The main aim of this dissertation is to investigate the open TSV technology regarding EM failure. This investigation includes the two phases of EM degeneration, where the first phase is the phase where stress is built up leading to cracking or void formation. In the case where a void is formed, the first phase is followed by the second phase where the void moves within the metal structure and expands leading to failure. In order to cover the full range of failure development, models for both phases are implemented into the Finite Elements Method (FEM) simulation software COMSOL [30] to provide the ability to analyze the lifetime of interconnect structures as well as to examine the impact of design changes on the lifetime. The validity of the implemented model is demonstrated. The first phase model is verified by comparing the results originating from the TCAD tool FEDOS developed at the institute. The results of the evaluation of the models for the second phase are qualitatively compared to results from experiments. This accomplishment is the basis for the EM failure assessment of interconnect structures taking also voiding into account.
This thesis is logically structured into 6 chapters. Chapter 2 gives a historic outline of the discovery of EM and the development of its modeling originating from the compact model developed by Black [10] relating current densities and lifetime. This is followed by Blech’s findings [14], a critical length times current product for the occurrence of EM induced voiding and hillock development. After the description of these simple models more advanced and complex models are introduced, where the impact of the vacancy dynamics and the stress build-up is gradually incorporated, finally resulting in the state of the art model for EM. The chapter closes with the introduction of the quantum mechanical EM force calculation giving a fundamental understanding of the different resistances of materials and interfaces against EM.
In Chapter 3 the models for a physics based EM assessment are presented in detail. To describe EM failure a variety of different physical phenomena have to be considered, resulting in the need to simulate a multiphysics problem. This includes in bulk regions the electro-thermal problem, as the current is the driving force for EM, the vacancy dynamics, and the continuum mechanical model. For interfaces, different behaviors are known including the fast diffusivity, differing equilibrium concentration, and segregation and are therefore described in this section. To detect the starting point of phase two a critical stress value is needed as a threshold value as explained in the void nucleation section. For the second phase, after a void is formed, the void evolution has to be tracked according to the phase field model explained in Section 3.6.
Subsequently Chapter 4 deals with the details of the essential implementations of the simulation models and how they are carried out for the two simulation tools employed for this work. First the required basis of FEM and its derivation are pointed out, building the principle method of computation for the simulations in FEDOS and COMSOL. Second the specifics concerning the process flow and the model implementation are presented.
Chapter 5 discuses the entire assessment of an interconnect structure with simulation results and their implications. After opening with an analysis of the current crowding effect, reasoning the locations of high current densities at corners of the conducting structure, the vacancies pile-up at the interface of two different metals due to a blocking behavior is surveyed. The typical characteristic of EM induced vacancy accumulation featuring a three phase behavior in time is used as an initial verification and is therefore discussed subsequently. Thereafter the stress build-up in the structure is simulated to evaluate the time until cracking or void formation for different current densities and is followed by the void evolution simulation. The assessment is completed by a fitting of the times resulting from the two phase simulations to Black’s equation.
Finally, the thesis is concluded in Chapter 6 giving an outline for possible further improvements by taking atomistic and micro structural properties into account.
« PreviousUpNext »Contents