Machine learning algorithms in combination with novel experimental data can accelerate the process of materials design and discovery for a wide range of applications. Using machine learning algorithms on experimental data one can discover new materials with desired properties, get ‘materials recipes’ for synthesizing materials, and build tools to automatically analyze new measurements to retrieve insights.
Combine data on materials structure (such as optical data, band gaps, layer thickness, etc) with function data (such as resistivity or performance) to model how materials structure affects functionality. Our database includes different functionalities (resistivity, Voc, Jsc, etc.), and data on materials structure can be used from our experimental database and/or calculated from publicly available databases such as materials proje
Use the large amount measurement data in our database (such as XRD, raman, optical) to develop machine learning based analysis tools which will enable automated fast analysis of measurements.
Use data on the materials process parameters (i.e how it was prepared. for example deposition temperature, oxygen levels, etc.) and machine learning algorithms to find the optimal process parameters for a desired functionality. We have implemented this for finding optimal process parameters for iron oxide absorber layer in a photovoltaic cell and with machine learning we achieved double the photovoltaic performance of the cells. Our database includes process parameters for all our samples and this method can be used for optimizing different functionalities (such as optical, resistivity, current-voltage, etc.)