How important are ESD mats?

[george graves]

George, you think that you've never lost a chip to static. But this can only be said if you are sure that none of your chips failed in the years after installation

I can!  But....I see what you're getting at.  I have no proof, other than my "educated guess" - That and of 1000's of *widgets* (automotive timed relay kind of thing - very basic - opamp/RC time/mosfet) that are running 24/7 since 2007,  none have failed other then water intrusion...Visible water damage (floods in the US during storm season)

But I agree it's "good practice" - I just fitted my lab with the same mat Dave uses.  Except - mine is green.  Whoops!  The leaded solder won't know what to do with one's self!

Most static damage does not result into immediate failure of components.

Not sure what to say....Most cheeseburgers don't result in heart attacks!  Do you have any more info on that? Static...not cheeseburgers. 

[shakilabanu]

I have a basic question about ESD mats - specifically, how important are they?

In my experience I've not noticed any difference with or without ESD ... in my earlier workplace there werent any and in my present work place they are used on most of the workbenches, but I havent yet noticed any difference. FYI I am in humid conditions, peninsular India.

[German_EE]

Opinions vary according to circumstances.

If it's a hobby thing then a grounded wrist strap is probably enough.

My last job was working on elevator control circuits where a product failure in a control board could be unfortunate. We therefore used wrist straps, mats AND had anti-static flooring in the workshop.

[schopi68]

Most static damage does not result into immediate failure of components.

Not sure what to say....Most cheeseburgers don't result in heart attacks!  Do you have any more info on that? Static...not cheeseburgers. 

Well electrostatic discharge typically produces punctually overheating in the semiconductors structure. This may result in an immediate failure if a connection is cut as a result of the overheating. But also it is possible that it will melt only partially. So the electrical characteristics (i.e. leakage currents) may change. This reduces the lifetime of the part. i.E. a component that had an original lifetime of 100 years may get a reduced lifetime of just 10 years or only a few weeks. For a manufacturer such a lifetime reduction may increase the returning devices during a guarantee period. It is normal that there are some failing devices, i.E. 1% during the first 2 years. If this increases to 2% due to static damage it will be a very expensive thing for larger production lines (double of the work).

One quote i found: "Analysis of non-conforming or defective devices showed that 60–75% were damaged by EOS (electrical overstress) or ESD. This rises to 90% for newer technologies. About 70% of these failures were attributed to damage from incorrectly grounded people.

Toshikazu Namaguchi, Hideka Uchida
EOS/ESD Symposium 1998 pp.245-251."

[DanielS]

Well electrostatic discharge typically produces punctually overheating in the semiconductors structure.

Overheating? With the amount of heatsinking provided by the rest of the device body, spontaneously injecting a few mJ in a relatively beefy IO pin (compared to other active elements within the rest of the IC) is not going to cause much local heating but raising a 2V gate's voltage to 20V might bust it.

I would be far more worried about dielectric breakdown within the chip or possibly blowing the clamping capacity of input/output diodes.

[george graves]

I guess that brings up the question of....how in the world would you ever test that?  Build 1 million units, and subject 10% to a static charge, and then wait 10 years for it to fail, and collect stats on it?

[schopi68]

You can do statistical projections of actual failures. Also you can do artificial aging of the devices (climate chamber).
The reason for an failure can be determined by doing microscope examination of the failed parts (and yes, i know that i.e. HP did such examinations already in the 1980s).

[schopi68]

Overheating? With the amount of heatsinking provided by the rest of the device body, spontaneously injecting a few mJ in a relatively beefy IO pin (compared to other active elements within the rest of the IC) is not going to cause much local heating but raising a 2V gate's voltage to 20V might bust it.

Here you can find a simplified description of this (and other) failure mechanisms.
www.semtech.com/images/datasheet/si96-11.pdf
As you can see almost every failure mechanism is the result of overheating/burnout.

[coppice]

Well electrostatic discharge typically produces punctually overheating in the semiconductors structure.

Overheating? With the amount of heatsinking provided by the rest of the device body, spontaneously injecting a few mJ in a relatively beefy IO pin (compared to other active elements within the rest of the IC) is not going to cause much local heating but raising a 2V gate's voltage to 20V might bust it.

I would be far more worried about dielectric breakdown within the chip or possibly blowing the clamping capacity of input/output diodes.

If you look at electron micrographs of zapped chips you seldom see pinholes punched around the I/O ring. They are normally somewhere deep in the logic.

[DanielS]

www.semtech.com/images/datasheet/si96-11.pdf
As you can see almost every failure mechanism is the result of overheating/burnout.

The document does not state a proportions between causes or links between them.

Oxide/Dielectric breakdown is listed as the first cause and is likely a factor in several instances of all three others.

[schopi68]

As described in the document dielectric breakdown also results in overheating and possible shorting. So the proportions are not relevant.

[george graves]

How about someone try to zap a part by rubbing your feet on a rug, and then do before and after tests?  Dave? 

[digital]

Why take a chance of zapping any component? It is industry practice to work in a static free environment.My last employer would instantly dismiss any employee not using static free equipment.Regards Wayne

[sotos]

How about someone try to zap a part by rubbing your feet on a rug, and then do before and after tests?  Dave?

I’ll try to do this. I will rub my feet on plastic on carpet, take a sensitive cmos chip, and touch all the pins and then test it if it’s still working.

[schopi68]

I’ll try to do this. I will rub my feet on plastic on carpet, take a sensitive cmos chip, and touch all the pins and then test it if it’s still working.

Well... if the device failes you have proven that static electricity may damage a device. If it does not fail immediately, you still do not know anything because:
- you do not know if you really had some static charge while doing the experiment
- you do not know if the chips characteristics are still in spec (without very detailed testing)
- you do not know if the chip will fail prematurely during the next months or years as a result of the discharge

Testing this way is a little bit like throwing lighted cigarettes into a forest: if it does not begin to burn you still cannot say that it will never burn. But you can trust the firefighters telling you that this sometimes will happen. 

Here is an interesting video with a real test (just ignore the product demo information  ):

In this video you will see a few snippets showing interesting facts (i am sorry, i could not find better videos with more in deep information for free):

[zapta]

My last employer would instantly dismiss any employee not using static free equipment.Regards Wayne

'last'? What happened? You didn't use static free equipment?

Just kidding.

[zapta]

[Here is an interesting video with a real test (just ignore the product demo information  ):

Wow! This is a damaged resistor. I always assumed that only CMOS and similar semi conductors are at risk.

[sotos]

What about clamping diodes, what are they for? They are for protection I know, so if somebody is charged, aren’t they for discharging him?

[schopi68]

Clamping diodes are mainly to protect the assembled device. The clamping diodes are designed to work by draining the charge into a larger external capacitor or to main's earth/the power rail. As long as the correct surrounding circuit is not connected, the charge will go anywhere... but not the intended way as this does not exist at this time.
(i.E. from the input through the clamping diode to the chips internal power rail and back into the other parts. )

This is why it is so much more important to take ESD-protection into account on disassembled devices or single components. A well designed device is able to drain/withstand normal ESD-Levels (see the Human Body Model) at the time it had been assembled. Before that its parts are susceptible for damage due to electrostatic discharge.

[Trinity]

You guys have convinced me to pick one up, and it's on the way.

As a quick followup question, I see all sorts of cleaners that claim to be especially designed for ESD mats, but is there any reason I can't just use Windex, isopropyl, or water?

[zapta]

You guys have convinced me to pick one up, and it's on the way.

Hopefully it's a rubber one ;-)

[SeanB]

Plain water works best, and if needed a drop of plain dishwash liquid followed by a rinse with plain water. Anything else runs the risk of degrading the rubber or leaving a film on it.

[zapta]

They are not expensive and last long time.

http://desco.descoindustries.com/DescoCatalog/WorksurfaceMats/MatAccessoriesCleaners/ReztoreAntistatic/10435/#.U_OgbXWx3UY

The foam makes it easier to wipe grippy rubbery mats (vs. plain water).

[schopi68]

As i can see in the product datasheet the Reztore Cleaner contains mainly water (60-100%) and isopropanol (5-25%).

As isopropanol does not affect most of the plastics used, i wouldn't have any concerns in using it.
(Isopropanol harms soft-pvc and Styrol-Acrylnitril as described here on page 15 - sorry for using the wrong language document - http://www.kuhnke.de/fileadmin/templates/content/Automation/Branchen/Medizintechnik/764343chemische_bestaendigkeit.pdf ). 

A mixture of at least 50% water and 50% isopropanol has a better cleaning result than using it pure.

[Trinity]

Hopefully it's a rubber one ;-)

3M multi-layer rubber, yes.