Answers are grounded in this paper and cite their sources.
Power electronic devices, which are becoming ubiquitous in myriads of emerging engineering systems, generate localized heating that exceeds 1 kW/cm². Conventional air cooling strategies are inadequate to dissipate such highly concentrated thermal loads. This has created a major challenge that has become a severe thermal management bottleneck. In this work, we experimentally investigated and numerically studied the true potential of direct immersion cooling, which is lauded as all-encompassing versatile thermal management solution for ultra-high heat flux power electronics. The experiments are compared with a conjugate heat transfer model that is developed using ANSYS Icepak. Our single-phase cooling approach with two dielectric fluids demonstrates a heat transfer coefficient ≈ 1100 W/m²·K, representing a 43 × enhancement over convective air cooling (≈26 W/m²·K). At nearly identical junction temperature, we managed to dissipate ≈ 26 W using single-phase immersion cooling, but only ≈ 1.4 W via convective air cooling, further reinforcing the superior heat removal potential of immersion cooling over air cooling. Importantly, our experiments show that ≈ 65% of the heat generated in silicon power electronics is dissipated through the drain tab while the remaining ≈ 35% is dissipated through the plastic encapsulant, an unexpected and surprising result given the general emphasis put on the metal drain tab for heat removal from the device footprint. Furthermore, unlike air cooling, immersion cooling shows minimal heat spreading from the power device to the printed circuit board, which we attribute to the proximity of the cooling solution to the source of heat. With superior heat transfer coefficient, immersion cooling reduced the overall thermal resistance by ≈ 75% compared to natural air cooling. After earning credence by validating the ANSYS Fluent model prediction with experiments, we employed the numerical framework to analyze the thermal management of traction inverters in electric vehicles. This analysis shows that unlike forced air cooling, immersion cooling maintains junction temperature well below the safe operating limit (≈150 °C) with minimal thermal crosstalk and heat spreading to neighboring components. Through experiments and modeling, this work not only improves current understanding in the field but shows the immense potential of state-of-the-art direct immersion cooling for the thermal management of silicon power electronics.
No takes yet. Share an insight, caveat, or question.
Jadhav et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: