QwykSylver
Mechanical
- Jan 2, 2014
- 2
Hello!
I have quite a problem on my hands trying to model a small pre-stressed MEMS device in a transient analysis. It is about 600x600um and is made of 500nm of SiO2 and 95nm of platinum. It is adhered to the substrate in the center only. The SiO2 has about 300 MPa of compressive stress which I emulate using a fictitious coefficient of thermal expansion (CTE) and a given temperature. This puts it in its buckled state which is easy to get results for. The hard part is using a bottom contact that is 10um below the surface (makes a significant difference in profile shape). Also actuating it using an input force while using contacts seems to be very difficult. It seems with this geometry that snap through occurs just as the device lifts off of the contact (which causes convergence difficulties).
I am using mid-side nodes btw since this has a lot of rotation. Also, this is a layered shell model.
Setup has 5 steps
1 apply perturbing pressure to give slight offset
2 apply temperature to pre-stress the model
3 remove perturbing pressure
4 apply force to cause model to snap-through to other stable state
5 set applied force to 0
The pre-stressing needs a lot of sub-steps to converge as does the removal of the perturbing force (else the model starts to become unsymmetrical, maybe due to compiled error). Even the first contact has a lot of penetration after pre-stress. With the setup I have now it chatters for eternity and does not converge.
Model
Mesh
Buckled State With no Contacts
Snap-Through With no Contacts
Contact Setup
Contact After Pre-stress
Setup
Any help or advice would be greatly appreciated! In the long run I would love to apply an input displacement and output the models total energy. I have that script written but I need the bad boy to converge first...
Daniel Porter
ME PhD Candidate
University of Louisville
I have quite a problem on my hands trying to model a small pre-stressed MEMS device in a transient analysis. It is about 600x600um and is made of 500nm of SiO2 and 95nm of platinum. It is adhered to the substrate in the center only. The SiO2 has about 300 MPa of compressive stress which I emulate using a fictitious coefficient of thermal expansion (CTE) and a given temperature. This puts it in its buckled state which is easy to get results for. The hard part is using a bottom contact that is 10um below the surface (makes a significant difference in profile shape). Also actuating it using an input force while using contacts seems to be very difficult. It seems with this geometry that snap through occurs just as the device lifts off of the contact (which causes convergence difficulties).
I am using mid-side nodes btw since this has a lot of rotation. Also, this is a layered shell model.
Setup has 5 steps
1 apply perturbing pressure to give slight offset
2 apply temperature to pre-stress the model
3 remove perturbing pressure
4 apply force to cause model to snap-through to other stable state
5 set applied force to 0
The pre-stressing needs a lot of sub-steps to converge as does the removal of the perturbing force (else the model starts to become unsymmetrical, maybe due to compiled error). Even the first contact has a lot of penetration after pre-stress. With the setup I have now it chatters for eternity and does not converge.
Model
Mesh
Buckled State With no Contacts
Snap-Through With no Contacts
Contact Setup
Contact After Pre-stress
Setup
Any help or advice would be greatly appreciated! In the long run I would love to apply an input displacement and output the models total energy. I have that script written but I need the bad boy to converge first...
Daniel Porter
ME PhD Candidate
University of Louisville