• chemometrics
• crystallization
• multivariate
• SMM
• spin crossover

Iron(II) and cobalt(II) bis-pyrazolilpyridyl (bpp-R) complexes, with general formula [M(bpp-R)2](X)2solv, M = Fe, Co; X = anion; solv = co-crystallized solvent, possess magnetic properties of interest. In particular, iron(II) derivatives can undergo spin transition from high spin (S = 2, HS) to low spin (S = 0, LS), showing the spin crossover (SCO) phenomenon, while cobalt(II) compounds can behave as single molecule magnets (SMMs) with consequent slow relaxation of the magnetization at low temperature. Both magnetic features are mainly influenced by the distortion of the octahedral coordination environment around the metal center. This is in turn governed by the crystal packing, which depends on the substituent R of the bpp ligands, the anion X and the possible co-crystallized solvent (solv). In this thesis all these co-acting factors were studied with an innovative multivariate approach with the help of chemometrics tools. Firstly, experimental design (DoE) was applied on the basis of the principal properties (from literature) of a batch of 38 anions to select the six most representative/diverse ones and perform a systematic experimental structural study on the influence of these anions. Indeed, only very few anions are usually employed in the crystallization of [M(bpp-R)2](X)2solv, namely ClO4, BF4 and PF6, with our approach the anion base can be exploited by a multivariate and rational approach. The bpp-COOMe was selected as representative ligand, and the crystallization reactions of [Co(bpp-COOMe)2](X)2solv complexes with the six selected anions SbF6, CH3COO, SCN, C(CN)3, N(CN)2 and CF3COO were then performed to map the structural effect and hence the SMM behavior. Secondly, a PLS regression model between the structural data, i.e. coordination bonds and angles, of the iron(II) complexes available in the literature, and the spin transition temperature T1/2 was obtained. It suitably highlighted the trend between the structural distortion of the iron(II) centers and the T1/2 by considering several structural parameters simultaneously. Structural data for iron(II) complexes with bpp-NO2 were found to be missing, so the ligand was experimentally obtained with a new synthetic path to confirm the PLS model results. These findings were then transposed to cobalt(II) derivatives, which are much less present in the literature, but the few ones known are isomorphous with the iron(II) ones, in order to obtain a prevision on the best compounds with maximized SMM behavior. Last but not least, a Quantitative Structure Property Relationships (QSPR) approach was also explored. A huge set of molecular descriptors were generated with the program AlvaDesc for a variety of metal(II) complexes of the bpp family, and the most informative were selected with the help of a sparse principal component like algorithm. This reduced set of molecular descriptors was then used to obtain a QSPR model, which in parallel to the one created with the structural data can be potentially implemented together with the structural parameters and the anion principal properties to generate a whole all-inclusive potent model suitable to show the synthetic directions to follow to obtain new complexes with maximized magnetic properties.