RESUMO
To expedite new molecular compound development, a long-sought goal within the chemistry community has been to predict molecules' bulk properties of interest a priori to synthesis from a chemical structure alone. In this work, we demonstrate that machine learning methods can indeed be used to directly learn the relationship between chemical structures and bulk crystalline properties of molecules, even in the absence of any crystal structure information or quantum mechanical calculations. We focus specifically on a class of organic compounds categorized as energetic materials called high explosives (HE) and predicting their crystalline density. An ongoing challenge within the chemistry machine learning community is deciding how best to featurize molecules as inputs into machine learning models-whether expert handcrafted features or learned molecular representations via graph-based neural network models-yield better results and why. We evaluate both types of representations in combination with a number of machine learning models to predict the crystalline densities of HE-like molecules curated from the Cambridge Structural Database, and we report the performance and pros and cons of our methods. Our message passing neural network (MPNN) based models with learned molecular representations generally perform best, outperforming current state-of-the-art methods at predicting crystalline density and performing well even when testing on a data set not representative of the training data. However, these models are traditionally considered black boxes and less easily interpretable. To address this common challenge, we also provide a comparison analysis between our MPNN-based model and models with fixed feature representations that provides insights as to what features are learned by the MPNN to accurately predict density.
Assuntos
Aprendizado de Máquina , Redes Neurais de ComputaçãoRESUMO
Herein we report the fabrication of ultralight gold aerogel monoliths with tunable densities and pore structures. Gold nanowires are prepared at the gram scale by substrate-assisted growth with uniform size, ultrathin diameters, high purity, and a high aspect ratio. Freeze-casting of suspensions of these nanowires produces free-standing, monolithic aerogels with tunable densities from 6 to 23 mg/cm3, which to the best of our knowledge represents the lowest density monolithic gold material. We also demonstrate that the pore geometries created during freeze-casting can be systematically tuned across multiple length scales by the selection of different solvents and excipients in the feedstock suspension. The mechanical behavior of porous materials depends on relative density and pore architectures.
RESUMO
Packing motifs-patterns in how molecules orient relative to one another in a crystal structure-are an important concept in many subdisciplines of materials science because of correlations observed between specific packing motifs and properties of interest. That said, packing motif data sets have remained small and noisy due to intensive manual labeling processes and insufficient labeling schemes. The most prominent labeling algorithms calculate relative interplanar angles of nearest neighbor molecules to determine the packing motif of a molecular crystal, but this simple approach can fail when neighbors are naively sampled isotropically around the crystal structure. To remedy this issue, we propose an optimization algorithm, which rotates the molecular crystal structure to find representative molecules that inform the packing motif. We package this algorithm into an automated framework-Autopack-which both optimally rotates the crystal structure and labels the packing motif based on the appropriate neighboring molecules. In this work, we detail the Autopack framework and its performance, which shows improvements compared to previous state-of-the-art labeling methods, providing the first quantitative point of comparison for packing motif labeling algorithms. Furthermore, using Autopack (available at https://ipo.llnl.gov/technologies/software/autopack), we perform the first large-scale study of potential relationships between chemicals' compositions and packing motifs, which shows that these relationships are more complex than previously hypothesized from studies that used only tens of polycyclic aromatic hydrocarbon molecules. Autopack's capabilities help pose next steps for crystal engineering research focusing not only on a molecule's adoption of a specific packing motif but also on new structure-property relationships.
Assuntos
Algoritmos , Análise por Conglomerados , Estrutura MolecularRESUMO
We demonstrate a scalable method to create metallic nanowire arrays and meshes over square-centimeter-areas with tunable sub-100 nm dimensions and geometries using the shear alignment of block copolymers. We use the block copolymer poly(styrene)-b-poly(2-vinyl pyridine) (PS-P2VP) since the P2VP block complexes with metal salts like Na2PtCl4, thereby enabling us to directly pattern nanoscale platinum features. We investigate what shear alignment processing parameters are necessary to attain high quality and well-ordered nanowire arrays and quantify how the block copolymer's molecular weight affects the resulting Pt nanowires' dimensions and defect densities. Through systematic studies of processing parameters and scanning transmission electron microscopy (STEM) tomography, we determine that the equivalent of 2-3 monolayers of PS-P2VP are required to produce a single layer of well-aligned nanowires. The resulting nanowires' widths and heights can be tuned between 11-27 nm and 9-50 nm, respectively, and have periodicites varying between 37 and 63 nm, depending on the choice of block copolymer molecular weight. We observe that the height-to-width ratio of the nanowires also increases with molecular weight, reaching a value of almost 2 with the largest dimensions fabricated. Furthermore, we demonstrate that an additional layer of Pt nanowires can be orthogonally aligned on top of and without disturbing an underlying layer, thereby enabling the fabrication of Pt nanowire meshes with tunable sub-100 nm dimensions and geometries over a cm2-area.
RESUMO
Low-density metal foams have many potential applications in electronics, energy storage, catalytic supports, fuel cells, sensors, and medical devices. Here, we report a new method for fabricating ultralight, conductive silver aerogel monoliths with predictable densities using silver nanowires. Silver nanowire building blocks were prepared by polyol synthesis and purified by selective precipitation. Silver aerogels were produced by freeze-casting nanowire aqueous suspensions followed by thermal sintering to weld the nanowire junctions. As-prepared silver aerogels have unique anisotropic microporous structures, with density precisely controlled by the nanowire concentration, down to 4.8 mg/cm3 and an electrical conductivity up to 51â¯000 S/m. Mechanical studies show that silver nanowire aerogels exhibit "elastic stiffening" behavior with a Young's modulus up to 16â¯800 Pa.
RESUMO
Graphene is an atomically thin, two-dimensional (2D) carbon material that offers a unique combination of low density, exceptional mechanical properties, thermal stability, large surface area, and excellent electrical conductivity. Recent progress has resulted in macro-assemblies of graphene, such as bulk graphene aerogels for a variety of applications. However, these three-dimensional (3D) graphenes exhibit physicochemical property attenuation compared to their 2D building blocks because of one-fold composition and tortuous, stochastic porous networks. These limitations can be offset by developing a graphene composite material with an engineered porous architecture. Here, we report the fabrication of 3D periodic graphene composite aerogel microlattices for supercapacitor applications, via a 3D printing technique known as direct-ink writing. The key factor in developing these novel aerogels is creating an extrudable graphene oxide-based composite ink and modifying the 3D printing method to accommodate aerogel processing. The 3D-printed graphene composite aerogel (3D-GCA) electrodes are lightweight, highly conductive, and exhibit excellent electrochemical properties. In particular, the supercapacitors using these 3D-GCA electrodes with thicknesses on the order of millimeters display exceptional capacitive retention (ca. 90% from 0.5 to 10 A·g(-1)) and power densities (>4 kW·kg(-1)) that equal or exceed those of reported devices made with electrodes 10-100 times thinner. This work provides an example of how 3D-printed materials, such as graphene aerogels, can significantly expand the design space for fabricating high-performance and fully integrable energy storage devices optimized for a broad range of applications.
RESUMO
Programmable positioning of 2 µm polystyrene (PS) beads with single particle precision and location selective, "on-demand", particle deposition was demonstrated by utilizing patterned electrodes and electrophoretic deposition (EPD). An electrode with differently sized hole patterns, from 0.5 to 5 µm, was used to illustrate the discriminatory particle deposition events based on the voltage and particle-to-hole size ratio. With decreasing patterned hole size, a larger electric field was required for a particle deposition event to occur in that hole. For the 5 µm hole, particle deposition began to occur at 10 V/cm where as an electric field of 15 V/cm was required for particles to begin depositing in the 2 µm holes. The likelihood of particle depositions continued to increase for smaller sized holes as the electric field increased. Eventually, a monolayer of particles began to form at approximately 20 V/cm. In essence, a voltage threshold was found for each hole pattern of different sizes, allowing fine adjustments in pattern hole size and voltage to control when a particle deposition event took place, even with the patterns on the same electrode. This phenomenon opens a route toward controlled, multimaterial deposition and assembly onto substrates without repatterning of the electrode or complicated surface modification of the particles. An analytical approach using the theories for electrophoresis and dielectrophoresis found the former to be the dominating force for depositing a particle into a patterned hole. Ebeam lithography was used to pattern spherical holes in precise configurations onto electrode surfaces, where each hole accompanied a polystyrene (PS) particle placement and attachment during EPD. The versatility of e-beam lithography was utilized to create arbitrary pattern configurations to fabricate particle assemblies of limitless configurations, enabling fabrication of unique materials assemblies and interfaces.
RESUMO
Recent ex situ observations of crystallization in both natural and synthetic systems indicate that the classical models of nucleation and growth are inaccurate. However, in situ observations that can provide direct evidence for alternative models have been lacking due to the limited temporal and spatial resolution of experimental techniques that can observe dynamic processes in a bulk solution. Here we report results from liquid cell transmission electron microscopy studies of nucleation and growth of Au, CaCO3, and iron oxide nanoparticles. We show how these in situ data can be used to obtain direct evidence for the mechanisms underlying nanoparticle crystallization as well as dynamic information that provide constraints on important energetic parameters not available through ex situ methods.
RESUMO
The materials science community has been increasingly interested in harnessing the power of deep learning to solve various domain challenges. However, despite their effectiveness in building highly predictive models, e.g., predicting material properties from microstructure imaging, due to their opaque nature fundamental challenges exist in extracting meaningful domain knowledge from the deep neural networks. In this work, we propose a technique for interpreting the behavior of deep learning models by injecting domain-specific attributes as tunable "knobs" in the material optimization analysis pipeline. By incorporating the material concepts in a generative modeling framework, we are able to explain what structure-to-property linkages these black-box models have learned, which provides scientists with a tool to leverage the full potential of deep learning for domain discoveries.
RESUMO
Bundles of multi-walled carbon nanotubes of uniform diameter decorated with Ni nanoparticles were synthesized using mesoporous silicates as templates. The ordered morphology and the narrow pore size distribution of mesoporous silicates provide an ideal platform to synthesize uniformly sized carbon nanotubes. In addition, homogeneous sub-10 nm pore sizes of the templates allow in situ formation of catalytic nanoparticles with uniform diameters which end up decorating the carbon nanotubes. The resulting carbon nanotubes are multi-walled with a uniform diameter corresponding to the pore diameter of the template used during the synthesis that are decorated with the catalysts used to synthesize them. They have a narrow size distribution which can be used in many energy related fields of research.
Assuntos
Nanopartículas/química , Nanotecnologia/métodos , Nanotubos de Carbono/química , Níquel/química , Silicatos/química , Nanopartículas/ultraestrutura , Nanotubos de Carbono/ultraestrutura , PorosidadeRESUMO
In this paper, we leverage predictive uncertainty of deep neural networks to answer challenging questions material scientists usually encounter in machine learning-based material application workflows. First, we show that by leveraging predictive uncertainty, a user can determine the required training data set size to achieve a certain classification accuracy. Next, we propose uncertainty-guided decision referral to detect and refrain from making decisions on confusing samples. Finally, we show that predictive uncertainty can also be used to detect out-of-distribution test samples. We find that this scheme is accurate enough to detect a wide range of real-world shifts in data, e.g., changes in the image acquisition conditions or changes in the synthesis conditions. Using microstructure information from scanning electron microscope (SEM) images as an example use case, we show that leveraging uncertainty-aware deep learning can significantly improve the performance and dependability of classification models.
RESUMO
Scanning Electron Microscopy (SEM) images provide a variety of structural and morphological information of nanomaterials. In the material informatics domain, automatic recognition and quantitative analysis of SEM images in a high-throughput manner are critical, but challenges still remain due to the complexity and the diversity of image configurations in both shape and size. In this paper, we present a generally applicable approach using computer vision and machine learning techniques to quantitatively extract particle size, size distribution and morphology information in SEM images. The proposed pipeline offers automatic, high-throughput measurements even when overlapping nanoparticles, rod shapes, and core-shell nanostructures are present. We demonstrate effectiveness of the proposed approach by performing experiments on SEM images of nanoscale materials and structures with different shapes and sizes. The proposed approach shows promising results (Spearman coefficients of 0.91 and 0.99 using fully automated and semi-automated processes, respectively) when compared with manually measured sizes. The code is made available as open source software at https://github.com/LLNL/LIST.
RESUMO
Recent advancements with the directed assembly of block copolymers have enabled the fabrication over cm2 areas of highly ordered metal nanowire meshes, or nanolattices, which are of significant interest as transparent electrodes. Compared to randomly dispersed metal nanowire networks that have long been considered the most promising next-generation transparent electrode material, such ordered nanolattices represent a new design paradigm that is yet to be optimized. Here, through optical and electrical simulations, we explore the potential design parameters for such nanolattices as transparent conductive electrodes, elucidating relationships between the nanowire dimensions, defects, and the nanolattices' conductivity and transmissivity. We find that having an ordered nanowire network significantly decreases the length of nanowires required to attain both high transmissivity and high conductivity, and we quantify the network's tolerance to defects in relation to other design constraints. Furthermore, we explore how both optical and electrical anisotropy can be introduced to such nanolattices, opening an even broader materials design space and possible set of applications.
RESUMO
This communication reports a new method to purify copper nanowires with nearly 100% yield from undesired copper nanoparticle side-products formed during batch processes of copper nanowire synthesis. This simple separation method can yield large quantities of long, uniform, high-purity copper nanowires to meet the requirements of nanoelectronics applications as well as provide an avenue for purifying copper nanowires in the industrial scale synthesis of copper nanowires, a key step for commercialization and application of nanowires.
RESUMO
Graphene is a two-dimensional material that offers a unique combination of low density, exceptional mechanical properties, large surface area and excellent electrical conductivity. Recent progress has produced bulk 3D assemblies of graphene, such as graphene aerogels, but they possess purely stochastic porous networks, which limit their performance compared with the potential of an engineered architecture. Here we report the fabrication of periodic graphene aerogel microlattices, possessing an engineered architecture via a 3D printing technique known as direct ink writing. The 3D printed graphene aerogels are lightweight, highly conductive and exhibit supercompressibility (up to 90% compressive strain). Moreover, the Young's moduli of the 3D printed graphene aerogels show an order of magnitude improvement over bulk graphene materials with comparable geometric density and possess large surface areas. Adapting the 3D printing technique to graphene aerogels realizes the possibility of fabricating a myriad of complex aerogel architectures for a broad range of applications.
RESUMO
Formation of biomineral structures is increasingly attributed to directed growth of a mineral phase from an amorphous precursor on an organic matrix. While many in vitro studies have used calcite formation on organothiol self-assembled monolayers (SAMs) as a model system to investigate this process, they have generally focused on the stability of amorphous calcium carbonate (ACC) or maximizing control over the order of the final mineral phase. Little is known about the early stages of mineral formation, particularly the structural evolution of the SAM and mineral. Here we use near-edge X-ray absorption spectroscopy (NEXAFS), photoemission spectroscopy (PES), X-ray diffraction (XRD), and scanning electron microscopy (SEM) to address this gap in knowledge by examining the changes in order and bonding of mercaptophenol (MP) SAMs on Au(111) during the initial stages of mineral formation as well as the mechanism of ACC to calcite transformation during template-directed crystallization. We demonstrate that formation of ACC on the MP SAMs brings about a profound change in the morphology of the monolayers: although the as-prepared MP SAMs are composed of monomers with well-defined orientations, precipitation of the amorphous mineral phase results in substantial structural disorder within the monolayers. Significantly, a preferential face of nucleation is observed for crystallization of calcite from ACC on the SAM surfaces despite this static disorder.