Discussion Forum

Hi all,

I'm hoping someone with experience in the heat transfer module can help me here. I'm modeling arrays of nano-structures on a substrate excited with normally-incident, linearly polarised light in the microwave heating module.

My emw module modeling is as I expect (I've used this a lot) but am having trouble with the boundary conditions for the heat module. On the vertical sides I'm using periodic boundary conditions for both emw/ht modules and they're working well. At the top, I've declared an extra wave equation (electric) and am exciting with a scattering BC from above. At the top of the structures I'm using convection heating nodes to let some heat escape.

My problem is the lower boundary. For all intents and purposes the substrate is an infinite heat sink- my structures are 50nm tall and the substrate is a standard 0.5mm sapphire slide.

COMSOL technical help suggested I use a temperature lower boundary condition with a substrate layer with extremely high thermal conductivity to model the substrate but I'm finding I can vary the resulting steady-state temperature by playing with these properties. In fact, I have so much freedom to mess with the end result that I can get any temperature I want.

I've tried infinite elements nodes and they weren't much help.

Has anyone else come up against this and if so, what have they done? I'm beginning to think I have to model the entire thickness of the wafer and use lower boundary convection to get around this!

PS I'm using reduced thermal conductivities for my metal layer and a thin thermally resistive layer to simulate the thermal conductivity discontinuity at the metal/sapphire interface.