Exploration of Chemical Space for Designing Functional Molecules Accounting for Geometric Stability

Abstract

The design of functional molecules is regarded as searching for molecules with desired functionalities in chemical space populated by candidate molecules. Considering the geometric stability of molecules during the search process is crucial for designing realistic molecules. Here, we propose a method for designing functional molecules by exploring chemical space while explicitly accounting for geometric stability based on computational quantum alchemy. The proposed design method allows the simultaneous prediction of functional molecule in the equilibrium structure and its target desired property in an inverse design fashion without preparing the molecular geometries and performing brute-force screening. The applicability of the design method is proven by obtaining molecules with the desired atomization and electronic energies in various chemical spaces: (BF, CO), (CH4, NH3), 18 BN-doped benzene derivatives, and 3.1 × 105 BN-doped phenanthrene derivatives.

Optimization of General Molecular Properties in the Equilibrium Geometry Using Quantum Alchemy: An Inverse Molecular Design Approach

“…The results are compared with those in Ref. 10. Second, we search for a molecule with high electric dipole strength (EDS) in simple (N2, CO)CS and realistic chemical space (BN-doped benzene derivatives BnNnC6−2nH6, n = 1-3)CS.…”

“…dp da change in the equilibrium geometry. 10 Assuming the equilibrium geometry, can be expressed as (5) The first term on the right-hand side represents the gradient in the iso-geometric chemical space and contains the elements of . The second term takes into account the displacement of the equilibrium geometry.…”

“…( 5) becomes , and their evaluations are avoided. 10 For the other functionalities, must be evaluated. That is, the computational cost of MDM becomes significantly higher when targeting general properties.…”

“…Therefore, in our previous study, MDM was only applied for optimizing energetic functionalities. 10 For the design of general properties, the present MDM avoids This procedure is repeated until the variation in a becomes sufficiently small (Figure 1(b)). The full optimization problem is solved under that condition since the equilibrium geometry change is…”

“…This design was also implemented in our previous study Ref. 10; the difference between this work and Ref. 10 is the algorithm for updating the molecular species.…”