Wire bond quality measurement

EDAboard.com Forum Index -> Analog IC Design & Layout -> Wire bond quality measurement

Author

Message

rikie_rizza

Joined: 05 Aug 2006
Posts: 334
Helped: 20
Location: Bikini Bottom, between a rock and a pineapple

29 Sep 2008 9:39 gold wire bond current limit

Friends,

Need help.

Which is the best method to measure wire bond (bonding quality), force with I or force with V?

Any other sensitive method in measuring wire bond, including delamination?

Thanks
Rikie_Rizza

Google
AdSense
Google Adsense

29 Sep 2008 9:39 Ads

erikl

Joined: 09 Sep 2008
Posts: 735
Helped: 138
Location: Germany

29 Sep 2008 18:12 bond wire current limit

Hi r_r,

standard method is to apply a fixed pulling force to the bond wire, and in parallel to it. The applied pulling force depends on the bonding method, the connected materials, and the wire thickness. A certain force will be guaranteed by the bonding/packaging fab.

Of course this is not a sensitive test. However, if a wire pulls out or breaks, the bonding wasn't correct anyway. Normally, this test is applied following a statistical method, never to 100%, of course.

HTH, erikl

rikie_rizza

Joined: 05 Aug 2006
Posts: 334
Helped: 20
Location: Bikini Bottom, between a rock and a pineapple

30 Sep 2008 6:21 wire bond current limit

Hi Erikl,

Thanks, but here at BE device testing, we must be sure that the wire bond quality is good and reliable enough for the device to perform at the most extreme condition within the specified limit. The IC, is already pack so any type of force such mold pressure, burn-in, and even ESD can damage the bonding invisibly.

That is why we need a sensitive non destructive test to, at least, know the device bond quality.

erikl

Joined: 09 Sep 2008
Posts: 735
Helped: 138
Location: Germany

30 Sep 2008 10:55 voltage drop in wire bond

If the device has already been subject to "the most extreme condition within the specified limit" like burn-in, temperature and humidity cycling and ESD test, the only "sensitive" test to my knowledge is standard testing of the electrical parameters, using the max. allowable current limit. Another non-destructive test is X-ray inspection of the bonding wires. Low X-ray dose if NV memory is involved!

rikie_rizza

Joined: 05 Aug 2006
Posts: 334
Helped: 20
Location: Bikini Bottom, between a rock and a pineapple

02 Oct 2008 8:45 Wire bond quality measurement

Mmmm, X-ray is not possible because we dont have the machine to inspect all hundreds thousand of the automatically. Beside, the wire is AL, so it wont be visible using X-ray.

erikl

Joined: 09 Sep 2008
Posts: 735
Helped: 138
Location: Germany

02 Oct 2008 13:23 Re: Wire bond quality measurement

rikie_rizza wrote:

Mmmm, X-ray is not possible because we dont have the machine to inspect all hundreds thousand of the automatically. Beside, the wire is AL, so it wont be visible using X-ray.

Sure, that's a pity Sad

(
With excellent X-ray equipment, however, one can differentiate even between Si & Al .

rfsystem

Joined: 25 Feb 2002
Posts: 858
Helped: 96

04 Oct 2008 22:06 Re: Wire bond quality measurement

Use 2-5x rated current and measure via 4-terminal the voltage drop decrease or increase. Both indicate a life-time issue with gold-palladium interface.

erikl

Joined: 09 Sep 2008
Posts: 735
Helped: 138
Location: Germany

05 Oct 2008 12:43 Re: Wire bond quality measurement

rfsystem wrote:

Use 2-5x rated current and measure via 4-terminal the voltage drop decrease or increase. Both indicate a life-time issue with gold-palladium interface.

Ingenious technique, but won't be quite easy at a packaged device! Wink

rikie_rizza

Joined: 05 Aug 2006
Posts: 334
Helped: 20
Location: Bikini Bottom, between a rock and a pineapple

07 Oct 2008 3:48 Wire bond quality measurement

Why it cannot work with molded device?

I manage to increase the sensitivity using micro Amps force, previously I use mV force and the device passes. Does it measure better in uA scale or higher current scale?

erikl

Joined: 09 Sep 2008
Posts: 735
Helped: 138
Location: Germany

07 Oct 2008 17:03 Re: Wire bond quality measurement

rikie_rizza wrote:

Why it cannot work with molded device?

Because you can't make a real 4-terminal-measurement over the bond wire itself. You always have to inject a current against GND (or VDD). If it's a gate input of a MOSFET, the current always will flow via a pn-junction of one of the input ESD-diodes in forward direction - so you'll essentially measure the forward voltage of the junction - not a meaningful result about the wire bond quality.

I manage to increase the sensitivity using micro Amps force, previously I use mV force and the device passes. Does it measure better in uA scale or higher current scale?

For a really meaningful conclusion on wire bond status, I think you would need at least tens - if not hundreds - of mA - and no pn-junction in series.

rfsystem

Joined: 25 Feb 2002
Posts: 858
Helped: 96

07 Oct 2008 19:17 Re: Wire bond quality measurement

I remember this technique for a packaged test but with MUXes designed in to verify in addition to BIST for the I/O's the reliability of the bonds.

Bad bonds show rapid electromigration under higher currents. So you need some 100mA. And here is the limit. The output driver should be bigger than typical and todays drivers are reduced slope drivers at low voltage.

It was considered as an option. I do not know if this option under many option was implemented. From the analog circuit perspective a 4-terminal measurement work.