Heat transfer via air to the substrate 1

[epd20]

Hi,
I have some problems in modeling the heat transfer of a standing beam (MEMS), which being thermal actuated involves temperature analysis.
I've already made heat transfer by convection, using the specified command, but I don't see how to make the heat transfer by conduction to the substrate, via air.
Have someone experienced the same problem? And how could I solve it?

Thanks in advance

 

[harry123456]

Hi
Whether or not heat conduction will occur to the substarte will depend uponw how your beam is fabricated also. If it is released 2microns or so above the substrate then you conduction will dominate. You could do two analyses then: Apply an equivalent heat transfer coefficient to the bottom surface of the beam taking into account thermal resistance between beam and substrate. This coefficent will be
Kair/tair = 0.026 W/mK/2 microns = 13,500 W/m2K approx. The beam might also conduct from side walls for which you will have to account for a shape factor.
Second method more involved will be to create a mesh of air around the beam using thermal elements.
Hopefully I havent gone too off track from your problem.
harry

 

[epd20]

Hi,
Thanks for your reply and excuse me for the delay.

I just don't have yet specification about the exact gap between my structure nad the substrate, so I will see later on. One more question about your post: you meant 13.5 W/(m^2 K) if the gap was 2 um, didn't you? Sorry for the silly question but I'm not sure of where did you take that value..
Second method: my model would become too complex and slow I think.

Thank you in advance,

Emile

 

[harry123456]

Hi,
No I meant 13,5000 which you can prescribe at the bottom surface of your beam. Notice that this value is very large. This value will depend on how small the gap is between your substrate and beam. This value is typical for 2 microns. As this air gap increases your heat transfer will switch to convection dominant.
This value you can compute from any standard heat transfer textbook. Just look up thermal resistance.
You can also look up papers on thermal actuators. The modes of heat transfer etc. and how to get the values are explained more in detail in those papers. Though you are modeling only a beam the physics would be the same. The thermal actuators operate on a applied voltage and deflect due to the resulting temp. Depening on the fabrication, substrate gap etc. the heat loss is computed for teh actuator and the final thermal profile which gives the deflection. This procedure would be same for your cantilever beam. Only tricky thing is getting all these heat losses correct because in the end they determine your final deflection.
Hope this helps.
harry

 

[epd20]

Hi,
that's the problem, I'm trying to get the same deflection measured in Lab with my simulations...
With your help I should succeed. All I have to do is to adjust the parameters. Thanks a lot!
Emile

 

[harry123456]

Yes thats true. But also be sure you can explain why you used what you used when you validate your simulations to other people. Then you will have really good results.
Harry

 

[epd20]

Hi again.
While trying to modify my model like you suggested me, I had a doubt. The fact is that the equivalent heat transfer coefficient

...equivalent heat transfer coefficient to the bottom surface of the beam taking into account thermal resistance between beam and substrate. This coefficent will be
Kair/tair = 0.026 W/mK/2 microns...

has the dimensions of a convection coefficient (W/m[COLOR=yellow red]²[/color] K), hasn't it? So, do I have to put it like a convection, with the CONV command, or like a layer with different thermal conductivity?
By the way, the gap between structure and substrate is 20 microns and yes, conduction through the substrate is absolutely the predominant heat transfer phenomenon (datas from Lab experiences...)

...I'm also trying the second solution, meshing a sort of air layer with SOLID70 thermal element, hope this will work


Cheers

Emile

http://www.master-nanotech.com/

http://www.polito.it/

 

[harry123456]

Hi Emile,
You said that the gap between the bottom surface of the structure and the substrate is 20 microns. Then, that changes the BC's drastically. In this case, there will be three modes of heat transfer:
1. Conduction- This will not occur through the air now since the gap is not as low as 2 microns. This will occur only to the anchors attached to the substrate. The beam will be fixed at one end to the substrate by an anchor pad. On this anchor pad you apply a room temperature BC since the substrate is a large thermal mas at ambient. This acts as a heat sink. But now there will be no conduction to the air. If your gap was 2 microns then you are right you specify a heat transfer coefficient kair/tair=13,500 to take care of the heat loss. But now the gap is 20 microns no need to do that.
2. Convection-the beam will convect from all faces. You can apply standard heat transfer coeffs for hot plates facing upward downard.
3.radiation- apply this using radiosty solver method in ansys.
I have recently written a paper on this. I could send you that which describes all this in detail. But I do not know how to do that since rule of eng--tips is not to give out any e-mail etc.
harry

 

[epd20]

Hi Harry
thanks a lot for your very fast response!

Well, I was thinking exactly like you do, but the Lab team told me that they (in some way, don't ask me how) know (!!) that conduction via air is the principal thermal transfer phenomenon, convection being only a small 10% of the overall transfer and radiation being just very very small that one can avoid to consider it. That sounds strage to me too, because 20 microns sounds really large.

You would be very kind to drop me your publication if possible, thank you very very much (

http://emile.clawz.com/contact.html

hope this is not against any forum rule)

Another question: I'm trying the second solution you proposed, but I'm not sure it's working. Do I have to use some sort of contact elements (see thread569-127482)

Cheers

Emile

http://www.master-nanotech.com/

http://www.polito.it/

 

[harry123456]

Hi Emile,
I will send you the paper. However, please note that I have submitted the paper and it is not yet published. The paper should make most of your heat transfer questions clear. If not drop me a line.
harry

 

[epd20]

Hi all,
as in my previous post:

--------------------
Nickel (SOLID69)
--------------------
Air (SOLID70) 20 microns
--------------------
Silicon (SOLID70) Substrate @ fixed Temp
--------------------

For a thermal analysis in which the Nickel is heated up to a certain temp by an applied voltage, will it be enough if I merge all the elements together, I mean, will the heat transfer take place between Nickel and Air or I will need to study contact between the surfaces with the appropriate elements?

Thanks again for the help

Emile

http://www.master-nanotech.com/

http://www.polito.it/

 

[harry123456]

Glue the volumes together before you mesh them. that should take care of your problem

 

[epd20]

Ok thanks. But then how can I define different material properites? I used to do that by the MAT command just beforethe meshing of each solid separately..

Hope I'm not abusing too much of your time.

Thanks

Emile

http://www.master-nanotech.com/

http://www.polito.it/

 

[epd20]

Thanks Harry,
I've done with the merging only; but first I had to mesh carefully the two volumes, to make sure the nodes of the upper one were coincidents with the ones of the lower one.
Now all the model works. Thanks again.
Emile

http://www.master-nanotech.com/

http://www.polito.it/

 

[thanhvan]

To Harry123456 and apd20,

I am very interested in this topic, would you please send me a copy of your paper to dauthanhvan@gmail.com. Or would you please tell me the name of your paper, so that I can look up at the library.
Thanks so much.
ThanhVan