Computational Study of the Mechanisms that Stabilize Organic Molecule‐Based Magnets

Abstract

The objective of this PhD thesis has been the study of the mechanisms that stabilize high-spin states in organic molecule-based magnetic materials. These materials require organic radicals with permanent magnetic moment as building units. Additionally, these molecules need to interact ferromagnetically, and expand that interaction along all three directions of the space.

We have carried out our studies using computational chemistry techniques, mostly ab- initio methods (MP2, CASSCF and CASMP2), and DFT-based methods (B3LYP or M06L). Besides, we have also used the hybrid Molecular Mechanics Valence Bond method (MMVB) for alternant hydrocarbons with high number of active electrons.

We have demonstrated that in organic radicals the energy gap between two spin states is usually higher when the stabilization of the spin centers occurs by means of through- bond (TB) instead of through-space (TS) interactions. As a result, alternant hydrocarbons (π-delocalized, with TB interactions) are more stable than non-alternant hydrocarbons (π-localized, with TS interactions). Therefore, alternant hydrocarbons would be preferable in the design of permanent molecular magnets.

Polymerization of high-spin radicals leads to high-spin systems. However, our research showed that the gap of energy between the first and second spin states decreases with the number of units bonded when these are alternant hydrocarbons. On the other hand, it has been proved that, when the synthesis of macro-radicals follows the SU-CU- SU methodology (SU=spin-containing unit; CU=coupling unit), the SU and CU units keep their multiplicity once coupled. In that case, the energy gap between the spin states of the system can be described considering the energy gap of the spin states of the constitutive units.

McConnell-I theory is widely applied to describe ferromagnetic intermolecular interactions. Our research has revised systematically this approach. We have explored the existence of a magneto-structural relationship using pairs of well-known radicals (H2NO·, ·CH3 and ·C3H5) at different geometrical orientations. We demonstrated that McConnell-I model predicts correctly the spin preference of the ground state when the interacting spin-containing radical centers are placed in parallel planes and there are mainly TS interactions between them. However, in other cases, the prediction of the spin preference becomes very complex, and more detailed quantum calculations are required. Overall, we have demonstrated that this model must be used carefully when predicting the multiplicity of the through-space interaction between two radicals. Further, we evaluated whether McConnell-I theory could be applied to assess the magnetic character of real crystals on the subset of experimentally FM crystals of the α- nitronyl nitroxide (α-NN) family. We analyzed the closest contacts between two intermolecular ONCNO groups (atoms where the spin densities are mainly located) for

each chosen crystal. We concluded that the ONCNO interactions do not describe entirely the observed macroscopic magnetic property for all the systems. Consequently, TS interactions not considered in the simplistic ONCNO model must play an important role defining the magnetic character. Secondly, we proved there is not a simple magneto-structural relationship, such as the one suggested in McConnell-I model, that can be applied to all through-space interactions in the crystals. This conclusion was reached after a twofold statistical analysis (namely, factor and cluster analyses) of the geometrical parameters as a function of the calculated energy gap ΔES-T.

Charge-transfer salts are successful examples of molecular magnets. However, the formation of diamagnetic dimers of the donor species, [D]22+, or the acceptor species, [A]22-, causes the loss of the magnetic properties. We studied the causes of this dimerization studying the formation of TCNE dimers, [TCNE]22-], as a prototypical example of an organic acceptor. The Eint of two charged molecules has two components: the Coulomb contribution (Ecoul > 0, for molecules with the same charge) and the bonding energy (Ebond < 0). If the repulsion energy is higher than the bonding energy in absolute value (|Ecoul|>|Ebond|) the two molecules will repel and the formation of the dimer will not be stable (Eint > 0). However, if there is any force that counterbalances the repulsion between the two charged molecules, the bonding energy could overcome the repulsion energy in absolute value (|Ecoul|<|Ebond|), and the metastable minima would become stable (Eint < 0). The calculations performed described three metastable minima that agree with those observed experimentally. Besides, the spectroscopic features of each class of these three dimers have been calculated and are in agreement with the

available experimental data. Extended calculations performed in the presence of cations or polar solvents resulted in the stabilization of the dimers, which demonstrates that counterbalance of the repulsive energy is needed for the formation of these long multicenter bonds. The two electrons - four centers (2e-/4c) bond described is unique since it involves 2e- and takes place among four carbon atoms chemically equivalent.

El objetivo de esta tesis ha sido estudiar computacionalmente las bases teóricas del magnetismo molecular para poder utilizar el conocimiento adquirido en el diseño de materiales magnéticos moleculares. Hemos analizado los mecanismos a través del enlace (TB: through-bond) y a través del espacio (TS: through-space) que estabilizan moléculas de alto spin (radicales) y sus interaccionan intermoleculares ferromagnéticas. Para llevar a cabo dichos estudios se han utilizado métodos híbridos como el Molecular Mechanics Valence Bond (MMVB), métodos DFT como el B3LYP y métodos ab-initio como MP2, CASSCF, y CASMP2.

Así pues, por un lado, se ha estudiado la estabilidad de moléculas orgánicas de alto spin y su posible polimerización manteniendo su alta multiplicidad de spin. Se ha llegado a la conclusión que el mecanismo TS es de menor coste energético que el TB. Por lo tanto, los radicales cuyos centros de spin se estabilizan a través del enlace TB son más estables. Asimismo, compuestos que presentan ambos mecanismos, los estados de spin de los estados fundamental y primer excitado vendrán determinados por el mecanismo TS.

Por otro lado, se estudiaron las interacciones intermoleculares entre radicales, con el objetivo de establecer las condiciones que favorecen las que son ferromagnéticas. En este contexto, se evaluó la teoría denominada McConnell-I. Tras metódicos estudios de la interacción entre dos radicales (H2NO·, ·CH3 y ·C2H6) en diferentes orientaciones en el espacio, se concluyó que el ámbito de aplicación de esta teoría está limitado a cuando los centros de spin interaccionan en planos paralelos y existe una interacción TS predominante. Estudios adicionales en cristales de la familia α-nitronil nitróxido demostraron que la teoría de McConnell-I no se puede aplicar de forma general a cualquier interacción intermolecular entre radicales. Se observó que esta teoría no predice correctamente el comportamiento magnético de cristales cuando se analiza sólo la interacción entre los átomos que contienen mayoritariamente la densidad de spin (ONCNO). Así pues, el estudio se debe ampliar a otros contactos entre las moléculas para poder describir correctamente el comportamiento magnético observado.

Finalmente hemos establecido que, en sales de transferencia de carga, se dan casos de dimerización de las especies constituyentes, por ejemplo tetracianoetileno (TCNE), cuando la repulsión entre especies de la misma carga se minimiza por la presencia de contra-iones o disolventes polares. De esta manera, se favorece la formación del enlace en el dímero al permitir la interacción de los electrones desapareados.