The development of neuromorphic computing architectures based on two terminal filamentary resistance switching devices is limited in part by the high degree of variability in resistance states and switching voltages. Because of the large role filament shape plays in directing thermal and electric fields around the filament (and thus switching parameters), unambiguous knowledge of filament morphology resulting from direct characterization of filament shape is essential to solve critical ongoing challenges of device switching variability. Here, we have utilized a conductive atomic force microscopy scalpel technique to simultaneously scribe through a polycrystalline dielectric layer in formed Cu/HfO/p+Si electrochemical metallization cell devices.
View Article and Find Full Text PDF