From Database to Diffractometer: Putting Structure Predictions to the Test

High-throughput computational databases now catalog thousands of predicted stable compounds, accelerating the search for new functional materials. But how reliably do these predictions translate to the lab? Motivated by attempts to synthesize computationally predicted superconductors that consistently formed disordered phases rather than the targeted structures, we launched a systematic experimental survey of candidates from OQMD/Materials Project/Alexandria. We attempted to synthesize a diverse set of compounds and used X-ray diffraction to assess whether the predicted phases actually form. In this talk, we present examples of both successful and unsuccessful synthesis attempts and discuss the emerging patterns. Our results reveal systematic patterns in prediction failures, often traceable to competing disordered phases or gaps in the underlying phase diagrams that affect calculated stability. This work offers an experimental perspective on high-throughput structure predictions, highlighting challenges that experimentalists should anticipate when selecting synthesis targets.