Pushing the limits for miniaturized cooling using single/two-phase embedded cooling

The recent trends in electronic systems have seen a surge in power density with power levels expected to reach 500-1000 W/cm². For maintaining operating temperatures within the permissible limit of electronic systems, various the state-of-the-art solutions have been proposed varying from single phase to two phase cooling in various configurations – microchannel, heat-pipe, jet-impingement etc.  [1]. Most of the proposed solutions have taken advantage of miniaturization where high surface area to volume ratio results in better heat transfer. However, this comes with challenges such as high pressure drop, flow instability in two-phase flow, etc. The cooling miniaturization so far has been primarily investigated at the order of several hundred µm or above. Recent developments in electronic device integration show a further increase in integration complexity and interconnect density, such as CMOS 2.0  [2] and backside power delivery networks, resulting in a further reduction of the Si substrate thickness and distance between interconnection structures. The shrinking length scales, and the increased risk of excessive self-heating thereby necessitate the need to develop liquid cooling strategies at even smaller dimensions.

The proposed PhD thus seeks to explore potential means – active or passive, single phase/two-phase or a combination of both, for very small-scale systems. The high-power levels at such small scales imply high heat fluxes (200 – 1000 W/cm²) following which it is inevitable to expect phase change. The proposed work thus seeks to understand the dynamics of phase change and the relevant phenomena – bubble dynamics, bubble growth, dry out, etc, as governed by various geometrical and heating parameters. At such small scale, interface phenomena such as capillary phenomenon, Marangoni convection would play an important role. How much positive impact they can have needs to be investigated. Further, can some surface modifications such as surface wettability be used to tune the desired performance. The proposed study further seeks to investigate this behavior not restricted to pure fluids but also binary fluid mixtures where the presence of mass transfer can further augment the local advection and thus heat transfer. The proposed system configuration includes simple microchannels and would be extended to stacked configuration to analyze the impact of conjugate heat transfer as well as contact angle dynamics. While addressing the above-mentioned fundamental questions, the technological implications would lie to push the limits of power that can be applied, through appropriate system design in terms of fluidic network. Initial investigations would pertain to current scale micro-scale systems with objective to see possibility of having hybrid cooling – both single and two-phase to seek advantages of both and eventually looking into the feasibility of miniaturizing this system.

The objectives are thus as follows: 

1. Develop fundamental understanding of phase change phenomenon at microscale (5- 50 um) including the impact of capillary, wettability, surface roughness on hydrodynamics and thermodynamics behavior and subsequent impact on heat transfer.
2. Modeling and simulating the physical phenomenon to understand the coupled thermo-hydro behavior as described to develop efficient fluidic networks facilitating evaporation and condensation.
3. Use machine learning methods (data driven, physics informed neural networks) for wider understanding of operating conditions.
4. Design and fabricate demonstrators to validate the modeling methodologies and to characterize the thermal and hydraulic performance of the developed cooling structures.