Optimized Bags of Artificial Neural Networks Can Predict the Viability of Organisms Exposed to Nanoparticles.

Prediction of organismal viability upon exposure to a nanoparticle in varying environments─as fully specified at the molecular scale─has emerged as a useful figure of merit in the design of engineered nanoparticles. We build on our earlier finding that a bag of artificial neural networks (ANNs) can provide such a prediction when such machines are trained with a relatively small data set (with ca. 200 examples). Therein, viabilities were predicted by consensus using the weighted means of the predictions from the bags. Here, we confirm the accuracy and precision of the prediction of nanoparticle viabilities using an optimized bag of ANNs over sets of data examples that had not previously been used in the training and validation process. We also introduce the viability strip, rather than a single value, as the prediction and construct it from the viability probability distribution of an ensemble of ANNs compatible with the data set. Specifically, the ensemble consists of the ANNs arising from subsets of the data set corresponding to different splittings between training and validation, and the different bags (k-folds). A k-1k machine uses a single partition (or bag) of k - 1 ANNs each trained on 1/k of the data to obtain a consensus prediction, and a k-bag machine quorum samples the k possible k-1k machines available for a given partition. We find that with increasing k in the k-bag or k-1k machines, the viability strips become more normally distributed and their predictions become more precise. Benchmark comparisons between ensembles of 4-bag machines and 34 fraction machines suggest that the 34 fraction machine has similar accuracy while overcoming some of the challenges arising from divergent ANNs in the 4-bag machines.

Atomically precise nanoclusters predominantly seed gold nanoparticle syntheses.

Seed-mediated synthesis strategies, in which small gold nanoparticle precursors are added to a growth solution to initiate heterogeneous nucleation, are among the most prevalent, simple, and productive methodologies for generating well-defined colloidal anisotropic nanostructures. However, the size, structure, and chemical properties of the seeds remain poorly understood, which partially explains the lack of mechanistic understanding of many particle growth reactions. Here, we identify the majority component in the seed solution as an atomically precise gold nanocluster, consisting of a 32-atom Au core with 8 halide ligands and 12 neutral ligands constituting a bound ion pair between a halide and the cationic surfactant: Au32X8[AQA+•X-]12 (X = Cl, Br; AQA = alkyl quaternary ammonium). Ligand exchange is dynamic and versatile, occurring on the order of minutes and allowing for the formation of 48 distinct Au32 clusters with AQAX (alkyl quaternary ammonium halide) ligands. Anisotropic nanoparticle syntheses seeded with solutions enriched in Au32X8[AQA+•X-]12 show narrower size distributions and fewer impurity particle shapes, indicating the importance of this cluster as a precursor to the growth of well-defined nanostructures.

Machine Learning-Assisted Carbon Dot Synthesis: Prediction of Emission Color and Wavelength.

Carbon dots (CDs) have attracted great attention in a range of applications due to their bright photoluminescence, high photostability, and good biocompatibility. However, it is challenging to design CDs with specific emission properties because the syntheses involve many parameters, and it is not clear how each parameter influences the CD properties. To help bridge this gap, machine learning, specifically an artificial neural network, is employed in this work to characterize the impact of synthesis parameters on and make predictions for the emission color and wavelength for CDs. The machine reveals that the choice of reaction method, purification method, and solvent relate more closely to CD emission characteristics than the reaction temperature or time, which are frequently tuned in experiments. After considering multiple models, the best performing machine learning classification model achieved an accuracy of 94% in predicting relative to actual color. In addition, hybrid (two-stage) models incorporating both color classification and an artificial neural network k-ensemble model for wavelength prediction through regression performed significantly better than either a standard artificial neural network or a single-stage artificial neural network k-ensemble regression model. The accuracy of the model predictions was evaluated against CD emission wavelengths measured from experiments, and the minimum mean average error is 25.8 nm. Overall, the models developed in this work can effectively predict the photoluminescence emission of CDs and help design CDs with targeted optical properties.

Nonradiative relaxation dynamics in the [Au25-nAgn(SH)18]-1 (n = 1, 12, 25) thiolate-protected nanoclusters.

Evaluation of the electron-nuclear dynamics and relaxation mechanisms of gold and silver nanoclusters and their alloys is important for future photocatalytic, light harvesting, and photoluminescence applications of these systems. In this work, the effect of silver doping on the nonradiative excited state relaxation dynamics of the atomically precise thiolate-protected gold nanocluster [Au25-nAgn(SH)18]-1 (n = 1, 12, 25) is studied theoretically. Time-dependent density functional theory is used to study excited states lying in the energy range 0.0-2.5 eV. The fewest switches surface hopping method with decoherence correction was used to investigate the dynamics of these states. The HOMO-LUMO gap increases significantly upon doping of 12 silver atoms but decreases for the pure silver nanocluster. Doped clusters show a different response for ground state population increase lifetimes and excited state population decay times in comparison to the undoped system. The ground state recovery times of the S1-S6 states in the first excited peak were found to be longer for [Au13Ag12(SH)18]-1 than the corresponding recovery times of other studied nanoclusters, suggesting that this partially doped nanocluster is best for preserving electrons in an excited state. The decay time constants were in the range of 2.0-20 ps for the six lowest energy excited states. Among the higher excited states, S7 has the slowest decay time constant although it occurs more quickly than S1 decay. Overall, these clusters follow common decay time constant trends and relaxation mechanisms due to the similarities in their electronic structures.

Electronic relaxation dynamics in [Au25(SR)18]-1 (R = CH3, C2H5, C3H7, MPA, PET) thiolate-protected nanoclusters.

We investigate the excited electron dynamics in [Au25(SR)18]-1 (R = CH3, C2H5, C3H7, MPA, PET) [MPA = mercaptopropanoic acid, PET = phenylethylthiol] nanoparticles to understand how different ligands affect the excited state dynamics in this system. The population dynamics of the core and higher excited states lying in the energy range 0.00-2.20 eV are studied using a surface hopping method with decoherence correction in a real-time DFT approach. All of the ligated clusters follow a similar trend in decay for the core states (S1-S6). The observed time constants are on the picosecond time scale (2-19 ps), which agrees with the experimental time scale, and this study confirms that the time constants observed experimentally could originate from core-to-core transitions and not from core-to-semiring transitions. In the presence of higher excited states, R = H, CH3, C2H5, C3H7, and PET demonstrate similar relaxations trends whereas R = MPA shows slightly different relaxation of the core states due to a smaller gap between the LUMO+1 and LUMO+2 gap in its electronic structure. The S1 (HOMO → LUMO) state gives the slowest decay in all ligated clusters, while S7 has a relatively long decay. Furthermore, separate electron and hole relaxations were performed on the [Au25(SCH3)18]-1 nanocluster to understand how independent electron and hole relaxations contribute to the overall relaxation dynamics.

Theoretical investigation of relaxation dynamics in the Au18(SH)14 thiolate-protected gold nanocluster.

Experimental findings of Au18(GSH)14 as a photosensitizer with the highest potential compared to other glutathione-protected clusters demand understanding the photophysics and relaxation dynamics of the Au18(SR)14 cluster. To this end, we perform ab initio real-time nonadiabatic molecular dynamics simulations on Au18(SH)14 to investigate its relaxation dynamics compared to the well-studied [Au25(SR)18]-1 relaxation dynamics. In this work, the excitations covering up to ∼2.6 eV in the optical absorption spectrum are analyzed to understand the electronic relaxation process of the Au18(SH)14 cluster. The ground state growth times of Au18(SH)14 are several orders of magnitude shorter than the growth times observed for the [Au25(SH)18]-1 nanocluster. The S1 (HOMO-LUMO) state gives the slowest decay time (∼11 ps) among all the states (S1-S30) considered similar to [Au25(SH)18]-1. However, the S1 state in Au18(SH)14 is a semiring-to-core charge transfer state, whereas S1 in the [Au25(SH)18]-1 cluster is a core-to-core transition. The remaining higher excited states have very short decay time constants less than 1.4 ps except for S2 which has the second slowest decay of 6.4 ps. The hole relaxations are faster than the electron relaxations in Au18(SH)14 due to the closely packed HOMOs in the electronic structure. Radiative relaxations are also examined using the time-dependent density functional theory method, and the excited state emission energy and lifetime are found to be in good agreement with experiment.

[Au18(dppm)6Cl4]4+: a phosphine-protected gold nanocluster with rich charge states.

A diphosphine-protected 18-gold-atom nanocluster was isolated via a facile reduction of an AuI precursor by NaBH4. Its composition was identified as {[Au18(dppm)6Cl4]·C6H6·3Cl·PF6} (SD/Au18, SD = SunDi; dppm = bis-(diphenylphosphino)methane) by X-ray single crystal structural analysis. This nanocluster possesses a prolate shape and is built from an Au10 kernel (bi-octahedral Au6 units sharing one edge) fused with two Au7 caps via sharing six gold atoms. The identity of the Au18 cluster is further demonstrated by ESI-MS. The number of valence electrons of [Au18(dppm)6Cl4]4+ is 10 (n* = 18-4-4), which does not match with the known magic numbers according to the spherical jellium model, and elongated models must be considered. The special stability of the Au18 cluster likely arises from geometrical factors in the metallic core. Two charge states are reported for this system. This work not only presents the structure elucidation of a diphosphine-protected Au18 nanocluster, but also provides an important insight into the growth pattern of gold nanoclusters and the charge states they can achieve.

Diphosphine-protected ultrasmall gold nanoclusters: opened icosahedral Au13 and heart-shaped Au8 clusters.

Due to distinctive quantum confinement effects, ultrasmall gold nanoparticles usually exhibit interesting electronic structure and molecular-like properties. However, the lack of atomically-precise structural information makes the understanding of them almost impossible, such as understanding the relationships between their compositions and unique properties. Herein, by reducing a diphosphine AuI precursor (Au2(dppm)2Cl2; dppm = Ph2PCH2PPh2) with or without a S2- releasing reagent, we enriched our knowledge of the members in the families of Au13 and Au8 by the structural determinations of two new dppm-protected gold nanoclusters, [Au13(dppm)6]5+ (SD/Au1) and [Au8(dppm)4S2]2+ (SD/Au2), respectively. Within SD/Au1, the Au13 kernel significantly deviates from the ideal Ih icosahedron by the elongation of three surface Au-Au bonds to ∼3.5 Å, giving it C3 symmetry, whereas SD/Au2 has a novel heart-shaped C2 symmetric Au8S2 core (central Au4 tetrahedron + two Au2S units) protected by four µ2-dppm ligands in the outer shell. Of note, SD/Au1 represents a rare Au13 nanocluster with an opened icosahedral geometry, and SD/Au2 shows a new edge-shared "core + 4exo" structure type that has never been observed before. The electronic structures and optical absorption spectra of these systems are correlated with time-dependent density functional theory (TDDFT) calculations. Based on the spherical jellium model, the stability of the Au13 and Au8 nanoclusters can be ascribed to 8- and 2-electron superatoms with 1S21P6 and 1S2 configurations, respectively. Interestingly, the cluster SD/Au2 exhibits bright yellow luminescence with an emission maximum at 591 nm that slightly hypsochromically shifts to 581 nm upon cooling to 93 K. Our findings not only enrich the family of diphosphine-protected ultrasmall gold nanoclusters, but also demonstrate the rich variations of gold kernels during the transformation from a simple AuI precursor to Au nanoclusters.