Time to thermally damage industrial ICs

[bacon4life]

For industrial ICs, does short term operation above 85C immediately damage the chip? Or do ICs follow the rule of thumb that every 10°C increase in temperature reduces life by half? I have a circuit board that might occasionally be exposed to temperatures above 85C. I am trying to understand the appropriate temperatures for setting the high temperature alarm and the high temperature trip. After receiving an alarm, it may take up to a week to investigate and remedy the high temperature situation.

 

[MacGyverS2000]

There's no way to tell with any general answer... it will be highly chip-dependent. Junction temp is what you really care about, not necessarily ambient temp... but letting a chip cook for a week is a recipe for disaster.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[3DDave]

The chips are exposed above that temperature for a short while when they are soldered - however there is no power applied.

I expect the failure is because there is increased migration of dopants under the electrical field during operation which will alter the fundamental characteristics of structures on the chip. This migration is dependent on the exact materials involved, the time, the temperature, and the current/voltage being applied.

If it's critical, run a test. There cannot be a rule of thumb that applies.

 

[IRstuff]

In general, the answer is no, since the relative reliability is modeled by the Arrhenius equation, which is an exponential with the absolute temperature and the activation energy of the defect(s). The caveat is that every type of defect has a different activation energy, and there are already latent defects on each chip to start with. If you're unluckily, one of those defects might reside on a transistor or contact, and a small overtemp will cause it to kill that part of the chip, while if you are lucky, the all the defects will be relatively far from anything active.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[itsmoked]

What IR sez is why they 'burn-in' parts that go into critical systems. A company may make a part that's rated 85C but someone wants them to work at 125C. If there is any possibility of them running that hot you can screen them by burning them in. Some will fail.

The standard engineering methodology is to make sure the parts don't exceed the published temperatures then no crap-shoot is in play.

You might check to see if that part is offered in 105 or 125. If so the odds of yours taking brief overages is more likely. A crap-shoot but with better odds. If none are available at those higher ratings I wouldn't go there on anything that would be a big problem if it choked.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

One option is to add heat sinks to the parts. Often, as I personally can attest to, the ambient directly adjacent to a part could be considerably hotter, but a heat sink could possibly reach and receive slightly cooler air that's closer to the apparent ambient temperature, in addition to providing some transient ride-through capability, since the thermal mass of the heat sink will stabilize the chip temperature a little bit.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[bacon4life]

Thanks everyone for the input.

The component doesn't generate much heat, so it will only be exposed to high temperature on the very rare occasion the overall enclosure overheats due to problems with other equipment in the enclosure. Measuring temperature is a new feature of the component, so even a very high trip setting will be provide an improvement.

 

[IRstuff]

See:

https://www.electronics-cooling.com...perature-really-reduce-life-electronics-half/

note that "the" activation energy is actually many:

https://www.eesemi.com/activation-energy.htm.

Only a failure with 0.8 eV/K activation energy results in a halving of the life with 10C change.

You can calculate a worst-case acceleration factor for the devices, assuming a plausibly worst-case activation energy, and a plausible lifetime exposure to some plausible higher temperature

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[waross]

Using a Hi-Temp alarm to kill all power to the chip will probably let it ride through an over temperature event.
When the chip is carrying current, the internal temperatures will be higher than the ambient.
Consider taking Keith's advice. Through a handful of chips into an oven and bake them at your target temperature for a few hours.
If all survive, you are probably safe at that temperature with the chips unpowered.
What matters more than the ambient temperature is the internal temperatures.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[Comcokid]

Down-hole oil well instruments sometimes use standard ICs that have been carefully qualified and the change in specs is carefully understood. Temperatures can reach 150C or higher. Because of the expense of shutting down a well to put an instrument down and get measurements such instruments might be rated for a single use. A secondary market exists for used instruments for smaller outfits willing to save money on the instrument and take on the risk of reuse.

 

[LionelHutz]

I'd think it is a crap shoot. You can calculate and speculate but in the end there are factors that will make it a crap shoot.

We had some qualifying testing done on a circuit board. After the normal test levels, they ran the boards up to 200*C while powered and operating a few times. As the temperature rose, the various parts of the board would shut down and stop working. The Y capacitor on the control power input failed on one sample but otherwise the boards all worked again once the temperature was brought back down.

 

[MacGyverS2000]

they ran the boards up to 200*C while powered and operating a few times. As the temperature rose, the various parts of the board would shut down and stop working.

I would be interested to know what material was used for the PCB... even for high-temp materials, the glass transition temp is in the mid- to high-100's, so unless you were working with a really good resin, that board was starting to flow at such a high temp (and that doesn't include the added temp of whatever component was operating on top of it). Definitely not a good environment to try and run a board in.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

I agree with the notion that calculation is very speculative, particularly since the activation energies aren't always precisely known and their relative contribution to chip failure aren't always known. So, you could plug in a 1-eV/K activation energy and calculate that 150C will result in reducing MTBF from 50,000 hr to 35 minutes, there's no guarantee that 1 eV correctly characterizes your system's aggregated defects

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[bacon4life]

Wow! I had not realized folks ran normal components at such high temperatures. That makes me a bit more comfortable using 85C as the alarm. After getting getting an 85C alarm, the whole board is going to be replaced when the overall apparatus is serviced within a week. If the board simply stops working above 85C, the situation will be similar to the legacy installation that lacks any thermal monitoring.

The board draws less than 0.3 watts, so I assumed the components would be quite close to the overall temperature of the enclosure. Chips are much smaller than when I last hand soldered something, so I may not have a good sense of how much heat a modern component can dissipate.

This is a COTS (commercial off the shelf) board from a vendor, so specific testing of ICs is not an option for me.

 

[MacGyverS2000]

bacon, to run components at those temps, the components need to be designed for those temps... no one is running a commercial part at 200C and expect it to work. Excepting unique parts, military grade is about the best you're going to get, and that only gets you to ~125C. Anything beyond that is a complete crap shoot on whether it will continue to work at all, forget about affecting its lifetime. My experience has been much about 125 and things just start going off the rails... usually +10C for more than 5-10 minutes and most chips just start failing completely (though they may come back to life after temps drop significantly... but that's in no way guaranteed).

Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

High temperatures manifest two possibilities, failure or performance degradation. The latter simply means that the part no longer meets its performance specification, but might otherwise be fully functional. However, if the rest of the system is not designed for a slower part, for example, then the system will appear as if it failed. The former is where hard failures can occur, but it may be soon, or it might be later; nevertheless, high temperature excursions, even if the part seemingly survives, might have induced latent damage that might show up later on, even if the part remains within its proper operating temperature.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[LionelHutz]

No, there was no expectation of the board working at 200*, they just wanted to see what happened to it doing that.