Thermal management and thermomechanical reliability of LTCC package for 24 GHz band: Master's thesis

This master’s thesis studies thermal management and thermomechanical reliability of radar sensor demonstrator package for the 24 GHz band. The package consists of a RFIC flip chip bonded on a LTCC module attached to a printed circuit board. The work includes fabricating test structures and performing measurements and tests on them. Flotherm simulation software is used to assist in improving thermal management and thermomechanical reliability. The usability of the software in this kind of study is also evaluated. Temperature measurements on test structures revealed that vias right under the flip chip bumps cool the chip most efficiently. The version in which signals and grounds were first distributed to the edge areas of the module and then connected through the LTCC layers with vias had higher chip temperature in both simulations and measurements. Underfill applied under the RFIC chip was also found to enhance the thermal management of the package. Thermomechanical reliability tests revealed that to ensure reliability of demonstrator package the LTCC module should be as small as possible. The fatigue life predictions made with Flostress software were substantially longer than real test results because fatigue damage was not due to crack in a solder joint as the simulator assumed. Instead the damage was located on printed circuit board. The discovery demonstrated that joints are not the most susceptible to thermally induced stresses and strains after all, an assumption which was taken for granted in whole study. Flotherm was concluded to be a useful tool for thermal management since it gave reasonable results and managed to imitate the thermal behaviour of the structure. In contrast thermomechanic simulations made with Flostress module gave over optimistic results. However when attributes were changed to investigate the effect of package dimensions and simulation circumstances, simulations followed real situation quite well. Thus it was concluded that even though Flostress was unable to predict fatigue life accurately it can be used for comparing different designs from thermomechanical reliability point of view.