Cohesive Modelling for Spontaneous Adhesion in Microelectromechanical Systems

The reliability of microelectromechanical systems (MEMS), namely microengines and miniaturized sensors, is often limited by phenomena of spontaneous adhesion between parts which should maintain the capability of relative motion. The issues of catastrophic adhesion, but also of dangerous increments of friction, are often referred to in the literature with the term “stiction”, contraction of “static friction”. Many research efforts are currently devoted to stiction phenomena, both from the experimental point of view and in the ambit of predictive simulations, and to the study of possible provisions apt to reduce their effects on MEMS operation (see the review papers [1] and [2]). The main source of adhesion, in the presence of humidity, is represented by the capillary tension in water droplets condensed around the asperities of the contacting surfaces. In dry conditions, the uncharged surfaces stick to each other because of different forces, among which the most important ones are represented by the shortrange intermolecular Van der Waals interactions [3]. This communication aims at the formulation of a computational model capable to simulate the adhesion in various situations. The model is based on the concepts of fracture mechanics, which are often used to relate the interface adhesion energy to the geometric configuration of the system, see e.g. [4]. Hence, when considering the finite element model, the adhesive behaviour is simulated by introducing some cohesive interface elements on the contacting surfaces. The fracture energy connected to the chosen interface model must equal the surface energy, which in turn can be experimentally measured in different environmental conditions. The cohesive law must be able to interpret the basic features of adhesion, which is basically controlled by what happens at the microscale. For this reason, a sort of multiscale procedure has been devised, in the sense that the macroscale interface model is tuned on the basis of detailed computations carried out at the microscale, on a representative part of the surface. In this way, it is possible to catch some essential aspects of stiction, most of which are related to the microscopic roughness. The geometric model of the rough surface can be either artificially generated on the