Matching atomistic simulations and in situ experiments to investigate the mechanics of nanoscale contact

Many emerging devices and technologies rely on contacts between nanoscale bodies. Recent analytical theories, experiments, and simulations of nanocontacts have made conflicting predictions about the mechanical response as these contacts are loaded and separated. The present investigation combined in situ transmission electron microscopy (TEM) and molecular dynamics (MD) simulation to study the contact between a flat diamond indenter and a nanoscale silicon tip. The TEM was used to pre-characterize the materials, such that an atomistic model tip could be created with identically matched materials, geometry, crystallographic orientation, loading conditions, and degree of amorphization. A large work of adhesion was measured in the experiment and attributed to unpassivated surfaces and a large compressive stress applied before separation, resulting in covalent bonding across the interface. The simulations modeled atomic interactions across the interface using a Buckingham potential in order to reproduce the experimental work of adhesion without explicitly modeling covalent bonds, thereby enabling larger time- and length-scale simulations than would be achievable with a reactive potential. Then, the experimental and simulation tips were loaded under similar conditions with real-time measurement of contact area and deformation, yielding three primary findings. First, the results demonstrated that significant variation in the value of contact area can be obtained from simulations, depending on the technique used to determine it. Therefore, care is required in comparing measured values of contact area between simulations and experiments. Second, the contact area and deformation demonstrated significant hysteresis, with larger values measured upon unloading as compared to loading. Therefore, continuum predictions, in the form of a Maugis-Dugdale contact model, could not be fit to full loading/unloading curves. Third, the load-dependent contact area could be accurately fit by allowing the work of adhesion in the continuum model to increase with applied force from 1.3 to 4.3 J/m^2. The most common mechanisms for hysteretic behavior—which are viscoelasticity, capillary interactions, and plasticity—can be ruled out using the TEM and atomistic characterization. Stress-dependent formation of covalent bonds is suggested as a physical mechanism to describe these findings, which is qualitatively consistent with trends in the areal density of in-contact atoms as measured in the simulation. The implications of these results for real-world nanoscale contacts are that significant hysteresis may cause significant and unexpected deviations in contact size, even for nominally elastic contact