Jae Yong Song

Transverse thermoelectric devices offer a promising alternative to conventional longitudinal architectures due to their simplified, scalable layouts and enhanced mechanical durability. However, their practical deployment has been fundamentally hindered by low output power. Here, we demonstrate a geometric design strategy that increases transverse output power by redistributing the internal temperature field through mirror-symmetry breaking, thereby activating the conventional longitudinal Seebeck response in the transverse direction. We term this phenomenon the geometry-induced transverse Seebeck effect (G-TSE) and develop a topology-optimization framework to systematically design thermoelectric leg architectures that maximize this effect. As a proof of concept, we experimentally validate this strategy using an additively manufactured polycrystalline Fe–50Ni alloy leg, achieving a 998-fold enhancement in transverse output power compared to a standard cuboid geometry under the same applied heat flux. Numerical simulations for conventional Seebeck materials further indicate that this approach can bring transverse devices close to representative longitudinal thermoelectrics in volumetric power density normalized by the square of the applied heat flux. These results establish geometry as a design degree of freedom for transverse thermoelectric energy conversion and provide a systematic route to devices adapted to heat sources of different shapes.

This study presents a physics-based, scale-invariant methodology for reconstructing high-resolution power density maps from infrared (IR) thermography of neural processing unit (NPU) chips. The proposed approach combines the Laplacian of the temperature field with discrete cosine transform (DCT)-based spatial filtering to simultaneously enhance spatial resolution and suppress measurement noise. The reconstructed power maps are validated against three-dimensional thermal simulations using an empirically calibrated effective thermal conductivity, achieving a mean absolute error (MAE) below 2 °C. To overcome the limitations of bare-die measurements under high-power operation, a regression-based extrapolation model is developed to predict power distributions at a high operating frequency of 700 MHz beneath a metal heat spreader, with an MAE of 0.77 °C. The scale-invariance and spatial resolution of the method are further validated using a micro-heater with fine-pitch serpentine structures. An optimal spatial filtering condition is determined via coefficient of variation (CV) minimization, yielding a normalized cut-off frequency (NCF) of 0.11. The proposed method provides a robust and computationally efficient approach for quantitative power mapping and thermal analysis of advanced semiconductor systems.

Black phosphorus (BP) has a low electrical conductivity and a high thermal conductivity, despite its narrow band gap and high Seebeck coefficient. Here, it is experimentally demonstrated that the Sn atoms are substitutionally doped into BP polycrystals and that multiscale microstructural defects, such as point defects, dislocations, and amorphous phases, as phonon scattering sites, are independently incorporated into them. The Sn doping into the BP polycrystal increases the carrier concentration up to 3.7 × 10¹⁸ cm^-3 at 300 K without a significant degradation of the Seebeck coefficient and effectively decreases the thermal conductivity due to the phonon-impurity scattering. The multiscale defects (point defects, dislocations, grain boundaries, amorphization) synergistically suppress the lattice thermal conductivity (from 13 to 6.3 W/m K) with a decoupling of the electronic transport. As a result, the thermoelectric figure-of-merit, ZT, is significantly enhanced more than 1 order of magnitude, by controlling the carrier concentration and the multiscale microstructural defects, independently.

In this work, hydrogen is introduced into deformed Ti alloys via cathodic electrolysis under various charging conditions, and the microstructures and mechanical properties of the specimens are examined. The presence of hydrides is verified using X-ray diffraction, scanning electron microscopy, and electron backscatter diffraction analysis based on a rescanning approach using local pattern averaging. Hydrogen charging induces changes in the microstructure, making it possible to determine the hydrogen penetration depth. Vickers hardness tests are then performed to assess the extent of hydrogen penetration in the Ti alloys. Compression tests are conducted to further understand the impacts of hydrides on the mechanical reliability of hydrogen-charged Ti alloys. To reinforce the experimental findings, numerical simulations are conducted utilizing experimental data to comprehend hydrogen diffusion in the Ti alloys. Subsequently, the hydrogen diffusion mechanism in the hydrogen-charged specimens is verified through the obtained results.

The corrosion mechanism and kinetics of the silver-coated conductive yarn (SCCY) used for wearable electronics were investigated under a NaCl solution, a main component of sweat. The corrosion occurs according to the mechanism in which silver reacts with chlorine ions to partly form sliver chloride on the surface of the SCCY and then the local silver chloride is detached into the electrolyte, leading to the electrical disconnect of the silver coating. Thus, the electrical conductance of the SCCY goes to zero after 2.7 h. The radial part-coating of gold, which is continuously electrodeposited in the longitudinal direction on the SCCY but is partly electrodeposited in the radial direction, extends the electrical conducting lifetime up to 192 h, despite the corrosion rate increasing from 129 to 196 mpy (mils per year). Results show that the gold partly-coating on the SCCY provides a current path for electrical conduction along the longitudinal direction until all the silver underneath the gold coating is detached from the SCCY strands, which creates the electrical disconnect. Based on the corrosion behavior, i.e., local oxidation and detachment of silver from the SCCY, the gold part-coating is more cost effective than the gold full-coating electrodeposited on the entire surface for electrically conducting SCCY.

Solution‐phase exfoliated graphene has always been an attractive material for flexible thermoelectric applications, but traditional oxidative routes suffer from poor flake quality and a lack of quality doping techniques to make complementary n‐type and p‐type films. Here, it is demonstrated that by changing the adsorbed surfactant during the intercalation‐exfoliation process (polyvinylpyrrolidone for n‐type, pyrenebutyric acid for p‐type), both extremely high electrical conductivity (3010 and 2330 S cm⁻¹) and high Seebeck coefficients (53.1 and −45.5 µV K⁻¹) can be achieved. The result is that both of these films show remarkable power factors, over 600 µW m⁻¹ K⁻² at room temperature, which is over an order of magnitude better than that in previous works demonstrating complementary n‐type and p‐type graphene thermoelectric films. Based on these films, a full all‐graphene thermoelectric device is constructed as a proof of concept, where a peak power of 5.0 nW is recorded at a temperature difference of 50 K.