Ionic liquids (ILs) are salts that are liquids at ambient conditions (typically below 373 K) and are known for their many unique properties, including low volatility and high thermal stability. Despite the promise of ILs, their targeted design is challenging for several reasons, including i) the vast number of candidate ions that could be synthesised as components of an IL, ii) the lack of predictive methods to determine what ion combinations will yield ILs with desired melting points, and iii) experimentally known ILs possess limited chemical diversity. In this work, we present a data-driven framework for designing novel low-melting-point ILs. We model ILs as bipartite graphs and apply a link prediction algorithm to identify promising cation-anion pairs, expanding the collected IL database more than 30- fold, while prioritising low melting points. To further explore chemical space, we trained variational auto-encoders (VAEs) to generate new IL candidates through learning a latent space that enables modelling the data distribution. A thermodynamics-inspired classification model is subsequently employed to filter out ILs predicted to melt above 373 K. Finally, molecular dynamics simulations validate our approach, confirming that 18 out of 20 generated ILs have melting points below 373 K.

Ionic liquids (ILs) are salts set apart by their low melting points and can act as highly tuneable solvents with broad application potential, for example as catalysts, in batteries, and for drug delivery. The potential chemical space of ILs is vast, with only a very small region having been explored to date. Machine learning offers a promising approach to advance into this vast space of unexplored ILs; however, existing IL databases contain limited ion diversity, constraining the performance of generative models. To address this, we introduce conditional variational autoencoders (CVAEs) and a novel ion scoring method as a conditioning factor. The ion score prioritises ions with a higher likelihood of forming low-melting-point ILs. Our CVAEs effectively generate novel and diverse cations and anions. Furthermore, we constructed a melting point prediction model to identify cation-anion pairs that are likely to yield ILs with low melting points. Visualisation of the generated ILs alongside existing ones reveals that our approach effectively expands the chemical space of ILs with novel structures. Molecular dynamics simulations further validate that 13/15 of the generated ILs possess desirable low melting points (<373 K). The associated code is available at https://github.com/fate1997/ILGen-ion.

Selective recognition of ammonium cations fundamentally relies on their degree of substitution. In biological systems, proteins can preferentially bind more substituted ammonium cations over less substituted homologues. To date, a general methodology to mimic this behavior remains elusive as synthetic hosts principally observe the inverse order (i.e., 1°>2°>3°>4°). Here we show that, through combining supramolecular recognition with solid-phase abstraction, we can overturn the canonical order of synthetic receptor selectivity across a diverse range of ammonium cation scaffolds. Quaternary ammonium cations access a lower energy solid-state than tertiary counterparts through multipoint binding to an adaptive array of isostructural BINOL·counterion networks. The preferential abstraction of quaternary ammonium cations from mixtures of homologous cations proceeds under thermodynamic control with excellent selectivity and remains operative even under aqueous conditions.

Porous liquids are a sub-class of porous materials that combine permanent porosity, typically associated with solids, with the fluidity and fast mass-transfer capabilities of liquids, making them ideal candidates for gas storage and separation applications. One strategy to form porous liquids is the dissolution of discrete and permanently porous molecular species at relatively high concentrations in cavity-excluded solvents, thus introducing permanent porosity into the liquid in which it is dissolved and ensuring a solution of reasonable porosity is obtained. To access high-performance porous liquids for target applications, the selection of both the porous molecular species and the cavity-excluded solvent is key to ensuring the solvent is permanently excluded and the pore carrier is highly soluble. Finding new solvents that fit both these requirements is challenging, often resulting in a trial-and-error approach. While predictive data-driven models may be attractive, the youth of the porous liquid field currently limits the availability of data necessary to train robust models. Here, we present a computational workflow for the discovery of new porous liquid solutions combining solubility prediction software and a size-exclusivity prediction algorithm that correctly predicts size-exclusivity; this is followed by experimental validation with a representative system. Our workflow yielded size-excluded solvent and soluble porous organic cage pairs, leading to the realisation of a new porous liquid with enhanced methane uptake compared to previous systems discovered in a purely experimental high-throughput brute-force manner, highlighting the advantages of incorporating a computational workflow in the discovery of new porous liquids.

Graphite is a layered material with applications in nuclear reactors, electronics, and batteries. As a consequence of the weak van der Waals interactions between the layers, their stacking arrangement is often disordered which in turn affects the mechanical, electronic and optical properties of graphite. In this work, we develop an atomistic approach to predict the structure of graphite and capture its stacking disorder. We first perform ab initio density-functional theory to determine the potential energy surface of graphite, which is then used to parametrize an effective model. Using this model, we perform large-scale Monte-Carlo simulations to predict the atomic structure of graphite at finite temperatures. We compare the resulting structures to those produced by an Ising model and show that our approach produces X-ray diffraction patterns in closer agreement with experimental measurements. In the future, this approach can be straightforwardly applied to other layered materials, such as covalent organic frameworks or transition metal dichalcogenides.

Membrane-based separation processes hold great promise for sustainable extraction of lithium from brines for the rapidly expanding electric vehicle industry and renewable energy storage. However, it remains challenging to develop high-selectivity membranes that can be upscaled for industrial processes. Here we report solution-processable polymer membranes with subnanometre pores with excellent ion separation selectivity in electrodialysis processes for lithium extraction. Polymers of intrinsic microporosity incorporated with hydrophilic functional groups enable fast transport of monovalent alkali cations (Li+, Na+ and K+) while rejecting relatively larger divalent ions such as Mg2+. The polymer of intrinsic microporosity membranes surpasses the performance of most existing membrane materials. Furthermore, the membranes were scaled up and integrated into an electrodialysis stack, demonstrating excellent selectivity in simulated salt-lake brines. This work will inspire the development of selective membranes for a wide range of sustainable separation processes critical for resource recovery and a global circular economy.

Redox flow batteries (RFBs) are promising for long-duration grid-scale sustainable energy storage. The ion-exchange membrane is a key component that determines energy efficiency and cycling stability. However, it remains challenging to develop membranes with high ionic conductivity and high selectivity toward redox-active electrolytes. We report the development of ion-conductive polymer membranes with record-breaking energy efficiency. By incorporating triptycene into poly(ether-ether-ketone) and controlled sulfonation, the resulting intrinsically microporous polymer membranes form highly interconnected water channels that facilitate transport of charge-balancing ions, particularly hydroxide anions. These microporous membranes showed high ionic conductivity without compromising the selectivity toward redox-active species. The membranes enabled excellent performance in alkaline aqueous organic and zinc-iron flow batteries, demonstrating long-term stability, high power density, and an operational current density up to 700 mA cm(-2). The membranes also improved performance in neutral pH aqueous RFBs with high capacity utilization and retention, enhanced energy efficiency, and boosted power density.

One of the main aspects in which artificial capsules and cages still differ from their biological counterparts is symmetry─a fundamental trait that gives natural systems exceptional selectivity in molecular recognition. While the symmetry challenge has been recently addressed within metallosupramolecular or covalent systems, the creation of purely noncovalent capsular assemblies with tunable symmetry remains elusive. One exciting avenue toward reducing symmetry is to avail chirality in chiral-sensitive self-assembly, where symmetry is altered upon component binding. Herein, we report new pathways for the self-assembly of purely noncovalent capsules, wherein both the assembly outcome and symmetry can be finely adjusted through the strategic design of hydrogen bonding motifs and chirality. Specifically, depending on the mixture composition and hydrogen bonding centers embedded on the amino acid-derived benzene-1,3,5-triamide components, the system yields purely noncovalent capsules, ranging from highly symmetrical dimers to reduced symmetry octamers and new heterochiral tetrameric assemblies of low symmetry. The self-assembly processes presented here can be selectively modulated between narcissistic self-sorting pathways and diastereoselective social mixing, as well as toward the creation of social heterocomponent libraries─all yielding distinct capsular products. By showcasing the unique ability for molecular recognition, we believe our approach will pave the way for creating new capsular-type assemblies with unique functionalities and provide a fresh perspective on intricate, nature-inspired molecular dynamics.

Metal-organic cages (MOCs) are a class of self-assembled materials with promising applications in chemical purifications, sensing, and catalysis. Their potential is, however, hampered by challenges in the targeted design of MOCs with desirable properties. MOC discovery is thus often reliant on trial-and-error approaches and brute-force manual screening, which are time-consuming, costly and material-intensive. Translating the synthesis and property screening of MOCs to an automated workflow is therefore attractive, to both accelerate discovery and provide the datasets crucial for data-led approaches to accelerate MOC discovery and to realize their targeted properties for specific applications. Here, an automated workflow for the streamlined assembly and property screening of MOCs was developed, incorporating automated high-throughput screening of variables pertinent to MOC synthesis, data curation and automated analysis, and development of a host:guest assay to rapidly assess binding behavior. Computational modelling supplemented this automated experimental workflow for post priori rationalization of experimental outcomes. This study lays the groundwork for future large-scale MOC screening: from a relatively modest screen of 24 precursor combinations under one set of reaction conditions, 3 clean MOC species were identified, and subsequent screening of their host:guest behavior highlighted trends in binding and the identification of potential applications in molecular separations.

With the pressing urgency to reduce carbon footprint, photocatalytic carbon dioxide reduction has attracted growing attention as a sustainable mitigating option. Considering the important role of catalytic active sites (CASs) in the catalytic processes, control and design of the density and environment of CASs could enhance the catalyst performance. Herein, we report a novel metal-covalent organic framework (MCOF), MCOF-Co-315, featuring earth-abundant Co cocatalysts and conjugation through a covalently bonded backbone. MCOF-Co-315 showed a CO production rate of 1616 μmol g-1 h-1 utilizing Ru(bpy)3Cl2 as photosensitizer and triethanolamine (TEOA) as sacrificial electron donor with a 1.5 AM filter, vis mirror module (390-740 nm), and irradiation intensity adjusted to 1 sun and an especially outstanding apparent quantum yield (AQY) of 9.13% at 450 nm. The photocatalytic reaction was studied with electron paramagnetic resonance (EPR) spectroscopy, X-ray absorption near-edge structure (XANES), and in situ synchrotron Fourier Transform Infrared (FT-IR) spectroscopy, and an underlying mechanism is proposed.

Porous liquids are a sub-class of porous materials that combine permanent porosity, typically associated with solids, with the fluidity and fast mass-transfer capabilities of liquids, making them ideal candidates for gas storage and separation applications. One strategy to form porous liquids is the dissolution of discrete and permanently porous molecular species at relatively high concentrations in cavity-excluded solvents, thus introducing permanent porosity into the liquid in which it is dissolved and ensuring a solution of reasonable porosity is obtained. To access high-performance porous liquids for target applications, the selection of both the porous molecular species and the cavity-excluded solvent is key to ensuring the solvent is permanently excluded and the pore carrier is highly soluble. Finding new solvents that fit both these requirements is challenging, often resulting in a trial-and-error approach. While predictive data-driven models may be attractive, the youth of the porous liquid field currently limits the availability of data necessary to train robust models. Here, we present a computational workflow for the discovery of new porous liquid solutions combining solubility prediction software and a size-exclusivity prediction algorithm that correctly predicts size-exclusivity; this is followed by experimental validation with a representative system. Our workflow yielded size-excluded solvent and soluble porous organic cage pairs, leading to the realisation of a new porous liquid with enhanced methane uptake compared to previous systems discovered in a purely experimental high-throughput brute-force manner, highlighting the advantages of incorporating a computational workflow in the discovery of new porous liquids.

An aminal tying method was applied to post-synthetically modify a flexible organic cage, RCC1, to construct a porous ionic liquid (PIL). The resulting PIL, [RCC1-IM][NTf2]6, displayed melting behaviour below 100 °C, a transition to a glass phase on melt-quenching, CO2 uptake, and its permanent porosity was confirmed using molecular dynamic simulations.

Organic cages are a robust class of molecular hosts with a myriad of applications in materials science. Despite this, there has been a paucity of explorations into the modification of their properties via external functionalization. In this work, [n]rotaxanes featuring unoccupied organic cages as stopper components and a small 2,2'-bipyridine macrocycle were constructed using the active metal template (AMT) approach. By exploiting a scrambling methodology, it was possible to synthesize cages with a defined number of interlocked components (n = 2-4). The gas uptake, solubility, and thermal properties of the interlocked systems were compared against those of their constituent, non-interlocked components. In this manner, we were able to demonstrate the potential of exploiting the mechanical bond for modulating the physiochemical properties of these molecular materials.

Here we report a two-step, hierarchical synthesis that assembles a trigonal prismatic organic cage into a more symmetric, higher-order tetrahedral cage, or 'cage of cages'. Both the preformed [2+3] trigonal prismatic cage building blocks and the resultant tetrahedral [4[2+3]+6]cage molecule are constructed using ether bridges. This strategy affords the [4[2+3]+6]cage molecule excellent hydrolytic stability that is not a feature of more common dynamic cage linkers, such as imines. Despite its relatively high molar mass (3,001 g mol-1), [4[2+3]+6]cage exhibits good solubility and crystallizes into a porous superstructure with a surface area of 1,056 m2 g-1. By contrast, the [2+3] building block is not porous. The [4[2+3]+6]cage molecule shows high CO2 and SF6 uptakes due to its polar skeleton. The preference for the [4[2+3]+6]cage molecule over other cage products can be predicted by computational modelling, as can its porous crystal packing, suggesting a broader design strategy for the hierarchical assembly of organic cages with synthetically engineered functions. A computationally guided approach was used to predict the hierarchical assembly of four trigonal-shaped organic cage compounds into a more symmetric, higher-order, tetrahedral-shaped 'cage of cages' that crystallizes into a porous superstructure.

Historically, the chemical discovery process has predominantly been a matter of trial-and-improvement, where small modifications are made to a chemical system, guided by chemical knowledge, with the aim of optimising towards a target property or combination of properties. While a trial-and-improvement approach is frequently successful, especially when assisted by the help of serendipity, the approach is incredibly time- and resource-intensive. Complicating this further, the available chemical space that could, in theory, be explored is remarkably vast. As we are faced with near infinite possibilities and limited resources, we require improved search methods to effectively move towards desired optima, e.g. chemical systems exhibiting a target property, or several desired properties. Bayesian optimisation (BO) has recently gained significant traction in chemistry, where within the BO framework, prior knowledge is used to inform and guide the search process to optimise towards desired chemical targets, e.g. optimal reaction conditions to maximise yield, or optimal catalyst exhibiting improved catalytic activity. While powerful, implementing BO algorithms in practice is largely limited to interfacing via various APIs - requiring advanced coding experience and bespoke scripts for each optimisation task. Further, it is challenging to seamlessly link these with electronic lab notebooks via a graphical user interface (GUI). Ultimately, this limits the accessibility of BO algorithms. Here, we present Web-BO, a GUI to support BO for chemical optimisation tasks. We demonstrate its performance using an open source dataset and associated emulator, and link the platform with an existing electronic lab notebook, datalab. By providing a GUI-based BO service, we hope to improve the accessibility of data-driven optimisation tools in chemistry; https://suprashare.rcs.ic.ac.uk/web-bo/. Improving accessibility of data-driven optimisation for chemical tasks via a graphical user interface.

Polymers play a crucial role in a wide array of applications due to their diverse and tunable properties. Establishing the relationship between polymer representations and their properties is crucial to the computational design and screening of potential polymers via machine learning. The quality of the representation significantly influences the effectiveness of these computational methods. Here, we present a self-supervised contrastive learning paradigm, PolyCL, for learning robust and high-quality polymer representation without the need for labels. Our model combines explicit and implicit augmentation strategies for improved learning performance. The results demonstrate that our model achieves either better, or highly competitive, performances on transfer learning tasks as a feature extractor without an overcomplicated training strategy or hyperparameter optimisation. Further enhancing the efficacy of our model, we conducted extensive analyses on various augmentation combinations used in contrastive learning. This led to identifying the most effective combination to maximise PolyCL's performance.