Predicting Spontaneous Orientational Self-Assembly: In Silico Design of Materials with Quantum Mechanically Derived Force Fields

Abstract

De novo design of self-assembled materials hinges upon our ability to relate macroscopic properties to individual building blocks, thus characterizing in such supramolecular architectures a wide range of observables at varied time/length scales. This work demonstrates that quantum mechanical derived force fields (QMD-FFs) do satisfy this requisite and, most importantly, do so in a predictive manner. To this end, a specific FF, built solely based on the knowledge of the target molecular structure, is employed to reproduce the spontaneous transition to an ordered liquid crystal phase. The simulations deliver a multiscale portrait of such self-assembly processes, where conformational changes within the individual building blocks are intertwined with a 200 ns ensemble reorganization. The extensive characterization provided not only is in quantitative agreement with the experiment but also connects the time/length scales at which it was performed. Realizing QMD-FF predictive power and unmatched accuracy stands as an important leap forward for the bottom-up design of advanced supramolecular materials.

Accurate Quantum-Mechanically Derived Force-Fields through a Fragment-Based Approach: Balancing Specificity and Transferability in the Prediction of Self-Assembly in Soft Matter

“…At 360 K, no clear indication of a stable nematic phase is found along the 0.5 μs simulations, beside a small drift of the order parameter in the last 30 ns, which confirms the long period of P 2 oscillations found for the similar 5CB mesogen near the transition. 24 Figure 5a shows the behavior of some relevant observables averaged along the production runs at different temperatures, comparing them with the available experimental values. Both the system density ρ and the order parameter show a nice agreement with the experiment at all considered temperatures, with a slight and somewhat systematic overestimation of the former property.…”

“…Therefore, general-purpose FFs cannot meet a key design principle of most nanotechnology devices, that is, tweaking properties by changing one atom at a time, be it to engineer complex bio-mimetic nanoscale motion 36−38 or to guide a de novo design of programmed assembly of complex supramolecular structures. 24,33,39 A sound and elegant route to achieve such specificity is to resort to quantum mechanical (QM) calculations, so as to derive the FF parameters from a higher-level QM description. 40 In this regard, the QMD-FF strategy has flourished 22,23,41−44 as it provides a robust route to derive accurate FF parameters in cases where transferability fails and experimental training data sets are lacking or totally missing.…”

“…70 To address such a major T NI error, which makes the MD de novo design of LC effectively useless, a number of strategies have been proposed, aimed to refine selected sets of FF parameters. 32,65,68 In this framework, only very recently, it was shown 24 that a full, QM-based (i.e., solely from the knowledge of the chemical structure of the composing molecules) parameterization of all FF parameters is able to accurately predict T NI with an unprecedented accuracy (δT NI < 2% ≈5 K). However, the evolved nature of QMD-FF parameterization hampers its widespread use as previously described.…”

Complexes of Alkaline and Ammonium Cations with Dopamine and Eumelanin Precursors: Dissecting the Role of Noncovalent Cation−π and Cation–Lone Pair (σ-Type) Interactions

“…The aim of such detailed 2D maps is 2-fold. First, they can serve as reference to tune for instance quantum mechanically derived force fields, − which in turn can be employed in classical MD simulations, thus taking into account larger systems and other fundamental ingredients as solvent, other ionic species or embedding, to reliably describe supramolecular phenomena . Next, the wide exploration of the Δ E multidimensional surface, achieved through the analysis of multiple landscapes, allows for detecting additional interactions that might be overlooked in approaches considering few conformers or even in more limited scans.…”

“…First, they can serve as reference to tune for instance quantum mechanically derived force fields, 49−53 which in turn can be employed in classical MD simulations, thus taking into account larger systems and other fundamental ingredients as solvent, other ionic species or embedding, to reliably describe supramolecular phenomena. 54 Next, the wide exploration of the ΔE multidimensional surface, achieved through the analysis of multiple landscapes, allows for detecting additional interactions that might be overlooked in approaches considering few conformers or even in more limited scans. As a matter of fact, upon phenolic substitution onto the aromatic core, mp2 mod calculations revealed 45,46 the insurgence of a strong cation−lone pair (σ-type) interaction, between the ion and the OH groups in the plane of the aromatic ring, which for catechol becomes dominant on the most considered cation−π pattern, where the ion lies on top of the ring plane.…”