How small a molecular crystal can be? Looking for Hidden Crystalline Aggregates in Dissolving Benzoic Acid Nanoparticles

Answering the title question would provide valuable information on the behaviour of matter at the molecular scale. The problem is far from trivial: while crystals can be experimentally detected through their physico-chemical properties, such as coherent X-ray scattering, nanocrystals emerging from their liquid matrix are too small to be detected unequivocally by standard crystallographic techniques. We face the problem by means of Molecular Dynamics (MD), as available in the Milano Chemistry Molecular Simulation (MiCMoS) package [1,2]. We exploit a “top-down” approach by investigating the dissolution of a small (~5 nm, 576 molecules) benzoic acid (BZA) nanoparticle embedded in a ~14 nm large nanodroplet of its own liquid at 300 K. By operating in high degree of undercooling (ΔT ≈ 100 K) the goal was to slow down the dissolution kinetics enough to detect any persistent supramolecular aggregate that could retain some degree of translational symmetry. Findings [3] indicate that the BZA nanoparticle invariably melts, as confirmed by the global loss of both rotational correlation and RDFs features compared to the experimental crystal structure. However, we developed a correlation-based criterion that allowed us to detect some metastable supramolecular aggregates that persist for more than 30 ps, which share some packing features with crystalline BZA, despite being fluxional and partly liquid structures. The similarities of these alleged “crystal embryos” to the bulk structure were assessed through geometrical analysis on intervals of the trajectory where local equilibrium conditions were roughly fulfilled. If partially symmetric aggregates have detectable cohesion while persisting for enough time, it is reasonable that they could represent true metastable intermediates for the inverse process – that is, when the crystal grows. Analysis of the molten region of the nanoparticle further indicates that the system is not fully liquid yet, as the dominant aggregation mode remains identical to only one found in the crystalline state. We also show that Bragg peaks could be recognizable in principle from partially crystalline structures down to ~3 nm, providing some suggestions on how an experiment should be designed to probe these elusive structures at their minimum length scale. [1] A. Gavezzotti, L. Lo Presti, and S. Rizzato, CrystEngComm, 24, 5, 922–930, 2022 [2] A. Gavezzotti, L. Lo Presti, MiCMoS 2.3 Users’ Manual https://sites.unimi.it/xtal_chem_group/index.php/research/5-micmos [3] M. Vacchini, L. Sironi, G. Macetti and L. Lo Presti, in preparation.