Machine learning of spectra-property relationship for imperfect and small chemistry data.

Machine learning (ML) is causing profound changes to chemical research through its powerful statistical and mathematical methodological capabilities. However, the nature of chemistry experiments often sets very high hurdles to collect high-quality data that are deficiency free, contradicting the need of ML to learn from big data. Even worse, the black-box nature of most ML methods requires more abundant data to ensure good transferability. Herein, we combine physics-based spectral descriptors with a symbolic regression method to establish interpretable spectra-property relationship. Using the machine-learned mathematical formulas, we have predicted the adsorption energy and charge transfer of the CO-adsorbed Cu-based MOF systems from their infrared and Raman spectra. The explicit prediction models are robust, allowing them to be transferrable to small and low-quality dataset containing partial errors. Surprisingly, they can be used to identify and clean error data, which are common data scenarios in real experiments. Such robust learning protocol will significantly enhance the applicability of machine-learned spectroscopy for chemical science.

Sensitive and Visual Detection of Phosgene by a TICT-Based BODIPY Dye with 8-(o-Hydroxy)aniline as the Active Site.

Phosgene and its substitutes (diphosgene and triphosgene) are widely utilized as chemical industrial materials and chemical warfare agents and pose a threat to public health and environmental safety due to their extreme toxicity. Research efforts have been directed to develop selective and sensitive detection methods for phosgene and its substitutes. In this paper, we have prepared two BODIPY-based fluorescent probes, o-Pah and o-Pha, which are two isomers with different active sites, ortho-aminohydroxy (3',4' or 4',3') phenyls at meso position of BODIPY, and compared their sensing performance toward triphosgene. The probe with o-(4'-amino-3'-hydroxyl), o-Pha, exhibits better sensing performance over the o-(3'-amino-4'-hydroxyl), o-Pah, for instance, a lower limit of detection (LOD) (0.34 nm vs. 1.2 nm), and more rapid response (10 s vs. 200 s). Furthermore, based on the above comparative studies, a red-fluorescence probe o-Phae has been constructed through extending 3,5-conjugation of o-Pha. The probe o-Phae displays rapid response (60 s), high sensitivity to triphosgene (LOD=0.88 nm), and high selectivity for triphosgene over relevant analytes including nitric oxide. Finally, a facile test strip for phosgene was fabricated by immobilizing o-Phae in a polyethylene oxide membrane for sensitive (<2 ppm) and selective detection of phosgene in the gas phase.

Modulating Charge Separation and Intersystem Crossing in Donor-Switch-Acceptor Systems: A Computational Study.

Charge separation and intersystem crossing play critical roles in various applications of organic long persistent luminescence materials, including light-emitting diodes, chemical sensors, theranostics, and many biomedical and information applications. Using first-principles calculations, we demonstrate that an azobenzene acting as a photoswitch can be used for altering the configuration of a donor-switch-acceptor (D-S-A) molecular system to ensure charge separation and promote intersystem crossing upon photoexcitation. The trans to cis photoisomerization of an azobenzene switch creates an electron trap that stabilizes the charge-separated state. The cis conformation further facilitates the singlet to triplet intersystem crossing in the excited state. Our theoretical study of the D-S-A system may help the design of long persistent luminescent organic devices.

Spectra-Based Machine Learning for Predicting the Statistical Interaction Properties of CO Adsorbates on Surface.

Theoretical analyses of small-molecule adsorption on heterogeneous catalyst surfaces often rely on simplified models of molecular adsorption with the most favorable configuration. Given that real-world experimental tests frequently entail multiple molecules interacting with the surface, there is a pressing need for a comprehensive multimolecule adsorption model to bridge the gap between theory and experiment. Using machine learning, we predict the average values of important adsorption properties from conformationally averaged, calculated infrared and Raman spectra and compare these values to those theoretically derived from the conformationally averaged ensemble. Remarkably, our approach yields excellent predictions even when faced with large and indeterminate numbers of surface molecules. These quantitative spectra-averaged property relationships provide a theoretical framework for extracting key interaction properties from the spectra of real chemical environments.

A stable triplet diradical emitter.

Molecules with luminescence have been extensively investigated, but the luminescence of a stable molecule with a triplet ground state has not been observed. Synthesis of boron-containing radicals has attracted lots of interest because of their unique electronic structures and potential applications in organic semiconductors. Though some boron-based diradicals have been reported, neutral boron-containing diradicals with triplet ground states are rare. Herein two borocyclic diradicals with different substituents (3 and 4) have been isolated. Their electronic structures were investigated by EPR and UV spectroscopy, and SQUID magnetometry, in conjunction with DFT calculations. Both experiment and calculation suggest that 3 is an open shell singlet diradical while 4 is a triplet ground state diradical with a large singlet-triplet gap (0.25 kcal mol-1). Both diradicals show multi fluorescence peaks (3: 414, 431, and 470 nm; 4: 420, 433, and 495 nm). 3 displays multiple redox steps and is a potential material towards the design of high-density memory devices. 4 represents the first example of a neutral triplet boron-containing diradical with a strong ferromagnetic interaction, and also is the first stable triplet diradical emitter.

Bridged Azobenzene Enables Dynamic Control of Through-Space Charge Transfer for Photochemical Conversion.

Through-space charge transfer (TSCT) has become a thriving strategy of modulating photogenerated charges in organic photoresponsive molecular systems for potential applications in luminescence, optoelectronics, and photochemical conversion. Yet fixed configuration between electron donor (D) and acceptor (A) is disadvantageous to mitigate charge recombination undermining their performances. By carrying out first-principle simulations, we proposed a protocol enabling dynamic control of TSCT within a D-A system by use of a bridged azobenzene (BAB), whose configuration is self-adaptive upon photoexcitation. While the Z-isomer of BAB facilitates π-π stacking of D-A pair with designated frontier orbital alignment to ensure TSCT, the E-isomer of BAB breaks that stacking and restrains charge recombination. Further, as a CO2 molecule is weakly bound to the anionic acceptor, the former goes bent as a result of charge transfer from the latter, suggesting a path for photodriven CO2 reduction aided by such a donor-switch-acceptor system. Our proof-of-concept study shows the potential of using specific photoswitch to adaptively steer spatial electron transfer within stacked π systems toward photochemical conversion.

A Machine-Learning Protocol for Ultraviolet Protein-Backbone Absorption Spectroscopy under Environmental Fluctuations.

Ultraviolet (UV) absorption spectra are commonly used for characterizing the global structure of proteins. However, the theoretical interpretation of UV spectra is hindered by the large number of required expensive ab initio calculations of excited states spanning a huge conformation space. We present a machine-learning (ML) protocol for far-UV (FUV) spectra of proteins, which can predict FUV spectra of proteins with comparable accuracy to density functional theory (DFT) calculations but with 3-4 orders of magnitude reduced computational cost. It further shows excellent predictive power and transferability that can be used to probe structural mutations and protein folding pathways.

Tandem Enzymatic Self-Assembly and Slow Release of Dexamethasone Enhances Its Antihepatic Fibrosis Effect.

Many chronic liver diseases will advance to hepatic fibrosis and, if without timely intervention, liver cirrhosis or even hepatocellular carcinoma. Anti-inflammation could be a standard therapeutic strategy for hepatic fibrosis treatment, but a "smart" strategy of hepatic fibrosis-targeted, either self-assembly or slow release of an anti-inflammation drug ( e.g., dexamethasone, Dex), has not been reported. Herein, we rationally designed a hydrogelator precursor Nap-Phe-Phe-Lys(Dex)-Tyr(H2PO3)-OH (1-Dex-P) and proposed a tandem enzymatic strategy of self-assembly and slow release of Dex, with which the precursor exhibited much stronger antihepatic fibrosis effect than Dex both in vitro and in vivo. Enzymatic and cell experiments validated that 1-Dex-P was first dephosphorylated by alkaline phosphatase to yield Nap-Phe-Phe-Lys(Dex)-Tyr-OH (1-Dex), which self-assembled into nanofiber 1-Dex. The nanofiber was then hydrolyzed by esterase to transform into nanofiber 1, accompanied by slow release of Dex. We anticipate that our "smart" tandem enzymatic strategy could be widely employed to design more sophisticated drug delivery systems to achieve enhanced therapeutic efficacy than free drugs in the future.

Immune Responsive Release of Tacrolimus to Overcome Organ Transplant Rejection.

Transplant rejection is the key problem in organ transplantation and, in clinic, immunosuppressive agents such as tacrolimus are directly administered to the recipients after surgery for T-cell inhibition. However, direct administration of tacrolimus may bring severe side effects to the recipients. Herein, by rational design of two hydrogelators NapPhePheGluTyrOH (1) and Nap d-Phe dPheGluTyrOH (2), a facile method of immune responsive release of tacrolimus is developed from their hydrogels to overcome organ transplantation rejection. Upon incubation with protein tyrosine kinase, which is activated in T cells after organ transplantation, the tacrolimus-encapsulating Gel 1 or Gel 2 is disassembled to release tacrolimus. Cell experiments show that both Gel 1 and Gel 2 have better inhibition effect on the activated T cells than free drug tacrolimus. Liver transplantation experiments indicate that, after 7 days of treatment of same dose tacrolimus, the recipient rats in the Gel 2 group show significantly extended median survival time of 22 days while the recipients treated with conventional tacrolimus medication have a median survival time of 13 days. It is expected herein that this "smart" facile method of immune responsive release of tacrolimus can be applied to overcome organ transplantation rejection in clinic in the near future.