A High-Voltage-Isolated MEMS Quad-Solenoid Transformer with Specific Insulation Barriers for Miniaturized Galvanically Isolated Power Applications.

The paper reports on high voltage (HV)-isolated MEMS quad-solenoid transformers for compact isolated gate drivers and bias power supplies. The component is wafer-level fabricated via a novel MEMS micro-casting technique, where the tightly coupled quad-solenoid chip consists of monolithically integrated 3D inductive coils and an inserted ferrite magnetic core for high-efficiency isolated power transmission through electromagnetic coupling. The proposed HV-isolated transformer demonstrates a high inductance value of 743.2 nH, along with a small DC resistance of only 0.39 Ω in a compact footprint of 6 mm2, making it achieve a very high inductance integration density (123.9 nH/mm2) and the ratio of L/R (1906 nH/Ω). More importantly, with embedded ultra-thick serpentine-shaped (S-shaped) SiO2 isolation barriers that completely separate the primary and secondary windings, an over 2 kV breakdown voltage is obtained. In addition, the HV-isolated transformer chips exhibit a superior power transfer efficiency of over 80% and ultra-high dual-phase saturation current of 1.4 A, thereby covering most practical cases in isolated, integrated bias power supplies such as high-efficiency high-voltage-isolated gate driver solutions.

A High-Performance MEMS Accelerometer with an Improved TGV Process of Low Cost.

High-performance MEMS accelerometers usually use a pendulum structure with a larger mass. Although the performance of the device is guaranteed, the manufacturing cost is high. This paper proposes a method of fabricating high-performance MEMS accelerometers with a TGV process, which can reduce the manufacturing cost and ensure the low-noise characteristics of the device. The TGV processing relies on laser drilling, the metal filling in the hole is based on the casting mold and CMP, and the packaging adopts the three-layer anodic bonding process. Moreover, for the first time, the casting mold process is introduced to the preparation of MEMS devices. In terms of structural design, the stopper uses distributed comb electrodes for overload displacement suppression, and the gas released by the packaging method provides excellent mechanical damping characteristics. The prepared accelerometer has an anti-overload capability of 10,000 g, the noise density is less than 0.001°/√Hz, and it has ultra-high performance in tilt measurement.

MEMS-Casting Fabricated Chip-Style 3D Metal Solenoidal Transformers towards Integrated Power Supply.

A silicon-chip-based 3D metal solenoidal transformer is proposed and developed to achieve AC-DC conversion for integrated power supply applications. With wafer-level micro electromechanical systems (MEMS) fabrication technique to form the metal casting mold and the following micro-casting technique to rapidly (within 6 min) fill molten ZnAl alloy into the pre-micromachined silicon mold, 45-turns primary solenoid and 7-turns secondary solenoid are fabricated in silicon wafers, where the two intertwining solenoids are located at inner deck and outer deck, respectively. Permalloy soft magnetic core is inserted into a pre-etched channel in the silicon chip, which is surrounded by the solenoids. The size of the chip-style transformer is as small as 8.5 mm × 6.6 mm × 2.5 mm. The internal resistance of the primary solenoid is 1.82 Ω and that of the secondary solenoid is 0.16 Ω. The working frequency of the transformer is 60 kHz. Combined with the testing circuit of the switch mode power supply, the DC voltage of 13.02 V is obtained when the input is 110 V at 50 Hz/60 Hz. Furthermore, the on-chip 3D solenoidal transformer is used for lighting four LEDs, which shows great potential for AC-DC power supply. The wafer-level fabricated chip-style solenoidal AC-DC transformer for integrated power supply is advantageous in uniform fabrication, small size and volume applications.

Double-Deck Metal Solenoids 3D Integrated in Silicon Wafer for Kinetic Energy Harvester.

A silicon-chip based double-deck three-dimensional (3D) solenoidal electromagnetic (EM) kinetic energy harvester is developed to convert low-frequency (<100 Hz) vibrational energy into electricity with high efficiency. With wafer-level micro electro mechanical systems (MEMS) fabrication to form a metal casting mold and the following casting technique to rapidly (within minutes) fill molten ZnAl alloy into the pre-micromachined silicon mold, the 300-turn solenoid coils (150 turns for either inner solenoid or outer solenoid) are fabricated in silicon wafers for saw dicing into chips. A cylindrical permanent magnet is inserted into a pre-etched channel for sliding upon external vibration, which is surrounded by the solenoids. The size of the harvester chip is as small as 10.58 mm × 2.06 mm × 2.55 mm. The internal resistance of the solenoids is about 17.9 Ω. The maximum peak-to-peak voltage and average power output are measured as 120.4 mV and 43.7 µW. The EM energy harvester shows great improvement in power density, which is 786 µW/cm3 and the normalized power density is 98.3 µW/cm3/g. The EM energy harvester is verified by experiment to be able to generate electricity through various human body movements of walking, running and jumping. The wafer-level fabricated chip-style solenoidal EM harvesters are advantageous in uniform performance, small size and volume applications.

Treatment of ill-balanced datasets of fMRI with Modified Fuzzy c-means Method.

In fMRI dataset, the population of actived voxels is always much less than the total population of the voxels, and that produced an ill-balanced dataset. Some methods, such as limiting the analysis to the gray matter voxels where the BOLD signal is expected and removing the voxels that is absolutely non-actived based on statistical criteria, have been used to treat the ill-balanced dataset. In this article, a new method, Modified Fuzzy c-means(MFc), has been proposed to treat the ill-balanced dataset of fMRI. The main difference from other statistical methods is that it is datadriven. iven. The MFc method is used to classify the voxels into two clusters with nearly the same population and all actived voxels are contained in one cluster. Thus we got nearly half voxels to analysis and the ill-balanced dataset can be treated. The efficiency of clustering analysis is also boosted.