Multi-Objective Analysis and Optimization of Integrated Cooling in Micro-Electronics With Hot Spots

Sohail R. Reddy

CHALLENGE - With computing power from electronic chips on a constant rise, innovative methods are needed for efficient thermal management. Arrays of micro pin-fins act not only as heat sinks, but also allow for the electrical interconnection between stacked layers of integrated circuits. This work performs a multi-objective optimization of three shapes of pin-fins to maximize the efficiency of this cooling system.

SOLUTION - For each of the three cross section shapes of the pin-fins, two optimization studies were performed to find an optimum configuration for each of the three shapes under two different operating conditions. The objectives of both studies were to minimize maximum temperature and minimize inlet pressure. An inverse design approach that allows for the design of cooling configurations without prior knowledge of thermal mapping was proposed and validated.  This inverse problem was solved using multi-objective optimization where two more objectives were added to the previous optimization study. The three simultaneous objectives were: maximize background heat flux, maximize hot spot heat flux, minimize inlet pressure (pumping power). This work used the NSGA-II (Non-Dominated Sorting Genetic Algorithm) to navigate through design space to search for optimum designs using optimization platform modeFRONTIER








BENEFITS -  An observation was made that even the smallest thin film thickness improves both overall and local temperature uniformity. It was noticed that for the circular cross section pin-fins, a thin film thickness of more than 5 µm is needed to create any measurable performance. The thin films significantly lower the maximum temperature for all three pin-fin cross section shapes.