Accelerated mapping of electronic density of states patterns of metallic nanoparticles via machine-learning.

Within first-principles density functional theory (DFT) frameworks, it is challenging to predict the electronic structures of nanoparticles (NPs) accurately but fast. Herein, a machine-learning architecture is proposed to rapidly but reasonably predict electronic density of states (DOS) patterns of metallic NPs via a combination of principal component analysis (PCA) and the crystal graph convolutional neural network (CGCNN). With the PCA, a mathematically high-dimensional DOS image can be converted to a low-dimensional vector. The CGCNN plays a key role in reflecting the effects of local atomic structures on the DOS patterns of NPs with only a few of material features that are easily extracted from a periodic table. The PCA-CGCNN model is applicable for all pure and bimetallic NPs, in which a handful DOS training sets that are easily obtained with the typical DFT method are considered. The PCA-CGCNN model predicts the R2 value to be 0.85 or higher for Au pure NPs and 0.77 or higher for Au@Pt core@shell bimetallic NPs, respectively, in which the values are for the test sets. Although the PCA-CGCNN method showed a small loss of accuracy when compared with DFT calculations, the prediction time takes just ~ 160 s irrespective of the NP size in contrast to DFT method, for example, 13,000 times faster than the DFT method for Pt147. Our approach not only can be immediately applied to predict electronic structures of actual nanometer scaled NPs to be experimentally synthesized, but also be used to explore correlations between atomic structures and other spectrum image data of the materials (e.g., X-ray diffraction, X-ray photoelectron spectroscopy, and Raman spectroscopy).

Facile aqueous-phase synthesis of Ag-Cu-Pt-Pd quadrometallic nanoparticles.

Ag-Cu-Pt-Pd quadrometallic nanoparticles which small Pt and Pd nanoparticles were attached on the surface of AgCu Janus nanoparticles were firstly synthesized by sequential reduction of Pt and Pd precursor in the presence of Janus AgCu bimetallic nanoparticles as seeds in an aqueous solution. Even though there was a small amount of Cu2O on the surface, the synthesized nanoparticles were mainly composed of four independent metallic part, not alloy parts. By theoretical calculation and growth mechanism study, we found that different reducing rate between Ag+ and Cu2+ and sequential reduction of Pt and Pd precursors would be key roles for the formation of the quadrometallic nanoparticles.

Pattern Learning Electronic Density of States.

Electronic density of states (DOS) is a key factor in condensed matter physics and material science that determines the properties of metals. First-principles density-functional theory (DFT) calculations have typically been used to obtain the DOS despite the considerable computation cost. Herein, we report a fast machine learning method for predicting the DOS patterns of not only bulk structures but also surface structures in multi-component alloy systems by a principal component analysis. Within this framework, we use only four features to define the composition, atomic structure, and surfaces of alloys, which are the d-orbital occupation ratio, coordination number, mixing factor, and the inverse of miller indices. While the DFT method scales as O(N3) in which N is the number of electrons in the system size, our pattern learning method can be independent on the number of electrons. Furthermore, our method provides a pattern similarity of 91 ~ 98% compared to DFT calculations. This reveals that our learning method will be an alternative that can break the trade-off relationship between accuracy and speed that is well known in the field of electronic structure calculations.

Reaction Mechanism of Area-Selective Atomic Layer Deposition for Al2O3 Nanopatterns.

The reaction mechanism of area-selective atomic layer deposition (AS-ALD) of Al2O3 thin films using self-assembled monolayers (SAMs) was systematically investigated by theoretical and experimental studies. Trimethylaluminum (TMA) and H2O were used as the precursor and oxidant, respectively, with octadecylphosphonic acid (ODPA) as an SAM to block Al2O3 film formation. However, Al2O3 layers began to form on the ODPA SAMs after several cycles, despite reports that CH3-terminated SAMs cannot react with TMA. We showed that TMA does not react chemically with the SAM but is physically adsorbed, acting as a nucleation site for Al2O3 film growth. Moreover, the amount of physisorbed TMA was affected by the partial pressure. By controlling it, we developed a new AS-ALD Al2O3 process with high selectivity, which produces films of ∼60 nm thickness over 370 cycles. The successful deposition of Al2O3 thin film patterns using this process is a breakthrough technique in the field of nanotechnology.

Simulation Protocol for Prediction of a Solid-Electrolyte Interphase on the Silicon-based Anodes of a Lithium-Ion Battery: ReaxFF Reactive Force Field.

We propose the ReaxFF reactive force field as a simulation protocol for predicting the evolution of solid-electrolyte interphase (SEI) components such as gases (C2H4, CO, CO2, CH4, and C2H6), and inorganic (Li2CO3, Li2O, and LiF) and organic (ROLi and ROCO2Li: R = -CH3 or -C2H5) products that are generated by the chemical reactions between the anodes and liquid electrolytes. ReaxFF was developed from ab initio results, and a molecular dynamics simulation with ReaxFF realized the prediction of SEI formation under real experimental conditions and with a reasonable computational cost. We report the effects on SEI formation of different kinds of Si anodes (pristine Si and SiOx), of the different types and compositions of various carbonate electrolytes, and of the additives. From the results, we expect that ReaxFF will be very useful for the development of novel electrolytes or additives and for further advances in Li-ion battery technology.

Atomistics of the lithiation of oxidized silicon (SiOx) nanowires in reactive molecular dynamics simulations.

Although silicon oxide (SiOx) nanowires (NWs) are recognized as a promising anode material for lithium-ion batteries (LIBs), a clear understanding of their lithiation mechanism has not been reported yet. We elucidate the lithiation mechanism of SiOx NWs at the atomic scale based on molecular dynamics (MD) simulations employing the ReaxFF reactive force field developed through first-principles calculations. SiOx NWs with crystalline Si (c-Si) core and amorphous SiO2 (a-SiO2) shell structures of ∼1 nm in thickness show smaller volume expansion than pristine Si NWs, as found in previous experiments. Lithiation into SiOx NWs creates two interfaces: c-Si/a-LixSi and a-LixSi/a-LiySiO2. The mobility of the latter, which is located farther toward the outside of the NW, is slower than that of the former, which is one of the reasons why the thin SiO2 layer can suppress the volume expansion of SiOx NWs during lithiation. Another reason can be found from the stress distribution, as the SiOx NWs show stress distribution different from the pristine case. Moreover, the lithiation of SiOx NWs leads to the formation of Li2O and Li4SiO4 compounds in the oxide layer, where several Li atoms (not a majority) in Li4SiO4 can escape from the compound and diffuse into the c-Si, in contrast to the Li2O case. However, Li atoms that pass through the SiO2 layer penetrate into the c-Si preferentially along the 〈110〉 or 〈112〉 direction, similar to the mechanism observed in pristine Si NWs. We expect that our comprehensive understanding of the lithiation mechanism of SiOx NWs will provide helpful guidance for the design of SiOx anodes to obtain better performing LIBs.

A comparative first-principles study of the lithiation, sodiation, and magnesiation of black phosphorus for Li-, Na-, and Mg-ion batteries.

Using first-principles calculations, we describe and compare atomistic lithiation, sodiation, and magnesiation processes in black phosphorous with a layered structure similar to graphite for Li-, Na-, and Mg-ion batteries because graphite is not considered to be an electrode material for Na- and Mg-ion batteries. The three processes are similar in that an intercalation mechanism occurs at low Li/Na/Mg concentrations, and then further insertion of Li/Na/Mg leads to a change from the intercalation mechanism to an alloying process. Li and Mg show a columnar intercalation mechanism and prefer to locate in different phosphorene layers, while Na shows a planar intercalation mechanism and preferentially localizes in the same layer. In addition, we compare the mechanical properties of black phosphorous during lithiation, sodiation, and magnesiation. Interestingly, lithiation and sodiation at high concentrations (Li2P and Na2P) lead to the softening of black phosphorous, whereas magnesiation shows a hardening phenomenon. In addition, the diffusion of Li/Na/Mg in black phosphorus during the intercalation process is an easy process along one-dimensional channels in black phosphorus with marginal energy barriers. The diffusion of Li has a lower energy barrier in black phosphorus than in graphite.

Development of the ReaxFFCBN reactive force field for the improved design of liquid CBN hydrogen storage materials.

Liquid CBN (carbon-boron-nitrogen) hydrogen-storage materials such as 3-methyl-1,2-BN-cyclopentane have the advantage of being easily accessible for use in current liquid-fuel infrastructure. To develop practical liquid CBN hydrogen-storage materials, it is of great importance to understand the reaction pathways of hydrogenation/dehydrogenation in the liquid phase, which are difficult to discover by experimental methods. Herein, we developed a reactive force field (ReaxFFCBN) from quantum mechanical (QM) calculations based on density functional theory for the storage of hydrogen in BN-substituted cyclic hydrocarbon materials. The developed ReaxFFCBN provides similar dehydrogenation pathways and energetics to those predicted by QM calculations. Moreover, molecular dynamics (MD) simulations with the developed ReaxFFCBN can predict the stability and dehydrogenation behavior of various liquid CBN hydrogen-storage materials. Our simulations reveal that a unimolecular dehydrogenation mechanism is preferred in liquid CBN hydrogen-storage materials. However, as the temperature in the simulation increases, the contribution of a bimolecular dehydrogenation mechanism also increases. Moreover, our ReaxFF MD simulations show that in terms of thermal stability and dehydrogenation kinetics, liquid CBN materials with a hexagonal structure are more suitable materials than those with a pentagonal structure. We expect that the developed ReaxFFCBN could be a useful protocol in developing novel liquid CBN hydrogen-storage materials.