Conducting atomic force microscopy for nanoscale tunnel barrier characterization

Increasing demands on nanometer-scale properties of oxide tunnel barriers necessitate a consistent means to assess them on these length scales. Conducting atomic force microscopy (CAFM) is a promising technique both for understanding connections between nanoscale tunnel barrier characteristics and macroscopic device performance as well as for rapid qualitative evaluation of new fabrication methods and materials. Here we report CAFM characterization of aluminum oxide (AlOx) barriers to be used in Josephson-junction qubits, with a particular emphasis on developing reproducible imaging conditions and appropriate interpretation. We find that control of the imaging force is a critical factor for reproducibility. We imaged the same sample on the same day with the same cantilever varying only the imaging force between scans. Statistical properties compiled from the resulting current maps varied approximately exponentially with imaging force, with typical currents increasing by two orders of magnitude for only a factor of 5 increase in imaging force. Given appropriate control of the imaging force, scan to scan variation of the current recorded at the same location was approximately ±0.5, which establishes a criterion for statistical reproducibility of CAFM measurements. We further find that the appropriate interpretation for CAFM (under most imaging conditions), is as a probe of local propensity for insulator breakdown. Samples stored in air for weeks before study showed current features with oxidation times of order minutes. This indicates that these features were created by the scanning of the tip, and thus represent local pinhole susceptible regions. We finally present results for several AlOx samples showing that under appropriate imaging conditions significant sample to sample variation is observed, thus demonstrating the potential of this technique to qualitatively assess and facilitate under standing of potential qubit tunnel barrier devices.