Conference Agenda

Advances in the growth of thick-epitaxial 4H-SiC materials, device fabrication, device design, and medium-voltage electronic packaging have been key to the successful demonstration of >20-kV SiC technology. In collaboration with industrial and academic partners, the U.S. Combat Capabilities Development Command (DEVCOM) Army Research Laboratory continues its mission to develop advanced, single-die, high-voltage/high-power switches and diodes. This work presents preliminary measurements of recently fabricated high-voltage PiN diodes rated for 20 kV and 2 A. These devices have a thick epitaxial drift region of 180 μm and a chip area of 0.25 cm². In this report, a small number of these high-voltage SiC diodes were packaged in-house and evaluated to determine their steady-state and dynamic characteristics under various loads. These high-voltage diodes are intended to be scaled up and used in various pulsed power, directed-energy, and power grid applications.

This paper is a review study summarizing the development of electrical insulation materials for turbogenerators over the past 15 years, focusing on the influence of oxide nanofillers (MgO, SiO₂, TiO₂, Al₂O₃, ZnO) on electrical, thermal, and mechanical properties of composite insulation systems. The aim is to analyze published results, identify trends, and evaluate key challenges in manufacturing and industrial application.

Literature indicates that adding oxide nanoparticles to epoxy and polyurethane matrices can significantly improve breakdown strength, reduce dielectric loss, and enhance thermal conductivity. MgO and Al₂O₃ are effective in suppressing space charge, while SiO₂ and TiO₂ improve resistance to electrical treeing. ZnO shows specific behavior depending on concentration and polymer type. The review also emphasizes the importance of nanoparticle surface treatment and homogeneous dispersion for stable performance.

The section on manufacturing risks discusses agglomeration, residual moisture, and surface chemistry incompatibility, which may lead to increased losses and accelerated aging. Recommended practices include drying, controlled mixing, sonication, and viscosity control.

Based on long-term diagnostic studies, reliability of insulation systems in turbogenerators is limited by multifactor stresses (temperature, humidity, electric field). The review concludes that oxide nanofillers offer a promising route to improve lifetime and operational reliability, but industrial implementation requires interface control, reproducible processing, and robust in-service diagnostics.

In recent years, long-distance GIL has been widely used. During the on-site withstand voltage test of GIL, breakdowns can be easily caused. However, the existing method cannot locate the breakdown accurately and requires a large number of sensors. To solve the problems, this paper proposes a novel breakdown location method suitable for long-distance GIL. This method is based on the transient voltage measurement and its natural frequency. The relationship between the breakdown position and the natural frequency is analyzed in detail. The measured voltage waveforms during the on-site withstand voltage test of a long-distance GIL in China are given, and the validation results show that the proposed method not only can locate the breakdowns with high accuracy but also requires very few sensors.

In recent years, increasing demand for semiconductor devices has led to demand for mass production equipment that is low cost, high yield, and stable in operation. Excimer lasers are widely used as the light source in semiconductor manufacturing equipment. Excimer lasers generate ultraviolet light using a mixture of noble gases and halogen gases. In semiconductor manufacturing equipment, fluorine is used as the halogen gas, and this fluorine reacts with carbon in the gas to generate CF4 gas. Because CF4 gas reduces the output of excimer lasers, it must be removed to ensure stable equipment operation.

In this study, we investigated the removal characteristics of CF4 gas using pulsed corona discharge and evaluated the effect of applied voltage conditions on the CF4 gas removal rate. As a result, it was found that the removal rate of CF4 gas increased by applying a high voltage and a high repetition rate of pulse voltage.

The charge distribution of a dielectric surface backed by a metallic ground plane is a critical parameter contributing to many insulation-relevant physical processes, such as electrostatic discharge, triboelectric charging, and surface flashover. However, direct measurement of the charge distribution is challenging. Non-contact probes can determine the local field or the potential above a charged surface. Translating these measured quantities into an underlying surface charge distribution is non-trivial and is influenced by factors such as probe spacing, spatial resolution, and the overall measurement geometry. In this work, we explore the relationship between the surface charge distribution, the electric field external and internal to a dielectric, and the potential developed on the surface. A first‑principles formulation based on Maxwell’s equations yields analytical solutions for idealized geometries. These include cases in which a charged film can be approximated as an infinite plane or a finite circular region of constant charge density, provided that the sensing probe does not perturb the field. Numerical simulation enables analysis of realistic, finite geometries with non-uniform charge density, where the measurement is further influenced by the probe's dimensions and surroundings. These studies inform a methodology for mapping non-uniform charge densities of charged films backed by a ground plane via non-contact measurements with a Trek 341B electrostatic voltmeter. The probe response function is initially obtained using charged foils with prescribed geometry and potentials; subsequently, maps of the surface charge distribution are obtained from corona-charged films under a priori unknown conditions. The electrical breakdown limits governed by the maximum achievable surface charge and probe characteristics are discussed.

Power transformers constitute a critical component of modern power systems, ensuring reliable transmission and distribution of electrical energy from generation sources to downstream utility networks. The operational reliability of these assets strongly influences the stability of the entire power grid. Therefore, continuous condition monitoring is essential for the early detection of incipient faults and abnormal operating conditions. Among the available diagnostic techniques, Dissolved Gas Analysis (DGA) is widely recognized as one of the most effective methods for assessing transformer insulation degradation and internal fault conditions. However, conventional DGA based diagnostic approaches, although simple and convenient, often exhibit limited diagnostic accuracy when addressing complex or simultaneous fault scenarios.

To overcome these limitations, this study proposes an Artificial Intelligence (AI) driven multi-fault diagnostic framework for power transformers. The Duval Triangle Method (DTM) is employed to generate synthetic data for single fault conditions, while an extended DTM-based approach is utilized to synthesize datasets representing multiple fault scenarios for model training. For fault classification, the CatBoost algorithm is adopted due to its robustness and its ability to efficiently handle structured and categorical data. Furthermore, a Tabular Generative Adversarial Network (TabGAN) is employed to augment the testing dataset for multiple fault scenarios. The framework is evaluated using simulated datasets reported in the literature together with the TC10 Transformer Fault Dataset to assess the generalization capability and reliability of the proposed system. The proposed model achieved a diagnostic accuracy of 91%, demonstrating its effectiveness in identifying diverse transformer fault conditions. Overall, this study contributes to the development of an AI assisted diagnostic strategy that enhances the accuracy and practicality of multi-fault identification in transformer condition monitoring.