Quantum Computers Now Model Complex Molecules with Greater Accuracy

Quantum calculations now routinely match the accuracy of established classical methods for complex chemical systems. A new multilevel embedding framework achieves agreement within approximately 1 kilocalorie per mole with classical references for a Menshutkin SN2 reaction occurring inside a carbon nanotube. This advance combines quantum computation, focused on strongly correlated parts of molecules, with classical high-level wave-function methods for the surrounding areas, bridged by quantum-selected configuration interaction and embedded within density functional theory

Tuan Minh Do of the University of Osaka and colleagues have created a method for modelling chemical systems by combining quantum computers with traditional computational techniques. The framework intelligently allocates computational effort, with a quantum algorithm focusing on complex electron interactions within a specific molecular area, while classical methods handle the rest. The approach achieved results within approximately one kilocalorie per mole of established references for a particular chemical reaction inside a carbon nanotube, demonstrating improved accuracy.

Tuan Minh Do and colleagues at the University of Osaka have developed a hybrid quantum-classical computing framework to model complex chemical systems with increased precision. This method tackles a longstanding challenge in quantum chemistry: accurately simulating large molecules where some electrons interact intensely and are difficult to predict. The framework strategically allocates computational resources, utilising a quantum algorithm for these intensely interacting electrons, while employing conventional methods for the remainder of the molecule. Results demonstrate agreement within approximately one kilocalorie per mole of established references for a Menshutkin SN2 reaction occurring within a carbon nanotube, a level of accuracy previously difficult to achieve. This approach, which uses quantum-selected configuration interaction combined with density functional theory, estimates electron behaviour by assessing the space they occupy.

Precision hybrid quantum-classical modelling of molecular systems using multilevel embedding

A landmark in hybrid quantum-classical computation has been achieved, with agreement within approximately 1 kilocalorie per mole with established classical references, a threshold previously unattainable for complex systems. Intelligent combination of quantum and classical computational techniques surpasses the accuracy of prior methods. The new multilevel embedding framework addresses limitations in capturing electron correlation, a key challenge in modelling molecular behaviour, while simultaneously managing the computational cost associated with large systems.

Bond dissociation and adsorption energies were calculated to assess chemical reactions and material interactions, utilising only a subset of the 144 qubits available on a superconducting quantum computer at the University of Osaka. Consistent results across organic, metal-organic, and metallic systems were obtained when computing reaction barriers, specifically for the Menshutkin SN2 reaction within a carbon nanotube. Efficient identification of key electronic wave function components streamlined the hybrid quantum-classical process, and coupling with advanced classical methods like coupled cluster theory refined accuracy beyond the active molecular area.

However, this level of precision does not yet translate to practical applications for exceedingly complex molecules or simulations extending beyond a few dozen atoms, highlighting the ongoing need for substantial hardware and algorithmic improvements. This framework offers a potential route to reliable calculations for larger molecules, bridging the gap between the precision of high-level wave-function theory and the scalability of density functional theory. Osaka University scientists have demonstrated promising results with a 144-qubit processor, yet full quantum accuracy for complex systems remains a distant goal. This hybrid approach represents a strong step forward, particularly in addressing the challenges of modelling systems where electrons interact in complex ways. The computational framework successfully integrates quantum algorithms with classical methods, efficiently managing computational demands by focusing quantum computation on strongly correlated active spaces, regions of molecules where electron interactions are particularly intense, acting as a bridge linking these quantum and classical treatments within a density functional theory environment, allowing for calculations on a subset of qubits from a 144-qubit superconducting processor, and providing a foundation for future expansion of the method to more complex molecular systems.

The significance of accurately modelling chemical systems lies in its broad applicability to fields such as drug discovery, materials science, and catalysis. Traditional quantum chemistry methods, while highly accurate, suffer from a computational cost that scales exponentially with system size, limiting their application to relatively small molecules. Density functional theory (DFT) offers a more scalable alternative, but often struggles to accurately describe strongly correlated systems, where electron interactions are particularly complex. This new multilevel embedding framework aims to overcome these limitations by strategically combining the strengths of both approaches. The framework’s core innovation lies in its ability to treat the strongly correlated “active space”, the region of a molecule where electron correlation is most critical, using a quantum algorithm, while employing classical methods to describe the less critical surrounding environment.

The quantum algorithm employed is a sampling-based approach known as quantum-selected configuration interaction (QSCCI). Configuration interaction (CI) is a method for approximating the many-body Schrödinger equation by expressing the wave function as a linear combination of Slater determinants, representing different electronic configurations. QSCCI leverages the power of quantum computation to efficiently sample these configurations, focusing on those that contribute most significantly to the overall wave function. This sampling process is then integrated with classical high-level wave-function methods, such as coupled cluster theory (CCSD(T)) or multireference perturbation theory (MRPT), to refine the description of the surrounding molecular region. The entire calculation is embedded within a DFT framework, providing a consistent and efficient way to handle the overall system.

The Menshutkin SN2 reaction, chosen as a benchmark for this method, is a well-studied nucleophilic substitution reaction. Performing this calculation within the confines of a carbon nanotube introduces additional complexity due to the unique electronic properties of the nanotube structure and its influence on the reaction mechanism. Achieving agreement within approximately 1 kilocalorie per mole with established classical references for this reaction demonstrates the framework’s ability to accurately capture the subtle energetic details of a complex chemical process. The use of only 144 qubits on the superconducting quantum computer highlights the potential for near-term quantum devices to tackle challenging chemical problems. While 144 qubits represent a significant advancement, scaling to even larger qubit numbers and improving qubit coherence times remain crucial for tackling truly complex molecular systems.

The researchers meticulously validated their approach by computing bond dissociation and adsorption energies for a diverse range of systems, including organic molecules, metal-organic frameworks, and metallic surfaces. This broad validation confirms the robustness and general applicability of the framework. The ability to consistently obtain accurate results across different chemical environments is a testament to the careful design and implementation of the hybrid quantum-classical algorithm. Future work will focus on extending the framework to larger molecules and more complex chemical reactions, as well as exploring alternative quantum algorithms and classical methods to further enhance its accuracy and efficiency. The ultimate goal is to develop a computational tool that can reliably predict the properties of complex chemical systems, accelerating the discovery of new materials and technologies.

The research successfully demonstrated a new multilevel embedding framework combining quantum and classical computational methods. This approach allows for more accurate quantum chemistry calculations on complex systems by treating strongly correlated parts of molecules with a quantum computer, while utilising classical methods for the remainder. Researchers achieved agreement of approximately 1 kcal/mol with classical references when modelling a Menshutkin S N2 reaction inside a carbon nanotube using 144 qubits. The authors intend to extend this framework to larger molecules and more complex reactions to improve its accuracy and efficiency.