Deductive Machine Learning Challenges and Opportunities in Chemical Applications.

Contemporary machine learning algorithms have largely succeeded in automating the development of mathematical models from data. Although this is a striking accomplishment, it leaves unaddressed the multitude of scenarios, especially across the chemical sciences and engineering, where deductive, rather than inductive, reasoning is required and still depends on manual intervention by an expert. This review describes the characteristics of deductive reasoning that are helpful for understanding the role played by expert intervention in problem-solving and explains why such interventions are often relatively resistant to disruption by typical machine learning strategies. The article then discusses the factors that contribute to creating a deductive bottleneck, how deductive bottlenecks are currently addressed in several application areas, and how machine learning models capable of deduction can be designed. The review concludes with a tutorial case study that illustrates the challenges of deduction problems and a notebook for readers to experiment with on their own.

Graph to Activation Energy Models Easily Reach Irreducible Errors but Show Limited Transferability.

Activation energy characterization of competing reactions is a costly but crucial step for understanding the kinetic relevance of distinct reaction pathways, product yields, and myriad other properties of reacting systems. The standard methodology for activation energy characterization has historically been a transition state search using the highest level of theory that can be afforded. However, recently, several groups have popularized the idea of predicting activation energies directly based on nothing more than the reactant and product graphs, a sufficiently complex neural network, and a broad enough data set. Here, we have revisited this task using the recently developed Reaction Graph Depth 1 (RGD1) transition state data set and several newly developed graph attention architectures. All of these new architectures achieve similar state-of-the-art results of ∼4 kcal/mol mean absolute error on withheld testing sets of reactions but poor performance on external testing sets composed of reactions with differing mechanisms, reaction molecularity, or reactant size distribution. Limited transferability is also shown to be shared by other contemporary graph to activation energy architectures through a series of case studies. We conclude that an array of standard graph architectures can already achieve results comparable to the irreducible error of available reaction data sets but that out-of-distribution performance remains poor.

Reductive Doping Inhibits the Formation of Isomerization-Derived Structural Defects in N-doped Poly(benzodifurandione) (n-PBDF).

Recently, solution-processable n-doped poly(benzodifurandione) (n-PBDF) has been made through in-situ oxidative polymerization and reductive doping, which exhibited exceptionally high electrical conductivities and optical transparency. The discovery of n-PBDF is considered a breakthrough in the field of organic semiconductors. In the initial report, the possibility of structural defect formation in n-PBDF was proposed, based on the observation of structural isomerization from (E)-2H,2'H-[3,3'-bibenzofuranylidene]-2,2'-dione (isoxindigo) to chromeno[4,3-c]chromene-5,11-dione (dibenzonaphthyrone) in the dimer model reactions. In this study, we present clear evidence that structural isomerization is inhibited during polymerization. We reveal that the dimer (BFD1) and the trimer (BFD2) can be reductively doped by several mechanisms, including hydride transfer, forming charge transfer complexes (CTC) or undergoing an integer charge transfer (ICT) with reactants available during polymerization. Once the hydride transfer adducts, the CTC, or the ICT product forms, structural isomerization can be effectively prevented even at elevated temperatures. Our findings provide a mechanistic understanding of why isomerization-derived structural defects are absent in n-PBDF backbone. It lays a solid foundation for the future development of n-PBDF as a benchmark polymer for organic electronics and beyond.

Reactive Vapor-Phase Inhibitors for Area-Selective Depositions at Tunable Critical Dimensions.

Area-selective depositions (ASD) take advantage of the chemical contrast between material surfaces in device fabrication, where a film can be selectively grown by chemical vapor deposition on metal versus a dielectric, for instance, and can provide a path to nontraditional device architectures as well as the potential to improve existing device fabrication schemes. While ASD can be accessed through a variety of methods, the incorporation of reactive moieties in inhibitors presents several advantages, such as increasing thermal stability and limiting precursor diffusion into the blocking layer. Alkyne-terminated small molecule inhibitors (SMIs)─propargyl, dipropargyl, and tripropargylamine─were evaluated as metal-selective inhibitors. Modeling these SMIs provided insight into the binding mechanism, influence of sterics, and complex polymer network formed from the reaction between inhibitors consisting of alkene, aromatic, and network branchpoints. While a significant contrast in the binding of the SMIs on copper versus a dielectric was observed, residual amounts were detected on the dielectric surfaces, leading to variable ALD growth rates dependent on pattern-critical dimensions. This behavior can be controlled and utilized to direct film growth on patterns only above a critical threshold dimension; below this threshold, both the dielectric and metal features are protected. This method provides another design parameter for ASD processes and may extend its application to broader-ranging device fabrication schemes.

Large Room-Temperature Magnetoresistance in a High-Spin Donor-Acceptor Conjugated Polymer.

Open-shell conjugated polymers (CPs) offer new opportunities for the development of emerging technologies that utilize the spin degree of freedom. Their light-element composition, weak spin-orbit coupling, synthetic modularity, high chemical stability, and solution-processability offer attributes that are unavailable from other semiconducting materials. However, developing an understanding of how electronic structure correlates with emerging transport phenomena remains central to their application. Here, the first connections between molecular, electronic, and solid-state transport in a high-spin donor-acceptor CP, poly(4-(4-(3,5-didodecylbenzylidene)-4H-cyclopenta[2,1-b:3,4-b']dithiophen-2-yl)-6,7-dimethyl-[1,2,5]-thiadiazolo[3,4-g]quinoxaline), are provided. At low temperatures (T < 180 K), a giant negative magnetoresistance (MR) is achieved in a thin-film device with a value of -98% at 10 K, which surpasses the performance of all other organic materials. The thermal depopulation of the high-spin manifold and negative MR decrease as temperature increases and at T > 180 K, the MR becomes positive with a relatively large MR of 13.5% at room temperature. Variable temperature electron paramagnetic resonance spectroscopy and magnetic susceptibility measurements demonstrate that modulation of both the sign and magnitude of the MR correlates with the electronic and spin structure of the CP. These results indicate that donor-acceptor CPs with open-shell and high-spin ground states offer new opportunities for emerging spin-based applications.

Δ2 machine learning for reaction property prediction.

The emergence of Δ-learning models, whereby machine learning (ML) is used to predict a correction to a low-level energy calculation, provides a versatile route to accelerate high-level energy evaluations at a given geometry. However, Δ-learning models are inapplicable to reaction properties like heats of reaction and activation energies that require both a high-level geometry and energy evaluation. Here, a Δ2-learning model is introduced that can predict high-level activation energies based on low-level critical-point geometries. The Δ2 model uses an atom-wise featurization typical of contemporary ML interatomic potentials (MLIPs) and is trained on a dataset of ∼167 000 reactions, using the GFN2-xTB energy and critical-point geometry as a low-level input and the B3LYP-D3/TZVP energy calculated at the B3LYP-D3/TZVP critical point as a high-level target. The excellent performance of the Δ2 model on unseen reactions demonstrates the surprising ease with which the model implicitly learns the geometric deviations between the low-level and high-level geometries that condition the activation energy prediction. The transferability of the Δ2 model is validated on several external testing sets where it shows near chemical accuracy, illustrating the benefits of combining ML models with readily available physical-based information from semi-empirical quantum chemistry calculations. Fine-tuning of the Δ2 model on a small number of Gaussian-4 calculations produced a 35% accuracy improvement over DFT activation energy predictions while retaining xTB-level cost. The Δ2 model approach proves to be an efficient strategy for accelerating chemical reaction characterization with minimal sacrifice in prediction accuracy.

Anomalous magnetoresistance in a nonconjugated radical polymer glass.

Macromolecules bearing open-shell entities offer unique transport properties for both electronic and spintronic devices. This work demonstrates that, unlike their conjugated polymer counterparts, the charge carriers in radical polymers (i.e., macromolecules with nonconjugated backbones and with stable open-shell sites present at their pendant groups) are singlet cations, which opens significant avenues for manipulating macromolecular design for advanced solid-state transport in these highly transparent conductors. Despite this key point, magnetoresistive effects are present in radical polymer thin films under applied magnetic fields due to the presence of impurity sites in low (i.e., <1%) concentrations. Additionally, thermal annealing of poly(4-glycidyloxy-2,2,6,6- tetramethylpiperidine-1-oxyl) (PTEO), a nonconjugated polymer with stable open-shell pendant groups, facilitated better electron exchange and pairwise spin interactions resulting in an unexpected magnetoresistance signal at relatively low field strengths (i.e., <2 T). The addition of 4-hydroxy-2,2,6,6-tetramethylpiperidin-N-oxy (TEMPO-OH), a paramagnetic species, increased the magnitude of the MR effect when the small molecule was added to the radical polymer matrix. These macroscopic experimental observables are explained using computational approaches that detail the fundamental molecular principles. This intrinsic localized charge transport behavior differs from the current state of the art regarding closed-shell conjugated macromolecules, and it opens an avenue towards next-generation transport in organic electronic materials.

Emerging Two-Dimensional Organic Semiconductor-Incorporated Perovskites─A Fascinating Family of Hybrid Electronic Materials.

Halide perovskites have attracted a great amount of attention owing to their unique materials chemistry, excellent electronic properties, and low-cost manufacturing. Two-dimensional (2D) halide perovskites, originating from three-dimensional (3D) perovskite structures, are structurally more diverse and therefore create functional possibilities beyond 3D perovskites. The much less restrictive size constraints on the organic component of these hybrid materials particularly provide an exciting platform for designing unprecedented materials and functionalities at the molecular level. In this Perspective, we discuss the concept and recent development of a sub-class of 2D perovskites, namely, organic semiconductor-incorporated perovskites (OSiPs). OSiPs combine the electronic functionality of organic semiconductors with the soft and dynamic halide perovskite lattice, offering opportunities for tailoring the energy landscape, lattice and carrier dynamics, and electron/ion transport properties for various fundamental studies, as well as device applications. Specifically, we summarize recent advances in the design, synthesis, and structural analysis of OSiPs with various organic conjugated moieties as well as the application of OSiPs in photovoltaics, light-emitting devices, and transistors. Lastly, challenges and further opportunities for OSiPs in molecular design, integration of novel functionality, film quality, and stability issues are addressed.

Deep reaction network exploration of glucose pyrolysis.

Resolving the reaction networks associated with biomass pyrolysis is central to understanding product selectivity and aiding catalyst design to produce more valuable products. However, even the pyrolysis network of relatively simple [Formula: see text]-D-glucose remains unresolved due to its significant complexity in terms of the depth of the network and the number of major products. Here, a transition-state-guided reaction exploration has been performed that provides complete pathways to most significant experimental pyrolysis products of [Formula: see text]-D-glucose. The resulting reaction network involves over 31,000 reactions and transition states computed at the semiempirical quantum chemistry level and approximately 7,000 kinetically relevant reactions and transition states characterized with density function theory, comprising the largest reaction network reported for biomass pyrolysis. The exploration was conducted using graph-based rules to explore the reactivities of intermediates and an adaption of the Dijkstra algorithm to identify kinetically relevant intermediates. This simple exploration policy surprisingly (re)identified pathways to most major experimental pyrolysis products, many intermediates proposed by previous computational studies, and also identified new low-barrier reaction mechanisms that resolve outstanding discrepancies between reaction pathways and yields in isotope labeling experiments. This network also provides explanatory pathways for the high yield of hydroxymethylfurfural and the reaction pathway that contributes most to the formation of hydroxyacetaldehyde during glucose pyrolysis. Due to the limited domain knowledge required to generate this network, this approach should also be transferable to other complex reaction network prediction problems in biomass pyrolysis.

Δ-Learning applied to coarse-grained homogeneous liquids.

Coarse-grained molecular dynamics (CGMD) simulations address lengthscales and timescales that are critical to many chemical and material applications. Nevertheless, contemporary CGMD modeling is relatively bespoke and there are no black-box CGMD methodologies available that could play a comparable role in discovery applications that density functional theory plays for electronic structure. This gap might be filled by machine learning (ML)-based CGMD potentials that simplify model development, but these methods are still in their early stages and have yet to demonstrate a significant advantage over existing physics-based CGMD methods. Here, we explore the potential of Δ-learning models to leverage the advantages of these two approaches. This is implemented by using ML-based potentials to learn the difference between the target CGMD variable and the predictions of physics-based potentials. The Δ-models are benchmarked against the baseline models in reproducing on-target and off-target atomistic properties as a function of CG resolution, mapping operator, and system topology. The Δ-models outperform the reference ML-only CGMD models in nearly all scenarios. In several cases, the ML-only models manage to minimize training errors while still producing qualitatively incorrect dynamics, which is corrected by the Δ-models. Given their negligible added cost, Δ-models provide essentially free gains over their ML-only counterparts. Nevertheless, an unexpected finding is that neither the Δ-learning models nor the ML-only models significantly outperform the elementary pairwise models in reproducing atomistic properties. This fundamental failure is attributed to the relatively large irreducible force errors associated with coarse-graining that produces little benefit from using more complex potentials.

Enhanced Electron Transport in Nonconjugated Radical Oligomers Occurs by Tunneling.

Incorporating temperature- and air-stable organic radical species into molecular designs is a potentially advantageous means of controlling the properties of electronic materials. However, we still lack a complete understanding of the structure-property relationships of organic radical species at the molecular level. In this work, the charge transport properties of (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO) radical-containing nonconjugated molecules are studied using single-molecule charge transport experiments and molecular modeling. Importantly, the TEMPO pendant groups promote temperature-independent molecular charge transport in the tunneling region relative to the quenched and closed-shell phenyl pendant groups. Results from molecular modeling show that the TEMPO radicals interact with the gold metal electrodes near the interface to facilitate a high-conductance conformation. Overall, the large enhancement of charge transport by incorporation of open-shell species into a single nonconjugated molecular component opens exciting avenues for implementing molecular engineering in the development of next-generation electronic devices based on novel nonconjugated radical materials.

Design Rules for Two-Dimensional Organic Semiconductor-Incorporated Perovskites (OSiP) Gleaned from Thousands of Simulated Structures.

Two-dimensional (2D) halide perovskites are an attractive class of hybrid perovskites that have additional optoelectronic tunability due to their accommodation of relatively large organic ligands. Nevertheless, contemporary ligand design depends on either expensive trial-and-error testing of whether a ligand can be integrated within the lattice or conservative heuristics that unduly limit the scope of ligand chemistries. Here, the structural determinants of stable ligand incorporation within Ruddlesden-Popper (RP) phase perovskites are established by molecular dynamics (MD) simulations of over ten-thousand RP-phase perovskites and training of machine learning classifiers capable of predicting structural stability based solely on generalizable ligand features. The simulation results show near-perfect predictions of positive and negative literature examples, predict trade-offs between several ligand features and stability, and ultimately predict an inexhaustibly large 2D-compatible ligand design-space.

Bridging the Monomer to Polymer Gap in Radical Polymer Design.

Radical polymers bearing open-shell moieties at pendant sites exhibit unique redox and optoelectronic properties that are promising for many organic electronic applications. Nevertheless, gaps remain in relating the electronic properties of repeat units, which can be easily calculated, to the condensed-phase charge transport behaviors of these materials. To address this gap, we have performed the first quantum chemical study on a broad swathe of radical polymer design space that explicitly includes the coupling between polymer constraints and radical-mediated intramolecular charge transfer. For this purpose, a chemical space of 64 radical polymer chemistries was constructed based on varying backbone units, open-shell chemistries, and spacer units between the backbone and the radical groups. For each combination of backbone, radical, and spacer, comprehensive conformational sampling was used to calculate expected values of intrachain charge transport using several complementary metrics, including the end-to-end thermal Green's function, Delta-Wye transformed inverse resistance, and the Kirchhoff transport index. We observe that charge transport in radical polymers is primarily driven by the choice radical chemistry, which influences the optimal choice of backbone chemistry and spacer group that mediate radical alignment and avoid the formation of undesired trap states. Given the limited exploration of radical chemistries beyond the TEMPO radical for this class of materials, these findings suggest tremendous opportunities exist for synthetic exploration in radical polymers.

Comprehensive exploration of graphically defined reaction spaces.

Existing reaction transition state (TS) databases are comparatively small and lack chemical diversity. Here, this data gap has been addressed using the concept of a graphically-defined model reaction to comprehensively characterize a reaction space associated with C, H, O, and N containing molecules with up to 10 heavy (non-hydrogen) atoms. The resulting dataset is composed of 176,992 organic reactions possessing at least one validated TS, activation energy, heat of reaction, reactant and product geometries, frequencies, and atom-mapping. For 33,032 reactions, more than one TS was discovered by conformational sampling, allowing conformational errors in TS prediction to be assessed. Data is supplied at the GFN2-xTB and B3LYP-D3/TZVP levels of theory. A subset of reactions were recalculated at the CCSD(T)-F12/cc-pVDZ-F12 and ωB97X-D2/def2-TZVP levels to establish relative errors. The resulting collection of reactions and properties are called the Reaction Graph Depth 1 (RGD1) dataset. RGD1 represents the largest and most chemically diverse TS dataset published to date and should find immediate use in developing novel machine learning models for predicting reaction properties.

Unraveling two distinct polymorph transition mechanisms in one n-type single crystal for dynamic electronics.

Cooperativity is used by living systems to circumvent energetic and entropic barriers to yield highly efficient molecular processes. Cooperative structural transitions involve the concerted displacement of molecules in a crystalline material, as opposed to typical molecule-by-molecule nucleation and growth mechanisms which often break single crystallinity. Cooperative transitions have acquired much attention for low transition barriers, ultrafast kinetics, and structural reversibility. However, cooperative transitions are rare in molecular crystals and their origin is poorly understood. Crystals of 2-dimensional quinoidal terthiophene (2DQTT-o-B), a high-performance n-type organic semiconductor, demonstrate two distinct thermally activated phase transitions following these mechanisms. Here we show reorientation of the alkyl side chains triggers cooperative behavior, tilting the molecules like dominos. Whereas, nucleation and growth transition is coincident with increasing alkyl chain disorder and driven by forming a biradical state. We establish alkyl chain engineering as integral to rationally controlling these polymorphic behaviors for novel electronic applications.

Tailoring Molecular-Scale Contact at the Perovskite/Polymeric Hole-Transporting Material Interface for Efficient Solar Cells.

Perovskite solar cells (PSCs) have delivered a power conversion efficiency (PCE) of more than 25% and incorporating polymers as hole-transporting layers (HTLs) can further enhance the stability of devices toward the goal of commercialization. Among the various polymeric hole-transporting materials, poly(triaryl amine) (PTAA) is one of the promising HTL candidates with good stability; however, the hydrophobicity of PTAA causes problematic interfacial contact with the perovskite, limiting the device performance. Using molecular side-chain engineering, a uniform 2D perovskite interlayer with conjugated ligands, between 3D perovskites and PTAA is successfully constructed. Further, employing conjugated ligands as cohesive elements, perovskite/PTAA interfacial adhesion is significantly improved. As a result, the thin and lateral extended 2D/3D heterostructure enables as-fabricated PTAA-based PSCs to achieve a PCE of 23.7%, improved from the 18% of reference devices. Owing to the increased ion-migration energy barrier and conformal 2D coating, unencapsulated devices with the new ligands exhibit both superior thermal stability under 60 °C heating and moisture stability in ambient conditions.

Thermally Accessible Prebiotic Pathways for Forming Ribonucleic Acid and Protein Precursors from Aqueous Hydrogen Cyanide.

The search for prebiotic chemical pathways to biologically relevant molecules is a long-standing puzzle that has generated a menagerie of competing hypotheses with limited experimental prospects for falsification. However, the advent of computational network exploration methodologies has created the opportunity to compare the kinetic plausibility of various channels and even propose new pathways. Here, the space of organic molecules that can be formed within four polar or pericyclic reactions from water and hydrogen cyanide (HCN), two established prebiotic candidates for generating biological precursors, was comprehensively explored with a state-of-the-art exploration algorithm. A surprisingly diverse reactivity landscape was revealed within just a few steps of these simple molecules. Reaction pathways to several biologically relevant molecules were discovered involving lower activation energies and fewer reaction steps compared with recently proposed alternatives. Accounting for water-catalyzed reactions qualitatively affects the interpretation of the network kinetics. The case-study also highlights omissions of simpler and lower barrier reaction pathways to certain products by other algorithms that qualitatively affect the interpretation of HCN reactivity.

Evidence That Less Can Be More for Transferable Force Fields.

Graph-based parameter assignment has been the basis for developing transferable force fields for molecular dynamics simulations for decades. Nevertheless, transferable force fields vary in how specifically terms are defined with respect to the molecular graph and the procedures for generating parametrization data. More-specific force-field terms increase the complexity of the force field, theoretically increasing accuracy but also increasing training data requirements. In contrast, less-specific force fields can be reused across larger regions of chemical space, theoretically reducing accuracy but also reducing the number of parameters and training data requirements. Here, the tradeoffs between force-field specificity and accuracy are quantified by parametrizing three new sets of force fields with varying levels of graph specificity, using a shared procedure for generating training data. These force fields are benchmarked for their ability to reproduce the structural features and liquid properties of 87 organic molecules at 146 distinct state points. The overall accuracy for properties that were directly trained on rapidly saturates as the graph specificity of the force-field increases. From this, we conclude there is at best a marginal benefit of using less transferable and more complex force fields with common sources of quantum-chemically derived training data. When looking at properties unseen during training, there is some evidence that the more-complex force fields even perform slightly worse. These results are rationalized by the fortuitous regularization of force fields based on less-specific and more-transferable atom types. Both the saturation in the accuracy of training properties and the marginally worse performance on off-target properties fundamentally contradict the expectation that bespoke force fields are generally more accurate, given their larger number of parameters, and suggests that increasing force-field complexity should be carefully justified against performance gains and balanced against available training data.