Keynote Speeches

Using solar-driven photocatalytic seawater splitting process to produce simultaneously hydrogen and oxygen has attracted enormous attention as a promising prospect in the future to create green chemical fuels due to its low cost, friendly environment, and especially taking advantage of abundant unlimited resources from sunlight and seawater. Heterojunction Rh@Cr₂O₃/SrTiO₃:Al (RCSTOA) photocatalyst synthesized through flux method combined with photo deposition treatment possessed unique physiochemical properties and high seawater-splitting performance. The important thin layers of Rh@Cr₂O₃:Al core-shell structure (5-20 nm) strongly intimate with a core-photocatalytic SrTiO₃:Al resulted in the vigorous increase of the seawater splitting process without using sacrificial agents. The separated H₂ evolution from O₂ through the photocatalytic seawater splitting was also carried out in the twin photoreactor under simulated sunlight irradiation as shown in Fig. 1. The highest performance RCSTOA-CBS-IO₃^-/I^--(Neosepta or FAA) system exhibited at pH 10.0 for reduction side and pH 4.0 for oxidation side achieved of 544/183 µmol.g^-1 and 721/280 µmol.g^-1, respectively (Fig. 2). These results are expected to extend the advanced seawater splitting engineering and the H₂/O₂ separation in this field. This study stimulates further exploration of the heterojunction photocatalysts and constructs the high efficiency of the artificial photosynthesis system.

Keywords: Seawater Splitting, Photocatalysis, SrTiO₃, Solar Hydrogen, Sunlight Harvesting

Coming soon.

Electrochemical deionization (ECDI) systems are attractive desalination devices that achieve relatively high salt adsorption capacity (SAC) with acceptable energy consumption, which can replace or be a promising complement to other desalination techniques. The ion-removing mechanisms of the ECDI system can be divided into two distinct categories [1]: (1) formation of the electric double-layer (EDL) on the surface of the electrodes to store ions and (2) utilizing Faradaic reactions through charge compensation, ion intercalation, surface redox binding, and conversion reactions. Based on these mechanisms, three dissimilar materials can be categorized, leading to six different combinations of the ECDI systems which are EDL//EDL, pseudocapacitive//pseudocapacitive, EDL//pseudocapacitive, EDL//battery, battery//battery and pseudocapacitive//battery. In general, the systems utilizing Faradaic reactions on one or both electrodes exhibit higher SAC in comparison with the pure EDL systems. In addition, Faradaic materials with the unique “memory effect” [2] and high reversibility generally exhibit high salt-removal capacity and rate which are beneficial to the practical application. However, most ECDI systems are only suitable for brackish water desalination, and the energy consumption is significantly increased with the salt concentration of brine water without considering their short cycle life. Here, we will give the general guidelines for constructing electrochemical deionization systems with high capacity and high rate of salt removal but without any membranes. The examples will include MnO₂//PPy (PPy: polypyrrol) [1,2], CuHCF@CNT//PPy (CuHCF: copper hexacyanoferrate@carbon nanotube) [3], and PPy-p-TS//PPy-ClO₄ (PPy-p-TS: 4-methylbenzene-sulfonic acid-doped PPy) [4] systems.

Keywords: energy storage materials, electrochemical desalination, faradaic reaction, memory effect, membrane-free deionization devices

References

1. C.-C. Hu, et al., “A highly efficient faradaic desalination system utilizing MnO₂ and polypyrrole-coated titanium electrodes”, Desalination 498 (2021) 114807.
2. C.-C. Hu, et al., “Construction of an inverted-capacitive deionization system utilizing pseudocapacitive materials”, Electrochemistry Communications 104 (2019) 106486.
3. C.-C. Hu, et al., “A high-capacity hybrid desalination system using battery type and pseudocapacitive type electrodes”, Desalination 545 (2023) 116160.
4. C.-C. Hu, et al., “Dopant-designed conducting polymers for constructing a high-performance, electrochemical deionization system achieving low energy consumption and long cycle life”, Chemical Engineering Journal 457 (2023) 141373.

To construct a new academic discipline of nano/micro-device science, we are establishing the following five elemental technologies: "MEMS," "nano/micro fabrication," "3D electronics packaging technology," "surface modification technology," and "biomaterials. In this conference, the following research results and applications related to these elemental technologies will be presented.

”Development of Cu hemispherical filter with many micro-sized through holes for application to high-precision soft X-ray analyzer”. It was installed in the Spring-8 soft x-ray analyzer, and the detection sensitivity was 200 times higher than that of conventional measurements. The grid is made of Cu and has a hemisphere with a diameter of 90 mm, a film thickness of 80 μm, and 60 μm diameter micro holes with 800,000 through-holes.

”Research on low-dislocation GaN substrate fabrication technology combining UV (Ultraviolet) nanoimprinting and crystal growth”. To increase the external quantum efficiency of LED light, we focused on the dislocation of GaN and proposed and developed a method to reduce the dislocation during crystal growth. We have succeeded in creating thin GaN substrates with fewer dislocation by combining nanoimprinting, dry etching, MOCVD, and HVPE methods. Since it is now possible to reduce dislocation, this method is now being used in research and development of GaN substrates for power devices.

”Research on TGV (Through Glass Via) substrate fabrication technology for I-type structure with high hermeticity”. MEMS packaging technology with through glass via (TGV) on a transparent glass substrate with I-type electrode structure was studied. First, through holes were fabricated by laser processing. Next, submicron Au particles were embedded in the through-hole to form the electrode. The substrate with the TGV structure and the lid substrate were bonded in a vacuum to ensure that the device was hermetically sealed.

”Research on Au-Au bonding technology by combining VUV (Vacuum Ultraviolet) surface treatment and diffusion methods”. We confirmed that Au-to-Au diffusion bonding at low temperatures was possible by removing the carbon remaining on the surface of the Au film using VUV/O3.

”Development of an actuator with two functions using soft materials”. A soft PDM (dimethylpolysiloxane) structure was fabricated using a mold made by a 3D printer. Next, electrodes were fabricated on the paper using an inkjet printer, and an actuator was fabricated by bonding the electrodes and the structure using plasma. The actuator has two functions: proximity sensing and strain gauge.

”Development of Augmented Reality (AR) Glass with 60° FOV (Field of View)”. The developed hands-free AR glasses have a nanostructured light guide path on the surface of the glass. In addition, there is a projector with a micro-LED as a light source. The nanostructures were fabricated using nanoimprinting, resulting in a large FOV of 60°.

”Research on regenerative medicine using fish scale collagen and MEMS technology”. Research on regenerative medicine technology for human oral epithelial mucosa using fish scale collagen, which has no effect on the human body, was successfully performed in vitro by combining MEMS technology with fish scale collagen to produce oral epithelial mucosa.

Keywords: MEMS, nano-micro fabrication, Electronics packaging, Surface treatment, bio-material

Two-dimensional (2D) semiconducting transition metal dichalcogenides (TMDs) represent the ultimate thickness for scaling down channel materials. They provide a tantalizing solution to push the limit of semiconductor technology nodes in the sub-1 nm range. One key challenge with 2D semiconducting TMD channel materials is to achieve large-scale batch growth on insulating substrates of single crystals with spatial homogeneity and compelling electrical properties. Recent studies have claimed the epitaxy growth of wafer-scale, single-crystal 2D TMDs on a c-plane sapphire substrate with deliberately engineered off-cut angles. It has been postulated that exposed step edges break the energy degeneracy of nucleation and thus drive the seamless stitching of mono-oriented flakes. Here we show that a more dominant factor should be considered: in particular, the interaction of 2D TMD grains with the exposed oxygen-aluminium atomic plane establishes an energy-minimized 2D TMD-sapphire configuration. Reconstructing the surfaces of c-plane sapphire substrates to only a single type of atomic plane ( plane symmetry) already guarantees the single-crystal epitaxy of mono layer TMDs without the aid of step edges. Electrical results evidence the structural uniformity of the monolayers. Our findings elucidate a long-standing question that curbs the wafer-scale batch epitaxy of 2D TMD single crystals-an important step towards using 2D materials for future electronics. Experiments extended to perovskite materials also support the argument that the interaction with sapphire atomic surfaces is more dominant than step-edge docking.

Keywords: 2D Semiconductors, epitaxy growth, single crystal

Figure 1. Computer-generated atomic structure of a c-plane sapphire. Each A and B atomic plane with an exposed surface exhibits a mirror-symmertic, threefold atomic arrangement. Surface planes of the A and B slabs stack alternatingly along the a-axis whereas the facetd step has the height of strictly one slab (c₀/7 = 2.17 Å)

Recently, the transient liquid phase (TLP) bonding process has become a promising method in advanced electronic packaging. Full intermetallic compounds (IMCs) joints provide good strength and reliable high-melting-point phase after bonding. Furthermore, full intermetallic compound bonding is more resistant to electromigration and shear fracture than Sn-rich solder. The integrity of IMCs can make sure of fixed space between stacked chips and avoid collapse due to solder re-melting. However, several issues may emerge when the structure of joint was dominated by IMCs such as coarse grains, phase boundaries, voids formation, preferred orientation and brittle nature of IMCs. Thus, controlling the growth and the properties of IMCs is critical to tackle these issues.

In this study, the composition of under bump metallization was optimized to alter the microstructure and to enhance mechanical and thermal properties of micrbump. Cu-Ni, Cu-Zn and Cu-Ni-Zn substrates were used to influence the grain structure and the associated strengthening mechanism would be addressed. After utilization of different alloyed substrate, grain sizes were significantly refined and orientation was diversified, which can reduce the effect from coarse grains and preferred orientation. Ni and Zn atoms would act as nucleation sites and more grains would form during reflow. In addition, the substitution of lattice position in Cu₆Sn₅ by solute atoms may cause lattice distortion, which can diverse the orientation of Cu₆Sn₅. The random orientation of grain can reduce stress concentration and hinder crack propagation. Moreover, in Cu-Zn or Cu-Ni-Zn substrate, a Zn-rich layer can inhibit the interdiffusion between substrate and solder. The void formation from Kirkendall voids and phase boundary from phase transformation can be thus reduced.

The mechanical test showed that the reliability increased after appropriate element addition in substrates. The strengthening mechanisms were mainly attributed to grain refinement and c-axis diversification. After shear testing, fractured samples were further analyzed to reveal weak points in the joints. With the assistance of EPMA and EBSD, The correlation between fracture paths and grain structure is established. In conventional Cu/Solder/Cu structure, the columnar grains with preferred orientation would demonstrated intragranular fracture. However, after grain refinement, large area of grain boundaries would obstruct the crack propagation. The fracture paths revealed partly intergranular fracture and partly intragranular fracture. Furthermore, the existence of voids also severely affect the fracture behavior. Thus, it is crucial to understand the weak points in the joints and to eliminate them. To sum up, the modification in grain structure has exhibited potential in strengthening the full intermetallic joints, and such method may be further utilized in other solder systems.

Keywords: Micro-bump; Microstructure; Intermetallic compounds; Grain orientation; EBSD

One key challenge with two-dimensional semiconducting transition metal dichalcogenides (2D-TMD) channel materials or insulating hexagonal boron nitride (hBN) for industry applications is the large-scale batch growth on substrates with continuous single crystallinity, and spatial homogeneity. To achieve large-area and grain boundary-free growth of 2D materials, it is critical to understand the fundamentals of the substrate-2D interaction to enable the mono-orientation of these 2D domains.

Taking a few examples including 2D TMD, hBN, and perovskites, we like to discuss the factors that control the epitaxy. Detailed analysis will be provided to illustrate the following points. (1) In the situation that 2D seeds strongly bind to the atomic substrates edges (eg. hBN on Cu(111); TMD on Ga₂O₃(100)), seed nucleation and orientation are controlled by the step edges. (2) In the circumstance that 2D seeds do not strongly bind to the step edges (eg. TMD on C- or M-sapphire; TMD on Ga₂O₃ (001)), the epitaxy of 2D is governed by the atomic structures (and symmetry) of the substrates. Hence, the reconstruction the surfaces of substrates to only a single type (symmetry) is necessary for the single-crystal epitaxy of monolayer TMDs without the aid of step edges. (3) Precursors may react with the substrate surfaces to change the substrate atomic structures and thus the orientation of 2D materials. For example, the surface sulfurization of sapphire or surface metallization of Ga₂O₃ drastically change the orientation preference. In summary, the substrate symmetry, step edges and chemical modification of substrates (or edges) need to be taken into consideration when reconstructing the substrates for successful epitaxial growth of 2D materials.

Keywords: 2D materials, CVD, Growth, Transition Metal Dichalcogenides, Epitaxy

In this presentation, we examine the implications of high entropy alloys (HEAs) for sustainable development. Our goal is to clarify how HEAs could be designed to reduce adverse societal and environmental impacts, while also providing insight for informed alloy development. Utilizing a palette of 18 elements, we evaluate their potential for nine distinct societal impacts, employing artificial intelligence to supplement gaps in the data. This enables us to scrutinize a dataset of over 340 pre-existing HEA compositions for their relative benefits and drawbacks. For those designed to withstand high temperatures, we evaluate their credentials against Nickel-based superalloys, the current benchmarks. For alloys aimed at room-temperature strength and ductility, we set them against traditional steels that are often employed for their load-bearing and damage tolerance characteristics. Our findings not only shed light on HEAs that excel in sustainability metrics but also those that lag behind, thereby aiding in the development of materials designed with societal impact in mind. Our results shed light not only those HEAs with promising sustainability profiles but also those falling short, thus offering a roadmap for the conscientious development of future materials.

High entropy alloys (HEAs) have opened a vast alloy composition field for materials design, and some alloys have demonstrated excellent mechanical properties with reasonable ductility. In addition, lower mobility of alloying elements in the alloys could maintain microstructural stabilities and mechanical properties at elevated temperatures. Our research group have designed and evaluated some HEAs for high temperature applications, which are refractory high entropy alloys (RHEAs), platinum-group-metals-based high entropy alloys (PGM-HEAs) and high entropy superalloys (HESAs) in collaboration with National Tsing Hua University (NTHU), University of Bordeaux (UB), and National Institute for Materials Science (NIMS).
In this talk, some research highlights in terms of HEAs designed for high temperature use are introduced as follows.

(i) Development of Ni-rich HESAs with superior mechanical properties.
HESAs have been proposed to design alloy composition to lie in the high entropy field and to have coherent g and g’ two-phase microstructure. Some alloys demonstrated higher yield stress with higher ductility at 750°C. Structure-property relationship of HEAs will be discussed.
(ii) Development of PGM-HEAs

High temperature hardness and oxidation behavior of Ir-rich PGM-HEAs will be introduced. The designed Ir-rich PGM-HEAs possess FCC and L1₂ two-phase structure and interestingly, much improved oxidation resistance compared to conventional Ir-based alloys.

(iii) Elemental Effect on oxidation behavior of RHEAs.
The effect of alloying elements on the oxidation behavior of bcc-structured NbTi-rich RHEAs was investigated. For instance, Mo can enhance the oxidation resistance once the protective oxide layer which can retard the diffusion of oxygen is formed, due to its low oxygen solubility. On the other hand, Mo itself forms volatile oxides thereby the presence of Mo on the surface would cause catastrophic damage to the alloys.

Keywords: high entropy alloys, gamma prime, high temperature strength, oxidation, platinum group metals.

Phase diagrams provide fundamental materials phase equilibria information, which is crucial for materials design and materials manufacturing. Henry-Louis Le Chatelier published the first carbon dioxide phase diagram in 1873. The phenomenological approach to phase diagram calculation, Calphad (Calculation of Phase Diagram) method, arose in the 1970s and has made significant advances since then. Reliable calculated phase diagrams rely upon appropriate thermodynamic modeling, which depends on accurate experimental results [1,2]. First-principles calculations can determine thermodynamic data and are helpful when experimental data is lacking [3]. The fields of machine learning and artificial intelligence are rapidly expanding [4,5]. A machine learning approach was employed to predict formation enthalpy over a large scale of structure types. A surrogate model fitted to the existing density functional theory data provides a mathematical relationship between the structure and thermodynamic properties, thus accelerating the development of the thermodynamic description for a given materials system. Chen and Chang published a paper [6] on solidification path prediction, which is the first one to combine thermodynamic and kinetic models. Phase diagram calculation has gained further emphasis since it sits at the base of ICME (Integrated computational materials engineering), and the "materials genome" efforts have again put phase diagram calculation in the center of the stage. ICME integrates thermodynamic and various kinetic models to predict the materials properties in various application environments and manufacturing processes. Machine learning and artificial intelligence are employed in recent efforts of ICME, leading to significant improvements in kinetic modeling. With the improvement of kinetic modeling, accurate descriptions of phase diagrams become more critical in predicting materials properties. However, due to the complicated nature and lack of general data of materials Gibbs energies, reliable calculated phase diagrams still require accurate experimental measurements.

Keywords: Phase diagrams; Calphad; Experimental results; Applications: AI

References:

[1] Y. A. Chang, “Phase Diagram Calculations in Teaching, Research, and Industry”, Metallurgical and Materials Trasactions B, Vol. 37B, pp. 7-39, (2003).

[2] Y. A. Chang, S. Chen, F. Zhang, X. Yan, F. Xie, R. Schmid-Fetzer and W. A. Oates, “Phase diagram calculation: past, present and future”,  Progress in Materials Science, Vol. 49(3–4), pp. 313-345, (2004).

[3] S.-K. Lin, C.-K. Yeh, W. Xie, Y.-C. Liu and M. Yoshimura, "Ab initio-aided CALPHAD thermodynamic modeling of the Sn-Pb binary system under current stressing", Scientific Reports , Vol. 3, 2731(1-4), (2013).

[4] D, Morgan and R. Jacobs. "Opportunities and challenges for machine learning in materials science" , Annual Review of Materials Research, Vol. 50, pp. 71-103, (2020).

[5] W. Y. Wang, J. Li, W. Liu and Z.-K. Liu, “Integrated computational material sengineering for advanced materials:A brief review “, Computational Materials Science, Vol. 158, pp. 42-48, (2019).

[6] S.-W. Chen and Y. A. Chang, “Application of thermodynamic models in the calculation of solidication paths of Al-rich Al-Li alloys”, Meallurgical and materiasl transactions A, Vol. 23A, pp. 267-271, (1991).

Coming soon.

High-entropy alloys (HEAs) are a novel class of materials formed by mixing four or more different elements, each of which constitutes at least 5% of the alloy composition, resulting in a solid solution phase. In contrast to traditional alloys, HEAs offer a wide range of compositional adjustments, which can potentially lead to significant variations in their physical and chemical properties. Consequently, they have gained considerable attention in recent years. Due to the multi-component nature of HEAs, high mechanical strength is an expected property, and much research has focused on understanding the mechanisms and applications related to their mechanical properties. However, the electrical and optical properties of HEAs should also exhibit intriguing phenomena, yet there is limited literature exploring these aspects in depth. In this presentation, we will discuss the results of first-principles calculations applied to investigate the mechanical, electrical, and optical properties of HEAs. This analysis aims to shed light on the role of high-entropy effects in shaping the mechanical, electrical, and optical characteristics of these materials.

Innovation in materials is pivotal in numerous engineering applications. The synergy between experiments, computation, and artificial intelligence (AI) offers abundant possibilities for enhancing our comprehension and utilization of materials. Once trained by data from computation and experiments, machine learning models can achieve predictive power at unparalleled speeds. These rapid surrogate models learned from the data provide fast-forward modeling and make inverse design possible. The surge in generative AI and foundation models offer further possibility for accelerating materials innovation. In this talk, I will present our three ongoing efforts using artificial intelligence to improve how we model and design materials. The first focuses on the Deep Material Network (DMN), which aims to resolve a longstanding challenge in multiscale simulation of structures with heterogeneous materials (for example, composite materials). DMN, as a rapid surrogate model for materials homogenization, bridges the gap between the micro-scale, where deformation mechanisms occur, and the macro-scale, which is crucial for engineering and structural applications. I will present how the DMN can be generalized to encompass microstructural characteristics using a Graph Neural Network (GNN). The GNN-DMN extension allows us to model composite materials with different morphologies and polycrystalline materials such as metals. The second part of my talk focuses on generating 3D bioinspired microstructures with target properties. Distinct from engineering materials, biological structural materials are often composites of hard/brittle minerals and soft/ductile proteins arranged into complex microstructures with remarkable mechanical properties. Learning from these microstructures is critical for the design of high-performance materials with multiple functions. I will introduce a deep generative model based on transformer architecture. This model enables us to discern inherent patterns in intricate, ambiguously defined biological materials. Furthermore, the model allows us to create novel 3D microstructures with desired properties from limited high-resolution X-ray micro-computed tomography data. Finally, I will address how to adapt self-supervised techniques and millions of polymer data to train a foundation model for polymer design. The embedding and in-context training of the foundation model hold significant promise in addressing the longstanding challenge of scarce materials data.

Keywords: Deep Learning, Generative Artificial Intelligence, Composite, Bioinspired Materials, Polymers

The landscape of material science and manufacturing has transformed significantly with the advent of machine learning (ML). This toolkit of data driven methods accelerated the discovery and production of new materials by accurately predicting the complicated physical processes and mechanisms that are not fully described by existing material theories. Yet, with an array of intricate ML models at our disposal, the pressing question remains: Which ML algorithm is best suited for our needs? In this presentation, we aim to provide insights to strategically select appropriate models aligned with specific design challenges. We further segment material design challenges into: 1) deep learning based interpolation problem: ample training data capturing design space trends. 2) deep learning-based extrapolation problem: immense design space demanding more than just the initial training dataset. 3) limited data scenario: instances where only handful of dataset is available. 4) multi-fidelity datasets: a combination of concise, precise datasets and expansive, approximate ones. The most successful machine learning-based surrogate models and design approaches will be discussed for each case along with pertinent literature.

Fig. 1. Machine Learning Based Design Algorithms: A Diagrammatic Overview Categorized by Data Characteristics and Design Space Size

Keywords: machine learning, optimization, deep learning, Bayesian optimization, materials design

We employed first-principles calculations and MEAM modeling to investigate the short-range chemical ordering and dislocation behaviors in the NbTaTiV RHEA. Our first-principles calculations showed that the thermodynamic stability of this HEA can be greatly enhanced by increasing the number of Ta-V pairs while decreasing the number of V-V and Ta-Ti pairs within the alloy, indicating the tendency to form short-range chemical ordering in the NbTaTiV RHEA. Our MEAM simulations further showed that the average core energy of dislocations in the NbTaTiV RHEA is substantially lower than their constituent elements, indicating a relatively higher dislocation density with a reduced mobility in RHEA than in other BCC metals. It was also found that the average dislocation core energy in RHEA with local chemical ordering is comparable to that within the SQS structure. However, some core structures with extremely low energies were identified, which can effectively hinder the motion of dislocation to enhance the mechanical strength of the NbTaTiV RHEA.

Keywords：NbTaTiV RHEA, short-range chemical ordering, first-principles calculations, MEAM modeling