An astounding fact about water, a small molecule, is that it remains liquid at room temperature. A new simulation technique, which originates from brain research, offers insights into the causes. A team of researchers from Bochum and Vienna have analyzed the atomic interactions of water molecules using artificial neural networks.
In light of their discoveries, they describe the melting temperature of ice and the density maximum at 4ºC which is based exclusively on computer simulations. The recently created technique is as exact as quantum mechanical calculations; however it is 100,000 times quicker. The teams of Prof. Dr. Christoph Dellago of the University of Vienna and PD Dr. Jörg Behler of the Ruhr-Universität Bochum explains the research in the journal "Procedures of the National Academy of Sciences" or PNAS.
Water has various properties that are difficult to understand merely based on its chemical composition. It reaches its maximum density at 4ºC, such that ice floats on liquid water. Furthermore, it is surprising that such a small molecule is in liquid state at room temperature and not gaseous. Hydrogen bonds play a major role in these phenomena. The analyses revealed that van der Waals interactions are conclusive for the adaptability and geometry of these hydrogen bonds. Through this, they estimate the properties of water, though they exert only very weak forces, weaker, for instance, than electrostatic interactions.
Method from brain research
Jörg Behler formulated the technique according to a process that was initially designed for brain research. The neural networks discover the forces between each atom as a function of their geometric alignment. "We can along these lines do PC reproductions that would not be conceivable with traditional quantum mechanical strategies, on the grounds that the computational exertion would be too high notwithstanding for a supercomputer", says the Head of an Independent Junior Research Group at the Bochum Chair for Theoretical Chemistry.
Dr. Tobias Morawietz applied the technique for the first time in his doctoral work to analyze the properties of water. The simulations were performed in relation to Bochum's Cluster of Excellence Resolv, in partnership with Andreas Singraber in the group of Christoph Dellago at the University of Vienna. Additionally, Tobias Morawietz conducted some of his simulations there. He is currently proceeding with his research in Vienna as a post-doctoral researcher.
The research received financial aid from the DFG (German Research Foundation) in relation to the Cluster of Excellence Resolv (EXC 1069) and other projects such as Heisenberg-Fellowship Be3264/6-1, Emmy-Noether-Project Be3264/3-1, and Project Be3264/5-1).
Funding also was provided by the Ruhr-University Research School Plus (DFG GSC 98/3), the Studienstiftung des Deutschen Volkes in Germany and Austria’s Österreichischer Wissenschaftsfonds FWF (P24681-N20, SFB Vicom, F41). The estimations were performed in parts at the Vienna Scientific Cluster.