College of Chemistry

Aflatoxin contamination of staple food commodities is one of the most critical food safety threats in sub-Saharan Africa. Yet, systematic surveillance data from the Democratic Republic of the Congo (DRC) remain exceedingly scarce. In this study, we assessed the prevalence of aflatoxin B1 (AFB1), B2 (AFB2), G1 (AFG1), and G2 (AFG2) in 43 food samples comprising maize (Zea mays L.) raw grain (n = 18), cassava (Manihot esculenta Crantz) raw tuber (n = 11), and groundnut (Arachis hypogaea L.) raw grain (n = 14), collected from two retail markets in Kinshasa, DRC. Analyses performed by reverse-phase HPLC coupled with fluorescence detection and Kobra® electrochemical post-column derivatization revealed the presence of aflatoxins in 41 of 43 samples, with AFB1 being the most prevalent aflatoxin (95.35%), followed by AFB2 (55.81%), AFG1 (44.19%), and AFG2 (34.88%). Total aflatoxin concentrations ranged from below the detection limit to 191.20 μg/kg, with a mean of 16.40 μg/kg. Most samples exceeded the EU AFB1 regulatory limit of 2 μg/kg and the EU total aflatoxin limit of 4 μg/kg. Temperature and relative humidity recorded across eight food storage depots at both retail markets ranged from 29.60 to 30.90 °C and 34.00 to 56.00%, respectively, and serve as indicative of the prevailing storage environment, which may contribute to Aspergillus proliferation and aflatoxin biosynthesis at retail markets in Kinshasa. Future work that combine dietary exposure assessment, quantitative risk modeling, and multi-mycotoxin profiling across Kinshasa and all provinces of the DRC is needed to fully characterize the public health burden and inform evidence-based intervention strategies.

Fe(0) dinitrogen complexes [SiP₃ ^R]Fe-N[triple bond, length as m-dash]N-Fe[SiP₃ ^R] (1-R; [SiP₃ ^R] = PhSi(CH₂PR₂)₃; R = ⁱ Pr, Ph) are accessed by one- or two-electron reduction of Fe(i) and Fe(ii) chloride complexes [SiP₃ ^{i Pr}]FeCl or κ²-[SiP₃ ^R]FeCl₂ (R = ⁱ Pr, Ph), respectively, and fully characterized by means of multinuclear NMR spectroscopy, single-crystal X-ray diffraction (SC-XRD), and Mössbauer spectroscopy. Complexes 1-R represent rare examples of bimetallic Fe(0) compounds with dinitrogen as a bridging ligand and have an overall spin of S = 0. Heating 1-R in the presence of SiH₄ gas resulted in four-fold Si-H activation to furnish diiron complexes [SiP₃ ^R](H)₂Fe[double bond, length as m-dash]Si[double bond, length as m-dash]Fe(H)₂[SiP₃ ^R] (2-R; R = ⁱ Pr, Ph), which were characterized by multinuclear NMR spectroscopy, SC-XRD, Mössbauer spectroscopy, and density functional theory calculations. Complexes 2-R exhibit the shortest iron-silicon bonds to date and nearly linear Fe[double bond, length as m-dash]Si[double bond, length as m-dash]Fe cores. This work extends a previously developed strategy for SiH₄ activation from cobalt to iron, while offering insight into the silicon-atom extrusion from SiH₄ in molecular systems.

Macrolide antibiotics are a class of structurally diverse polyketides including essential medicines like erythromycin and azithromycin. Structurally minimal macrolides such as narbomycin lack much of the functionality common to other members of the class and exhibit a distinct mode of action wherein they completely inhibit bacterial growth despite only moderate inhibition of global protein synthesis. Unfortunately, lack of availability of these natural products has prevented a thorough investigation of their unusual mechanisms of action. Here we report a 10-step synthesis of narbomycin enabled by a Liebeskind-Srogl fragment coupling, a novel macrolactonization by direct displacement of an Evans auxiliary, and a stereoselective late-stage enone reduction. In addition to accessing the natural product, the route enables access to C8,19-unsaturated variants which could in principle serve as useful precursors for late-stage functionalization by conjugate addition.

A deeper understanding of the electronic structure of the lanthanides (Ln) is essential to harness their technological applications and enable responsible recovery/separation. Here, an isostructural series of tris(3,5-dimethyl-1-pyrazolyl)borate (Tp*)-supported LnIII adducts of a redox noninnocent 3,5-di-tert-butyl-o-semiquinone (DTBSQ) ligand, (Tp*)2Ln(DTBSQ), is characterized experimentally and computationally for all the early lanthanides La-Gd, except Pm. Computational simulations, together with spectroscopic measurements on selected systems, show the metals retain the trivalent oxidation state despite semiquinone binding with weak metal-radical coupling. While LnIII-DTBSQ bond distances decrease with decreasing LnIII ionic radii, theory suggests increased 4f-orbital mixing driven by the energetic accessibility of the f-shell, with significant radical contributions to the bonding, occurs for Sm/Eu. Orbital analyses and multireference calculations confirm these effects originate from metal-ligand orbital energy matching rather than through-space overlap. The LnIII 4f interactions are more covalent with the radical ligand than the Tp*, showing remarkably low 6s character. This contrasts with other common ligand-Ln studies and suggests that the DTBSQ-based orbitals have the appropriate symmetry and energy to preferentially interact with the 4f orbitals over the 6s orbital. This work provides new insights into Ln-radical interactions, revealing bonding characteristics for potential selective f-element chelators and extension into actinide chemistry.

The rational development of transition-metal catalysts, even when guided by theory and mechanistic knowledge, involves significant trial and error. Although machine learning offers the potential to accelerate catalyst discovery and optimization, accurately modeling the complex structures of catalysts and the multistep mechanisms by which they react remains challenging, given the limited sets of data available. Olefin hydroformylation is a quintessential example of this challenge: its catalytic cycle involves many, often reversible, steps, and decades of study have not yielded reliable structure-selectivity relationships. We report Libra-ML, a 3D structure-based deep learning approach for predicting experimental outcomes of transition-metal-catalyzed reactions. To demonstrate the ability of Libra-ML to model the outcomes of complex catalytic reactions, we predicted the regioselectivity of hydroformylation with terminal olefins catalyzed by rhodium complexes. Comparisons to existing methods demonstrate the strong performance of Libra-ML and illustrate the importance of encoding 3D structures to predict experimental outcomes with molecular catalysts.

Chiroptical properties in metal halide perovskites typically arise from asymmetric crystal structures induced by chiral organic molecules during synthesis or chiral additives. However, the synthesis of two-dimensional chiral halide perovskites with sufficiently high dissymmetry factors for practical applications remains challenging. Herein, we present a two-step ionic gel-mediated nucleation and recrystallization reaction, enabling the synthesis of nanostructured chiral halide perovskite films exhibiting large intrinsic circular dichroism dissymmetry factors (gCD) up to 10−2. The ionic gel mediates the reaction between lead(II) iodide (PbI2) and R/S-methylbenzylammonium iodide (R/S-MBAI), to facilitate the formation of higher-order iodoplumbate PbI3- and oligomeric Pb-I-MBA species, which recrystallize during heat treatment as anisotropic nanostructures without preferential orientation. Ionic gel-mediated films exhibit a very pronounced peak splitting originating from a well-defined exciton fine structure and narrow photoluminescence linewidth. This is indicative of the presence of electron-hole exchange interactions and minimized structural disorder in the material leading to significantly amplified chiroptical response. This work demonstrates a synthesis route for nanostructured chiral MBA2PbI4 films, whose unique structural feature with reduced crystalline imperfection enables amplified circular dichroism and gCD response.

We report machine-learning (ML) models that predict with high accuracy the enantioselectivity of ketone hydrogenation with Noyori-type diphosphine-diamine catalysts. To mitigate bias in published data, we combined a data set curated from the literature with >1,000 examples from our own high-throughput experimentation. We demonstrate that 3D graph neural networks, trained on 3D representations derived from the enantiodetermining transition state, outperform models based on 2D graphs, molecular fingerprints, and descriptors derived from semiempirical quantum calculations. These results indicate that models trained on 3D structural representations that reflect catalyst-substrate interactions relevant to enantioselectivity can accurately extrapolate to structures distinct from those in the training set.

The reliable determination of transition states (TSs) benefits from second-order information for robust convergence and validation, but the computational expense of Hessians prohibits their routine use in TS optimization. Here, we present a machine-learning-driven TS optimizer that directly predicts the leftmost Hessian eigenvector (LMHE), which is the critical mode that locally approximates the reaction coordinate encompassing the TS. We demonstrate that our LMHE optimizer recovers TS solutions at the same rate as full-Hessian optimizers and robustly from degraded initial guess geometries, thereby eliminating the excessively long wall times characteristic of full-Hessian approaches and reducing total gradient evaluations compared to standard quasi-Newton methods. We further improve accuracy and robustness using uncertainty quantification for identifying occasional LMHE prediction failures, where we then fall back to a full-Hessian update from the machine-learned potential at that optimization step, avoiding expensive active learning. Overall, our methodology and semiautomated workflow deliver second-order stability at first-order computational expense to provide a highly efficient engine for high-throughput reaction discovery.

The homogeneous conversion of ambient dinitrogen to amine products via the N₂ reduction reaction (N₂RR) remains a prized yet challenging feat for d-block complexes and is scarcely reported for f-block complexes. New, mononuclear Ti^IV and Zr^IV aryloxide complexes Ti(DP)₂ (1Ti), Zr(DP)₂ (1Zr), and DP = [2-(OC₆H₂-2-^tBu,4-Me)₂CHPh] produce up to 51 eq. and up to 7.0 eq. of HN(SiMe₃)₂ per Ti/Zr, from N₂, K⁰, weak acid, and chlorotrimethylsilane. Complex 1Ti exhibits more than double the activity toward N₂-silylation of any previously reported Ti N₂RR catalyst and can also catalyze the formation of up to 19 eq. of NH₃, a new feature in early metal N₂RR chemistry. The mononuclear 1Zr is the most active Zr catalyst for N₂-silylation to date. [KSm(DP)₂(THF)₃] (1Sm), the mononuclear analogue of our previously reported dinuclear A-Sm complex, displays only stoichiometric N₂-silylation due to its vulnerability to deleterious side reactions. DFT calculations confirm the catalysis can proceed via a monomeric Ti complex with a terminally bound, activated N₂, agreeing with experimental ¹H DOSY NMR measurements; the N-H bond is formed first, directing the catalyst selectivity. The isolable reduction product [K₃(THF)Ti(DP)(DP^-)(N₂)] is also an active catalyst, and an intermediate in the calculated cycle.

The dehydrogenation of alkanes to alkenes is an appealing strategy for upgrading abundant hydrocarbons, yet it is constrained by the inherent challenge of cleaving two inert C(sp³)-H bonds with selectivity and without overoxidation. We report a cooperative photocatalytic dehydrogenation of unactivated cycloalkanes under visible light irradiation enabled by a new dinuclear uranyl complex supported by an oxidatively stable germanotungstate, [NBu ⁿ ₄]₈[(UO₂)₂(GeW₁₀O₃₄(µ₂-OH)₂)₂]·(CH₃)₂CO (1). The uranyl complex catalytically converts cyclooctane to cyclooctene under ambient conditions with a TON per molecule of 44 and 9,10-dihydrophenanthrene to phenanthrene with a TON of 73 per molecule, using 1 mol% 1 in MeCN solution, under 427 nm irradiation, using [S₂O₈]^2- or chloranil (C₆Cl₄O₂) as a sacrificial oxidant. The value of preorganization of two uranyl centers to yield the product of double hydrogen atom abstraction (HAA) is discussed in comparison with uranyl nitrate, [UO₂(NO₃)₂(H₂O)₂]·4H₂O, the most widely studied uranyl photocatalyst, which is inactive for this reaction. A direct hydrogen atom transfer (d-HAT) mechanism with two HAA processes is proposed as trans-di-deuterated substrate (9S, 10S)-9,10-dihydrophenanthrene-9,10-d₂ (C₁₄H₁₀D₂) exclusively forms d₁ phenanthrene from abstraction of an H and a D atom from the same face.