Modern industry's demand for semiconductor process of ultra-thin films has encountered technological limits due to quantum tunneling effects, necessitating new process technologies. In these extreme situations, we can discover endless opportunities as an engineer. We can solve product reliability problems by creating a very thermodynamically stable environment for semiconductor materials, and conversely, we can maximize performance by controlling materials at the atomic level.

• Current top-down nanopatterning techniques for the fabrication of semiconductor devices are typically comprised of many sequential processes including film deposition, lithography, and etch steps. However, this approach faces fundamental limitations primarily due to pattern-misalignment errors that are critical to the viability of advanced technology nodes. In addition, current integration strategies have focused on the implementation of 3D structured devices such as 3D FinFETs and 3D V-NANDs in place of the conventional downscaling of planar devices. With this regard, there is a growing need for technologies that enable patterning in the horizontal and vertical direction for versatile nanopatterning in 3D cell structures. 

• In recent years, area-selective atomic layer deposition (AS-ALD) has attracted considerable attention as an alternative bottom-up approach for the patterning of nanoscale dimensions enabling the selective growth of thin films in conjunction with surface modification. The AS-ALD method enables self-aligned fabrication by limiting deposition to specific areas and also reducing the number of processing steps compared with the top-down patterning specified above, indicative of time- and cost-effective methodologies. For this purpose, surface modification with alkylating agents is the most important prerequisite for the successful enablement of AS-ALD, since ALD operates with a strong dependence of the surface character of the employed substrates by virtue of a surface-reaction-controlled regime with an alternating exposure of precursor and co-reactant molecules.

• Growing films using atomic layer deposition (ALD) is the most successful process among elemental metal ALD processes. Although numerous experimental approaches have been explored and some have demonstrated worthwhile results, the fundamental nature and the interfacial interaction mechanism of the film–substrate interface are not yet fully understood. The complex phenomena of interfacial interactions in the tungsten/alumina heterostructure systems inspired us to apply deep computational evaluations using first-principles calculations to explore the interfacial interactions at the atomic level, where the phenomena cannot be directly observed through experimental analysis. Underlying surfaces can significantly affect the subsequent film layers, altering device performance. 

• We demonstrate such a framework by combining density functional theory (DFT) calculations with ab initio molecular dynamics (AIMD) to determine the preference between amorphous and crystalline phases. We present ab initio evidence for the thin film growth mechanism of the amorphous phase over crystalline phases from the perspective of thermodynamic stabilization at the nanoscale. The computational framework described here can be widely applied to determining strategies to design effective conductive films in more complex interfacial systems and synthesize complex alloy films. Our computations offer a novel strategy for designing thin films with outstanding thermal and chemical stability as well as other new chemical and physical characteristics.

• A detailed computational fluid dynamics (CFD) model of the ALD reactor is developed using a finite-volume-based code and validated. It accounts for the transport processes within the feeding system and reaction chamber. The simulated precursor spatiotemporal profiles assuming no ALD reaction are used as boundary conditions in modeling diethylzinc reaction/diffusion, the predictions of which agreed with experimental electron microscopy measurements. Further simulations can confirm that the present deposition flux is much less than the upper limit of flux, below which the decoupling of reactor/substrate is an accurate assumption. The modeling approach demonstrated here allows for the design of ALD processes for thin-film formation including the synthesis of metal–organic framework.

• Neuromorphic devices and systems have attracted great attention as next-generation computing due to their high efficiency in processing complex data. So far, they have been demonstrated using both machine-learning software and complementary metal-oxide-semiconductor-based hardware. The emerging paradigm of neuromorphic computing is inspired by neural networks of the brain and based on energy-efficient hardware for information processing. To create devices that mimic what occurs in our brains’ neurons and synapses, the scientific community must overcome a fundamental molecular engineering challenge: how to design devices that exhibit controllable and energy-efficient transition between different resistive states triggered by incoming stimuli.

• The revolutionary development of machine learning (ML), data science, and analytics, coupled with its application in material science, stands as a significant milestone of the scientific community over the last decade. Investigating active, stable, and cost-efficient catalysts is crucial for oxygen evolution reaction owing to the significance in a range of electrochemical energy conversion processes. In this work, we have demonstrated an efficient approach of high-throughput screening to find stable transition metal oxides under acid condition for high-performance oxygen evolution reaction (OER) catalysts through density functional theory (DFT) calculation and a machine learning algorithm. A methodology utilizing both the Materials Project database and DFT calculations was introduced to assess the acid stability under specific reaction conditions.  Through these approaches, we not only streamline the choice of the promising electrocatalysts but also offer insights for the design of varied catalyst models and the discovery of superior catalysts.  

• Using the Google Cloud Platform, tens of millions of documents are extracted and the desired information is statistically analyzed and presented through natural language processing analysis. We plan to classify battery materials with excellent performance, change trends in main topics in battery literature each year, and battery synthesis recipe extraction techniques using the latest natural language processing technique (Google BERT).