2D-Material-Assisted GaN Growth on GaN Template by MOCVD and Its Exfoliation Strategy.

The production of freestanding membranes using two-dimensional (2D) materials often involves techniques such as van der Waals (vdW) epitaxy, quasi-vdW epitaxy, and remote epitaxy. However, a challenge arises when attempting to manufacture freestanding GaN by using these 2D-material-assisted growth techniques. The issue lies in securing stability, as high-temperature growth conditions under metal-organic chemical vapor deposition (MOCVD) can cause damage to the 2D materials due to GaN decomposition of the substrate. Even when GaN is successfully grown using this method, damage to the 2D material leads to direct bonding with the substrate, making the exfoliation of the grown GaN nearly impossible. This study introduces an approach for GaN growth and exfoliation on 2D material/GaN templates. First, graphene and hexagonal boron nitride (h-BN) were transferred onto the GaN template, creating stable conditions under high temperatures and various gases in MOCVD. GaN was grown in a two-step process at 750 and 900 °C, ensuring exfoliation in cases where the 2D materials remained intact. Essentially, while it is challenging to grow GaN on 2D material/GaN using only MOCVD, this study demonstrates that with effective protection of the 2D material, the grown GaN can endure high temperatures and still be exfoliated. Furthermore, these results support that vdW epitaxy and remote epitaxy principle are not only possible with specific equipment but also applicable generally.

AI-driven robotic chemist for autonomous synthesis of organic molecules.

The automation of organic compound synthesis is pivotal for expediting the development of such compounds. In addition, enhancing development efficiency can be achieved by incorporating autonomous functions alongside automation. To achieve this, we developed an autonomous synthesis robot that harnesses the power of artificial intelligence (AI) and robotic technology to establish optimal synthetic recipes. Given a target molecule, our AI initially plans synthetic pathways and defines reaction conditions. It then iteratively refines these plans using feedback from the experimental robot, gradually optimizing the recipe. The system performance was validated by successfully determining synthetic recipes for three organic compounds, yielding that conversion rates that outperform existing references. Notably, this autonomous system is designed around batch reactors, making it accessible and valuable to chemists in standard laboratory settings, thereby streamlining research endeavors.

Exceptional Thermochemical Stability of Graphene on N-Polar GaN for Remote Epitaxy.

In this study, we investigate the thermochemical stability of graphene on the GaN substrate for metal-organic chemical vapor deposition (MOCVD)-based remote epitaxy. Despite excellent physical properties of GaN, making it a compelling choice for high-performance electronic and light-emitting device applications, the challenge of thermochemical decomposition of graphene on a GaN substrate at high temperatures has obstructed the achievement of remote homoepitaxy via MOCVD. Our research uncovers an unexpected stability of graphene on N-polar GaN, thereby enabling the MOCVD-based remote homoepitaxy of N-polar GaN. Our comparative analysis of N- and Ga-polar GaN substrates reveals markedly different outcomes: while a graphene/N-polar GaN substrate produces releasable microcrystals (µCs), a graphene/Ga-polar GaN substrate yields nonreleasable thin films. We attribute this discrepancy to the polarity-dependent thermochemical stability of graphene on the GaN substrate and its subsequent reaction with hydrogen. Evidence obtained from Raman spectroscopy, electron microscopic analyses, and overlayer delamination points to a pronounced thermochemical stability of graphene on N-polar GaN during MOCVD-based remote homoepitaxy. Molecular dynamics simulations, corroborated by experimental data, further substantiate that the thermochemical stability of graphene is reliant on the polarity of GaN, due to different reactions with hydrogen at high temperatures. Based on the N-polar remote homoepitaxy of µCs, the practical application of our findings was demonstrated in fabrication of flexible light-emitting diodes composed of p-n junction µCs with InGaN heterostructures.

Stability of Graphene and Influence of AlN Surface Pits on GaN Remote Heteroepitaxy for Exfoliation.

Remote epitaxy is a promising technology that has recently attracted considerable attention, which enables the growth of thin films that copy the crystallographic characteristics of the substrate through two-dimensional material interlayers. The grown films can be exfoliated to form freestanding membranes, although it is often challenging to apply this technique if the substrate materials are prone to damage under harsh epitaxy conditions. For example, remote epitaxy of GaN thin films on graphene/GaN templates has not been achieved by a standard metal-organic chemical vapor deposition (MOCVD) method due to such damages. Here, we report GaN remote heteroepitaxy on graphene/AlN templates by MOCVD and investigate the influence of surface pits in AlN on the growth and exfoliation of GaN thin films. We first show the thermal stability of graphene before GaN growth, based on which two-step growth of GaN on graphene/AlN is developed. The GaN samples are successfully exfoliated after the first step of the growth at 750 °C, whereas the exfoliation failed after the second step at 1050 °C. In-depth analysis confirms that the pits in AlN templates lead to the degradation of graphene near the area and thus the alteration of growth modes and the failure of exfoliation. These results exemplify the importance of chemical and topographic properties of growth templates for successful remote epitaxy. It is one of the key factors for III-nitride-based remote epitaxy, and these results are expected to be of great help in realizing complete remote epitaxy using only MOCVD.

Understanding the 2D-material and substrate interaction during epitaxial growth towards successful remote epitaxy: a review.

Remote epitaxy, which was discovered and reported in 2017, has seen a surge of interest in recent years. Although the technology seemed to be difficult to reproduce by other labs at first, remote epitaxy has come a long way and many groups are able to consistently reproduce the results with a wide range of material systems including III-V, III-N, wide band-gap semiconductors, complex-oxides, and even elementary semiconductors such as Ge. As with any nascent technology, there are critical parameters which must be carefully studied and understood to allow wide-spread adoption of the new technology. For remote epitaxy, the critical parameters are the (1) quality of two-dimensional (2D) materials, (2) transfer or growth of 2D materials on the substrate, (3) epitaxial growth method and condition. In this review, we will give an in-depth overview of the different types of 2D materials used for remote epitaxy reported thus far, and the importance of the growth and transfer method used for the 2D materials. Then, we will introduce the various growth methods for remote epitaxy and highlight the important points in growth condition for each growth method that enables successful epitaxial growth on 2D-coated single-crystalline substrates. We hope this review will give a focused overview of the 2D-material and substrate interaction at the sample preparation stage for remote epitaxy and during growth, which have not been covered in any other review to date.

Exploring Optimal Reaction Conditions Guided by Graph Neural Networks and Bayesian Optimization.

The optimization of organic reaction conditions to obtain the target product in high yield is crucial to avoid expensive and time-consuming chemical experiments. Advancements in artificial intelligence have enabled various data-driven approaches to predict suitable chemical reaction conditions. However, for many novel syntheses, the process to determine good reaction conditions is inevitable. Bayesian optimization (BO), an iterative optimization algorithm, demonstrates exceptional performance to identify reagents compared to synthesis experts. However, BO requires several initial randomly selected experimental results (yields) to train a surrogate model (approximately 10 experimental trials). Parts of this process, such as the cold-start problem in recommender systems, are inefficient. Here, we present an efficient optimization algorithm to determine suitable conditions based on BO that is guided by a graph neural network (GNN) trained on a million organic synthesis experiment data. The proposed method determined 8.0 and 8.7% faster high-yield reaction conditions than state-of-the-art algorithms and 50 human experts, respectively. In 22 additional optimization tests, the proposed method needed 4.7 trials on average to find conditions higher than the yield of the conditions recommended by five synthesis experts. The proposed method is considered in a situation of having a reaction dataset for training GNN.

Scalable graph neural network for NMR chemical shift prediction.

Graph neural networks (GNNs) have been proven effective in the fast and accurate prediction of nuclear magnetic resonance (NMR) chemical shifts of a molecule. Existing methods, despite their effectiveness, suffer from high space complexity and are therefore limited to relatively small molecules. In this work, we propose a scalable GNN for NMR chemical shift prediction. To reduce the space complexity, we sparsify the graph representation of a molecule by regarding only heavy atoms as nodes and their chemical bonds as edges. To better learn from the sparsified graph representation, we improve the message passing and readout functions in the GNN. For the message passing function, we adapt the attention mechanism and residual connection to better capture local information around each node. For the readout function, we use both node-level and graph-level embeddings as the local and global information to better predict node-level chemical shifts. Through the experimental investigation using 13C and 1H NMR datasets, we demonstrate that the proposed method yields higher prediction accuracy and is more scalable to large molecules having many heavy atoms.

Tracking Efficiency Improvement According to Incident Beam Size in QPD-Based PAT System for Common Path-Based Full-Duplex FSO Terminals.

Free space optical (FSO) communication can support various unmanned aerial vehicles' (UAVs) applications that require large capacity data transmission. In order to perform FSO communication between two terminals, it is essential to employ a pointing, acquisition, and tracking (PAT) system with an efficient and optimal performance. We report on the development of a common optical-path-based FSO communication system, tailored for applications in UAVs. The proposed system is equipped with a quadrant photodiode (QPD)-based PAT system without an additional beacon beam subsystem. The presented approach reduced the structural complexity and improved the tracking efficiency for the same size, weight, and power (SWaP). To achieve a robust FSO link in a dynamic UAV environment, the observability and controllability were obtained based on the linearized control according to the incident beam size on the QPD, which was verified by optical simulation and experiments. As a result, the QPD-based PAT system for implementing FSO links demonstrated an up to 4.25 times faster tracking performance. Moreover, the FSO link experimentally confirmed the 1.25 Gbps full-duplex error-free communication at a 50 m distance.

The Role of the Graphene Oxide (GO) and Reduced Graphene Oxide (RGO) Intermediate Layer in CZTSSe Thin-Film Solar Cells.

Cu2ZnSn(S,Se)4 (CZTSSe) solar cells with low cost and eco-friendly characteristics are attractive as future sources of electricity generation, but low conversion efficiency remains an issue. To improve conversion efficiency, a method of inserting intermediate layers between the CZTSSe absorber film and the Mo back contact is used to suppress the formation of MoSe2 and decomposition of CZTSSe. Among the candidates for the intermediate layer, graphene oxide (GO) and reduced GO have excellent properties, including high-charge mobility and low processing cost. Depending on the type of GO, the solar cell parameters, such as fill factor (FF), were enhanced. Thus, the conversion efficiency of 6.3% was achieved using the chemically reduced GO intermediate layer with significantly improved FF.

Interplay of sidewall damage and light extraction efficiency of micro-LEDs.

This Letter describes the impact of shape on micro light-emitting diodes (µLEDs), analyzing 400 µm2 area µLEDs with various mesa shapes (circular, square, and stripes). Appropriate external quantum efficiency (EQE) can yield internal quantum efficiency (IQE) which decreases with increasing peripheral length of the mesas. However, light extraction efficiency (ηe) increased with increasing mesa periphery. We introduce analysis of Jpeak (the current at peak EQE) since it is proportional to the non-radiative recombination. Etching the sidewalls using tetramethylammonium hydroxide (TMAH) increased the peak EQE and decreased the sidewall dependency of Jpeak. Quantitatively, the TMAH etching reduced non-radiative surface recombination by a factor of four. Hence, shrinking µLEDs needs an understanding of the relationship between non-radiative recombination and ηe, where analyzing Jpeak can offer new insights.

Incidence of preventable cardiopulmonary arrest in a mature part-time rapid response system: A prospective cohort study.

BACKGROUND: The purpose of a rapid response system (RRS) is to reduce the incidence of preventable cardiopulmonary arrests (CPAs) and patient deterioration in general wards. The objective of this study is to investigate the incidence and temporal trends of preventable CPAs and determine factors associated with preventable CPAs in a hospital with a mature RRS. METHODS: This was a single-center prospective cohort study of all CPAs occurring in the general ward between March 2017 and June 2020. The RRS operates from 07:00 to 23:00 on weekdays and from 07:00 to 12:00 on Saturdays. All CPAs were reviewed upon biweekly conference, and a panel of intensivists judged their preventability. Trends of preventable CPAs were analyzed using Poisson regression models and factors associated with preventable CPAs were analyzed using multivariable logistic regression. RESULTS: There were 253 CPAs over 40 months, and 64 (25.3%) of these were preventable. The incidence rate of CPAs was 1.07 per 1000 admissions and that of preventable CPAs was 0.27 per 1000 admissions. The number of preventable CPAs decreased by 24% each year (incidence rate ratio = 0.76; p = 0.039) without a change in the total CPA incidence. The most common contributor to the preventability was delayed response from physicians (n = 41, 64.1%). A predictable CPA with a pre-alarm sign had increased odds in the occurrence of preventable CPAs, while a cardiac cause of CPAs and RRS operating hours had decreased odds in terms of occurrence of preventable CPA. CONCLUSION: Our study showed that one-fourth of all CPAs occurring in the general wards were preventable, and these arrests decreased each year. A mature RRS can evolve to reduce preventable CPAs with regular self-evaluation. Efforts should be directed at improving physicians' response time since a delay in their response was the most common cause of preventable CPAs.

Uncertainty-aware prediction of chemical reaction yields with graph neural networks.

In this paper, we present a data-driven method for the uncertainty-aware prediction of chemical reaction yields. The reactants and products in a chemical reaction are represented as a set of molecular graphs. The predictive distribution of the yield is modeled as a graph neural network that directly processes a set of graphs with permutation invariance. Uncertainty-aware learning and inference are applied to the model to make accurate predictions and to evaluate their uncertainty. We demonstrate the effectiveness of the proposed method on benchmark datasets with various settings. Compared to the existing methods, the proposed method improves the prediction and uncertainty quantification performance in most settings.

Data undersampling models for the efficient rule-based retrosynthetic planning.

Computer-aided retrosynthetic planning for organic molecules, which is based on a large synthetic database, is a significant part of the recent development of autonomous robotic chemists. As in other AI fields, however, the class imbalance problem in the dataset affects the prediction performance of retrosynthetic paths. Here, we demonstrate that applying undersampling models to the imbalanced reaction dataset can improve the prediction of retrosynthetic templates for target molecules. We report improvements in the top-1 and top-10 prediction accuracies by 13.8% (13.1, 5.4%) and 8.8% (6.9, 2.4%) for undersampling based on the similarity (random, dissimilarity) clustering of molecular structures of products, respectively. These results demonstrate the importance of deep understanding of the statistical distribution, internal structure, and sampling for the training dataset. For practical applications, the target-oriented undersampling model is proposed and confirmed by the improved prediction performance of 9.3 and 4.2% for the top-1 and top-10 accuracies, respectively.

Molecular search by NMR spectrum based on evaluation of matching between spectrum and molecule.

Inferring molecular structures from experimentally measured nuclear magnetic resonance (NMR) spectra is an important task in many chemistry applications. Herein, we present a novel method implementing an automated molecular search by NMR spectrum. Given a query spectrum and a pool of candidate molecules, the matching score of each candidate molecule with respect to the query spectrum is evaluated by introducing a molecule-to-spectrum estimation procedure. The candidate molecule with the highest matching score is selected. This procedure does not require any prior knowledge of the corresponding molecular structure nor laborious manual efforts by chemists. We demonstrate the effectiveness of the proposed method on molecular search using 13C NMR spectra.

The stability of graphene and boron nitride for III-nitride epitaxy and post-growth exfoliation.

A challenging approach, but one providing a key solution to material growth, remote epitaxy (RE)-a novel concept related to van der Waals epitaxy (vdWE)-requires the stability of a two-dimensional (2-D) material. However, when graphene, a representative 2-D material, is present on substrates that have a nitrogen atom, graphene loss occurs. Although this phenomenon has remained a hurdle for over a decade, restricting the advantages of applying graphene in the growth of III-nitride materials, few previous studies have been conducted. Here, we report the stability of graphene on substrates containing oxygen or nitrogen atoms. Graphene has been observed on highly decomposed Al2O3; however, graphene loss occurred on decomposed AlN at temperatures over 1300 °C. To overcome graphene loss, we investigated 2-D hexagonal boron nitride (h-BN) as an alternative. Unlike graphene on AlN, it was confirmed that h-BN on AlN was intact after the same high-temperature process. Moreover, the overgrown AlN layers on both h-BN/AlN and h-BN/Al2O3 could be successfully exfoliated, which indicates that 2-D h-BN survived after AlN growth and underlines its availability for the vdWE/RE of III-nitrides with further mechanical transfer. By enhancing the stability of the 2-D material on the substrate, our study provides insights into the realization of a novel epitaxy concept.

Improving Ultraviolet Responses in Cu2ZnSn(S,Se)4 Thin-Film Solar Cells Using Quantum Dot-Based Luminescent Down-Shifting Layer.

Quantum dot (QD)-based luminescent down-shifting (LDS) layers were deposited on Cu2ZnSn(S,Se)4 (CZTSSe) solar cells via the drop-casting method. The LDS layers can easily widen the narrow absorption wavelength regions of single-junction solar cells and enable improvement of the short-circuit current. The optical properties of LDS layers deposited on glass and containing different QD contents were analyzed based on their transmittance, reflectance, and absorbance. The absorber films to be used in the CZTSSe solar cells were determined by X-ray diffraction measurements and Raman spectroscopy to determine their crystal structures and secondary phases, respectively. The completed CZTSSe solar cells with LDS layers showed increased ultraviolet responses of up to 25% because of wavelength conversion by the QDs. In addition, the impact of the capping layer, which was formed to protect the QDs from oxygen and moisture, on the solar cell performance was analyzed. Thus, a maximal conversion efficiency of 7.3% was achieved with the 1.0 mL QD condition; furthermore, to the best of our knowledge, this is the first time that LDS layers have been experimentally demonstrated for CZTSSe solar cells.

Toward Large-Scale Ga2O3 Membranes via Quasi-Van Der Waals Epitaxy on Epitaxial Graphene Layers.

Epitaxial growth using graphene (GR), weakly bonded by van der Waals force, is a subject of interest for fabricating technologically important semiconductor membranes. Such membranes can potentially offer effective cooling and dimensional scale-down for high voltage power devices and deep ultraviolet optoelectronics at a fraction of the bulk-device cost. Here, we report on a large-area ß-Ga2O3 nanomembrane spontaneous-exfoliation (1 cm × 1 cm) from layers of compressive-strained epitaxial graphene (EG) grown on SiC, and demonstrated high-responsivity flexible solar-blind photodetectors. The EG was favorably influenced by lattice arrangement of SiC, and thus enabled ß-Ga2O3 direct-epitaxy on the EG. The ß-Ga2O3 layer was spontaneously exfoliated at the interface of GR owing to its low interfacial toughness by controlling the energy release rate through electroplated Ni layers. The use of GR templates contributes to the seamless exfoliation of the nanomembranes, and the technique is relevant to eventual nanomembrane-based integrated device technology.

Valid, Plausible, and Diverse Retrosynthesis Using Tied Two-Way Transformers with Latent Variables.

Retrosynthesis is an essential task in organic chemistry for identifying the synthesis pathways of newly discovered materials, and with the recent advances in deep learning, there have been growing attempts to solve the retrosynthesis problem through transformer models, which are the state-of-the-art in neural machine translation, by converting the problem into a machine translation problem. However, the pure transformer provides unsatisfactory results that lack grammatical validity, chemical plausibility, and diversity in reactant candidates. In this study, we develop tied two-way transformers with latent modeling to solve those problems using cycle consistency checks, parameter sharing, and multinomial latent variables. Experimental results obtained using public and in-house datasets demonstrate that the proposed model improves the retrosynthesis accuracy, grammatical error, and diversity, and qualitative evaluation results verify its ability to suggest valid and plausible results.

Impact of Na Doping on the Carrier Transport Path in Polycrystalline Flexible Cu2ZnSn(S,Se)4 Solar Cells.

It is well-known that the alkali doping of polycrystalline Cu2ZnSn(S,Se)4 (CZTSSe) and Cu(In,Ga)(Se,S)2 has a beneficial influence on the device performance and there are various hypotheses about the principles of performance improvement. This work clearly explains the effect of Na doping on the fill factor (FF) rather than on all of the solar cell parameters (open-circuit voltage, FF, and sometimes short circuit current) for overall performance improvement. When doping is optimized, the fabricated device shows sufficient built-in potential and selects a better carrier transport path by the high potential difference between the intragrains and the grain boundaries. On the other hand, when doping is excessive, the device shows low contact potential difference and FF and selects a worse carrier transport path even though the built-in potential becomes stronger. The fabricated CZTSSe solar cell on a flexible metal foil optimized with a 25 nm thick NaF doping layer achieves an FF of 62.63%, thereby clearly showing the enhancing effect of Na doping.