Investigations into Droplet-Based Microelectrofluidic Technology for Hot Spot Cooling and Thermal Management in Integrated Circuits

Duke University, Durham NC

Investigators

Abstract

Decreasing feature sizes and increasing package densities are making thermal issues extremely important in integrated circuit (IC) design. Uneven thermal maps and hot spots in ICs cause physical stress and performance degradation, and they also reduce reliability. This research project is focused on embedded cooling methods and an IC design flow for thermal management. The goal of the project is to develop new techniques for chip cooling based on droplet microfluidics, and to relate this technology to the integrated circuit (IC) design flow. It leverages a recent DARPA-funded research effort at Duke University that led to a novel electrowetting-based droplet transport technology. It is based on high-speed electrowetting transport of discrete microliter- and nanoliter-sized droplets under voltage control. Cooling liquids can be transported, in user-defined patterns, under clocked-voltage control over a 2D-array of electrodes without the need for external pumps and valves.In addition, liquid flow is inherently increased with increasing temperature. Thus, local hot areas on a chip will have increased cooling applied locally without the need for external sensors. The project is also focused on the design of the architecture for fluidic pathways, which will transport droplets for convective cooling. While the droplet paths are statically allocated during design time, the rate of droplet transport is dynamically determined during run-time. Simulation and profiling tools are being developed to determine the rate of droplet transport on different channels and at different times during field operation. Additional topics being investigated include protocols for droplet routing and optimization techniques for droplet layout. This project will result in a self-contained, self-regulated, and low power (< a microwatt) chip cooling technique. The insights gained from this work are expected to pave the way for a new generation of embedded systems for computing and communications, as well as high-end, high-speed computation platforms. Power management and thermal problems are being widely seen as major roadblocks in the evolution of ICs; this project is expected to help overcome these roadblocks. This project also serves as an important bridge between two diverse research communities. It provides researchers from the microfluidics area with a major application driver for continued research into droplet-flow methods at the nanoliter and picoliter scales. On the other hand, it enables IC designers to leverage advances in droplet-flow technology for chip cooling.