viscoelastic material/cyclic loading/ damage description/ansys/element

[sartor_bjk]

Dear all,
I have a similar problem to:
Viscoelastic fatigue?
Started by cmc123‎, 2011-08-19 02:05
thread.

I have an epoxy based, silver filled(highly) die attach material which is used in integrated circuits. They fail due to long-term temperature cycling.

I would like to describe the damage and implement it to ANSYS.

I have thought of such a way, i would be really happy if you could comment on whether it is a solid start or not:

1)Investigating the mechanical behavior of material (stress - strain curve) at room temperature.

2)Applying cyclic loads at relatively high loads and after several cycles, stop the test, and load the specimen in TENSILE mode and OBSERVE ELASTIC MODULUS change.

3)Once I see the elastic modulus change over a complete cycle range, I try to describe it mathematically. At this moment, I am having difficulty to imagine how I can introduce this to ANSYS code basically ( i am not a software expert, or coder).

p.s.: I am able to use usermat subroutine in ansys. I have all elastic modulus, bulk modulus, and shear modulus data (cure % dependent) for my materials.
p.s.2: Is there any element in ansys which will support such an activity, elastic modulus change (like softening, and killing the element at a certain level, it will be like a crack at that point, temperature and frequency dependent material properties)
I would be really happy to discuss over this problem with you.

[cmc123]

Hi,

I think the theory you have got there is fine. However...

I ran a project doing something similar (but different) to do with materials at temperature. One problem with this was with the results. Thankfully we had enough results to implement what we wanted to do. But for higher accuracy I think either a much bigger range of results, and repeat readings would be required - so a LOT more time input, or alternatively better quality results (ie a much better tensile machine).

I dont use ansys myself but I do use FEA software. Can you not input the Young's modulus as a function of something else? eg. E = mX + c, where X is the variable you basing E on. And variable X can be defined as a parameter at the start.

Hope this helps.

[sartor_bjk]

Thank you very much your answer.
I can input E change as a function of a variable that I define (in other words, damage).
Normally, my E is Temperature and time dependent. I have the master curves for E and K, and thus get G and v.

Besides that T and t, I would like to introduce a damage parameter, which will be D. so, E(D)=mD+c , this is the form that I will be trying to get. Of course, this function is linear. I do expect a non-linear change of E with the damage as cyclic loading increases (needs to be clarified though). I have also found some literature which describe the non-linear change of E by simple formulas which can be implemented into FE code relatively easily.

Which FE software are you using?

[cmc123]

I use COMSOL FE-software. Out of interest, how do you define damage, D? I'm guessing it's a function of cyclic strain and number of cycles? Do you have a reference?