Search

Abstract This article explores the deep learning approach towards approximating the effective electrical and thermal conductivities of copper (Cu)-carbon nanotube (CNT) composites with CNTs aligned to the field direction. Convolutional neural networks (CNN) are trained to map the two-dimensional images of stochastic Cu-CNT networks to corresponding conductivities. The CNN model learns to estimate the Cu-CNT composite conductivities for various CNT volume fractions, interfacial electrical resistances, R c = 20 Ω–20 kΩ, and interfacial thermal resistances, R ″ t,c = 10−10–10−7 m2K/W. For training the CNNs, the hyperparameters such as learning rate, minibatch size, and hidden layer neurons are optimized. Without iteratively solving the physical governing equations, the trained CNN model approximates the electrical and thermal conductivities within a second with the coefficient of determination (R 2 ) greater than 98%, which may take longer than 100 min for a convectional numerical simulation. This work demonstrates the potential of the deep learning surrogate model for the complex transport processes in composite materials.

Abstract Large discrepancies between well-mixed reaction rates and effective reactions rates estimated under fluid flow conditions have been a major issue for predicting reactive transport in porous media systems. In this study, we introduce a framework that accurately predicts effective reaction rates directly from pore structural features by combining 3D pore-scale numerical simulations with machine learning (ML). We first perform pore-scale reactive transport simulations with fluid–solid reactions in hundreds of porous media and calculate effective reaction rates from pore-scale concentration fields. We then train a Random Forests model with 11 pore structural features and effective reaction rates to quantify the importance of structural features in determining effective reaction rates. Based on the importance information, we train artificial neural networks with varying number of features and demonstrate that effective reaction rates can be accurately predicted with only three pore structural features, which are specific surface, pore sphericity, and coordination number. Finally, global sensitivity analyses using the ML model elucidates how the three structural features affect effective reaction rates. The proposed framework enables accurate predictions of effective reaction rates directly from a few measurable pore structural features, and the framework is readily applicable to a wide range of applications involving porous media flows.

The geometrical design of thermoelectric legs in modules is key for sustainable power generation but can be hardly achieved by traditional fabrication process. Here, the authors develop an extrusion-based 3D printing process of Cu2Se thermoelectric materials for efficient power generation.

Thermoelectric devices are often rigid and do not adapt conformally to surfaces. Here, Park et al. prepare a Bi2Te3-based thermoelectric paint, containing a Sb2Te3chalcogenidometalate additive, that can be paint-brushed onto curved surfaces and form thermoelectric modules with good efficiencies.

Grain boundaries in polycrystalline materials may offer the opportunity to explore physical phenomena that do not normally occur within the crystal grains. Here, the authors show that twin boundaries in Bi2Te3works as an electron supply for the whole bulk material.

We present the deep learning model for internal forced convection heat transfer problems. Conditional generative adversarial networks (cGAN) are trained to predict the solution based on a graphical input describing fluid channel geometries and initial flow conditions. Without interactively solving the physical governing equations, a trained cGAN model rapidly approximates the flow temperature, Nusselt number (Nu), and friction factor (f) of a flow in a heated channel over Reynolds number ranging from 100 to 27,750. For an effective training, we optimize the dataset size, training epoch, and a hyperparameter ν. The cGAN model exhibited an accuracy up to 97.6% when predicting the local distributions of Nu and f. We also show that the trained cGAN model can predict for unseen fluid channel geometries such as narrowed, widened, and rotated channels if the training dataset is properly augmented. A simple data augmentation technique improved the model accuracy up to 70%. This work demonstrates the potential of deep learning approach to enable cost-effective predictions for thermofluidic processes. [ABSTRACT FROM AUTHOR]

We measure and model the dynamic temperature response of electrocaloric (EC) multilayer capacitors (MLCs) which have been recently highlighted as novel solid-state refrigerators. The MLC temperature responses depend on the operation voltage waveform, thus we consider three types of voltage waveforms, which include square, triangular, and trapezoidal. Further, to implement an effective refrigeration cycle, the waveform frequency and duty cycle should be carefully chosen. First, our model is fitted to the measurements to evaluate an effective EC power and thermal properties, and calculates an effective cooling power for an EC MLC. The prediction shows that for a MLC with a thermal relaxation time for cooling, trc, a square voltage waveform with a duty cycle of 0

Harman method is a rapid and simple technique to measure thermoelectric properties. However, its validity has been often questioned due to the over-simplified assumptions that this method relies on. Here, we quantitatively investigate the influence of the previously ignored parasitic thermal effects on the Harman method and develop a method to determine an intrinsic ZT. We expand the original Harman relation with three extra terms: heat losses via both the lead wires and radiation, and Joule heating within the sample. Based on the expanded Harman relation, we use differential measurement of the sample geometry to measure the intrinsic ZT. To separately evaluate the parasitic terms, the measured ZTs with systematically varied sample geometries and the lead wire types are fitted to the expanded relation. A huge discrepancy (~28%) of the measured ZTs depending on the measurement configuration is observed. We are able to separately evaluate those parasitic terms. This work will help to evaluate the intrinsic thermoelectric property with Harman method by eliminating ambiguities coming from extrinsic effects. [ABSTRACT FROM AUTHOR]

Controlling crystalline phases in polymorphic materials is critical not only for the fundamental understanding of the physics of phase formation but also for the technological application of forbidden, but potentially useful physical properties of the nominally unstable phases. Here, using tin oxide (SnO2) as a model system, we demonstrate a new way to enhance the mechanical hardness of an oxide by stabilizing a high-pressure dense phase through nitrogen integration in the oxide. Pristine SnO2 has a tetragonal structure at the ambient pressure, and undergoes phase transitions to orthorhombic and cubic phases with increasing pressure. Leveraging the enhanced reactivity of nitrogen in plasma, we are able to synthesize tin oxynitride (SnON) thin films with a cubic phase same as the high-pressure phase of SnO2. Such nitrogen-stabilized cubic SnON films exhibit a mechanical hardness of ~23 ± 4 GPa, significantly higher than even the nitride counterpart (Sn3N4) as the result of the shortened atomic distance of the denser, high-pressure cubic phase. Moreover, SnON has a heavily doped, n-type semiconducting property with a controllable optical bandgap. Our work will provide new opportunities to search for and to utilize beneficial, but hidden physical properties that exist in a particular phase stable only at extreme conditions. [ABSTRACT FROM AUTHOR]