Electromigration:- basic rule to be followed

Electromigration

Can anybody tell me the typical electromigration rule for metal 1 in the RF analog design? Thanks.

Re: Electromigration

I use 1micron width for 1mA .
But it varies from technology to technology.
you can refer design rule manual in order to get the current carrying capability of individual metal.

is this the answer you are looking for?

--Sree

Re: Electromigration

sree_lakshmi said:

... 1micron width for 1mA .

Click to expand...

Depends very much on your expected max. chip temperature. At automotive max. (150 .. 175°C) this current density value may decrease by an order of magnitude.

Re: Electromigration

sree_lakshmi said:

I use 1micron width for 1mA .
But it varies from technology to technology.
you can refer design rule manual in order to get the current carrying capability of individual metal.

is this the answer you are looking for?

--Sree

Click to expand...

Thank you. It is exactly what I am looking for.

Added after 5 minutes:

erikl said:

sree_lakshmi said:

... 1micron width for 1mA .

Click to expand...

Depends very much on your expected max. chip temperature. At automotive max. (150 .. 175°C) this current density value may decrease by an order of magnitude.

Click to expand...

Very helpful. Thank you.

Electromigration

AC electromigration is almost nil at radio-range frequencies.
Your foundry should provide you an AC current density limit.
Even a low frequency, balanced reversing waveform should
be 10X the DC limit at least.

But foundries are not always interested in these finer
points, since they are not on the hook for the efficiency
of your layouts.

Electromigration

it is very much temperature dependent....... same metal layer handle 2mA/um @ upto 110 degree......... but only 0.8mA/um @125 degre C..

reference : UMC90nm.

Hi,

Can anyone explain abt the relation b/w temp & current density which plays a role in Electromigration in detail.

My undertanding is that, when Temp increases - delay increases (Ignore Temperature inversion concept), so the resistance will be more as delay increase, hence the flow of current will be slow.

I accept, Increase in voltage increases Curent flow.

Thanks
Kicha

Kicha10 said:

Hi,

Can anyone explain abt the relation b/w temp & current density which plays a role in Electromigration in detail.

My undertanding is that, when Temp increases - delay increases (Ignore Temperature inversion concept), so the resistance will be more as delay increase, hence the flow of current will be slow.

I accept, Increase in voltage increases Curent flow.

Thanks
Kicha

Click to expand...

[Trying]

In Simple terms - current density plays the key role - as that relates directly to the probability of a Cu or Al atom gets a kick to - every collision leads to finite momentum transfer -> more current density --> more number of electrons [finite momentum transfer] per second. Second important is when a Cu or Al atom gets the momentum transferred - what does the molecule do? are these dynamics phonon oscillation, linear motion, more energy loss - temperature dependent?

It not only has to do with how electrons bombard/collide with conductor atoms/molecules, [surface-scattering and grain-boundary scattering], it has to do with increased phonon-scattering and conductor atoms become more mobile as the temperature increases.

The EM problem is somewhat cumulative in nature - in the sense - as weaker [necking] regions develop in the conductor tracks - the rate of increase of all 3 scattering increases right at the weaker spot. The weak region offers more resistance, and more IR drop, and more current density compared to other regions. Temperature acts as a catalyst to open the interconnect majorly. I am not saying to disregard other factors - that adds to make it even worst as the thermal conductivity on low-k porous dielectric + mechanical stress at the Cu & liner boundary and so on...

Thanks Sat for the reply ))