ACS Central Science最新文献

Rapid determination of molecular structures can greatly accelerate workflows across many chemical disciplines. However, elucidating structure using only one-dimensional (1D) NMR spectra, the most readily accessible data, remains an extremely challenging problem because of the combinatorial explosion of the number of possible molecules as the number of constituent atoms is increased. Here, we introduce a multitask machine learning framework that predicts the molecular structure (formula and connectivity) of an unknown compound solely based on its 1D ¹H and/or ¹³C NMR spectra. First, we show how a transformer architecture can be constructed to efficiently solve the task, traditionally performed by chemists, of assembling large numbers of molecular fragments into molecular structures. Integrating this capability with a convolutional neural network, we build an end-to-end model for predicting structure from spectra that is fast and accurate. We demonstrate the effectiveness of this framework on molecules with up to 19 heavy (non-hydrogen) atoms, a size for which there are trillions of possible structures. Without relying on any prior chemical knowledge such as the molecular formula, we show that our approach predicts the exact molecule 69.6% of the time within the first 15 predictions, reducing the search space by up to 11 orders of magnitude.

We introduce a multitask machine learning framework that rapidly predicts both the molecular structure and molecular fragments of an unknown compound using only one-dimensional ¹H and ¹³C NMR spectra.

The self-assembly of Tau into filaments, which mirror the structures observed in Alzheimer’s disease (AD) brains, raises questions about the role of AD-specific post-translational modifications (PTMs) in the formation of paired helical filaments (PHFs). To investigate this, we developed a synthetic approach to produce Tau(291–391) featuring N-acetyllysine, phosphoserine, phosphotyrosine, and N-glycosylation at positions commonly modified in post-mortem AD brains. Using various electron and optical microscopy techniques, we discovered that these modifications generally hinder the in vitro assembly of Tau into PHFs. Interestingly, while acetylation’s effect on Tau assembly displayed variability, either promoting or inhibiting phase transitions in cofactor-free aggregation, heparin-induced aggregation, and RNA-mediated liquid–liquid phase separation (LLPS), phosphorylation uniformly mitigated these processes. Our observations suggest that PTMs, particularly those situated outside the rigid core, are pivotal in the nucleation of PHFs. Moreover, with heparin-induced aggregation leading to the formation of heterogeneous aggregates, most AD-specific PTMs appeared to decelerate aggregation. The impact of acetylation on RNA-induced LLPS was notably site-dependent, whereas phosphorylation consistently reduced LLPS across all proteoforms examined. These insights underscore the complex interplay between site-specific PTMs and environmental factors in modulating Tau aggregation kinetics, highlighting the role of PTMs located outside the ordered filament core in driving the self-assembly.

Synthetic peptide fragments of Tau self-assemble into fibrils mirroring paired helical filaments. Post-translational modifications in the ordered core control aggregation and phase transitions.

X-ray detection technology is essential in various fields, including medical imaging and security checks. However, exposure to large doses of X-rays poses considerable health risks. Therefore, it is crucial to reduce the radiation dosage without compromising detection efficiency. To address this concern, we propose an innovative cascade-engineered approach that uses two interconnected single-crystal devices to mitigate dark current and enhance the detection limit. Using laboratory-grown methylammonium lead bromide (MAPbBr₃) perovskite single crystals, we engineered devices that significantly reduced detection thresholds and improved signal-to-noise ratios (SNRs). The detection threshold dropped from 590 nGy·s^-1 with the conventional method to 100 nGy·s^-1 using the cascade approach, surpassing the most recent record of 500 nGy·s^-1 achieved for MAPbBr₃ devices under nearly identical conditions. The dark current was halved compared to that of conventional devices, and spatial resolution improved from 5.6 to 8.5 lp·mm^-1. Imaging trials confirmed improved resolution and effectiveness at low doses, highlighting the approach's potential for medical diagnostics that prioritizes reducing radiation exposure without compromising image quality. The groundbreaking nature of this approach is highlighted by its adaptability across diverse electrical environments and crystal types, as evident in CdTe crystals, indicating its potential for widespread utilization in low-dose leakage monitoring and commercial X-ray devices.

X-ray detection technology is essential in various fields, including medical imaging and security checks. However, exposure to large doses of X-rays poses considerable health risks. Therefore, it is crucial to reduce the radiation dosage without compromising detection efficiency. To address this concern, we propose an innovative cascade-engineered approach that uses two interconnected single-crystal devices to mitigate dark current and enhance the detection limit. Using laboratory-grown methylammonium lead bromide (MAPbBr₃) perovskite single crystals, we engineered devices that significantly reduced detection thresholds and improved signal-to-noise ratios (SNRs). The detection threshold dropped from 590 nGy·s^–1 with the conventional method to 100 nGy·s^–1 using the cascade approach, surpassing the most recent record of 500 nGy·s^–1 achieved for MAPbBr₃ devices under nearly identical conditions. The dark current was halved compared to that of conventional devices, and spatial resolution improved from 5.6 to 8.5 lp·mm^–1. Imaging trials confirmed improved resolution and effectiveness at low doses, highlighting the approach’s potential for medical diagnostics that prioritizes reducing radiation exposure without compromising image quality. The groundbreaking nature of this approach is highlighted by its adaptability across diverse electrical environments and crystal types, as evident in CdTe crystals, indicating its potential for widespread utilization in low-dose leakage monitoring and commercial X-ray devices.

This study explores a novel cascade-engineering device designed for low-dose X-ray detection.

Rapid determination of molecular structures can greatly accelerate workflows across many chemical disciplines. However, elucidating structure using only one-dimensional (1D) NMR spectra, the most readily accessible data, remains an extremely challenging problem because of the combinatorial explosion of the number of possible molecules as the number of constituent atoms is increased. Here, we introduce a multitask machine learning framework that predicts the molecular structure (formula and connectivity) of an unknown compound solely based on its 1D ¹H and/or ¹³C NMR spectra. First, we show how a transformer architecture can be constructed to efficiently solve the task, traditionally performed by chemists, of assembling large numbers of molecular fragments into molecular structures. Integrating this capability with a convolutional neural network, we build an end-to-end model for predicting structure from spectra that is fast and accurate. We demonstrate the effectiveness of this framework on molecules with up to 19 heavy (non-hydrogen) atoms, a size for which there are trillions of possible structures. Without relying on any prior chemical knowledge such as the molecular formula, we show that our approach predicts the exact molecule 69.6% of the time within the first 15 predictions, reducing the search space by up to 11 orders of magnitude.

Proton exchange membrane (PEM) water electrolyzers stand as one of the foremost promising avenues for acidic water splitting and green hydrogen production, yet this electrolyzer encounters significant challenges. The primary culprit lies in not only the requirements of substantial platinum-group-metal (PGM)-based electrocatalysts (e.g., IrO _x ) at the anode where sluggish oxygen evolution reaction (OER) takes place, but also the harsh high overpotential and acidic environments leading to severe performance degradation. The key points for obtaining accurate stability/durability information on the OER catalysts have not been well agreed upon, in contrast to the oxygen reduction reaction fields. In this regard, we herein reviewed and discussed the pivotal experimental variables involved in stability/durability testing (including but not limited to electrolyte, impurity, catalyst loading, and two/three-electrode vs membrane-electrode-assembly), while the test protocols are revisited and summarized. This outlook is aimed at highlighting the reasonable and effective accelerated degradation test procedures to unravel the acidic OER catalyst instability issues and promote the research and development of a PEM water electrolyzer.

Proton exchange membrane (PEM) water electrolyzers stand as one of the foremost promising avenues for acidic water splitting and green hydrogen production, yet this electrolyzer encounters significant challenges. The primary culprit lies in not only the requirements of substantial platinum-group-metal (PGM)-based electrocatalysts (e.g., IrO_x) at the anode where sluggish oxygen evolution reaction (OER) takes place, but also the harsh high overpotential and acidic environments leading to severe performance degradation. The key points for obtaining accurate stability/durability information on the OER catalysts have not been well agreed upon, in contrast to the oxygen reduction reaction fields. In this regard, we herein reviewed and discussed the pivotal experimental variables involved in stability/durability testing (including but not limited to electrolyte, impurity, catalyst loading, and two/three-electrode vs membrane-electrode-assembly), while the test protocols are revisited and summarized. This outlook is aimed at highlighting the reasonable and effective accelerated degradation test procedures to unravel the acidic OER catalyst instability issues and promote the research and development of a PEM water electrolyzer.

The key experimental parameters/protocols as well as their effects on the stability tests of catalysts for acidic oxygen evolution are critically reviewed and discussed in this outlook.

The self-assembly of Tau into filaments, which mirror the structures observed in Alzheimer's disease (AD) brains, raises questions about the role of AD-specific post-translational modifications (PTMs) in the formation of paired helical filaments (PHFs). To investigate this, we developed a synthetic approach to produce Tau(291-391) featuring N-acetyllysine, phosphoserine, phosphotyrosine, and N-glycosylation at positions commonly modified in post-mortem AD brains. Using various electron and optical microscopy techniques, we discovered that these modifications generally hinder the in vitro assembly of Tau into PHFs. Interestingly, while acetylation's effect on Tau assembly displayed variability, either promoting or inhibiting phase transitions in cofactor-free aggregation, heparin-induced aggregation, and RNA-mediated liquid-liquid phase separation (LLPS), phosphorylation uniformly mitigated these processes. Our observations suggest that PTMs, particularly those situated outside the rigid core, are pivotal in the nucleation of PHFs. Moreover, with heparin-induced aggregation leading to the formation of heterogeneous aggregates, most AD-specific PTMs appeared to decelerate aggregation. The impact of acetylation on RNA-induced LLPS was notably site-dependent, whereas phosphorylation consistently reduced LLPS across all proteoforms examined. These insights underscore the complex interplay between site-specific PTMs and environmental factors in modulating Tau aggregation kinetics, highlighting the role of PTMs located outside the ordered filament core in driving the self-assembly.

The synthesis of polymeric thermoset materials with spatially controlled physical properties using readily available resins is a grand challenge. To address this challenge, we developed a photoinitiated polymerization method that enables the spatial switching of radical and cationic polymerizations by controlling the dosage of monochromatic light. This method, which we call Switching Polymerizations by Light Titration (SPLiT), leverages the use of substoichiometric amounts of a photobuffer in combination with traditional photoacid generators. Upon exposure to a low dose of light, the photobuffer inhibits the cationic polymerization, while radical polymerization is initiated. With an increased light dosage, the buffer system saturates, leading to the formation of a strong acid that initiates a cationic polymerization of the dormant monomer. Applying this strategy, patterning is achieved by spatially varying light dosage via irradiation time or intensity allowing for simple construction of multimaterial thermosets. Importantly, by the addition of an inexpensive photobuffer, such as tetrabutylammonium chloride, commercially available resins can be implemented in grayscale vat photopolymerization 3D printing to prepare sophisticated multimodulus constructs.

We employ weakly basic anionic photobuffers to decouple two photopolymerizations in one pot. The photobuffer enables tuning of multimaterials by adjusting the dosage of a single wavelength of light.