Combining Artificial Intelligence with Combinatorial X-ray Diffraction Enables Rapid Phase Mapping of New Materials

JCAP’s high-throughput team has partnered with computer scientists from Cornell University to develop a new method for rapid construction of phase diagrams using data from JCAP’s collaboration with Stanford Synchrotron Radiation Lightsource Laboratory.

Suram, S. K. et al. Automated Phase Mapping with AgileFD and its Application to Light Absorber Discovery in the V-Mn-Nb Oxide System. ACS Combinatorial Science, 19(1), 37-46 DOI: 10.1021/acscombsci.6b00153 (2017)

Discovery of new materials and understanding their structure-property relationships are needed to realize the promise of generating solar fuels from carbon dioxide and water. JCAP scientists in the high-throughput group have made significant progress in identifying functional photoabsorbers, catalysts, and their assemblies using variety of accelerated synthesis of combinatorial libraries and screening methods. X-ray diffraction is an analytical technique that is routinely used for identification of the crystallographic phases present in a material. Rapid construction of phase diagrams from X-ray characterization of combinatorial composition libraries is a grand challenge in materials discovery research and is being addressed by JCAP’s joint project with the Computational Sustainability Network (

JCAP ACS Combinatorial Science Suram et al 2017 cover 19(1)

A recent article led by S. Suram details the development of an algorithm that captures the complexity of material phase behavior. The algorithm is called AgileFD and is the first scalable phase mapping algorithm for combinatorial XRD data that models alloying-based peak shifting and adheres to Gibbs’ phase rules. AgileFD enabled characterization of the previously-unknown phase behavior of V-Mn-Nb oxides where the identification of eight phases, including phases with significant alloying, generated the first phase map for this pseudoternary system. When the AgileFD solutions are combined with bad-gap energies from automated Tauc analysis and high-throughput UV-Vis spectroscopy data, the researchers could identify band-gap tuning of nearly 0.2 eV in the MnV2O6 crystal structure as a function of lattice parameters and vanadium composition in the energy range suitable for solar applications.

The article was recognized as the ACS Editor’s Choice®, is the cover article in January’s issue of ACS Combinatorial Science, and demonstrates how the multidisciplinary effort in material and computer science accelerates discovery of materials that are critical to emerging technologies such solar-fuel generation.


Related Links: