Thermal testing of a 3-die stacked chip scale package including evaluation of simplified and complex package geometry finite element models

Thermal performance testing was conducted on a 16/spl times/16 mm, 2-metal layer, 591-ball, 0.50 mm pitch, 1.20 mm overall height chip scale package (CSP) containing an offset pyramid configuration of three stacked Delco thermal test die. Die sizes from bottom-to-top were 10.16/spl times/10.16 mm (Delco PST-6), 6.35/spl times/6.35 mm (Delco PST-4), and 3.81/spl times/0.81 mm (Delco PST-2). Testing was carried out using eight different multi-die power configurations in a natural convection environment to highlight the effects of radiant and convective heat transfer. Measured data was obtained on a sample size of five packages to calculate Theta JA, Psi JT, and Psi JB values at each of the eight different multi-die power configurations. Furthermore, Theta JC and Theta JB cold plate measurements were also obtained. For the purposes of thermal testing, each of the five CSP test samples was mounted on a JEDEC standard 101.5/spl times/114.1/spl times/1.60 mm 1S2P thermal test board. Measured results are used to suggest a methodology for the generation of linear superposition matrix equations as a means to present multi-die package thermal test data such that it may account for changes in. thermal cross talk between die at varying die power configurations. The ANSYS finite element analysis modeling software was used to simulate the eight aforementioned thermal test configurations for the purpose of verifying the acquired test data. Both simplified and complex package substrate metal layer trace patterns were evaluated for simulation accuracy. The simplified patterns consisted of conductor traces that do not follow the detailed routing of the actual design, but instead extend straight outward towards the substrate edge. Alternatively, the complex patterns consisted of the detailed trace layers exactly as they are physically routed on the CSP substrate. In both the complex and simplified metal layer trace pattern finite element models, vias that connect the top and bottom trace layers were represented by two-dimensional thermal conduction elements. Simulated results for both the simplified and complex metal layer trace pattern models are compared to the acquired test data. The CSP package structure, thermal test data, and finite element models are presented and discussed.

[1]  B. M. Guenin,et al.  Methodology for thermal evaluation of multichip modules , 1995, Proceedings of 1995 IEEE/CPMT 11th Semiconductor Thermal Measurement and Management Symposium (SEMI-THERM).

[2]  A. Agarwal,et al.  Thermal modeling and analysis of pin grid arrays and multichip modules , 1991, 1991 Proceedings, Seventh IEEE Semiconductor Thermal Measurement and Management Symposium.

[3]  F. Carson,et al.  Development of improved thermal performance embedded heat slug plastic ball grid array package , 2003, 53rd Electronic Components and Technology Conference, 2003. Proceedings..

[4]  B. A. Zahn Steady state thermal characterization and junction temperature estimation of multichip module packages using the response surface method , 2000 .

[5]  B. A. Zahn Steady state thermal characterization and junction temperature estimation of multi-chip module packages using the response surface method , 1998, ITherm'98. Sixth Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (Cat. No.98CH36208).

[6]  Avram Bar-Cohen Thermal Management of Air- and Liquid-Cooled Multichip Modules , 1987 .

[7]  B. A. Zahn,et al.  Evaluation of isothermal and isoflux natural convection coefficient correlations for utilization in electronic package level thermal analysis , 1997, Thirteenth Annual IEEE. Semiconductor Thermal Measurement and Management Symposium.

[8]  G.B. Kromann Thermal management for ceramic multichip modules: experimental program , 1992, Proceedings 1992 IEEE Multi-Chip Module Conference MCMC-92.