Student project

Deformation in thin SiN membranes due to residual stresses

For the patterning on 'unconventional' surfaces and/or non-IC applications new micro- and nanopat- terning technologies are developed. One of these techniques is the localized material deposi- tion through ultra-miniature shadow masks (nanostencils). Nanostencil lithography is a pattern method, which uses a membrane to shield the substrate from material °ux during deposition on a substrate. In this project a partial solution will be given for one of the known problems within this stencil lithography method. This report focusses on the problem of the existence of residual stresses in the SiN membranes and ¯lm material after an evaporation process. These residual stresses are causing deformations in the membrane through which the gap size be- tween substrate and membrane will increase. These di®erent gap sizes are in°uencing the sharpness and sizes of the deposited structures on the substrate. To decrease the deformations in the mem- brane stabilization techniques are developed. In this project the deformations in a membrane, caused by residual stresses, are described within a ¯nite element model (FEM). Second the stabi- lization techniques are implemented in the model to make an optimized design for the nanostencils. First the experiments, done on a SiN membrane, are analyzed. From these experiments the values of the residual stresses present in the materials are calculated. The results of an experiment on a membrane, wherein cantilever structures are present, are used as veri¯cation when developing 2D and 3D models. Cantilever shapes are used, because theoretical veri¯cation and relative simple modeling is possible with these structures. To obtain a model that is describing the experiments accurate, di®erent methods for the imple- mentation of residual stresses into the models are developed and compared with the experimental results. Four methods are evaluated (an ANSYS option, surfacestress and two di®erent temper- ature di®erence methods). The method wherein the stresses are described by implementing a temperature di®erence between two materials has got the smallest error comparing the results with the experiments (§20%). The 2D cantilever models are expanded to a 3D model. In this 3D model an element type is used that is still giving an accurate solution when using large sizes and aspect ratio's in the models. The behavior of the element type is veri¯ed. An optimalization of the geometry of the model is done and the optimized model is compared with the experimental values. The di®erences of the 3D models are within 10% of the experimental results. A 2D silicon stabilization cantilever model is developed to see the reduction of the de°ections com- pared to the unstabilized cantilevers. In this stabilized model a simple calculation of the stress working on the silicon is made and this stress is implemented by calculating the temperature dif- ferences. The approach is veri¯ed with the use of di®erent simulations. The veri¯ed 2D models are expanded to 3D models. An optimalization is done for 2D and 3D models. This optimal design data can be used as guideline for the optimalization of whole membranes. The developed modeling techniques are used in the models of whole membranes. Tetrahedral elements are used to mesh the complex geometry of the membranes. Because of a to large mem- brane size (compared to the element size) the element solutions aren't accurate enough. A silicon stabilized model is made to verify the reduction values between stabilized and unstabilized sim- ulations with the experiments. The absolute values obtained in the simulations aren't the same as the experiments, but the reduction values when stabilizing are giving a good description. The developed guideline isn't veri¯ed with the results from the simulations, because a lack of time. However ¯rst simulations are giving a good impression for the veri¯cation of this guideline.


    • LMIS1-STUDENT-2005-004

    Record created on 2005-11-02, modified on 2016-08-08


  • There is no available fulltext. Please contact the lab or the authors.

Related material