Here is the abstract you requested from the Thermal_2012 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|Heat Spreading Implications for Hybrid Silicon-Encapsulant Systems-in-Package|
|Keywords: System-in-Package, Heat Spreading, Thermal Conductivity|
|We constructed an integrated ultra-high density (iUHD) multi-chip module assembly with resistive heaters that simulate hot chips with large power requirements. RTD sensors were patterned next to the heaters to monitor the temperature variations across module surface. The iUHD has patterned electrical interconnects on both the front and the back side. Connections from the front to the back side of the module are made with through-substrate vias. The front has a BGA and is mounted to a PCB that can be probed for resistive continuity. A thermocouple was inserted into the base of the PCB, planar with the board BGA to measure the thermal gradients across the iUHD/PCB assembly. The module assembly was mounted to a thermal plate to maintain constant temperature at the base of the PCB during testing. The assembly was tested with and without finned heat sinks on top to measure the through-board thermal resistance versus the thermal resistance directly to ambient atmosphere. Lateral heat spreading in the iUHD configuration did not augment heat transfer and package-to-junction thermal resistance when compared to an equivalent single-die package. While the functionality of the multi-die system cannot be obtained in a stand-alone COTS ASIC, it is important to note the unique thermal challenges inherent in the multi-die integration. The assembly without underfill had a thermal resistance of 7.6 W/°C. The underfill provided insulation that dissipated the heat generated at the surface of the module and reduced the thermal resistance of the assembly to 4.2 W/°C. These values are independent of the power applied to the module heaters. The Cu and Al thermal sink fins increased the power capacity of the assembly by factors of 1.9 and 2.4 respectively. Though Cu has a higher thermal conductivity than Al, the Al fin was more effective dissipating the thermal energy in the assembly because the fins were 1.6 times longer than the fins on the Cu sink providing more dissipative mass. Thermal resistance of the assembly with the heat sink fins is a reciprocal function of power, where R(P) = [2.96 + 1.95/P] W/°C for Al, and R(P) = [3.73 + 1.97/P] W/°C. Temperature gradients across the assembly during testing led to module deformation and eventually interconnect failure. Results from this experimental test bed will be used to determine the accuracy and detail needed to define geometric models for thermal finite element simulations for future product design rules that can resist thermally induced stresses.|
Charles Stark Draper Laboratory