Molecular dynamics simulations of the structure and dynamics of confined polymer melts

We performed a series of molecular dynamics simulations investigating the static and dynamic properties of polymer melts confined between planar solid surfaces. The solid–melt interface was found to be very narrow (approximately two segment diameters) and independent of chain length. Inside the interface the segment density profile was oscillatory, the bond orientation altered between directions parallel and normal to the solid surface, and the chain ends accumulated very close to the wall (in the absence of strong wall–segment attraction). The oscillations of the segment density profile were weaker and were dampened faster than those of a simple fluid density profile next to the same solid surface. This reflected the reduced ability of sequences of connected segments (chains) to layer themselves against a solid surface because of restrictions on their configurations imposed by the chain connectivity requirement. This effect made the solid–melt interface even narrower than that of a simple fluid. Only the chain portions lying inside the interface had their shape affected by the wall. Chain statistical segments inside the interface assumed orientations parallel to the wall. In the absence of wall–segment attraction, the size of the statistical segments inside the interface was unaffected. This situation resulted in an apparent decrease of the radius of gyration normal to the wall an apparent increase of the radius of gyration parallel to the wall and spatial independence of the total radius of gyration. The wall effect was gradually diminished and chains assumed their bulk dimensions when their center-of-mass was so far from the solid surface that no portions of the chain could reach the interface (i.e., at a distance comparable to the bulk radius of gyration). The microscopic dynamics of chain portions inside the interface were strongly anisotropic. The mobility increased in the direction parallel to the wall and decreased normal to the wall. This fact was caused by the angular asymmetry of the segment–segment collisions inside the interface, i.e., by the same mechanism that induces the segment layering. The total mobility inside the neutral wall–melt interface was identical with that in the bulk reflecting the fact that the average segment density inside the interface had essentially the bulk value. The presence of strong wall–segment attraction increased the average interfacial density above the bulk value and lowered the mobility of the interfacial chain portions in all directions. The mean-square displacement of the chain center-of-mass during a certain time interval was affected by the solid only if the chain had a portion of itself inside the interface for a fraction of this time interval. The longest relaxation time of the chains, a property that cannnot be localized properly on a length scale smaller than the interfacial width, exhibited a weak and strongly diminishing with chain length spatial dependence.