Description
With nanotechnology, miniaturization of appliances for fluidics has reached molecular dimensions. This leads to qualitatively novel, unsolved challenges to modeling and simulation. Classical approaches to the description of flow processes are based on the continuum assumption and fail on the nanoscale. Molecular dynamics (MD) simulation, however, permits a realistic description of these systems. In the past, the MD simulation method could only be applied to small systems with a few thousand particles. Consequently, a large gap between MD simulation results on one hand and, on the other hand, results from experiments as well as results based on continuum methods was present.
It is the objective of this subproject to close that gap. For this purpose, MD simulations are carried out for multiphase systems and flows of real liquids in nanoscopic geometries with real wall materials. By consistently employing massively parallel supercomputing facilities, extremely large numbers of particles are considered up to system sizes on the microcsopic scale. Thus, multiphase systems and their flow behavior are studied for characteristic lengths between a nanometer and a micrometer. From this approach, results with fundamental importance for many areas of nanotechnology can be expected, which can also serve as a basis for the development of hybrid models in Project A.2.
For the present subproject vapor-liquid phase transitions and interfaces are of particular importance. E.g., the distinction between the highly curved surface of a nanoscale droplet and a microscopic phase boundary which is quasi-planar on the molecular level can hardly be studied by experimental methods in a reliable fashion. However, massively parallel MD simulation of systems with a large number of particles permits the precise determination of the droplet formation energy for a wide range of droplet sizes as well as its influence on nucleation in supersaturated vapors [1-4]. By confining multiphase systems to a nanoscopic volume it is possible to stabilize otherwise unstable phase equilibria [4].
The required computational progress is realized in cooperation with Project D.1 [1]. For this purpose, the Tersoff multi-body potential was implemented as a model for real wall materials such as silicon, graphite and carbon nanotubes. Thus, two geometric scenarios are considered: planar flow between parallel walls, e.g. of graphite, and flow through nanotubes. This will provide the basis for higher-level descriptions of nanoscale flow phenomena that are directly suitable for practical applications.
[1] M. Bernreuther, C. Niethammer, M. Horsch, J. Vrabec, S. Deublein, H. Hasse & M. Buchholz: Innovative HPC methods and application to highly scalable molecular simulation. Innovatives Supercomputing in Deutschland 7(1): 50-53 (2009).
[2] M. Horsch, J. Vrabec, M. Bernreuther, S. Grottel, G. Reina, A. Wix, K. Schaber & H. Hasse: Homogeneous nucleation in supersaturated vapors of methane, ethane, and carbon dioxide predicted by brute force molecular dynamics. The Journal of Chemical Physics 128: 164510 (2008).
[3] J. Vrabec, M. Horsch & H. Hasse: Molecular dynamics based analysis of nucleation and surface energy of droplets in supersaturated vapors of methane and ethane. ASME Journal of Heat Transfer 131: 043202 (2009).
[4] M. Horsch, J. Vrabec & H. Hasse: A modification of the classical nucleation theory based on molecular simulation data for surface tension, critical nucleus size, and nucleation rate. Physical Review E 78: 011603 (2008).