Investigation of Nanopillar Wicking Capabilities for Heat Pipes Applications

Since Moore’s prediction in 1965, transistor count density on computer chips has grown exponentially and roadmaps for future industry growth still project exponential development for the next decade. With higher transistor densities, greater heat flux dissipation is required in order for performance to keep par with chip development. However, it is theorized that current cooling systems would not be able to cope with heat fluxes of future computer chips. Microchip heat management systems can be either active or passive. Active systems require an external driving component that increases the system’s complexity and ultimately power consumption. Heat pipes are passive fluidic systems, which are more robust and easier to implement than their active counterparts. Recirculation of the coolant in a heat pipe is done passively by means of a wicking structure that induces capillary flow from the condenser to the evaporator. However, there are many limiting factors associated with heat pipes based on the wick dimensions, fluid selection and orientation. At CPU chip operating temperatures the most significant limitation is the capillary limit. This limitation must be addressed in order to meet future computer chip heat dissipation requirements. In order to find an optimal geometry that would maximize the capillary flow, a theoretical model was developed using a rectangular pillar array. Surface tension forces induce a capillary flow that is opposed by viscous stresses from the pillars. Due to the regular and well-defined geometry of the pillar array, an ab initio approach can be used to model this flow, rather than resorting to Darcy’s flow and empirical permeability correlations. Predicted values of maximum flow rate were obtained from this theoretical model. This model and its results are directly applicable to carbon nanotube (CNT) and nanowire (NW) based wick structures. To validate the merit of nanostructure wicks for use in heat pipes, experimental data was collected to show the capillary limits of various nanowicks. The capillary limit of a wick was associated with the heat flux at which the wick cannot sustain the fluid flow necessary for heat removal and burnout occurs. When a baseline wick was experimentally compared to a nanowick, it was found that due to the difference in thickness of the wicks, the baseline wick provided higher flow rates. However, when the data were normalized to produce velocity values, the nanowick was found to have a higher velocity than the baseline wicks.