Designing crystalline solids with specific functionalities is a challenge due to the complexity of predicting how molecules will assemble in the solid state. Traditional methods for discovering new functional materials rely heavily on trial and error, which is time-consuming and often ineffective.

Porous materials have shown great promise for applications such as carbon capture, fuel storage, and pollutant sequestration. Well-studied materials such as metal-organic frameworks (MOFs) and covalent organic frameworks (COFs) have limitations such as instability under certain conditions and high production costs. Molecular crystals are relatively unexplored as porous materials because molecules show a strong preference for close packing in the solid state. This case study demonstrates the value of a computation-led approach to design and guide the synthesis of a new class of porous organic salts. The work built on previous results performed using the ARCHER HPC, which demonstrated the computation-led discovery of, at the time, the most porous molecular crystals [1]. However, a limitation of these porous molecular crystals, which were held together by intermolecular hydrogen bonds, was that the materials were metastable, so that there was an energetic driving force for their transformation to non-porous structures during applications. In this work, published in Nature [2],  we sought to discover porous molecular crystals that correspond to their thermodynamically most stable structure, meaning that they should be stable to collapse of their pores.

Crystal structure prediction landscapes of three trigonal organic amines (top row), showing crystal packing of the predicted porous crystal structures (middle row) and comparison of simulated vs observed X-ray diffraction from the materials. Comparison of predicted structures to the X-ray diffraction patterns was used to determine the structures of the materials. Reproduced from [2].

Our research group develops crystal structure prediction (CSP) methods for molecular crystals. These were used to explore the hypothesis that porous crystal packing could be stabilised by using the interaction between net charges in molecular salts. CSP was used to study a series of halide salts of organic molecules that were designed with amine at the ends of long, rigid arms. These are some of the most complex molecular crystals studied to date using crystal structure prediction.

The computational results demonstrated that, for trigonal organic amines, the attraction of the protonated amines to the negatively charged halides promoted crystal packing that left large, open channels in the structure. Lattice energy calculations predicted that these porous packings were the most stable possible structures for a series of trigonal amines, as either chloride or bromide salts. Having obtained these promising predictions, the materials were synthesised and confirmed the predicted structures.

We explored applications of this new class of porous materials, which have been named non-metal organic frameworks, because they have some similar structural features to metal organic frameworks, but without metal atoms linking the organic molecules. The isoreticular salts exhibited exceptional capacity for trapping iodine, outperforming all but five known MOFs reported for iodine capture. This capability is crucial for applications such as radioactive iodine capture in the nuclear industry. Moreover, the solids were shown to be structurally stable over several cycles of adsorption and desorption, making them suitable for practical applications.

The materials are currently being studied for further applications that are relevant for energy applications, such as high proton conductivity that is aided by the channels within the structures.

Outcomes:

• Demonstrated the predictive capability of crystal structure prediction methods for complex salt systems, with predictions verified experimentally;
• Applied the computational approach to design salt systems where the thermodynamic global minimum structures are porous and, therefore, stable over repeated cycles of guest uptake and release;
• Discovered materials with better properties than existing materials in areas of importance for energy-related applications: iodine capture, with importance in nuclear energy generation, and proton conductivity, for use in fuel cells, batteries and sensors.[3]

References:

[1] Functional materials discovery using energy–structure–function maps. Nature 543, 657–664 (2017).

[2] Porous isoreticular non-metal organic frameworks. Nature 630, 102–108 (2024)

[3] Non-Metal Organic Frameworks Exhibit High Proton Conductivity, Journal of the American Chemical Society, accepted for publication (2025).

Collaborators:

Southampton, School of Chemistry and Chemical Engineering: Graeme M. Day, Joseph Glover, Roohollah Hafizi.

Experimental collaborators, Liverpool Chemistry: Megan O’Shaughnessy, Andrew I. Cooper, Mounib Barhi, Rob Clowes, Samantha Y. Chong, Stephen P. Argent, Jungwoo Lim, Alex R. Neale, Laurence J. Hardwick.

Funding:

• European Research Council, Horizon2020, Autonomous Discovery of Advanced Materials (ADAM, grant agreement 856405)

Contact:

Prof. Graeme Day; g.m.day@soton.ac.uk