Better and faster design of organic light-emitting materials is within reach, thanks to the combination of machine learning and quantum computing. A collaborative research team has devised a new approach that employs a hybrid quantum-classical computational molecular design to accelerate the discovery of efficient OLED emitters, as detailed in a recent publication in Intelligent Computing.
Over the past decade, academia and industry have recognized the potential of organic luminescent materials for flexible and versatile optoelectronic devices like OLED displays. However, finding materials that exhibit optimal efficiency has been a challenge.
To tackle this obstacle, the research team developed a unique workflow that harnesses both machine learning and quantum computing. By utilizing a combination of classical and quantum computers, they were able to expedite the calculations necessary for designing deuterated OLED emitters, which are organic materials that replace hydrogen atoms with deuterium atoms in the emitter molecules.
The workflow begins with quantum chemistry calculations conducted on a classical computer to determine the quantum efficiencies of various deuterated Alq molecules. These efficiency data are then used to create training and test datasets for constructing a machine learning model. This model predicts the quantum efficiencies of different deuterated Alq molecules.
Next, the machine learning model is employed to build an energy function known as a Hamiltonian, which enables quantum optimization. Two quantum variational optimization algorithms called the variational quantum eigensolver (VQE) and the quantum approximate optimization algorithm (QAQA) are employed on a quantum computer to aid in the discovery of molecules with optimal quantum efficiencies. Additionally, a synthetic constraint is introduced during the quantum optimization process to ensure the synthesizability of the optimized molecule.
To enhance the accuracy of predictions on quantum devices, the authors implemented a noise-robust technique called recursive probabilistic variable elimination (RPVE). This technique enabled the authors to identify the optimal deuterated molecule with a high degree of accuracy using a quantum device. Furthermore, they noted that combining this noise-robust technique with the selected quantum optimization algorithms could lead to quantum advantage in calculations performed on near-term quantum devices.
The authors anticipate that their innovative approach, which merges quantum chemistry, machine learning, and quantum optimization, will open up new avenues for generating and optimizing essential molecules for material informatics. This advancement has the potential to revolutionize the design and development of organic light-emitting materials, paving the way for enhanced OLED displays and other optoelectronic devices in the future.
