Bao N. Nguyen

The increasing use of AI in chemical research requires reliable reaction data to support efficient synthetic route design and reaction discovery. We present OPRD-100, an expert-curated and cross-validated dataset of 3,878 reactions extracted from 100 Organic Process Research & Development articles published between 1999 and 2024. Each entry captures detailed reactivity information, including yield, stereochemistry, temperature, time, solvents, and reagents. Analysis of the dataset reveals the relative importance of reaction data reported in tables versus text—particularly for reactions with low or no yields—and highlights the challenges of consistently capturing information from complex studies. Comparison with Reaxys showed systematic omissions and a strong bias toward high-yield reactions, underscoring the risks of relying on uncurated data. OPRD-100 thus provides a robust benchmark for training and evaluating AI tools in industrially relevant organic synthesis. Reaction data, data mining, benchmarking dataset

We present graphRC, a method for rapid transition state (TS) mode analysis using internal coordinates derived from molecular graphs. The imaginary mode of a TS describes the direction of atomic motion at the saddle point, providing a local approximation to the reaction coordinate, while Intrinsic Reaction Coordinate (IRC) and Quick Reaction Coordinate (QRC) calculations trace the full pathway to adjacent minima. In all cases, displacements are expressed in Cartesian coordinates and do not directly describe changes in bonding. By translating these into internal coordinate changes (bonds, angles, and dihedrals), graphRC provides chemical insight into the TS mode and reaction coordinate trajectories without prior knowledge of reactant and product structures. Molecular connectivity is determined using xyzgraph, a flexible graph builder validated across 4846 structures spanning 61 elements and 490 element-pair bond types, with close agreement to DFT-derived bonding. Initial validation on 16 diverse TS achieved 100% identification of bond changes, rotations, and inversions, with zero false positives compared to IRC and QRC connectivity. Across 395 TS covering organic, organometallic, and catalytic transformations, normal-mode analysis alone detects the primary bond change in every case, with high agreement to IRC-derived connectivity. This enables programmatic TS verification at a fraction of the cost of formal reaction coordinate calculations, complementing more rigorous methods with rapid, interpretable analysis.

We present graphRC, a method for rapid transition state (TS) mode analysis using internal coordinates derived from molecular graphs. The imaginary mode of a TS describes the direction of atomic motion at the saddle point, providing a local approximation to the reaction coordinate, while Intrinsic Reaction Coordinate (IRC) and Quick Reaction Coordinate (QRC) calculations trace the full pathway to adjacent minima. In all cases, displacements are expressed in Cartesian coordinates and do not directly describe changes in bonding. By translating these into internal coordinate changes (bonds, angles, dihedrals), graphRC provides chemical insight into the TS mode and reaction coordinate trajectories without prior knowledge of reactant and product structures. Molecular connectivity is determined using xyzgraph, a flexible graph builder validated across 4846 structures spanning 61 elements and 490 element-pair bond types with close agreement to DFT-derived bonding. Initial validation on 16 diverse TS achieved 100% identification of bond changes, rotations and inversions with zero false positives compared to IRC and QRC connectivity. Across 395 TS covering organic, organometallic and catalytic transformations, normal mode analysis alone detects the primary bond change in every case, with high agreement to IRC-derived connectivity. This enables programmatic TS verification at a fraction of the cost of formal reaction coordinate calculations, complementing more rigorous methods with rapid, interpretable analysis.

DFT benchmarking on 127 base metal complexes revealed unreliable spin state predictions by B3LYP, PBE0, TPSSh and M06-L functionals, especially when energy gaps between the lowest energy spin states were...

We present graphRC, a graph-based method for rapid transition state (TS) mode analysis that provides chemical insight along normal mode displacements and reaction coordinate trajectories by translating Cartesian displacements into meaningful internal coordinate changes. Internal coordinates are constructed using xyzgraph, a flexible graph-based approach that achieves 98.4% agreement with DFT and 99.8% with existing methods across 4846 structures. Initial validation on 16 diverse TS achieved 100% identification of bond changes, rotations and inversions with zero false positives. Expansion to 395 diverse TS demonstrated robust performance, with 100.0% detection of the primary bond change and an overall agreement of 88.4% with Intrinsic Reaction Coordinate (IRC) connectivity across 984 vibrational bonds. This approach enables programmatic TS verification and rapid reaction coordinate analysis in seconds, complementing more rigorous IRC and Quick Reaction Coordinate (QRC) methods while delivering robust, interpretable TS analysis across diverse chemical systems.

A principal components map was used to optimise on-DNA Suzuki–Miyaura coupling, reductive amination and reverse amide coupling reactions.

One use of CO₂ as a starting material in organic transformations is in the synthesis of cyclic carbonates and polycarbonates. Due to the low reactivity of CO₂, this transformation must be carried out in the presence of an efficient catalyst. Although several catalytic systems have been developed in the past decade, reducing the CO₂ pressure at which the reaction is carried out remains one of the main challenges of the process. In this context, in the present work, we describe the catalytic activity of mixed metal oxides in the synthesis of cyclic carbonates from CO₂ (1 atm) and epoxides at 70 °C. Using these materials as catalysts represents significant benefits since they are very stable, cost-effective, and can be reused in several reaction cycles.

The knowledge of the reactivity of N-nitrosamines (NSAs) with common organic reagents in synthesis is essential in determining their presence in pharmaceutical products, if formed and retained during synthesis. In this study, we carried out a comprehensive survey of the Reaxys database for all reactions in which the NSA functional group is consumed. Very different reactivities for different classes of NSAs, e.g., N,N-dialkylnitrosamines and N,N-diphenylnitrosamine, were identified, suggesting substrates which should be included in any future reactivity screening. A classification of NSAs based on their reactivities, and corresponding reagents and transformations, was drawn up based on the data. Furthermore, the survey identified missing areas in the reported reactivities of NSAs with different reagents. This led to an experimental reactivity screening of 8 commercial NSAs with common synthetic reagents in the Mirabilis tool for purge assessment. The results showed Na₂S₂O₄ in 1 M aqueous NaOH at 50 °C to be highly effective at destroying NSAs without damaging other organic compounds.

One use of CO2 as a starting material in organic transformations is in the synthesis of cyclic carbonates and polycarbonates. Due to the low reactivity of CO2, this transformation must be carried out in the presence of an efficient catalyst. Although several catalytic systems have been developed in the last decade, reducing the CO2 pressure at which the reaction is carried out remains one of the main challenges of the process. In this context, in the present work, we describe the catalytic activity of mixed metal oxides (MMOs) in the synthesis of cyclic carbonates from CO2 (1 atm) and epoxides at 70 °C. The use of these materials as catalysts represents a great advantage since they are highly stable and economical and can be reused in several reaction cycles.

A principal component surfactant_map was developed for 91 commercially surfactants for use in surfactant-enabled organic reactions in water. The map led to activation of fluoride anion, usually strongly hydrated, as an active nucleophile in water.