A molecular crystal does not have a single stiffness value. Its molecules are arranged differently along different crystallographic directions, so the same crystal can resist force strongly on one morphological facet, an exposed crystal surface, and deform or crack more easily on another. In pharmaceutical manufacturing, this can affect milling, powder handling and tablet compression, influencing whether a medicine can be produced consistently. In electronic materials, facet-specific stiffness can affect durability during manufacture and use. Laboratory nanoindentation remains essential, but some facets are too small or difficult to test reliably.
We developed the MechaPredict software to identify promising facets and predict their mechanical response before committing to expensive experiments. MSc in Applied Chemistry student, Mubarak Almehairbi, wrote and tested the code in work conceived by Prof. Sharmarke Mohamed, Associate Professor, Chemistry and Theme Leader at the Center for Catalysis and Separations.
The work was published in Chemistry – A European Journal in 2024. A public GitHub project page provides releases, an installer, documentation and license-request information. MechaPredict already has several freely licensed academic users around the world. Copyright protection is under review by the relevant UAE bodies, with support from Khalifa University Enterprises Company (KUEC), the University’s technology transfer platform. We are exploring licensing routes for for-profit customers in the pharmaceutical industry who have expressed interest in licensing the code from Khalifa University. The software is being marketed and maintained by Solid Form Innovations, a spin-out company founded and led by Prof. Sharmarke Mohamed.
MechaPredict is a software that calculates the facet-specific Young’s modulus of molecular crystals, showing how stiff a chosen crystal facet is when force is applied. It turns quantum-mechanical calculations into a mechanical property that can be tested in the laboratory. During development, we used Khalifa University’s high-performance computing (HPC) cluster to study molecular crystals systematically. The method was subsequently published, and the software was made available to licensed academic users.
The calculations use dispersion-corrected density functional theory, a quantum-mechanical method for predicting how molecules interact in a crystal. We compared three models: PBE-D3, PBE-TS and PBE-MBD. MechaPredict uses the calculated stiffness data and the index of a selected morphological facet to determine its Young’s modulus, a measure of how strongly that facet resists deformation.
MechaPredict’s tensor transformations are completed in an instant; producing dependable input data is the demanding part. Each crystal must be geometrically optimized, and calculating the elastic constants that describe its response to stress requires many repeated quantum-mechanical calculations. Khalifa University鈥檚 HPC cluster allowed us to apply the same protocol across 15 crystals, 24 facet-specific experimental measurements and three models. We validated the code by comparing predictions with those measurements. For the (020) facet of L-aspartic acid, PBE-D3 predicted a stiffness of 24.59 gigapascals (GPa) before the experiment; the later measurement was 24.53 卤 0.56 GPa. All three models fell within the reported experimental uncertainty.
