SUPERLASER has been awarded over 3 million HPC hours for the computational prediction of perovskite materials capable of emitting superradiance.

Katherine Inzani and Connor Williamson from project partner University of Nottingham have been awarded more than 3 million High-Performance Computing (HPC) hours for the SUPERLASER project. These are provisioned on the regional supercomputer Sulis, managed by the HPC Midlands+ consortium, and the UK’s national supercomputer Archer2, via the Materials Chemistry Consortium. This significant resource award will enable high-throughput simulations that aim to unlock the potential of room-temperature super-radiant perovskites.

Super-radiant perovskites have been synthesised and utilised in experimental devices based on nanocrystal superlattices. However, their coherent emission has been limited to low temperatures due to structural imperfections at the nanocrystal grain boundaries.

A central objective of the SUPERLASER project is the prediction of suitable perovskite materials, that have the potential for superradiance at room temperature. Computationally, this requires high-throughput calculations of the optical properties of crystal structures to ascertain a map of key properties: band gaps, density of states, phonon density of states and Franck-Condon factors.

SUPERLASER focuses on halide perovskites with the general formula ABX₃, where A is a cation (organic or inorganic), B is a heavy metal, and X is a halide. The study also explores double perovskites (A₂BB′X₆), which form when two different metals occupy the B site. The primary goal is to predict new materials capable of exhibiting superradiance at room temperature—a phenomenon that has not yet been theoretically demonstrated through ab initio simulations.

To achieve this, the team developed a novel predictive workflow that goes beyond conventional tolerance factor-based stability models. Using this approach, they will generate a comprehensive map of over one million perovskite compounds, with stable phases and strong spin-orbit-coupling, focusing on non-radioactive, non-critical halide-based materials.

A Funnel-Like Materials Screening Strategy

The screening process follows a funnel-like strategy: starting with a broad pool of candidates and progressively narrowing it down to the most promising ones through increasingly complex computational methods. Each stage of the funnel includes increasingly rigorous methods:

1. Compositional Feasibility
   • Screening complex compositions (e.g., AA′BB′X₃X′₃) to assess likelihood of forming a perovskite structure
2. Phase Identification
   • Predicting structural phases (cubic, tetragonal, orthorhombic, rhombohedral) across temperature ranges
3. Thermodynamic Stability
   • Determining whether the material is likely to decompose into simpler compounds
4. Dynamic Stability
   • Using phonon spectra (calculated via machine-learned interatomic potentials and validated by DFT) to assess distortions
5. Electronic Structure Analysis
   • Evaluating properties like band gaps and Rashba splitting using hybrid DFT functionals with spin-orbit coupling—essential for systems with heavy atoms
6. Optical Emission Prediction
   • Assessing potential for superradiance through a three-pronged test of computable optical properties, including: many-body GW+BSE calculations, electron-phonon coupling analyses and high-precision electronic and optical modeling

Role of HPC Resources

The Archer2 and Sulis computing resources are indispensable for every stage of this pipeline. Early stages involve a large amount of broad but relatively simple calculations. As the pool of candidates narrows, more intensive simulations—such as DFT, phonon modeling, and many-body GW-BSE theory—are used to achieve high-fidelity predictions. So far this year, 2.2 million hours have been awarded on Archer2, which are being used for the high-throughput screening of a large number of candidate materials, whilst 1 million hours have been awarded on Sulis for the high-precision electronic and optical modeling. In addition, 34,000 GPU hours have been awarded for acceleration of the screening strategy via machine learning.

This systematic, scalable approach allows the team to efficiently pinpoint the most promising candidates for room-temperature superradiance, pushing the boundaries of what is currently possible in materials prediction and quantum optoelectronics.

© Stuart Hollis / Hollis Photography