Keywords: multi-physics model, structural reliability analysis, FORM, SORM, importance sampling, microelectronic package.

The design for reliability of structures such as microelectronic devices is a very challenging issue. Microelectronics components experience a complex structural behaviour. This behaviour introduces coupling between several physical phenomenons (e. g. electric, thermal and mechanic). In fact, the electrical signal due to power transistor switching in the microelectronics packages causes a Joule effect power loss in the structure, which is consequently warmed and experiences thermo-mechanical behaviour. The cyclic nature of the electrical signal may cause thermal fatigue damage in the weak parts of the package such as solder joints. In addition to the fact of bringing closer various physics, the structural behaviour of microelectronics packages is affected by numerous uncertainties due to:

• natural variations of material properties and manufacturing process,
• unpredictable operating conditions (for example mission profile, environmental factors),
• and the lack of adequate model parameters for many of the materials being applied.

Therefore, more coherent prediction of the performance and the reliability of these devices would be made using realistic numerical simulations which incorporate both the coupling between the various existing physics and the uncertainties in the design parameters. In this way, it would be possible to assess the reliability of the given structure and to quantify the influence of the various input parameters on the structural reliability.

The aim of the present paper is to integrate a global reliability approach in a multi-physics finite element computation scheme. A finite element code devoted to multi-physics computing (e. g. MulPhyDo) has been built in the laboratory [1]. This code can perform weak couplings between electrical, thermal and mechanical physics, including floating subdomains and parallel architecture to gain computational time. Beside, it makes it possible to have different time-steps in the multi-physics integration. The purpose of my current research is to make use of the existing structural reliability methods (e.g. FORM, SORM, Importance sampling) to carry out reliability-based analysis of structures, which operate in an electro-thermo-mechanical context. The computation of failure probability in this study are based on the principle of coupling a stochastic model (e. g. the probabilistic characterization of the input random variables) and a mechanical model (e.g. the procedure to compute the limit state function) [2].

An academic example of the reliability analysis of an element of microelectronic package whose solder joints are subjected to thermo-mechanical fatigue damage is treated in this study. The results show a relative large discrepancy between FORM and SORM failure probability, which could means that the limit state surface has strong curvatures. The obtained sensitivity factors, by their respective signs, are in agreement with the expected effect of the associated variable. Both substrate and chip CTE appear negligible for the reliability, thus one can check the case where only their discrepancy is taken as a random variable. However, the given conclusions are not definitive until the FORM and SORM results are compared to simulation result which is not yet available.

1

L. Menanteau, O. Pantalé and S. Caperaa, "A coupled electro-thermo-mechanical fem code for large scale problems including multi-domain and multiple time-steps aspects", In Coupled Problems, ISBN:84-959999-71-4, Santorini Island, 2005.

2

M. Lemaire, "Fiabilité des structures - couplage mécano-fiabiliste statique", Hermes Science Publication, 2005.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £140 +P&P)