Posted by Pauljnoonan: Thu May 29, 2008 5:31 pm    Post subject: THB v HAST

Does anyone have any input on why THB is still more effective than HAST at accelerating certain failure mechanisms such as migration even though HAST is theoretically a more "severe" environment?

Posted by Paulpang: Sun Jun 01, 2008 4:20 pm    Post subject:

I also want to know it....pls advise.

Posted by FARel Engr: Tue Jun 03, 2008 11:48 am    Post subject:

Hi,

If we’re just talking about electromigration, and you’re seeing e-mig failures at THB and not HAST, and the electrical excitation used by the THB and HAST boards are identical, then I would surmise that the big difference between the duration of THB (1000 hrs.) and HAST (96 hrs.) is the main reason for your observation. The difference in temp at which THB and HAST are done (85C vs. 130C) may not be significant for the device tested in terms of e-mig, especially if it is subjected to high current densities during THB and HAST biasing.

E-mig should take a very long time to emerge under normal THB/HAST biasing, so the 96-H window of HAST is really too short for this mechanism to manifest. In fact, normal 1000-H THB should also not result in e-mig failures, unless there is a fab issue like a metal line masking problem at hand.

This is just my opinion. I hope others can share their views too.

Posted by Paula: Sat Jun 07, 2008 10:54 am    Post subject:

I think, THB and HAST are used to test moisture and corrosion resistance of the package. There is static bias during these test to enhance ionic induce corrosion. If the device you are testing is not design for static bias, then I think whatever e-mig (or EOS) related failures is not valid.

E-mig is a device failure mechanism thus I think it is more appropriate to use Dynamic Lifetest. THB and HAST are package related test.

To analyze THB/HAST failures, we do SAT, mechanical decap and just inspect the die for corrosion.

My thoughts.

Posted by FARel Engr: Sat Jun 07, 2008 2:57 pm    Post subject:

THB and HAST are indeed designed primarily for corrosion testing, but I think what Paul wants to know is why THB seems to bring out e-mig more in his device than HAST. So the phenomenon is what we are trying to establish here, not the industry rel testing practices.

Actually, a static bias is better at inducing e-mig, so dynamic burn-in is not suitable to test for electromigration. Long-term static burn-in would be better. But if you really want to test the e-mig characteristics of your fab process, then you really need to do electromigration testing on samples designed for e-mig testing.

Posted by Paula: Thu Jun 12, 2008 4:22 pm    Post subject:

I read on this link: ---------/emig2.htm

"Electromigration must not be confused with EOS-induced metal reflow, which is a different phenomenon. Electromigration occurs gradually whereas EOS-induced metal reflow is gross and abrupt."

EOS is one of the most common failure observed during THB and HAST, because of several reason - wrong biasing or test board problem or putting a device design to function dynamically in static bias, etc.

I wonder if there is only a confusion between emig and eos.

My thoughts.

Posted by Pauljnoonan: Thu Jun 12, 2008 5:18 pm    Post subject:

Just to clear things up here. When I say migration I don't mean electromigration. I mean the migration of a material in the package i.e. the movement of ions towards a negative potential in the package.

Posted by FARel Engr: Fri Jun 13, 2008 2:59 pm    Post subject:

Hi Paul,

Thanks for clearing that up. Sorry for the confusion - we just got used to a different name for this mechanism - mobile ionic contam. With its high temp, HAST should be more severe than THB for this mechanism. If this is not the case, the first thing I'll do is compare the THB and HAST board circuits. Do they subject the device to exactly the same bias? I remember having the same problem a long time ago only to find out that someone has redesigned the HAST boards to simplify the circuit and cut costs.

Good luck!

Posted by Pauljnoonan: Fri Jun 13, 2008 6:27 pm    Post subject:

Thanks. The HAST/THB boards are identical and receive the same bias conditions for each. the only difference is HAST is at a higher temperature and the chamber is pressurized (2 atmospheres).

The only explanation I can think of is that for the ionic contaminants present in your samples, the longer time of the THB has a greater failure accelerating effect than the higher temp of HAST.  This is probably why the longer (but less stressful) THB is showing more failures than the shorter (but more stressful) HAST.