siggi84
Electrical
- Mar 19, 2009
- 11
I am having a problem with thin layers in ansys. Simply speaking I have a Pt nanowire (8 um length, 1 um width and 0.05 um thick) on top of a thin layer (100 um length, 100 um width and 0.1 um thick) of SiO2. The SiO2 layer is on top of a very thick layer of Si (100 um length, 100 um width and 100 um thick). I am ignoring the air interface.
Pt is a good heat- and electric conductor, SiO2 is excellent heat- and electric insulator and Si is good heat conductor but poor electric conductor.
I have fixed the temperature at 300 K on the backside of the Si layer and I apply 2.6 V over the nanowire. The voltage creates current and the nanowire heats up (because of resistivity). I want to find the temperature distribution around the wire.
I have already managed to mesh the whole thing using mapped meshing, but of course I get some warnings regarding aspect ratios of the elements because of the thin layers. I am using SOLID69 (3-D Coupled Thermal-Electric Solid) elements for the whole geometry.
I get reasonable results when I use a thick SiO2 layer (~1 um), but when I use a thin layer (~0.1 um) I get obvious numerical errors like 25 K close the the heater, which is colder than any boundary condition (The number of elements through the layer is the same). I think I can not use shell elements (like SHELL157) because the temperature distribution changes a lot through the SiO2 layer because of how good insulator it is.
Do you have any ideas how to solve this problem? What is the reason for this numerical error? At the moment I am totally stuck
I have attached the DB log file and a datafile. If you want to run it you can change the SiO2 layer thickness by changing the value of the OT variable.
Pt is a good heat- and electric conductor, SiO2 is excellent heat- and electric insulator and Si is good heat conductor but poor electric conductor.
I have fixed the temperature at 300 K on the backside of the Si layer and I apply 2.6 V over the nanowire. The voltage creates current and the nanowire heats up (because of resistivity). I want to find the temperature distribution around the wire.
I have already managed to mesh the whole thing using mapped meshing, but of course I get some warnings regarding aspect ratios of the elements because of the thin layers. I am using SOLID69 (3-D Coupled Thermal-Electric Solid) elements for the whole geometry.
I get reasonable results when I use a thick SiO2 layer (~1 um), but when I use a thin layer (~0.1 um) I get obvious numerical errors like 25 K close the the heater, which is colder than any boundary condition (The number of elements through the layer is the same). I think I can not use shell elements (like SHELL157) because the temperature distribution changes a lot through the SiO2 layer because of how good insulator it is.
Do you have any ideas how to solve this problem? What is the reason for this numerical error? At the moment I am totally stuck
I have attached the DB log file and a datafile. If you want to run it you can change the SiO2 layer thickness by changing the value of the OT variable.