Deep generative design of porous organic cages via a variational autoencoder

Porous organic cages (POCs) are a class of porous molecular materials characterised by their tunable, intrinsic porosity; this functional property makes them candidates for applications including guest storage and separation. Typically formed via dynamic covalent chemistry reactions from multifunctionalised molecular precursors, there is an enormous potential chemical space for POCs due to the fact they can be formed by combining two relatively small organic molecules, which themselves have an enormous chemical space. However, identifying suitable molecular precursors for POC formation is challenging, as POCs often lack shape persistence (the cage collapses upon solvent removal with loss of its cavity), thus losing a key functional property (porosity). Generative machine learning models have potential for targeted computational design of large functional molecular systems such as POCs. Here, we present a deep-learning-enabled generative model, Cage-VAE, for the targeted generation of shape-persistent POCs. We demonstrate the capacity of Cage-VAE to propose novel, shape-persistent POCs, via integration with multiple efficient sampling methods, including Bayesian optimisation and spherical linear interpolation.

Industrial Separation Challenges: How Does Supramolecular Chemistry Help?

The chemical industry and the chemical processes underscoring it are under intense scrutiny as the demands for the transition to more sustainable and environmentally friendly practices are increasing. Traditional industrial separation systems, such as thermally driven distillation for hydrocarbon purification, are energy intensive. The development of more energy efficient separation technologies is thus emerging as a critical need, as is the creation of new materials that may permit a transition away from classic distillation-based separations. In this Perspective, we focus on porous organic cages and macrocycles that can adsorb guest molecules selectively through various host-guest interactions and permit molecular sieving behavior at the molecular level. Specifically, we summarize the recent advances where receptor-based adsorbent materials have been shown to be effective for industrially relevant hydrocarbon separations, highlighting the underlying host-guest interactions that impart selectivity and permit the observed separations. This approach to sustainable separations is currently in its infancy. Nevertheless, several receptor-based adsorbent materials with extrinsic/intrinsic voids or special functional groups have been reported in recent years that can selectively capture various targeted guest molecules. We believe that the understanding of the interactions that drive selectivity at a molecular level accruing from these initial systems will permit an ever-more-effective "bottom-up" design of tailored molecular sieves that, in due course, will allow adsorbent material-based approaches to separations to transition from the laboratory into an industrial setting.

Identifying porous cage subsets in the Cambridge Structural Database using topological data analysis

As rationally designable materials, the variety and number of synthesised metal-organic cages (MOCs) and organic cages (OCs) are expected to grow in the Cambridge Structural Database (CSD). In this regard, two of the most important questions are, which structures are already present in the CSD and how can they be identified? Here, we present a cage mining methodology based on topological data analysis and a combination of supervised and unsupervised learning that led to the derivation of - to the best of our knowledge - the first and only MOC dataset of 1839 structures and the largest experimental OC dataset of 7736 cages, as of March 2022. We illustrate the use of such datasets with a high-throughput screening of MOCs and OCs for xenon/krypton separation, important gases in multiple industries, including healthcare.

Digitization of the electron shell via the localized orbital locator formalism: trends in the size and electronegativity changes of atoms across the periodic table

The parameters of the (3,−3) critical point in the topology of the localized orbital locator inside the electron shell reflect regularity in the change of basic atom properties across the periodic table.

Evaluation of packing single and multiple atoms and molecules in the porous organic cage CC3-R

The absorption of multiple atoms and molecules, including Kr, Xe, CH₄, CO₂, C₂H₂, H₂O, and SF₆, within CC3-R, a Porous Organic Cage (POC), was calculated and analyzed. The CC3-R molecule has one central cavity and four window sites. Most adsorbents were modeled with either one unit in the central cavity, four units in the window sites, or with five units in both sites. For Xe, the most favorable site was the central one. The CO₂ molecule binds about 3 kcal mol^-1 in free energy more strongly than CH₄ in the central cavity of CC3-R at 300 K which may be enough to allow useful discrimination. Four C₂H₂ units and four CO₂ units are calculated to bind similarly inside CC3-R (ΔH(298 K) = -8.6 and -7.7 kcal mol^-1 per unit, respectively). Since H₂O is smaller, more waters can easily fit inside. For twelve water molecules, the binding enthalpy per water is ΔH(298 K) = -16.4 kcal mol^-1. For comparison, the binding enthalpy of (H₂O)₁₂ at the same level of theory (B3LYP/6-31G(d,p)-D3BJ//M06-2X/6-31G(d)) is predicted to be -12.3 kcal mol^-1 per water. Finally, the dimerization of CC3-R and the association of CC3-R with CC3-S was studied as well as 3 to 9 iodine atoms enclosed in CC3-R.

Machine Learning Meets with Metal Organic Frameworks for Gas Storage and Separation

The acceleration in design of new metal organic frameworks (MOFs) has led scientists to focus on high-throughput computational screening (HTCS) methods to quickly assess the promises of these fascinating materials in various applications. HTCS studies provide a massive amount of structural property and performance data for MOFs, which need to be further analyzed. Recent implementation of machine learning (ML), which is another growing field in research, to HTCS of MOFs has been very fruitful not only for revealing the hidden structure-performance relationships of materials but also for understanding their performance trends in different applications, specifically for gas storage and separation. In this review, we highlight the current state of the art in ML-assisted computational screening of MOFs for gas storage and separation and address both the opportunities and challenges that are emerging in this new field by emphasizing how merging of ML and MOF simulations can be useful.

Robust Supramolecular Nano-Tunnels Built from Molecular Bricks*

Herein we report a linear ionic molecule that assembles into a supramolecular nano-tunnel structure through synergy of trident-type ionic interactions and π-π stacking interactions. The nano-tunnel crystal exhibits anisotropic guest adsorption behavior. The material shows good thermal stability and undergoes multi-stage single-crystal-to-single-crystal phase transformations to a nonporous structure on heating. The material exhibits a remarkable chemical stability under both acidic and basic conditions, which is rarely observed in supramolecular organic frameworks and is often related to structures with designed hydrogen-bonding interactions. Because of the high polarity of the tunnels, this molecular crystal also shows a large CO₂ -adsorption capacity while excluding other gases at ambient temperature, leading to high CO₂ /CH₄ selectivity. Aggregation-induced emission of the molecules gives the bulk crystals vapochromic properties.

Toward crystalline porosity estimators for porous molecules

Our data-mining of crystalline molecular materials reveals the correlations between the molecular and crystalline porosity.

Toward Automated Tools for Characterization of Molecular Porosity

The emerging advanced porous materials, e.g. extended framework materials and porous molecular materials, offer an unprecedented level of control of their structure and function. The enormous possibilities for tuning these materials by changing their building blocks mean that, in principle, optimally performing materials for a variety of applications can be systematically designed. However, the process of finding a set of optimal structures for a given application requires computational high-throughput tools to analyze and sieve through many candidate materials. In particular, in the case of porous molecular materials, the analysis and selection of a molecule is one of the key aspects as the structure of the molecule determines the structure of the resulting material, and very often the porosity of the molecule significantly contributes to the porous properties of the resulting material. In this work, we introduce definitions and algorithms to characterize porosity at the molecular level, along with a software implementation of these algorithms. We demonstrate applications of the software tool in the discovery and characterization of porous molecules among ca. 94 million molecules currently enlisted in the PubChem database.

Fluorinated porous molecular crystals: vapor-triggered on–off switching of luminescence and porosity

Fluorinated porous molecular crystals (PMCs) were fabricated from platinum(ii) dihalide complexes with 9-pentafluorophenyl-9-arsafluorene. The diiodide complex formed a PMC exhibiting open–close switching of porosity as well as on–off switching of luminescence.

Eigencages: Learning a Latent Space of Porous Cage Molecules

Porous organic cage molecules harbor nanosized cavities that can selectively adsorb gas molecules, lending them applications in separations and sensing. The geometry of the cavity strongly influences their adsorptive selectivity. For comparing cages and predicting their adsorption properties, we embed/encode a set of 74 porous organic cage molecules into a low-dimensional, latent "cage space" on the basis of their intrinsic porosity. We first computationally scan each cage to generate a three-dimensional (3D) image of its porosity. Leveraging the singular value decomposition, in an unsupervised manner, we then learn across all cages an approximate, lower-dimensional subspace in which the 3D porosity images congregate. The "eigencages" are the set of orthogonal, characteristic 3D porosity images that span this lower-dimensional subspace, ordered in terms of importance. A latent representation/encoding of each cage follows by approximately expressing it as a combination of the eigencages. We show that the learned encoding captures salient features of the cavities of porous cages and is predictive of properties of the cages that arise from cavity shape. Our methods could be applied to learn latent representations of cavities within other classes of porous materials and of shapes of molecules in general.

An evolutionary algorithm for the discovery of porous organic cages

The chemical and structural space of possible molecular materials is enormous, as they can, in principle, be built from any combination of organic building blocks. Here we have developed an evolutionary algorithm (EA) that can assist in the efficient exploration of chemical space for molecular materials, helping to guide synthesis to materials with promising applications. We demonstrate the utility of our EA to porous organic cages, predicting both promising targets and identifying the chemical features that emerge as important for a cage to be shape persistent or to adopt a particular cavity size. We identify that shape persistent cages require a low percentage of rotatable bonds in their precursors (<20%) and that the higher topicity building block in particular should use double bonds for rigidity. We can use the EA to explore what size ranges for precursors are required for achieving a given pore size in a cage and show that 16 Å pores, which are absent in the literature, should be synthetically achievable. Our EA implementation is adaptable and easily extendable, not only to target specific properties of porous organic cages, such as optimal encapsulants or molecular separation materials, but also to any easily calculable property of other molecular materials.

High-Throughput Screening of Metal-Organic Frameworks for Macroscale Heteroepitaxial Alignment

The ability to align porous metal-organic frameworks (MOFs) on substrate surfaces on a macroscopic scale is a vital step toward integrating MOFs into functional devices. But macroscale surface alignment of MOF crystals has only been demonstrated in a few cases. To accelerate the materials discovery process, we have developed a high-throughput computational screening algorithm to identify MOFs that are likely to undergo macroscale aligned heterepitaxial growth on a substrate. Screening of thousands of MOF structures by this process can be achieved in a few days on a desktop workstation. The algorithm filters MOFs based on surface chemical compatibility, lattice matching with the substrate, and interfacial bonding. Our method uses a simple new computationally efficient measure of the interfacial energy that considers both bond and defect formation at the interface. Furthermore, we show that this novel descriptor is a better predictor of aligned heteroepitaxial growth than other established interface descriptors, by testing our screening algorithm on a sample set of copper MOFs that have been grown heteroepitaxially on a copper hydroxide surface. Application of the screening process to several MOF databases reveals that the top candidates for aligned growth on copper hydroxide comprise mostly MOFs with rectangular lattice symmetry in the plane of the substrate. This result indicates a substrate-directing effect that could be exploited in targeted synthetic strategies. We also identify that MOFs likely to form aligned heterostructures have broad distributions of in-plane pore sizes and anisotropies. Accordingly, this suggests that aligned MOF thin films with a wide range of properties may be experimentally accessible.

Computational modelling of solvent effects in a prolific solvatomorphic porous organic cage

Crystal structure prediction methods can enable the in silico design of functional molecular crystals, but solvent effects can have a major influence on relative lattice energies, sometimes thwarting predictions. This is particularly true for porous solids, where solvent included in the pores can have an important energetic contribution. We present a Monte Carlo solvent insertion procedure for predicting the solvent filling of porous structures from crystal structure prediction landscapes, tested using a highly solvatomorphic porous organic cage molecule, CC1. Using this method, we can understand why the predicted global energy minimum structure for CC1 is never observed from solvent crystallisation. We also explain the formation of three different solvatomorphs of CC1 from three structurally-similar chlorinated solvents. Calculated solvent stabilisation energies are found to correlate with experimental results from thermogravimetric analysis, suggesting a future computational framework for a priori materials design that factors in solvation effects.

High-Throughput Screening Approach for the Optoelectronic Properties of Conjugated Polymers

We propose a general high-throughput virtual screening approach for the optical and electronic properties of conjugated polymers. This approach makes use of the recently developed xTB family of low-computational-cost density functional tight-binding methods from Grimme and co-workers, calibrated here to (Time-Dependent) Density Functional Theory ((TD)DFT) data computed for a representative diverse set of (co)polymers. Parameters drawn from the resulting calibration using a linear model can then be applied to the xTB derived results for new polymers, thus generating near DFT-quality data with orders of magnitude reduction in computational cost. As a result, after an initial computational investment for calibration, this approach can be used to quickly and accurately screen on the order of thousands of polymers for target applications. We also demonstrate that the (opto)electronic properties of the conjugated polymers show only a very minor variation when considering different conformers and that the results of high-throughput screening are therefore expected to be relatively insensitive with respect to the conformer search methodology applied.

On flexible force fields for metal-organic frameworks: Recent developments and future prospects

Classical force field simulations can be used to study structural, diffusion, and adsorption properties of metal-organic frameworks (MOFs). To account for the dynamic behavior of the material, parameterization schemes have been developed to derive force constants and the associated reference values by fitting on ab initio energies, vibrational frequencies, and elastic constants. Here, we review recent developments in flexible force field models for MOFs. Existing flexible force field models are generally able to reproduce the majority of experimentally observed structural and dynamic properties of MOFs. The lack of efficient sampling schemes for capturing stimuli-driven phase transitions, however, currently limits the full predictive potential of existing flexible force fields from being realized. This article is categorized under: Structure and Mechanism > Computational Materials ScienceMolecular and Statistical Mechanics > Molecular Mechanics.

Dissecting Porosity in Molecular Crystals: Influence of Geometry, Hydrogen Bonding, and [π···π] Stacking on the Solid-State Packing of Fluorinated Aromatics

Porous molecular crystals are an emerging class of porous materials that is unique in being built from discrete molecules rather than being polymeric in nature. In this study, we examined the effects of molecular structure of the precursors on the formation of porous solid-state structures with a series of 16 rigid aromatics. The majority of these precursors possess pyrazole groups capable of hydrogen bonding, as well as electron-rich aromatics and electron-poor tetrafluorobenzene rings. These precursors were prepared using a combination of Pd- and Cu-catalyzed cross-couplings, careful manipulations of protecting groups on the nitrogen atoms, and solvothermal syntheses. Our study varied the geometry and dimensions of precursors, as well as the presence of groups capable of hydrogen bonding and [π···π] stacking. Thirteen derivatives were crystallographically characterized, and four of them were found to be porous with surface areas between 283 and 1821 m² g^-1. Common to these four porous structures were (a) rigid trigonal geometry, (b) [π···π] stacking of electron-poor tetrafluorobenzenes with electron-rich pyrazoles or tetrazoles, and

Porous Molecular Solids and Liquids

Until recently, porous molecular solids were isolated curiosities with properties that were eclipsed by porous frameworks, such as metal-organic frameworks. Now molecules have emerged as a functional materials platform that can have high levels of porosity, good chemical stability, and, uniquely, solution processability. The lack of intermolecular bonding in these materials has also led to new, counterintuitive states of matter, such as porous liquids. Our ability to design these materials has improved significantly due to advances in computational prediction methods.

Functional materials discovery using energy-structure-function maps

Molecular crystals cannot be designed in the same manner as macroscopic objects, because they do not assemble according to simple, intuitive rules. Their structures result from the balance of many weak interactions, rather than from the strong and predictable bonding patterns found in metal-organic frameworks and covalent organic frameworks. Hence, design strategies that assume a topology or other structural blueprint will often fail. Here we combine computational crystal structure prediction and property prediction to build energy-structure-function maps that describe the possible structures and properties that are available to a candidate molecule. Using these maps, we identify a highly porous solid, which has the lowest density reported for a molecular crystal so far. Both the structure of the crystal and its physical properties, such as methane storage capacity and guest-molecule selectivity, are predicted using the molecular structure as the only input. More generally, energy-structure-function maps could be used to guide the experimental discovery of materials with any target function that can be calculated from predicted crystal structures, such as electronic structure or mechanical properties.

A publicly available crystallisation data set and its application in machine learning

A publicly available crystallisation database for clusters of highly similar compounds is used to build machine learning models.

Application of computational methods to the design and characterisation of porous molecular materials

Composed from discrete units, porous molecular materials (PMMs) possess properties not observed for conventional, extended solids. Molecular simulations provide crucial understanding for the design and characterisation of these unique materials.