Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Register Log in

Wire bond quality measurement

Status
Not open for further replies.
R

rikie_rizza

Full Member level 6
Joined
Aug 5, 2006
Messages
331
Helped
29
Reputation
58
Reaction score
8
Trophy points
1,298
Location
Bikini Bottom, between a rock and a pineapple
Activity points
3,173
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
 

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
 

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.
 

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 said:
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.
Click to expand...

Sure, that's a pity :-((
With excellent X-ray equipment, however, one can differentiate even between Si & Al .
 

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.
 

rfsystem said:
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.
Click to expand...

Ingenious technique, but won't be quite easy at a packaged device! ;-)
 

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?
 

rikie_rizza said:
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?
Click to expand...

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.
 

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.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top