Researchers use hybrid model on quantum computers to predict molecular behavior in solvents

A team led by Kenneth Merz, PhD, Staff in Cleveland Clinic's Center for Computational Life Sciences, showed how quantum computers can be used for investigating how molecules act in aqueous solutions. 

One of the most essential components of chemistry is understanding how molecules react in specific liquids. For example in water-based solvent solutions, a molecule interacts with the water molecules around it, causing it to change its behavior. If researchers can predict how a molecule will change when it is in a liquid, they can create more effective drugs and treatments.  

Calculating all the possible reactions and outcomes of each molecule in solvent is too time-consuming and costly for current methods on classical computers. To overcome this, the Merz lab completed the first study to demonstrate implicit solvent simulations using Sample-Based Quantum Diagonalization (SQD) on real quantum hardware showing a new way forward to better characterize molecules in solution.  

The paper, published in a special issue of The Journal of Physical Chemistry, is also the first study to test SQD in the solvent phase. The paper was featured as the cover story for this issue of the journal.  

To understand how molecules behave in liquids, Dr. Merz and his team used IBM Quantum System One to run SQD, which selects electronic configurations “samples” of a molecule that helps to characterize the energy of a molecule. The samples are then sent to a classical computer that analyzes them and selects the most likely outcomes. This process is repeated and improved until the most accurate prediction is made.  As part of a partnership with IBM, Cleveland Clinic is home to IBM Quantum System One, the first quantum computer dedicated to healthcare research.

To validate the model, the team tested it on four polar molecules that are commonly observed in chemistry and biology; methanol, ethanol, methylamine and water. Up to 52 qubits were utilized on each test, which achieved a chemical accuracy of less than 1 kcal/mol. These results demonstrate the model’s ability to predict the molecules’ energies and solvation free energy.  

“This study is a significant stride towards practical quantum chemistry on quantum computers,” Dr. Merz says. “Quantum hybrid models are still largely unexplored and very few are tested on quantum hardware. By testing this model on quantum hardware, we are demonstrating its abilities to advance chemical research using quantum computers.”