Electron density-based GPT for optimization and suggestion of host-guest binders.

Here we present a machine learning model trained on electron density for the production of host-guest binders. These are read out as simplified molecular-input line-entry system (SMILES) format with >98% accuracy, enabling a complete characterization of the molecules in two dimensions. Our model generates three-dimensional representations of the electron density and electrostatic potentials of host-guest systems using a variational autoencoder, and then utilizes these representations to optimize the generation of guests via gradient descent. Finally the guests are converted to SMILES using a transformer. The successful practical application of our model to established molecular host systems, cucurbit[n]uril and metal-organic cages, resulted in the discovery of 9 previously validated guests for CB[6] and 7 unreported guests (with association constant Ka ranging from 13.5 M-1 to 5,470 M-1) and the discovery of 4 unreported guests for [Pd214]4+ (with Ka ranging from 44 M-1 to 529 M-1).

An integrated self-optimizing programmable chemical synthesis and reaction engine.

Robotic platforms for chemistry are developing rapidly but most systems are not currently able to adapt to changing circumstances in real-time. We present a dynamically programmable system capable of making, optimizing, and discovering new molecules which utilizes seven sensors that continuously monitor the reaction. By developing a dynamic programming language, we demonstrate the 10-fold scale-up of a highly exothermic oxidation reaction, end point detection, as well as detecting critical hardware failures. We also show how the use of in-line spectroscopy such as HPLC, Raman, and NMR can be used for closed-loop optimization of reactions, exemplified using Van Leusen oxazole synthesis, a four-component Ugi condensation and manganese-catalysed epoxidation reactions, as well as two previously unreported reactions, discovered from a selected chemical space, providing up to 50% yield improvement over 25-50 iterations. Finally, we demonstrate an experimental pipeline to explore a trifluoromethylations reaction space, that discovers new molecules.

Exploring and mapping chemical space with molecular assembly trees.

The rule-based search of chemical space can generate an almost infinite number of molecules, but exploration of known molecules as a function of the minimum number of steps needed to build up the target graphs promises to uncover new motifs and transformations. Assembly theory is an approach to compare the intrinsic complexity and properties of molecules by the minimum number of steps needed to build up the target graphs. Here, we apply this approach to prebiotic chemistry, gene sequences, plasticizers, and opiates. This allows us to explore molecules connected to the assembly tree, rather than the entire space of molecules possible. Last, by developing a reassembly method, based on assembly trees, we found that in the case of the opiates, a new set of drug candidates could be generated that would not be accessible via conventional fragment-based drug design, thereby demonstrating how this approach might find application in drug discovery.

Modeling UV-Vis spectra of low dimensional materials using electrostatic embedding: The case of CdSe.

We present a generalization of a self-consistent electrostatic embedding approach (SC-Ewald) devised to investigate the photophysical properties of 3D periodic materials, to systems in one- or two-dimensional (2D) reduced periodicity. In this approach, calculations are carried out on a small finite molecular cluster extracted from a periodic model, while the crystalline environment is accounted for by an array of point charges which are fitted to reproduce the exact electrostatic potential (at ground or the excited state) of the infinite periodic system. Periodic density functional theory (DFT) calculations are combined with time dependent DFT calculations to simulate absorption and emission properties of the extended system under investigation. We apply this method to compute the UV-Vis. spectra of bulk and quantum-confined 0D quantum dots and 2D extended nanoplatelets of CdSe, due to their relevance as sensitizers in solar cells technologies. The influence of the size and shape of the finite cluster model chosen in the excited state calculations was also investigated and revealed that, although the long-range electrostatics of the environment are important for the calculation of the UV-Vis, a subtle balance between short- and long-range effects exists. These encouraging results demonstrate that this self-consistent electrostatic embedding approach, when applied in different dimensions, can successfully model the photophysical properties of diverse material classes, making it an attractive low-cost alternative to far more computationally demanding electronic structure methods for excited state calculations.

Digitizing Chemistry Using the Chemical Processing Unit: From Synthesis to Discovery.

The digitization of chemistry is not simply about using machine learning or artificial intelligence systems to process chemical data, or about the development of ever more capable automation hardware; instead, it is the creation of a hard link between an abstracted process ontology of chemistry and bespoke hardware for performing reactions or exploring reactivity. Chemical digitization is therefore about the unambiguous development of an architecture, a chemical state machine, that uses this ontology to connect precise instruction sets to hardware that performs chemical transformations. This approach enables a universal standard for describing chemistry procedures via a chemical programming language and facilitates unambiguous dissemination of these procedures. We predict that this standard will revolutionize the ability of chemists to collaborate, increase reproducibility and safety, as we all as optimize for cost and efficiency. Most importantly, the digitization of chemistry will dramatically reduce the labor needed to make new compounds and broaden accessible chemical space. In recent years, the developments of automation in chemistry have gone beyond flow chemistry alone, with many bespoke workflows being developed not only for automating chemical synthesis but also for materials, nanomaterials, and formulation production. Indeed, the leap from fixed-configuration synthesis machines like peptide, nucleic acid, or dedicated cross-coupling engines is important for developing a truly universal approach to "dial-a-molecule". In this case, a key conceptual leap is the use of a batch system that can encode the chemical reagents, solvent, and products as packets which can be moved around the system, and a graph-based approach for the description of hardware modules that allows the compilation of chemical code that runs on, in principle, any hardware. Further, the integration of sensor systems for monitoring and controlling the state of the chemical synthesis machine, as well as high resolution spectroscopic tools, is vital if these systems are to facilitate closed-loop autonomous experiments. Systems that not only make molecules and materials, but also optimize their function, and use algorithms to assist with the development of new synthetic pathways and process optimization are also possible. Here, we discuss how the digitization of chemistry is happening, building on the plethora of technological developments in hardware and software. Importantly, digital-chemical robot systems need to integrate feedback from simple sensors, e.g., conductivity or temperature, as well as online analytics in order to navigate process space autonomously. This will open the door to accessing known molecules (synthesis), exploring whether known compounds/reactions are possible under new conditions (optimization), and searching chemical space for unknown and unexpected new molecules, reactions, and modes of reactivity (discovery). We will also discuss the role of chemical knowledge and how this can be used to challenge bias, as well as define and expand synthetically accessible chemical space using programmable robotic chemical state machines.

Electron Storage System Based on a Two-Way Inversion of Redox Potentials.

Molecular-level multielectron handling toward electrical storage is a worthwhile approach to solar energy harvesting. Here, a strategy which uses chemical bonds as electron reservoirs is introduced to demonstrate the new concept of "structronics" (a neologism derived from "structure" and "electronics"). Through this concept, we establish, synthesize, and thoroughly study two multicomponent "super-electrophores": 1,8-dipyridyliumnaphthalene, 2, and its N,N-bridged cyclophane-like analogue, 3. Within both of them, a covalent bond can be formed and subsequently broken electrochemically. These superelectrophores are based on two electrophoric (pyridinium) units that are, on purpose, spatially arranged by a naphthalene scaffold. A key characteristic of 2 and 3 is that they possess a LUMO that develops through space as the result of the interaction between the closely positioned electrophoric units. In the context of electron storage, this "super-LUMO" serves as an empty reservoir, which can be filled by a two-electron reduction, giving rise to an elongated C-C bond or "super-HOMO". Because of its weakened nature, this bond can undergo an electrochemically driven cleavage at a significantly more anodic-yet accessible-potential, thereby restoring the availability of the electron pair (reservoir emptying). In the representative case study of 2, an inversion of potential in both of the two-electron processes of bond formation and bond-cleavage is demonstrated. Overall, the structronic function is characterized by an electrochemical hysteresis and a chemical reversibility. This structronic superelectrophore can be viewed as the three-dimensional counterpart of benchmark methyl viologen (MV).

Mapping the optoelectronic property space of small aromatic molecules.

Small aromatic molecules and their quinone derivatives find use in organic transistors, solar-cells, thermoelectrics, batteries and photocatalysts. These applications exploit the optoelectronic properties of these molecules and the ease by which such properties can be tuned by the introduction of heteroatoms and/or the addition of functional groups. We perform a high-throughput virtual screening using the xTB family of density functional tight-binding methods to map the optoelectronic property space of ~250,000 molecules. The large volume of data generated allows for a broad understanding of how the presence of heteroatoms and functional groups affect the ionisation potential, electron affinity and optical gap values of these molecular semiconductors, and how the structural features - on their own or in combination with one another - allow access to particular regions of the optoelectronic property space. Finally, we identify the apparent boundaries of the optoelectronic property space for these molecules: regions of property space that appear off limits for any small aromatic molecule.

Insight into the self-assembly of water-soluble perylene bisimide derivatives through a combined computational and experimental approach.

We use a combination of computational and experimental techniques to study the self-assembly and gelation of water-soluble perylene bisimides derivatised at the imide position with an amino acid. Specifically, we study the likely structure of self-assembled aggregates of the alanine-functionalised perylene bisimide (PBI-A) and the thermodynamics of their formation using density functional theory and predict the UV-vis spectra of such aggregates using time-dependent density functional theory. We compare these predictions to experiments in which we study the evolution of the UV-Vis and NMR spectra and the rheology and neutron scattering of alkaline PBI-A solutions when gradually decreasing the pH. Based on the combined computational and experimental results, we show that PBI-A self-assembles at all pH values but that aggregates grow in size upon protonation. Hydrogel formation is driven not by aggregate growth but reduction of the aggregation surface-charge and a decrease in the colloidal stability of the aggregation with respect to agglomeration.

Mapping binary copolymer property space with neural networks.

The extremely large number of unique polymer compositions that can be achieved through copolymerisation makes it an attractive strategy for tuning their optoelectronic properties. However, this same attribute also makes it challenging to explore the resulting property space and understand the range of properties that can be realised. In an effort to enable the rapid exploration of this space in the case of binary copolymers, we train a neural network using a tiered data generation strategy to accurately predict the optical and electronic properties of 350 000 binary copolymers that are, in principle, synthesizable from their dihalogen monomers via Yamamoto, or Suzuki-Miyaura and Stille coupling after one-step functionalisation. By extracting general features of this property space that would otherwise be obscured in smaller datasets, we identify simple models that effectively relate the properties of these copolymers to the homopolymers of their constituent monomers, and challenge common ideas behind copolymer design. We find that binary copolymerisation does not appear to allow access to regions of the optoelectronic property space that are not already sampled by the homopolymers, although it conceptually allows for more fine-grained property control. Using the large volume of data available, we test the hypothesis that copolymerisation of 'donor' and 'acceptor' monomers can result in copolymers with a lower optical gap than their related homopolymers. Overall, despite the prevalence of this concept in the literature, we observe that this phenomenon is relatively rare, and propose conditions that greatly enhance the likelihood of its experimental realisation. Finally, through a 'topographical' analysis of the co-polymer property space, we show how this large volume of data can be used to identify dominant monomers in specific regions of property space that may be amenable to a variety of applications, such as organic photovoltaics, light emitting diodes, and thermoelectrics.

Accelerated Discovery of Organic Polymer Photocatalysts for Hydrogen Evolution from Water through the Integration of Experiment and Theory.

Conjugated polymers are an emerging class of photocatalysts for hydrogen production where the large breadth of potential synthetic diversity presents both an opportunity and a challenge. Here, we integrate robotic experimentation with high-throughput computation to navigate the available structure-property space. A total of 6354 co-polymers was considered computationally, followed by the synthesis and photocatalytic characterization of a sub-library of more than 170 co-polymers. This led to the discovery of new polymers with sacrificial hydrogen evolution rates (HERs) of more than 6 mmol g-1 h-1. The variation in HER across the library does not correlate strongly with any single physical property, but a machine-learning model involving four separate properties can successfully describe up to 68% of the variation in the HER data between the different polymers. The four variables used in the model were the predicted electron affinity, the predicted ionization potential, the optical gap, and the dispersibility of the polymer particles in solution, as measured by optical transmittance.

Photocatalytically active ladder polymers.

Conjugated ladder polymers (cLaPs) are introduced as organic semiconductors for photocatalytic hydrogen evolution from water under sacrificial conditions. Starting from a linear conjugated polymer (cLiP1), two ladder polymers are synthesized via post-polymerization annulation and oxidation techniques to generate rigidified, planarized materials bearing dibenzo[b,d]thiophene (cLaP1) and dibenzo[b,d]thiophene sulfone subunits (cLaP2). The high photocatalytic activity of cLaP1 (1307 µmol h-1 g-1) in comparison to that of cLaP2 (18 µmol h-1 g-1) under broadband illumination (λ > 295 nm) in the presence of a hole-scavenger is attributed to a higher yield of long-lived charges (µs to ms timescale), as evidenced by transient absorption spectroscopy. Additionally, cLaP1 has a larger overpotential for proton reduction and thus an increased driving force for the evolution of hydrogen under sacrificial conditions.

Computational high-throughput screening of polymeric photocatalysts: exploring the effect of composition, sequence isomerism and conformational degrees of freedom.

We discuss a low-cost computational workflow for the high-throughput screening of polymeric photocatalysts and demonstrate its utility by applying it to a number of challenging problems that would be difficult to tackle otherwise. Specifically we show how having access to a low-cost method allows one to screen a vast chemical space, as well as to probe the effects of conformational degrees of freedom and sequence isomerism. Finally, we discuss both the opportunities of computational screening in the search for polymer photocatalysts, as well as the biggest challenges.

Emissive Azobenzenes Delivered on a Silver Coordination Polymer.

Azobenzene has become a ubiquitous component of functional molecules and polymeric materials because of the light-induced trans → cis isomerization of the diazene group. In contrast, there are very few applications utilizing azobenzene luminescence, since the excitation energy typically dissipates via nonradiative pathways. Inspired by our earlier studies with 2,2'-bis[ N,N'-(2-pyridyl)methyl]diaminoazobenzene (AzoAM oP) and related compounds, we investigated a series of five aminoazobenzene derivatives and their corresponding silver complexes. Four of the aminoazobenzene ligands, which exhibit no emission under ambient conditions, form silver coordination polymers that are luminescent at room temperature. AzoAE pP (2,2'-bis[ N,N'-(4-pyridyl)ethyl]diaminoazobenzene) assembles into a three-dimensional coordination polymer (AgAAE pP) that undergoes a reversible loss of emission upon the addition of metal-coordinating analytes such as pyridine. The switching behavior is consistent with the disassembly and reassembly of the coordination polymer driven by displacement of the aminoazobenzene ligands by coordinating analytes.

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.

Revealing the Origins of Mechanically Induced Fluorescence Changes in Organic Molecular Crystals.

Mechanofluorochromic molecular materials display a change in fluorescence color through mechanical stress. Complex structure-property relationships in both the crystalline and amorphous phases of these materials govern both the presence and strength of this behavior, which is usually deemed the result of a mechanically induced phase transition. However, the precise nature of the emitting species in each phase is often a matter of speculation, resulting from experimental data that are difficult to interpret, and a lack of an acceptable theoretical model capable of capturing complex environmental effects. With a combined strategy using sophisticated experimental techniques and a new theoretical approach, here the varied mechanofluorochromic behavior of a series of difluoroboron diketonates is shown to be driven by the formation of low-energy exciton traps in the amorphous phase, with a limited number of traps giving rise to the full change in fluorescence color. The results highlight intrinsic structural links between crystalline and amorphous phases, and how these may be exploited for further development of powerful mechanofluorochromic assemblies, in line with modern crystal engineering approaches.

Communication: Evaluating non-empirical double hybrid functionals for spin-state energetics in transition-metal complexes.

The computationally assisted, accelerated design of inorganic functional materials often relies on the ability of a given electronic structure method to return the correct electronic ground state of the material in question. Outlining difficulties with current density functionals and wave function-based approaches, we highlight why double hybrid density functionals represent promising candidates for this purpose. In turn, we show that PBE0-DH (and PBE-QIDH) offers a significant improvement over its hybrid parent functional PBE0 [as well as B3LYP* and coupled cluster singles and doubles with perturbative triples (CCSD(T))] when computing spin-state splitting energies, using high-level diffusion Monte Carlo calculations as a reference. We refer to the opposing influence of Hartree-Fock (HF) exchange and MP2, which permits higher levels of HF exchange and a concomitant reduction in electronic density error, as the reason for the improved performance of double-hybrid functionals relative to hybrid functionals. Additionally, using 16 transition metal (Fe and Co) complexes, we show that low-spin states are stabilised by increasing contributions from MP2 within the double hybrid formulation. Furthermore, this stabilisation effect is more prominent for high field strength ligands than low field strength ligands.

Multiconfiguration Pair-Density Functional Theory Predicts Spin-State Ordering in Iron Complexes with the Same Accuracy as Complete Active Space Second-Order Perturbation Theory at a Significantly Reduced Computational Cost.

The spin-state orderings in nine Fe(II) and Fe(III) complexes with ligands of diverse ligand-field strength were investigated with multiconfiguration pair-density functional theory (MC-PDFT). The performance of this method was compared to that of complete active space second-order perturbation theory (CASPT2) and Kohn-Sham density functional theory. We also investigated the dependence of CASPT2 and MC-PDFT results on the size of the active-space. MC-PDFT reproduces the CASPT2 spin-state ordering, the dependence on the ligand field strength, and the dependence on active space at a computational cost that is significantly reduced as compared to CASPT2.

Modelling photophysical properties of metal-organic frameworks: a density functional theory based approach.

Design of optical properties within metal-organic frameworks (MOFs) is a subject of ever increasing attention in recent years with theoretical approaches poised to play a key role alongside experiment in both the understanding of fundamental mechanisms and the further development of high performance materials. We have developed and applied a simple and computationally affordable protocol rooted in density functional theory (DFT) and its time dependent counterpart (TD-DFT) to two isostructural MOFs based on a 4,4'-bis((3,5-dimethyl-1H-pyrazol-4-yl)methyl)-biphenyl (H2DMPMB) linker. These systems show a remarkable dependence of photoluminescence properties on the interchange of zinc and cadmium cations as building units. Our investigation was able to successfully rationalize the subtle change in the photoluminescence mechanism experimentally observed responsible for the large (0.88 eV) red shift (from 335 nm to 441 nm) observed when going from the cadmium to the zinc based structure. More generally, this computational protocol seems well adapted for the characterization and rationalization of the absorption and emission behaviour of such complex extended materials.

Electrostatic Embedding To Model the Impact of Environment on Photophysical Properties of Molecular Crystals: A Self-Consistent Charge Adjustment Procedure.

A case study of 1,8-dihydroxy-2-napthaldehyde (DHNA)-exhibiting an excited-state intramolecular double proton transfer resulting in photophysical properties sensitive to the surrounding environment-has been used to assess the performance of electrostatic embedding approaches designed to accurately recover the effects of a bulk crystalline environment on calculated photophysical properties. The first approach, based on time-dependent density functional theory (TD-DFT) applied in a QM/QM' scheme, makes use of a background point charge distribution which can accurately reproduce the exact ground-state Ewald potential of the bulk crystal. The second approach seeks to "optimize" these charges in a self-consistent manner in order to reproduce the electrostatic field produced by the environment at the excited state. Using these two approaches, both absorption and emission properties of molecular crystals, such as the position and the relative shift in the emission bands in the solid state with respect to solution, can be accurately reproduced. More generally, the results obtained show how these computationally affordable approaches can be used to predict the excited-state behavior of molecules in condensed phases, thus allowing their employment to predict or design new molecular materials with enhanced photophysical properties.

Describing excited state intramolecular proton transfer in dual emissive systems: a density functional theory based analysis.

The excited state intramolecular proton transfer (ESIPT) reaction taking place within 2-(2-hydroxyphenyl)benzoxazole (HBT) and two recently experimentally characterized napthalimide derivatives-known as N-1 and N-4-has been investigated in order to identify and test a possible protocol for the description and complete mechanistic and electronic characterization of the reaction at the excited state. This protocol is based on density functional theory, time-dependent density functional theory, and a recently proposed electron density based index (DCT). This method is able to identify all stable species involved in the reaction, discriminate between possible reaction pathways over potential energy surfaces (PES), which are intrinsically very flat and difficult to characterize, and quantitatively measure the excited state charge transfer character throughout the reaction. The photophysical properties of the molecules (i.e., absorption and emission wavelength) are also quantitatively determined via the implicit inclusion of solvent effects in the case of toluene and, the more polar, tetrahydrofuran. The accuracy obtained with this protocol then opens up the possibility of the ab initio design of molecules exhibiting ESIPT for tailored applications such as highly selective molecular sensors.