Driven by the demands of AI computing and green data centers, AI-integrated AC-DC immersion liquid-cooled energy storage systems have become a cornerstone for high-efficiency, high-density power management. Their power conversion and management subsystems, serving as the "energy heart" of the entire unit, must provide efficient, reliable, and precise power delivery and switching for critical loads such as PFC stages, DC-DC converters, pump drivers, and auxiliary circuits. The selection of power MOSFETs is pivotal in determining the system's conversion efficiency, power density, thermal performance under immersion, and long-term reliability. Addressing the stringent requirements of immersion cooling systems for ultra-high power density, exceptional thermal conductivity, electrical safety, and EMI control, this article reconstructs the MOSFET selection logic centered on scenario-based adaptation, providing an optimized, ready-to-implement solution.
I. Core Selection Principles and Scenario Adaptation Logic
Core Selection Principles
Voltage & Safety Margin: For AC-DC stages (e.g., PFC) handling rectified high voltage, MOSFET voltage ratings must withstand peak line voltages and transients with ample margin (e.g., ≥800V for 3-phase AC). For DC-DC bus and pump drives, margins should accommodate switching spikes and fault conditions.
Ultra-Low Loss is Paramount: Prioritize devices with minimal combined conduction loss (low Rds(on)) and switching loss (low Qg, Qrr). This is critical for maximizing efficiency in always-on systems and minimizing heat generation within the immersion fluid.
Package & Immersion Compatibility: Select packages (TO-220, TO-247, TO-220F, DFN) that offer robust construction, reliable isolation, and compatibility with immersion coolant. Thermal performance must be evaluated in conjunction with the liquid cooling medium.
Reliability Under Continuous Stress: Devices must be rated for 24/7 operation at elevated case temperatures possible before heat is carried away by coolant. Consider avalanche ruggedness, gate oxide reliability, and resistance to corrosion or electrochemical migration in immersed environments.
Scenario Adaptation Logic
Based on the core power flow and load types within the AI immersion energy storage system, MOSFET applications are divided into three primary scenarios: High-Voltage AC-DC Front-End (Power Intake), Intermediate Bus & Pump Drive (Power Distribution & Thermal Management), and Low-Voltage, High-Current Point-of-Load (POL) Conversion (Core Power Delivery). Device parameters and characteristics are matched accordingly.
II. MOSFET Selection Solutions by Scenario
Scenario 1: High-Voltage AC-DC Front-End (e.g., 3-Phase PFC Stage) – Power Intake Device
Recommended Model: VBM18R10S (Single-N, 800V, 10A, TO-220)
Key Parameter Advantages: Utilizes SJ_Multi-EPI (Super Junction) technology, achieving a balanced performance with 800V breakdown voltage and Rds(on) of 600mΩ at 10V Vgs. The 10A current rating is suitable for multi-parallel configurations in high-power PFC circuits.
Scenario Adaptation Value: The 800V rating provides a strong safety margin for 3-phase applications (rectified voltage ~565V). The TO-220 package offers excellent creepage and clearance distances, beneficial for high-voltage PCB design, and its robust mechanical structure is suitable for immersion. Low gate charge inherent to SJ technology facilitates high-frequency switching, improving PFC stage power density and efficiency.
Scenario 2: Intermediate DC Bus Conversion & Coolant Pump Drive – Distribution & Thermal Management Device
Recommended Model: VBP1606S (Single-N, 60V, 150A, TO-247)
Key Parameter Advantages: Features an extremely low Rds(on) of 5mΩ at 10V Vgs and a high continuous current rating of 150A using advanced Trench technology.
Scenario Adaptation Value: The ultra-low conduction loss minimizes heat generation in the critical power path of the DC-DC intermediate bus converter (e.g., 48V to 12V). The high current capability and robust TO-247 package make it ideal for driving high-power immersion coolant pumps, ensuring reliable thermal management. Its parameters support high-frequency synchronous rectification or half-bridge topologies, optimizing converter efficiency.
Scenario 3: Low-Voltage, High-Current Point-of-Load (POL) Conversion – Core Power Delivery Device
Recommended Model: VBQF1202 (Single-N, 20V, 100A, DFN8(3x3))
Key Parameter Advantages: Employs Trench technology to achieve an ultra-low Rds(on) of 2mΩ at 10V Vgs and 2.5mΩ at 4.5V Vgs. Rated for 100A continuous current in a compact DFN package.
Scenario Adaptation Value: The ultra-low Rds(on) is critical for minimizing loss in high-current POL converters (e.g., 12V/48V to sub-1V for AI processors/ASICs). The tiny DFN8(3x3) footprint enables extremely high power density on the PCB, which is essential for immersion systems where space near processors is premium. Its low gate threshold voltage (0.6V) allows for efficient drive from low-voltage controller ICs.
III. System-Level Design Implementation Points
Drive Circuit Design
VBM18R10S: Requires a dedicated high-side gate driver IC with sufficient drive capability and isolation where needed. Careful attention to minimizing parasitic inductance in the high-voltage switching loop is crucial.
VBP1606S: Pair with a high-current gate driver. Use low-inductance gate drive paths and consider using a negative turn-off voltage for faster switching and better noise immunity in pump drive applications.
VBQF1202: Optimize layout for minimal power loop inductance. Use a driver placed very close to the MOSFET. Multi-phase interleaving is recommended for POL applications to manage current and thermal stress.
Thermal Management within Immersion
Graded Heat Transfer Strategy: All packages rely on the immersion coolant as the primary heat sink. Ensure PCB designs facilitate efficient heat transfer from the device package to the board and then to the coolant. The TO-247 and TO-220 packages offer good thermal mass.
Derating in Liquid: While immersion cooling is highly effective, derating guidelines should still be followed based on the estimated junction-to-coolant thermal resistance. Monitor for potential local fluid heating.
EMC and Reliability Assurance
Immersion-Specific EMI: The dielectric fluid can alter parasitic capacitances. Careful snubber design (RC across drain-source for VBM18R10S) and input filtering are essential to meet EMI standards.
Protection for Immersion: Ensure all gate drive circuits are protected against transients. Use gate resistors and TVS diodes. Select conformal coatings or potting materials compatible with the specific immersion coolant to prevent long-term degradation or galvanic corrosion.
IV. Core Value of the Solution and Optimization Suggestions
The power MOSFET selection solution for AI immersion liquid-cooled energy storage systems, based on scenario adaptation logic, achieves optimal device matching from high-voltage intake to ultra-high-current POL delivery. Its core value is threefold:
Maximized System Efficiency for PUE Reduction: By selecting specialized devices—SJ MOSFETs for high-voltage switching, ultra-low Rds(on) Trench MOSFETs for bus conversion and POL—conduction losses are minimized at every stage. This contributes directly to a superior Power Usage Effectiveness (PUE), reducing operational costs and heat load on the immersion system.
Enabling Ultra-High Power Density and Reliability: The use of compact, high-performance packages like DFN for POL stages allows for more power stages in less volume, crucial for compute-dense AI racks. The selected high-voltage and pump-drive MOSFETs offer robust electrical and mechanical characteristics, ensuring stable operation in the unique environment of immersion cooling, leading to higher system MTBF.
Balanced Performance and Cost-Effectiveness: This solution leverages mature, high-volume MOSFET technologies (SJ, Trench) that offer an excellent balance of performance and cost. Compared to emerging wide-bandgap solutions, it provides a more immediately cost-effective and supply-chain-stable path to building high-performance immersion-cooled energy storage systems, accelerating time-to-market.
In the design of AI-integrated immersion-cooled energy storage systems, power MOSFET selection is a foundational element for achieving unprecedented levels of efficiency and power density. This scenario-based selection solution, by aligning device capabilities with specific subsystem requirements and incorporating design considerations for the immersion environment, provides a comprehensive technical blueprint. As AI computing demands escalate, future exploration will likely focus on the integration of silicon carbide (SiC) MOSFETs for the highest voltage/hardest switching stages and the development of intelligent, immersion-optimized power modules, paving the way for the next generation of ultra-efficient, high-density data center infrastructure.

Detailed Topology Diagrams