In Situ Characterization of Local Thermal–Hydraulic Characteristics in a Liquid-Cooling Cold Plate for AI Servers Using an Embedded Flexible MEMS Microsensor

A flexible MEMS microsensor integrating temperature, thermal-flow, and capacitive-pressure sensing was developed for embedded monitoring in liquid-cooling systems. The sensing elements were fabricated on a 50-μm polyimide substrate and protected by a patterned PI 9320 layer. Two microsensors were installed near the inlet and outlet of a commercial cold plate and evaluated at nominal coolant flow rates of 2–4 L/min. The measured local temperature difference decreased from approximately 4.8 to 3.2 °C as the nominal flow rate increased, with a temperature difference of approximately 4.4 °C at 3 L/min. Temperature-dependent calibration curves were established for the flow and pressure channels. In a separate 120 h circulation test using a dedicated flow channel, all three sensing channels remained operational. The temperature, flow, and pressure sensitivities decreased by 7.2%, 5.4%, and 4.7%, respectively. These results demonstrate the feasibility of integrating flexible multiparameter microsensors into existing liquid-cooling hardware and provide an experimental basis for embedded cooling-system monitoring.