Various factors affect the interfacial thermal resistance (ITR) between two materials, making ITR prediction a high-dimensional mathematical problem. Machine learning is a cost-effective method to address this. Here, we report ITR predictive models based on experimental data. The physical, chemical, and material properties of ITR are categorized into three sets of descriptors, and three algorithms are used for the models. Those descriptors assist the models in reducing the mismatch between predicted and experimental values and reaching high predictive performance of 96%. Over 80,000 material systems composed of 293 materials were inputs for predictions. Among the top-100 high-ITR predictions by the three different algorithms, 25 material systems are repeatedly predicted by at least two algorithms. One of the 25 material systems, Bi/Si achieved the ultra-low thermal conductivity in our previous work. We believe that the predicted high-ITR material systems are potential candidates for thermoelectric applications. This study proposed a strategy for material exploration for thermal management by means of machine learning.

Presently, the most popular and economical commercial process of H 2 production is steam-methane reforming, which uses fossil fuels as the raw material and produces comparable amounts of CO 2 as the by-product. [1] It is definitely an environmental unfriendly and a non-sustainable H 2 production process, and development of green H 2 production is in urgent need. In this regard, renewable energy driven electrolytic water splitting has been gaining increasing popularity and is considered the most promising green H 2 production process for future hydrogen economy infrastructure. For electrolytic water splitting, H 2 is generated at cathodes from water reduction, the hydrogen evolution reaction (HER), and O 2 is generated at anodes from water oxidation, the oxygen evolution reaction (OER). To drive HER and OER at room temperature, a minimum thermodynamic cell voltage of 1.23 V is required. In practice, a cell voltage significantly higher than 1.23 V is required to overcome extra resistances existing in the water splitting system. [2] The extra cell voltage needed is the overpotential (η), which is the main target for electrocatalyst development to reduce its value for more efficient and thus more economically competitive hydrogen production for large-scale applications.Over the years, transition metals, particularly Fe, Ni, and Co based electrocatalysts, including alloys, [3] oxides, [4] phosphides, [5] sulfides, [6] etc. have been widely studied and exhibited excellent electrocatalytic performances toward water splitting reactions. Some of the metallic electrocatalysts showed excellent electrocatalytic activities for both HER and OER, and can be applied as bifunctional electrocatalysts for water splitting. Bifunctional electrocatalysts offer the advantage of material application convenience and many of them are multi-metal based materials such as NiFe and NiFeM. [7][8][9] Huang et al. fabricated NiFe nanotubes via a novel bubble-releasing assisted pulse electrodeposition method. These NiFe nanotubes exhibited outstanding bifunctional features in alkaline media with η 10 , η at 10 mA cm −2 , of 236 and 100 mV for the OER and HER, respectively. [7] Qin et al. used a hydrogen reduction method to coat NiFeMo alloy onto Ni foams, which showed impressive η 10 of 238 and 45 mV for the OER and HER, respectively. [8] Khalid et al. fabricated non-noble tri-metallic nanoparticles embedded Twinning is demonstrated to be an effective way of enhancing efficiencies of metallic catalysts toward electrolytic water splitting. Dendritic Cu possessing dense coherent nanotwin (NT) boundaries (NTCu-5nm) is successfully prepared with an organic-assisted electrodeposition at high pulse current densities. NT boundaries significantly improve electrocatalytic efficiencies and stability of NTCu-5nm over nanocrystalline Cu (NCCu), reducing overpotentials at 10 mA cm −2 for the oxygen evolution reaction (OER) from 378 to 281 mV and from 235 to 88 mV for the hydrogen evolution reaction (HER), with a small chronoamperometric decay of 5% after 100 ...

Temperature increase in the continuously narrowing interconnects accelerates the performance and reliability degradation of very large scale integration (VLSI). Thermal boundary resistance (TBR) between an interconnect metal and dielectric interlayer has been neglected or treated approximately in conventional thermal analyses, resulting in significant uncertainties in performance and reliability. In this study, we investigated the effects of TBR between an interconnect metal and dielectric interlayer on temperature increase of Cu, Co, and Ru interconnects in deeply scaled VLSI. Results indicate that the measured TBR is significantly higher than the values predicted by the diffuse mismatch model and varies widely from 1 × 10 −8 to 1 × 10 −7 m 2 K W −1 depending on the liner/barrier layer used. Finite element method simulations show that such a high TBR can cause a temperature increase of hundreds of degrees in the future VLSI interconnect. Characterization of interface properties shows the significant importance of interdiffusion and adhesion in TBR. For future advanced interconnects, Ru is better than Co for heat dissipation in terms of TBR. This study provides a guideline for the thermal management in deeply scaled VLSI.

The lithium-ion capacitor (LIC) is a novel energy storage device, pairing battery-type anodes with capacitor-type cathodes, capable of delivering high energy and power densities. The anode materials of excellent high-rate capability, however, are required to resolve the common kinetics imbalance issue for high-performance LICs. A simple one-step solvothermal, metal–organic framework (MOF) evolved process was developed to synthesize hollow porous α-Fe2O3 nanoparticles (α-Fe2O3 HPNPs) as an anode material of excellent high-rate capability for high-performance LICs. The α-Fe2O3 HPNP anode achieved an excellent high-rate capability and cycling stability, through accelerating lithium-ion diffusions with the porous shell and shortening lithium-ion diffusion paths and buffering large volume variations during cycling with the confined hollow space. The quantitative kinetic analyses showed that capacitive processes are the main contributor to the capacity generation of the α-Fe2O3 HPNP anode, making the α-Fe2O3 HPNP an excellent match with capacitor-type cathodes, glucose-derived carbon nanospheres (GCNS) of high specific surface areas, for the assembly of LICs. The α-Fe2O3 HPNP//GCNS LIC delivered a high energy density of 107 Wh kg–1 at 0.24 W kg–1 and maintained an adequate energy density of 86 Wh kg–1 at an extremely high power density of 9.68 kW kg–1. Moreover, it exhibited a high capacity retention of 84% after 2500 cycle operations at 1 A g–1. Both materials and nanostructure of electrodes play a key role for high-performance LICs, and the hollow porous nanoparticulate structure is proven to be an advantageous nanostructure for the anode materials of LICs.

No abstract