Abstract
Resistive switching phenomenon is normally studied through fabricating two types of metal/insulator/metal (MIM) test structures: i) using a shadow mask to pattern dot-like top electrodes (on an insulator/metal sample) that also serve as pad to connect the probe station tips and ii) using photolithography to pattern metallic wires rotated 90 degrees that cross at one point to sandwich an insulator, and that are attached to large pads to connect the probe station tips. The second method is the most recommendable because the size of the MIM-like memristor can be reduced; however, many researchers prefer to use the first configuration because it is easier to fabricate. The pressure exerted by probe station tips on the top electrode of MIM-like memristors can alter the electrical characteristics of these devices, leading to lower dielectric breakdown (DB) voltages, higher post-DB currents, lower variability of the post-DB currents, erratic cycling, and lower yield is shown here. These observations have been confirmed statistically by measuring >900 devices made with different electrodes (Ni, Au, Cu, and Ag) and dielectric stacks (HfO2 and Al2O3). This indicates that the electrical characteristics reported in previous works using dot-like memristors may be unreal, as they could have been affected by the force exerted by the probe station tips.