To develop components of electrical devices at the scale of a few atoms for supercomputer, communications and health care technologies, a quantitative understanding of materials at the nanoscale scale is required. For materials relevant to MGI, we are developing statistical models to predict the contribution of topographic variations in a material to its measured near-field scanning probe microwave microscopy (NSSM) image.
These empirical statistical models are critically important for quantitative characterization of electronic and magnetic materials and their properties at the nanometer scale because NSSM images of materials are influenced by both topographic variations in the sample and material property variations in the sample. With such a prediction model, one can estimate the contribution to the observed NSSM image due solely to material property variations.
Correcting NSSM images of materials for topographic backgrounds is a major challenge. Based on the Atomic Force Microscopy (AFM) topography images, we predict the signature of topographic variations to any NSSM image. The difference between the observed NSSM image and this prediction is our estimate of the contribution due to material property variations in the sample. Our statistical approach has the advantage of being fully empirical compared to current background-correction schemes based on physical models that are often overly simplistic.
We developed a prediction model for a 2-d photovoltaic material. The predictors in the model are sample elevation and an estimate of sample edge strength. We determine the model parameters with a robust regression regression method. Ongoing work includes development of methods to select adjustable parameters in a refined version our approach, and comparison of different model calibration schemes. Currently we are developing a similar method for a GaN nanowire material.