Posted by Paul2008: Wed Apr 09, 2008 4:20 pm    Post subject: HTRB burn-in failure

Hello,

[HTRB burn-in test failed @168hrs. Would any expert here please advise what's the major material properties that we need to consider for HTRB test?

Is the failure related to the properties of molding compound (eg. Tg or volume resistivity), the Si die itself or other factor?

Thanks.

Posted by FARel Engr: Thu Apr 10, 2008 4:31 pm    Post subject:

Hi Paul2008,

The HTRB test is designed to catch weaknesses in the die's transistors when these are put in the blocking/cut-off/non-conducting state. Such weaknesses may result in voltage shifts or current leakages in the transistors over time. Thus, HTRB is primarily a wafer fab rel test.

However, HTRB failures can be due to a variety of reasons, not just wafer fab issues. You need to do a thorough FA on the failing parts to determine what caused the failures. The cause may be very simple, such as lead oxidation, or very complex, such as one that involves marginal transistor degradations that are hard to detect. Unfortunately, a thorough FA is the only way to go forward.

If you need a more specific advice on how to proceed, please provide the failure rate, the failure mode, and FA steps already performed (and results of course).

Thanks and good luck!

Posted by Paul2008: Fri Apr 11, 2008 5:39 pm    Post subject:

Hi FARel Engr,

Thanks a lot for your advices and information.

Here are some more information:
(1) Failure rate: 3/25
(2) Failure mode: walking Bvdss, Bv starts from about -20V and moves rapidly to about -250V.
(3) After 64hrs baking @ 150C, the failing parts didn't recover.
(4) Other similar devices have failed the test but pass by switching to another mold compound.
(5) Same device applied into non-mold compound encapsulated package eg.TO39 pass HTRB.

As we know, Bvdss problem is the common failure in HTRB. It is caused by the presence of foreign materials or ionic contaminants. But we are not sure whether those contaminants are came from wafer fab process or mold compound.

At this moment, we haven't performed any destructive FA yet. Is there any method we can use to check for those ionic contaminants? Or we can just simply decap the parts and have visual inspection?

P.S. Item (4) & (5) above are the observation that we suspect the failure may also be related to mobile ionic contaminants from mold compound at high temperature.

Thanks again.

Posted by FARel Engr: Wed Apr 16, 2008 7:44 am    Post subject:

Hi Paul2008,

Ionic contam failures usually improve, if not fully recover, after bake. This is because the ions are dispersed by the thermal energy. When you say that the failures did not recover after bake, do you mean that the reading is still -250V, or there's an improvement but not enough to make the units pass? Can you datalog them again and see what the readings now are? Are there other parameters in the test program failed (the entire test must be run on the units)?

It is interesting that you don't see the failure in a TO package. Another angle to look at is the possibility that the molding compound exerts too much pressure/stress on the die surface. Is this a new compound? Have you used this compound on this device before, or is this the first time (qual)?

Have you decapsulated at least one sample already (after all the non-destructive tests have been done)? What are the results? Did it recover completely?

FARel Engr