an isosurface at a low level (see Figure 65b). The first area showing a low RDG and relatively high ED (≤ 0.225 a.u. / 1.5 e∙Å^-3) refers to the covalent interactions depicted in blue. The donor-acceptor bonds between the carbene carbon atoms and the central silicon atom (0.1 a.u. < ρ < 0.125 a.u. (green)) are clearly separated from these interactions. The resuming interactions showing an ED lower than 0.05 a.u.

(~0.34 e∙Å^-3) are not represented in the molecular graph. They indicate non-covalent inter-actions between the lone pairs of the silicon atom and the four hydrogen atoms (H20B, H23A, H43B, H46A).

(a) (b)

Figure 65: Plots of the RDG versus the ED in 3 (a). Isosurface of the RDG s = 0.05 (b). The surface is coloured according to ρ(r).

In order to distinguish between attractive and repulsive interaction it is common to map the sign(λ2)∙ρ onto a surface of the RDG at a reasonable level. For attractive interactions the second eigenvalue of the Hessian matrix λ2 is negative; for repulsive interactions the opposite is true. In Figure 66 the RGD on an isolevel of 0.3 a.u. is depicted. For the sake of

Topological Analysis of the EDD

clarity only points are shown where ρ(r) ≤ 0.05 a.u. The surface is coloured according to sign(λ2)∙ρ; green colour indicates attractive interactions, red repulsive. Additionally, the VSCCs at the silicon atom are indicated by the Laplacian at an isolevel of 2.5 e∙Å (blue). This representation of the intramolecular non-covalent interaction clearly reveals an attractive interaction between four hydrogen atoms of the isopropyl groups (H20B, H23A, H43B, H46A) and the lone pairs of the silicon atom. This explains the deviation of the Dipp-groups from the planar geometry.

Figure 66: Non-covalent intramolecular interaction around the silicon atoms in 3. Isosurface of the RDG s = 0.3. The surface is coloured according to sign(λ2)∙ρ; green attractive interactions, red repulsive. In blue an isosurface at a level of 0.25 e∙Å^-5 of the Laplacian is shown indicating the two lone pairs at the silicon atom.

Additionally, Figure 66 visualises effective shielding of the silylone by the Dipp groups. This is in good accordance with the reactivity found for the silylone 3. As reported by Roesky et al.^[232] the crystalline silylone (cAAC)2Si is stable in an inert atmosphere for about two years without any decomposition. Moreover, it does not react with molecular hydrogen, ammonia, and carbon dioxide unlike e.g. silylenes. To date the only reported reaction of (cAAC)2Si is the electron-induced conversion into a six-membered cyclic silylene (Figure 67).

Figure 67: Conversion of a silylone (cAAC)2Si into cyclic silylene via an intramolecular proton transfer.

Conclusion

5.6 Conclusion

The experimental charge density investigation presented here shows that the refinement of a MM against high-resolution X-ray data can lead to results of excellent quality even for complex molecules such as of the silylone (cAAC^cy)2Si. Although the low symmetry and the large number of atoms in the molecule lead to a very time consuming data collection and model refinement, the investigation of 3 could show that the experimental investigation of the EDD can give insights into the structure of a molecule, which are not always possible via quantum chemical calculations.

However, in order to obtain these results, it is essential to collect data of extraordinary quality. The investigation of 3 confirmed the importance of an accurate collection of the low-order data especially for the refinement of the subtle bonding features. A limiting factor here is the low dynamic range of the CCD detectors, which are often used in charge density investigations. As a result, the most intense low-order reflections very often exceed the counting rate of the detector. It could be proven in this investigation, that the use of an attenuator recollecting the frames, which exceed the dynamic range, is not useful as it intro-duces scaling problems. Therefore, the automatic attenuation should not be used for charge density data collections. Thus most attention should also be paid to the crystal selection, in order to find a crystal that allows the measurement of reflection up to high resolution and at the same time allows a measurement of the intense reflection within the counting rate of the detector.

A suitable alternative to circumvent these problems could be the collection of ‘fast’ scans, in which the scan interval per frame is increased or the reduction of the incident beam for complete runs. Future studies will be necessary to test the suitability of these strategies. In the present study the error arising from the bad scaling of the attenuated reflection could be minimised by rejection of 22 reflections affected by this error. The completeness of the data was not affected by this since 21 reflections were also collected without attenuation.

Similar to the refinements shown in Section 4 model 3 showed signs of resolution dependent errors. Positive residual density was found in close proximity to the silicon atom and the course of the scale factor mimics a u-shape. Parallel to what was found in Section 4 the model could be improved by a refinement with resolution-dependent scaling, by reduction of the size of the integration box or by the application of an empirical determined correction factor. This strongly suggests TDS to be the reason for the observed errors, although the temperature dependence of these errors could not be proven because only one dataset of 3 is available. The best model was obtained from the refinement against the empirical corrected dataset, which was used for the topological analysis.

Conclusion

The topological analysis of the EDD could clearly prove that the bonding situation in 3 is best described as a central silicon of formally oxidation state zero that is stabilised by to donor-acceptor bonds of the cAAC ligands. Thus 3 should be named silylone. The singlet state of the carbenes could be proven by the Laplacian distribution that resembles that of the ¹A1 CH2 molecule and that found for Fischer-type carbenes. Additionally, it was shown that the nitrogen atom in the cAAC tends to donate lone pair density into the N–Ccarbene bond, which results in higher s-character of the lone pair. Since this donation is only feasible for a singlet carbene, the bonding geometry of the nitrogen can be used as an indicator for the electronic state of the nitrogen. The properties of the Si–C bonds are in good accordance with those found for other donor bonds. They reveal an ED that is equally or even slightly lower than in normal Si–C single bonds, which is in strict contrast to what is found for covalent double bonds with triplet states, such as C=C double bonds. The ellipticities found for the two Si–C bonds indicate a different amount of π-backdonation, which is also mirrored in the bond length as well as the Bader charges of the carbene carbon atoms. Via periodic solid state calculations, it can be shown, that this is due to the different geometry of the two cAACs caused by weak intermolecular interactions. In order to obtain more information about these weak interactions and their role for the different back donation further quantitative studies of the silylones with cylcohexyl- and dimethyl groups at the Cα

carbon would be needed.

However, both bonds show a significant π-backdonation. Therefore, a double bond could be drawn in a Lewis diagram of 3, but it should always be kept in mind that the bond is not formed between two fragments in triplet state. The carbenes and the silicon atom both are in singlet state. Thus the interaction is better described by a donor-acceptor bond indicated by two arrows, as only this description also includes the lone pair density remaining at the silicon atom.

Unveiling Disorder in [Ge8{N(SiMe3)2}6]