The ongoing miniaturization beyond the fundamental limitations of nowadays information technologies is essential to keep up recent years’ development rate in the performance of our devices. Since the unit sizes of the latest generation integrated circuits approach 10 nm, the next breakthrough must challenge the atomic length-scales. In the recent decade several devices were demonstrated where the active volume approaches this ultimate size limit mostly relying on electrochemical metallization cells, where the voltage driving with proper polarity and amplitude induces the formation or dissolution of a metallic nanofilament within a solid electrolyte matrix. Our research group was also active in the investigation of silver-sulfide based resistive switching memory junctions. Along these studies we have found that at the ultimate single-atom dimension the embedding matrix is not necessary, and solely pure Ag single atom contacts can exhibit reproducible resistive switching due to the current induced rearrangement of the central atoms [Geresdi et al, Nanoscale 3, 1504 (2011)]. Later a research group in Konstanz has demonstrated the trainable, extremely stable memory operation of a similar current driven atomic switch using nanofabricated aluminum break junction devices [Schirm et al, Nature Nanotechnology. 8, 645, 2013]. In the latter study superconducting subgap spectroscopy was also applied to confirm the single atom diameter of the junction. In this topic, however, only limited proof of principle experimental results are available using a few different metals, and the detailed electrical characterization of the switching properties is lacking. The PhD work targets the detailed study of current induced single-atom resistive switching in a broad range of different metals using notched-wire break junction devices. A further goal is the thorough study of the switching characteristics, like the investigation of the voltage dependence of the switching speed,the study of the noise characteristics, and the search for multilevel programming possibilities and further neuromorphic functionalities.